Content uploaded by Tanja Wintrich
Author content
All content in this area was uploaded by Tanja Wintrich on Apr 23, 2020
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tjsp20
Journal of Systematic Palaeontology
ISSN: 1477-2019 (Print) 1478-0941 (Online) Journal homepage: https://www.tandfonline.com/loi/tjsp20
A new cymbospondylid ichthyosaur
(Ichthyosauria) from the Middle Triassic (Anisian)
of the Augusta Mountains, Nevada, USA
Nicole Klein, Lars Schmitz, Tanja Wintrich & P. Martin Sander
To cite this article: Nicole Klein, Lars Schmitz, Tanja Wintrich & P. Martin Sander (2020):
A new cymbospondylid ichthyosaur (Ichthyosauria) from the Middle Triassic (Anisian)
of the Augusta Mountains, Nevada, USA, Journal of Systematic Palaeontology, DOI:
10.1080/14772019.2020.1748132
To link to this article: https://doi.org/10.1080/14772019.2020.1748132
View supplementary material
Published online: 20 Apr 2020.
Submit your article to this journal
View related articles
View Crossmark data
A new cymbospondylid ichthyosaur (Ichthyosauria) from the Middle Triassic
(Anisian) of the Augusta Mountains, Nevada, USA
Nicole Klein
a
, Lars Schmitz
b,c
, Tanja Wintrich
a,d
and P. Martin Sander
a,c
a
Division of Paleontology, Institute of Geoscience, University of Bonn, Nußallee 8, 53115 Bonn, Germany;
b
W. M. Keck Science
Department, Claremont McKenna, Scripps, and Pitzer Colleges, Keck Science Center, 925 N. Mills Avenue, Claremont,
CA 91711-5916, USA;
c
Dinosaur Institute, Natural History Museum of Los Angeles County, 900 Exposition Boulevard, Los
Angeles, CA 90007, USA;
d
Institute of Anatomy, University of Bonn, Nussallee 10, 53115 Bonn, Germany
(Received 11 March 2019; accepted 24 March 2020)
A new taxon, Cymbospondylus duelferi sp. nov., is described from the late Anisian of the Fossil Hill Member of the Favret
Formation of the Augusta Mountains, Pershing County, Nevada, USA. The holotype and only specimen consists of a fairly
complete skull associated with articulated and disarticulated postcranial material. Body length reconstruction revealed a
medium-sized individual with an estimated body length of 4.3 m. Dorsal vertebrae as well as the left humerus of this
specimen are associated with three strings of articulated tail and posterior dorsal vertebrae, which are on average 68% smaller
than the vertebrae of the medium-sized individual. Due to the small size of these vertebrae and their position within the trunk
region, they most likely represent fetuses. Cymbospondylus duelferi sp. nov. provides the second-oldest evidence for
viviparity in ichthyosaurs. In the course of this study, the skull morphology of C. petrinus and C. nichollsi was reviewed,
resulting in revised character scorings for both. Phylogenetic analyses demonstrate a sister-taxon relationship of the new taxon
with C. petrinus, which was found in the same stratigraphic unit, the Fossil Hill Member, that also crops out at the classical
Fossil Hill locality in the Humboldt Range, where the member is part of the Prida Formation. Cymbospondylus duelferi sp.
nov. shares a very similar skull morphology with C. petrinus and C. nichollsi, whereas the shoulder girdle morphology
differs. Because there is no evidence for ontogenetic differences or sexual dimorphism in the new specimen, a third species
of Cymbospondylus –neglecting the type species C. piscosus that is only known from five vertebrae –is recognized from the
Fossil Hill Member. Further occurrences of the genus in the Lower and Middle Triassic of Svalbard (Boreal Ocean) and
Europe (Tethys) point to a very fast radiation and dispersal of cymbospondylids during the Middle Triassic.
https://zoobank.org:pub:906F5F32-9090-496D-8FC7-680D25EED5EF
Keywords: skull morphology; phylogeny; new species; fetus; cymbospondylid diversity
Introduction
Ichthyosaurs appeared by the Early Triassic (Olenekian/
Smithian), as documented by occurrences from Japan
(Utatsusaurus –Shikama et al. 1978), China
(Chaohusaurus –summarized in Zhou et al. 2017),
Svalbard (Grippia –Wiman 1929) and North America
(Utatsusaurus and Grippia in British Columbia, Canada
–Cuthbertson et al. 2013; cf. Grippia and cf.
Utatsusaurus in Nevada, USA –Kelley et al. 2016),
demonstrating a widespread distribution at least in the
Northern Hemisphere. Data from the Southern
Hemisphere are limited due to the lack of adequate mar-
ine localities. Most of this material is disarticulated and/
or highly fragmentary, except for the basal ichthyosauri-
form Chaohusaurus from China. Chaohusaurus is so far
the best known of any early ichthyosauriform (Zhou
et al. 2017), and among these specimens there is a
pregnant female (Motani et al. 2014), which represents
the oldest known evidence of viviparity in ichthyosaurs.
In addition, the large number of individuals of
Chaohusaurus known from a single locality allowed for
study of morphological variation in the light of sexual
dimorphism (Motani et al. 2018).
Nevada yields one of the most diverse Middle
Triassic ichthyosaur faunas. Fossiliferous sediments of
the Fossil Hill Member, which is part of the Prida
Formation that crops out in the West Humboldt Range
of central Nevada, yield the classical Middle Triassic
ichthyosaur fauna described by Merriam (1908). The
Fossil Hill Member is also part of the Favret Formation,
which crops out in the Augusta Mountains (Nichols &
Silberling 1977) farther east. In the Augusta Mountains
the sediments of the Fossil Hill Member are much
thicker (up to 300 m) compared to the Humboldt Range
Corresponding author. Email: nklein@posteo.de
#The Trustees of the Natural History Museum, London 2020. All rights reserved.
Journal of Systematic Palaeontology, 2020
http://dx.doi.org/10.1080/14772019.2020.1748132
Published online 20 Apr 2020
and consist of black shales, indicating anoxic or dysoxic
bottom waters (Bucher 1988;Fr
€
obisch et al. 2013).
The fauna of the Fossil Hill Member in the Favret
Formation (Augusta Mountains) includes numerous and
abundant invertebrates (i.e. Ammonoidea, Daonella,
belemnites; Bucher 1988) whereas fishes
(Chondrichthyes, Actinopterygii, Coelacanthidae) are
rare (Fr€
obisch et al. 2013). The most common vertebrate
fossils are ichthyosaurs: the mixosaurian ichthyosaurs
Phalarodon callwayi and P. fraasi (Sander & Bucher
1990; Schmitz et al. 2004; Schmitz 2005; unpublished
material collected in 2005–2017), the medium-sized
Omphalosaurus (unpublished material collected in
2011), which is possibly an ichthyosaur as well, the
large Cymbospondylus nichollsi (Fr€
obisch et al. 2006)
and the giant Thalattoarchon, an apex predator with
large cutting teeth (Fr€
obisch et al. 2013). Sauropterygia,
while very common in the Tethys, are represented only
by a single individual, of the pistosauroid
Augustasaurus (Sander et al. 1997; Rieppel et al. 2002).
The most common ichthyosaur in the Fossil Hill Member
is Cymbospondylus petrinus, which was described in detail
by Merriam (1908). Another species of Cymbospondylus,C.
nichollsi, from the Fossil Hill Member was erected substan-
tially later (Fr€
obisch et al.2006) based on a new find con-
sisting of an incomplete skull articulated to the anterior trunk
region. Cymbospondylus has also been described from the
Early and Middle Triassic of Svalbard (Sander 1992;
Engelschiøn et al. 2018), from the Middle Triassic of the
southern Alps (C. buchseri from the Monte San Giorgio
locality, Switzerland: Sander 1989) and from the Middle
Triassic of Germany (Germanic Basin: Huene 1916;Sander
1989), suggesting –on the basis of the currently known fos-
sil record –a Northern Hemisphere distribution.
Cymbospondylus was a large pelagic form that must have
reached around 9.3 m in body length (Merriam 1908), with
numerous small, conical teeth indicating a diet consisting of
fish and cephalopods (Rieber 1970; Massare 1987).
Continued fieldwork over the last decade in the Augusta
Mountains has produced a new cymbospondylid ichthyosaur
specimen with three partially preserved, articulated vertebral
columns of fetuses that is described herein. The specimen
wasbrieflymentionedbyFr
€
obisch et al. (2013), and the
histology of the vertebral centra of both the mother and
fetuses was described in Houssaye et al. (2018).
Material and methods
Institutional abbreviations
FMNH: Field Museum of Natural History, Chicago,
Illinois, USA; LACM: Natural History Museum of Los
Angeles County, Los Angeles, California, USA;
UCMP: University of California Museum of
Paleontology, Berkeley, California, USA.
Discovery
LACM DI 158109 is a partial skeleton that was found dur-
ing the 2011 field season by one of the authors (PMS) and
excavated in 2014. The skeleton was originally contained in
a large early diagenetic carbonate concretion, part of which
was heavily damaged by weathering; another part was lost
to weathering. Early diagenetic concretion formation ensured
three-dimensional preservation of the skeleton with little
crushing. The concretion was bisected by an east-west trend-
ing fault. The northern half was downthrown by several cen-
timetres. Combined with the southerly exposition of the
slope, this led to the partial destruction of the southern,
uplifted half of the concretion. The sediment encasing the
concretion is a greyish-brown shale, whereas the concretion
consists of dense, dark grey limestone with abundant ammo-
noids. These ammonoids released partially liquid hydrocar-
bon from their phragmocones upon preparation, as also
described by Nichols & Silberling (1977).
The northern part of the concretion is continuous but, dur-
ing collection, separated into several blocks (blocks #1, 3–5;
Fig. 1A). These blocks comprise a nearly complete right half
of the skull exposed in ventrolateral view (block #1; Fig. 2)
associated with somewhat disarticulated cervical vertebrae
and ribs (block #3) as well as the right shoulder girdle (Figs
1A, 3A, C) and a right articulated humerus, radius and ulna
exposed in ventral view (block #4; Figs 1A, B, 4A, 5A, D).
Towards the posterior, the skeleton is highly incomplete due
to the fault, which had uplifted these parts, exposing them to
weathering. Here, only some strings of dorsal vertebrae, dis-
tal ends of mid-dorsal ribs and parts of gastral ribs are asso-
ciated with the main concretion (block #5; Fig. 1A). The
fault also extends across the skull, cutting off the dorsal parts
of the right upper temporal opening and most of the left half
of the skull. Two of the isolated blocks found in the scree on
the slope below the main concretion contain additional ele-
ments of the skull, i.e. the right upper temporal region (block
#2) and the ventral part of the ?right parietal.
Parts of the dorsal vertebral column, as well as the com-
plete left humerus and the incomplete left coracoid, were
found preserved in numerous small blocks from the scree
which are the remains of the southern part of the concre-
tion (Fig. 1A). Some of the scree blocks could be joined
together to form larger blocks (e.g. block #6; Figs 1A,5),
but no connection could be found with the main concre-
tion except for block #2, which contains the skull roof
(Fig. 1A) and connects with block #1 (the main skull
block). However, it is clear from the distribution of the
bones, their size and the field situation that all blocks rep-
resent the same individual. The scree block containing the
left humerus and coracoid (block #6) shows on its reverse
2 N. Klein et al.
Figure 1. Field top view of all collected blocks of the specimen (LACM DI 158109), the holotype of Cymbospondylus duelferi sp.
nov., with the component blocks labelled. A, blocks of LACM DI 158109 assembled as found during excavation and before detailed
preparation. The top of the pictures shows the northern part of the concretion where blocks still fit together. Numbers indicate block
numbers. From left to right: (block #1) skull in ventrolateral view; (block #2) the left upper temporal region; (block #3) the
disarticulated neck region with cervical vertebrae and cervical ribs; (block #4) parts of the shoulder girdle, humerus, radius and ulna
of the right side of the body; (block #5) ribs and gastral ribs of the trunk region; (block #6) the left humerus and distal part of the
left coracoid in dorsal view, of which the reverse side shows three strings of small vertebrae (see Fig. 5) as well as dorsal vertebrae
and ribs of the mother; and (block #7) vertebrae and ribs of the mother as well as gastral ribs. B, partially articulated right forearm
with some disarticulated and incomplete autopodial elements and the distal part of the scapula and coracoid after preparation in
ventral view (block #4 in Fig. 1A). C, left humerus and distal part of left coracoid in dorsal view (block #6 in Fig. 1A). D, isolated
phalanx from block #5.
A new cymbospondylid ichthyosaur 3
side two strings of articulated vertebrae that belong to much
smaller individuals (Figs 1C,5). Another string of small
articulated vertebrae in block #6 is closely associated with and
parallels an articulated string of larger vertebrae. Due to the
size of the smaller vertebrae, which is roughly 32% of the
size of the larger vertebrae, and their position within the trunk
region of the larger individual, these three strings of small ver-
tebrae are interpreted as the remains of three fetuses.
Preparation
The blocks of the main concretion were prepared
mainly from the field top except for the skull, which was
exposed from all sides in one large block (Fig. 2B, D, E),
hereafter called the main skull block. One scree block
preserves the right upper temporal region in visceral
view. After a cast was taken of this surface, the block
was prepared down to the outside of the bones surround-
ing the temporal opening (Fig. 2A, C). A fit of this scree
block with the main skull block is possible (Fig. 2A–E)
based on two features. One is the posterior-most lateral
part of the squamosal, which is preserved as an imprint
on the scree block. The other feature is two distinct
calcite veins that are continuous in the main skull block
and the scree block across the upper temporal region.
The other skull block containing the ventral part of the
parietal is, however, too fragmentary to show a fit.
Body length estimations
In the past, skull proportions had been used to estimate
body length in incomplete Cymbospondylus skeletons
(Sander 1992;Fr
€
obisch et al. 2006): however, these
might not yield reliable body length estimates because it
is unlikely that skull length scales isometrically with
total length. Scheyer et al. (2014) plotted body length vs
humerus length of Triassic ichthyosaur skeletons for
which body sizes are known, resulting in a well-supported
regression line (Scheyer et al. 2014,fig.3;R
2
¼0.9899).
Adding the humerus length (117 mm; see Table 1)
of the mother (LACM DI 158109a) to this regression
results in a body length estimate of around 4.3 m.
Systematic palaeontology
Class Reptilia Linnaeus, 1758
Subclass Diapsida Osborn, 1903
Order Ichthyosauria de Blainville, 1835
Genus Cymbospondylus Leidy, 1868
Genus diagnosis based on C. petrinus according to
McGowan & Motani (2003). Anterodorsal margin of
scapula emarginated; anterior flange of humerus slightly
concave; 55 or more presacral vertebrae; carpals round;
metacarpals retaining shaft, at least partially; probably
pentadactyl; diapophyses reaching anterior margin of
vertebral centra; pronounced deepening of amphicoelous
excavation of vertebral centra near centre.
Genus diagnosis according to Fr€
obisch et al. (2006).
Medium- to large-sized ichthyosaur; oval to egg-shaped
orbit; nasal extends far posteriorly; large anterior tem-
poral terrace; nasal and frontal reach anterior margin of
temporal opening; back of the skull formed by a large,
paired skull bone (postparietal?, neomorph?), positioned
posteriorly to the parietal; occipital condyle concave;
elongate, anteriorly slanted diapophyses truncated by
anterior margin of centrum in middle and posterior dor-
sal region; ribs single-headed, except in cervical region;
coracoid with coracoid foramen.
Table 1. Skull and postcranial measurements (mm) of LACM
DI 158109.
Feature Measurement
Length of skull as preserved 610
Estimated total skull length 650–700
Skull height >94.5
Skull length posterior to
nares/postnarial length
314
Skull length anterior to
nares/prenarial length
>245
Preorbital length as preserved 335
Length of cheek region 107
Length of orbit 140
Height of orbit 75
Postorbital length 120
Length of upper temporal opening 150
Width of upper temporal opening 59.2
Length of lateral embayment 70
Width of lateral embayment 47.8
Length of maxilla 190
Preserved length of lower jaw >635
Postdentary length of lower jaw 225
Number of scleral plates 12
Length of humerus (left/right) 132/120
Length of radius 59.7
Length of ulna 60.5
Coracoid –mediolateral width 99
Coracoid –anteroposterior length 73
Scapula –mediolateral width 81
Scapula –anteroposterior length 91
Vertebrae of mother
Cervical (height) 37
Length of four dorsal centra 27.4/26.6/24.4/24
Height and length of six dorsal centra 35.8 21.5
44.4 22.5
41.4 22.5
45.6 20.1
47.1 25.2
44.0 22.0
4 N. Klein et al.
Apomorphy-based diagnosis of the Cymbospondylus
clade from Nevada based on this study. In both analy-
ses in TNT (Traditional Search and New Technology
Search), the synapomorphy unifying C. nichollsi and C.
petrinus with C. duelferi is character 190 (state 0), i.e. the
absence of a medial facet for the scapula at the coracoid,
which is a derived state. The following autapomorphies
are present in Cymbospondylus from Nevada: C. petrinus
and C. duelferi both have tooth rows that end below the
external naris. Cymbospondylus petrinus,C. nichollsi and
C. duelferi share the boomerang-shaped lacrimal, the rela-
tively large scleral rings and eye socket, and the squamosal
and postorbital being excluded from the upper temporal
fenestra by a posterior process of either the nasal or the
postfrontal. The European C. buchseri is excluded from
the Nevada clade, making a revision of the genus
Cymbospondylus necessary, which is, however, beyond the
scope of the current paper.
Remarks. In the generic diagnosis of McGowan &
Motani (2003, pp. 63 and 146), the first two characters
(“anterodorsal margin of scapula emarginated; anterior
flange of humerus slightly concave”) are apomorphies
found in the phylogenetic analysis of Motani (1999).
According to the diagnosis by McGowan & Motani
(2003), the new specimen LACM DI 158109 pertains to
Cymbospondylus because it shows the character “anterior
flange of humerus slightly concave”but differs in its fan-
shaped (and not emarginated) anterodorsal margin of the
scapula. The remaining characters in their diagnosis were
provided “solely for convenience”(McGowan & Motani
2003, p. 63). Among these is one present in the new speci-
men: “carpals round”. The remaining characters listed in
the diagnosis of McGowan & Motani (2003) cannot be
observed in the new specimen due to its state of preserva-
tion: “55 or more presacral vertebrae; metacarpals retain-
ing shaft, at least partially; probably pentadactyl;
diapophyses reaching anterior margin of vertebral centra;
pronounced deepening of amphicoelous excavation of
vertebral centra near center”.
AsnotedbyFr
€
obisch et al. (2006), their diagnosis of the
genus Cymbospondylus represents a synthesis of Motani
(1999), Sander (2000), Maisch & Matzke (2000)and
McGowan & Motani (2003), as well as their own observa-
tions. Except for the reinterpretation of the large, paired skull
bone presented therein (see below), this diagnosis remains
applicable. Following the diagnosis by Fr€
obisch et al.
(2006), LACM DI 158109 pertains to Cymbospondylus
because it is a medium-sized to large ichthyosaur with an
oval to egg-shaped orbit. Its nasal extends far posteriorly,
there is a large anterior temporal terrace, the nasal and
frontal reach the anterior margin of the temporal opening,
the occipital condyle is concave and the ribs are single-
headed, except in the cervical region. A character related to
the diapophyses of the middle and posterior dorsal vertebrae
cannot be observed in LACM DI 158109 due to its state of
preservation. The only character of Fr€
obisch et al. (2006)
clearly absent from the new specimen is a coracoid with a
coracoid foramen.
In conclusion, the diagnosis of McGowan & Motani
(2003) is of limited use in determining the affinities of
the new specimen, with one apomorphy supporting
inclusion in the genus Cymbospondylus and the other
contradicting it. Because the other characters in this
diagnosis were provided “solely for convenience,”they
should not be used to identify the new specimen. The
diagnosis by Fr€
obisch et al. (2006), on the other hand,
provides strong support for including the new specimen
in the genus Cymbospondylus. Finally, one synapo-
morphy found in the current study also assigns the spe-
cimen to this genus (see above). In addition, there are
other potentially diagnostic skull characters in
Cymbospondylus specimens, such as the nasal reaching
so far posteriorly as to establish contact with the supra-
temporal (Figs 2C, D,6). However, a detailed revision
of the genus, including a revised data set for testing its
ingroup relationships, is beyond the scope of this study.
Cymbospondylus duelferi sp. nov.
(Figs 1–5)
Etymology. The species is named in honour of our pre-
parator Olaf D€
ulfer, for his many practical contributions
to Mesozoic marine reptile research.
Diagnosis. Cymbospondylus duelferi is unique in pos-
sessing the following combination of characters derived
from the phylogenetic analysis in this study: the post-
orbital skull length compared to the length of the orbit
is shorter than in C. petrinus and C. nichollsi but not as
short as in mixosaurs (character 84, state 0; see Tables
1,2); the rib midshaft cross sections are oval, rather
than figure-eight shaped (character 178, state 0); and the
coracoid possesses an anterior notch (character 185,
state 0), a longer anterior extension and a shorter poster-
ior one (character 187, state 0), and very small glenoid
and scapular facets (character 189, state 0).
Differential diagnosis. The following characters distin-
guish Cymbospondylus duelferi from other species of
Cymbospondylus. The length of the postorbital skull
compared to the orbital diameter is less in C. duelferi
than in C. petrinus (character 84, state 0). The partici-
pation of the splenial in the mandibular symphysis is
present but restricted to the posterior half in C. duel-
feri, whereas the dentary contributes more extensively
in C. petrinus (character 122, state 1).
Cymbospondylus duelferi has a sub-thecodont dentition
A new cymbospondylid ichthyosaur 5
(implantation with labially open pits for functional/
large teeth), whereas the dentition for C. petrinus is
described as thecodont (character 130, state 0).
Cymbospondylus duelferi has prominent grooves and
ridges basal to the crown in the apical third of the
root, but striations are not externally visible basal to
the crown in the apical third of the root in C. petrinus
(character 145, state 1). The maxilla is shorter and the
number of scleral plates (12) is lower in C. duelferi
compared to that in C. petrinus and C. nichollsi (14
and 16–18, respectively). The number of tooth posi-
tions is lower in C. duelferi (minimum number of
functional teeth is 16), even when the missing tip of
the snout is taken into account, than in C. petrinus
(30–35). The scapula of C. duelferi has no prominent
acromion process, in contrast to that of C. petrinus
(character 192, state 1). The anterodorsal margin of the
scapula is fan-shaped in C. duelferi, but emarginated
in C. petrinus (character 193, state 0). The size of the
glenoid contribution is at least as large as the coracoid
facet in C. duelferi but smaller in C. petrinus (charac-
ter 194, state 0). The humerus has a protruding delto-
pectoral crest in C. duelferi, which is absent in C.
petrinus (character 204, state 1; character 205, state 0).
The posterior margin of the ulna is convex in C. duel-
feri but notched in C. petrinus (character 223, state 2).
Holotype. LACM DI 158109, an incomplete and par-
tially disarticulated skeleton preserving an anteriorly
incomplete skull embedded in ventrolateral view, associ-
ated with shoulder girdle elements, parts of both fore-
limbs, ribs and vertebrae. The specimen also includes
three partial vertebral columns of probable fetuses
located in the trunk region of the large individual.
Horizon and locality. Fossil Hill Member of the Favret
Formation, Favret Canyon, Augusta Mountains,
Pershing County, Nevada, USA. The type locality,
LACM 8031, is on the northern slope of the rear of
Favret Canyon at an altitude of 1840 m. Exact coordi-
nates are on file at LACM. LACM DI 158109 originates
from the same general level as the holotype of C. nich-
ollsi (Fr€
obisch et al. 2006) and Phalarodon callawayi
(Schmitz et al. 2004) in the upper third of the Fossil
Hill Member, and pertains to the Rotelliformes or Meeki
zones, both of which are of late Anisian (Middle
Triassic) age.
Description
Skull and mandible
General comments. The preserved right half of the
skull is still articulated with the right lower jaw. The
Table 2. Comparison of measurements (mm), ratios and percentages, and element counts in Cymbospondylus duelferi (LACM DI
158109), C. petrinus (UCMP 9950) and C. nichollsi (FMNH PR2251). Total skull length is estimated for C. nichollsi and C.
duelferi.Abbreviations: nm, not measurable; UTF, upper temporal fenestra.
LACM DI 158109 C. nichollsi C. petrinus
Body length 4300 7600 9300
Skull length >610 (650) 975 1166
Total skull length based on % of postnarial
length of C. petrinus (1170/530)
684 978 1170
Humerus length 117 nm 325
Ratios
Skull length/preorbital length 1.78 nm 1.63
Skull length/postorbital length 6.2 4.97 5.69
Skull length/orbital length 5 4.9 5.9
Skull length/UTF length 4.3 4.3 6.1
Skull length/maxilla length 3.4 nm 2.84
Postnarial length/postorbital length 2.95 2.26 2.56
Postnarial length/maxilla length 1.63/61.3% nm 1.54/65%
Postnarial length/UTF length 2.06/48.4% 1.95/51.24% 2.8/35.85%
Postnarial length/orbital length 2.2 2.2 2.7
Postnarial length/length of cheek region 2.93 1.85 2.11
Orbital length/orbital height 1.73 1.56 1.72
Orbital length/postdentary length 0.9 0.55 0.62
Orbital length/postorbital length 1.17 0.7 0.9/0.64
Orbital length/preorbital length 0.38 nm 0.28
Orbital length/maxilla length 0.74 nm 0.5
Preorbital length/maxilla length 1.93/52% nm 2.08/48%
Orbital length/length of cheek region 1.31 0.83 0.79
Distance between anterior orbit margin and external naris/preorbital length 0.19 nm 0.06
Number of scleral plates 12 14 16–18
6 N. Klein et al.
left side of the skull and left lower jaw are not pre-
served. The tip of the long and narrow snout was lost to
weathering (Fig. 2B, D). From the taper of the snout, it
is clear that not much more than a few centimetres is
missing (see Table 1). The ventral side of the preserved
right skull half provides insights into the morphology of
the medial side of the right lower jaw as well as into
the palate. In this view, the basioccipital and basisphe-
noid are exposed, which are the only visible elements of
the occipital region.
In lateral view, the preorbital region forms an elong-
ate and slender rostrum that is dominated by the pre-
maxillae and nasals, as is typical for ichthyosaurs. The
maxilla is relatively long (51% of estimated preorbital
length). The maxillary tooth row reaches posterior to the
external naris. The total number of preserved teeth in
the upper and lower jaw is low when compared to that
preserved in C. petrinus. The postorbital region (38% of
postnarial length) is shorter than the orbit (45% of post-
narial length) and less than one-third of the preorbital
length (>107% of postnarial length) (Tables 1,2).
LACM DI 158109 preserves 12 well-developed scleral
plates outlining the eyeball which, however, did not fill
the entire orbit, because the scleral ring is surrounded
by sediment except for its dorsal-most part (Fig. 2B, D).
The orbit is nearly as long as the upper temporal open-
ing (48% of postnarial length) in an anteroposterior
orientation (Table 1). The anterior orbital margin is
irregularly concave. The upper temporal opening is ante-
romedially bordered by the parietal, anterolaterally by
the frontal, laterally by an elongated rectangular element
(see below), and posterolaterally as well as posteriorly
by the supratemporal. An anterior terrace (a flat-floored
depression in front of the upper temporal fenestra: sensu
Motani 1999) was present, although it is now largely
incomplete due to damage by faulting. The terrace is
formed medially by the dorsally ascending parietal and
laterally by the small frontal (Fig. 2B, D). The terrace is
laterally bordered by the posterior process of the nasal
and medially by the ascending parietals. The parietals
seem to have formed a low sagittal crest. The lower
temporal embayment is of moderate size, distinctly
arched, and dorsally bordered by the quadratojugal. The
jugal forms the anterior margin of the lower temporal
embayment, and the quadrate the posteroventral part.
For measurements see Table 1 and for skull ratios see
Table 2.
Premaxilla. The premaxillae meet at a skull midline
suture anteriorly. Posteriorly, the premaxillae are div-
ided by the anterior processes of the nasals (Fig. 2B,
D). The right premaxilla is complete except for its anter-
ior-most part. The premaxilla is tallest dorsoventrally in
lateral view where it meets the anterior process of the
nasal dorsally. Lateral to the nasal, the premaxilla nar-
rows again. The premaxilla surrounds the anterodorsal
part of the external naris. Posterolaterally, it has a
straight suture with the maxilla. The posteromedial
extension of the premaxilla is obscured due to damage
of the bone surface. However, the lateral part of the pre-
maxilla is shorter than the medial part. The right pre-
maxilla carries eight functional teeth and three
replacement teeth (Fig. 2B, D, G). In palatal view, the
left premaxilla exhibits three (possibly four) large pits
open to the labial side with broken-off teeth and five
replacement teeth (Fig. 2E, F). In ventral view, each
premaxilla is narrow and rod-like. The premaxillae meet
anteriorly and are divided posteroventrally by
the vomers.
Maxilla. The maxilla is relatively large, with an antero-
posterior extension of 52% of the preserved preorbital
length. Overall, the maxilla resembles an elongate tri-
angle, with the ventral part forming the longest side and
the dorsal part having the widest angle (Fig. 2B, D).
The anterodorsal part shares a suture with the premaxilla
and forms the ventral margin of the external naris. Its
mediodorsal suture is obscured by damage of the bone
surface. Posteroventrally, the maxilla contacts the lacri-
mal, and ventrally it has a very short suture with the
jugal, barely contacting it, at least in lateral view.
Posterior to the external naris, a rounded dorsal process
is indicated but is incompletely preserved due to dam-
age. Combining information from the dorsal and ventral
view of the skull (which exposes further teeth in the
right maxilla), the right maxilla has seven functional
tooth positions and three replacement teeth. The maxil-
lary tooth row extends posterior to the external naris but
not below the orbit (Fig. 2B, D–F).
Lacrimal. The lacrimal is a massive element, spanning
the anterior margin of the orbit and extending anteriorly
halfway to the external naris (Fig. 2B, D).
Anteroventrally, the lacrimal shares a long suture with
the maxilla, and posteroventrally it encases the tip of
the jugal. Dorsally, the lacrimal contacts the nasal in a
long, straight suture, while the posterodorsal contact
with the prefrontal is short. The surface of the lacrimal
was damaged by preparation.
Nasal. The nasal is a large element that makes up most
of the preorbital skull in dorsal view and contributes to
the lateral surface of the skull. Its anterior process
reaches well anterior to the external naris and divides
the premaxillae (Fig. 2B, D) but does not extend to the
preserved tip of the snout. The nasal might have bor-
dered the posterior margin of the external naris, but this
is unclear due to damage. The nasal significantly broad-
ens posterior to the external naris by forming a
A new cymbospondylid ichthyosaur 7
8 N. Klein et al.
ventrolateral process. The nasal must have contacted the
maxilla here, but a suture cannot be identified due to
damage (Fig. 2B, D). Posteriorly, the nasal extends
beyond the orbit as a rectangular posterior process that
extends along the dorsal margin of the skull. The nasal
is separated from the orbital margin by the pre- and
postfrontal, forming long anteroposteriorly extending
sutures with both. The scree block containing the right
upper temporal fenestra contains a large, rectangular
element that extends parallel to the anteromedial process
of the supratemporal and the posterolateral margin of
the upper temporal fenestra. According to our interpret-
ation (see comparison below; Fig. 6), this large, rect-
angular element is the continuation of the posterior
process of the nasal.
Prefrontal and postfrontal. The prefrontal is anteriorly
divided into a short and narrow dorsal process and a
long, slender, ventrally descending process, the latter
bordering the anterodorsal margin of the orbit.
Posteriorly, the prefrontal has a rectangular shape and
merges with the postfrontal (Fig. 2B, D). A suture
between the pre- and postfrontal is not apparent, and
they are fused. The postfrontal is also rectangular and
forms the posterodorsal margin of the orbit. The post-
frontal contacts the squamosal posteriorly and the post-
orbital ventrally. Dorsally, the pre- and postfrontal are
both bordered by the posterior process of the nasal.
Frontal and parietal. The frontal and parietal are
incompletely preserved in one of the two scree blocks,
which connects to the rest of the skull via a tiny piece
of the squamosal. The fit is corroborated by two distinct
calcite veins (Fig. 2A, B). The frontal borders the
anterolateral part of the upper temporal fenestra
(Fig. 2A, C), but is excluded from the orbit. Medially,
the frontal contacts the parietal. The parietal forms the
anteromedial and entire medial margin of the upper
temporal fenestra (Fig. 2A, C). Posteriorly, the parietal
is bordered by the thin dorsal process of the supratem-
poral. A low medial parietal crest is apparent, although
this region is incomplete.
Supratemporal. A thin, narrow element forms the post-
erolateral margin of the upper temporal fenestra. This
element is here interpreted as the anterolateral process
of the supratemporal (Fig. 2). The posteromedial margin
of the upper temporal fenestra is damaged and incom-
plete (Fig. 2A, C). Dorsally, the preserved part of the
supratemporal contacts the rectangular posterior process
of the nasal, and ventrally it contacts the squamosal.
Jugal. The jugal is a narrow, elongate element that
forms the ventral margin of the orbit. Anteriorly, the
jugal ascends slightly, but it is separated from the anter-
ior margin of the orbit by the lacrimal (Fig. 2B, D).
Posteriorly, the jugal is separated from the posterior
margin of the orbit by the postorbital. A long ascending
posterior process of the jugal forms the anterior margin
of the lateral embayment. It cannot be clarified whether
the jugal contacts the quadratojugal or if the postorbital
separates both elements.
Postorbital. The triradiate postorbital forms the entire
posterior margin of the orbit, slightly contributing to its
dorsal and ventral margins also. Ventrally, the post-
orbital contacts the jugal and dorsally the postfrontal
and squamosal. The posterior process of the postorbital
has a suture that extends along the squamosal and
finally meets the quadratojugal. The postorbital is
excluded from the upper temporal fenestra as well as
from the lateral embayment (Figs. 2B, D).
Squamosal. A small part of the squamosal is preserved
on the scree block containing the right upper temporal
fenestra, below the supratemporal and the rectangular
posterior process of the nasal (Fig. 2A, C). On the main
skull block, in lateral view, the squamosal is a massive
element forming most of the posterolateral part of the
skull (Fig. 2B, D). Ventrally, it limits the quadratojugal
along a curved suture. Anterodorsally, the squamosal
contacts the postfrontal and anteroventrally the
postorbital. The squamosal is excluded from the upper
temporal opening by the rectangular posterior process of
the nasal and the supratemporal (Fig. 2,6).
Quadratojugal. The rectangular quadratojugal is antero-
posteriorly longer than tall, and its exposure is exten-
sive. The quadratojugal spans the dorsal arch of the
lateral embayment. The ventral margin is concave; the
dorsal margin is convex (Fig. 2B, D). Dorsally, the
quadratojugal is bordered by the squamosal. Anteriorly,
Figure 2. LACM DI 158109, holotype of Cymbospondylus duelferi sp. nov. A, block #2, disarticulated right upper temporal region
after preparation in dorsal view; and B, block #1, skull in right lateral view. Note the two distinct calcite veins on the main skull
block that fit to those on the upper temporal region. C, outline sketch of the right upper temporal region (block #2) with elements
labelled. D, outline sketch of block #1, skull in right lateral view. E, block #1, skull in ventrolateral view, vertically mirrored. F,
outline sketch of the same with elements labelled. G, premaxillary teeth. H, maxillary tooth. I, details of premaxillary tooth pits that
open towards the labial side. Abbreviations: ang, angular; ar, articular; d, dentary; ecpt, ectopterygoid; fr, frontal; j, jugal; la,
lacrimal; mx, maxilla; na, nasal; pa, parietal; pal, palatine; po, postorbital; pofr, postfrontal; preart, prearticular; prefr, prefrontal;
prmx, premaxilla; pt, pterygoid; qj, quadratojugal; sple, splenial; squa, squamosal; st, supratemporal; surang, surangular.
A new cymbospondylid ichthyosaur 9
it contacts a ventral process of the squamosal, the post-
orbital and maybe the jugal. The posteroventral suture is
bordered by the quadrate.
Quadrate. The quadrate is only exposed in lateral view,
revealing the condyle at the posteroventral margin of
the skull and forming here a continuation of the quadra-
tojugal (Fig. 2B, D). The quadrate borders the posterior-
most part of the lateral embayment and contacts the
squamosal dorsally along a very short suture.
Occipital region. The massive, deeply concave basioc-
cipital is visible on the palatal aspect of LACM DI
158109. The basioccipital and basisphenoid are exposed
due to the anterior shift of the left pterygoid (Fig. 2E,
F). The occipital condyle has a thick and convex margin
surrounding a deeply concave, funnel-shaped area,
which both form the articular surface. Only the posterior
part of the basisphenoid is exposed. The posterior mar-
gin has a deep notch that divides the basisphenoid but
no evidence for the carotid arteries. Anteriorly, the basi-
sphenoid tapers slightly, overall resulting in a heart-
shaped outline in ventral view (Fig. 2E, F).
Palatal region. The ventral parts of the premaxillae
contribute to the palate. They join anteriorly and are
posteriorly divided by the vomers. The vomers are rod-
shaped elements that are dislocated and incomplete (Fig.
2E, F). The right pterygoid is in situ whereas the left
has shifted anteriorly and is slightly twisted (Fig. 2E,
F). The posterior part of the pterygoid has the wing-like
shape typical for ichthyosaurs with the end divided into
a straight posteriorly directed medial process and a
somewhat angled lateral process. The posterior part ori-
ginally covered the basioccipital and basisphenoid.
Along its midlength, the pterygoid widens before it
tapers into a thin anterior process. The interpterygoid
vacuity is narrow. The pterygoid has a transverse flange,
although it is not very prominent. Lateral to the mid-
length region of the right pterygoid, there is an element
that might represent an ectopterygoid. The palatine is
located lateral to the anterior process of the pterygoid;
the palatine ends before the widest expansion of the
pterygoid (Fig. 2E, F). The region between the possible
ectopterygoid and the palatine is covered by an unidenti-
fied bone, and it cannot be clarified whether this part
represents one element or two, because a suture cannot
be identified.
Lower jaw. The right lower jaw is still articulated with
the skull (Fig. 2B, D). As with the tip of the rostrum,
the anterior portion of the lower jaw is missing. The
ventral border of the preserved lower jaw forms a dis-
tinctive concave line. The lower jaw extends posteriorly
beyond the back of the skull in a retroarticular process.
The lower jaw shows the typical ichthyosaur construc-
tion with an elongation of elements (Romer 1956). The
dentary reaches posterior to the middle of the orbit, with
the eight preserved teeth being restricted to the anterior
half of the dentary (Fig. 2B, D). The surangular is rela-
tively large and is, besides the dentary, the main elem-
ent forming the lateral side of the lower jaw (Fig. 2B,
D). The surangular extends posterodorsally beyond the
lateral embayment to the posterior end of the lower jaw
and starts anteroventrally at the middle margin of the
external naris. The angular forms the posteroventral
margin of the lower jaw. The angular is excluded from
contact with the dentary by the surangular. The articular
provides the articulation to the quadrate. The articular
faces the posterior half of the lateral embayment and
has a small ascending crest (Fig. 2B, D). The palatal
view of the skull provides insights into the medial side
of the lower jaw and exposes the prearticular as well as
the splenial (Fig. 2E, F). The splenial is fully exposed.
Its anterior end takes part in the symphysis, sending
three prongs anteriorly. The posterior end is ventrally
underlain by the angular and reaches the Meckelian
fossa dorsally. The prearticular ends anteriorly in a thin
splint that does not reach this fossa. Posteriorly, it is
greatly expanded and forms the dorsal part of the medial
wall of the lower jaw. The prearticular extends to the
caudal end of the lower jaw (Fig. 2E).
Dentition. The left premaxilla provides insights into
tooth implantation (Fig. 2E). Functional teeth are set in
shallow pits (one-quarter of tooth length). These pits
are not closed alveoli but are open towards the labial
side (Fig. 2E, F, I), here having a more ligamentous
tooth attachment to the jaw. The lingually closed side is
made of a massive bulge of bone, which also connects
to the next tooth anteriorly and posteriorly. Thus, there
is a clear bony separation between each individual tooth
and its roots. It seems as if the tooth, during the process
of becoming functional, resorbed the bone tissue on the
labial side. This is because replacement teeth have their
own pits that enclose the entire tooth, and each is
located between two functional teeth. Thus, replacement
and functional teeth alternate, with functional teeth set
at a relatively wide distance from each other. The teeth
are relatively large and typically conical with bluntly
pointed tips. They are slightly curved lateromedially.
The crown, making up the apical third of the teeth, is
striated. Prominent grooves and ridges basal to the
crown in the apical third of the root are visible at the
base of the crown, indicating the presence of pliciden-
tine (Fig. 2H). The tooth cross section, of both root and
crown, is circular. The low number of teeth in the max-
illa and dentary is striking, but we cannot exclude the
possibility of taphonomic tooth loss. However, if teeth
10 N. Klein et al.
were lost after death, they likely were not ankylosed
to the jaw.
Postcranial skeleton
Vertebrae and ribs. Due to the partially disarticulated
and incomplete nature of LACM DI 158109, no verte-
bral count can be given. Most parts of the vertebral col-
umn were found in blocks in the scree, forming short
strings of articulated vertebrae. They all represent either
cervical or anterior dorsal centra, as suggested by their
circular transverse outlines. However, no vertebra could
be freed sufficiently from the matrix for a proper
description. On the main skull block, posterior to the
basisoccipital, there are five disarticulated vertebrae.
The first two are most likely the atlas and axis
(Merriam 1908, pl. 8, figs 1–3) or intercentra (see
Fr€
obisch et al.2006, fig. 7). Both elements are clearly
separated, not fused and most likely incomplete. No
intervertebral articular facets are identifiable, except for
that of the first element, which is slightly convex. It is,
however, not as convex as would be expected if it were
the atlas (in order to fit against the deeply concave
occipital condyle). The second element is approximately
half the height of the first, but this might be at least
partly owing to incompleteness. The third vertebra in
the row is a complete centrum in lateral view. It is
clearly a cervical vertebra because it displays two well-
separated articular facets for the bicephalic cervical ribs.
The parapophysis is elliptical and smaller than the dia-
pophysis which is slanted anteroventrally. It cannot be
determined whether the diapophysis contacts the anterior
margin of the centrum. Dorsal to the diapophysis is a
well-developed articular facet for the neural arch. In no
other centra are the articular facets of the ribs visible
because of incompleteness or sediment cover; most ver-
tebrae are cut in half, exhibiting a sagittal section. In
these sections, it can be seen that the intervertebral
articular facets do not slope evenly towards the centre.
For a description of the histology of the vertebral centra
of both mother and fetuses, see Houssaye et al. (2018).
Neural arches are not clearly identifiable; only a few
partially incomplete neural spines of the dorsal vertebrae
are preserved. The postzygapophysis is large and pro-
nounced, and the area below it is deeply constricted.
The anteroposteriorly broad neural spines are relatively
low and rectangular with a straight dorsal tip. The
neural arch seems to have been oriented dorsoventrally
and not inclined posteriorly.
Numerous preserved ribs form the remains of a well-
articulated rib cage. The extensive degree of articulation
is also seen in several small scree blocks in which ribs
of the left and right sides encase dorsal vertebrae.
Because of weathering, all ribs are incomplete except
for three well-preserved cervical ribs (Fig. 1A, block
#3). One is very short but the other two are longer. One
of these two has a straight shaft, whereas the shaft of
the other is slightly curved. The shaft is slender, taper-
ing and pointed in all three ribs. The articular facets are
clearly divided into a dorsal and a ventral head. Dorsal
ribs are single headed and have a round oval cross sec-
tion at midshaft. The microstructure of the ribs is
exposed in natural breaks and reveals large central
medullary cavities (exceeding 50% of rib diameter) sur-
rounded by an inner ring of resorption followed by com-
pact cortex. The preserved parts of the ribs are not
distinctly curved and have a striated surface. Gastral
ribs occur as fragments, except for one that is still
articulated and is divided into two steeply angled
branches and a median part with a short process point-
ing anteriorly (Fig. 1A, block #5).
Shoulder girdle. The right scapula partially covers the
coracoid and is itself overlain by an element interpreted
as the clavicle. These three elements are preserved in two
blocks that were part of the main concretion (Fig. 1A, B,
block #4). The bones overlie each other, which, in add-
ition to breakage, obscures their morphology. The left
scapula and coracoid are also preserved in a scree block
but both are rather incomplete (Figs 1A, C,4,block#6).
The putative clavicle is massive, with one end expanded
to a triangular region and the other end tapering and
pointed (Fig. 3A, C). The glenoid facet of the scapula is
damaged and incomplete. The scapula lacks a medial
articular facet for the coracoid, has a fan-shaped anterior
margin, and a pointed lateral elongated process with a
shorter, concave posterior margin (Fig. 3A, C). The pos-
terior end is very flat, and the medioposterior margin is
convex (Figs 1B, C, 3A, C). The outline is similar to,
although much larger than that of, a mixosaurid scapula
(Motani 1999, fig. 4B). The roughly crescent-shaped cor-
acoid is divided into an anteroposteriorly long and broad
part and a mediolaterally short, posterolaterally tapering
process (Fig. 3A, C). The anterior margin is slightly con-
vex, and the posterior margin is deeply concave.
Forelimb. The right humerus is preserved with the dis-
articulated shoulder girdle elements and with the radius
and ulna still in anatomically correct position (Figs 1A,
B,4A, block #4). All of the bones on this block are vis-
ible in ventral view. A proximal protrusion on the
humerus, interpreted as the deltopectoral crest, is visible.
The ventral view of this limb is consistent with the ven-
tral exposure of the skull. The left humerus is preserved
in dorsal view, together with the incomplete left scapula
and coracoid (Figs 1A, C,5A, block #6). The humeri
have a constricted midshaft and a triangular proximal
head in proximal view and are distally convex. They are
A new cymbospondylid ichthyosaur 11
broader distally than proximally (Figs 1B, 4A, 5A). The
anterior margin of the shaft is shorter and shallower
than the posterior margin. The dorsal side of the
humerus is evenly smooth.
The radius is a flat bone that has equally expanded
ends and a constricted shaft (Figs 1B,4A). The ulna is
also flat but lunate in shape, with the posterior margin
being evenly convex without a posterior notch (Figs 1B,
4A). The anterior margin is concave. The humerus, radius
and ulna all have a shaft consisting of cortical periosteal
bone (‘perichondral bone’sensu Caldwell 1997).
Distal to the radius and ulna are two round and one
lunate-shaped (most likely incomplete) carpals. An isolated
phalanx is associated with the dorsal and gastral ribs (Fig.
1A, E, block #5). The phalanx is relatively short, and one
side is nearly straight, whereas the other side is slightly
concave. The proximal articular facet of the phalanx is
nearly straight, whereas the distal end is convex.
Foetal remains
The scree blocks exposing the left humerus, girdle ele-
ments and trunk ribs contains three separate but articu-
lated strings of vertebrae (Fig. 5) that are much smaller
than the vertebrae belonging to the ribs and other post-
cranial bones (Tables 1,3). Measurements of these ver-
tebrae are difficult due to poor preservation, but all of
the dorsal vertebrae of the small columns are within the
same size range (Table 3) and are roughly 32% of the
height of the dorsal vertebrae of the putative mother.
Figure 3. Shoulder girdle bones of Triassic ichthyosaurs. A, photograph of the right disarticulated shoulder girdle of LACM DI
158109, holotype of Cymbospondylus duelferi sp. nov. The elements are partially overlying each other. B, composite shoulder girdle
of C. petrinus modified from Merriam (1908, pl. 11). C, sketch of the shoulder girdle of LACM DI 158109; anterior is to the top. D,
sketch of shoulder girdle of C. petrinus, modified from McGowan & Motani (2003, fig. 70). Abbreviations: cl, clavicle; co,
coracoid; scap, scapula.
12 N. Klein et al.
One of these strings (fetus 1) extends approximately
parallel to the vertebrae of the mother and consists of
16 vertebrae (Fig. 5A, C), of which one could be identi-
fied as a tail vertebra due to the presence of a rounded
rib facet located at the anterior margin of the centrum
(Fig. 5E). This string reveals the direction of the fetus:
the anterior end points towards the rear of the mother.
Fetus 2 consists of an uninterrupted series of 13
articulated dorsal vertebrae (Fig. 5B, D). A third string
of vertebrae (fetus 3) lies angled, approximately parallel
to the humerus and rib long axes, and has nine articu-
lated dorsal vertebrae preserved (Fig. 5B, D). All of the
strings of small vertebrae lay ventral to the dorsal verte-
brae of the larger individual.
Thus, LACM DI 158109 consists of one large indi-
vidual associated with the remains of at least three
Figure 4. Forelimb of Triassic ichthyosaurs. A, photograph of right humerus of LACM DI 158109, holotype of Cymbospondylus
duelferi sp. nov., with radius and ulna in ventral view. B, drawing of proximal right forelimb of UCMP 9950, holotype of C.
petrinus, in ventral view, modified from Merriam (1908, pl. 11). C, photograph of same right humerus as in (B) of C. petrinus
(UCMP 9950) in ventral view. D, sketch of forelimb of LACM DI 158109. E, sketch of forelimb of C. petrinus (UCMP 9950),
modified from Merriam (1908, pl. 11). F, sketch of forelimb of C. buchseri, modified from Sander (1989, fig. 7). G, sketch of
forelimb of Besanosaurus leptorhynchus, modified from McGowan & Motani (2003, fig. 70). Abbreviations: hu, humerus; ra,
radius; ul, ulna.
A new cymbospondylid ichthyosaur 13
14 N. Klein et al.
smaller individuals in its trunk region. Considering the
position of the small vertebral columns within the trunk
region of the larger specimen, we here interpret LACM
DI 158109 as a pregnant female with a minimum num-
ber of three fetuses preserved.
Comparisons
General comparison with Triassic ichthyosaurs
Early, Middle and Late Triassic ichthyosaur faunas
share little overlap at the genus level, especially if
named species are considered (see calibrated phylogeny
in Fr€
obisch et al. 2013), and there is no indication that
the new species pertains to any of the well-known Early
or Late Triassic taxa. We also exclude a number of
Middle Triassic genera from consideration that are only
known from a small fraction of the skeleton (as docu-
mented by the data matrix in Moon 2019), such as
Pessopteryx and Wimanius, which are of doubtful valid-
ity (see comments in McGowan & Motani [2003] and
Fr€
obisch et al. [2013]). The new species clearly does
not have any affinity with the diverse and common
mixosaurs of the Middle Triassic, because it lacks the
large eyes, heterodont dentition, angular carpals and
large anterior extent of the coracoid seen in these ani-
mals and is also greater in size (all known mixosaurs
are small, not exceeding 2 m in length) (McGowan &
Motani 2003). Xinminosaurus is a medium-sized Middle
Triassic taxon known from complete skeletons (Jiang
et al. 2008). This taxon possibly might be a junior syno-
nym of Tholodus (Maisch 2010) and differs from the
new species in its pronounced crushing dentition with
multiple tooth rows and plesiomorphic fin morphology
(Jiang et al. 2008). Other Middle Triassic medium-sized
monospecific genera are Besanosaurus (Dal Sasso &
Pinna 1996) and Phantomosaurus (Maisch & Matzke
2000). The two share a more derived morphology of the
pectoral and pelvic girdles with rounded, plate-like ele-
ments, and of the forelimbs and hind limbs, with short
and wide autopodial and zeugopodial elements
(McGowan & Motani 2003; Maisch 2010).
Besanosaurus also differs from all species of
Cymbospondylus in its extremely slender and elongated
snout with small teeth (Dal Sasso & Pinna 1996).
Finally, there is Thalattoarchon (Fr€
obisch et al. 2013),
likely exceeding in size the largest species of
Cymbospondylus,C. petrinus. Its giant skull (posterior
width 820 mm) must have been much larger than the
largest C. petrinus (skull length 1170 mm) (Fr€
obisch
et al. 2013). Thalattoarchon differs from all species of
Cymbospondylus in its very large, bicarinate trenchant
teeth (Fr€
obisch et al. 2013). Thus, apart from the char-
acter-based assignment of LACM DI 158109 to the
genus Cymbospondylus, the evidence presented in this
brief review is also consistent with the assignment of
the specimen to this genus.
The shoulder girdle of LACM DI 158109 is unique
among the material available for comparison (see
Differential diagnosis, above). LACM DI 158109 differs
from the European C. buchseri in the morphology of the
humerus and scapula (Sander 1989;Fig. 4). Except for
the vertebrae, the cymbospondylid material from
Svalbard is too fragmentary for detailed comparisons
with other taxa (Sander 1992; Engelschiøn et al. 2018).
However, the vertebrae of LACM DI 158109 are them-
selves too fragmentary to be compared to the material
from Svalbard. Body size is also an important criterion,
because despite its relatively small size, LACM DI
158109 is of reproductive age and not a juvenile
(see below).
Figure 5. LACM DI 158109, holotype of Cymbospondylus duelferi sp. nov. Block #6 preserving articulated strings of vertebrae of
three fetuses associated with the mother. A, block in field-top view displaying the left humerus, parts of the shoulder girdle, the
dorsal vertebrae of the mother and fetus 1. B, the same block in field-bottom displaying all three fetuses. Arrowheads indicate the
direction of the skull in individuals where we can determine direction (mother and fetus 1). C, enlargement of fetus 1. D, close-up of
B, horizontal row of vertebrae of fetus 2, with 13 articulated vertebrae, and an obliquely oriented string of vertebrae of fetus 3
consisting of nine articulated vertebrae. Each arrow marks a single vertebra. E, enlargement of fetus 1. The arrow indicates the round
rib articulation facet, which is at the anterior margin of the caudal vertebra.
Table 3. Measurements (mm) of foetal vertebrae
(height length) associated with LACM DI 158109 (see Fig.
5). Abbreviation: nm, not measurable.
Fetus 1 Fetus 2 Fetus 3
11.2 9.8 10 11 15 7
12.5 9.9 10 8188
13.5 8.7 12 9157
11 7.9 13 7157
13.5 9148.5 (break) 15 7
15.7 9.5 16 9.5 15 7
16.6 9.9 16 9157
15.1 9.9 14 8157
>14.4 >5.4 15 nm 14 8
14.9 7.3 14 9
18.7 11.4 14 9
13.2 10.2 12 10
12.5 9.6 12 11
13.6 11.8
14.8 7.2
14 8
9.2 8.6
A new cymbospondylid ichthyosaur 15
Re-interpretation of the skull morphology of C.
petrinus and C. nichollsi
Description of the skull morphology of Cymbospondylus
duelferi led us to reconsider the skull osteology of C.
petrinus and C.nichollsi. As several authors have
pointed out, the number of different interpretations of C.
petrinus skull morphology is remarkable (summarized in
Maisch & Matzke [2004] and Fr€
obisch et al. [2006]; see
further references in both). These different interpreta-
tions have influenced hypotheses of phylogenetic rela-
tionships. We studied first-hand the holotype (UCMP
9950) and referred skull (UCMP 9913) of C. petrinus as
well as the holotype and only known specimen of C.
nichollsi (FMNH PR2251). This led us to different
interpretations of some aspects of skull morphology for
both taxa from published ones (Fig. 5). Our new inter-
pretation of the skull morphology of C. petrinus and C.
nichollsi results in recoding of the character matrix for
both taxa (see Supplementary material). In the case of
C. petrinus, we largely agree with the interpretation of
Merriam (1908) but we identified sutures between the
postfrontal and postorbital as well as to the squamosal
and supratemporal (Fig. 6B, D), whereas he hesitated to
draw them (Fig. 6A, C). For the latter two elements, we
reverse the assignment of elements compared to those
of Merriam (1908), as did Huene (1916). In C. petrinus,
the supratemporal is a slender triradiate element, with a
thin and narrow dorsal part bordering the posterior mar-
gin of the upper temporal opening and a broad ventrally
extending part contributing to the occipital region and
contacting the squamosal in this region (Merriam 1908,
fig. 65). For C. nichollsi, we were not able to identify a
postparietal but conclude that the element marked in
Fr€
obisch et al. (2006) as a postparietal is the ventrally
bulging part of the parietal (Fig. 6E–H). In contrast to
Fr€
obisch et al. (2006), we did not find a contribution of
the postorbital to the upper temporal fenestra.
The rectangular element in the posterior part of the
skull roof in C. petrinus, lateral to the upper temporal
fenestra, has had many different interpretations.
Merriam (1908, pls 2, 3) recognized the characteristic
form of this element and identified it as the posterior
process of the postfrontal (Merriam 1908, p. 108) but
did not label it (Merriam 1908, figs 3, 4). Some later
researchers agreed with Merriam’s interpretation (Romer
1956,1968; Maisch & Matzke 2000) but others inter-
preted this element differently. Sander (1989,2000) and
Motani (1999) interpreted it as part of the squamosal
but disagreed on the general form of the latter. Ji et al.
(2016) labelled it as a postorbital, also favouring a com-
pletely different organization of posterior skull elements.
An elongated rectangular element is also obvious in the
skull of C. buchseri but was interpreted here as being
part of two different elements: the postfrontal and
squamosal (Sander 1989, fig. 3). In C. nichollsi,
Fr€
obisch et al. (2006, fig. 4) divided the element into
postorbital and supratemporal portions. Personal obser-
vation indicates that in C. nichollsi the rectangular elem-
ent is formed by a posterior process of the nasal (NK,
pers. obs.; Fig. 6G, H). In C. petrinus, the sutures are
not as clear, and the rectangular element could be part
of the postfrontal, as interpreted by Merriam (1908), or
part of the nasal, in accordance with the morphology of
C. nichollsi. The rectangular element is excluded from
the posterior margin of the upper temporal opening by
the elongated but slender anterolateral process of the
supratemporal in C. petrinus (Fig. 6A–D). In C. nich-
ollsi, the supratemporal is dislocated, and only a small
portion of this element is visible in lateral view (Fig.
6G). The disarticulated skull roof of LACM DI 158109
also possesses this rectangular element (Fig. 2A, D), but
whether it represents a posterior continuation of the
postfrontal or of the nasal cannot be determined, due to
poor preservation. However, we follow our interpret-
ation of the skull morphology of C. nichollsi for LACM
DI 158109 (Figs 2, 6E,F) in concluding that the rect-
angular element is most likely formed by the nasal.
We are aware that the nasal reaching back to the pos-
terior part of the skull is a highly unusual and unique
feature. However, even if our interpretation is incorrect,
in C. nichollsi the nasal unequivocally enters the upper
temporal fenestra (Fr€
obisch et al. 2006), and in C. petri-
nus it reaches as far back as the anterior margin of the
upper temporal fenestra, which demonstrates the unique
posterior elongation of the nasal in Cymbospondylus
(see genus diagnosis in Fr€
obisch et al. 2006), entering
the upper temporal opening. Similarly, it was noted by
Romer (1956, p. 161) that the nasals extend very far
back and override the frontals in ichthyosaur taxa more
derived than mixosaurs.
The shape and location of the supratemporal has also
been under continuous debate, despite being well pre-
served in C. petrinus (Merriam 1908, figs 3, 4, 6: labelled
as the squamosal; Fig. 6A–D). The supratemporal is a trir-
adiate element, forming the posterior margin of the dorsal
skull and the posterior margin of the upper temporal fen-
estra. An elongated slender anteromedial process extends
parallel to the parietal, forming the posteromedial margin
of the upper temporal fenestra. A slender anterolateral
process excludes the rectangular element from the upper
temporal fenestra. A broad ventral process of the supra-
temporal contacts the quadrate and forms most of the pos-
terior skull wall (Merriam 1908, fig. 6). Because of the
slender nature of the dorsal processes, the supratemporal
is easily shifted, damaged or lost, as was likely the case in
C. buchseri in which Sander (1992) did not detect the
16 N. Klein et al.
presence of a supratemporal. In C. nichollsi, the ventral
process of the supratemporal is preserved (visible in lat-
eral view; Fig. 6G) but no dorsal processes could be reli-
ably identified.
Comparison of C. duelferi with C. nichollsi and
C. petrinus
Skull. In comparison to the skulls of C. nichollsi and C.
petrinus, the skull of C. duelferi appears much more gra-
cile, with thinner bones and a less massive appearance,
but its skull ratios are very similar (Table 2). Although
difficult to capture qualitatively, the skull shape of LACM
DI 158109 varies from that of C. nichollsi and C. petrinus.
The skull of C. duelferi appears relatively short, as in C.
petrinus (Merriam 1908), reaching approximately one-
quarter to one-third of the trunk length. By comparison,
skull length in mixosaurs is about two-thirds of trunk
length, and in Ichthyosaurus it is about 80% (Merriam
1908). In addition, C. duelferi differs in skull proportions:
it possesses a relatively shorter postorbital skull region
and a shorter maxilla compared to C. petrinus.InC. duel-
feri, the facial region makes up 55.5% of the preserved
skull length, but this would have been longer considering
the incomplete snout. The maxilla is 51% of the (incom-
plete) preorbital length. All these ratios are comparable
with C. petrinus, in which the facial region makes up 60%
of the entire skull length and the maxilla 48% of the pre-
orbital length. The premaxilla in C. duelferi does not
reach as far posteriorly as it does in C. petrinus. Whether
the premaxillae are generally shorter in C. duelferi than in
C. petrinus cannot be clarified due to the incomplete snout
in LACM DI 158109. The distance between the anterior
orbit margin and external naris is relatively shorter in C.
petrinus than it is in C. duelferi (5.6% of preorbital length
vs 9.25%). The shape and course of sutures is, however,
similar in both taxa. Cymbospondylus petrinus and C.
duelferi both have tooth rows that end below the external
naris. Cymbospondylus petrinus,C. nichollsi and C. duel-
feri share the boomerang-shaped lacrimal, the relatively
large scleral rings and eye socket, and the squamosal and
postorbital being excluded from the upper temporal fenes-
tra by a posterior process of either the nasal or the post-
frontal. Due to the more arched lower temporal
embayment, the quadratojugal is more ventrally curved in
C. duelferi than in C. petrinus; however, this might be an
artefact of preservation. Cymbospondylus duelferi has a
lower number of scleral plates (12) when compared to C.
nichollsi (14) and C. petrinus (14–18). However, there is
no comparative data set on scleral plate numbers in ich-
thyosaurs that would allow us to understand the distribu-
tion of this character across Ichthyosauria. The number of
scleral plates mainly serves in distinguishing the new spe-
cies from its congeners. The sagittal crest in C. duelferi
and C. nichollsi is low, whereas it is distinctly elevated in
C. petrinus. However, in the latter it is more distinct in the
holotype (UCMP 9950) than in UCMP 9913. The shape
of the upper temporal fenestra is oval in C. nichollsi and
C. duelferi but triangular in C. petrinus.
The premaxilla does not border the external naris in C.
nichollsi but does in C. duelferi. The maxilla has a distinct
posterodorsal process contacting the prefrontal in C. nich-
ollsi. This process seems unlikely to have been present in
C. duelferi, although this is not clearly visible in LACM
DI 158109 due to its state of preservation. The lacrimal is
encompassed dorsally, anteriorly and ventrally by the
maxilla in C. nichollsi, whereas in C. duelferi the maxilla
contacts –as far as it is visible –only the ventral margin
of the lacrimal. The prefrontal forms the dorsal half of the
anterior orbit margin in C. nichollsi but only the dorsal
margin in C. duelferi.InC. petrinus the jugal is more
massive, with a nearly triradiate shape because it pos-
sesses a posterior process, whereas it is a slender lunate
element with no posterior process in C. duelferi and C.
nichollsi.InC. duelferi, the jugal contacts the quadratoju-
gal in a very short suture, whereas this suture is broad in
C. petrinus.InC. nichollsi no jugal-quadratojugal contact
exists. However, the presence of a jugal-quadratojugal
contact seems to depend on preservation, because the
slender posterior process of the jugal can be easily cov-
ered by the postorbital due to a slight shift of elements
during fossilization. In C. nichollsi, the anterior part of the
prefrontal is broad and reaches ventrally, but in C. duelferi
it has only a narrow anterior process that does not descend
ventrally. The parietals and frontals are not comparable
because they are incomplete and disarticulated in LACM
DI 158109.
Dentition. Preliminarily, we interpret the mode of tooth
implantation as sub-thecodont, although further studies
are necessary. The total number of teeth in C. petrinus
is higher (30/35) than in C. duelferi (>16). As figured
in Merriam (1908, fig. 3), the maxillary tooth row
approaches the anterior external naris in C. petrinus,
whereas it extends beyond the external naris in C. duel-
feri. The teeth are relatively smaller in C. petrinus,
although it is the larger taxon. The base of the dentine
cone is strongly folded in C. petrinus (Merriam 1908),
as in C. duelferi.Cymbospondylus petrinus is usually
interpreted as having a thecodont dentition, and we
retained this coding in the phylogenetic analysis pre-
sented below because the teeth are situated in com-
pletely enclosed bone pits (Merriam 1908). However,
Motani (1997) described the dentition of C. petrinus as
a peculiar form of thecodont because the teeth seem to
be fused to the bottom of the socket. Sander (1989)
described poorly preserved conical teeth with relatively
coarse striations in C. buchseri, which are set into
A new cymbospondylid ichthyosaur 17
18 N. Klein et al.
discrete pits in a shallow groove. Although the only
published specimen of C. buchseri is incompletely and
poorly preserved, its tooth row also does not extend into
the posterior part of the maxilla (Sander 1989). Further
morphological studies as well as micro-computer-tomog-
raphy data are necessary to understand the implantation
and replacement of teeth in Cymbospondylus.
Postcranium. Vertebral proportions (height and length)
are difficult to assess in LACM DI 158109 because the
vertebrae are incomplete or cut in half (sagittal plane
exposed). In the sagittal plane, the centra of C. duelferi
are comparable in shape and proportions (but not in size)
to those of C. petrinus (Merriam 1908). The neural
arches of C. nichollsi differ from those of C. duelferi in
having less pronounced postzygapophyses and a
straighter and relatively longer neural spine. Fr€
obisch
et al. (2006) noted tightly articulated pre- and postzyga-
pophyses, resulting in a strongly stiffened vertebral col-
umn, which may not represent a taxonomic feature but
could be the result of age and/or a pathology of that par-
ticular specimen (Fr€
obisch et al. 2006, p. 529). The
neural arches of C. duelferi appear more loosely con-
nected to the centra, which may be due to ontogeny or
preservation and might not necessarily be related to tax-
onomy. However, the neural arches of C. duelferi closely
resemble the morphology in C. petrinus and C. piscosus.
Cymbospondylus piscosus, the type species of the
genus, is only known from dorsal vertebrae, which dif-
fer from those of C. petrinus in size and in the shape of
the articular surface of the vertebral centra (Merriam
1908). In addition, the diapophyses in C. piscosus are
elongated ventrally, and the neural arches are massive
and possess large zygapophyses (Merriam 1908). The
size and shape of the neural arches, including the large
postzygapophyses, of LACM DI 158109 are comparable
to Merriam’s(1908) figures of C. piscosus.AsinC.
piscosus, the centra have a sloping, funnel-shaped
articular surface, but those of C. duelferi are less
amphicoelous and are thus more similar to those of
C. petrinus in having only a small ‘funnel’in the centre
of the articular surface. The vertebral morphology of
themuchlarger‘C. nevadanus’(Merriam 1908; but see
McGowan & Motani [2003], who regarded ‘C. nevada-
nus’as a species inquirendae) differs greatly from that of
C. duelferi.Thecentrumof‘C. nevadanus’is deeply
amphicoelous, exhibiting a small hole for the notochordal
canal. The neural arch of ‘C. nevadanus’is more massive
and less differentiated and differs in proportions, in that
it is not as high dorsoventrally but is much broader trans-
versely. Based on vertebral morphology, it is unlikely
that the differences between LACM DI 158109, C. pisco-
sus and ‘C. nevadanus’are due only to size. In contrast
to the condition in C. petrinus and C. nichollsi,where
the neural spines are inclined posteriorly (Merriam 1908;
Fr€
obisch et al. 2006), the neural arches of C. duelferi
seem to have been oriented upright, as figured for C. pis-
cosus (Merriam 1908,fig.137).
The dorsal ribs of C. duelferi differ from those of C.
petrinus and C. nichollsi in the cross-sectional shape of
the midshaft. In the latter two taxa, the ribs have a fig-
ure-eight-shaped cross section (Fr€
obisch et al. 2006)and
C. petrinus has a flattened dorsal surface (Merriam 1908,
pl. 10, fig. 6), whereas the rib midshafts of C. duelferi
have an even oval cross section where preserved.
The shoulder girdle elements of C. duelferi differ
markedly from those described for C. petrinus (Merriam
1908)andC. nichollsi (Fr€
obisch et al. 2006), as dis-
cussed above. Contrary to the roughly fan-shaped scapula
of C. duelferi, the scapula of C. petrinus and C. buchseri
has a triradiate form with an elongated dorsal blade, a
thick, massive glenoid articulation for the humerus, and
an anteriorly pointing rectangular process (Merriam 1908;
Sander 1989). The shape of the coracoid of C. duelferi is
comparable to the shape of the coracoid in C. petrinus
(Merriam 1908), C. nichollsi (Fr€
obisch et al. 2006)and
C. buchseri (Sander 1989), all of which have a roughly
crescent-shaped outline, but the posterolateral process is
much more pronounced in C. duelferi. The coracoid of
C. duelferi seems to lack a coracoid foramen.
The general shape of the humerus is similar in C.
duelferi and C. petrinus, but the proximal and distal
articular facets are more pronounced in C. petrinus. The
distal end of the humerus of C. petrinus is clearly sepa-
rated into two facets, which are less distinct in C. duel-
feri. The proximomedial margin of the humerus of C.
petrinus is expanded and convex, which is not the case
in C. duelferi. The deltopectoral crest and midshaft con-
striction are less pronounced in C. petrinus compared to
Figure 6. Interpretation of skull morphology of Cymbospondylus petrinus (A–D) and C. nichollsi (E–G). A, skull of C. petrinus in
lateral view and C, skull of C. petrinus in dorsal view. Both sketches are copied and modified from Merriam’s original drawings
(Merriam 1908, figs 3, 4). B, and D, our interpretations of the skull morphology of C. petrinus, basically agreeing with Merriam
(1908), except for including some sutures where he hesitated or was unclear. E, skull of C. nichollsi in lateral view, and F, skull of
C. nichollsi in dorsal view. Both sketches are copied and modified from Fr€
obisch et al. (2006, figs 3, 4). G, and H, our interpretation
of the skull morphology of C. nichollsi, differing mainly in the posterior part of the skull (see text). Abbreviations: ang, angular; ar,
articular; d, dentary; fr, frontal; j, jugal; la, lacrimal; mx, maxilla; na, nasal; pa, parietal; pal, palatine; po, postorbital; pofr,
postfrontal; preart, prearticular; prefr, prefrontal; prmx, premaxilla; pt, pterygoid; qj, quadratojugal; sple, splenial; squa,
squamosal; st, supratemporal; surang, surangular.
A new cymbospondylid ichthyosaur 19
C. duelferi. The radius in both taxa is constricted at
midshaft, but the proximal and distal facets are not well
pronounced in C. duelferi. The ulna of C. petrinus has a
clear notch on its posterior side, whereas this margin is
convex in C. duelferi.
Discussion
Ontogenetic stage of C. duelferi and
foetal remains
At a reconstructed body length of 4.3 m, C. duelferi is
small for a cymbospondylid from the Fossil Hill Member.
Cymbospondylus petrinus has a body length of 9.3 m
(Merriam 1908), and C. nichollsi waslargerthan7m
(Fr€
obisch et al. 2006)(Table 2). Judging from the diffi-
culty in identifying skull sutures, LACM DI 158109 had
achieved skeletal maturity, and its pregnant state docu-
ments that it was sexually and reproductively mature.
These inferences are in agreement with the vertebral hist-
ology of both the mother and fetuses (Houssaye et al.
2018). The large specimen is histologically mature, and
the sampled small specimen (the fourth vertebra from the
left in Fig. 5E) is a fetus as indicated by the trabecular
architecture and large notochordal canal (Houssaye et al.
2018). Unless we entertain the unparsimonious hypothesis
that the large individual ingested aborted fetuses of another
mother, the fetuses inside the rib cage must be the off-
spring of the large individual. Also, the high degree of
articulation argues against ingestion. We thus conclude
that the holotype of C. duelferi (LACM DI 158109) was
an adult individual.
Given its late Anisian age, LACM DI 158109 repre-
sents the second oldest pregnant ichthyosaur in the fossil
record, closely followed in time by the holotype of
Besanosaurus leptorhynchus from the latest
Anisian–earliest Ladinian (Grenzbitumenzone; Dal
Sasso & Pinna 1996). The oldest evidence of vivipary
in ichthyosaurs is from Chaohusaurus from the
Olenekian of China (Motani et al. 2014).
The size ratio estimated for the fetuses and the puta-
tive mother is based on the height of the dorsal
Figure 7. Phylogenetic relationships of Cymbospondylus duelferi sp. nov. Strict consensus tree of 12 most parsimonious trees of
1212 steps in length, based on the reduced taxon set (see text for further explanations). The values are the absolute and relative
Bremer supports. Consistency index ¼0.262; retention index ¼0.627.
20 N. Klein et al.
vertebrae. The vertebrae of the fetuses are, on average,
32% of the height of those of the large individual
(Tables 1,3). However, because only vertebral size is
available for the calculation of the size of the fetuses,
this estimate must be treated with caution. In other stud-
ies, body lengths of the fetuses and mother can be dir-
ectly measured or calculated from skull or humerus
length (summarized in O’Keefe & Chiappe 2012, Table
1). Comparing the estimated sizes of the fetuses in
LACM DI 158109, they are within the range of those
for other marine reptiles (O’Keefe & Chiappe 2011,
Table 1; Griebeler & Klein 2019). The ratio of foetal to
maternal size in LACM DI 158109 is about 9% larger
than that of the well-known Jurassic ichthyosaur
Stenopterygius (B€
ottcher 1990;O’Keefe & Chiappe
2011) but differs by only 3% from the fetal-maternal
size ratio in Chaohusaurus from the Early Triassic
(Motani et al. 2014). LACM DI 158109 (4.3 m recon-
structed body length) and Stenopterygius (3–4 m body
length) are both medium-sized ichthyosaurs, whereas
Chaohusaurus (1 m) was distinctly smaller.
The presence of three fetuses with LACM DI 158109 is
a minimum number since we cannot know whether add-
itional fetuses may have been lost taphonomically or due
to weathering before discovery. However, the large size of
the individual fetuses indicates that the original number
must have been relatively low and does not compare to
the maximum count known for Stenopterygius,wherea
single female can have as many as 11 fetuses (B€
ottcher
1990). The smaller size of the fetuses in Stenopterygius
allowed for a larger number per litter, maybe implying a
change in life history and reproduction strategy from the
Triassic (i.e. Cymbospondylus) to the Jurassic (i.e.
Stenopterygius). This is in agreement with the small size
of the fetuses compared to the adult in an indeterminate
ichthyosaur from the Early Jurassic of England (Boyd &
Lomax 2018). Fetus 1 at least of LACM DI 158109 indi-
cates a head-first birth position (Fig. 5E; see above) in
relation to the mother’s anatomy, with the tail lying close
to the mother’s shoulder girdle. Motani et al. (2014)
described a head-first position of fetuses for
Chaohusaurus.IntheEarlyJurassicStenopterygius,the
position of fetuses seems to be variable (B€
ottcher 1990),
although Motani et al. (2014) erroneously claimed that
Stenopterygius fetuses show a breach position.
Phylogenetic analysis and taxonomic affiliation
To determine the phylogenetic relationships of LACM
DI 158109, the specimen was coded in the phylogenetic
data matrix of Maxwell et al. (2019), which is a revised
version of the matrix of Moon (2019). Initially, we used
the same terminal taxa and characters as Maxwell et al.
(2019). Following Maxwell et al. (2019), we excluded
eight taxa from the analysis (Cymbospondylus piscosus,
Dearcmhara shawcrossi,Isfjordosaurus minor,
Ichthyosaurus acutirostris,Phalarodon major,
Suevoleviathan disinteger,Thaisaurus chonglakmanii and
Tholodus schmidi) because they are fragmentary or syn-
onymous with other taxa, and ran the original data matrix
to reproduce the tree figured therein. We obtained a simi-
lar, poorly resolved topology, but with minor differences,
such as recovering more most parsimonious trees (MPTs:
220 vs 150) of the same length (1634 steps). Secondly,
we partially recoded C. nichollsi and C. petrinus (see
Supplementary material) and added LACM DI 158109
(Supplementary material Fig. S1). We then re-ran the
analysis with our modified data matrix (see
Supplementary material). Thus, this analysis included 108
ingroup taxa, with Hupehsuchus nanchangensis as the
outgroup, and 287 characters. None of the characters was
treated as ordered. The matrix was analysed under max-
imum parsimony in TNT using a New Technology
Search (ratchet settings; Goloboff et al. 2008). We
obtained 199 MPTs with lengths of 1647 steps. Even the
50% majority rule tree was poorly resolved; it has a con-
sistency index (CI) of 0.195 and a retention index (RI) of
0.654 (Supplementary material Fig. S1).
To improve resolution, we deleted 49 of the Jurassic
neoichthyosaurian taxa, retaining only the best-known
species (see Supplementary material). The justification
for this was that the poorly known Jurassic taxa might
introduce unnecessary noise into the analysis. The final
matrix had 59 taxa. We then analysed this matrix with
the same settings as above using the script ‘aquickie.run’
supplied with TNT. This script identifies the MPTs for
the data, calculates the strict consensus and provides tree
support values if requested. Applying this script, we
obtained 12 MPTs of 1212 steps in length. The strict
consensus is fairly well resolved and supported (Fig. 7:
absolute and relative Bremer support values are shown),
with a CI of 0.262 and an RI of 0.627.
In both the initial analysis (with the taxon set of
Maxwell et al. 2019) and the analysis with the reduced
taxon set, LACM DI 158109 was nested with C. nich-
ollsi and C. petrinus. Thus, the three species from the
Fossil Hill Member always form a clade. In terms of
character scores, the terminal taxa of this clade differ
most obviously in their postcranial anatomy, but also in
many subtle aspects of cranial morphology. The clade is
supported by the absence of a medial facet on the corac-
oid for the scapula (character 190, state 0: absent). The
European C. buchseri is in an unresolved trichotomy
with this clade and more derived ichthyosaurs. The erec-
tion of a new species of the genus Cymbospondylus for
LACM DI 158109 is justified on the basis of its morph-
ology and the results of the phylogenetic analysis. It is
A new cymbospondylid ichthyosaur 21
unlikely that C. duelferi represents a juvenile of one of
the other two species, because of these differences and
also its considerably smaller adult body size (see
above). A time-calibrated phylogenetic hypothesis for
Triassic ichthyosaurs, including the new taxon, is
depicted in Figure 8. It is based on the strict consensus
tree from the reduced taxon set. Data on stratigraphic
ranges were taken from Bardet et al. (2014).
Cymbospondylus diversity and radiation
The new taxon expands the number of named species
within the genus to five (or four if ignoring C. piscosus:
see discussion above and McGowan & Motani [2003]
on the questionable validity of C. piscosus). With the
description of C. duelferi, three species of
Cymbospondylus are documented in the Anisian-aged
Fossil Hill Member of Nevada: one (C. petrinus) from
the Fossil Hill Member within the Prida Formation in
the Humboldt Range; and two (C. nichollsi, C. duelferi)
from the Fossil Hill Member within the Favret
Formation in the Augusta Mountains (see Nicholls &
Silberling [1977] for a review of the stratigraphy).
These species share a relatively uniform skull morph-
ology but differ in body size, postcranial anatomy and
robustness of the skull, and maybe also in vertebral
dimensions and proportions. Some of these differences
might be related to sexual dimorphism, which is not
easy to document in ichthyosaurs due to small sample
sizes, except maybe in Chaohusaurus for which over 40
individuals have been studied (Motani et al. 2018).
In addition to the Middle Triassic Nevada clade, there is
evidence of Cymbospondylus from the Early and Middle
Triassic of the Boreal Ocean (Svalbard: Sander 1992;
Engelschiøn et al. 2018) and the Middle Triassic of Europe
(Sander 1989). Middle Triassic sediments have yielded
high cymbospondylid diversity within the Pacific realm, an
as yet unidentified cymbospondylid species in the Boreal
Figure 8. Time-calibrated phylogenetic hypothesis for Triassic ichthyosaurs, identifying the new taxon described in this paper as a
member of the Cymbospondylus clade. The topology depicts the strict consensus tree of the reduced taxon set of 12 most
parsimonious trees. Time calibration is based on stratigraphic occurrence data (to stage level) from Bardet et al. (2014), computed
and visualized with the R-package ‘strap’(Bell & Lloyd 2014; R Core Team 2018).
22 N. Klein et al.
Ocean (Engelschiøn et al. 2018) and one cymbospondylid
taxon, C. buchseri (Sander 1989), in the Tethys, implying
either a Northern Hemisphere or global distribution of the
genus (although the lack of appropriate marine Triassic
localities in the southern hemisphere makes documenting a
global distribution difficult). The addition of a new species
to the clade underscores the notion that the cymbospondyl-
ids must have diversified and dispersed rapidly, considering
that the oldest records (from Svalbard) are late Olenekian in
age and that by the middle Anisian there is a species in
Nevada (C. petrinus), followed by two late Anisian species
(C. nichollsi and C. duelferi)inNevadaand,bythelatest
Anisian, a species in Europe (C. buchseri). The earliest
occurrence of the genus and the European occurrence are
separated by roughly 6 million years.
Conclusions
A new medium-sized ichthyosaur, Cymbospondylus
duelferi sp. nov., is described from the Fossil Hill
Member of the Favret Formation of the Augusta
Mountains, Nevada, USA. This extends the total number
of named species of the genus to four (excluding the
type species C. piscosus, which is known only from a
few vertebrae). The new species is part of a clade of
Anisian-aged Nevadan species, pointing to a diverse
local radiation. However, the genus Cymbospondylus is
widespread in the Middle Triassic, with additional
occurrences in Svalbard and Europe suggesting, at least,
a Northern Hemisphere distribution. The holotype and
only specimen of C. duelferi is a pregnant female bear-
ing at least three large fetuses, representing the second-
oldest record of a pregnant ichthyosaur known so far.
Phylogenetic relationships within Cymbospondylus are
well supported but its phylogenetic relationships with
other taxa remain problematic, as was pointed out previ-
ously (Moon 2019; Maxwell et al. 2019).
Acknowledgements
We are grateful to Jose Soler and Valeria Jamarillo for
preparation of the specimen, and to Maureen Walsh for
curation (all LACM). Erin Maxwell (SMNS) kindly
made the data matrix from Maxwell et al. (2019)
available before final publication and provided helpful
technical advice, as did Jos
e Luis Carballido (Museo
Egidio Feruglio, Trelew, Argentina). The specimen was
collected under BLM Paleontological Resources Use
Permit N-92625. The help of BLM staff at the
Winnemucca field office is gratefully acknowledged.
We thank the reviewers, A. Roberts (Bath) and L. L.
Delsett (Oslo), as well as the handling editor, Susannah
Maidment, for their constructive comments and
thorough review. Their comments greatly improved our
manuscript. Tom Young and his Great Basin Brewery
(Reno) generously supported our fieldwork. The study
was funded by grants from the Deutsche
Forschungsgemeinschaft (project numbers 388659338
and 264173172) to P. Martin Sander and the National
Geographic Society (grant number 9599-14) to
Lars Schmitz.
Supplemental material
Supplemental material for this article is available here:
http://dx.doi.org/10.1080/14772019.2020.1748132.
ORCID
Nicole Klein http://orcid.org/0000-0003-3638-1194
Lars Schmitz http://orcid.org/0000-0003-0210-4383
Tanja Wintrich http://orcid.org/0000-0002-1157-8604
P. Martin Sander http://orcid.org/0000-0003-
4981-4307
References
Bardet, N.,Falconnet, J.,Fischer, V.,Houssaye, A.,Jouve, S.,
Pereda-Suberbiola, X.,P
erez-Garc
ıa, A.,Rage, J.-C. &
Vincent, P. 2014. Mesozoic marine reptile
palaeobiogeography in response to drifting plates. Gondwana
Research 26,869–887. doi:10.1016/j.gr.2014.05.005
Bell, M. A. &Lloyd, G. T. 2014. strap: stratigraphic tree
analysis for palaeontology. R package version 1.4.
Updated at: https://CRAN.R-project.org/package=strap,
accessed 28 December 2019.
de Blainville, H. M. D. 1835. Description de quelques
esp
eces de reptiles de la Californie. Nouvelles Annales du
Museum d’Histoire Naturelle 4, 233–296.
B€
ottcher, R. 1990. Neue Erkenntnisse zur
Fortpflanzungsbiologie der Ichthyosaurier (Reptilia).
Stuttgarter Beitr€
age zur Naturkunde, Series B,164,1–51.
Boyd, M. J. &Lomax, D. R. 2018. The youngest occurrence
of ichthyosaur embryos in the UK: a new specimen from
the Early Jurassic (Toarcian) of Yorkshire. Proceedings of
the Yorkshire Geological Society,62,77–82. doi:10.1144/
pygs2017-008
Bucher, H. 1988. A new middle Anisian (Middle Triassic)
ammonoid zone from north-western Nevada (USA).
Eclogae Geologicae Helvetiae,81, 723–762.
Caldwell, M. W. 1997. Modified perichondral ossification and
the evolution of paddle-like limbs in ichthyosaurs and
plesiosaurs. Journal of Vertebrate Paleontology,17,
534–547. doi:10.1080/02724634.1997.10011000
Cuthbertson, R. S.,Russell, A. P. &Anderson, J. S. 2013.
Cranial morphology and relationships of a new grippidian
A new cymbospondylid ichthyosaur 23
(Ichthyopterygia) from the Vega–Phroso Siltstone Member
(Lower Triassic) of British Columbia, Canada. Journal of
Vertebrate Paleontology,33, 831–847. doi:10.1080/
02724634.2013.755989
Dal Sasso, C. &Pinna, G. 1996. Besanosaurus leptorhynchus
n. gen. n. sp., a new shastasaurid ichthyosaur from the
Middle Triassic of Besano (Lombardy, N. Italy).
Paleontologia Lombarda, nuova seria,4,1–23.
Engelschiøn, V. S.,Delsett, L. L.,Roberts, A. J. &Hurum,
J. H. 2018. Large-sized ichthyosaurs from the Lower
Saurian niveau of the Vikinghøgda Formation (Early
Triassic), Marmierfjellet, Spitsbergen. Norwegian Journal
of Geology,98, 239–265. doi:10.17850/njg98-2-05
Fr€
obisch, N. B.,Sander, P. M. &Rieppel, O. 2006. A new
species of Cymbospondylus (Diapsida, Ichthyosauria) from
the Middle Triassic of Nevada and a re-evaluation of the
skull osteology of the genus. Zoological Journal of the
Linnean Society,147, 515–538. doi:10.1111/j.1096-3642.
2006.00225.x
Fr€
obisch, N. B.,Fr€
obisch, J.,Sander, P. M.,Schmitz, L. &
Rieppel, O. 2013. Macropredatory ichthyosaur from the
Middle Triassic and the origin of modern trophic
networks. Proceedings of the National Academy of
Sciences of the United States of America,110, 1393–1397.
doi:10.1073/pnas.1216750110
Goloboff, P. A.,Farris, J. S. &Nixon, K. C. 2008. TNT, a
free program for phylogenetic analysis. Cladistics,24,
774–786.
Griebeler, E. M. &Klein, N. 2019. Life-history strategies
indicate live-bearing in Nothosaurus (Sauropterygia).
Paleontology,62(4), 1–17. doi:10.1111/pala.12425
Houssaye, A.,Nakajima, Y. &Sander, P. M. 2018.
Structural, functional, and physiological signals in
ichthyosaur vertebral centrum microanatomy and
histology. Geodiversitas,40, 161–170. doi:10.5252/
geodiversitas2018v40a7
Huene, F. von. 1916. Beitr€
age zur Kenntnis der
Ichthyosaurier im deutschen Muschelkalk.
Palaeontographica,62,1–68.
Ji, C., Jiang, D.-Y., Motani, R., Rieppel, O., Hao, W. &
Sun, Z.-Y. 2016. Phylogeny of the Ichthyopterygia
incorporating recent discoveries from South China.
Journal of Vertebrate Paleontology,36, e1025956. doi:
10.1080/02724634.2015.1025956
Jiang, D.-Y.,Motani, R.,Hao, W.,Schmitz, L.,Rieppel, O.,
Sun, Y. &Sun, Z. 2008. New primitive ichthyosaurian
(Reptilia, Diapsida) from the Middle Triassic of Panxian,
Guizhou, southwestern China and its position in the
Triassic biotic recovery. Progress in Natural Science,
18(10), 1315–1319. doi:10.1016/j.pnsc.2008.01.039
Kelley, N. P.,Motani, R.,Embree, P. &Orchard, M. J.
2016. A new Lower Triassic ichthyopterygian assemblage
from Fossil Hill, Nevada. PeerJ,4, e1626. doi:10.7717/
peerj.1626
Leidy, J. 1868. Notice of some reptilian remains from
Nevada. Proceedings of the Philadelphia Academy of
Science,20, 177–178.
Linnaeus, C. 1758. Systema naturae per regna tria naturae,
secundum classes, ordines, genera, species, cum
characteribus, differentiis, synonymis, locis.Vol. Tomus I.
Tenth edition. Laurentii Salvii, Holmiae [¼Stockholm],
i–ii, 1–824.
Maisch, M. W. 2010. Phylogeny, systematics, and origin of
the Ichthyosauria –the state of the art. Palaeodiversity,3,
151–214.
Maisch, M. W. &Matzke, A. T. 2000. The Ichthyosauria.
Stuttgarter Beitr€
age zur Naturkunde, Series B,298,1–159.
Maisch, M. W. &Matzke, A. T. 2004. Observations on
Triassic ichthyosaurs. Part XIII: new data on the cranial
osteology of Cymbospondylus petrinus (Leidy, 1868) from
the Middle Triassic Prida Formation of Nevada. Neues
Jahrbuch f€
ur Geologie und Pal€
aontologie, Monatshefte,
2004, 370–384.
Massare, J. A. 1987. Tooth morphology and prey preference
of Mesozoic marine reptiles. Journal of Vertebrate
Paleontology,7, 121–137.
Maxwell, E. E.,Cort
es, D. Y.,Patarroyo, P. &Parra Ruge,
M. A. 2019. A new specimen of Platypterygius
sachicarum (Reptilia, Ichthyosauria) from the Early
Cretaceous of Colombia and its phylogenetic implications.
Journal of Vertebrate Paleontology,39, e1577875. doi:10.
1080/02724634.2019.1577875
McGowan, C. &Motani, R. 2003. Ichthyopterygia.
Handbook of Paleoherpetology, 8. Dr. Friedrich Pfeil,
Munich, VIIIþ173 pp.
Merriam, J. C. 1908. Triassic Ichthyosauria, with special
reference to the American forms. Memoirs of the
University of California,1,1–155.
Moon, B. C. 2019. A new phylogeny of ichthyosaurs
(Reptilia: Diapsida). Journal of Systematic Palaeontology,
17(2), 129–155. doi:10.1080/14772019.2017.1394922
Motani, R. 1997. Temporal and spatial distribution of tooth
implantation in ichthyosaurs. Pp. 81–103 in J. M.
Callaway & E. L. Nicholls (eds) Ancient marine reptiles.
Academic Press, London and New York.
Motani, R. 1999. Phylogeny of the Ichthyopterygia. Journal
of Vertebrate Paleontology,19, 473–496. doi:10.1080/
02724634.1999.10011160
Motani, R.,Jiang, D.-Y.,Tintori, A.,Rieppel, O. &Chen,
G.-B. 2014. Terrestrial origin of viviparity in Mesozoic
marine reptiles indicated by Early Triassic embryonic
fossils. PLoS ONE,9(2), e88640. doi:10.1371/journal.
pone.0088640
Motani, R.,Huang, J.-D.,Jiang, D.-Y.,Tintori, A.,Rieppel,
O.,You, H.,Hu, Y.-C. &Zong, R. 2018. Separating
sexual dimorphism from other morphological variation in
a specimen complex of fossil marine reptiles (Reptilia,
Ichthyosauriformes, Chaohusaurus). Scientific Reports,
8(1), 14978. doi:10.1038/s41598-018-33302-4
Nichols, K. M. &Silberling, N. J. 1977. Stratigraphy and
depositional history of the Star Peak Group (Triassic),
northwestern Nevada. Geological Society of America,
Special Papers,178,1–73.
O’Keefe, F. R. &Chiappe, L. M. 2011. Viviparity and K-
selected life history in a Mesozoic marine plesiosaur
(Reptilia, Sauropterygia). Science,333, 870–873. doi:10.
1126/science.1205689
Osborn, H. F. 1903. On the primary division of the Reptilia
into two sub-classes, Synapsida and Diapsida. Science,17,
275–276. doi:10.1126/science.17.424.275-b
RCoreTeam.2018. R: A language and environment for statistical
computing. R Foundation for Statistical Computing. Vienna,
Austria. Updated at: https://www.R-project.org/,accessed18
August 2019.
Rieber, H. 1970. Phragmoteuthis ticinensis sp. nov., ein Coeloidea-
Rest aus der Grenzbitumenzone (Mittlere Trias) des Monte San
24 N. Klein et al.
Giorgio (Kanton Tessin, Schweiz). Pal€
aontologische Zeitschrift,
44,32–40. doi:10.1007/BF02989793
Rieppel, O.,Sander, P. M. &Storrs, G. 2002. The skull of
the pistosaur Augustasaurus from the Middle Triassic of
northwestern Nevada. Journal of Vertebrate Paleontology,
22, 577–592.
Romer, A. S. 1956. Osteology of the reptiles. University of
Chicago Press, Chicago, 800 pp.
Romer, A. S. 1968. An ichthyosaur skull from the Cretaceous
of Wyoming. Contributions to Geology,University of
Wyoming,7,27–41.
Sander, P. M. 1989. The large ichthyosaur Cymbospondylus
buchseri, sp. nov., from the Middle Triassic of Monte San
Giorgio (Switzerland), with a survey of the genus in
Europe. Journal of Vertebrate Paleontology,9, 163–173.
doi:10.1080/02724634.1989.10011750
Sander P. M. 1992. Cymbospondylus (Shastasauridae:
Ichthyosauria) from the Middle Triassic of Spitsbergen:
filling a paleobiogeographic gap. Journal of Paleontology,
66, 332–337. doi:10.1017/S0022336000033825
Sander, P. M. 2000. Ichthyosauria: their diversity,
distribution, and phylogeny. Pal€
aontologische Zeitschrift,
74,1–35. doi:10.1007/BF02987949
Sander, P. M. &Bucher, H. 1990. On the presence of
Mixosaurus (Ichthyopterygia: Reptilia) in the Middle
Triassic of Nevada. Journal of Paleontology,64, 161–164.
doi:10.1017/S0022336000042396
Sander, P. M.,Rieppel, O. C. &Bucher, H. 1997. A new
pistosaurid (Reptilia: Sauropterygia) from the Middle
Triassic of Nevada and its implications for the origin of
plesiosaurs. Journal of Vertebrate Paleontology,17,
526–533. doi:10.1080/02724634.1997.10010999
Scheyer, T.,Romano, C.,Jenks, J. &Bucher, H. 2014.
Early Triassic marine biotic recovery: the predators’
perspective. PLoS ONE,9(3), e88987. doi:10.1371/journal.
pone.0088987
Schmitz, L. 2005. The taxonomic status of Mixosaurus
nordenskioeldii (Ichthyosauria). Journal of Vertebrate
Paleontology,25, 983–985.
Schmitz, L.,Sander, P. M.,Storrs,G.W.&Rieppel, O.
2004. New Mixosauridae (Ichthyosauria) from
the Middle Triassic of the Augusta Mountains
(Nevada, USA) and their implications for mixosaur
taxonomy. Palaeontographica, Abteilung A,270,
133–162.
Shikama, T.,Kamei, T. &Murata, M. 1978. Early Triassic
ichthyosaurus, Utatsusaurus hataii, gen. et sp. nov., from
the Kitakami Massif, northeast Japan. The Science Reports
of the Tohoku University, Second Series, Geology, 48,
77–97.
Wiman, C. 1929. Eine neue Reptilordnung aus der Trias
Svalbards. Bulletin of the Geological Institution of the
University of Upsala,22, 183–196.
Zhou, M.,Jiang, D.-Y.,Motani, R.,Tintori, A.,Ji, C.,Sun,
Z.-Y.,Ni, P.-G. &Lu, H. 2017. The cranial osteology
revealed by three-dimensionally preserved skulls of the
Early Triassic ichthyosauriform Chaohusaurus
chaoxianensis (Reptilia: Ichthyosauromorpha) from Anhui,
China. Journal of Vertebrate Paleontology,37, e1343831.
doi:10.1080/02724634.2017.1343831
Associate Editor: Susannah Maidment
A new cymbospondylid ichthyosaur 25