The unusual tail of tethyshadros insularis (Dinosauria, Hadrosauroidea) from the adriatic island of the european archipelago

Abstract

The basal hadrosauroid Tethyshadros insularis from the uppermost Cretaceous of NE Italy lived on an island of the European archipelago in the Tethys Ocean. The tail of this dinosaur presents several apomorphic traits respect to the tails of other coeval hadrosauroids of the archipelago and of hadrosauroids in general. The estimated total length of the tail of the holotypic specimen shows that the tail was long, accounting for at least 56% of the total body length, relatively stiff and deep proximally, whereas it was whip-like distally. The reconstruction of the tail musculature by comparison with that of living archosaurs and other dinosaurs suggests that the posterior shift of the first haemapophysis affected the size and shape of the M.m. caudofemorales with important consequences on the locomotion of T. insularis. Somewhat peculiar stance and gait for this dinosaur are suggested also by limb features. The posterior shift of the vent and consequent longer distal tract of the intestine or a longer cloaca could increase the space for urine storage and urinary water reabsorption. The posterior shift of the vent could imply also longer oviducts and plausibly an increased number of eggs per clutch. Tail apomorphies of T. insularis may be related to the rugged and water-depleted karst landscape where the Italian dinosaur lived. The two main specimens of T. insularis differ in robustness possibly because of sexual dimorphism, ontogeny or high intraspecific variability.

Full text

THE UNUSUAL TAIL OF TETHYSHADROS INSULARIS (DINOSAURIA,
HADROSAUROIDEA) FROM THE ADRIATIC ISLAND OF THE EUROPEAN
ARCHIPELAGO
FABIO MARCO DALLA VECCHIA
Institut Català de Paleontologia Miquel Crusafont (ICP), Carrer de l’Escola Industrial 23, E-08201, Sabadell, Spain.
E-mail: fabio.dallavecchia@icp.cat
To cite this article: Dalla Vecchia F.M. (2020) - The unusual tail of Tethyshadros insularis (Dinosauria, Hadrosauroidea) from the Adriatic Island
of the European archipelago. Riv. It. Paleontol. Strat., 126(3): 583-628.
Rivista Italiana di Paleontologia e Stratigraa
(Research in Paleontology and Stratigraphy)
vol. 126(3): 583-628. November 2020
Abstract. The basal hadrosauroid Tethyshadros insularis from the uppermost Cretaceous of NE Italy lived on an
island of the European archipelago in the Tethys Ocean. The tail of this dinosaur presents several apomorphic traits
respect to the tails of other coeval hadrosauroids of the archipelago and of hadrosauroids in general. The estimated
total length of the tail of the holotypic specimen shows that the tail was long, accounting for at least 56% of the
total body length, relatively stiff and deep proximally, whereas it was whip-like distally. The reconstruction of the tail
musculature by comparison with that of living archosaurs and other dinosaurs suggests that the posterior shift of
the rst haemapophysis affected the size and shape of the M.m. caudofemorales with important consequences on the
locomotion of T. insularis. Somewhat peculiar stance and gait for this dinosaur are suggested also by limb features.
The posterior shift of the vent and consequent longer distal tract of the intestine or a longer cloaca could increase
the space for urine storage and urinary water reabsorption. The posterior shift of the vent could imply also longer
oviducts and plausibly an increased number of eggs per clutch. Tail apomorphies of T. insularis may be related to the
rugged and water-depleted karst landscape where the Italian dinosaur lived. The two main specimens of T. insularis
differ in robustness possibly because of sexual dimorphism, ontogeny or high intraspecic variability.
Received: April 29, 2020; accepted: July 23, 2020
Keywords: Axial skeleton; caudal vertebrae; functional morphology; insularity; latest Cretaceous; Karst.
IntroductIon
Complete and articulated caudal segments of
the vertebral column of dinosaurs are extremely
rare (Hone 2012). As a consequence, the actual tail
length is usually unknown in most dinosaur taxa
and it is object of interpretation in the skeletal re-
constructions. Some complete and articulated skel-
etons of small-sized dinosaurs have shown that,
when completely preserved, the tail is much longer
than previously supposed (e.g., Herner 2009).
Traditionally, little attention has been given
to vertebral column in the osteological and phy-
logenetic study of hadrosauroid dinosaurs (e.g.,
Wright & Lull 1942; Norman 2004; Horner et al.
2004). This because the vertebral morphology was
considered to be very conservative within the had-
rosauroids and the iguanodontians in general, with
variability limited manly to the shape of the neural
spine of the axis, degree of zygapophyseal develop-
ment in the cervicals, number of cervical and sacral
vertebrae, and elongation of the neural spines and
chevrons (e.g., Horner et al. 2004; Prieto-Marquez
2010; McDonald et al. 2012b; Norman 2015). This

Fabio Marco Dalla Vecchia
584
lack of information is particularly true for the cau-
dal vertebrae, which have usually been summarily
described in the literature. However, caudal ver-
tebrae are the most numerous elements in a had-
rosauroid skeleton (with the possible exclusion of
teeth) and therefore they are the ones that have the
highest probability to be found in the fossil record
as isolated elements.
Tail vertebrae can give information about the
development of tail musculature, allowing hypoth-
eses about locomotion and its ecological implica-
tions in extinct tetrapods (e.g., Persons & Currie
2011a-b, 2012 and 2014).
The preserved portion of the caudal vertebral
column is articulated in the holotype of the non-
hadrosaurid hadrosauroid Tethyshadros insularis Dalla
Vecchia, 2009c from the basal part of the Liburnian
Formation (upper Campanian or lower Maastrich-
tian) of NE Italy. Part of the tail is articulated also in
a second referred specimen from the same locality
and horizon. Furthermore, the tail of this dinosaur
revealed to be an exception to the traditional view
of the conservative morphology of the vertebrae,
presenting several apomorphic traits (Dalla Vecchia
2009c).
Here, the tail features of T. insularis are de-
scribed in detail, the total length of the tail of the
holotype is estimated and the caudal vertebrae are
compared with those of the other hadrosauroids
from the latest Cretaceous European archipelago.
The tail peculiarities of T. insularis are discussed and
hypotheses about their functional and adaptive sig-
nicance are advanced.
Institutional abbreviations – AMNH, American Museum
of Natural History, New York, U.S.A.; IPS, Institut de Paleontolo-
gia Dr. Miquel Crusafont (currently Institut Català de Paleontologia
“Miquel Crusafont,” ICP), Sabadell, Spain; IRSNB, Institut Royal
des Sciences Naturelles de Belgique, Brussels/Bruxelles, Belgium;
MB.R., Museum für Naturkunde, Berlin, Germany; MCD, Museu
de la Conca Dellà, Isona, Spain; MCSNT, Museo Civico di Storia
Naturale, Trieste, Italy; MDE, Musée des Dinosaures d’Esperaza,
Esperaza, France; MPZ, Museo Paleontológico de la Universidad de
Zaragoza, Zaragoza, Spain; NHMUK PV, The Natural History Mu-
seum (former British Museum of Natural History), London, U.K.;
ROM, Royal Ontario Museum, Toronto, Canada; SC, Italian State
collections; TMP, Royal Tyrell Museum of Palaeontology, Drum-
heller, Canada.
Materials, terminology and methods
The material object of this study consists of
the two specimens of T. insularis that preserve ar-
ticulated segments of the vertebral column, name-
ly the holotype SC 57021 (nicknamed “Antonio”)
and the referred specimen SC 57247 (nicknamed
“Bruno”). In this paper, I use the numbers of the
State collections that I attributed to the fossils in
1999 charged by the Ministero per i Beni Culturali
ed Ambientali, and have already been reported in
Dalla Vecchia (2008, 2009c). Personal observations
on SC 57247 were made mainly on the unprepared
specimen. SC 57247 was prepared in 2019, but it
has not been made available for this study. How-
ever, photographs of the prepared specimen could
be seen in the internet and the fossil was exhibited
to the public in the winter 2019-2020 at the Duino
Castle (Trieste), allowing some comparison.
The comparison with the material from the
uppermost Cretaceous of Romania referred to
Telmatosaurus transsylvanicus and deposited at the
NHMUK, and with the hadrosauroid vertebrae
from the Maastrichtian of Spain (Basturs Poble and
Sant Romà d’Abella localities and partly Els Nerets)
and France was based on the direct observation of
the specimens. The specimen TMP 1998.058.0001
was also observed personally. Information about
the specimens from the Maastrichtian localities of
Costa de les Solanes, Arén and partly Els Nerets
(Spain) is based on Prieto Márquez et al. (2019),
Cruzado-Caballero (2012) and Conti et al. (2020),
respectively. Information about the specimens from
Limburg and Germany was also taken from the lit-
erature.
Following Sereno (1998), the Hadrosauroidea
are all dinosaurs more closely related to Parasauro-
lophus than Iguanodon. In the phylogenetic analyses
published during the last 12 years, T. insularis falls
always within the Hadrosauroidea and outside the
Hadrosauridae, but its phylogenetic afnities are
somewhat variable. T. insularis forms a trichotomy
with Telmatosaurus transsylvanicus and the Hadrosau-
ridae in the strict consensus tree by Dalla Vecchia
(2009c). It is nested in a more basal position in the
Adams consensus tree of McDonald et al. (2012b:
g. 10), forming a trichotomy with Levnesovia transo-
xiana and the more advanced hadrosauroids (Telma-
tosaurus transsylvanicus results to be a hadrosaurid in
this analysis). T. insularis is lower in the tree than the
Hadrosauridae and their sister taxon Telmatosaurus
transsylvanicus in the strict consensus tree of Norman
(2015: g. 50), being nested between Probactrosaurus
gobiensis and Levnesovia transoxiana as the basal taxon
of the Hadrosauromorpha (sensu Norman 2014).

The tail of Tethyshadros insularis 585
T. insularis is the sister taxon of Telmatosaurus trans-
sylvanicus and is lower in the tree than Nanyangosau-
rus zhugeii and Zhanghenglong yangchengensis, which are
lower than the Hadrosauridae, in the simplied strict
consensus tree of Wang et al. (2015: g. 4); it has a
similar relationships also in the strict consensus tree
by Xing et al. (2014: g. 13). T. insularis is the sister
taxon of the Hadrosauridae and is higher in the tree
than Telmatosaurus transsylvanicus in the strict consen-
sus trees by Prieto-Márquez et al. (2016, 2019). T. in-
sularis is the sister taxon of Telmatosaurus transsylvani-
cus, and T. insularis + Telmatosaurus transsylvanicus is the
sister group of the Hadrosauridae, in the single most
parsimonious tree of the analysis by Xu et al. (2018).
Finally, T. insularis is placed between Claosaurus agilis
and Eotrachodon orientalis, outside the Hadrosauridae
and lower in the tree than Telmatosaurus transsylvanicus,
in the strict consensus tree by Conti et al. (2000).
“Anterior” and “posterior” are used here in
preference to “cranial” and “caudal” to avoid confu-
sion, because this paper deals mainly with caudal ver-
tebrae. Zygapophyses, parapophysis and diapophysis
are considered in their original meaning of process-
es, not as synonym of zygapophyseal, parapophy-
seal and diapophyseal facets. Following Norman
(1980: 43), the proximal caudal vertebrae are those
with laterodorsal pleurapophyses (sensu Wild 1973:
62) on the centrum. Pleurapophyses are reported in
literature also as “transverse processes” or “caudal
ribs”, but the transverse process and the caudal rib
are two distinct components of the pleurapophysis
which cannot be identied as such when they are
fused to each other. The mid-caudal vertebrae are
those without the pleurapophyses, but bearing hae-
mapophyses (haemal arches or chevrons), whereas
the distal caudal vertebrae lack pleurapophyses and
haemapophyses. The neural arch is composed of the
pedicels, zygapophyses and neural spine. The haem-
apophysis is composed of the pedicels and the hae-
mal spine. The backward slope of the neural spines
with respect to the vertical is the angle made by the
line connecting the mid-point of the basal part of
the spine and the mid-point of the apical part of
the spine (in lateral view) with the elongation axis
of the centrum minus 90°. The height of the cen-
trum is measured about at mid-centrum from the
ventral point of minimum central depth to the cor-
responding dorsal end of the centrum. Angles have
been measured on digital images with the software
ImageJ.
Following Romer (1977), the large intestine of
extant archosaurs (crocodilians and birds), i.e., the
straight portion of the intestinal tract between the
ileum and the cloaca, is here reported as the colon.
The theoretical background of using the re-
lationships between skeletal elements and soft tis-
sues in extant archosaurs to infer these relationships
in extinct dinosaurs is summarized in Ibiricu et al.
(2014). Crocodilians and birds represent the two
poles of the extant phylogenetic bracket (sensu Wit-
mer 1995) for non-avian dinosaurs. Although birds
share a more recent common ancestor with hadro-
sauroids than do crocodilians, they have an extreme-
ly reduced tail and a consequently specialized tail
musculature (Gatesy 1995). Crocodilian tails more
closely resemble those of the hadrosauroids in rela-
tive size, number of vertebrae, and development of
processes for muscle attachment, but crocodilians
are obligate quadrupeds with a prevailing sprawling
posture. This must be considered when using the
relationships between skeletal elements and soft tis-
sues in extant archosaurs to infer these relationships
in extinct dinosaurs, mainly when the latter are spe-
cialized taxa.
Specimens from Blasi site of Spain may be in-
dicated with a Museum inventory number (acronym
MPZ, Museo Paleontológico de la Universidad de
Zaragoza) or a eld number (acronym BLA).
The latest Cretaceous interval is considered to
be composed of the Campanian and Maastrichtian
Ages, 83.6-66 Ma.
SyStematIc palaeontology
ORNITHOPODA Marsh, 1881
IguanodontIa Sereno, 1986
Hadrosauroidea Cope, 1869 sensu Sereno, 1986
Tethyshadros Dalla Vecchia, 2009c
Tethyshadros insularis Dalla Vecchia, 2009c
Figs. 1-6 and 16, SI gs. 1-4 and 6.
Holotype - SC 57021, nearly complete and articulated skel-
eton lacking part of the tail.
Referred specimens - SC 57022, SC 57023, SC 57025, SC
57026, SC 57247, SC 57256 (see Dalla Vecchia 2009c). All the speci-
mens are at the MCSNT.
Distribution – Upper Campanian or lower Maastrichtian,
Villaggio del Pescatore locality, Trieste Province, north-eastern Italy.

Fabio Marco Dalla Vecchia
586
Description
The caudal vertebrae of the holotype of
Tethyshadros insularis have been described by Dalla
Vecchia (2009c). Only details useful for the com-
parison with the caudal vertebrae of the specimen
SC 57247 and the other latest Cretaceous hadrosau-
roids of the European archipelago and for the dis-
cussion about the functional implications of the
peculiar tail of T. insularis are considered here.
The tail of the holotype (Fig. 1) preserves
the rst 33 vertebrae, which are articulated, and is
incomplete distally. The missing segment of the
vertebral column remained in situ because the layer
containing it was twisted by a fault and its extrac-
tion was difcult. The preserved segment of the
caudal vertebral column is ca. 1650 mm long. It
is disturbed and slightly dislocated at the caudal/
sacral transition by a synsedimentary fault (see Fig.
1). The rst caudal vertebra was considered the rst
one with a tongue-shaped pleurapophysis like those
of the following proximal caudals. There are 13 or
14 proximal caudal vertebrae (it is unclear whether
caudal 14 has a very small bump that can be consid-
ered a pleurapophysis or no process at all), followed
by 20 or 19 mid-caudal vertebrae. A haemapophy-
sis occurs between caudal vertebrae 32 and 33, thus
the whole distal segment of the tail is missing and
plausibly some mid-caudal vertebrae are also miss-
ing (probably as many vertebrae as the preserved
ones, see below). Therefore, the tail of T. insularis
is not “complete or near-complete” as suggested by
Hone (2012: caption of tab. 2).
If not specied otherwise, the description of
the anatomical details of the tail below is based on
the holotype.
The specimen SC 57247 is a portion of an
articulated and probably complete skeleton that was
seriously damaged and partly destroyed during the
eld work in 1998. When recovered in 1999, it was
composed of a main block (SC 57247A, Fig. 2), sev-
eral smaller limestone blocks bearing various por-
tions of the skeleton and hundreds of small bone
fragments. Parts of the specimen remained exposed
in situ and have been collected over 15 year later.
According to press releases, blocks and fragments
were later (2018) assembled and prepared by formic
acid for exhibition purposes. As recovered in 1999,
Fig. 1 - Caudal segment of the vertebral column of the holotype of T. insularis (SC 57021). A) Photograph; B) interpretative drawing. The red
lines mark the synsedimentary fault displacing the caudal segment from most of the sacrum. Abbreviations: csv, caudosacral vertebra;
cv1-30, caudal vertebrae 1-30; fe, right femur; hm 5-25, haemal arches 5-25; il, right ilium; is, right ischium; pl, pleurapophysis; sv1-9,
sacral vertebrae 1-9. The scale bar equals 500 mm.

The tail of Tethyshadros insularis 587
the surface of the main block contained the articu-
lated proximal portion of the caudal vertebral col-
umn, which was composed of nine or ten vertebrae,
part of the sacrum, part of the pelvic girdle ad the
proximal portion of the articulated left femur (Fig.
2). Unlike the holotype, the ischia of this specimen
are not stretched by a synsedimentary fault (Dalla
Vecchia 2009c, g. 5D), although the distal portions
of the shafts are slightly displaced by a small fault.
Photographs made available in the internet and oth-
er mass-media after preparation show that the prox-
imal portion of the tail, from caudal 1 to 9 is com-
posed of perfectly articulated vertebrae exposed
in left lateral view. The rst vertebra posterior to
the fused sacral vertebrae is possibly a caudosacral.
The tail is ~180° bent at level of vertebrae 9-10 and
the tail segment posterior to vertebra 10 lies below
and nearly parallel to the proximal segment. This
unusual bending of the tail was caused some time
after burial by the folding of the still plastic car-
bonate layer containing the specimen (see below).
Caudal 10 is on the hinge of the fold and shows its
left lateral side. The bent segment is made of 13 ar-
ticulated vertebrae with their haemapophyses. Cau-
dals 11-23 are ventrally and ventrolaterally exposed
and dorsoventrally attened. Neural arches are dor-
soventrally compressed, thus the neural spines of
caudals 11-18 and 21 project laterally from the left
side of the vertebra and can be seen in left lateral
view. This tail segment is slightly disarticulated at
caudals 14-15; caudal 20 is just partially preserved
because is damaged by a fracture. Between caudal
23 and the following vertebra (vertebra 24) there is
a gap, probably because of the disarticulation of the
tail. In fact, also the following caudal (vertebra 25)
is isolated and separated from the most posterior
vertebrae. The latter are located on a further fold of
the layer (i.e., the specimen-bearing layer is folded
S-wise). This distal portion of the preserved part of
the tail is made, in anteroposterior order, of a single
isolated vertebra like caudal 25 (exposed in ventral
view), a series of 13 articulated vertebrae in ventral
Fig. 2 - Part of the specimen SC
57247 (SC 57247A) pre-
serving part of the pelvic
girdle and the articulated
proximal portion of the
caudal segment of the ver-
tebral column. A) The enti-
re SC 57247A; B) particular
of caudal vertebrae 5 to 8.
Abbreviations: ce5-8, centra
5-8; cv1-9, caudal vertebrae
1-9; fe, left femur; hm1, rst
haemal arch; is, ischium (the
right ischium is in parenthe-
ses); pu, pubis; ot, ossied
tendons; scr, synsacrum.
The scale bars equal 50 mm.

Fabio Marco Dalla Vecchia
588
view, a detached segment of four articulated and
smaller vertebrae possibly in lateral view and at least
two scattered and more distal elements. Therefore,
the total number of preserved vertebrae is at least
45 in SC 57247. The different sizes of the centra of
the two posterior tail segments suggest that some
vertebrae may be missing in between them.
Centra
Caudal centra are all longer than high along
the whole series in both specimens (measurements
of the holotype vertebrae are reported in SI, Tab.
1). This was considered an autapomorphy of T. in-
sularis by Dalla Vecchia (2009c). From caudal 16 on,
the length/height ratio (hereafter l/h) is ≥2 in the
holotype. The shape of the centra changes along
the tail. Centra bear pleurapophyses that project lat-
erally at level of the centrum-neural arch boundary
(Fig. 3) at least up to vertebra 13, possibly up to ver-
tebra 14. A sharp longitudinal ridge runs along the
whole dorsal part of each of the lateral surfaces of
the centrum in the following vertebrae 15-19 (Fig.
4B-C; SI, Fig. 1). These lateral ridges make hexago-
nal the centrum in anterior and posterior views. In
vertebrae 20-25, the lateral ridge gradually migrates
dorsally toward the dorsolateral margin of the cen-
trum; in these centra, a groove separated the ridge
from the dorsolateral margin of the centrum (SI,
Fig. 2A). The lateral ridge and groove disappear in
caudal 26 (SI, Fig. 2A). Practically, the upper lateral
side of the hexagon gradually fades into the dorsal
side. The centra of caudal vertebrae 23 to 33 have
the shape of a hemicylinder with the convex side
ventral, the wide at side dorsal and semicircular ar-
ticular facets (SI, Fig. 2). The hemicylindrical centra
are twice broader than high (SI, Tab. 1), as it can
be appreciated in vertebra 32, where the posterior
articular facet is 27 mm wide and 13 mm high (not
considering the projecting processes for the hae-
mapophyses; SI, Fig. 2B). The centra are probably
broader than high at least starting from caudal 15.
Centra are amphicoelous at least starting
from vertebra 8; the articular facets of the last 15
vertebrae are deeply concave (SI, Fig. 2B-D). They
are such also in SC 57247. The posterior articular
facet of centrum 32 apparently shows a small and
circular facet on its upper left corner (SI, Fig. 2B-C),
but it is unclear whether it is a real vertebral feature
or an artefact of preparation. The lateral surfaces
of the centra appear to be nearly at in the rst ver-
tebrae; they are concave from at least caudal 9 to 14.
They are concave ventral to the longitudinal ridge in
vertebrae 15 to 17 or 18. The ventral margin of the
centra in lateral view is markedly concave in centra
18-33.
The posterior processes for the chevrons are
well developed in all of the preserved centra start-
ing from caudal vertebra 7. Articular facets for the
pedicels of the haemal arches occur also at the an-
teroventral corner of the centrum at least up to ver-
tebra 22.
Intercentral spaces are evident posterior to
caudal 8 (Fig. 4) and range from 0 (between centra
14-15) to ca. 7.5 mm (e.g., between centra 23-24)
in length. Intercentral spaces are variable because
of the slight displacements of the centra, although
all of the vertebrae are articulated at level of the
zygapophyses.
The centra of the rst caudals of SC 57247
are also longer than high, but their elongation is
Fig. 3 - Proximal caudal vertebrae of
the holotype of Tethyshadros
insularis (SC 57021) with meat
cleaver-like neural spines and
tongue-like pleurapophyses.
A) Photograph; B)
interpretative drawing
of caudal vertebra 4.
Abbreviations: ce, centrum;
cv4, caudal vertebra 4; ne,
‘neck’; ns, neural spine;
pl, pleurapophysis; poz,
postzygapophysis; prz,
prezygapophysis; ot, ossied
tendons. The scale bars
equal 50 mm.

The tail of Tethyshadros insularis 589
minor and centra appear to be more massive than
those of the holotype. The centra of caudal ver-
tebrae 5-7 were 60-61 mm long and 53-53.5 mm
high (l/h was 1.13-1.15, before preparation), where-
as caudal 5 is 65 mm long with l/h = 1.51 in the
holotype (SI, Tab. 1). Also the centra of the bent
segment of the tail of SC 57247, corresponding to
vertebrae 10-23, and those of the following isolated
caudal vertebrae 24 and 25, appear as more mas-
sive than the corresponding centra of the holotype.
However, these vertebrae are exposed in ventral
view in SC 57247, whereas those of the holotype
are visible in lateral view, thus differences may be
just apparent. Caudal centrum 12 of SC 57247 has
two small and paired neurovascular foramina set in
the middle on the ventral side. There is practically
no intervertebral space between the centra in the
perfectly articulated rst 13 caudal vertebrae of SC
57247, indicating that intervertebral disks were thin,
if present.
Pleurapophyses
In the holotype of T. insularis, the pleur-
apophyses occur at least up to vertebra 13. Vertebra
14 is damaged, thus the presence or the absence of
the pleurapophysis cannot be checked. The pleur-
apophyses of caudal vertebrae 1-5 are tongue-
shaped, very broad anteroposteriorly, attened dor-
soventrally and directed laterally (Figs. 1, 3 and 5).
The pleurapophyses of caudals 2-5 had broke in
the proximal part and the distal fragment is turned
ventrally (Figs. 3 and 5). The pleurapophyses of the
following vertebrae 6 and 8-11 (the pleurapophysis
of caudal 7 is broken and missing) are not much re-
duced in lateral extent respect to the preceding, but
they are much narrower anteroposteriorly and with
a marked constriction at mid-shaft (SI, Fig. 3). It
is impossible to establish how much restoration af-
fected the shape of these pleurapophyses, which are
partly reconstructed (SI, Fig. 3B). Vertebrae 12 and
13 (possibly vertebra 14 too) present only a small,
wing-like, lateral knob (Fig. 4B; SI, Fig. 1A).
The pleurapophyses of the exposed left side
of SC 57247 were all missing in the original block
SC 57247A (Fig. 2), but they are preserved in the
prepared specimen. They can be identied at least
up to caudal 12. Assuming that they originated from
the preparation of the counter-block of SC 57247A
and are not totally reconstructed, they appear to
be slightly different from those of the holotype. In
fact, the pleurapophyses of SC 57247 are spatula-
shaped (narrow proximally and much expanded dis-
Fig. 4 - The articulated caudal vertebrae of the holotype of Tethyshadros insularis (SC 57021) from caudal 6 to 33. A) Segment 6-11; B) segment
11-18; C) segment 15-22; D) segment 23-33. Abbreviations: cv6-33, caudal vertebrae 6-33; hm1-26, haemal arches 1-26; is, ischium; ot,
ossied tendons. The scale bars equal 50 mm.

Fabio Marco Dalla Vecchia
590
tally) instead of tongue-shaped, at least up to caudal
5. In the rst ve caudal vertebrae, the caudal ribs
are apparently unfused to their transverse processes;
some of them have been wrongly glued to the centra
ventral to the short transverse processes. All of the
pleurapophyses posterior to caudal 5 and the single
caudal ribs (if they are not broken pleurapophyses)
are damaged or partly preserved and reconstructed;
thus, their apparent shape may be misleading and
should be considered carefully.
Neural spines
The neural arches and neural spines of the
most proximal caudal vertebrae are the tallest of the
entire vertebral column, but are not much higher
than those of the distal dorsal vertebrae. In cau-
dal vertebra 4 of the holotype (Fig. 3), the heights
of the neural arch and neural spine are 2.7 and 2.1
times the centrum height, respectively. The tallest is
the neural arch of caudal 3, posterior to which the
height decreases gradually. However, the neural arch
of caudal 7 is slightly lower than those of caudals
5 and 8 and the neural arches of caudals 11 and
12 are slightly higher than that of caudal 10. This
originates two shallow depressions along the dorsal
prole of the tail on caudals 7 and 10 (Fig. 1). The
decrease in height of the neural arches is greater
from caudals 12 to17 where the tail reduces much
of its depth. The neural arch of vertebrae 20 to 30
is very low and small (Fig. 4D; SI, Fig. 2A); it is
broken and missing in caudal 32, where the narrow
neural canal is exposed, and fragmentary in caudal
31 (SI, Fig. 2C-D). In these vertebrae, the neural
arch is set on the middle of a at and broad dorsal
side of the centrum (SI, Fig. 2C-D).
The neural spines of caudal vertebrae 1 to 5
or 6 (the neural spine of caudal 6 is damaged) are
anteroposteriorly wide, laterally at and meat cleav-
er-shaped (wrongly reported as “hatched-shaped”
[sic] in Dalla Vecchia 2009c: 1107), with a ‘neck’ just
dorsal to the postzygapophyses (Figs. 1, 3 and 5).
These neural spines are slightly recurved forward.
The neural spines of the last dorsal and sacral ver-
tebrae are probably similar to those of the rst cau-
dals (Dalla Vecchia 2009c). The neural spines of
caudal vertebrae 7 to 17 are fan- or spatula-shaped
in lateral view, aring apically with a slightly con-
vex dorsal margin, inclined posteriorly and slight-
ly recurved (Figs. 1 and 4A-C; SI, Fig. 1 and 3A).
The apical expansion of the neural spine practi-
cally ends with caudal 17 and the expansion gradu-
ally decreases after caudal 14 (Fig. 4B-C; SI, Fig. 1),
which is possibly the last vertebra with a hint of
a pleurapophyses. The neural spines of the follow-
ing caudals have a ‘petaloid’ outline with a rounded
apex and a maximum anteroposterior width which
is reached before the apex (Fig. 4C-D; SI, Fig. 2A).
These petaloid neural spines are straight and in-
clined backward (Figs. 1 and 4B-D). Neural spine
18 has a transitional shape (SI, Fig. 1B). The poste-
rior slopes of the neural spines with respect to the
vertical in caudals 4, 8, 15, 19 and 26 are about 24°,
47°, 45°, 36° and 60°, respectively.
In SC 57247 (Fig. 2), the neural spines of the
caudal vertebrae are similar to those of the holotype,
although they show some minor differences. Cau-
dals 5-7 are better preserved than in the holotype,
showing that the transition from the meat cleaver-
shaped and spatula-shaped neural spines is gradual
from caudal 4 to 6. The dorsal margin of spines 5
to 8 is concave in the middle (in lateral view) making
heart-shaped the expanded dorsal portion of the
spine. This is not the case of the holotype (Figs. 4a
and 5). The neural spines of caudals 5-9 are slightly
more inclined posteriorly than the corresponding in
the holotype. The spine of caudal 9 of SC 57247 is
less expanded apically than those of the preceding
caudals and its dorsal margin is rounded; the apical
portion of the spine of caudal 9 of the holotype is
more expanded and ‘cut’. In SC 57247, the spines
of the following caudals 10-17 are similar to that of
caudal 9, with a slight constriction at mid-spine in
lateral view.
Zygapophyses
Zygapophyses are still articulated each other
in all of the vertebrae except in caudals 4 and 5. The
prezygapophyses are relatively long and slender,
whereas the postzygapophyses are extremely short.
The rounded articular facet of the postzygapophy-
sis of caudal 4 faces lateroposteriorly and slightly
ventrally (Fig. 3). The articular facets of the zyg-
apophyses are nearly vertical, at least through caudal
26, thus a limited lateral movement was allowed to
the tail. The articular facets of the zygapophyses are
sub-vertical at least up to caudal vertebra 10 in SC
57247A.
Haemal arches (chevrons)
The rst haemal arch articulates between cau-

The tail of Tethyshadros insularis 591
dal vertebrae 7 and 8 (Fig. 1) in the holotype. The
position of this haemal arch corresponds to the
rst fully fan-shaped neural spine. All of the fol-
lowing 25 haemal arches are articulated to the cor-
responding centra (Fig. 4). Haemal arch 3 and partly
also haemal arch 1 are rotated and shows their Y-
like shape in anteroposterior view; the others are
exposed in right lateral view, excluded haemal arch
2 that is overlapped by haemal arches 1 and 3. The
haemal arches change in shape along the vertebral
series. Haemal arches 1-5 have long and rod-like
haemal spines which are only slightly compressed
transversely and taper distally. Haemal arches 1-4
are slightly recurved. Haemal arches 1-5 slope pos-
teriorly in a way that each one contacts the follow-
ing (Figs. 1 and 4A-B). The higher backward slope
of the rst haemal arches with respect to the fol-
lowing others corresponds to the slope of the distal
end of the ischia (i.e., the rst haemal arches and
distal end of the ischia are more or less parallel to
each other; Fig. 1).
Haemal arch 1 is the longest (250 mm); length
decreases gradually from haemal arches 6 (143 mm
long) to 26. The distal ends of the haemal spines
of haemal arches 6 to 26 are anteroposteriorly ex-
panded (Fig. 6). The distal expansion is asymmetri-
cal and skewed backward in haemal arches 6 to 13,
giving the haemal arch a boot-like shape in lateral
view (Fig. 6A-F; they are asymmetric chevrons, ac-
cording to the denition by Otero et al. 2012). This
asymmetry is incipient in haemal arch 6 and increas-
es up to haemal arch 12 (Figs. 4C and 6A-E), whose
backward expansion of the spine is longer than the
rest of the haemapophysis and contacts the anterior
side of the following haemal arch (Fig. 6E). Haemal
arch 13 (which is articulated to caudals 19 and 20) is
similar to haemal arch 12 (Fig. 6F) and shows a pat-
tern of ne diagonal scarring on the lateral surface
of the spine (Fig. 6F); this scarring can be observed
also in chevrons 8-11 and probably is present in all
distally expanded haemapophyseal spines. Haemal
arches 14 to 26 are shorter than the preceding ones
and have a somewhat bilobate (forked, according to
the denition by Otero et al. 2012) distal expansion
of the spine (Fig. 6G-K). This expansion is slightly
asymmetrical in haemal arch 14, which bears a lon-
ger posterior lobe (Fig. 6G), whereas the lobes are
more or less symmetrical in the successive haemal
arches; however the posterior lobe is longer again in
the last preserved haemal arch (Fig. 6K). The rst
haemal arches are much longer than the neural arch-
es of the corresponding vertebrae, then the dispro-
portion decreases gradually and lengths are similar
in corresponding neural arches and haemal arches
starting from about haemal arch 14 on.
The haemal arches were not evident in the un-
prepared SC 57247A (Fig. 2). After assemblage and
preparation, haemal arches are preserved at least up
to caudal 24. Haemal arches are exposed in lateral
view, except haemal arches 9, 11-13, and ?19 which
are exposed in anterolateral or posterolateral views,
showing both pedicels. The rst three haemal arches
Fig. 5 - Proximal caudal vertebrae 1-7 of the holotype of Tethyshadros insularis (SC 57021) and their broad pleurapophyses. Abbreviations: cv1-7,
caudal vertebrae 1-7; pl, pleurapophysis. The scale bars equal 100 mm.

Fabio Marco Dalla Vecchia
592
occur in the hinge zone of the layer fold. Haemal
arches 2 and 3 are articulated to caudals 9/10 and
10/11, respectively. Haemal arch 1 crosses haemal
arch 2 at mid-shaft and is detached from its articu-
lation with caudals 8/9. Therefore, the rst haemal
arch appears to occur in a slightly more distal posi-
tion respect to the holotype. However, the caudo-
sacral was counted as the rst caudal vertebra in
SC 57247A. In the holotype, the partially preserved
vertebra before the rst caudal (Fig. 1) could be
the last sacral (as supposed by Dalla Vecchia 2009c)
or a caudosacral; in the latter case, the position of
the rst haemal arch would be the same in the two
specimens. The rst three haemal arches are short-
er and more robust than those of the holotype,
but their spine is incompletely preserved distally;
Fig. 6 - Haemapophyses of the holotype of Tethyshadros insularis (SC 57021), right lateral view. Haemapophyses A) 3-8; B) 9; C) 10; D) 11; E)
12; F) 13; G) 14-16; H) 20; I) 21; J) 22; and K) 26. Larger arrows point to the ridges on haemapophyses 12 and 13; smaller arrows point
to the ne scarring on haemapophysis 13. Not drawn to scale. As for scale, see Figure 4.

The tail of Tethyshadros insularis 593
they are rod-like and their spines have an elliptical
cross-section. Haemal arches 4 and 5 are articulat-
ed with caudals 11/12 and 12/13, respectively. The
following haemal arches are slightly detached and
shifted from their anatomical articulation with the
centra. As there are seven intervertebral joints and
eight haemal arches, there is one in excess haemal
arch (possibly one of the two below the centrum
of caudal 14, i.e. haemal arches 7 and 8), which
can be explained as an artefact of preparation.
This should be kept in mind when reading the fol-
lowing description. Haemal arch 4 appears to be
complete and is the longest among the completely
preserved haemal arches, which gradually reduce in
length along the tail. It is rod-like with a straight
spine that is slightly and symmetrically expanded
at its distal extremity. Haemal arches 5 and 6 are
like haemal arch 4, but shorter. Haemal arch 6 is
slightly shifted anteriorly relative to its anatomical
position and the distal extremities of haemal arches
4-6 contact each other. The spine of haemal arch 7
is slightly recurved backward and is symmetrically
expanded distally resembling a spoon, whereas the
distal end of haemal arches 8 and 9 is asymmetri-
cal expanded posteriorly (boot-like). The following
haemal arches 10-13 have rounded and symmetri-
cally expanded extremities, therefore they are very
unlike the boot-like haemal arches of the holotype
(Fig. 6C-F).
Ossied Tendons
The ossied tendons are wire-like and oc-
cur epaxially in the proximal and mid-tail, as well
as in the dorsum and sacrum, on both sides of
the vertebral column. The posteriormost ossied
tendon ends on the anterior margin of the neural
spine of caudal 20 at mid-spine; it is located at
the beginning of the dorsoventral lowering of the
tail to a whip-like aspect and in correspondence
of the passage from boot-shaped to bilobate chev-
rons (Fig. 4C). Ossied tendons cross diagonally
the neural spines in caudal vertebrae 1 to 6, ar-
ranged in an irregular rhomboidal cross-lattice pat-
tern (Fig. 5; SI, Fig. 4). The thin ossied tendons
have probably been affected by the preparation by
acid, partly splitting away during preparation; the
preserved sample is only a part of the original lat-
tice. This is suggested by the scars left by the miss-
ing tendons on the right lateral side of the neural
spines.
Three sets of ossied tendons can be identi-
ed on the right lateral side of the proximal neural
spines of caudal vertebrae 1 and 2, whereas only
two occur in the following caudals 3-6 (SI, Fig.
4). One set runs posterodorsally inclined of 17.5-
22° with respect to the axis of the tail (number
of measurements, N = 5; mean, M = 19°) over-
lapping a set running posteroventrally and slightly
less inclined (9°-16.5°; N = 3; M = 11.5°). These
are identied as the tendons of the M. semispinalis
of the M. transversospinalis system in other iguan-
odontoideans (Organ 2006); specically, the me-
dial set belongs to the M. articulospinalis, whereas
the lateral set to the M. tendinoarticularis. The third
set is the most medial of the three and is repre-
sented by a few and thinner tendons running pos-
terodorsally at a low angle (~10°); these tendons
can be identied only on the rst and second cau-
dal vertebrae (SI, Fig. 4) and are plausibly those
of the M. spinalis, which also belongs to the M.
transversospinalis system, and are medial to the other
ossied tendons (Organ 2006). Apparently, the M.
spinalis tendons are scarcely ossied or not ossied
at all in the holotype of T. insularis. From dorsals 7
to 14 on the right side of the tail, ossied tendons
are only posterodorsally directed. At the base of
the neural spines of caudals 8 to 13, just above
the zygapophyses, a bundle of at least four ossied
tendons runs posterodorsally nearly parallel to the
axis of the tail (SI, Fig. 3A). Two tendons running
posterodorsally across spines 7-9 (SI, Fig. 3A) are
also inclined at very low angles (ca. 2.6° and 3.2°).
Because of their development, these are plausibly
the lateral tendons of the M. semispinalis (i.e., those
of the M. tendinorticularis; cf. Organ 2006).
The two tendons spanning vertebrae 15-
18 run posteroventrally (Fig. 4C; SI, Fig. 1), thus
they plausibly are the medial tendons of the M.
semispinalis. One of these two tendons crosses the
middle of the spines, whereas the other runs in a
more apical position and possibly originates from
the posterodorsal corner of neural spine 15. They
are inclined 8° and 12.5°, respectively. A detached
short tendon segment occurs near neural spine 10;
another one crops out from the left side of neural
spine 18 (Fig. 4A and C; SI, Fig. 1B). The last pre-
served tendon runs at the level of the lower part
of the neural spine at least along vertebrae 18-20
(Fig. 4C). It slopes posteroventrally, although it is
nearly parallel to the axis of the tail, thus it is a me-

Fabio Marco Dalla Vecchia
594
dial tendon of the M. semispinalis. Ossied tendons
run posteroventrally at low angle also along the left
side and can be seen in the gaps between the neural
spines. The irregular arrangement of the ossied
tendons was probably caused by muscular decay, as
suggested by the presence of detached elements.
Two bundles of long ossied tendons were
visible in the proximal part of the tail of the un-
prepared SC 57247A up to caudal 9 (Fig. 2). One
bundle ran along the dorsal half of the neural
spines of the posterior sacrals and rst caudals (up
to about caudal 3). The apparent direction of the
tendons was posteroventral at very low angle. The
other bundle ran along the middle of the neural
spines starting about at caudal 2 and reaching cau-
dal 8. Only a single tendon crossed the neural spine
of caudal 9. The apparent direction of the tendons
of this second bundle was initially posterodorsal at
low angle, then tendons curved to be nearly parallel
to the axis of the tail and nally they were directed
posteroventrally at low angle. One single tendon
crossed neural spine 4 posterodorsally at high an-
gle, possibly inserting to the posterior apical corner
of neural spine 5 (Fig. 2). After preparation, the
pattern is the same, although some missing seg-
ments of the tendons have probably been recon-
structed. The two bundles meet in correspondence
of caudals 2-3. At least six ossied tendons can
be counted on the lateral surface of neural spine
3. The most posterior ossied tendon occurs on
spine 10 and is single and thinner than the proxi-
mal ones. The longest tendon spans vertebrae 2 to
8, but portions may have been reconstructed. One
tendon apparently inserts on the anterior side of
the base of neural spine 9. The other insertions are
unclear due to the overlapping of the tendons, but
they are apparently located at mid-spine. Therefore,
ossied tendons do not form a regular rhomboidal
lattice in SC 57247; they are more numerous proxi-
mally (mainly on the sacrum and rst caudals) and
decreases in size and diameter moving posteriorly.
Orders of overlapping tendons cannot be reliably
identied.
Distal to caudal 10, i.e., posterior to the ex-
ure of the tail, no ossied tendons can be observed
in the prepared specimen; this could be a preser-
vational bias or a consequence of the preparation
by acid, which makes unstable and very fragile thin
bones as the ossied tendons are, leading to their
destruction.
eStImate of the total taIl length
of the holotype of TeThyshadros
insularis
According to Hone (2012), caudal vertebrae
count in ornithischian dinosaurs is probably less
variable than in the saurischians and it appears to
be generally consistent within hadrosaurs. Hadro-
sauroids are considered to have 50 to 70 tail verte-
brae (Norman 2004; Horner et al. 2004). However,
very few specimens preserve the complete distal
portion of the caudal segment of the vertebral
column (Hone 2012). Therefore, the lower count
of 50 vertebrae is probably taphonomically biased
or underestimated. For example, the specimen
IRSNB 1551 of Mantellisaurus athereldensis (see
Norman 2015) preserves only 33 caudal vertebrae
and its tail was reported as “almost complete” by
Norman (1986: 308). However, the last preserved
vertebra bears a chevron (Norman 1986: g. 2),
thus it is a mid-caudal and at least the entire distal
segment of the tail is missing in that specimen.
Specimen IRSNB 1551 preserves 15 proximal
caudals and only 18 mid-caudals (Norman 1986).
Therefore, also some mid-caudal vertebrae are
missing.
The most complete tail of the basal iguan-
odontian Tenontosaurus tilletti, which preserves 60
vertebrae but lacks the last distal elements, has
eight proximal caudals, 22-32 mid-caudals and 28-
38 distal caudals (uncertainty in the count of mid
and distal caudals is due to the uncertainty on the
position of the last chevron; Forster 1990). One
specimen of the hadrosaurid Edmontosaurus regalis
(ROM 5851) from the lower part of the Horse-
shoe Canyon Formation (uppermost Campanian)
near Drumheller (Alberta, Canada) was reported
to have 76 tail vertebrae (Lull & Wright 1942), but
unfortunately it was never described. The speci-
men TMP 1998.058.0001 (Fig. 7), belonging to a
small indeterminate hadrosaurid from the Dino-
saur Provincial Park (Alberta, Canada) (Brinkman
2014; Persons & Currie 2014) preserves one of
the most complete hadrosauroid tails ever discov-
ered. It comes from the Dinosaur Park Formation
(upper Campanian, underlying the Horseshoe
Canyon Formation), where Edmontosaurus is not
reported (Campione & Evans 2011), therefore it
plausibly belongs to a distinct genus. The articu-
lated tail of TMP 1998.058.0001 was broken into

The tail of Tethyshadros insularis 595
two segments during eld work and about ten of
the rst mid-caudal vertebrae are damaged (Fig.
7). This tail is composed of 77 vertebrae (pers.
obs., 2011). There are a caudosacral vertebra (with
pleurapophyses contacting the postacetabular
process of the ilium), at least 14 proximal caudal
vertebrae (vertebrae at the mid-proximal transi-
tion are damaged and unprepared) and over 25
mid-caudals. The rst chevron is located between
caudals 3 and 4. This specimen is used here as a
reference to estimate the length of the missing
portion of the tail of SC 57021, under the as-
sumption that both had originally the same total
count of caudal vertebrae.
The ratios between the length of each
caudal centrum and the length of the preceding
centrum in caudals 9 to 33 were calculated in SC
57021 and range 0.86-1.07 (SI, Tab. 1). The mean
ratio from caudals 9 to 33 (0.98; it means that the
centrum of vertebra x+1 is 98% the length of the
centrum of the preceding vertebra x) is similar
to the mean of the last 12 vertebrae (vertebrae
22-33; 0.9775), thus there is no evident increase
in shortening in the distal part of the segment.
I used this ratio (0.98) to estimate the length of
each of the missing centra 34-77 (SI, Tab. 2, es-
timate A). The sum of these lengths is 1009.94
mm. As the spaces left by the intervertebral discs
are short in the articulated segment, this sum well
approximates the total length of the missing seg-
ment, but is anyway a minimum estimated length
that is somewhat lower than the actual length.
The ratio between the length of each caudal
centrum and the length of the preceding one in
the tail of TMP 1998.058.0001 was also calculat-
ed (SI, Tab. 3). The mean of this ratio calculated
in caudal vertebrae 25 to 77 is 0.99, but the ratio
ranges 0.79-1.12 and an increase in centra short-
ening is evident in the distal part of the tail as the
mean ratio of the last 11 preserved centra is 0.92
(SI, Tab. 3).
I used the ratios of vertebrae 34-77 of TMP
1998.058.0001 to estimate the length of each of
missing centra 34-77 of SC 57021 (SI, Tab. 2,
Fig. 7 - The complete tail of the
indeterminate hadrosaurid
TMP 1998.058.0001, which
was divided into an anteri-
or and a posterior segment
when excavated. A) Ante-
rior segment; B) posterior
segment. Abbreviations: csv,
caudosacral vertebra; cv1,
rst caudal vertebra; hm1,
rst haemal arch. The scale
bars equal 100 mm.

Fabio Marco Dalla Vecchia
596
estimate B). Since the lengths of centra 68, 72
and 76 of TMP 1998.058.0001 are unknown, the
mean ratio of the last ten measurable centra of
TMP 1998.058.0001 (0.93) was used to calculate
the missing ratios. The sum of those lengths is
885.78 mm (nearly 125 mm less than the estimate
based on the mean ratio of caudals 9 to 33 of SC
57021). This estimate of the missing portion of
the tail of SC 57021 is probably more accurate
than that based only on the ratios from the pre-
served part of the tail, because accounts for the
increase in shortening of the centra that probably
occurred in the terminal portion of the tail. When
added to the preserved portion (1650 mm; Dalla
Vecchia 2009c), the tail of the holotype of T. insu-
laris results to be 2535.78 mm long (1650+885.78
mm) and appears to be somewhat whip-like (Fig.
8). The total body length measured from the tip
of the snout to the distal end of the tail results to
be 4505.78 mm (the preserved part of the skel-
eton is 3620 mm long; Dalla Vecchia 2009c). The
tail accounts for 56% of the total body length and
is over 3.5 times the length of the trunk.
The tail/femur ratio would be 6.04, much
higher than the ratio reported by Hone (2012) for
T. insularis and much higher than that of all other
hadrosauroids in table 2 of Hone (2012), ranging
3.7-3.1 (included TMP 1998.058.0001; 3.7). This
rather high ratio is due also to the comparatively
shorter femur of SC 57021, as suggested by the
low femur/humerus ratio (1.46, whereas this ratio
ranges 1.66-2.10, with mean 1.90, in a sample of
15 North American hadrosaurid specimens; Lull
& Wright 1942).
caudal vertebrae of other lateSt
cretaceouS hadroSauroIdS from the
european archIpelago
During latest Cretaceous times, most of pres-
ent-day central and southern Europe was an archi-
pelago of islands in the Tethys Ocean between the
Afro-Arabian continent to south, the North Euro-
pean landmass to north and the Asian continent to
east (Philip et al. 2000; Csiki-Sava et al. 2015). Ac-
cording to the palaeogeographic reconstruction by
Philip et al. (2000; Fig. 9), the Adriatic Island where
T. insularis lived was much smaller than the Ibero-
Armorican Island, which was the largest island of
the archipelago and corresponds to the present day
Iberian Peninsula plus most of France. The Adriat-
ic Island was possibly slightly larger than the closer
Tisia-Dacia Island (corresponding to the present
day north-western part of Romania).
No complete and articulated hadrosauroid
skeletons comparable to those from the Villaggio
del Pescatore site have yet been discovered in the
uppermost Cretaceous of Europe. Anyway, cau-
dal vertebrae of hadrosauroid dinosaurs allowing
comparison with the T. insularis material are report-
ed from the uppermost Cretaceous of Romania,
Spain, France, Belgium and the Netherlands, and
Germany. Most of this material was still undiscov-
ered, unpublished or had been summarily described
in the literature when T. insularis was published in
2009.
Romania
Several caudal vertebrae from the Densuş-
Fig. 8 - Reconstruction of the skeleton of the holotype of Tethyshadros insularis with the silhouette of the portion of the tail that is probably
missing in the fossil. The proximal caudals and their haemapophyses are evidenced in red; the ischium is yellow; the femur is green;
and the tibia is pale blue. Artwork by M. Auditore (modied). The scale bar equals 500 mm.

The tail of Tethyshadros insularis 597
Ciula and Sâmpetru Formations of the Haţeg
Basin, Transylvania, have been referred to the
basal hadrosauroid Telmatosaurus transsylvanicus (see
Nopcsa 1915, 1928; Dalla Vecchia 2014; pers. obs.).
All of the hadrosauroid material from the upper-
most Cretaceous of Transylvania has been tradi-
tionally referred to Telmatosaurus transsylvanicus (e.g.,
Therrien 2005; Benton et al. 2010). However, the
fossil-bearing units span the whole Maastrichtian
and possibly part of the upper Campanian (Csiki-
Sava et al. 2016); furthermore, most of the material
referred to Telmatosaurus transsylvanicus do not show
any diagnostic features of this taxon (see Dalla
Vecchia 2006, 2009a). Dalla Vecchia (2009a) sug-
gested referring the not-typical hadrosauroid mate-
rial from the uppermost Cretaceous of Transylva-
nia to Telmatosaurus transsylvanicus with reserve; this
suggestion was later followed by Prieto-Márquez
(2010) and Csiki-Sava et al. (2016). The only nearly
complete caudal vertebra to have been be referred
to Telmatosaurus transsylvanicus in the literature is a
proximal caudal (NHMUK PV R. 4915; Fig. 10A)
from the Vălioara locality of the Densuş-Ciula
Formation (Nopcsa 1928; Dalla Vecchia 2009b).
The type material of Telmatosaurus transsylvanicus is
from Sacel locality of the Sâmpetru Formation and
does not include caudal vertebrae (Nopcsa 1900;
Dalla Vecchia 2006, 2009b), thus NHMUK PV R.
4915 cannot be a priori referred to Telmatosaurus
transsylvanicus. Like the caudals of T. insularis, the
amphicoelous centrum of NHMUK PV R. 4915 is
longer than high (l/h at mid-centrum is ca. 1.35).
It has anterior and posterior processes for the
haemapophyses, and the posterior processes are
more prominent than the anterior. There are pleur-
apophyses, although they are broken. These fea-
tures indicate that NHMUK PV R. 4915 is a proxi-
mal element, but posterior to the vertebra bearing
the rst chevron. The zygapophyseal articular fac-
ets are nearly vertical. The neural spine is laterally
compressed, slightly recurved and nearly upright.
The neural spine is incipiently meat cleaver-shaped,
with a ‘neck’ at the base and the upper portion that
is moderately expanded anteroposteriorly, but not
much expanded apically as those of the rst proxi-
mal caudals of T. insularis. The upper 4/5 of the
spine has rough posterolateral and apical surfaces.
The ‘neck’ is basoapically longer than in the meat
cleaver-shaped neural spines of T. insularis; the neu-
ral arch is comparatively as tall as that of caudal 4
of T. insularis holotype but the neural spine is pro-
portionally lower (the heights of the neural arch
and neural spine are 2.67 and 1.68 times the cen-
trum height, respectively).
A second caudal vertebra, NHMUK PV
R.4973 (Fig. 10B), from the Sâmpetru Formation
near Sâmpetru, lacks most of the neural spine.
It was gured in Nopcsa (1928: pl. 5, g. 5) with
Fig. 9 - Location of the latest Cre-
taceous localities that yiel-
ded the caudal vertebrae
of hadrosauroid dinosaurs
mentioned in the text. The
late Maastrichtian palaeoge-
ographic map of the Euro-
pean archipelago is based on
Philip et al. (2000), modied.
1) Villaggio del Pescatore
(Italy); 2) Haţeg Basin (Ro-
mania); 3) the Tremp Syn-
cline localities (Spain); 4) the
Southern France localities;
5) the Limburg localities (the
Netherlands and Belgium);
and 6) Bad Adelholzen lo-
cality (Germany). White in-
dicates emergent land; pale
blue, shallow sea; pale blue
with brick fabric, carbonate
platforms; lilac, the Chalk
Sea; blue, deep sea; dark
blue, oceanic basins (oceanic
crust). Present day coastlines
of Europe, northern Africa
and the Arabian Peninsula
are reported as reference.

Fabio Marco Dalla Vecchia
598
the neural spine reconstructed on the basis of the
neural spine of NHMUK PV R. 4915. This second
vertebra is similar to NHMUK PV R. 4915, with a
‘necked’ basal part of the neural spine and asym-
metrically developed anterior and posterior process-
es for the chevrons on the centrum, but it is larger,
less elongated (l/h at mid-centrum is ca. 1.20) and
the neural spine is more inclined backward. Both
pleurapophyses lack their distal ends; the preserved
part is anteroposteriorly narrow, dorsoventrally at
and slightly dorsoventrally arched (Fig. 10C).
In the sample of caudal centra referred to
Telmatosaurus transsylvanicus at the NHMUK and at
the Muzeul Civilizaţiei Dacice şi Romane of Deva
(Romania), there are neither mid-caudal centra as
elongated as centra 23 to 33 of T. insularis nor cen-
tra with the shape of amphicoelous hemicylinders
(pers. obs.).
Spain
The oldest hadrosaur remains from the Con-
ca Dellà area, in the Tremp Syncline of NE Spain,
come from the Els Nerets site of the La Posa For-
mation, which is referred to the magnetochron
C31r and dated to the early Maastrichtian (Conti et
al. 2020). Ten hadrosaurid caudal vertebrae are re-
ported from this site (Conti et al. 2020: 3) and three
of them are gured (Conti et al. 2020: g. 5I-P).
These remains were referred to an unnamed lam-
beosaurine hadrosaurid showing afnity with the
Tsintaosaurini, although the phylogenetic analysis
produced to test the afnities of this unnamed tax-
on does not nd support for a clade Tsintaosaurini
(Conti et al. 2020: g. 9).
The three gured specimens are nearly com-
plete proximal caudal vertebrae (MCD-8638, IPS-
NE-13 and MCD-6690), whereas the other seven
Fig. 10 - Proximal caudal vertebrae from the uppermost Cretaceous of Transylvania (Romania). A) NHMUK PV R. 4915, right lateral view;
B) NHMUK PV R.4973, right lateral view; and C) NHMUK PV R.4973, right anterodorsal view. Abbreviations: hmaf, articular facet
for the haemal arch; ne, ‘neck’; ns, neural spine; pl, pleurapophysis; poz, postzygapophysis; prz, prezygapophysis. The scale bars equal
10 mm.

The tail of Tethyshadros insularis 599
are fragmentary specimens (MCD-61 to 63, MCD-
65 and 66, MCD-5209 and MCD-7095), which
are neither gured nor described. Although IPS-
NE-13 and MCD-6690 are reported as mid-caudals
by Conti et al. (2020), they preserve traces of the
pleurapophyses and hence are proximal caudals.
The three gured specimens have similar small sizes
(centra are 40, 36 and 40 mm long in MCD-8638,
IPS-NE-13 and MCD-6690, respectively). Accord-
ing to the measurements reported in Conti et al.
(2020: tab. 1) and g. 5J, L and M, the centra of the
nearly complete vertebrae are much taller than long,
especially in MCD-8638 and IPS-NE-13.
MCD-8638 is probably the rst caudal ver-
tebra because it has an anteroposteriorly short and
opisthocoelous centrum with sub-circular articular
facets (Conti et al. 2020: g. 5I), there is no refer-
ence to the presence of articular facets for the chev-
rons in its description and it presents a wing-like
proximodorsal part of the anteroposteriorly com-
pressed pleurapophyses (see vertebra MDE-Les1-5
from Lestaillats 1 locality of France below). The
neural spine is twice as tall as the centrum, ellip-
tical in cross-section and “thicker proximally than
distally” (Conti et al. 2020: 7). The neural spine is
upright, but it is backward recurved; in lateral view,
it enlarges slightly in the basal segment, maintains a
moderate anteroposterior width for most of its ba-
soapical length and tapers slightly apically. The apex
of the neural spine is blunt but not club-shaped (the
spine is laterally at and is not thickened apically).
IPS-NE-13 (bearing also the acronym IPS-
845; Fig. 11) is proportionally more elongated than
MCD-8638 and has an amphicoelous centrum with
hexagonal articular facets and facets for the chev-
rons (Conti et al. 2020: 7). Therefore, it is from a
more distal position in the tail. Conti et al. (2020:
7) retains that the caudal ribs were not fused to the
transverse processes in this vertebra, but what they
interpret as “square facets for attachment of the
transverse processes” are probably the worn-out
proximal part of the broken pleurapophyses. The
latter have sub-rectangular cross sections that are
craniocaudally longer than dorsoventrally high. The
pleurapophyses of this vertebra were more massive
than those of T. insularis. The tall neural spine of
IPS-NE-13, which lacks the apical portion, is slight-
ly recurved and anteroposteriorly wider than that
of MCD-8638 but is unlike the apically expanded
spines of T. insularis. This neural spine also lacks
the basal ‘neck’ of the rst caudals of T. insularis
and the Transylvanian taxon (Fig. 10A-B).
The deformed MCD-6690 has an amphicoe-
lous centrum like IPS-NE-13 (Conti et al. 2020: 7),
but it is proportionally more elongated. Further-
more, its neural spine is proportionally much taller
(over three times as tall as the centrum) and antero-
posteriorly narrower and it “expands and thickens
distally” (Conti et al. 2020: 7). Apparently, MCD-
6690 belongs to a different taxon with respect to the
other two proximal caudals from Els Nerets locality.
The type material of the lambeosaurine Para-
rhabdodon isonensis come from the Tossal de la Doba
locality (reported as Sant Romà d’Abella in the lit-
erature) of the Conca Dellà zone (Dalla Vecchia
et al. 2014). It is from the uppermost Maastrich-
tian Talarn Formation, which originated during the
magnetochron C29r (Fondevilla et al. 2019). The
sample includes a nearly complete proximal caudal
Fig. 11 - Proximal caudal vertebra IPS-NE-13 from the lower Maa-
strichtian Els Nerets locality of Conca Dellà, Spain. A) left
lateral view; B) posterior view. Abbreviations: hmaf, articu-
lar facet for the haemal arch; nc, neural canal; ns, neural spi-
ne; pl, pleurapophysis; poz, postzygapophysis; prz, prezyga-
pophysis. The scale bar equals 10 mm.

Fabio Marco Dalla Vecchia
600
vertebra (IPS-695-5, centrum height is 115 mm;
Fig. 12A-B), four proximal caudal centra and ve
mid-caudal centra (see Dalla Vecchia et al. 2014 for
the list of the museum numbers). These vertebrae
had been found associated and were referred to a
single individual (Casanovas-Cladellas et al. 1999),
but the provenance of some of the caudal centra
is dubious (Dalla Vecchia et al. 2014). Because of
its long pleurapophyses and the sub-circular articu-
lar facets of the centrum, IPS-695-5 is one of the
rst proximal caudals. However, it is not one of the
chevron-less most proximal caudals, having well-
Fig. 12 - Caudal vertebrae from the Maastrichtian Tossal de la Doba (A-B; Pararhabdodon isonensis) and Basturs Poble (C-H; Lambeosaurinae
indet., lateral views) localities of Conca Dellà, Spain. A-B) IPS-695-5, proximal caudal in right lateral and left posterodorsal view,
respectively; C) MCD-5117, proximal caudal centrum; D) MCD-4928, nearly complete neural arch of a proximal caudal vertebra; E)
MCD-4854, posterior proximal caudal; F) MCD-5112, one of the last proximal caudal vertebrae, right lateral view, but mirrored; G)
MCD-4862, mid-caudal vertebra; and H) MCD-4827, mid-caudal vertebra. Abbreviations: hmaf, articular facet for the haemal arch;
na, neural arch; ns, neural spine; pl, pleurapophysis; poz, postzygapophysis; prz, prezygapophysis. The scale bars equal 10 mm.

The tail of Tethyshadros insularis 601
developed posterior facets for the haemal arch. It is
probably the rst chevron-bearing vertebra of the
tail. Its centrum is much higher than long (l/h at
mid-centrum is ca. 0.63). The pleurapophyses have
an anteroposterior constriction in the middle and
a distal expansion (Fig. 12B) like those of caudals
9-11 of the T. insularis holotype, but they are com-
paratively longer, slender and dorsoventrally thicker
than those of T. insularis, and have a circular proxi-
mal cross section. Unlike the proximal caudals of T.
insularis, the preserved basal portion of the neural
spine is anteroposteriorly narrow in lateral view, has
parallel anterior and posterior margins and lacks a
basal ‘neck’ (Fig. 12B). The backward slope of the
neural spine is 40°.
Remains of at least 50 disarticulated and
scattered caudal vertebrae (including also isolated
centra and neural arches) belonging to several dis-
tinct individuals have been found in the Basturs
Poble bonebed of the Conca Dellà zone. Only a
few of them preserve the centrum associated with
the neural arch (pers. obs.). The Basturs Poble
bonebed occurs in the Conques Formation and is
stratigraphically lower than the close Tossal de la
Doba locality, but higher than the Els Nerets lo-
cality (Conti et al. 2020: g. 2); it is plausibly late
early Maastrichtian in age (magnetochron C31r;
Fondevilla et al. 2018). The hadrosauroid remains
from this bonebed have been tentatively referred
to Pararhabdodon isonensis by Fondevilla et al. (2018),
because of the stratigraphic range of this taxon,
which was reported also from the very close Ser-
rat del Rostiar-1 locality by Prieto-Márquez et al.
(2013; based on the maxilla MCD-4919). The Ser-
rat del Rostiar-1 bone-bearing horizon is approxi-
mately the lateral equivalent to the Basturs Poble
bonebed (see Fondevilla et al. 2018). However,
Prieto-Márquez et al. (2019: 33) have subsequently
referred MCD-4919 to “Tsintaosaurini indetermi-
nate” and have erected the new basal lambeosau-
rine Adynomosaurus arcanus from the same horizon
at the nearby Costa de les Solanes locality. Conse-
quently, the Basturs Poble bonebed sample should
now be referred to an indeterminate lambeosau-
rine. Most of the caudal vertebrae from the Basturs
Poble bonebed are small-sized, the centra ranging
25.3-65.5 mm in length, with 80% of the centra
that are less than 45 mm long. Evidence of im-
maturity is common within the sample: the neural
arches are unfused to the centra in some vertebrae
(e.g., MCD-4764, MCD-4832, MCD-4853, MCD-
4883a,c, MCD-4961, MCD-4975, MCD-4979 and
MCD-4997; probably this is also the case of most
of the isolated neural arches), whereas the neuro-
central suture is still partly visible in others (MCD-
4774, MCD-4824, MCD-4862, MCD-4890 and
MCD-4921).
The proximal centra are rather short and tall
(see for example, MCD-5117 with l/h at mid-cen-
trum < 0.74; Fig. 12C). In lateral view, the neural
spines of the proximal caudals are tall, ribbon-like,
slightly recurved and have prespinal and postspi-
nal laminae (Figs. 12D-E). They gradually broaden
apically, but the apical anteroposterior expansion is
moderate. The neural spines are laterally attened
and not club-shaped (i.e., they are not much trans-
versely thickened apically) and there is no ‘neck’ at
their bases. The centrum of the posterior proximal
vertebra MCD-4854 (Fig. 12E) is higher than long
(l/h is 0.93) and bears anterior and posterior ar-
ticular facets for the pedicels of the haemal arches;
the posterior facets are set on prominent processes.
The backward slope of the neural spine is about
45°. The pleurapophysis of the vertebra MCD-
5112 (Fig. 12F) is a thin longitudinal ridge, suggest-
ing that this is one of the last proximal caudals. Its
centrum is higher than long (l/h is ca. 0.90). The
anterior mid-caudals (Fig. 12D-E) have centra that
are moderately longer than high (l/h is 1.15 and
1.18 in MCD-4862 and MCD-4827, respectively).
Their neural spines are long and very slender and
their apical portions are slightly recurved. The
basal part of the neural spine has a sub-circular
cross-section, whereas the apical portion is slightly
expanded anteroposteriorly and attened laterally.
The anterior margin of the expanded portion is
thicker than the blade-like posterior margin. The
backward slope of the neural spine is about 60°.
These caudal vertebrae differ from those of T. in-
sularis in the minor elongation of their centra and
shape and size of their neural spines.
The few chevrons from the Basturs Poble
bonebed are mostly from the proximal part of the
tail and are fragmentary. The most complete haemal
arches (SI, Fig. 5A-D) are ribbon-like (nger-like;
Persons et al. 2014), with anteroposteriorly nar-
row spines that are attened laterally and extremely
long dorsoventrally. They are straight to moderate-
ly recurved in lateral view and there is no evidence
of a distal expansion. One ribbon-like chevron is

Fabio Marco Dalla Vecchia
602
recurved backward (SI, Fig. 5E-F), but its curva-
ture could be augmented by taphonomic factors;
its spine is unexpanded distally. This may be one
of the rst haemal arches, which are shorter, more
inclined backward and sometimes more recurved
than those immediately following along the tail in
hadrosaurids (cf. Paul 2010: gs at pp. 299, 304,
and 307-309). No boot-shaped or bilobate chev-
rons have been collected from the Basturs Poble
bonebed.
The Costa de les Solanes locality has yielded
the holotype and referred material of Adynomo-
saurus arcanus, including three fragmentary caudal
vertebrae (Prieto-Márquez et al. 2019). Specimens
have been identied by Prieto-Márquez et al. (2019)
as a proximal caudal centrum (MCD-7141; not g-
ured), a mid-caudal (MCD-7127, Prieto-Márquez et
al. 2019: g. 4I) and a distal caudal (MCD-7142,
Prieto-Márquez et al. 2019: g. 4G and H) verte-
bra. The proximal caudal centrum is described
as “craniocaudally compressed” (Prieto-Márquez
et al. 2019: 23). The mid-caudal vertebra is com-
posed of a centrum that is higher than long (l/h is
about 0.79, based on g. 4I) and an incomplete and
separated neural spine with the postzygapophyses.
It is unclear whether the neural spine is apically
complete or not. If it is complete, it is moderately
tall (twice as tall as the centrum), nearly straight,
laterally attened and moderately aring apically
in lateral view like the posterior proximal caudal
MCD-4854 from Basturs Poble site (Fig. 12E), al-
though the postzygapophyses appear to be much
larger. Because the rest of the neural arch is not
preserved, the slope of the spine is unknown. The
centrum of the purported distal caudal vertebra is
also higher than long (l/h is about 0.76, based on
g. 4H) and is larger than the centrum of MCD-
7127 according to the scales in gure 4 of Prieto-
Márquez et al. (2019). Presence or absence of the
facets for the articulation of the chevron cannot
be established because the vertebra is gured only
in left lateral view, this detail is not mentioned in
the description by Prieto-Márquez et al. (2019) and
the posteroventral portion of the centrum may
be damaged. The basal part of the neural spine is
sloping posteriorly at 22°. MCD-7142 is plausibly
a mid-caudal vertebra too. These caudal vertebrae
differ from those of T. insularis in the minor elon-
gation of their centra; also the shape of the neural
spine is unlike those of the neural spines of the
caudals of T. insularis.
Level 3 of the Blasi locality of the Tremp
Syncline near Arèn (Aragòn, NE Spain), base of the
Tremp Formation, late Maastrichtian in age (upper
part of magnetochron 30n; Pereda-Suberbiola et
al. 2009), yielded relatively complete remains of at
least 38 caudal vertebrae of hadrosaurids (Cruzado-
Caballero 2012, suppl. 9.4, tab. XIII; many inven-
tory numbers are repeated in this list) plus those
referred to the lambeosaurine Arenysaurus ardevoli.
According to Cruzado-Caballero (2012),
the type material of A. ardevoli includes 14 associ-
ated caudal vertebrae with chevrons (MPZ2006/20;
Cruzado-Caballero 2012: g. 4.90; Fig. 13, partim),
a pathological caudal vertebra (MPZ2004/480), 31
other isolated caudals and 13 isolated chevrons (see
Cruzado-Caballero 2012: 180-181 for the museum
inventory numbers), but the 31 isolated caudals are
referred to an indeterminate lambeosaurine in the
same work (Cruzado-Caballero 2012: suppl. 9.4,
tab. XIII). In the paratype material of A. ardevoli,
Pereda-Suberbiola et al. (2009) included only the
pathological vertebra, the 14 associated caudal ver-
tebrae with chevrons (MPZ2006/20), two other
caudal vertebrae (MPZ2008/272 and 313) and two
haemal arches (MPZ2008/314 and 330). All these
A. ardevoli specimens were supposed to belong to a
single individual (Pereda-Suberbiola et al. 2009). The
vertebral association MPZ2006/20 includes the last
four proximal caudals and the rst nine or ten mid-
caudals, if their relationship is the original one and
not an artefact of restoration. The pleurapophysis
of the pathologic vertebra MPZ2004/480 is solely a
knob (“It has a small transverse process in the pos-
terodorsal part”, Canudo et al. 2005: 11), therefore
the vertebra is one of the last proximal caudals. The
centra of the other proximal vertebrae are much
higher than long (l/h ranges 0.52-0.55; Cruzado-
Caballero 2012: suppl. 9.4, tab. XIII), bear facets for
the haemal arches and their articular facets are hex-
agonal and at (Cruzado-Caballero 2012). In lateral
view, the neural spines of these proximal caudals are
ribbon-like, tall, nearly straight, and slightly inclined
backward (Cruzado-Caballero 2012: g. 4.91). All
of the pleurapophyses appear to be broken. The
mid-caudal vertebrae of MPZ2006/20 possess cen-
tra that are higher than long (l/h at mid-centrum is
0.94 in BLA3/10; Fig. 13) and their neural spines
slope backward (this slope is 45° in BLA3/10). The
centra bear anterior and posterior articular facets for

The tail of Tethyshadros insularis 603
the chevrons borne by processes that are equally de-
veloped. The postzygapophyses are posterior to the
posterior margin of the centrum. In lateral view, the
neural spines are recurved and gradually are api-
cally. They are transversely thick apically and some-
what club-shaped. The cross-section of the basal
part of the spine is sub-circular. This is evident in
the pathological vertebra MPZ2004/480, which is
morphologically like the mid-caudals although it has
a hint of a pleurapophysis.
All of the 12 chevrons referred to A. arde-
voli by Cruzado-Caballero (2012: 203-204) have
long, ribbon-like haemal spines that are laterally
attened, have an elliptical cross-section and lack a
distal expansion and anteroposterior processes. The
only gured haemal arch (BLA3/327; Cruzado-Ca-
ballero 2012: g. 4.97) has a recurved haemal spine
and is unexpanded distally, resembling the chevron
MCD-4880 from the Basturs Poble bonebed (SI,
Fig. 5E-F).
The caudal vertebrae referred to A. ardevoli
differ from those of T. insularis in the minor elon-
gation of their centra, shape and size of the neural
spines and shape of the chevrons. The mid-caudal
vertebrae differ also from those from the Basturs
Poble bonebed in the higher curvature, higher ro-
bustness and thicker apical part of the neural spines
(compare Figs. 12E-H and 13; apically thick and
club-shaped neural spines have not been found in
the Basturs Poble bonebed; per. obs.). This sup-
ports that the specimens from Basturs Poble bone-
bed belong to a lambeosaurine distinct from A. ar-
devoli (see Fondevilla et al. 2018), in agreement with
the different ages of the two sites.
A sample of 11 partially preserved and iso-
lated proximal and mid-caudal vertebrae from Blasi
3 are summarily described by Cruzado-Caballero
(2012; see p. 103 for the inventory numbers) and
referred to an indeterminate hadrosaurid. Unfortu-
nately, only a few of them have been gured. The
centra of the proximal elements are higher than
long (Cruzado-Caballero 2012: 103; suppl. 9.4,
tab. XIII). The only nearly complete specimen (the
proximal caudal BLA3/212d, Cruzado-Caballero
2012: g. 4.13A-B) has l/h = 0.59. Its damaged
pleurapophyses are long, dorsoventrally attened
but thick proximally and slightly arched dorsally. In
lateral view, the neural spine is straight with anterior
and posterior margins that slightly diverge apically,
and has a posterior slope at 33°. There is no ‘neck’
at the base of the neural spine and no transverse
thickening apically. Unfortunately, ventral and dor-
sal views of this vertebra have not been published.
BLA3/50 (Cruzado-Caballero 2012: g. 4.13C-D;
Fig. 13 - MPZ2006/20, 14 associated caudal vertebrae of Arenysaurus ardevoli, base of the Tremp Formation, upper Maastrichtian, Blasi 3 level,
Spain. The posterior segment with the rst seven mid-caudal vertebrae. The scale bar equals 50 mm.

Fabio Marco Dalla Vecchia
604
it bears the same number as one of the caudals
referred to Arenysaurus ardevoli) is a mid-caudal
vertebra, as indicated by the absence of the pleur-
apophyses and presence of posterior facets for the
chevron. The damaged centrum is higher than long
(Cruzado-Caballero 2012: suppl. 9.4, tab. XIII). The
segment of the neural arch between the pedicels and
the postzygapophyses is long and nearly horizontal,
so that the postzygapophyses are located well beyond
the posterior end of the centrum, like in the mid-
caudal MCD-4854 from Basturs Poble locality (Fig.
12E).
Ten fragmentary and isolated chevrons are
reported from the Blasi 3 hadrosaurid sample (see
Cruzado-Caballero 2012: 99 for the inventory num-
bers). According to Cruzado-Caballero (2012: 107),
they have laterally attened haemal spines. The only
gured chevron of this sample (BLA3/327) is re-
ferred to Hadrosauridae indet. in Cruzado-Caballero
(2012: g. 4.17) but is referred to A. ardevoli in g.
4.97 of the same work. Comparison between these
chevrons and those of T. insularis is impossible, but it
is unlikely that the Spanish specimens are like those
of the Italian taxon, otherwise it would have noticed
by Cruzado-Caballero (2012).
A sample of 26 among the isolated hadrosau-
rid caudal vertebrae from Blasi 3 listed in Cruzado-
Caballero (2012: suppl. 9.4, tab. XIII) includes ve
proximal caudals, at least 13 mid-caudals and at least
three distal caudals. Centra range 63-40.5 mm in
length and are all higher than long (l/h ranges 0.48-
0.99), except the centrum of the mid-caudal vertebra
MPZ2008/310 (l/h = 1.11) and that of the distal cau-
dal vertebra BLA3/231 (l/h = 1.15) which is referred
to an indeterminate lambeosaurine. Two other isolat-
ed vertebrae, which are reported as proximal caudals
of an indeterminate lambeosaurine by Cruzado-Ca-
ballero (2012; MPZ2008/275 and MPZ2008/276),
have facets for the haemal arches, thus they are not
the rst vertebrae of the tail; their centra are much
higher than long (l/h is 0.58 and 0.55, respectively;
Cruzado-Caballero 2012: suppl. 9.4, tab. XIII).
All of these caudals of indeterminate hadro-
saurids from Blasi 3 differ from those of T. insularis
in the minor elongation of the centra, shape and size
of the neural spines and shape of the pleurapophy-
ses, when preserved.
France
The upper Maastrichtian dinosaur-bearing
sites of southern France have yielded several isolated
caudal vertebrae of hadrosaurids (Laurent 2003; Dal-
la Vecchia 2014), but only those from the Lestaillats
1 locality (Plagne Anticline, Haute Garonne, Petites
Pyrénées) are adequately preserved for comparison.
This site, which is located in the early late Maastrich-
tian Marnes de Lestaillats (Laurent 2003: g.7), yielded
three complete proximal caudals (MDE-Les1-4, 5 and
20). They were referred to indeterminate “hadrosau-
rids” by Laurent (2003). Their centra are higher than
long with nearly at articular facets and bear long
pleurapophyses; their neural spines are tall and rib-
bon- shaped in lateral view (Laurent 2003; Fig. 14).
These caudals differ in details, suggesting that they are
from different positions of the proximal part of the
tail.
The centrum of MDE-Les1-5 (Fig. 14A-B) is
rather anteroposteriorly short (l/h = 0.63), has sub-
circular articular facets, and no facets for the haemal
arch. The pleurapophyses are directed laterally, taper
distally and have a robust, wing-like dorsoproximal
lamina like the vertebra MCD-8638 from the Spanish
Els Nerets locality. Similar pleurapophyses occur also
in a vertebra of the holotype of Hypacrosaurus altispi-
nus that Brown (1913b: gs. 1 and 3) identied as the
rst caudal. Therefore, MDE-Les1-5 is probably the
rst caudal, as noticed by Laurent (2003). Apically, the
neural spine (which is deformed and was restored) is
slightly recurved, does not are anteroposteriorly and
is not much thickened transversely; it slopes posterior-
ly at 18.5° and its lateral sides show basoapical striae.
The centrum of MDE-Les1-4 (Fig. 14C-E)
has also sub-circular articular facets, but it presents
posteroventral processes bearing the articular facets
for the haemal arch and no anterior processes and
articular facets. The centrum is markedly higher than
long (l/h = 0.75). The pleurapophyses are rod-like,
anteroposteriorly narrow (nger-like), slightly at-
tened dorsoventrally and directed lateroposteriorly
and lateroventrally. The pleurapophyses do not taper,
do not expand distally and do not have a constriction
in the middle. The neural spine is straight, ribbon-like
with parallel anterior and posterior margins and does
not expand apically in lateral view; its posterior slope
is 18°. The neural spine is laterally attened and not
club-shaped (it is only slightly transversely thickened
apically). A prespinal lamina is developed in the anteri-
or basal half of the spine. There is a hint of a ‘neck’ in
the posterior basal portion of the spine just above the
postzygapophysis, but it is much less marked than in

The tail of Tethyshadros insularis 605
T. insularis and in the two proximal caudals from Tran-
sylvania (NHMUK PV R. 4915 and NHMUK PV
R.4973). The height of the neural spine is about 2.2
times the centrum height. The shape of the articular
facets of the centrum, the length of the pleurapophy-
ses, the shape and low posterior slope of the neural
spine and the development of the articular structures
for the chevron suggest that MDE-Les1-4 is the rst
chevron-bearing caudal vertebra of the tail.
MDE-Les1-20 (Fig. 14F-H) has articular facets
of the centrum that are much higher than wide and
well-developed anterior and posterior articular struc-
tures for the chevrons. The centrum is higher than
long (l/h = 0.82). The pleurapophyses are like those
Fig. 14 - The caudal vertebrae of an indeterminate hadrosaurid (probably a lambeosaurine) from the upper Maastrichtian Lestaillats 1 locality,
France. A-B) MDE-Les1-5, rst caudal vertebra, anterior (A) and ventral (B) views; C-E) MDE-Les1-4, probably the rst chevron-
bearing caudal vertebra, right lateral (C), posterior (D) and ventral (E) views; F-H) MDE-Les1-20, from a middle position in the
proximal segment of the tail, left lateral (F), posterior (G) and ventral (H) views. Abbreviations: fo, nutritive foramen; hmaf, articular
facet for the haemal arch; pl, pleurapophysis; poz, postzygapophysis; prspl, prespinal lamina; prz, prezygapophysis; wlpl, wing-like
proximodorsal lamina of the pleurapophysis. The scale bars equal 50 mm.

Fabio Marco Dalla Vecchia
606
of MDE-Les1-4, but are directed laterally instead of
lateroventrally and are proportionally shorter. The
neural spine is more inclined posteriorly (38°) than
that of MDE-Les1-4. In its apical half, the neural
spine slightly ares anteroposteriorly and is recurved;
it is attened laterally, but its apical part is thickened
transversely. The height of the neural spine is about
2.5 times the centrum height. This caudal vertebra is
from a middle position in the proximal segment of
the tail.
The caudals from Lestaillats 1 are relatively
small-sized and the presence of the neurocentral su-
ture and large nutritive foramina on the ventral side
of the centra (Fig. 14B, E and H) suggests that they
belong to immature individuals.
The associated ischia (MDE-Les1-1, 2, and
19) have slender and sigmoid shafts that resemble
the shafts of the ischia of SC 57247 (Laurent 2003:
pl. 40). However, all of the Lestaillats 1 vertebrae are
rather different from those of T. insularis in the l/h
ratios of the centra, shape of the pleurapophyses,
relative size and morphology of the neural spines and
probably also in the position within the caudal verte-
bral column of the rst chevron-bearing vertebra (i.e.,
the rst chevron-bearing vertebra was much more
proximal in the French taxon than in the Italian one).
Furthermore, the articular facets of the zygapophyses
appear to be less inclined than those of T. insularis (i.e.,
the articular facets are not nearly vertical in the French
taxon; see Laurent 2003).
Belgium and the Netherlands
A few isolated caudal vertebrae of hadrosau-
rid dinosaurs come from the marine Maastricht For-
mation (“Chalk”; late Maastrichtian in age) of the
Limburg (Belgium and the Netherlands). Two caudal
vertebrae, which are deposited at the IRSNB, were
described and gured by Dollo (1883), although mea-
surements and localities of provenance are not given
in that paper. The neurocentral suture is still visible
in both vertebrae, indicating that they belong to im-
mature individuals (but see the comment by Buffetaut
2009: 3). One specimen is the right half of an amphi-
coelous centrum with the base of the right pedicel
of the neural arch and the proximal part of the right
pleurapophysis (Dollo 1883: gs. 1-2). The centrum
appears to be about as high as long or slightly higher
than long, if it is ventrally complete. It is a proximal
vertebra that Dollo (1883) considered as the rst or
the second caudal because it appeared to lack the ar-
ticular facets for the haemal arch.
The other specimen is a nearly complete proxi-
mal caudal (Fig. 15A). The amphicoelous centrum
has both anterior and posterior articular facets for the
haemal arches and is as long as high (l/h = 1.00). The
pleurapophyses project from the basal part of the
pedicels of the neural arch, which expand ventrally
on the lateral side of the centrum, as it is shown by
the neurocentral suture. The distally incomplete pleur-
apophysis is dorsoventrally attened (Dollo 1883: g.
6) and has an elliptical cross-section. The apically in-
complete neural spine is nearly straight, laterally at-
tened, inclined posteriorly at ca. 50° and ares apicad
in lateral view, but its extremity is not much thickened
transversely.
A partial proximal caudal (MB.R.4450; Fig.
15B) from the Dutch part of Limburg was described
and gured by Buffetaut (2009: g. 1). Its centrum is
incomplete, but it was clearly higher than long when
entire and its posterior articular facet is concave. The
Fig. 15 - The proximal caudal vertebrae of indeterminate hadrosau-
rids from the upper Maastrichtian of Limburg (Belgium and
the Netherlands). A) specimen stored at IRSNB (without
number) and gured by Dollo (1883: g. 5); B) MB.R.4450.
They are shown in right lateral view. Abbreviations: hmaf,
articular facet for the haemal arch; ns, neural spine; pl, pleu-
rapophysis; poz, postzygapophysis; prz, prezygapophysis;
su, neurocentral suture. A is from Dollo (1883), modied;
B is redrawn from Buffetaut (2009). Scale bar in B equals
50 mm. The size of the IRSNB vertebra was not reported
in Dollo (1883).

The tail of Tethyshadros insularis 607
only preserved facet for the haemal arch is posterior;
this facet is wide and deep, but not set on a project-
ing process as in the IRSNB vertebra. The broken
pleurapophysis has a sub-circular cross section. The
neurocentral suture is open and occurs dorsal to the
pleurapophysis (i.e., the pleurapophysis projects from
the centrum, unlike the IRSNB vertebra). The neural
spine is very inclined backward (the basal preserved
part slopes at 62°) and is slightly attened laterally,
with an elliptical cross-section and nearly parallel an-
terior and posterior margins. The postzygapophyses
are posterior to the posterior margin of the centrum
and their articular facets are inclined at about 45°.
The segment of the neural arch from the top of the
pedicels to the postzygapophyses is nearly horizontal.
The l/h ratio, morphology of the facet for the haemal
arch and the robust pleurapophysis would suggest it is
from a more proximal position than the IRSNB ver-
tebra, but the high posterior slope and slenderness of
the neural spine is in contrast with this interpretation.
Furthermore, the cross-section of the pleurapophysis
is different in the two vertebrae and the pleurapophy-
sis of MB.R.4450 is placed on the centrum instead of
the neural arch as in the IRSNB vertebra, although
the position of the pleurapophysis with respect to the
dorsal margin of the centrum is the same. This may
indicate that MB.R.4450 and the IRSNB vertebra be-
longed to two distinct taxa.
The three Limburg vertebrae are rather unlike
the proximal caudals of T. insularis in the l/h ratios of
the centra, shape of the pleurapophyses, and size and
morphology of the neural spines.
Germany
A small (ca. 25 mm long) and amphicoelous
vertebral centrum from the marine Gerhartsreiter
Schichten of southern Germany (Maastrichtian in
age; Wellnhofer 1994; López-Martínez et al. 2001)
was reported as a “sub-terminal caudal vertebra” of
an indeterminate hadrosaurid by Wellnhofer (1994:
228). It lacks haemapophyseal facets and belongs to a
distal caudal vertebra. Unlike the centra of T. insularis,
the German specimen is higher than long (l/h is ca.
0.80; Wellnhofer 1994: g. 5I-L). It is associated with
other bones from a single, small-sized hadrosaur, in-
cluding a neural arch that was identied as the “neural
arch of a proximal caudal vertebra” (Wellnhofer 1994:
229). However, the high position of the “transverse
process” (Wellnhofer 1994: 229) on the neural arch
suggests it belongs to a dorsal vertebra.
the lIfe and burIal envIronment of
TeThyshadros insularis
Like the other hadrosauroids from the upper-
most Cretaceous of Europe, T. insularis was an insu-
lar dweller (Fig. 9), but it lived in a peculiar environ-
ment. Unlike hadrosauroids from Romania, Spain,
France and probably also Belgium-the Netherlands
and Germany, T. insularis lived on the emergent part
of a shallow water carbonate platform, which was
a karst like that existing today in the same area (the
Carso/Kras Plateau of Italy and Slovenia, from
which the Anglo-Saxon term “karst” originates),
but under a tropical or sub-tropical climate, because
the region was at a latitude of about 30°N during
the latest Cretaceous (Philip et al. 2000). Thin baux-
ite levels found just below the Liburnian Formation
and coal seams in the lower half of this formation
(Jurkovšek et al. 1996; Venturini et al. 2008) may
indicate deposition under a relatively humid climate.
A modern analogous could be the tabular lime-
stone platform of the Yucatán Peninsula, which
is an emergent carbonate plateau within the Cen-
tral American tropical belt (Finch 1965). In karsts,
surface streams are rare because runoff is directed
underground through openings and fractures in the
limestone. Because the water ow is essentially un-
derground into carbonate dissolution cavities, sur-
face water supply is limited.
At the Villaggio del Pescatore site, dinosaur
remains were found in a ~10 m-thick body of black,
well-bedded and mostly thinly laminated limestone
(hereafter reported as laminites) (Tarlao et al. 1994;
Palci 2003; Dalla Vecchia 2008). The remains of
the other European hadrosauroids are usually pre-
served into terrigenous uvial deposits and palaeo-
soils or, less frequently, into sediments deposited in
open marine settings (Therrien 2005; Dalla Vecchia
2006, 2014). The body of laminites originated into
a depression within older shallow marine limestone
with karstied sides and bottom, and is surrounded
by breccia bodies (Tarlao et al. 1994; Palci 2003); the
main dimension of its cropping out portion is 80 m
(Palci 2003).
The collected dinosaur sample is represented
by remains of at least six distinct individuals, all re-
ferable to T. insularis and all with a modest body size
as the holotype (Dalla Vecchia 2009c). T. insularis
fossils are represented by isolated scattered bones
at the base of the laminites (most of these bones

Fabio Marco Dalla Vecchia
608
partially cropped out of the rock layer and were
not collected during the eld work), while articu-
lated skeletons occur in the middle-upper part of
the lithosome (Dalla Vecchia 2008, g. 1; pers. obs.).
The laminites have yielded also remains of
small crocodyliforms (including the holotype and
paratype of the new species Acynodon adriaticus; Del-
no et al. 2008); a single pterosaur bone (Dalla Vec-
chia 2018); at least 120 small shes (2-5 cm long;
pers.obs.), apparently monospecic; a few fragmen-
tary shes of larger size (less than 20 cm, anyway;
pers. obs.); several crustaceans (a few large shrimps
and many small shrimp-like crustaceans concen-
trated on a few bed surfaces; pers.obs.); some small
coprolites; and a few and poorly preserved plant
remains including a conifer branch (Dalla Vecchia
2008). According to Palci (2003), foraminifers are
represented by a few miliolids and rotaliids, which
are organisms that have been reported also from
brackish environments. Remains of freshwater
plants (characeans) also occur. The δ13C and δ18O
isotopic values measured from samples taken in the
laminites and in the surrounding breccias are always
negative, thus they do not support a marine origin
of the sediments (Palci 2003).
Circulation was restricted in the water-lled
depression, developing uctuating dysoxic to anox-
ic conditions allowing the development of an oligo-
typical (monospecic?) small sh community and
the presence of a few other organisms. The good
state of anatomical articulation of the two dinosaur
specimens considered in this paper suggests a very
limited transport after death and an undisturbed
deposition in a low energy protective environment.
Since lamination is sub-millimetric, the good state
of preservation of the specimens cannot be due to
rapid burial under a thick sediment mantle, but to
the peculiar conditions of the water at the bottom.
Although the depression has been consid-
ered basically of tectonic origin (Tarlao et al. 1994;
Palci 2003), an important role in its formation was
probably played also by the carbonate dissolu-
tion, because of the karst setting and by compari-
son with modern analogues. The depression could
have been a cenote or a similar karst structure, i.e.
a vertical-walled sink-hole like those characteristic
of the present day Yucatán plateau (Finch 1965)
and emergent parts of carbonate platform islands
in the tropical belt (the latter are often reported in
the English literature as blue holes). Cenotes can
be the main source of fresh water on the surface
during the dry season. Water supply is a problem
for large tetrapods living in a karst, mainly during
the dry season, because there is no supercial water
drainage as waters ow or stay in the ssures and
cavities produced by carbonate dissolution inside
the rock body. A cenote could also attract animals
for this reason. Cenotes are traps for terrestrial tet-
rapods because of their vertical walls that prevent
to go out once fallen into and because of the thick
vegetation surrounding and concealing them. Sub-
fossil remains of crocodilians, tortoises, birds and
a caviomorph rodent dated to 3900-2500 years BP
have been collected in the peat formed at bottom
of a blue hole with anoxic water in the Abaco island
of The Bahamas (Steadman et al. 2007).
The perfectly articulated skeleton of the ho-
lotype of T. insularis could be the remains of an
individual who had fallen, drowned and deposited
at the bottom of a cenote-like sink-hole developed
into the Adriatic carbonate platform. Preservation
was favoured by dysoxic to anoxic conditions of the
bottom waters due to their stagnation. The unusual
folding of the tail of SC 57247 was caused by the
slump fold of the unstable ne carbonate sediment
partially lling the sinkhole, covering the steep and
coarser talus deposits and sometimes mixing with
them. Evidences of synsedimentary faulting and
displacement of the sediment involving the skel-
etons occur also in the holotype (Fig. 1) and speci-
mens SC 57022 (Fig. 16C-D) and SC 57026.
dIScuSSIon
Gracile and robust morphs
The holotype and SC 57247 share the same
morphology of the neural spines in the caudals,
both possess proximal caudal centra that are longer
than high, their rst chevron is very distal in the tail
and their ischia are extremely long and slender. Fur-
thermore, they come from the same outcrop and
SC 57247 was preserved only a few metres strati-
graphically above the holotype (Dalla Vecchia 2008:
g. 1) in the same body of laminites, corresponding
to a few thousand years of deposition at maximum,
if Arbulla et al. (2006) are correct in their interpre-
tation of the cyclicity of the deposition. Therefore,
it is assumed that the holotype and SC 57247 belong
to the same species (Dalla Vecchia 2009c).

The tail of Tethyshadros insularis 609
SC 57247 is not signicantly larger than the
holotype, as it is evident from the similar lengths
of the skulls and proximal caudal vertebral centra,
but it is more robust (Dalla Vecchia 2009c: 1102).
The specimens SC 57247 and SC 57022 (two mani
and right radius and ulna; Fig. 16C-D) belong to a
robust morph, while the holotype belong to a grac-
ile morph (Dalla Vecchia 2009c). The presence of
associate gracile and robust individuals within a
sample of a dinosaur species is not uncommon: it
is reported in the basal sauropodomorphs Plateosau-
rus engelhardti (see Galton 1997) and Thecodontosaurus
antiquus (see Benton et al. 2000), the theropods Coe-
lophysis bauri (see Colbert 1990; Rinehart et al. 2001),
Syntarsus rhodesiensis (see Raath 1990) and Tyrannosau-
rus rex (see Carpenter 1990; Larson 2008), the basal
iguanodontian Zalmoxes robustus (see Weishampel et
al. 2003) and an indeterminate lambeosaurine (see
Fondevilla et al. 2018). The presence of gracile and
robust individuals within a sample of supposedly
adult members of a non-avian dinosaur species has
often been related to sexual dimorphism (Raath
1990; Carpenter 1990; Galton 1997; Larson 2008),
but without denitive evidence (Mallon 2017). Any-
way, size dimorphism is widespread throughout ex-
tant animal populations (Fairbairn et al. 2007). Most
of the extant animal species are sexually dimorphic
rather than monomorphic (Andersson 1994; Isles
2009), including birds (Lezana et al. 2000; Campos
et al. 2005; Isles 2009; Delgado Castro et al. 2013)
and crocodiles (Lang 1987; Allsteadt & Lang 1995;
Larson 2008; Platt et al. 2009; Isles 2009). There-
fore, the presence of sexual dimorphism in T. in-
sularis would not be unexpected. The specimens of
T. insularis lived in the same geographic spot and
can be considered as coeval for the standards of
the geological time, thus it is improbable that their
morphological differences may be due to geograph-
ic and temporal factors (Mallon 2017).
Different robustness may be indicative of
a different ontogenetic stage, but the robust SC
57247 shows more macro-evidences of osteologi-
cal immaturity than the gracile holotype. In fact, al-
though the sacral vertebrae of SC 57247 are fused
into a synsacrum and the sacral ribs are fused to
the transverse processes of the sacral vertebrae, the
ilia are unfused to the synsacrum. If not an artefact
of preparation, the rst caudal ribs were unfused to
their transverse processes. Also some skull bones
are evidently unfused each other in SC 57247. Only
a skeletochronological age determination of the
two specimens, which has not yet been attempted,
could shed light on this aspect.
The shape of the haemapophyses is unlike
in the two T. insularis specimens. If this is a real
feature and not an artefact of preparation, it is,
together with the different robustness of the centra
and possibly the shape of the pleurapophyses of
the rst seven caudal vertebrae, the most striking
difference between the two tails. Romer (1956: 267)
considered the variation in the position of the rst
haemapophysis in living turtles and crocodilians
as related to sex. Based on the presumed different
position of the rst haemapophysis in males and
Fig. 16 - Mani of Tethyshadros insula-
ris and their different robu-
stness. A) Right manus of
SC 57021 (holotype); B) left
manus of SC 57021 (holot-
ype); C) left manus of SC
57022; D) right manus of
SC 57022. Abbreviations:
II-1 to 3, phalanges 1 to 3 of
digit II; III-1 to 3, phalanges
1 to 3 of digit III; IV-1 to 2,
phalanges 1 to 2 of digit IV;
c, carpal; mcII-IV, metacar-
pals II-IV; ra, radius; ul, ulna.
The scale bars equal 50 mm.

Fabio Marco Dalla Vecchia
610
females of Alligator mississippiensis, Larson & Frey
(1992) and Larson (1994) had presumed the sex
of specimens of the theropod Tyrannosaurus rex.
Later, Brochu (2003) and Erickson et al. (2005)
have shown that the position of the rst haemal
arch is not related to sex in Alligator mississippiensis.
Persons et al. (2015) have considered as a sexual
feature the different shape and robustness of
the rst haemal arches of two specimens of the
oviraptorid theropod Khaan mckennai. However,
only the rst chevrons are differently robust in
the two specimens of K. mckennai, whereas all of
the preserved chevrons are different in the two
specimens of T. insularis. Furthermore, Persons et
al. (2015) related the purported sexual differences
in the haemal arches of K. mckennai to tail-feather
displays in courtship behavior. As the tail of T.
insularis was plausibly featherless, analogy between
the condition in T. insularis and K. mckennai is not
possible. However, this does not exclude that the
differences in the tails of the two specimens of T.
insularis are due to sexual dimorphism.
These morphological differences might also
be related to the possible condition of T. insularis as
a r-strategist species (see Pianka 1970 for a deni-
tion) in an insular setting, because insular r-strategist
tetrapods are supposed to show wide intraspecic
morphological variability (Raia et al. 2003: 304).
The T. insularis sample is still limited, but it
could be increased in the future because more spec-
imens can be extracted from the Villaggio del Pes-
catore site. Furthermore, the ontogenetic inuence
on the morphological variability will hopefully be
tested by the skeletochronological age determina-
tion of the already available specimens. This will
shed more light on the possible dimorphism of T.
insularis, hopefully allowing establishing whether it
is sexual or not.
Tethyshadros and the other hadrosauroids of
the European archipelago
Tethyshadros insularis is, with the taxon from
Romania, the only taxon in the sample of European
latest Cretaceous hadrosauroids to have proximal
caudal centra that are longer than high. Proximal
caudal centra are usually higher than long also in the
continental hadrosauroids (e.g., Parks 1920; Lull &
Wright 1942; Norman 1986, 2002, 2004; Godefroit
et al. 1998; Horner et al. 2004). The two caudal ver-
tebrae from Romania are the most similar to those
of T. insularis within this sample. This is in agree-
ment with the close phylogenetic relationship of T.
insularis and Telmatosaurus transsylvanicus found by the
phylogenetic analysis of Dalla Vecchia (2009c), Xing
et al. (2012, 2014) and Wang et al. (2015) and sup-
ports the belonging of those caudals to the Roma-
nian taxon. T. insularis and Telmatosaurus transsylvanicus
have possibly a similar age and lived in palaeogeo-
graphically close islands (see Dalla Vecchia 2009c
and Csiki-Sava et al. 2015; Fig. 9).
However, the caudals from Transylvania dif-
fer from the proximal caudals of the holotype of
T. insularis in the minor elongation of the centrum,
neural spines that are less expanded apically in lat-
eral view and pleurapophyses that are not tongue-
like or petaloid. The caudal vertebrae with upright
and meat cleaver-shaped neural spines in lateral view
are vertebrae 1 to 5 in T. insularis and they do not
have haemapophyseal processes, because chevrons
start from caudal 7 or 8; chevron-bearing proximal
caudals of T. insularis have fan-shaped spines with
an higher posterior slope than that of the only com-
plete Transylvanian caudal. This suggests that the
chevrons started in a more proximal position of the
tail in the Transylvanian taxon with respect to T. in-
sularis, as it is usual in hadrosauroids.
All of the hadrosauroid vertebrae from the
Maastrichtian localities of the Conca Dellà and Blasi
(Spain), France, Belgium and the Netherlands dif-
fer from T. insularis caudals also in the shape of the
neural spines and pleurapophyses. Furthermore,
the haemal arches from Spain are all of the simple
nger-like type. The morphology of the proximal
caudal vertebra from type material of P. isonensis
suggests that the rst chevrons was placed in the
rst caudals in this taxon, unlike T. insularis and like
all others hadrosauroids; this appears to be the case
also for the caudal vertebrae from France.
The peculiarity of T. insularis’ tail cannot be
explained as a consequence of insularity alone, be-
cause all of the hadrosauroids from the European
archipelago were insular dwellers.
Implications of the tail morphology of
Tethyshadros insularis
Unlike other hadrosauroid dinosaurs, the di-
agnosis of T. insularis includes seven apomorphic
features related to the caudal vertebral column
(Dalla Vecchia 2009c: 1100 and 1102). Comparison
with the hadrosauroid caudal vertebrae from the up-

The tail of Tethyshadros insularis 611
permost Cretaceous of Europe reported above sup-
ports the peculiarity of those features.
The caudal series of T. insularis was modied
from the relatively conservative hadrosauroid tail
bauplan for some particular functions. Theses func-
tions may be an important factor in the successful
adaptation of this dinosaur to an unusual mode of
life, as it has been hypothesized for non-avian the-
ropods (Ostrom 1969). Atypical traits may reect an
adaptation to the peculiar habitat of T. insularis in a
karst island.
Skeletal features of the tail are strictly con-
nected with tail musculature in living reptiles and are
supposed to be such also in extinct non-avian di-
nosaurs by inference (Romer 1923, 1927; Frey et al.
1989; Gatesy 1990; Arbour 2009; Mallison 2011; Per-
sons & Currie 2011a, 2014; Otero et al. 2012; Ibiri-
cu et al. 2014). However, the relationships between
tail skeleton and tail muscles and even the presence
of some tail muscles are taxonomically variable in
extant reptiles (e.g., Frey et al. 1989; Gatesy 1990;
Wilhite 2003; Schwarz-Wings et al. 2009; Arbour
2009; Mallison 2011; Persons & Currie 2011a; Otero
et al. 2010; Ibiricu et al. 2014) and the reconstruc-
tions of the tail musculature in non-avian dinosaurs
are consequently somewhat different to each other.
Anyway, hypotheses about the relationships between
the peculiar skeletal features in the tail of T. insularis
and their related soft tissues can be formulated only
by inference from extant reptiles and by inferences
made on other non-avian dinosaurs. Speculation on
the adaptive importance of such features is also pos-
sible by inference based on the functional morphol-
ogy and physiology of extant archosaurs and other
reptiles.
Elongation of the vertebral centra. As in the other
hadrosauroids, the centra become more elongated
moving distally along the tail in T. insularis. However,
starting from an condition of centra longer than
high already in the rst caudal vertebrae (SI, Tab. 1),
the mid-caudal vertebrae of the holotype of T. in-
sularis are comparatively much more elongated than
in other hadrosauroids. This peculiar elongation of
the caudal centra has the following consequences:
1) number of caudal vertebrae being equal, it makes
for a longer tail; 2) location of the rst haemapophy-
sis being equal, it contributes to shift posteriorly its
position; 3) number of proximal caudal vertebrae
being equal, it shifts posteriorly the position of the
last pleurapophyses. The elongation of the posterior
proximal and mid-caudal centra of the holotype of
T. insularis is comparable to that of the correspond-
ing caudal centra of Tenontosaurus tilletti (see Forster
1990). The long tail, which accounts for 67% of the
total body length, is considered a diagnostic feature
of this North American taxon (Forster 1990: 274;
Norman 2004), but its functional morphology has
never been investigated.
The elongation of the mid-caudal centra, as
well as, plausibly, of the distal centra, and the lower-
ing of the corresponding chevrons and neural spines,
made the distal half of the holotype tail whip-like,
with a marked shallowing just after the proximal seg-
ment (Fig. 8).
The tail of the robust individual (SC 57247)
has relatively less elongated proximal caudal centra,
suggesting that the elongation of these centra is
somewhat intraspecically variable.
Centra of caudal vertebrae 23 to 33 with the shape
of amphicoelous semicylinders. The centra of vertebrae
23 to 33 are “platycoelous” (= amphiplatyan) and
“hexagonal in cross section” in Tenontosaurus tilletti
(Forster 1990: 278). In Iguanodon bernissartensis (see
Norman 1980: 43) and Mantellisaurus athereldensis
(see Norman 1986: 310), these centra are also am-
phiplatyan and “hexagonal cylinders” i.e., they have
a hexagonal cross-section. Caudal centra are amphi-
platyan in the non-hadrosaurid hadrosauroid Bactro-
saurus johnsoni and mid-tail centra have a hexagonal
cross-section (Godefroit et al. 1998). The articular
facets of the posterior mid-caudal centra of the
non-hadrosaurid hadrosauroid Eolambia caroljonesa
are slightly concave and hexagonal (McDonald et
al. 2012a, 2017). Caudal centra 23-35 of the non-
hadrosaurid hadrosauroid Nanyangosaurus zhugeii are
“platycoelous” and have “regular hexagonal...out-
lines in anterior or posterior view” (Xu et al. 2000:
37-38). In hadrosaurids, the “outline” of the cau-
dal centra is hexagonal according to Horner et al.
(2004: 453). At least in the saurolophine Edmonto-
saurus annectens (see Sternberg 1926), Brachylophosaurus
canadensis (see Prieto-Márquez 2001) and Saurolophus
angustirostris (see Maryánska & Osmólska 1984), mid-
caudal centra are amphiplatyan. The description of
the articular facets of the centra in the many other
hadrosaurid species with preserved mid-caudal ver-
tebrae was omitted in the papers dedicated to their
osteology.

Fabio Marco Dalla Vecchia
612
The peculiar shape of the mid-caudal centra
of the holotype of T. insularis had probably a specic
function, which is not immediately clear because of
the absence of extant analogues.
Number, shape and elevation of the pleurapophyses. In
reptiles, “generally the proximal caudals, to the num-
ber of half a dozen to a dozen or more, bear fused
ribs” (Romer 1956: 268). The holotype of T. insularis
has at maximum 14 pleurapophyses-bearing (i.e.,
proximal) caudal vertebrae. The saurolophine had-
rosaurid Edmontosaurus annectens (see Lull & Wright
1942: 79), Saurolophus osborni (see Brown 1913a: 389)
and Brachylophosaurus canadensis (see Prieto-Márquez
2001: 148), and the non-hadrosaurid hadrosau-
roid Nanyangosaurus zhugeii (see Xu et al. 2000: 188)
have 19 proximal caudals. N. zhugeii is higher in the
phylogenetic tree than T. insularis according to Mc-
Donald et al. (2012b). A specimen of the saurolo-
phine hadrosaurid Gryposaurus notabilis has at least
17 proximal caudals, but its tail is distally incomplete
(Parks 1920: 30). Among the lambeosaurine hadro-
saurids, Corythosaurus casuarius has 15 (Brown 1916:
711) or “anywhere from 13 to 16” (Ostrom 1963:
154) proximal caudals and Barsboldia sicinskii has at
least 16 (Maryánska & Osmólska 1981: 247). The
non-hadrosaurid hadrosauroid Xuwulong yueluni has
apparently 13 or 14 proximal caudals (You et al.
2001: g. 2). X. yueluni is more basal than T. insu-
laris according to McDonald et al. (2012b). Iguanodon
bernissartensis and Mantellisaurus athereldensis have 14
(see Norman 1980: 43) and 15 or 16 (Norman 1986:
310) proximal caudals, respectively. In the dryosau-
rid Dryosaurus altus, the pleurapophyses are present
up to caudal 12 (Galton 1981: 280). Tenontosaurus til-
letti has eight proximal caudals (Forster 1990). The
small basal iguanodontian Gasparinisaura cincosaltent-
sis has 11 proximal caudals (Coria & Salgado 1996:
448). Therefore, the number of caudals with pleur-
apophyses appears to be somewhat phylogenetically
biased and T. insularis retains a plesiomorphically low
count. Despite to this, the higher elongation of the
centra compensates for the low number of proximal
caudals, and the proximal segment of the tail of T.
insularis is proportionally longer than that of a had-
rosaurid.
The pleurapophyses of the rst ve caudals
are tongue-shaped in the holotype (Fig. 5), where-
as they appear to be spatula-shaped in SC 57247.
Therefore, the shape of the pleurapophyses may be
somewhat intraspecically variable, if this is not an
artefact of preparation.
Broad pleurapophyses offer a larger surface for
the origin or insertion of the tail musculature than
slender pleurapophyses. The epaxial M. longissismus
caudae (LCA) inserts dorsally on the pleurapophyses
in extant crocodilians (Frey 1982; Persons & Cur-
rie 2011a; Mallison 2011) and it is supposed to have
inserted there also in non-avian dinosaurs (Arbour
2009: g. 9a-c; Persons & Currie 2011a: g. 2a-b;
Mallison 2011: g. 4). LCA is supposed to originate
from the postacetabular process of ilium in non-
avian dinosaurs (Dilkes 2000: 103; Arbour 2009: 7)
and to extend along the full length of the tail. The
pleurapophyses have a relationship also with the
hypaxial M. caudofemoralis longus (CFL), which origi-
nates from the lateral sides of the haemapophyses
and the lateral and ventral sides of the centra of the
haemapophyses-bearing vertebrae in extant crocodil-
ians (Ibiricu et al. 2014: tab. 3) and supposedly also in
dinosaurs (e.g., Gatesy 1990; Wilhite 2003; Persons
& Currie 2011a; Otero et al 2012; Ibiricu et al. 2014).
In extant crocodilians, the CFL does not originate
from the pleurapophyses (Wilhite 2003: 83; Persons
& Currie 2011a: 125) or it does it minimally only in
the “rst few” pleurapophyses (Ibiricu et al. 2014:
462). Nevertheless and despite the fact that the CFL
ends before the disappearance of the pleurapophy-
ses in some extant reptiles (Persons & Currie 2011a),
a correlation between the decreasing in size of the
pleurapophyses and the distal tapering of the CFL
is hypothesized in many reconstructions of the tail
musculature of non-avian dinosaurs (Gatesy 1990;
Wilhite 2003; Persons & Currie 2011a; Otero et al.
2012; Ibiricu et al. 2014) “because the transverse pro-
cesses form a “shelf ” under which M. caudofemoralis
longus runs” (Wilhite 2003: 93). Being the CFL dor-
sally constrained by the pleurapophyses, the position
and orientation of these processes are indicative of
the dorsal extent of the muscle (Wilhite 2003; Per-
sons & Currie 2011a, 2014). As in the other hadro-
sauroids, the pleurapophyses of T. insularis are not
elevated above the centrum, unlike the condition in
non-avian theropods, where elevation of the pleur-
apophyses above the centrum indicates a higher dor-
sal development of the hypaxial musculature than in
hadrosauroids (Persons & Currie 2011a, b, 2014).
Shape and size of the neural spines. Although
there appears to be a slight difference in shape

The tail of Tethyshadros insularis 613
among the proximal neural spines in the holotype
and SC 57247, the upper part of the spine in lateral
view is apomorphically expanded in both of them.
As seen above, the neural spines of the proximal
caudal vertebrae of T. insularis are comparatively
much broader anteroposteriorly than those of the
other hadrosauroids from the European archipela-
go. The neural spines are comparatively broader in
basal hadrosauroids and more basal iguanodontians
(for example, Iguanodon bernissartensis; Norman 1980:
g. 47) than in hadrosaurids and became peculiarly
taller in the lambeosaurines (Horner et al. 2004).
Therefore, the shape and size of the neural spines
appear to be somewhat biased phylogenetically.
Taller neural spines indicate a higher development
of the epaxial musculature and have therefore a
functional signicance (Schwarz-Wings et al. 2009),
which has never been investigated in hadrosauroids.
In the holotype of T. insularis, the neural
spines are broader in correspondence with the larg-
er and broader pleurapophyses of the chevron-less
rst proximal vertebrae. Broader neural spines give
a wider surface for the origin and insertion of mus-
cles of the epaxial M. transversospinalis system (Organ
2006; Arbour 2009: g. 9; Persons & Currie 2011a:
g. 2; Mallison 2011: g. 4), which is associated with
the ossied tendons (Organ 2006) occurring up to
caudal vertebra 20 in the holotype of T. insularis.
The anteroposteriorly long apical margin of
the neural spines of T. insularis suggests the pres-
ence of a comparatively long attachment for the
ligamentum supraspinale (Schwarz-Wings et al. 2009).
However, the transversely attened spines indicate
that the diameter of this ligament was reduced.
The comparatively low and broad proximal
neural spines of T. insularis might be better suited to
avoid injuries during mating than the taller and slen-
der spines of the hadrosaurids (Rothschild 1994);
in that case, the spinal shape might be sexually di-
morphic.
Shape and size of the haemal arches. According
to Horner et al. (2004: 453), the rst ve haemal
arches are more inclined posteriorly than the fol-
lowing ones and often touch each other in hadro-
saurids, while, according to Lull & Wright (1942:
81), “the anterior three or four chevrons are smaller
and are bent strongly backward to clear the essen-
tial passages which must have lain above the ischia”.
For Romer (1956: 273), “in hadrosaurs... the rst
few chevrons may be bent backward, presumably
to clear the pelvic outlet”. This higher backward
slope of the rst haemapophyses can be observed
in Brachylophosaurus canadensis (see Prieto-Márquez
2001: g. 52); Gryposaurus notabilis (see Parks 1920:
pl. 1); Saurolophus osborni (see Brown 1913a: pl. 63)
and S. angustrostris (see Maryánska & Osmólska
1984: 126); Edmontosaurus regalis (see Lull & Wright
1942: pl. 12B); and possibly E. annectens (see Lull &
Wright 1942: pls 13 and 15 [it is shown in mount-
ed skeletons] and Prosaurolophus maximus (see Parks
1924: pl. 5 [also a mounted skeleton]). This is the
condition in the holotype of T. insularis too (Figs. 1
and 4A-B). The posterior inclination of the proxi-
mal chevrons occurs also in Iguanodon bernissartensis
(see Norman 1980: g. 47). Only the rst chevron
has a different slope in Xuwulong yueluni (see You et
al. 2011: g. 2), Mantellisaurus athereldensis (see Nor-
man 1986: g. 39) and Corythosaurus casuarius (see
Lull & Wright 1942: pl. 27A). In the indeterminate
hadrosaurid TMP 1998.058.0001, only the rst two
chevrons are shorter and more inclined backward
than the others (Fig. 7A). In Tenontosaurus tilletti, the
rst two chevrons are only much shorter than the
others but are not more inclined backward and do
not touch each other (Forster 1990: g. 5). Non-
avian theropods (e.g., Gilmore 1920: pls 16, 22, 29,
30 and 35; Russell 1972: gs 1, 4 and pl. 1; Holtz
2004: g. 5.8; Holtz et al. 2004: g.4.8; Carpenter
et al. 2005: g. 3.14; Paul 2010; Persons et al. 2015);
basal sauropodomorphs (Galton & Upchurch 2004:
g. 12.1; Paul 2010); sauropods (Upchurch et al.
2004: g. 13.1; Paul 2010; Otero et al. 2012: gs.
4B-C and 5B); basal thyreophoran (Norman et al.
2004: g. 15.4; Paul 2010); stegosaurs (Paul 2010);
ankylosaurs (Carpenter 1997: g. 22.2; Paul 2010);
basal iguanodontians (Norman 2004: g. 19.13;
Paul 2010); heterodontosaurids (Paul 2010); pachy-
cephalosaurs (Paul 2010); basal neoceratopsians
(Forster & Sereno 1997: g. 23.4; Hailu & Dodson
2004: g. 22.2; Paul 2010); and ceratopsids (Forster
& Sereno 1997: g. 23.5; Paul 2010) are not recon-
structed with the rst ve haemapophyses that are
more backward inclined than the following ones
and contacting each other. Therefore, this arrange-
ment of the rst haemapophyses seems to occur
only within derived iguanodontians.
Unlike sauropods (Wilhite 2003; Otero et
al. 2012) and some non-avian theropods (Persons
& Currie 2011a), there is little change of form

Fabio Marco Dalla Vecchia
614
throughout the chevron series in hadrosaurids (Lull
& Wright 1942: 81). The boot-like and the follow-
ing bilobate chevrons of the holotype of T. insularis
are not reported in any other hadrosauroids, whose
haemal arch spines are just laterally compressed,
nger-like rods without any anterior or posterior
processes (e.g., Lull & Wright 1942; Norman 1986,
2004; Horner et al. 2004).
A change in chevron shape along the tail simi-
lar to that observed in the holotype of T. insularis
occurs in some sauropods (e.g., Mamenchisaurus, Shu-
nosaurus and Diplodocus; Otero et al. 2012: g. 4B-C)
and in ornithomimid (Russell 1972: gs. 1 and 4;
Persons & Currie 2011a: gs. 5-6) and tyrannosau-
rid theropods (Lambe 1917: gs. 5, 14 and 19; Mat-
thew & Brown 1923: gs. 2 and 4; and Brochu 2003:
g. 68). A specimen of the tyrannosaurid Gorgosau-
rus has chevrons very similar to the boot-like chev-
rons 12-13 of the holotype of T. insularis (with an
extremely long posterior process of the spine; Per-
sons & Currie 2011a: g. 7). The extant crocodilians
Caiman and Paleosuchus also show a similar pattern
(Frey 1982: g. 36). The dromaeosaurid theropod
Deinonychus antirrhopus has bilobate ‘proximal’ chev-
rons that resemble the distal chevrons of T. insularis
(Ostrom 1969: g. 40A). Therefore, the shape of
the haemal arches of T. insularis is unusual within
the hadrosauroids but not within the Dinosauria.
In theropod and sauropod dinosaurs, the
morphology of the spine of the boot-like and bi-
lobate haemal arches is supposed to reect the re-
lationships between the CFL and the M. ilioischio-
caudalis (M. iliocaudalis + M. ischiocaudalis) and to be
indicative of size and extent of the CFL (Wilhite
2003; Persons & Currie 2011a; Otero et al. 2012). In
these dinosaurs, the change in the shape of the hae-
mal spine from digit-like to asymmetrical (i.e., boot-
like) is supposed to be related to the tapering of
the CFL, which is observed in extant crocodilians
(Persons & Currie 2011a). In dinosaurs, this taper-
ing would be recorded by “diagonal scarring” (Per-
sons & Currie 2011a: 119) on the lateral surface of
the haemal spine (“blade”) of a few asymmetrical
(i.e., boot-like) and forked (i.e., bilobate) chevrons
(Persons & Currie 2011a: gs. 5-7; Otero et al. 2012:
g. 5B). This “scarring” is identied as the trace of
the septum that divided the CFL from the M. iliois-
chiocaudalis (Wilhite 2003; Persons & Currie 2011a;
Otero et al. 2012) and occurs in haemal arches lo-
cated just anterior the disappearance of the pleur-
apophyses in theropods (Persons & Currie 2011a:
119), whereas it is found in the haemapophyses set
in correspondence of the reduction “in size and
development” of the pleurapophyses in the sauro-
pod Diplodocus (Otero et al. 2012: 251). According to
Persons & Currie (2011a: 125), “the anteroposterior
ascent and eventual posterior disappearance of this
scar is, therefore, taken to mark the M. ilioischiocauda-
lis’ gradual dorsal intrusion and eventual usurpation
of the M. caudofemoralis”.
As seen above in the description of T. in-
sularis, a “diagonal scarring” like that reported by
Persons & Currie (2011a) and Otero et al. (2012)
cannot be clearly identied in the haemal arches of
the holotype of T. insularis. The spines of haemal
arches 12 and 13 show faint and posteroventrally to
anterodorsally oriented ridges (Fig. 6E-F); distinct
ne striations with a similar direction are spread
over the spine of haemal arch 13 (Fig. 6F). Diago-
nal ne striations occur also in the distal part of the
spine of the other boot-like haemal arches (Fig. 6B-
D). However, the orientation of these striations is
opposite to that of the “diagonal scarring” reported
by Persons & Currie (2011a) and Otero et al. (2012).
Therefore, a relationship between the shape of the
haemal arch and the relative position of the CFL
and M. ilioischiocaudalis is not clearly established in T.
insularis. Furthermore, there is no correspondence
between the disappearance of the pleurapophyses,
change in shape of the haemal arches and their re-
duction in size in T. insularis. In the holotype, the
haemal arches gradually decrease in size from hae-
mal arch 6 to 14 (caudals 12/13 to 20/21), pleur-
apophyses disappear at caudal 13 or 14 and the
boot-like haemal arches 12-13 articulate on caudals
18-19 and 19-20, thus they are posterior to the last
pleurapophyses.
Ibiricu et al. (2014: 469) hypothesized that
“ventrally broad haemal arches [of the proximal
caudals] may correspond to an increased surface of
attachment for the CFL” in large titanosaurian sau-
ropods.
As in crocodilians, the broad haemal spines
with anterior and posterior processes gave more
room also for the insertion of the interhaemal
ligaments (Frey 1988: g. 17; Schwarz-Wings et al.
2009).
The distally expanded haemal spines could
have provided more surface for the attachment of
the M. ischiocaudalis (see Arbour 2009: g. 9; Persons

The tail of Tethyshadros insularis 615
& Currie 2011a: g. 2B), not only in correspondence
of the tapering of the CFL, but also posterior to its
disappearance.
Tail stiffness. The caudal vertebrae of T. insularis
show two features that have often been considered
to increase the tail rigidity: nearly vertical articular
facets of the zygapophyses and presence of epaxial
ossied tendons (e.g., Forster 1990; Maxwell & Os-
trom 1995).
The nearly vertical articular facets of the zyg-
apophyses limit long axis rotation and lateral exibil-
ity of the tail, although exibility increases posterior-
ly as zygapophyses reduce in size (Schwarz-Wings et
al. 2009). However, steeply inclined articular facets
of the zygapophyses alone do not totally hinder the
lateral exibility of the tail, as it is demonstrated by
the curled tails of living and fossil archosaurs with
steeply inclined articular facets of the zygapophyses
(Frey 1982; Maxwell & Ostrom 1995: g. 2; Ford &
Martin 2010: 334).
The ossied epaxial tendons are an ornith-
ischian synapomorphy (Sereno 1986). In the basal
iguanodontian Tenontosaurus tilletti, ossied tendons
“extend to the end of the tail” (Forster 1990: 280)
and occur also hypaxially. They end at about caudal
20 in Iguanodon bernissartensis (see Norman 1980) and
Mantellisaurus athereldensis (see Norman 1986). Ac-
cording to Horner et al. (2004: 453), ossied ten-
dons extend “caudally to about the midsection of
the tail”. In the saurolophine hadrosaurid Brachylo-
phosaurus canadensis, the lattice of ossied tendons is
developed up to caudal 15, then ossied tendons are
arranged parallel to the curvature of the tail (Prieto-
Márquez 2001: 154). Ossied tendons seem to reach
at about caudal 25 in the lambeosaurine Corythosaurus
casuarius (Brown 1916: pls. 13-14). Their function in
T. insularis is plausibly the same as in the other iguan-
odontians, not a specialization. The ossied tendons
trellis of the hadrosaurids restricted the dorsoventral
tail oscillation increasing spinal rigidity in the dorso-
ventral plane (Horner et al. 2004; Persons & Currie
2012) and reinforced proximally the tail against dor-
soventral shear (Schwarz-Wings et al. 2009). Howev-
er, they did not prevent the lateral motion of the tail
(Organ 2006). An important feature of ossied ten-
dons is to store and release elastic energy in a more
effective way than non-mineralised tendons (Organ
2006), thus these structures helped in the quick re-
cover after lateral bending of the tail.
Nearly vertical articular facets of the zyg-
apophyses and presence of epaxial ossied tendons
occur also in other hadrosauroids and did not make
the tail of T. insularis peculiarly rigid. However, T.
insularis presents other features suggesting a com-
paratively higher stiffness of the proximal portion
the tail. The passive rigidity of this portion of the
vertebral column was increased by the unusually
broad pleurapophyses and neural spines (Hildeb-
rand & Goslow 1998). The anteroposteriorly ex-
panded neural spines favoured the apical insertion
of the supraspinous ligaments, which stiffen the tail
dorsally, and reduced interspinal lengths; likewise,
the anteroposteriorly broad haemal spines stiffened
the tail ventrally (Schwarz-Wings et al. 2009). The
elongation of the centra reduced the density of in-
tervertebral exure points lowering the degree of
tail exibility per unit of absolute tail length (Per-
sons et al. 2014).
Thus, the deeper proximal part of the tail of
T. insularis was relatively rigid and quick in recover-
ing from lateral oscillations, whereas the low distal
part, with small zygapophyseal surfaces and without
ossied tendons, was probably freer to move on the
vertical and lateral planes. The low and more mo-
bile distal part of the tail might have been used as a
whip, possibly for courtship, intraspecic signalling
and sexual display (Myhrvold & Currie 1997).
Tail musculature. As in crocodilians and in the
muscular reconstructions of other non-avian dino-
saurs, the principal epaxial muscles of the tail of
T. insularis were the M. transversospinalis system and
the LCA, the latter extending along the whole tail.
The rst inserted onto the lateral side of the neural
spine, the other onto the lateral side of the neural
pedicels (the base of the neural arch) and the dorsal
side of the pleurapophyses in the proximal caudals
and on the dorsal side of the centrum in the middle
and distal caudals (Frey 1988; Arbour 2009; Persons
& Currie 2011a; Mallison 2011). Therefore, these
muscles had broad insertion surfaces in T. insularis.
Actually, the M. semispinalis and M. spinalis of the M.
transversospinalis system were anchored to the bone
by tendons that originated and attached to the mar-
gin of the spine (Organ 2006), thus it is the shape
of the spine more than its anteroposterior surface
to favour their insertion. Up to caudal 20 of SC
57021, tendons of M. semispinalis were ossied, re-
vealing the presence of this muscle.

Fabio Marco Dalla Vecchia
616
The anteroposterior breadth of the neural
spines and pleurapophyses are much less important
than their proximodistal elongation in assessing the
development of the epaxial musculature (Persons &
Currie 2011a,b, 2014). As the caudal neural spines
of T. insularis are comparatively lower than those of
lambeosaurine hadrosaurids and the pleurapophy-
ses are not particularly long proximodistally, the
proximal caudal epaxial musculature of T. insularis
was comparatively less developed (see Persons &
Currie 2014) than in lambeosaurine hadrosaurids,
which were the most common hadrosauroids in the
latest Cretaceous European Archipelago. However,
the peculiar broadness of the neural spines and
pleurapophyses of SC 57021 imply that intrinsic
Mm. interspinales and Mm. intertransversarii were short-
er than in other hadrosauroids.
Posterior to the disappearance of the pleur-
apophyses, the ventral limit of the LCA was prob-
ably marked by the sharp longitudinal ridge located
in the dorsal part of the lateral surface of the cen-
trum (caudals 15-22; SI, Fig. 1), whereas this muscle
had probably a broad ventral insertion on the at
dorsal side of the hemicylindrical centra of caudals
23-33 (cf. Frey 1988: g. 46C; SI, Fig. 2C and D).
Plausibly, the M. transversospinalis system sensibly re-
duced its size or disappeared with the neural spine
reduction around caudal 20 (cf. Frey 1988: g. 46;
Schwarz-Wings et al. 2009: g. 9E).
The main hypaxial muscles in the tail were the
Mm. caudofemorales (CFL [see above] and M. caudofem-
oralis brevis, CFB), M. ischiocaudalis and M. iliocaudalis.
In extant crocodilians, the CFL inserts via ten-
don onto the 4th trochanter of the femur, whereas
the CFB inserts onto a slightly more proximal posi-
tion (Frey et al. 1989; Otero et al. 2010).
In extant crocodilians, the origin of the CFL
is strictly related to the haemal arches (Frey 1982;
Wilhite 2003; Persons & Currie 2011a; Otero et al.
2012; Ibiricu et al. 2014). The point of origin of the
CFL that is closer to its femoral insertion is onto
the rst haemal arch, whereas the insertion onto
the ventrolateral surfaces of the centra starts more
posteriorly (Frey 1982). The strict relationship of
the CFL with the rst chevrons is observed also in
limbed squamates (e.g. Russell et al. 2001; Persons
& Currie 2011a; Ibiricu et al. 2014). The CFL is the
main femoral retractor in extant reptiles (Gatesy
1990; Otero et al. 2010; Persons & Currie 2011a;
Ibiricu et al. 2014) and is considered to have had a
pivotal importance in the locomotive power stroke
of non-avian dinosaurs (Romer 1927; Gatesy 1990;
1995; Persons & Currie 2011a, 2014).
In the extant crocodilian Caiman latirostris, the
CFB originates on the posteroventral portion of the
ilium and on the lateral sides of the centrum and
ventral side of the pleurapophyses of the rst caudal
vertebra (Otero et al. 2010: 176), which is the only
caudal vertebra not to bear a chevron in crocodil-
ians, as the rst chevron usually articulates between
caudals 2/3 (Erickson et al. 2005). This location of
the CFB origin has been reported also from other
crocodilians (Ibiricu et al. 2014: 456, 462 and tab. I).
In non-avian theropods and sauropods, the CFB has
consequently been reconstructed as originating from
the proximal, chevron-less caudal vertebrae (Persons
Fig. 17 - Reconstruction of the M.m caudofemorales and M. transversus perinei in Tethyshadros insularis. Reconstruction is based on the relationships
of M.m caudofemorales and skeletal elements in extant crocodilians and limbed squamates. Red, M. caudofemoralis longus; green, M. caudo-
femoralis brevis; orange, M. transversus perinei. Skeletal reconstruction is by M. Auditore, muscle reconstruction is by the author. The scale
bar equals 500 mm.

The tail of Tethyshadros insularis 617
& Currie 2011a: g. 9; Ibiricu et al. 2014: g. 6). The
CFB is a femoral retractor as the CFL but it is con-
sidered to have had a secondary role in the locomo-
tive power stroke of non-avian dinosaurs (Romer
1927; Gatesy 1990; Persons & Currie 2011a).
Inferring this origin for the Mm. caudofemorales
in T. insularis, the chevron-less vertebrae 1 to 6 or 7
and their broad pleurapophyses were the place of
origin of the CFB, whereas the origin of CFL was
displaced posteriorly in the tail (Fig. 17). Because
the rst chevron has shifted to a more distal posi-
tion and consequently the rst chevrons are shorter
than they would be if they were located more proxi-
mally, the proximal transversal section of the CFL,
where the muscle is deeper, is lesser in T. insularis
than in other hadrosauroids with a more proximally
placed rst chevron.
The “transition point” (sensu Gauthier
1986:19) and the distal-most extent of the CFL ori-
gin (Gatesy 1995) can be located around caudal ver-
tebra 20 (rather posterior to the last pleurapophy-
ses) in the holotype of T. insularis.
Maintaining the proportions, the shape and
the relationships with the skeletal elements that the
reconstruction of the CFL has in other dinosaurs
(Wilhite 2003: g. 5.13; Gallina & Otero 2009: g. 7;
Persons & Currie 2011a: gs. 9 and 11, 2011b: g. 3,
2012: g. 5, 2014: g. 26.2; Persons et al. 2014: gs.
7-9; Otero et al. 2012: g. 5; Ibiricu et al. 2014: g.
6), this muscle results to be comparatively shorter
and much more slender in T. insularis than in other
dinosaurs and its place of origin extends along the
vertebral column less than half the length of the
whole muscle and its tendon (Fig. 17). Conversely,
the CFB results be comparatively much more devel-
oped and its role in femur retraction would be more
relevant than in other reptiles.
It might be possible that, unlike extant rep-
tiles but like the M. caudofemoralis pars caudalis of the
ostrich (Ibiricu et al. 2014), the CFL of T. insularis
originated also from the lateroventral sides of the
centra of the chevron-less caudals, and possibly
also from the ventral side of their pleurapophyses.
In that case, the CFB might be relegated to cau-
dals 1-2 and/or the posteroventral side of the long
postacetabular process of ilium; the CFL would
be larger than in the reconstruction of gure 16
(which is based on its origin from the chevron-
bearing vertebrae only), although its maximum
cross-section would not be as deep as the muscle
had originated from longer chevrons articulated to
the rst caudals. However, the development of M.
caudofemoralis pars caudalis and M. caudofemoralis pars
pelvica (which are the avian homologue of crocodil-
ian CFL and CFB, respectively) is a consequence of
the extremely short caudal segment of the vertebral
column of the Pygostylia and the function of con-
trol over the tail feathers and is related to the abil-
ity to y (Gatesy 1995; Ibiricu et al. 2014), which is
obviously not the case of T. insularis.
In the crocodilian Alligator sinensis, the dorsal
venter of M. transversus perinei (TP) originates from
the surface of the centra of caudals 1- 4 and is con-
nected by an aponeurosis to a ventral venter that is
split into two parts, one inserting onto the distal end
of the ischium and the other into the cloacal carti-
lage (Cong et al. 1998). According to Frey (1982),
the origin of the dorsal venter of TP is on vertebrae
1-2 (i.e., the chevron-less vertebrae) in Caiman and
Paleosuchus. TP wraps the CFL in extant crocodyl-
ians. If TP was present in T. insularis, it had to be
much displaced posteriorly, as origin and insertion
as well, with respect the other dinosaurs (Fig. 17).
Locomotion. The epaxial M. transversospinalis
system and LCA and the hypaxial M. ischiocaudalis
and M. iliocaudalis would function to bend the tail
(Ostrom 1969; Arbour 2009). Synchronous contrac-
tions of the epaxial musculature produced tail ex-
tension in the vertical plane, whereas synchronous
contractions of the hypaxial musculature resulted in
ventral exion (Schwarz-Wings et al. 2009). How-
ever, synchronous contractions of the epaxial and
hypaxial musculature of both sides could stiffen the
tail, avoiding oscillations and stabilizing it during lo-
comotion. This may have helped T. insularis in main-
taining balance during the irregular movements (cf.
Ostrom 1969) involved in locomotion on a rough
ground as that of a karst landscape.
As seen above, the CFL is considered to
be the major contributor to the power stroke of
the hind limbs in all gaits in non-avian dinosaurs
(Gatesy 1990, 1995; Persons & Currie 2011a, 2014).
Considering the reconstruction of the Mm. caudo-
femorales of T. insularis inferred from the relation-
ship of these muscles with the skeletal elements in
extant crocodilians (Fig. 17), the CFL of T. insularis
has a comparatively shorter origin area, a much
longer segment from the insertion on the femur to
the closer origin on the rst chevrons and a lesser

Fabio Marco Dalla Vecchia
618
maximum cross-sectional area than other dinosaurs
with the tail musculature reconstructed in the same
way, whereas it has a much larger CFB (Persons &
Currie 2011a: g. 9, 2014: g. 26.2B and E; Persons
et al. 2014: 7B1; Otero et al. 2012: g. 5; Ibiricu et
al. 2014: g. 6). This has probably an impact in the
role of those two muscles in locomotion and casts
doubts about the analogy of their functions in croc-
odilians and T. insularis.
The cross-sectional area of the CFL at
its greatest width is the most important param-
eter linked to the locomotive contribution of the
muscle (Persons & Currie 2014). A comparatively
lesser maximum cross-sectional area of the CFL
than other dinosaurs would suggest a comparatively
lesser contractile force (Ft) and torque exerted by
this muscle in T. insularis with respect to the same
muscle in a hadrosauroid with a more proximally
placed rst chevron (moment arm length being
the same for the torque) (Persons & Currie 2011a,
2014). Furthermore, the torque of the CFL would
be proportionally lower in T. insularis than in an av-
erage hadrosauroid, because the femur of T. insularis
is much shorter than tibia (Fig. 17), whereas femur
is usually slightly longer than tibia in hadrosauroids
(Lull & Wright 1942; Norman 2004). In T. insularis,
the point of insertion of the tendon of the CFL
(and that of CFB as well) was proportionally closer
to the joint’s centre of rotation (acetabulum) than
in those hadrosauroids. Therefore, the moment arm
length (R, measured from the joint’s centre of rota-
tion and the muscle’s insertion) was comparatively
shorter in T. insularis than that of a hadrosauroid
with a hind limb of the same length but with a pro-
portionally longer femur, assuming that the 4th tro-
chanter, which is not preserved in any T. insularis
specimens, was in the same position in T. insularis as
it is in the other hadrosauroids. According to lever
mechanics, muscles inserting closer to a joint work
at higher velocities; CFL insertion is more proxi-
mally located to the joint’s centre of rotation in taxa
that move more quickly (Persons & Currie 2011a,
2014; Maidment et al. 2012). Because of the pos-
terior displacement of the origin of the CFL, the
angle between the vector line of pull (equivalent to
the orientation of the CFL tendon) and the vector
orthogonal to the moment arm (θ) would be close
to 0 (Fig. 17). As the effective force (Fe) that the
muscle exerts is calculated with the equation Fe =
Ft·cosθ (Persons & Currie 2011a: 122), the Fe ex-
erted by the CFL of T. insularis was close to the Ft
and the effectiveness of the joint was close to the
maximum (Ibiricu et al. 2014: 15). This somewhat
compensated the proportionally shorter moment
arm length, as the potential torque generation (τm)
is calculated with the equation τm = R Fe·sinΦ (Φ is
the angle between the vectors of the moment arm
and the effective force, which is assumed to be 90°
when the femur is positioned perpendicular to the
ground; Persons & Currie 2011a: 122).
The shortening of the moment arm length,
the posterior displacement of the CFL origin, with
consequent elongation of the muscle but lower-
ing of its maximum cross-sectional area, and the
comparatively larger CFB suggest that the locomo-
tor abilities of T. insularis were somewhat different
from those of the other hadrosauroids (Persons &
Currie 2011a, 2014). This may be an adaptation to
its peculiar life environment: locomotion on the un-
even and rugged karst ground instead of that at
and regular of the ooding and coastal plains. It
may be a modication to climb, jump or walk on
rugged stony grounds rather than to reach a more
or less fast or sustained locomotion. As an alterna-
tive, the tail musculature of T. insularis might just be
reshaped as a consequence to the posterior shift of
the vent, which would be the main adaptation of
this dinosaur, in order to obtain the same locomo-
tor performance as other hadrosauroids.
However, also limb features support an ad-
aptation of T. insularis to a peculiar style of loco-
motion. The manus of T. insularis is similar to that
of the hadrosaurids in its overall structure (Brown
1912; Lambe 1913; Parks 1920: g. 13). Humerus,
radius and ulna and their articulation (Dalla Vec-
chia 2009c: gs 1 and 6) is like that of other hadro-
sauroids (Norman 1986, 2004; Horner et al. 2004;
Brett-Surman & Wagner 2007), with the radius par-
allel to ulna and rmly set in front of it and the
manus with the palm facing mediocaudally (Senter
2012). Manus/forelimb length ratio (metacarpal III
length/radius+humerus length ratio) is 0.226 and
0.210 (because of the different length of the right
and left bones; Dalla Vecchia 2009c: tab.1) in the
upper range of the Hadrosauridae, whose elonga-
tion of the manus is possibly an autapomorphic
feature (Maidment & Barrett 2014: 58). Radius/hu-
merus ratio (0.922 and 0.975) is within the range of
basal and saurolophine hadrosaurids (Lull & Wright
1942: tabs 4-8; Maidment & Barrett 2014: 60).

The tail of Tethyshadros insularis 619
Forelimb/hind limb length ratio (radius+humerus
length/femur + tibia length ratio, 0.584) is that usu-
al for hadrosauroids (Maidment & Barrett 2014: tab.
2). However, the hands of T. insularis are tridactyl
because they lost digits V, which were not involved
in support during locomotion, anyway (Dilkes 2000:
g. 3). The remaining digits II-IV of T. insularis
had no grasping abilities, because the distal articu-
lar surfaces of the metacarpals are at (Fig. 16B)
and phalanges II-2 and III-2 are short and wedge-
shaped (Fig. 16A-B) (see also Maidment et al. 2012:
6-7). Extension and exion were not possible at the
metatarsal-phalangeal joints and at the phalangeal
joints, with the possible exception of the phalangeal
joint of the phalanx III-1. However, phalanx III-1
seems to be fused with the metacarpal III in the
right manus of SC 57022 (Fig. 16D). Furthermore,
the elongated metacarpals are tightly appressed to
form a single functional unit, the phalangeal por-
tion of the manus is very short (Fig. 16) and the car-
pus is reduced to a single, very small and rounded
carpal (Fig. 16A). Ungual phalanges of digits II and
III are hoof-like, but digit IV lacks a hoof-like un-
gual phalanx (Fig. 16B-D). The manus of T. insularis
has on the whole the aspect of a pillar.
Unlike all other hadrosauroids, the tibia is
much longer than femur in T. insularis (femur/tibia
length ratios is 0.76; Dalla Vecchia 2009c). The pes/
hind limb length ratio (metatarsal III/femur+tibia
length ratio, 0.199; see Dalla Vecchia 2009c: tab.1) is
slightly higher than in hadrosaurids (Lull & Wright
1942: tabs 4-7; Maidment & Barrett 2014: 63), but
this ratio is negatively size-correlated in ornithischi-
ans (Maidment & Barrett 2014: 65). The pedes have
very short phalangeal portions (SI, Fig. 6A), even
shorter than those already short of the hadrosaurids
(e.g., Maiasaura peeblesorum, see Dilkes 2000: g.12G-
H; Saurolophus sp., Moreno et al. 2006: g. 3E; SI,
Fig. 6B). Non-ungual phalanges distal to those of
the rst ray are disc-like and ungual phalanges are
hoof-like and much wider than long. Digit mobility
was possible only at the metatarso-phalangeal joint
and between phalanges 1 and 2.
Modications from the basal hadrosauroid
pattern occurred in T. insularis’ manus and pes re-
duced the mass of the limbs (Hildebrand & Gos-
low 1998) and are a further development toward the
subunguligrade posture shown by the hadrosaurids
(Moreno et al. 2006). T. insularis shows a puzzling
mixture of features suggesting both quadrupedal-
ity (subunguligrade posture, hoof-like manual un-
gual phalanges and broad supracetabular process of
ilium) and bipedality (femur much shorter than tibia
and comparatively high pes/hind limb length ratio)
(Maidment & Barrett 2014).
Reduction of the mass of the limbs, verte-
bral column rigidity reducing oscillating motions,
short and robust femur, long and slender tibia, long
and compact metatarsus, CFL attachment located
proximal along the leg, high radius/humerus ratio
(slender radius-ulna) and digitigrade manus with
long and tightly appressed metacarpals are ‘curso-
rial’ features (Coombs 1978; Hildebrand & Goslow
1998; Maidment et al. 2012). However, according
to Maidment et al. (2012: 3) “cursorial morphology
has not always been found to closely correlate with
maximum running speed in extant mammals, and
it may correspond with other features of locomo-
tor performance, such as stamina or locomotor ef-
ciency at slow speeds”.
According to Brown (1912: 107), the manus
of the hadrosaurid Edmontosaurus “was no longer
used to any extent in progression” because of “the
extreme elongation of the metacarpals, the loose
articulation of the phalanges and the reduction of
the unguals to two functional hoofs”. Gracility of
the mani of the holotype of T. insularis and their
small bearing surfaces (Fig. 16A) suggest that they
might not be used primarily for weight-bearing (a
hypothesis already advanced by Ostrom 1964: 994
for hadrosaurs based on the reduced carpus). Mani
might be employed for steering and to balance the
vertical and lateral oscillations of the anterior part
of the body during locomotion, which would be
particularly useful on an uneven rugged ground.
Position of the rst haemal arch and concomitant
elongation of the ischium. In reptiles, haemal arches
usually begin at about the third to seventh caudal
(Romer 1956: 267). According to Horner et al.
(2004: 453) the rst haemal arch is distal to caudal
2 or 3 in lambeosaurines and distal to the fourth or
fth caudal in the other hadrosaurids. The rst hae-
mal arch occurs between caudals 4 and 5 in the sau-
rolophine Edmontosaurus annectens (see Lull & Wright
1942: 79), Gryposaurus notabilis (see Parks 1920: pl. 1)
and Saurolophus osborni (see Brown 1913a: 389) and
in the lambeosaurine Nipponosaurus sachalinensis (see
Suzuki et al. 2004: 152); it is located between cau-
dals 3 and 4 in the indeterminate hadrosaurid TMP

Fabio Marco Dalla Vecchia
620
1998.058.0001 (not considering the caudosacral in
the count of the caudals; see above). The rst hae-
mal arch occurs between caudal 2 and 3 in Iguanodon
bernissartensis (see Norman 1980: g. 47), Mantellisau-
rus athereldensis (see Norman 1980: 308, g. 39) and
the non-hadrosaurid hadrosauroid Xuwulong yueluni
(see You et al. 2011: g. 2); it occurs between cau-
dal 1 and 2 in the basal iguanodontians Tenontosaurus
tilletti (see Forster 1990: g. 5), Dryosaurus altus and
Dysalotosaurus lettowvorbecki (see Galton 1981: 280).
Therefore, the position of the rst haemal
arch appears to be somewhat biased phylogeneti-
cally and it is more distal in the tail in saurolophine
hadrosaurids than in most other iguanodontians.
However, T. insularis has apomorphically an even
more distal position of the rst chevron and a con-
sequent elongation of the ischiadic shaft, unlike all
other iguanodontians and dinosaurs in general.
The absence of the haemapophyses in the
most proximal caudals is related to the presence of
the cloaca and associated structures (Williston &
Gregory 1925: 110; Romer 1956: 267). The poste-
riorly displaced rst haemal arch and the concomi-
tant elongation of the centra displace posteriorly
the vent and copulatory organs, lengthening the
terminal tract of the intestine (colon) and possibly
the cloaca too. This is the main morphological pe-
culiarity of T. insularis’ tail and has important conse-
quences also on the tail musculature. As this feature
occurs in both specimens SC 57021 and SC 57247,
it is unrelated to their different body robustness and
is plausibly linked to a function that was advanta-
geous in the environment where these dinosaurs
lived.
The three main sources of body water loss in
reptiles are respiration, nitrogenous excretion and
transpiration through the body surface (Minnich
1982). In reptiles, the cloaca and colon modify the
urine produced by the kidneys and regulate renal
water (Minnich 1982). In crocodilians, which lack a
bladder, the cloaca and colon are the main urinary
storage sites (Minnich 1982). The reptilian cloaca
usually absorbs water, sodium and chlorine and se-
cretes urates (Minnich 1982); the urodaeum of the
cloaca is the primary site for urine modication in
crocodilians and is capable of storing large quanti-
ties of urine (Kuchel & Franklin 2000). In birds, the
colon is not only important for water absorption
from food, but also for its reabsorption (together
with caeca) from ureteral urine (post-renal water)
(McLelland 1990; Duke et al. 1995; Musara et al.
2002), although this is variable in different birds. In
the ostrich (Struthio camelus), there is no post-renal
water reabsorption (Skahdauge et al. 1984; Duke et
al. 1995) and the large size of the colon is an im-
portant anatomical adaptation to maximise the ab-
sorption of the ingested water (Musara et al. 2002).
In birds where the colon is relatively short “colonic
water absorption is maximised by retrograde ow
of urine into the lower gastrointestinal tract”; Mu-
sara et al. 2002: 318). Possibly, the longer colon of
T. insularis would maximise water absorption or a
longer colon and/or cloaca would store more urine
and allow maximising post-renal water reabsorption
in case of dehydration. The elongation of the colon
and possibly cloaca suggested by the posterior shift
of the vent could be an adaptation to a fresh water
depleted environment, as is the case of an emergent
carbonate platform with its prevailing underground
water ow, and subject to drought events affecting
also the development of the vegetation (possibly
the main source of water for this vegetarian dino-
saur).
The posterior shift of the vent and the con-
sequent distal genital position (McLelland 1990;
Ziegler & Olbort 2007) might have also reproduc-
tive implications. For example, it might make breed-
ing easier, although no clear advantage seems to
derive from the posterior shift of the vent, if hadro-
sauroids copulated as reconstructed by Isles (2009).
This aspect cannot be further investigated, because
we do not have much information about hadrosau-
roid genitals and copulation (Isles 2009).
The posterior shift of the vent implies also
the elongation of the oviducts in females, if the
shift does not correspond to a proportional elonga-
tion of the cloaca posterior to the opening of the
oviducts into the urodeum. Extant crocodilians have
two functional oviducts, ovulated all the eggs at the
same time and “eggshells are calcied in assembly-
line fashion along the oviduct, followed by simul-
taneous oviposition” (Schweitzer et al. 2007: 1156,
see also g. 1; see Fox 1977, and Zelenitsky & Hills
1997). Extant birds have a single functional oviduct
and produce one egg at a time (Varricchio et al.
1997; Zelenitsky & Hills 1997). At least some non-
avian theropods had two functional oviducts and
produce one egg per oviduct at a time (Varricchio
et al. 1997; Sato et al. 2005; Zelenitsky 2006; Yang
et al. 2019). Hadrosaurids are supposed to have had

The tail of Tethyshadros insularis 621
two functional oviducts and to produce more than
one egg per oviduct at a time like crocodilians (Zel-
enitsky & Hills 1997; Yang et al. 2019). Under these
assumptions, longer oviducts meant more room for
a comparatively higher number of fertilized eggs;
T. insularis would produce clutches composed of
a comparatively higher number of eggs respect to
other hadrosauroids with a proportionally shorter
oviduct, the egg size remaining comparable. This
might be an adaptation of an insular r-strategist
taxon (see Raia et al. 2003) to the variable and harsh
environmental conditions (mainly about water sup-
ply) of the Adriatic island.
concluSIonS
The minimum estimated total length of the
distally incomplete tail of the holotype of T. insu-
laris, based on the proportional centrum-by-cen-
trum scaling respect to the complete tail of a had-
rosaurid from the Upper Campanian of Canada, is
over 2.5 metres and accounts for 56% of the total
body length. The tail of T. insularis was rather long
and its portion posterior to vertebra 20 was prob-
ably whip-like.
The two main specimens of T. insularis differ
in robustness, in the shape of the haemapophyses
and possibly in the shape of the pleurapophyses.
These differences may represent sexual dimor-
phism (as it has been suggested for other dino-
saurs), although ontogeny or high intraspecic
variability due to the insular condition might be
alternative explanations.
Comparison between the caudal vertebrae of
T. insularis and those of the latest Cretaceous had-
rosauroids that lived in the European archipelago
shows that the vertebrae of the Italian taxon differ
from those of the other hadrosauroids, although
the rst caudals show some resemblance with those
from Romania, which may belong to Telmatosaurus
transsylvanicus (a taxon that is found to be close to T.
insularis in some phylogenetic analyses). The pecu-
liar features of the caudals of T. insularis (vertebral
centra longer than high also in the proximal ele-
ments; centra of the posterior mid-caudals with the
shape of amphicoelous semicylinders; pleurapoph-
yses of the rst ve caudals broad and tongue- or
fan-shaped; apically broad neural spines in lateral
view, which gradually change in shape from meat
cleaver-like in the rst proximal caudals, to fan- or
spatula-shaped in the posterior proximal and rst
mid-caudals, to ‘petaloid’ in the following mid-
caudals; haemal arches changing in shape along the
tail from rod-like to boot-like to bilobate; and rst
haemal arch set between caudals 7-8 or 8-9, with a
long, chevron-less proximal segment of the tail) are
not found in the other insular hadrosauroids of the
European archipelago as well as in continental had-
rosauroids and are conrmed as apomorphic traits
of the Italian taxon.
These apomorphies cannot be a only a con-
sequence of insularity, because all of the hadrosau-
roids from the European archipelago were insular
dwellers but they are more conservative than T.
insularis. The tail apomorphies of T. insularis may
be related to the karst landscape where the Italian
dinosaur lived. Skeletal features suggest that the
long tail was relatively stiff and deep proximally,
whereas it was whip-like and more movable distally.
The reconstruction of the tail musculature based
on comparison with that of living archosaurs and
the reconstructed musculature in other non-avian
dinosaurs, shows that the posterior shift of the
rst chevron affected the size and shape of the
M. caudofemoralis longus, which is usually the major
contributor to the power stroke of the hind limbs
in reptiles, and of the M. caudofemoralis brevis. This
different development of the hypaxial musculature
would probably have important consequences on
the locomotion of T. insularis, and may be related
to its life habits too. The posterior displacement
of the origin of the M. caudofemoralis longus, with
consequent elongation of the muscle but lowering
of its maximum cross-sectional area, the shorten-
ing of its moment arm length, and the compara-
tively more developed M. caudofemoralis brevis, sug-
gest that the locomotor abilities of T. insularis were
somewhat unlike those of the other hadrosauroids.
This is indicated also by some peculiar limb fea-
tures (including a compact tridactyl manus without
grasping ability and with a small bearing surface,
short phalangeal portions of the pedes and tibia
much longer than femur) suggestive of quadrupe-
dality as well bipedality and a cursorial locomotor
mode. These may be adaptations to an effective
locomotion (by climbing, jumping or walking) on
the stony and rugged karst ground rather than to
reach a fast or sustained locomotion. Stiffening of

Fabio Marco Dalla Vecchia
622
the tail, increased by synchronous contractions of
the epaxial and hypaxial musculature of both sides,
would avoid its oscillations and stabilize it during
locomotion on the rough ground. The forelimbs,
with their pillar-like mani, may be used for steering
and to balance the oscillations of the anterior part
of the body.
The posterior shift of the vent suggested by
the posterior position of the rst haemapophysis
implies a longer distal tract of the intestine or a
longer cloaca, which could increase the space for
water absorption or urine storage and urinary water
reabsorption, an advantage in the water-depleted
karst where the animal lived. In the case of a nor-
mally-sized cloaca, the posterior shift of the vent
would imply also longer oviducts and plausibly an
increased number of eggs per clutch, also a pos-
sible advantage in the stressing life environment of
T. insularis.
Acknowledgements: I would like to thank D. Arbulla (MCSNT)
for the access to the Tethyshadros material; A. Milner, B. Strilisky, J.
Le Loeuff, R. Gaete and J. Galindo for the support at the NHMUK,
TMP, MDE, MCD and IPS/ICP, respectively. I am grateful to A.
Prieto-Márquez (ICP) for the discussion and the information on
hadrosauroid osteology; H. Mallison for the discussion and the in-
formation on the tail muscles and their functions; A. Cau for the
discussions on the signicance of the tail peculiarities of Tethyshadros;
and the reviewers R. Coria and A. Prieto-Márquez for comments on
the manuscript and the revision of the English text. Thanks to G.
Tunis and S. Venturini for the information and comments on the
geology and palaeoenvironment of the Villaggio del Pescatore site. I
am grateful to D. Brinkman (TMP) and M. Fabbri (Yale University)
for the measurements of the specimen TMP 1998.058.0001 and to
M. Auditore for the drawing of the skeletal reconstruction of T. in-
sularis. The Soprintendenza Regionale ai Beni ed Attività Culturali del
Friuli Venezia Giulia granted the 1998–1999 eld work and permit-
ted the study of the Italian specimens in years 2007-2008. Research
at the NHM (2007) received support from the SYNTHESYS Project,
which was nanced by the European Community Research Infra-
structure Action under the FP6 “Structuring the European Research
Area” Programme. A. Magri and G. Damiani supported the realisa-
tion of this paper.
RefeRences
Allsteadt J. & Lang J.W. (1995) - Sexual dimorphism in the
genital morphology of young American alligators, Al-
ligator mississippiensis. Herpetologica, 51: 314-325.
Andersson M. (1994) - Sexual size dimorphism. In: Krebs J.R.
& Clutton-Brock T. (Eds) - Sexual Selection. Mono-
graphs in Behaviour and Ecology: 246-293. Princeton
University Press, Princeton.
Arbour V.M. (2009) - Estimating impact forces of tail club
strikes by ankylosaurid dinosaurs. PLoS ONE, 4:e6738.
doi:10.1371/journal.pone.0006738.
Arbulla D., Cotza F., Cucchi F., Dalla Vecchia F.M., De Giusto
A., Flora O., Masetti D., Palci A., Pittau P., Pugliese N.,
Stenni B., Tarlao A., Tunis G. & Zini L. (2006) - Escur-
sione nel Carso Triestino, in Slovenia e Croazia. 8 giug-
no. Stop 1. La successione Santoniano–Campaniana del
Villaggio del Pescatore (Carso Triestino) nel quale sono
stati rinvenuti i resti di dinosauro. In: Melis R., Romano
R. & Fonda G. (Eds) - Guida alle escursioni/excursions
guide, Società Paleontologica Italiana – Giornate di Pa-
leontologia 2006: 20-27. EUT Edizioni Università di
Trieste, Trieste.
Benton M.J., Juul L., Storrs G.W. & Galton P.M. (2000) -
Anatomy and systematics of the prosauropod dino-
saur Thecodontosaurus antiquus from the Upper Triassic
of southeast England. Journal of Vertebrate Paleontology,
20(1): 77-108.
Benton M.J., Csiki-Sava Z., Grigorescu D., Redelstorff R.,
Sander P.M., Stein K. & Weishampel D.B. (2010) - Di-
nosaurs and the island rule: the dwarfed dinosaurs from
Haţeg Island. Palaeography, Palaeoclimatology, Palaeoecology,
293: 438-454.
Brett-Surman M.K. & Wagner J.R. (2006) - Discussion on
character analysis of the appendicular anatomy in Cam-
panian and Maastrichtian North American hadrosaurids
– variation and ontogeny. In: K. Carpenter (Ed.) - Horns
and beaks – Ceratopsian and ornithopod dinosaurs:
135-169. Indiana University Press, Bloomington and In-
dianapolis.
Brinkman D.B. (2014) - The size-frequency distribution of
hadrosaurs from the Dinosaur Park Formation of Al-
berta, Canada. In: Eberth D.A. & Evans D.C. (Eds) -
Hadrosaurs: 416-421. Indiana University Press, Bloom-
ington and Indianapolis, Indiana.
Brochu C.A. (2003) - Osteology of Tyrannosaurus rex: Insights
from a nearly complete skeleton and high resolution
computed tomographic analysis of the skull. Society of
Vertebrate Paleontology Memoirs, 7: 1-138.
Brown B. (1912) - The osteology of the manus in the Family
Trachodontidae. Bulletin of the American Museum of Natu-
ral history, 31: 105-107.
Brown B. (1913a) - The skeleton of Saurolophus, a crested duck-
billed dinosaur from the Edmonton Cretaceous. Bulletin
of the American Museum of Natural history, 32: 387-392.
Brown B. (1913b) - A new trachodont dinosaur, Hypacrosaurus,
from the Edmonton Cretaceous of Alberta. Bulletin of
the American Museum of Natural history, 32: 395-406.
Brown B. (1916) - Corythosaurus casuarius: Skeleton, muscula-
ture, and epidermis. Bulletin of the American Museum of
Natural history, 35: 709-716.
Buffetaut E. (2009) - An additional hadrosaurid specimen (Di-
nosauria: Ornithischia) from the marine Maastrichtian
deposits of the Maastricht area. Carnets de Géologie, Letter
2009/03 (CG2009_L03): 1-4.
Campione N.E. & Evans D.C. (2011) - Cranial growth and
variation in Edmontosaurus (Dinosauria: Hadrosauri-
dae): Implications for latest Cretaceous megaherbivore
diversity in North America. PLoS ONE, 6(9): e25186.

The tail of Tethyshadros insularis 623
doi:10.1371/journal.pone.0025186.
Campos F., Hernández M., Arizaga J., Miranda R. & Amezcua
A. (2005) - Sex differentiation of Corn Buntings Miliaria
calandra in Northern Spain. Ringing & Migration, 22: 159-
162.
Canudo J.I., Cruzado-Caballero P. & Moreno-Azanza M.
(2005) - Possible theropod predation evidence in hadro-
saurid dinosaurs from the Upper Maastrichtian (Upper
Cretaceous) of Aren (Huesca, Spain). Kaupia, 14: 9-13.
Carpenter K. (1990) - Variation in Tyrannosaurus rex. In: Car-
penter K. & Currie P.J. (Eds) - Dinosaur Systematics:
Perspectives and Approaches: 141-145, Cambridge
Univ. Press, Cambridge, UK.
Carpenter K. (1997) - Ankylosaurs. In: Farlow J.O. & Brett-
Surman M.K. (Eds) - The Complete Dinosaur: 307-316,
Indiana Univ. Press, Bloomington & Indianapolis.
Carpenter K., Miles C., Ostrom J. H. & Cloward C. (2005)
- Redescription of the small maniraptoran theropods
Ornitholestes and Coelurus from the Upper Jurassic Morri-
son Formation of Wyoming. In Carpenter K. (ed.) - The
Carnivorous Dinosaurs: 49-71, Indiana University Press,
Bloomington and Indianapolis.
Casanovas-Cladellas M.L., Pereda-Suberbiola X., Santafé J.V.
&. Weishampel D.B. (1999) - First lambeosaurine had-
rosaurid from Europe: palaeobiogeographical implica-
tions. Geological Magazine, 136: 205-211.
Colbert E.H. (1990) - Variation in Coelophysis bauri. In: Car-
penter K. & Currie P.J. (Eds) - Dinosaur Systematics:
Perspectives and Approaches: 81–90, Cambridge Univ.
Press, Cambridge, UK.
Cong L., Hou L., Wu X. & Hou J. (1998) - The gross anatomy
of Alligator sinensis Fauvel. Forestry Publishing House,
Beijing.
Conti S., Vila B., Sellés A.G., Galobart À., Benton M.J. &
Prieto-Márquez A. (2020) - The oldest lambeosaurine
dinosaur from Europe: insights into the arrival of Tsin-
taosaurini. Cretaceous Research, 107 (104286): 1-15.
Coombs W.P. jr. (1978) - Theoretical Aspects of Cursorial
Adaptations in Dinosaurs. The Quarterly Review of Biology,
53(4): 393-418.
Cope E.D. (1869) - Synopsis of the extinct Batrachia, Reptilia
and Aves of North America. Transactions of the American
Philosophical Society, 14: 1-235.
Coria R.A. & Salgado L. (1996) - A basal Iguanodontian (Or-
nithischia: Ornithopoda) from the Late Cretaceous of
South America. Journal of Vertebrate Palaeontology, 16: 445-
457.
Cruzado-Caballero P. (2012) - Restos directos de dinosaurios
hadrosáuridos (Ornithopoda, Hadrosauridae) del Maas-
trichtiense superior (Cretácico Superior) de Arén (Hues-
ca). Unpublished Ph.D. thesis. Universidad de Zaragoza.
Csiki-Sava Z., Buffetaut E., Ősi A., Pereda-Suberbiola X.
& Brusatte S.L. (2015) - Island life in the Cretaceous
- faunal composition, biogeography, evolution, and ex-
tinction of land-living vertebrates on the Late Creta-
ceous European archipelago. ZooKeys, 469: 1-161. doi:
10.3897/zookeys.469.8439.
Csiki-Sava Z., Vremir M., Vasile S., Brusatte S.L., Dyke G.,
Naish D., Norell M.A. & Totoianu R. (2016) - The East
Side Story - the Transylvanian latest Cretaceous conti-
nental vertebrate record and its implications for under-
standing Cretaceous - Paleogene boundary events. Creta-
ceous Research, 57: 662-698.
Dalla Vecchia f. M. (2006) - Telmatosaurus and the other had-
rosaurids of the Cretaceous European Archipelago. An
overview. Natura Nascosta, 32: 1-55.
Dalla Vecchia F. M. (2008) - I dinosauri del Villaggio del Pes-
catore (Trieste): qualche aggiornamento. Atti Museo Civi-
co di Storia Naturale di Trieste, num. spec.: 111-129.
Dalla Vecchia f.M. (2009a) - European hadrosauroids. Actas
de las IV Jornadas Internacionales sobre Paleontologia de Dino-
saurios y su Entorno, pp. 45-74, Colectivo Arqueológico-
Paleontológico de Salas, Salas de los Infantes.
Dalla Vecchia f.M. (2009b) - Telmatosaurus and the other had-
rosaurids of the Cretaceous European Archipelago. An
update. Natura Nascosta, 39: 1-18.
Dalla Vecchia f.M. (2009c) - Tethyshadros insularis, a new had-
rosauroid dinosaur (Ornithischia) from the Upper Cre-
taceous of Italy. Journal of Vertebrate Paleontology, 29(4):
1100-1116.
Dalla Vecchia F.M. (2014) - An overview of the latest Creta-
ceous hadrosauroid record in Europe. In: Eberth D.A.
& Evans D.C. (Eds) - Hadrosaurs: 268-297. Indiana
University Press, Bloomington, Indiana.
Dalla Vecchia F.M. (2018) - A wing metacarpal from Italy and
its implications for latest Cretaceous pterosaur diversity.
In: Hone D. W. E., Witton M. P. & Martill D. M. (Eds) -
New Perspectives on Pterosaur Palaeobiology. Geological
Society, London, Special Publications, 455: 209-219.
Dalla Vecchia F.M., Gaete R., Riera V., Oms O., Prieto-
Márquez A., Vila B., Garcia Sellés A. & Galobart À.
(2014) - The hadrosauroid record in the Maastrichtian
of the eastern Tremp Syncline (northern Spain). In: Eb-
erth D.A. & Evans D.C. (Eds) - Hadrosaurs: 298-314.
Indiana University Press, Bloomington & Indianapolis,
Indiana.
Delno M., Martin J.E. & Buffetaut E. (2008) - A new species
of Acynodon (Crocodylia) from the Upper Cretaceous
(Santonian–Campanian) of Villaggio del Pescatore, It-
aly, Palaeontology, 51(5): 1091-1106.
Delgado Castro G., Delgado J.D., González J. & Wink M.
(2013) - Sexual size dimorphism in the extreme SW
breeding population of the European Storm Petrel Hy-
drobates pelagicus (Aves: Procellariformes). Vertebrate Zool-
ogy, 63(3): 313-320.
Dilkes D.W. (2000) - Appendicular myology of the hadrosau-
rian dinosaur Maiasaura peeblesorum from the Late Creta-
ceous (Campanian) of Montana. Transactions of the Royal
Society of Edinburgh - Earth Sciences, 90: 87-125.
Dollo L. (1883) - Note sur les restes de dinosauriens recontrés
dans le Crétace supérieur de la Belgique. Bulletin du Musée
Royal d’Histoire Naturelle de Belgique, 2: 205-221.
Duke G.E., Degen A.A. & Reynhout J.K. (1995) - Movement
of urine in the lower colon and cloaca of ostriches. The
Condor, 97: 165-173.
Erickson G.M., Lappin A.K. & Larsson P. (2005) - Androgy-

Fabio Marco Dalla Vecchia
624
nus rex: the utility of chevrons for determining the sex
of crocodilians and non-avian dinosaurs. Zoology, 108:
277-286.
Fairbairn D.J., Blanckenhorn W.U., Szekely T. (2007) - Sex,
size, and gender roles – evolutionary studies of sexual
size dimorphism. Oxford University Press, Oxford
(UK), 266 pp.
Finch W.A. jr. (1965) - The karst landscape of Yucatan. U.S.
Department of Commerce/National Bureau of Stan-
dards/Institute for Applied Technology, 179 pp., Clear-
inghouse, Springeld (VA).
Fondevilla V., Dalla Vecchia F.M., Gaete R., Galobart À.,
Moncunill-Solé B., Köhler M. (2018) - Ontogeny and
taxonomy of the hadrosaur (Dinosauria, Ornithopoda)
remains from Basturs Poble bonebed (late early Maas-
trichtian, Tremp Syncline, Spain). PLoS ONE 13(10):
e0206287. doi.org/10.1371/journal.pone.0206287.
Fondevilla V., Riera V., Vila B., Garcia Sellés A., Dinares-Turell
J., Vicens E., Gaete R., Oms O. & Galobart À. (2019) -
Chronostratigraphic synthesis of the latest Cretaceous
dinosaur turnover in south-western Europe. Earth-Sci-
ence Reviews, 191: 168-189.
Ford T.L. & Martin L.D. (2010) - A semi-aquatic life habit
for Psittacosaurus. In: Ryan M.J., Chinnery-Allgeier B.J. &
Eberth D.A. (Eds) - New Perspectives on Horned Dino-
saurs: The Royal Tyrrell Museum Ceratopsian Sympo-
sium: 328-339, Indiana University Press, Bloomington
& Indianapolis.
Forster C.A. (1990) - The postcranial skeleton of the orni-
thopod dinosaur Tenontosaurus tilletti. Journal of Vertebrate
Paleontology, 10(3): 273-294.
Forster C.A. & Sereno P.C. (1997) - Marginocephalians. In:
Farlow J.O. & Brett-Surman M.K. (Eds) - The Complete
Dinosaur: 317-329, Indiana University Press, Blooming-
ton & Indianapolis.
Fox H. (1977) - The urogenital system of Reptiles. In: Gans C.
& Pough F.H. (Eds), Biology of the Reptilia. 6: 1-157,
Academic Press, New York.
Frey E. (1982) - Der Bau des Bewegungsapparates der Kroko-
dile und seine Funktion bei der aquatischen Fortbewe-
gung. Diploma Thesis, Tübingen, University of Tübin-
gen.
Frey E. (1988) - Anatomie des Körperstammes von Alligator
mississippiensis Daudin (Anatomy of the body stem of
Alligator mississippiensis Daudin). Stuttgarter Beiträge zur
Naturkunde, Series A (Biologie), 424: 1-106.
Frey E., Riess J. & Tarsitano S.F. (1989) - The axial tail muscu-
lature of recent crocodiles and its phyletic implications.
American Zoologist, 29: 857-862.
Gallina P.A. & Otero A. (2009) - Anterior caudal transverse
processes in sauropod dinosaurs: morphological, phylo-
genetic and functional aspects. Ameghiniana, 46: 165-176.
Galton P.M. (1981) - Dryosaurus, a hypsilophodontid dinosaur
from the Upper Jurassic of North America and Africa:
Postcranial skeleton. Paläontologische Zeitschrift, 55: 271-
312.
Galton P.M. (1997) - Comments on sexual dimorphism in the
prosauropod dinosaur Plateosaurus engelhardti (Upper Tri-
assic, Trossingen). Neues Jahrbuch für Geologie und Paläon-
tologie, Monatshefte, 1997(11): 674-682.
Galton P.M. & Upchurch P. (2004) - Prosauropoda. In:
Weishampel D.B., Dodson P. & Osmólska H. (Eds) -
The Dinosauria: 232–258. University of California
Press, Berkeley, California.
Gatesy S.M. (1990) - Caudofemoral musculature and the evo-
lution of theropod locomotion. Paleobiology, 16: 170-186.
Gatesy S.M. (1995) - Functional evolution of the hind limb
and tail from basal theropods to birds. In: Thomason J.J.
(Ed.) - Functional morphology in vertebrate paleontol-
ogy: 219–234. Cambridge University Press, Cambridge.
Gauthier J. (1986) - The origin of birds and the evolution of
ight. In: Padian K. (Ed.) - Saurischian monophyly and
the origin of birds. Memoirs of the California Academy of
Sciences, 8: 1-55.
Gilmore C.W. (1920) - Osteology of the carnivorous Dinosau-
ria in the United States National Museum, with special
reference to the genera Antrodemus (Allosaurus) and Cera-
tosaurus. Bulletin of the U.S. National Museum, 110: 1-154.
Godefroit P., Dong Z.-M., Bultynck P., Li H. & Feng L. (1998)
- Sino-Belgian Cooperation Program “Cretaceous Di-
nosaurs and Mammals from Inner Mongolia. 1. New
Bactrosaurus (Dinosauria: Hadrosauroidea) material from
Iren Dabasu (Inner Mongolia, P. R. China). Bulletin de
l’Institut Royal des Sciences Naturelles de Belgique - Sciences de
la Terre, 68 (suppl.): 3-70.
Hailu Y. & Dodson P. (2004) - Basal Ceratopsia. In: Weisham-
pel D. B., Dodson P. & Osmólska H. (Eds) - The Dino-
sauria: 478–493. University of California Press, Berke-
ley, California.
Herner M. (2009) - Postcranial osteology of Leaellynasaura ami-
cagraphica (Dinosauria: Ornithischia) from the Early Cre-
taceous of Southeastern Australia. Journal of Vertebrate
Paleontology, 29(3, Supplement): 113A.
Hildebrand M. & Goslow G. (1998) - Analysis of Vertebrate
Structure. 5th Edition. John Wiley & Sons, Hoboken.
660 p.
Holtz T.R. jr. (2004) - Tyrannosauroidea. In: Weishampel D.B.,
Dodson P. & Osmólska H. (Eds) - The Dinosauria:
111–136. University of California Press, Berkeley, Cali-
fornia.
Holtz T.R. jr., Molnar R.E. & Currie P.J. (2004) - Basal Tet-
anurae. In: Weishampel D.B., Dodson P. & Osmólska H.
(Eds) - The Dinosauria: 71–110. University of Califor-
nia Press, Berkeley, California.
Hone D.W.E. (2012) - Variation in the tail length of non-avian
dinosaurs. Journal of Vertebrate Paleontology, 32(5): 1082-
1089.
Horner J.R., Weishampel D.B. & Forster C.A. (2004) - Hadro-
sauridae. In: Weishampel D. B., Dodson P. & Osmólska
H. (Eds) - The Dinosauria: 438–463. University of Cali-
fornia Press, Berkeley, California.
Ibiricu L.M., Lamanna M.C. & Lacovara K.J. (2014) - The in-
uence of caudofemoral musculature on the titanosau-
rian (Saurischia: Sauropoda) tail skeleton: morphological
and phylogenetic implications. Historical Biology, 26(4):
454-471.

The tail of Tethyshadros insularis 625
Isles T.E. (2009) - The socio-sexual behaviour of extant ar-
chosaurs: implications for understanding dinosaur be-
haviour. Historical Biology, 21(3-4): 139-214.
Jurkovšek B., Toman M., Ogorelec B., Sribar L, Drobne K.,
Poljak M. & Sribar L. (1996) - Geological map of the
southern part of the Trieste–Komen Plateau. Creta-
ceous and Paleogene carbonate rocks. Scale 1:50.000.
Institut za Geologijo, geotehniko in geoziko, Ljubljana,
Slovenia, 143 pp. and map.
Kuchel L.J. & Franklin C.E. (2000) - Morphology of the clo-
aca in the estuarine crocodile, Crocodylus porosus, and its
plastic response to salinity. Journal of Mor phology, 245:
168-176.
Lambe L.M. (1913) - The manus in a specimen of Trachodon
from the Edmonton Formation of Alberta. The Ottawa
Naturalist, 27(2): 21-26.
Lambe L.M. (1917) - The Cretaceous theropodous dinosaur
Gorgosaurus. Memoir of the Geological Survey of Canada, 100:
1-84.
Lang J. W. (1987) - Crocodilian behaviour: implications for
management. In: Webb G.J.W., Manolis C. & Whitehead
P.J. (Eds) - Wildlife Management: Crocodiles and Alliga-
tors: 273-294. Surrey Beatty, Sydney.
Larson P.L. (1994) - Tyrannosaurus sex. In: Rosenberg G.D. &
Wolberg D.L. (Eds) - Dino Fest. The Paleontological Society
Special Publication, 7: 139-155. The Paleontological Soci-
ety, Knoxville.
Larson P.L. (2008) – Variation and sexual dimorphism in Ty-
rannosaurus rex. In: Carpenter K. & Larson P.L. (Eds)
- Tyrannosaurus rex, The Tyrant King: 103-128. Indiana
University Press, Bloomington.
Larson P.L. & Frey E. (1992) - Sexual dimorphism in the
abundant Upper Cretaceous theropod, Tyrannosaurus rex.
Journal of Vertebrate Paleontology, 12: 38A.
Laurent Y. (2003) - Les faunes de vertébrés continentaux du
Maastrichtien supérieur d’Europe: systematique et bio-
diversité. Strata, 41: 1-81.
Lezana L., Miranda R., Campos F. & Peris S.J. (2000) - Sex dif-
ferentiation in the spotless starling (Sturnus unicolor Tem-
minck, 1820). Belgian Journal of Zoology, 130(2): 139 -142.
López-Martínez N., Canudo J.I., Ardèvol L., Pereda-Suberbio-
la X., Orue-Etxebarria X., et al. (2001) - New dinosaurs
sites correlated with upper Maastrichtian pelagic depos-
its in the Spanish Pyrenees: implications for the dinosaur
extinction pattern in Europe. Cretaceous Research, 22: 41-
61.
Lull R.S. & Wright N.E. (1942) - Hadrosaurian dinosaurs of
North America. Geological Society of America - Special Pa-
pers, 40: 1-242.
Maidment S.C.R. & Barrett P.M. (2014) - Osteological corre-
lates for quadrupedality in ornithischian dinosaurs. Acta
Palaeontologica Polonica, 59(1): 53-70.
Maidment S.C.R., Linton D.H., Upchurch P., Barrett P.M.
(2012) - Limb-Bone Scaling Indicates Diverse Stance
and Gait in Quadrupedal Ornithischian Dinosaurs. PLoS
ONE, 7(5): e36904. doi:10.1371/journal.pone.0036904.
Mallison H. (2011) - Defense capabilities of Kentrosaurus aethi-
opicus Hennig, 1915. Palaeontologia Electronica, 14(2), 10A:
25 pp.
Mallon J.C. (2017) - Recognizing sexual dimorphism in the
fossil record: lessons from nonavian dinosaurs. Palaeobi-
ology, 43(3): 495-507.
Marsh O.C. (1881) - Classication of the Dinosauria. American
Journal of Sciences, 7: 81-86.
Maryánska T. & Osmólska H. (1981) - First lambeosaurine di-
nosaur from the Nemegt Formation, Upper Cretaceous,
Mongolia. Acta Palaeontologica Polonica, 26: 243-255.
Maryánska T. & Osmólska H. (1984) - Postcranial anatomy of
Saurolophus angustirostris with comments on other hadro-
saurs. Palaeontologia Polonica, 46: 119-141.
Matthew W.D. & Brown B. (1923) - Preliminary notices of
skeletons and skulls of Deinodontidae from the Creta-
ceous of Alberta. American Museum Novitates, 89: 1-9.
Maxwell W.D. & Ostrom J.H. (1995) - Taphonomy and palaeo-
biological implications of Tenontosaurus-Deinonychus asso-
ciations. Journal of Vertebrate Palaeontology, 15(4): 707-712.
McDonald A.T., Bird J., Kirkland J.I. & Dodson P. (2012a)
- Osteology of the basal hadrosauroid Eolambia carol-
jonesa (Dinosauria: Ornithopoda) from the Cedar Moun-
tain Formation of Utah. PLoS ONE, 7(10): e45712.
doi:10.1371/journal.pone.0045712.
McDonald A.T., Espilez E., Mampel L., Kirkland J.I. & Alcala
L. (2012b) - An unusual new basal iguanodont (Dino-
sauria: Ornithopoda) from the Lower Cretaceous of
Teruel, Spain. Zootaxa, 3595: 61-76.
McDonald A.T., Gates T.A., Zanno L.E. & Makovicky P.J.
(2017) - Anatomy, taphonomy, and phylogenetic impli-
cations of a new specimen of Eolambia caroljonesa (Dino-
sauria: Ornithopoda) from the Cedar Mountain Forma-
tion, Utah, USA. PLoS ONE, 12(5): e0176896. doi.org/
10.1371/journal.pone.0176896.
McLelland J. (1990) - A colour atlas of avian anatomy. Wolfe
Publishing Ltd., London, 127 pp.
Minnich J.E. (1982) - The use of water. In: Gans C. & Pough
F.H. (Eds) - Biology of the Reptilia. 12: 325-395, Aca-
demic Press, New York.
Moreno K., Carrano M.T. & Snyder R. (2006) - Morphologi-
cal changes in pedal phalanges through ornithopod di-
nosaur evolution: a biomechanical approach. Journal of
Morphology, 268 (1): 50-63.
Musara C., Chamuorwa J.P., Holtug K. & Skadhauge E. (2002)
- Water absorption in relation to fermentation in the co-
lon of the ostrich (Struthio camelus). Onderstepoort - Journal
of Veterinary Research, 69: 315-320.
Myhrvold N.P. & Currie P.J. (1997) - Supersonic sauropods?
Tail dynamics in the diplodocids. Paleobiology, 23(4): 393-
409.
Nopcsa F. (1900) - Dinosaurierreste aus Siebenbürgen (Schädel
von Limnosaurus transsylvanicus nov. gen. et spec.). Denk-
schriften der kaiserlichen Akademie der Wissenschaften in Wien
- Mathematisch-Naturwissenschaftliche Klasse, 68: 555-591.
Nopcsa F. (1915) - Die Dinosaurier der Siebenbürgischen
Landesteile Ungarns. Mitteilungen Jahrbuch des Königliche n
Ungarischen Geologischen Reichsanstalt, 23: 1-24.
Nopcsa f. (1928) - Dinosaurierreste aus Siebenbürgen. IV. Die
wirbelsäule von Rhabdodon und Orthomerus. Palaeontologica

Fabio Marco Dalla Vecchia
626
Hungarica, 1(1921-1923): 273-304.
Norman D.B. (1980) - On the ornithischian dinosaur Iguanodon
bernissartensis from the Lower Cretaceous of Bernissart
(Belgium). Institut Royal des Sciences Naturelles de Belgique –
Mémoires, 178: 1-100.
Norman D.B. (1986) - On the anatomy of Iguanodon athereld-
ensis (Ornithischia: Ornithopoda). Bulletin de l’Institut
Royal des Sciences Naturelles de Belgique - Sciences de la Terre,
56: 281-372.
Norman D.B. (2002). On Asian ornithopods (Dinosauria:
Ornithischia). 4. Probactrosaurus Rozhdestevensky, 1966.
Zoological Journal of the Linnean Society, 136: 113-144.
Norman D.B. (2004) - Basal Iguanodontia. In: Weishampel
D.B., Dodson P. & Osmólska H. (Eds) - The Dinosau-
ria: 413-437. University of California Press, Berkeley,
California.
Norman D.B. (2014) - Iguanodonts from the Wealden of Eng-
land: do they contribute to the discussion concerning
hadrosaur origins? In: Eberth D.A. & Evans D.C. (Eds)
- Hadrosaurs: 10-43. Indiana University Press, Bloom-
ington, Indiana.
Norman D.B. (2015) - On the history, osteology, and system-
atic position of the Wealden (Hastings Group) dinosaur
Hypselospinus ttoni (Iguanodontia: Styracosterna). Zoo-
logical Journal of the Linnean Society, 173: 92-189.
Norman D.B., Witmer L.M. & Weishampel D.B. (2004) - Basal
Thyreophora. In: Weishampel D.B., Dodson P. & Os-
mólska H. (Eds) - The Dinosauria: 335-342. University
of California Press, Berkeley, California.
Organ C.L. (2006) - Thoracic Epaxial Muscles in Living Ar-
chosaurs and Ornithopod Dinosaurs. The Anatomical Re-
cord, 288A: 782-793.
Ostrom J. H. (1963) - Parasaurolophus cyrtocristatus, a crested
hadrosaurian dinosaur from New Mexico. Fieldiana - Ge-
ology, 14: 143-168.
Ostrom J. H. (1964) - A reconsideration of the paleoecology
of hadrosaurian dinosaurs. American Journal of Science,
262: 975-997.
Ostrom J. H. (1969) - Osteology of Deinonychus antirrhopus, an
unusual theropod from the Lower Cretaceous of Mon-
tana. Bulletin of the Peabody Museum of Natural History, 30:
1-165.
Otero A., Gallina P.A. & Herrera Y.L. (2010) - Pelvic mus-
culature and function of Caiman latirostris (Crocodylia,
Alligatoridae). Herpetological Journal, 20: 173-184.
Otero A., Gallina P.A., Canale J.I. & Haluza A. (2012) - Sauro-
pod haemal arches: morphotypes, new classication and
phylogenetic aspects. Historical Biology, 24(3): 243-256.
Palci A. (2003) - Ricostruzione paleoambientale del sito fos-
silifero senoniano del Villaggio del Pescatore (Trieste).
Unpublished PhD thesis, pp.134, Università degli Studi
di Trieste.
Parks W.A. (1920) - The osteology of the trachodont dinosaur
Kritosaurus incurvimanus. University of Toronto Studies - Geo-
logical Series, 11: 1-74.
Parks W.A. (1924) - Dyoplosaurus acutosquameus, a new genus and
species of armored dinosaur; with notes on a skeleton
of Prosaurolophus maximus. University of Toronto Studies -
Geological Series, 18: 1-35.
Paul G. (2010) - The Princeton eld guide to dinosaurs. Princ-
eton University Press, Princeton and Oxford, 320 pp.
Pereda-Suberbiola X., Canudo J.I., Cruzado-Caballero P., Bar-
co J.L., López-Martínez N., Oms O. & Ruiz-Omeñaca
J.I. (2009) - The last hadrosaurid dinosaurs of Europe: a
new lambeosaurine from the uppermost Cretaceous of
Arén (Huesca, Spain). Comptes rendus Palevol, 8: 559-572.
Persons W.S.IV & Currie P.J. (2011a) - The tail of Tyrannosau-
rus: reassessing the size and locomotive importance of
the M. caudofemoralis in non-avian theropods. The Ana-
tomical Record, 294: 119-131.
Persons W.S.IV & Currie P.J. (2011b) - Dinosaur speed demon:
the caudal musculature of Carnotaurus sastrei and impli-
cations for the evolution of South American abelisau-
rids. PLoS ONE, 6(10): e25763. doi:10.1371/journal.
pone.0025763.
Persons W.S.IV & Currie P.J. (2012) - Dragon tails: convergent
caudal morphology in winged archosaurs. Acta Geologica
Sinica, 86(6): 1402-1412.
Persons W.S.IV & Currie P.J. (2014) - Duckbills on the run,
the cursorial abilities of hadrosaurs and implications for
tyrannosaur hunting strategies. In: Eberth D.A. & Evans
D.C. (Eds) - Hadrosaurs: 449-458. Indiana University
Press, Bloomington, Indiana.
Persons W.S.IV, Currie P.J. & Norell M.A. (2014) - Oviraptoro-
saur tail forms and functions. Acta Palaeontologica Polonica,
59 (3): 553-567.
Persons W.S.IV, Funston G.F., Currie P.J. & Norell M.A.
(2015) - A possible instance of sexual dimorphism in
the tails of two oviraptorosaur dinosaurs. Scientic Re-
ports, 5, 9472: 1-4.
Philip J., Floquet M., Platel J.P., Bergerat F., Sandulescu M.,
Baraboshkin E., Amon E. O., Poisson A., Guiraud
R., Vaslet D., Le Nindre Y., Ziegler M., Bouaziz S. &
Guezou J.C. (2000) - Map 16. - Late Maastrichtian (69.5-
65 Ma). In: Dercourt J., Gaetani M., Vrielynck B., Barrier
E., Biju-Duval B., Brunet M.F., Cadet J.P., Crasquin S.
& Sandulescu M. (Eds) - Atlas Peri-Tethys, Palaeogeo-
graphical Maps. CCGM/CGMW, Paris.
Pianka E. (1970) - On r and k selection. The American Natural-
ist, 104: 592-597.
Platt S.G., Rainwater T.R., Thorbjarnarson J.B., Finger A.G.,
Anderson T.A. & McMurry S.T. (2009) - Size estima-
tion, morphometrics, sex ratio, sexual size dimorphism,
and biomass of Morelet’s crocodile in northern Belize.
Caribbean Journal of Science, 45: 80-93.
Prieto-Márquez A. (2001) - Osteology and variation of Brachy-
lophosaurus canadensis (Dinosauria, Hadrosauridae) from
the Upper Cretaceous Judith River Formation of Mon-
tana. Unpublished Master Thesis. Montana State Uni-
versity.
Prieto-Márquez A. (2010) - Global phylogeny of Hadrosau-
ridae (Dinosauria: Ornithischia) using parsimony and
Bayesian methods. Zoological Journal of the Linnean Society-
London, 15: 435-502.
Prieto-Márquez A., Dalla Vecchia F.M., Gaete R. & Galobart
À. (2013) - Diversity, relationships, and biogeography

The tail of Tethyshadros insularis 627
of the Lambeosaurine dinosaurs from the European
archipelago, with description of the new Aralosaurin
Canardia garonnensis. PLoS ONE, 8: e69835. doi:10.1371/
journal.pone.0069835.
Prieto-Márquez A., Erickson G.M. & Ebersole J.A. (2016) - A
primitive hadrosaurid from southeastern North Ameri-
ca and the origin and early evolution of ‘duck-billed’ di-
nosaurs. Journal of Vertebrate Paleontology, 36(2): e1054495.
Prieto-Márquez A., Fondevilla V., Garcia Sellés A., Wagner
J.R. & Galobart À. (2019) - Adynomosaurus arcanus, a
new lambeosaurine dinosaur from the Late Cretaceous
Ibero-Armorican Island of the European archipelago.
Cretaceous Research, 96: 19-37.
Raath M.A. (1990) - Morphological variation in small thero-
pods and its meaning in systematics: Evidence from
Syntarsus rhodesiensis. In: Carpenter K. & Currie P.J. (Eds)
- Dinosaur Systematics: Perspectives and Approaches:
91–105, Cambridge Univ. Press, Cambridge, UK.
Raia P., Barbera C. & Conte M. (2003) - The fast life of a
dwarfed giant. Evolutionary Ecology, 17: 293-312.
Rinehart L.F., Lucas S.G. & Heckert A.B. (2001) - Preliminary
statistical analysis dening the juvenile, robust and grac-
ile forms of the Triassic dinosaur Coelophysis. Journal of
Vertebrate Paleontology, 21: 93A.
Romer A.S. (1923) - Crocodilian pelvic muscles and their avian
and reptilian homologues. Bulletin of the American Museum
of Natural History, 48: 533-552.
Romer A.S. (1927) - The pelvic musculature of the ornithis-
chian dinosaurs. Acta Zoologica, 8: 225-275.
Romer A.S. (1956) - Osteology of Reptiles. 3rd Edition (1976).
University of Chicago Press, Chicago, 772 pp.
Romer A.S. (1977) - Anatomia comparata dei Vertebrati. Pic-
cin Editore, Padova, 539 pp. Italian edition of: The ver-
tebrate body (1949, 1st edition), W.B. Saunders Com-
pany, Philadelphia, 643 pp.
Rothschild B.M. (1994) - Paleopathology and the sexual habits
of dinosaurs, as derived from study of their fossil re-
mains. In: Rosenburg G.D. & Wolberg D.L. (Eds) - Dino
fest, Paleontological Society Special Publication, 7: 275-283.
Russell A.P., Bergmann P.J. & Barbadillo L.J. (2001) - Maximal
caudal autotomy in Podarcis hispanica (Lacertidae): the
caudofemoralis muscle is not sundered. Copeia, 1: 154-
163.
Russell D. (1972) - Ostrich dinosaurs from the Late Creta-
ceous of western Canada. Canadian Journal of Earth Sci-
ences, 9: 375-402.
Sato T., Cheng Y.N., Wu X.C., Zelenitsky D.K. & Hsiao Y.F.
(2005) - A pair of shelled eggs inside a female dinosaur.
Science, 308: 375.
Schwarz-Wings D., Frey E. & Martin T. (2009) - Reconstruc-
tion of the bracing system of the trunk and tail in hy-
posaurine dyrosaurids (Crocodylomorpha; Mesoeucro-
codylia). Journal of Vertebrate Paleontology, 29(2): 453-472.
Schweitzer M.H., Elsey R.M., Dacke C.G., Horner J.R. &
Lamm E.-T. (2007) - Do egg-laying crocodilian (Alliga-
tor mississippiensis) archosaurs form medullary bone? Bone,
40: 1152-1158.
Senter P. (2012) - Forearm orientation in Hadrosauridae (Di-
nosauria: Ornithopoda) and implications for museum
mounts. Palaeontologia Electronica, 15(3), 30A, 10 pp.
Sereno P. (1986) - Phylogeny of the bird-hipped dinosaurs
(Order Ornitischia). National Geographic Research, 2: 234-
256.
Sereno P. (1998) - A rationale for phylogenetic denitions,
with application to higher-level taxonomy of Dinosau-
ria. Neues Jahrbuch für Geologie und Paläontologie, Abhandlun-
gen, 210: 41-83.
Skahdauge E., Warui C.N., Kamau J.M.Z. & Maloiy G.M.O.
(1984) - Function of the lower intestine and osmoregu-
lation in the ostrich: preliminary anatomical and physi-
ological observations. Quarterly Journal of Experimental
Physiology, 69: 809-818.
Steadman D. W., Franz R., Morgan G. S., Albury N. A., Kakuk
B., Broad K., Franz S. E., Tinker K., Pateman M. P., Lott
T. A., Jarzen D. M. & Dilcher D. L. (2007) - Exception-
ally well preserved late Quaternary plant and vertebrate
fossils from a blue hole on Abaco, The Bahamas. Proceed-
ings of the National Academy of Sciences, 104(50): 19897-
19902.
Sternberg C.H. (1926) - A new species of Thespesius from the
Lance Formation of Saskatchewan. Bulletin of the Geologi-
cal Survey of Canada, 44: 73-84.
Suzuki D., Weishampel D.B. & Minoura N. (2004) - Nippono-
saurus sachalinensis (Dinosauria: Ornithopoda): anatomy
and systematic position within Hadrosauridae. Journal of
Vertebrate Paleontology, 24 (1): 145-164.
Tarlao A., Tentor M., Tunis G. & Venturini S. (1994) - Eviden-
ze di una fase tettonica nel Senoniano inferiore dell’area
del Villaggio del Pescatore (Trieste). Gortania-Atti Museo
Friulano di Storia Naturale, 15(1993): 23-34.
Therrien F. (2005) - Palaeoenvironments of the latest Creta-
ceous (Maastrichtian) dinosaurs of Romania: insights
from uvial deposits and paleosols of the Transylvanian
and Haţeg basins. Palaeography, Palaeoclimatology, Palaeoecol-
ogy, 218: 15-56.
Upchurch P., Barrett P.M. & Dodson P. (2004) - Sauropoda.
In: Weishampel D. B., Dodson P. & Osmólska H. (Eds)
- The Dinosauria: 259–322. University of California
Press, Berkeley, California.
Varricchio D.J., Jackson F., Borkowskl J.J. & Horner J.R. (1997)
- Nest and egg clutches of the dinosaur Troodon formosus
and the evolution of avian reproductive traits. Nature,
385: 247-250.
Venturini S., Tentor M. & Tunis G. (2008) - Episodi conti-
nentali edulcicoli ed eventi biostratigraci nella sezione
campaniano-maastrichtiana di Cotici (M.te San Michele,
Gorizia). Natura Nascosta, 36: 6-23.
Wang R.-F., You H.-L., Wang S.-Z., Xu S.-C., Yi J., Xie L.-J., Jia
L. & Xing H. (2015) - A second hadrosauroid dinosaur
from the early Late Cretaceous of Zuoyun, Shanxi Prov-
ince, China. Historical Biology, 29: 17-24.
Wellnhofer P. (1994) - Ein Dinosaurier (Hadrosauridae) aus
der Oberkreide (Maastricht, Helvetikum-Zone) des
bayerischen Alpenvorlandes. Mitteilungen der Bayerischen
Staatssammlung für Paläontologie und historisches Geologie, 34:
221-238.

Fabio Marco Dalla Vecchia
628
Weishampel D.B., Jianu C.M., Csiki Z. & Norman D.B. (2003)
- Osteology and phylogeny of Zalmoxes (n. g.), an un-
usual Euornithopod dinosaur from the latest Cretaceous
of Romania. Journal of Systematic Palaeontology, 1: 65-123.
Wild R. (1973) - Die Triasfauna der Tessiner Kalkalpen. XXIII.
Tanystropheus longobardicus (Bassani) (Neue Ergebnisse).
Schweizerische Paläontologische Abhandlungen, 95: 1-162.
Wilhite R. (2003) - Biomechanical reconstruction of the ap-
pendicular skeleton in three North American Jurassic
Sauropods. PhD dissertation, Eunice, Louisiana State
University.
Williston S.W. & Gregory W.K. (1925) - The osteology of the
reptiles. Harvard University Press, Cambridge, 300 pp.
Witmer L.M. (1995) - The Extant Phylogenetic Bracket and
the importance of reconstructing soft tissues in fossils.
In: Thomason J.J. (Ed.) - Functional Morphology in
Vertebrate Paleontology: 19-33. Cambridge University
Press, New York.
Xing H., Prieto-Márquez A., Gu W. & Yu T-X. (2012) - Re-
evaluation and phylogenetic analysis of the hadrosau-
rine dinosaur Wulagasaurus dongi from the Maastrichtian
of northeast China. Vertebrata PalAsiatica, 50: 160-169.
Xing H., Wang D., Han F., Sullivan C., Ma Q., He Y., Hone
D.W.E., Yan R., Du F. & Xu X. (2014) - A New Basal
Hadrosauroid Dinosaur (Dinosauria: Ornithopoda)
with Transitional Features from the Late Cretaceous
of Henan Province, China. PLoS ONE, 9(6): e98821.
doi:10.1371/journal.pone.0098821.
Xu X., Zhao X.-J., Lü J.-C., Huang W.-B., Li Z.-Y. & Dong
Z.-M. (2000) - A new iguanodontian from Sangping
Formation of Nexiang, Henan and its stratigraphical
implication. Vertebrata PalAsiatica, 38: 176-191.
Xu X., Tan Q., Gao Y., Bao Z., Yin Z., Guo B., Wang J., Tan
L., Zhang Y. & Xing H. (2018) - A large-sized basal an-
kylopollexian from East Asia, shedding light on early
biogeographic history of Iguanodontia. Science Bulletin,
63: 556-563.
Yang T.-R., Engler T., Lallensack J.N., Samathi A., Makowska
M., Schillinger B. (2019) - Hatching asynchrony in ovi-
raptorid dinosaurs sheds light on their unique nesting
biology. Integrative Organismal Biology, 1(1): obz030, doi.
org/10.1093/iob/obz030
You H., Li D. & Liu W. (2011) - A New Hadrosauriform di-
nosaur from the Early Cretaceous of Gansu Province,
China. Acta Geologica Sinica, 85: 51-57.
Zelenitsky D.K. (2006) - Reproductive traits of non-avian the-
ropods. Journal of the Paleontological Society of Korea, 22(1):
209-216.
Zelenitsky D.K. & Hills L.V. (1997) - Normal and pathological
eggshells of Spheroolithus albertensis, oosp. nov., from the
Oldman Formation (Judith River Group, Late Campan-
ian), southern Alberta. Journal of Vertebrate Paleontology,
17(1): 167-171.
Ziegler T. & Olbort S. (2007) - Genital structures and sex iden-
tication in crocodiles. Crocodile Specialist Group Newslet-
ter, 26: 16-17.