PreprintPDF Available

The dual origin of turtles from pareiasaurs

Preprints and early-stage research may not have been peer reviewed yet.


The origin of turtles (traditional clade: Testudines) has been a vexing problem in paleontology. New light was shed with the description of Odontochelys, a transitional specimen with a plastron and teeth, but no carapace. Recent studies nested Owenetta (Late Permian), Eunotosaurus (Middle Permian) and Pappochelys (Middle Triassic) as turtle ancestors with teeth, but without a carapace or plastron. A wider gamut phylogenetic analysis of tetrapods nests Owenetta, Eunotosaurus and Pappochelys far from turtles and far apart from each other. Here dual turtle clades arise from a clade of stem turtle pareiasaurs. Bunostegos (Late Permian) and Elginia (Late Permian) give rise to dome/hard-shell turtles with late-surviving Niolamia (Eocene) at that base, inheriting its Baroque horned skull from Elginia. In parallel, Sclerosaurus (Middle Triassic) and Arganaceras (Late Permian) give rise to flat/soft-shell turtles with Odontochelys (Late Triassic) at that base. In all prior phylogenetic analyses taxon exclusion obscured these relationships. The present study also exposes a long-standing error. The traditional squamosal in turtles is here identified as the supratemporal. The actual squamosal remains anterior to the quadrate in all turtles, whether fused to the quadratojugal or not. 3
The dual origin of turtles from pareiasaurs
311 Collinsville Avenue, Collinsville, IL 62234, USA,
ABSTRACT— The origin of turtles (traditional clade: Testudines) has been a vexing
problem in paleontology. New light was shed with the description of Odontochelys, a
transitional specimen with a plastron and teeth, but no carapace. Recent studies nested
Owenetta (Late Permian), Eunotosaurus (Middle Permian) and Pappochelys (Middle
Triassic) as turtle ancestors with teeth, but without a carapace or plastron. A wider gamut
phylogenetic analysis of tetrapods nests Owenetta, Eunotosaurus and Pappochelys far
from turtles and far apart from each other. Here dual turtle clades arise from a clade of
stem turtle pareiasaurs. Bunostegos (Late Permian) and Elginia (Late Permian) give rise
to dome/hard-shell turtles with late-surviving Niolamia (Eocene) at that base, inheriting
its Baroque horned skull from Elginia. In parallel, Sclerosaurus (Middle Triassic) and
Arganaceras (Late Permian) give rise to flat/soft-shell turtles with Odontochelys (Late
Triassic) at that base. In all prior phylogenetic analyses taxon exclusion obscured these
relationships. The present study also exposes a long-standing error. The traditional
squamosal in turtles is here identified as the supratemporal. The actual squamosal
remains anterior to the quadrate in all turtles, whether fused to the quadratojugal or not.
Turtle workers trying to find the ancestors of turtles keep moving further afield as
more disparate candidates are proposed. Over sixty years ago, Gregory (1946) wrote:
“The gigantic known pareiasaurs seem to present almost ideal conditions for the
derivation of the primitive chelonian characters... stem chelonians may have been derived
from some small pareiasaurs related to Elginia.” Later, Reisz and Laurin (1991) proposed
Owenetta as a turtle relative. Rieppel and deBraga (1996) and deBraga and Rieppel
(1997) argued for a vague diapsid/sauropterygian ancestry. Lee (1997) renewed the
argument for a pareiasaur ancestry, but unfortunately used traditional pareiasaurs (see
below). Li et al. (2008) made headlines with their announcement of Odontochelys, a Late
Triassic turtle with teeth. When they nested it basal to Proganochelys, they failed to
report an outgroup genus and assumed that turtles had a single origin. Schoch and Sues
(2015) promoted Middle Triassic Pappochelys as a diapsid ancestor to anapsid turtles like
Odontochelys. A summary by Joyce (2015) supported the placement of Eunotosaurus as
a turtle ancestor and discussed the domination of molecular data that confusingly placed
turtles as sisters to a wide variety of taxa throughout the Amniota, settling recently on the
Archosauria. Joyce also noted molecular studies frequently recover family tree topologies
that do not match those of morphological studies, or other molecular studies. Foth and
Joyce (2016) reported the turtle lineage extended back to Odontochelys, Pappochelys
(Schoch and Sues, 2015) and Eunotosaurus (Seeley, 1892; Lyson et al., 2010; Bever et
al., 2015). Lyson et al., (2016) explained how stem turtles used broader ribs while
burrowing, employing Eunotosaurus as an example.
Unfortunately, all prior studies excluded relevant taxa in the ancestry of turtles
and in the turtle clade. Some workers (Reisz and Laurin, 1991; Schoch and Sues, 2015;
Seeley, 1892; Lyson et al., 2010; Bever et al., 2015; Lyson et al., 2016) introduced or
employed unrelated taxa that converged with turtles. Some trusted DNA analyses (e.g.
Field et al., 2014) without confirming the results with morphological studies. Based on
those missteps, the ancestral turtle problem continues to vex paleontologists, whether
they realize it or not.
An irrefutable turtle cladogram should recover a series of ancestral turtle taxa in
which a gradual accumulation of traits is readily visible in the entire skeleton (not just the
ribs) of every taxon. In such a cladogram ghost lineages would be minimized and all prior
ancestral candidates would be tested against a substantial number of turtle taxa with
primitive traits, not just Proganochelys.
An online morphological study (;
subset in Fig. 1) recovers a novel dual origin of turtles after testing all prior candidates
for turtle ancestry and hundreds more (Supp. Data; Figs. 2, 3). Commonly known as the
large reptile tree (= LRT), this analysis has been adding taxa for the last eight years and
currently documents the near and far interrelationships of 1165 taxa ranging from stem
tetrapods to mammals, birds and turtles.
Traditionally firsthand access has been a stringent requirement in paleontology.
Many prior workers had firsthand access to fossils, but omitted relevant taxa in their
analyses. Here published photographs and drawings provide most of the data used in the
present analysis. As results attest, omitting relevant taxa (Graybeal, 1998) is clearly the
larger problem here.
The present list of 1165 genus-based taxa minimizes bias and tradition in the
process of selecting ingroup and outgroup taxa for smaller, more focused studies because
all major and many minor clades are well established here. All taxa in the LRT are
generic, specific or species-based. Chimaeras are not employed.
No characters used in the LRT are specific to the clades that include turtles. Traits
specific to turtles would have been useless on birds and tree shrews, and possibly useless
on stem turtles and their ancestors. Generalized characters were chosen or invented for
their ability to lump and split clades and for their visibility in a majority of tetrapod taxa,
many of which had never been tested together. Although some characters are similar to
those from various prior analyses, the present list (see Supporting Data; to be completed when the ms. is accepted) was largely built from
scratch. All taxon subsets of the LRT (e.g. Fig. 1) raise the character/taxon ratio.
At present, the 231 multi-state character set has proven sufficient to lump and
separate 1164 taxa ( the incomplete fossil of Maelestes is the exception), typically with
high Bootstrap scores. All derived taxa document a gradual accumulation of traits in
ancestral taxa going back to stem tetrapods in the Devonian. That’s a strong sign that this
character list is either ideal or good enough for the task at hand. In the past, certain
workers considered 231 characters too small for the number of tested taxa—when the
taxon list was a quarter of the size it is now. Others thought the characters themselves
were less than optimally fashioned. Not all opinions can be accommodated given the
constraints of a single lifetime. Complete resolution in the LRT and high Bootstrap scores
falsify any blackwashing levied against the present character list. For all of its faults, real
or imagined, the LRT continues to lump and separate every new taxon as more taxa are
added every week.
The fault in all prior studies has been taxon exclusion (Graybeal, 1998). That fault
has been minimized here with a wide gamut taxon list.
Taxa and characters were compiled in MacClade 4.08 (Maddison and Maddison,
1990), then imported into PAUP* 4.0b (Swofford, 2002) and analyzed using parsimony
analysis with the heuristic search algorithm. All characters were treated as unordered and
no character weighting was used. Bootstrap support figures were calculated for 100
replicates. The cladogram, character list and data matrix accompany this manuscript and
will be available in permanent repository here: xxxxx and (to be completed when the ms. is accepted).
The LRT nests 1165 taxa in near-complete resolution (Supp. Data, see above). In
the more focused and completely resolved turtles and kin subset of the LRT (Fig. 1; 46
taxa), the first split separates the (Diadectes + Bolosauridae + Procolophonidae) clade
from the (Stephanospondylus + Pareiasauridae) clade. Next the Pareiasauridae splits into
a traditional pareiasaur clade and a stem turtle clade. The stem turtle clade splits
Bunostegos from Sclerosaurus. The Bunostegos clade produces Elginia, Niolamia and
other hard-shell turtles (Fig. 2). The Sclerosaurus clade produces Arganceras,
Odontochelys and other soft-shell turtles (Fig. 3).
In order to reunite turtles as a monophyletic clade, Sclerosaurus and Arganaceras
need to be removed from the inclusion set. Then the remaining soft-shell clade members
nest between Elginia mirabilis and Niolamia.
Looking beyond the turtle clades in the LRT, Pappochelys nests at the base of the
placodonts with Palatodonta. Eunotosaurus nests with Acleistorhinus, Microleter and
Delorhynchus, also far from the two turtle clades. Owenetta does not nest with
Procolophon, or with turtles, but closer to the origin of lepidosauriformes (contra Reisz
and Laurin, 1991). These prior candidates for turtle ancestry essentially dismiss
themselves when more attractive sisters become available in a larger taxon list.
The LRT sheds new light on the dual origin of turtles and invalidates all other
candidates for turtle ancestry (Owenetta, Pappochelys, Eunotosaurus) that nest far from
the two turtle clades. Taxa within the LRT, going back to basal tetrapods, document a
gradual accumulation of traits in the turtle lineage (and every other included clade). Long
ghost lineages are minimized here.
Based on the nineteenth century discoveries of Elginia, Meiolania and Niolamia,
the present insights into turtle origins could have been recovered anytime in the last
hundred years. Unfortunately these taxa were never tested together in a phylogenetic
analysis. Gregory (1946) and Lee (1997) were correct in identifying pareiasaurs as turtle
ancestors, but they did not correctly identify meiolaniids as basalmost turtles or recover a
dual origin for turtles due to taxon exclusion. Prior workers mistakenly assumed that
Proganochelys quenstedti (Baur, 1887; Gaffney, 1990) was the basalmost turtle. Lee
(1997) included Elginia in a pareiasaur/turtle study, but focused on pareiasaurs and did
not include Niolamia and Meiolania. Joyce (2015) was confident in using Proganochelys
as the turtle outgroup taxon instead of using the stem turtles with teeth, Elginia and
Stephanospondylus and the Pareiasaur Clades
Derived from smaller millerettids, the bulky, basalmost pareiasaur,
Stephanospondylus pugnax (Early Permian; Geinitz and Deichmüller, 1882;
Stappenbeck, 1905; Romer, 1935), was omitted from all prior pareiasaur and turtle
studies because it was considered a diadectid and diadectids were considered stem
amniotes (Kissel, 2010). Provided with more taxa to be attracted to, diadectids nest with
Stephanospondylus and the pareiasaurs in the LRT (Fig. 1). Like diadectids,
Stephanospondylus lacks the large quadratojugal cheeks that characterize traditional
pareiasaurs like Scutosaurus. The external nares are taller than wide and oriented
anteriorly. The teeth are sharp cones with constricted bases. The palatine has short sharp
teeth. The neural spines are expanded. At least some of the anterior dorsal ribs have
expanded costal plates, as in certain Diadectes specimens, while other dorsal ribs extend
3x more laterally than ventrally, creating a wide, shallow, disk-like torso. Tiny ribs close
to the sacrum are narrow and straight. The pectoral girdle includes a robust clavicle,
scapula, coracoid, procoracoid and a pentagonal interclavicle. The humerus is robust with
a thick waist and axial torsion. The radius and ulna are short and graviportal. The carpus
is robust with ten large elements. The pelvis has a tall pareiasaur-like ilium and a
ventrally separated pubis and ischium. The femur has a long, angled neck offset from the
main axis, a trait retained by pareiasaurs and turtles.
At the first dichotomy following Stephanospondylus, pareiasaurs split between
traditional pareiasaurs, like Pareiasaurus + Scutosaurus, and the stem turtle clade. At the
base of the stem turtle clade, knob-skulled Bunotegos (Late Permian; Tsuji et al., 2013)
splits from spike-skulled Sclerosaurus (late survivor in the Middle Triassic, Fig. 3; von
Meyer, 1859; Sues and Reisz, 2008).
The Bunostegos Clade
Hard-shell turtles, like Meiolania and Niolamia (Fig. 2), are derived from the
Bunostegos clade of pareiasaurs (Fig. 1). The skull of Bunostegos is ornamented with
knobs anteriorly (over the nares), laterally (over the orbits and at the jaw joints) and
posteriorly (at the upper skull corners). The postparietal is surrounded by the parietals
and conjoined tabulars. The supratemporal descends to the quadratojugal, isolating the
squamosal in the middle of the cheek. The quadratojugal descends relative to the tooth
row. The small teeth have dull, slightly expanded points. The torso remains wide and
disk-like. None of the dorsal ribs have large costal plates. The scapula is vertically
oriented and located anterior to the dorsal ribs. The pelvis is also taller than wide with a
posteriorly descending ilium, a large circular acetabulum and a widely separated, but
short pubis and ischium.
Recently discovered and only partly preserved, Elginia wuyongae (Late Permian;
Liu and Bever, 2018) and better preserved, Elginia mirabilis (Late Permian; Fig. 2;
Newton, 1893), appear next. E. mirabilis is elaborately horned. The teeth are short and
shaped like human incisors. The orbits are located further anteriorly. The pineal opening
is large. The formerly expanded quadratojugal cheeks are reduced to three sharp lateral
Post-crania is unknown in E. mirablis, so whether it had a carapace and plastron,
like its phylogenetic successors, Niolamia and Meiolania or not, like its proximal
ancestors, Bunostegos and E. wuyongae, remains unknown. However, in E. wuyongae,
small osteoderms, apparently the genesis of the carapace, line the dorsal vertebrae.
Phylogenetic bracketing indicates the carapace and plastron first appeared in hard shell
turtles near the E. mirabilis grade in the Late Permian. The Meiolania plastron has a large
central fenestra, so it is a not the result of expanding gastralia (contra Schoch and Sues,
2015, who considered Pappochelys a stem turtle). Gastralia are absent in pareiasaurs.
Basalmost hard-shell turtles
Here (Fig. 1) the larger horned turtles, Niolamia (late-surviving in the Eocene;
Ameghino, 1899; Sterli and de la Fuente, 2011) and Meiolania (late-surviving in the
Oligocene; Owen, 1882, 1886; Gaffney, 1983, 1985, 1996), are recovered for the first
time as basalmost hard-shell turtles close to their only other ‘match’ in the entire LRT,
the small pareiasaur, Elginia (Fig. 2). In all prior studies meiolaniids were considered
aberrant late arrivals. With its larger headdress, Niolamia is the more primitive of the two
tested meiolaniids. In Niolamia small parts of a Meiolania-like carapace and a spiked tail
ring were present among the few post-cranial scraps known for this taxon (Sterli and de
la Fuente, 2011).
The skull of Niolamia (Fig. 2) is similar to Elginia in every view with the
following slightly derivations: The premaxillae lack an ascending process, making the
nares confluent and invisible in lateral view. The orbit is located further anteriorly. The
notch at the posterior maxilla located below the posterior orbit in Elginia moves forward,
below the anterior orbit in Niolamia. Teeth are absent, but tiny empty alveoli remain. The
many small anterior cranial spikes of Elginia are absent in Niolamia, but the posterior
tabular spikes are greatly expanded to form a dorsal shield. On the supratemporal the
posterior two spikes of Elginia are enlarged in Niolamia. The base of the lower spike
descends toward the quadratojugal perhaps without touching it. The quadratojugal is no
longer laterally expanded or spiked. The postorbital is overlapped by the postfrontal from
above and by the jugal from below. The squamosal descends to overlap the
quadratojugal, contacting the jawline. The pineal opening is reduced to absent in
Niolamia on a parietal that is now no longer than the frontal.
In a smaller, likely juvenile specimen related to Niolamia, Crossochelys
(Simpson, 1938), the postparietals are dorsally exposed and bordered by fenestrae. In
Niolamia the tabulars expand to fill those fenestrae and extend medially to cover the
postparietals creating a solid shield.
In the more derived Meiolania (Fig. 2) cranial spikes are reduced. Spikes are
absent in all descendant taxa. The descending ramus of the supratemporal creates a suture
with the quadratojugal leaving a lateral temporal fenestra (the incisura columellae auris)
posterior to the squamosal. Distinct from known pareiasaurs, the tail in Meiolania is long,
robust, armored and provided with a club tip, traits inherited only by later
proganochelids. Confirming their primitive status, meiolaniids are the only known hard-
shell turtles in which the forelimbs can still extend laterally, as in most tetrapods. In all
derived hard-shell turtles, the humerus extends more or less anteriorly from the opening
formed between the carapace and plastron.
The Sclerosaurus Clade
A separate, but parallel clade of turtles with ‘soft’ shells arose from the small
horned pareiasaur, Sclerosaurus (a late survivor in the Middle Triassic, Fig. 3; von
Meyer, 1859; Sues and Reisz, 2008) and Arganaceras vacanti (Late Permian; Jalil and
Janvier, 2005). Arganaceras is the larger of the two, but is more derived in having
smaller horns and other skull traits lacking in ancestral pareiasaurs (see below).
Phylogenetically Arganaceras pushes the origin of the more primitive, but late surviving
Sclerosaurus to the early Late Permian. Workers with firsthand access to the specimen
(Sues and Reisz, 2008), but without the current taxon list, considered Sclerosaurus a
Overall smaller and distinct from Bunostegos, the incomplete skull of
Sclerosaurus elongates the supratemporal horns without producing a descending ramus to
the quadratojugals. That leaves the small squamosal exposed posteriorly, anterior to the
quadrate. The orbit is taller than wide, relatively larger and exposed in dorsal view due to
narrower dorsal skull elements. These traits are retained by soft shell turtles and readily
distinguish them from the hard shell lineage. The temple, measured at the squamosal, is
anteroposteriorly much shorter. The laterally expanded quadratojugal has two short,
robust spikes. The teeth are shorter and more robust. The torso and limbs are shaped
similar to those of Bunostegos, but are more gracile due to the smaller size. The scapula
is relatively smaller, no larger than the paired coracoids. In lateral view the pelvis is
symmetrical with tiny fused pubes and ischia. A small hypoischium is present. A narrow
quilt of tiny osteoderms protects the dorsal vertebrae. No gastralia are present, so the
plastron found in soft-shell turtles, like Odontochelys, is also a novel ossification (contra
Schoch and Sues, 2015).
In Arganaceras (Late Permian; Jalil and Janvier, 2005) the supratemporal and
quadratojugal spikes of Sclerosaurus are much reduced, but a nasal bump remains.
Distinct from hard-shell turtles, the naris is elongate and lateral in position. The
quadratojugal is much reduced with a single posterior spike. The dorsal process of the
published quadratojugal (in Jalil and Janvier, 2005) is actually the ventral squamosal in
long contact with the jugal as in Bunostegos and Sclerosaurus. Arganaceras has smaller
teeth than in Sclerosaurus.
The loose bones of the disarticulated skull can be arranged to create a long gap
between the medial crania and the lateral supratemporal. A comparable cranial gap in
Odontochelys and Trionyx is a skull depression, the posttemporal fenestra, extending
from the occiput to the jugal and postorbital.
Arganaceras (Fig. 3) lacks post-cranial data. If it had a plastron, like
Odontochelys, it might be considered a turtle. These are transitional taxa.
Odontochelys (Late Triassic; Li et al., 2008; Fig. 3) was preserved more
completely and is known from several specimens. In dorsal view the skull is lower and
wider with large openings for the nares and orbits. Posttemporal depressions (fenestrae)
extend anteriorly to the jugal. Originally (Lie et al., 2008) the skull was mistakenly
illustrated without posttemporal fenestrae, following the morphology of the turtle
previously considered the basalmost representative of the clade, Proganochelys. In
Odontochelys the premaxillary ascending processes continues to separate the nares and
contact the nasals. The pineal opening is absent. In ventral view the palate is essentially
solid with robust elements surrounding the tiny medial nares on either side of a robust set
of wide, fused and toothy vomers. In lateral view the jugal and squamosal are reduced in
height. The quadratojugal is a vestige below the jugal. With that loss, the quadrate is now
laterally exposed and located posterior to the squamosal. The supratemporal continues to
sit on top of the quadrate and remains pointed posteriorly, as shown in Trionyx (where the
supratemporal is also traditionally mislabeled the squamosal). The marginal teeth are
small, simple cones in Odontochelys.
The five cervicals of Odontochelys are elongate with tiny ribs, a trait retained by
soft-shell turtles, distinct from meiolaniids. The ten dorsals are also long with laterally
extended paddle-like ribs. Three sacrals converge laterally on the small ilium. The
slender and tapering tail is subequal to the cervicals + dorsals in length. The ventral view
is dominated by a radiating plastron composed of seven laterally paired elements between
the interclavicle and pubis, all without a central fenestra, distinct from meiolaniids. The
pectoral girdle is relatively smaller with a narrower scapula. The coracoid has a stem-like
process and the clavicle is much shorter. The humerus is relatively slender, more gracile
than the short, robust antebrachium and broad symmetrical manus. Due to the great
breadth of the plastron the humerus is restricted in motion to an anterior-to-lateral
quadrant. The ilium is tiny compared to the robust pubis and ischium. The hypoischium is
larger than in Sclerosaurus. The hind limb is comparatively gracile with a narrower
asymmetric pes.
In post-Odontochelys tested taxa, like Trionyx, all teeth and the premaxillae are
absent. In Trionyx the supraoccipital develops a long posterior process with a dorsal crest.
The parietals have a complex shape, flat dorsally, concave laterally with lateral wings
forming the base of the posttemporal fenestra. Completing the floor of the posttemporal
fenestra, the postparietal and tabular are lower than the laterally framing supratemporal.
The squamosal, still between the jugal and quadrate, is the smallest lateral element. The
laterally hollow quadrate is larger, dipping down to meet the descending posterior
mandible. The cervicals and caudals are shorter. The dorsal ribs extend slightly beyond a
complete set of interlocking osteoderms forming a low, flat carapace. The humerus
extends anteriorly, convergent with derived hard-shell turtles. Distinct from
Odontochelys, the limbs are gracile with elongate metapodials and phalanges.
Other Turtle Ancestor Candidates
In the LRT, Pappochelys (Middle Triassic) nests with basal placodonts close to
Palatodonta (Middle Triassic). The basal pachypleurosaur/sauropterygian,
Diandongosaurus (Early Triassic; Shang et al., 2011), is a proximal out-group taxon.
Placodonts have dorsal and lateral temporal fenestrae inherited from Pennsylvanian basal
diapsids like Petrolacosaurus. Such temporal openings are not found in any turtle or
pareiasaur. Several other placodonts, not closely related to Pappochelys, but derived from
Placodus, develop a carapace universally considered convergent with that of turtles.
Also converging with turtles, Eunotosaurus had fewer and broader ribs than its
closest relatives in the LRT that preserve post-crania, like Eocasea. Unfortunately, the
closest tested relatives of Eunotosaurus: Acleistorhinus, Microleter and Delorhynchus,
are known chiefly from skulls. All have a lateral temporal fenestra that opens ventrally
and other traits not found in basal turtles or pareiasaurs.
Bever et al. (2015) reported an upper temporal fenestra in a juvenile
Eunotosaurus and so concluded it was a diapsid with implications for turtle ancestry. In
the LRT no diapsid and no turtle nests near Eunotosaurus. The supratemporal that should
have covered that opening was taphonomically dislocated closer to the midline of the
skull, but was not reported by Bever et al. The left supratemporal remained in place, but
with a small hole punched in the center, either incompletely ossified or taphonomically
damaged. As Bever et al. acknowledged: adult Eunotosaurus specimens do not have an
upper temporal fenestra. A supratemporal covers it.
Triassic Turtle Tracks
Lichtig et al. (2017) described Chelonipus ichnites from the early Middle Triassic
and the late Early Triassic. The authors matched those tracks to the pes of hard-shell taxa
like Proganochelys, distinct from Odontochelys. They concluded, “the resemblance of
these tracks to pareiasaur tracks (Pachypes) supports arguments of the origin of turtles
from pareiasaurs.”
Nomenclature Issues
Joyce et al. (2004) reported on the changing and confusing history of naming
turtle clades. Unfortunately their publication preceded the later discoveries of
Odontochelys, Bunostegos and Arganaceras. Joyce et al. (2004) also omitted Elginia and
Sclerosaurus. Based on the absence of these key taxa, Joyce et al. nested Meiolania and
Trionyx in the clade Cryptodira, even though neither has the ability to pull the head back
under the carapace.
The dual origin of turtles from pareiasaurs was not anticipated in prior
nomenclature. Turtle clade nomenclature needs to be updated to reflect the evolutionary
events and clades recovered here.
The traditional clade, Pantestudines, is defined only by extant taxa and the precise
composition remained unclear even to Joyce et al. (2004). Pantestudines has neither
utility, nor monophyly in the LRT.
The traditional clade, Testudines (Batsch, 1788; Chelonia + Chelus and
descendants, Joyce et al., 2004), remains monophyletic in the LRT, but no longer
includes several basal hard-shell turtles or any soft-shell turtles.
The traditional clade, Testudinata (Klein, 1760; first member of Pantestudines
with a complete turtle shell that is homologous with the shell present in Chelonia, Joyce
et al., 2004), continues to include Meiolania, Niolamia their last common ancestor and its
descendants. However, soft-shell turtles are now excluded from this clade.
The traditional clade Trionychia (Hummel, 1929; Trionyx and kin, Joyce et al.
2004) remains monophyletic in the LRT.
A new clade, Protestudinata (“before Testudo and kin”), is proposed for
Bunostegos, Elginia, their last common ancestor and all of its descendants, the hard-shell
A new clade Protrionychia (“before Trionyx and kin”) is proposed for
Sclerosaurus, Arganaceras, their last common ancestor and all of its descendants, the
soft-shell turtles.
A new clade Keratospiti (“horned house”) is proposed for Bunostegos,
Sclerosaurus, their last common ancestor and all of its descendants, which includes all
stem turtles and turtles. Not all clade members have a “horned house.”
A new clade Propareiasauria (“before pareiasaurs”) is proposed for
Stephanospondylus, Pareiasaurus, their last common ancestor and all of its descendants,
which includes all pareiasaurs, stem turtles and turtles.
All current candidates for turtle ancestry are tested here. The addition of relevant
taxa nests turtles with stem-turtle pareiasaurs and nests other candidates elsewhere,
confirming the pre-cladistic observations and assessment of Wm. King Gregory (1946),
who linked Elginia to turtles. Two clades of pareiasaurs arise from a sister to
Stephanospondylus. One clade produced large traditional pareiasaurs. The other clade
produced the stem turtles, Bunostegos, Elginia and Sclerosaurus. Spiky Elginia was basal
to a clade of hard-shell turtles starting with the spiky grade Meiolaniidae. Sclerosaurus
descendants lose their spikes more quickly, giving rise to Arganaceras, Odontochelys and
other soft-shell turtles, like Trionyx. The pareiasaur ancestry of turtles sheds light on the
identity of traditionally misidentified turtle cranial bones. The upper corner cranial bone
is no longer the squamosal, but the supratemporal. The real squamosal continues to form
the posterior rim of the skull, anterior to the quadrate, even if the quadrate is not visible
Ameghino, F. 1899. Sinopsis geológica paleontológica. Suplemento (adiciones y
correcciones). Censo Nacional, La Plata:1–13.
Batsch, A. J. G. C. 1788. Versuch einer Anleitung, zur Kenntniß und Geschichte der
Thiere und Mineralien. Akademische Buchhandlung, Jena:1–528.
Baur, G. 1887. On the phylogenetic arrangement of the Sauropsida. Journal of
Morphology 1:93–104.
Bever, G. S., T. R. Lyson, D. J. Field, and B. -A. S. Bhular. 2015. Evolutionary origin of
the turtle skull. Nature 525:239–242, doi:10.1038/nature14900
deBraga, M., and O. Rieppel. 1997. Reptile phylogeny and the interrelationships of
turtles. Zoological Journal of the Linnean Society 120:281–354.
Field, D. J., J. A. Gauthier, B. L. King, D. Pisani, T. R. Lyson, and K. J. Peterson. 2014.
Toward consilience in reptile phylogeny: miRNAs support an archosaur, not
lepidosaur, affinity for turtles. Evolution & Development 16:189–196.
Foth, C., and W. G. Joyce. 2016. Slow and steady: the evolution of cranial disparity in
fossil and recent turtles. Proceedings of the Royal Society B 283:
Gaffney, E. S. 1983. The cranial morphology of the extinct horned turtle, Meiolania
platyceps, from the Pleistocene of Lord Howe Island, Australia. Bulletin of the
American Museum of Natural History 175, article 4:361–480.
Gaffney, E. S. 1985. The cervical and caudal vertebrae of the cryptodiran turtle,
Meiolania platyceps, form the Pleistocene of Lord Howe Island, Australia.
American Museum Novitates 2805:1–29.
Gaffney, E. S. 1990. The comparative osteology of the Triassic turtle Proganochelys.
Bulletin of the American Museum of Natural History 194:1–263.
Gaffney, E. S. 1996. The postcranial morphology of Meiolania platyceps and a review of
the Meiolaniidae. Bulletin of the American Museum of Natural History 229:1–
Geinitz, H. B., and J. V. Deichmüller. 1882. Die Saurier der unteren Dyas von Sachsen.
Paleontographica, N. F. 9:1–46.
Graybeal, A. 1998. Is it better to add taxa or characters to a difficult phylogenetic
problem? Systematic Biology 47:9–17.
Gregory, W. K. 1946. Pareiasaurs versus placodonts as near ancestors to the turtles.
Bulletin of the American Museum of Natural History 86: article 6:275–326.
Hummel, K. 1929. Die fossilen Weichschildkro¨ten (Trionychia). Geologische und
Palaeontologische Abhandlungen 16:359–487.
Jalil, N. -E., and P. Janvier. 2005. Les pareiasaures (Amniota, Parareptilia) du Permien
supérieur du Bassin d’Argana, Maroc. Geodiversitas 27:35–132.
Joyce, W. G. 2007. Phylogenetic relationships of Mesozoic turtles. Bulletin of the
Peabody Museum of Natural History 48:3–102.
Joyce, W. G. 2015. The origin of turtles: a paleontological perspective. Journal of
Experimental Zoology 324B:181–193. (doi:10.1002/jez.b.22609)
Joyce, W. G., J. F. Parham, and J. A. Gauthier. 2004. Developing a protocol for the
conversion of rank-based taxon names to phylogenetically defined clade names,
as exemplified by turtles. Journal of Paleonotology. 78 (5):989–1013.
Kissel, R. 2010. Morphology, Phylogeny, and Evolution of Diadectidae (Cotylosauria:
Diadectomorpha) PhD. thesis. Toronto: University of Toronto Press:1–185.
Klein, I. T. 1760. Klassification und kurze Geschichte der VierfüßigenThiere (translation
by F. D. Behn). Jonas Schmidt, Lübeck:1–381.
Lee, M. S. Y. 1997. Pareiasaur phylogeny and the origin of turtles. Zoological Journal of
the Linnean Society 120:197–280.
Li , C., X. Wu, O. C. Rieppel, L. Wang, and L. Zhao. 2008. An ancestral turtle from the
Late Triassic of southwestern China. Nature 456:497–501.
Lichtig, A. J., S. G. Lucas, H. Klein, and D. M. Lovelace. 2017. Triassic turtle tracks and
the origin of turtles. Historical Biology DOI: 10.1080/08912963.2017.1339037
Lyson, T. R., G. S. Bever, B.-A. S. Bhullar, W. G. Joyce, and J. A. Gauthier. 2010.
Transitional fossils and the origin of turtles. Proceedings of the Royal Society B
6:830–833. (doi:10.1098/rsbl.2010.0371)
Lyson, T. R., B. S. Rubidge, T. M. Scheyer, K. de Queiroz, E. R. Schrachner, R. M.
Smith, J. Botha-Brink, and G. S. Bever. 2016. Fossorial origin of the turtle shell.
Current Biology 26:1887–1894.
Maddison, D. R., and W. P. Maddison. 1990. MacClade 4: Analysis of Phylogeny and
Character Evolution. Sinauer Associates, Inc., Sunderland, MA.
Meyer, H. von. 1859. Sclerosaurus armatus aus dem bunten Sandestein von Rheinfelsen.
Palaeontographica 7:35–40.
Newton, E. T. 1893. On some new reptiles from the Elgin Sandstone: Philosophical
Transaction of the Royal Society B 184:473–489.
Owen, R. 1882. Description of fossil remains of two species of Megalanian genus
(Meiolania) from "Lord Howe Island." Philosophical Transactions of the Royal
Society London, series B 177:471–480.
Owen, R. 1888. On parts of the skeleton of Meiolania platyceps (Owen). Philosophical
Transactions of the Royal Society London, series B, 179:181–191.
Rieppel, O., and M. deBraga. 1996. Turtles as diapsid reptiles. Nature 384:453–455.
Romer, A. S. 1925. Permian amphibian and reptilian remains described as
Stephanospondylus. Journal of Geololgy 33:447–463.
Schoch, R. R., and H.-D. Sues. 2015. A Middle Triassic stem-turtle and the evolution of
the turtle body plan. Nature 523:584–587. (doi:10.1038/nature14472)
Seeley, H. G. 1892. On a new reptile from Welte Vreden (Beaufort West), Eunotosaurus
africanus (Seeley). Quarterly Journal of the Geological Society of London
Shang, Q. –H., X. –C. Wu and C. Li. 2011. A new eosauropterygian from Middle
Triassic of Eastern Yunnan Province, Southwestern China. Vertebrata PalAsiatica
Simpson, G. G. 1938. Crossochelys. Eocene horned turtle from Patagonia. Bulletin of the
American Museum of Natural History 74:221–254.
Stappenbeck, R. 1905. Uber Stephanospondylus n. g. und Phanerosaurus H. v. Meyer:
Zeitschritt der Deutschen Geologlschen Gesellschnft 57:380–437.
Sterli, J., and M. de la Fuente M. 2011. Re-description and evolutionary remarks on the
Patagonian horned turtle Niolamia argentina Ameghino, 1899 (Testudinata,
Meiolaniidae). Journal of Vertebrate Paleontology 31:1210–1229.
Sues, H. –D., and R. R. Reisz. 2008. Anatomy and phylogenetic relationships of
Sclerosaurus armatus (Amniota: Parareptilia) from the Buntsandstein (Triassic)
of Europe. Journal of Vertebrate Paleontology 28:1031–1042. doi: 10.1671/0272-
Swofford, D. 2002. PAUP*: Phylogenetic Analysis Using Parsimony (*And Other
Methods). Version 4.0b10. Sinauer Associates, Inc., Sunderland, MA.
Tsuji, L. A., C. A. Sidor, J. S. B. Steyer, R. M. H. Smith, N. J. Tabor, and O. Ide. 2013.
The vertebrate fauna of the Upper Permian of Niger—VII. Cranial anatomy and
relationships of Bunostegos akokanensis (Pareiasauria). Journal of Vertebrate
Paleontology 33:747–763. doi:10.1080/02724634.2013.739537
End section with: Submitted February 23, 2018; accepted Month DD, YYYY
Figure captions
FIGURE 1. Subset of the large reptile tree (
focusing on the dual turtle clades and their proximal outgroups. Intended for 1 column.
FIGURE 2. The skulls of Elginia, Niolamia, Meiolania and Proganochelys in three
views. Squamosal and supratemporal are reidentified here based on Elginia. Intended for
page width.
FIGURE 3. Select views of the skulls of Sclerosaurus, Arganceras, Odontochelys and
Trionyx. Squamosal and supratemporal are reidentified here based on Elginia. Intended
for page width.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Turtle (Testudines) tracks, Chelonipus torquatus, reported from the early Middle Triassic (Anisian) of Germany, and Chelonipus isp. from the late Early Triassic (Spathian) of Wyoming and Utah, are the oldest fossil evidence of turtles, but have been omitted in recent discussions of turtle origins. These tracks provide significant clues as to how early the turtle Bauplan originated. Turtle trackways are quite distinctive: the manus and pes form tracks nearly parallel to the midline and indicate an unusually wide gait in which the trackway width is nearly equal to the stride length. These tracks do not fit what would be expected to be made by Triassic Pappochelys or Odontochelys, a supposed prototurtle and an early turtle, respectively. In contrast, these tracks are consistent with what would be expected from the Triassic turtles Proganochelys and Palaeochersis. The features inferred to be present in Triassic turtle tracks support the notion that Odontochelys is a derived aquatic branch of the turtle stem lineage rather than the ancestral state of all turtles. Chelonipus also resembles the Permian track Pachypes dolomiticus, generally assigned to a pareiasaur trackmaker. These revelations highlight the need to consider all available evidence regarding turtle origins, rather than just the body fossils.
Full-text available
The turtle shell is a complex structure that currently serves a largely protective function in this iconically slow-moving group. Developmental and fossil data indicate that one of the first steps toward the shelled body plan was broadening of the ribs (approximately 50 my before the completed shell). Broadened ribs alone provide little protection and confer significant locomotory and respiratory costs. They increase thoracic rigidity, which decreases speed of locomotion due to shortened stride length, and they inhibit effective costal ventilation. New fossil material of the oldest hypothesized stem turtle, Eunotosaurus africanus (260 mya) from the Karoo Basin of South Africa, indicates the initiation of rib broadening was an adaptive response to fossoriality. Similar to extant fossorial taxa the broad ribs of Eunotosaurus provide an intrinsically stable base on which to operate a powerful forelimb digging mechanism. Numerous fossorial correlates are expressed throughout Eunotosaurus’ skeleton. Most of these features are widely distributed along the turtle stem and into the crown clade, indicating the common ancestor of Eunotosaurus and modern turtles possessed a body plan significantly influenced by digging. The adaptations related to fossoriality likely facilitated movement of stem turtles into aquatic environments early in the groups’ evolutionary history, and this ecology may have played an important role in stem turtles surviving the Permian/Triassic extinction event.
Full-text available
Transitional fossils informing the origin of turtles are among the most sought-after discoveries in palaeontology. Despite strong genomic evidence indicating that turtles evolved from within the diapsid radiation (which includes all other living reptiles), evidence of the inferred transformation between an ancestral turtle with an open, diapsid skull to the closed, anapsid condition of modern turtles remains elusive. Here we use high-resolution computed tomography and a novel character/taxon matrix to study the skull of Eunotosaurus africanus, a 260-million-year-old fossil reptile from the Karoo Basin of South Africa, whose distinctive postcranial skeleton shares many unique features with the shelled body plan of turtles. Scepticism regarding the status of Eunotosaurus as the earliest stem turtle arises from the possibility that these shell-related features are the products of evolutionary convergence. Our phylogenetic analyses indicate strong cranial support for Eunotosaurus as a critical transitional form in turtle evolution, thus fortifying a 40-million-year extension to the turtle stem and moving the ecological context of its origin back onto land. Furthermore, we find unexpected evidence that Eunotosaurus is a diapsid reptile in the process of becoming secondarily anapsid. This is important because categorizing the skull based on the number of openings in the complex of dermal bone covering the adductor chamber has long held sway in amniote systematics, and still represents a common organizational scheme for teaching the evolutionary history of the group. These discoveries allow us to articulate a detailed and testable hypothesis of fenestral closure along the turtle stem. Our results suggest that Eunotosaurus represents a crucially important link in a chain that will eventually lead to consilience in reptile systematics, paving the way for synthetic studies of amniote evolution and development.
Turtles (Testudinata) are a diverse group of amniotes that have a rich fossil record that extends back to the Late Triassic, but little is known about global patterns of disparity through time.We here investigate the cranial disparity of 172 representatives of the turtle lineage and their ancestors grouped into 20 time bins ranging from the Late Triassic until the Recent using two-dimensional geometric morphometrics. Three evolutionary phases are apparent in all three anatomical views investigated. In the first phase, disparity increases gradually from the Late Triassic to the Palaeogene with only a minor perturbation at the K/Textinct event. Although globalwarming may have influenced this increase, we find the Mesozoic fragmentation of Pangaea to be a more plausible factor. Following its maximum, disparity decreases strongly towards the Miocene, only to recover partially towards the Recent. The marked collapse in disparity is likely a result of habitat destruction caused by global drying, combined with the homogenization of global turtle faunas that resulted from increased transcontinental dispersal in the Tertiary. The disparity minimum in the Miocene is likely an artefact of poor sampling.
A series of partial shells and disarticulated shell elements is the basis for a nearly complete restoration of the carapace and plastron of Meiolania platyceps from the Pleistocene of Lord Howe Island, New South Wales. The carapace is broadly domed, with C-shaped bridge peripherals, but is not as arched as in living testudinids. A small cervical scale lies on a protuberant nuchal bone that has two midline articulation facets on its ventral surface for the 7th and 8th cervical vertebrae. The bones of the shell are usually fused with sutures present in a small number of disarticulated specimens. The shell bones tend to be thin in the central areas with the peripherals and other marginal areas thicker. The bone surface is coarsely textured with irregular grooves, pits, and foramina. The scale sulci are usually poorly defined, shallow troughs. In most specimens the posterior margin is strongly serrated with no caudal notch. There are 8 costals and 11 peripherals, but the number of neurals and the presence of a pygal and suprapygals are unknown. The first thoracic centrum faces directly anteriorly and the first thoracic rib reaches the axillary buttress on the third peripheral. In the plastron the epiplastra meet on the midline and bear a short median process, apparently not homologous to that in Proganochelys and Kayentachelys, that bifurcates dorsally and articulates with the scapula. The epiplastra are relatively broad and bear two pairs of scales, about equal in size, with moderate anterior projections. No other plastral scale sulci are determinable. The entoplastron is a large, ovoid bone with a long posterior median process that separates the hyoplastra for much of their length on the dorsal surface. The axillary and inguinal buttresses do not extend onto the costals, and contact peripherals 3 and 8, respectively. The hyoplastron and hypoplastron send digitate processes to the peripherals producing a thin, partially ligamentous contact with numerous fontanelles. Mesoplastra are absent. A large, irregular, median fontanelle is present at the junction of hyoplastra and hypoplastra. The posterior plastral lobe tapers strongly in contrast to the squared off end of the anterior lobe. Small articulation facets are present on the xiphiplastra for the pubis. The available shell material of Meiolania platyceps shows a great range of variation. Most shells have an extensive peripheral overhang but some specimens have almost no overhang. The shell sulci are usually broad and indistinct but in some specimens they are finely incised with raised edges. The limbs and girdles are completely known in Meiolania platyceps and are represented by both articulated and disarticulated specimens. The shoulder girdle is a stout element with a wide scapular-acromion angle and without a glenoid neck. The coracoid is wide and flat, flaring medially. The pelvis has large but widely separated thyroid fenestrae and a small epipubic process. The ilium flares slightly dorsally. There are two sacral ribs and, usually, the rib of the first caudal is fused to the second sacral. The humerus in Meiolania is expanded distally and proximally which contrasts to the narrower condition in nearly all cryptodires, but is similar to Proganochelys. The articular and surface morphology, however, is more similar to baenids and other primitive cryptodires. The ulna, radius, tibia, fibula, and femur of Meiolania are similar to those bones in Proganochelys and primitive cryptodires, except that in Meiolania they are generally stockier and more robust, with wider ends. The carpus of Meiolania has seven carpal bones: ulnare, intermedium, medial centrale, and four distal carpals. The manus formula is 2-2-2-2-2 with broad, flat unguals. The tarsus of Meiolania has an astragalocalcaneum showing no sign of sutures or fusion. Two distal tarsals are definitely known, but four were probably present. The pedal formula of Meiolania is 2-2-2-2-0. A revision of the family Meiolaniidae recognizes four genera: Niolamia (Eocene, Argentina), Ninjemys (Pleistocene, Queensland), Warkalania (Miocene, Queensland), and Meiolania (Miocene to Pleistocene, Northern Territory, Queensland, Lord Howe Island, New Caledonia). A PAUP analysis of 22 characters results in one cladogram: (Niolamia (Ninjemys (Warkalania, Meiolania))). The relationships of meiolaniids are analyzed using 17 taxa and 40 characters. Within the eucryptodires the shortest cladogram in a PAUP analysis is as follows: (Plesiochelyidae (Xinjiangchelys (Meiolaniidae (((Sinemys, Dracochelys) Ordosemys) (Chelydridae (Chelonioidea (Trionychoidea, Testudinoidea))))))). The data matrix consists of 19 cranial characters, 12 vertebral characters, and 9 shell characters. The cervical vertebrae prove to be particularly significant in resolving the extinct eucryptodires. In this analysis the biconcave caudal, ligamentous bridge attachment, and narrow epiplastra are seen to originate within the extinct eucryptodires. The family level name Sinemydidae is expanded to include Sinemys, Dracochelys, and Ordosemys.