National Library Bibliothèque nationale .+. of Canada du Canada Acquisitions and Direction des acquisitions et Bibliographie Services Branch des services bibliographiques 395 Wellinglon Street 395. rue Wellington Ottawa, Ontario Ottawa (Onlario) K1A ON4 K1A ON4
NOTICE AVIS
The quality of this microform is la qualité de cette microforme heavily dependent upon the dépend grandement de la qualité quality of the original thesis Je la thèse soumise au submitted for microfilming. microfilmage. Nous avons tout Every effort has been made to fait pour assurer une qualité ensure the highest quality of supérieure de reproduction. reproduction possible.
If pages are missing, contact the S'il manque des pages, veuillez university which granted the communiquer avec l'université degree. qui a conféré le grade.
Some pages may have indistinct La qualité d'impression de print especially if the original certaines pages peut laisser à pages were typed with a poor désirer, surtout si les pages typewriter ribbon or if the originales ont été university sent us an inferior dactylographiées à l'aide d'un photocopy. ruban usé ou si l'université nous a fait parvenir une photocopie de qualité inférieure.
Reproduction in full or in part of La reproduction, même partielle, this microform is governed by de cette microforme est soumise the Canadian Copyright Act, à la Loi canadienne sur le droit R.S.C. 1970, c. C-30, and d'auteur, SRC 1970, c. C-30, et subsequent amendments. ses amendements subséquents.
CaTI, ad···a • Functional Morphology and Phylogeny
of Keichousaurus hui
(Sauropterygia, Reptilia)
Kebang Lin
Department of Biology
McGill University, Montreal • July, 1994
A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfilment
of the requirements for the degree of Doctor of Philosophy.
Our hie Noue ffJl6,-,nctt
THE AUTHOR RAS GRANTED AN L'AUTEUR A ACCORDE UNE LICENCE IRREVOCABLE NON-EXCLUSIVE IRREVOCABLE ET NON EXCLUSIVE LICENCE ALLOWING THE NATIONAL PERMETTANT A LA BffiLIOTHEQUE LffiRARY OF CANADA 1'0 NATIONALE DU CANADA DE REPRODUCE, LOAN, DISTRmUTE OR REPRODUIRE, PRETER, DISTRIBUER SELL COPIES OF mS/HER THESIS BY OU VENDRE DES COPIES DE SA ANY MEANS AND IN ANY FORM OR THESE DE QUELQUE MANIERE ET FORMAT, MAKING TffiS THESIS SOUS QUELQUE FORME QUE CE SOlT AVAILABLE 1'0 INTERESTED POUR METTRE DES EXEMPLAIRES DE PERSONS. CETTE THESE A LA DISPOSITION DES PERSONNE INTERESSEES.
THE AUTHOR RETAlNS OWNERSlllP L'AUTEUR CONSERVE LA PROPRIETE OF THE COPYRIGHT IN mS/HER DU DROIT D'AUTEUR QUI PROTEGE THESIS. NEITHER THE THESIS NOR SA THESE. NI LA THESE NI DES SUBSTANTIAL EXTRACTS FROM 11' EXTRAITS SUBSTANTIELS DE CELLE MAY BE PRINTED OR OTHERWISE CI NE DOIVENT ETRE IMPRIMES OU REPRODUCED WITHOUT mS/HER AUTREMENT REPRODUITS SANS SON PERMISSION. AUTORISATION,
ISBN 0-612-00110-5
Canadti • PREFACE
This thesis is a description of the osteology, ontogeny and functional a.Jatomy
of the Middle Triassic Chinese pachypleurosaurid reptile Keichousaurus hui. The
preliminary description published soon after its discovery (Young, 1958) was sketchy
in nature and based on a small number of incomplete1y prepared specimens. More
thorough preparation of sorne of the original specimens, as weil as further discoveries
and preparations of new specimens from the same locality, made it possible for me to
completely ilIustrate and redescribe the species. An embryo of Keichousaurus hui was
described for the tirst time. The ontogenetic changes of Keichousaurus hui from
embryo, through juvenile, subadult to adult stages were detailed. A functional analysis • concerning the locomotion pattern of Keichousaurus hui was carried out, and although literature is used to obtain background data, the conclusions drawn were original unless
explicitly stated otherwise. The musculature reconstructions of the locomotive
apparatus of Keichousaurus hui were presented, also for the tirst time. After a review
of the literature, original conclusions are given. It is hcped that this study will provide
a more complete picture of morphology, life history and phylogeny of Keichousaurus
hui.
The text of this thesis was prepared for submission to the Philosophical
Transactions ofthe Royal Society ofLondon B., which has published several papers on
nothosaurs, and follows the format of that journal.
• j • UNIQUE CONTRIBUTIONS
This study is the tirst detailed description of skeletal morphology of
Keichousaurus hui Young, 1958, the Chinese member of the Family
Pachypleurosauridae of diapsid reptiles.
Sexual dimorphism of KeichousGurus hui was described.
The growth pattern of Keichousaurus hui was studied by comparing the
specimens of different developmental stages. Both morphologica! changes and the
proportional changes were described. Statistical procedures were used in the study of
the proportional changes. • Functional analysis was carried Oùt to detelmine the functions of limbs, the neck and the tail in the aquatic locomotion of Keichousaurus hui. 1t was shown that, unlike
the European pachypleurosaurs, Keichousaurlls hui was an underwater rowing swimmer,
using the foreJimbs for propulsion.
Musculature reconstruction of the limbs was attempted, based on the marks on
the bones and the proposed swimming pattern of the animal.
A phylogenetic analysis was also carried out.
This study documents the diversity both in the morphology and in the
locomotion pattern among pachypleurosaw·ids.
• ii • TABLE OF CONTENTS
Preface .
Umque·Contn'b'utlOns Il..
Table of Contents iii
• LIst· 0 fF'Igures...... Vil..
Listing of Tables ix
Abstract x
Résumé xi
Acknowledgements xiii .) iii • Introduction . Systematic Paleontology ...... 6
Description . II
Axial skeleton . II
Skull . II
Lower Jaw . 23
Tooth Implantation ...... 24
Vertebrae and Ribs ...... 25
Gastralia . 36
Appendicular Skeleton 37 • Pectoral Girdle 37 Humerus . 41
Dina and radius . 41
Carpus and Manus . 42
Pelvic Girdle . 46
Femur . 49
Tibia and fibula ...... 51
Tarsus and Pes . 51
Measurements and Proportions 53 • iv • Sexual Dimorphism o...... 57 Ontogenj o...... 61
Description of GXD835û02 o...... 61
The developmental stage of GXD835002 o...... 64
The ontogenetic stage of V7917 o...... 67
Subadult specimens o...... 69
The adults ...... " 69
Allometric growth of Keichousaurus hui 71
Locomotion . 76
Function of the forelimbs of Keichousaurus hui o •••••••••••••••• 78 • Reconstruction of the muscular system of the shoulder girdle 84 Hindlimb and the pelvic girdle o •••••••••••••••••••••••••••• 89
Muscle reconstruction of pelvic girdle and hindlimbs . 91
The tail . 93
The neck o ••••••••••••••••••••••••••••••••••••••••••• 95
The swinuning pattern of Keichousaurus o ••••••••••••••••••••• 96
Phylogenetic Analysis o...... 98
Conclusions o...... 112 • v • References 114 Abbreviations 122
•
• VI • LIST OF FIGURES
Figure 1 Location of Keichousaurus hui. 8
Figure 2 Reconstruction of Keichousaurus hui. 12
Figure 3 Reconstruction of the skull of Keichousaurus hui. 13
Figure 4 The skull of Keichousaurus hui. 14
Figure 5 Palatal view of Keichousaurus hui skull, GXD838028. 22
Figure 6 Dorsal aspect of Keichousaurus hui, GXD7613. 26
Figure 7 Ventral aspect of Keichousaurus hui, GXD838028. 27
Figure 8 The atlas-axis complex of Keichousaurus hui. 30 • Figure 9 The accessory articulation complex of Keichousaurus hui. 34 Figure 10 Pectoral girdle of Keichousaurus hui. 38
Figure II Humerus of Keichousaurus hui. 43
Figure 12 Lower arm of Keichousaurus hui. 44
Figure 13 Pelvic girdle of Keichousaurus 47
Figure 14 Hindlimb of Keichousaurus hui. 50
Figure 15 An embryo of Keichousaurlls hui, GXD835002. 62
Figure 16 A juvenile specimen of Keichousaurus hui, V7917. 68
Figure 17 Swimming speeds of modern aquatic vertebrates. 81
Figure 18 Cladogram of sauropterygians. 110
• vii • LISTING OF TABLES
Table 1 Measurements of Keichousaurus hui 54
Table Il Proportions of measurements of Keichousaurus hui 56
Table III AHometric coefficients 74
Table IV Character states of sauropterygians 109
•
• viii • ABSTRACT
Keichousaurus hui Young, 1958, from the Middle Triassic of Guizhou, China
is a small sauropterygian reptile. It has short snout and elongated temporal openings
resembling the European pachypleurosaurid Dactylosaurus. Unlike ail the other
sauropterygians, the parietal opening is anteriorly positioned. The neck is long and
flexible. The body is rigid and the bones are pachyostotic. It has two or threc sacral
vertebrae. The most striking feature of Keichousaurus is its broad ulna. The entire
forelimb has the outline of a paddle or an oar, and may have been functioned like one.
There is noticeable sexual dimorphism, as is the case for Alpine pachypleurosaurids. • The growth of the humerus is highly positive allometric, indicating an important role of the forelimb in locomotion. The horizontal orientation ofthe pectoral girdle indicate
that Keichousaurus, as well as other pachypleurosaurids, was not a subaquatic flyer.
Instead, a drag-based regime was followed in locomotion. The symmetrical rowing of
the forelimbs precludes lateral undulatory movement of the body. However, vertical
undulation is theoretically possible. The reassessment of the phylogenetic position of
Keichousaurus hui confirmed that it is a member of the monophyletic group
Pachypleurosauroidea.
• ix • RÉSUMÉ
Keichousaurus hui Young est un petit reptile sauropterygien qui a été découvert
à Guizhou en Chine, dans des roches datant du Trias moyen. Il ressemble au
Dactylosaurus, un pachypleurosauridé européen, par son museau court et ses ouvertures
temporales allongées. Contrairement aux autres sauropterygiens, l'ouverture pariétale
est située antérieurement. Son êou est long et flexible et son corps est rigide et son
squelette est pachyostique. Ses vertèbres sacrées sont au nombre de deux ou trois. La
caractéristique la plus étonnante du Keichousaurus est son cubitus très élargi. Ses
membres antérieurs montrent un pourtour qui évoque celui d'une rame ou d'une • paggaie, et ils fonctionnaient peut-être ainsi. Comme dans le cas des pachypleurosauridé s de Alpes, nous notons un important dimorphisme sexuel. La
croissance de l'humérus présente une importante allométrie positive indiquant le rôle
important des membres antérieurs dans la locomotion. La position horizontale de la
ceinture pectorale du Keichousaurus, tout comme chez les autres pachypleurosauridé s,
indique qu'il n'était pas un "volant subaquatique". Au contraire, la locomotion était
largement influencée par la friction. La symétrie latérale du mouvement natatoire des
membres antérieurs exclue toute possibilité de mouvements ondulatoires latéraux du
corps. Cependant il serait théoriquement possible qu'il pouvait se déplacer par
mouvements ondulatoires verticaux. La réévaluation de la position phylogénétique du
• x Keichousaurus hui confirme qu'i! fait partie du groupe monophylétique de • Pachypleurosauroidea.
•
• xi • ACKNOWLEDGEMENTS
lt is a great pleasure to acknowledge the great help and support of Dr. Robert
1. Carroll of the Departrnent of Biology, McGill University, without whose
encouragement and patience this study could not have been fini shed. 1 would like to
express my appreciation to the other members ofmy dissertation committee -- Drs. M.J.
Dunbar, D.M. Green, V. Pasztor, and R. Peters -- for their time and effort. 1would also
like to thank Professors Mingzhen Zhou, Meeman Zhang, AHin Sun and Zhiming Dong,
of the Institute of Vertebrate Paleontology and Palaeoanthropology (IVPP), Chinese
Academy of Sciences, Beijing, China, for their encouragements and suggestions. • The specimens of this study were loaned to me by the IVPP and the Guizhou Museum, Guizhou, China. 1 would especially like to thank Mr. Huiyang Cai of the
Guizhou Museum to assist me in my field trip and allow me to study the specimens
under his care.
1have benefitted greatly from conversations with Drs. H.-D. Sues (Royal Ontario
Museum, Toronto), O. Rieppel (Field Museum of Natural History, Chicago) and G.
Storrs (University of Bristol, UK) on the functional anatomy and phylogeny of
sauropterygians. The numerous discussions with Drs. Xiaochun Wu (Royal Ontario
Museum), Robert Holmes (McGill University), Stephen Godfrey (Royal Tyrrell
Museum, Drurnheller, Alberta), Denis Walsh and Mrs. Michael Debraga and Edward
• xii Hitchcock were also very stimulating. • The French translation of the abstract was kindly provided by Dr. Michel Di Vergilio.
Figures 4a, 6, 7 and 15 were drawn by Mrs. Pamela Gaskill. 1 am grateful to
her effort as weil as her initial guidance in drafting the figures.
1 have received considerable assistance from the staff at the Redpath Museum,
McGill University, especially Misses. Delise Alison, Ingrid Birker, Marie La Ricca and
Barbara Lawson.
Financial assistance was provided by Natural Sciences and Engineering Research
Council of Canada grants to Dr. R.L. Carroll, the Clifford Wong fellowship
administered by the Faculty of Graduate Studies, McGill University, and teaching • assistantship to the author from the Department of Biology, McGill University.
• xiii 1 INTRODUCTION
In the history of tetrapod evolution, reinvasion of the aquatic environment
occurred time after time. ln every tetrapod class - the amphibians, reptiles, birds and
mammals - there have been lineages that adapted themselves to a life in the water.
The Mesozoic sauropterygians is one of such lineages. Claudiosaurus, a
primitive diapsid from the Upper Permian of Madagascar, shares with the
sauropterygians a series ofderived characters, while maintaining primitive diapsid body
proportions and limb structure (Carroll, 1981). Although it is hard to prove that
Claudiosaurus is the ancestor of sauropterygians, it has been shown that Claudiosaurus • is the most closely related sister-group of sauropterygians amongst known primitive diapsids (Sues, 1987; Rieppel, 1989; Storrs, 1991; see Rieppel, 1993 for a different
opinion).
Three major morphological groups are traditionally recognized among the
sauropterygians. Nothosaurs of the Triassic were small to medium in size (20 cm to
3 meters) and may have lived in a lagoonal or continental shelf environment. Their
limb structure, although modified somewhat from the primitive condition, is still
comparable to terrestrial forms. Plesiosaurs, a prominent element in Jurassic and
Cretaceous open seas, were larger in size, up to 15 meters in length. They were highly
specialized in body proportions and structures, apparently weil adapted to the open sea
• 1 environment. Their trunk region was rigid. The limb girdles and limbs were massive,
1 and the manus and pes were elongated through hyperdactyly. Placodonts are a group
of bizarre aquatic animais which are often associated with sauropterygians. Their
massive and weil consolidated skull, crushing teeth and heavy limbs indicate a bottom
dwel1ing, mollusc-eating habitat. The distribution of placodonts is restricted to the
Middle and Upper Triassic. The affinity of placodonts with other sauropterygians is
still problematic.
The monophyly ofthe Sauropterygia is supported by a series of synapomorphies,
including the loss of the lower temporal bar and the lacrimal, and the scapula being
partially lateral to the clavicle, and the interclavicle ventral to the clavicles. Of the
thret: morphological groups, plesiosaurs and placodonts are almost certainly
monophyletic groups. The synapomorphies of plesiosaurs include massive and highly • modified limbs and limb girdles and large size. Nothosaurs, however, are defined mainly by plesiomorphic characters, such as small to medium size and limbs not as
highly specialized as in plesiosaurs. There are few possible synapomorphies proposed
to define the traditional Nothosauria, which are functionally significant and structurally
complex, such as closed interpterygoid vacuity (Carroll & Gaskill, 1985; Sues, 1987)
and closed occiput region. However, recent phylogenetic studies shows that there are
other characters indicating that nothosaurs might he paraphyletic (Rieppel, 1989; Storrs,
1991, 1993). Even so, nothosaurs represent a distinctive level of adaptation and in the
discussion of functional anatomy, the name "nothosaur" is still used, which means
"sauropterygians minus placodonts and plesiosaurs." Stratigraphically, nothosaurs are • 2 never found above the Norian-Rhaetioo boundary of Upper Triassic, while typical • plesiosaurs are never found below it. Among "nothosaurs", two morphological groups are traditionally recognized en
the basis of size and skull proportions: nothosaurids and pachypleurosaurids (again, of
uncertain taxonomical significance). Nothosaurids are larger of the two (adult size
usually longer than 2 meters). Their skull is relatively larger compared to their body,
and the upper temporal openings on the skull table are larger than the orbits. If
plesiosaurs originated from among nothosaurids, which might weil be the case, this
group would be paraphyletic.
Pachypleurosaurids, on the other hand, are smaller in size. They have a small
skull and their upper temporal openings are usually smaller than the orbits. Sorne
authors think that small temporal openings are plesiomorphic and hence cannot be used • to define pachypleurosaurids as a natural group. However, comparison shows that the temporal openings of pachypleurosaurids are either much smaller or have different
shape that those of more primitive diapsids. Another unique feature of
pachypleurosaurids is the structure of the quadrate which probably supported a
tympanum. Associated with the quadrate is the quadrate fossa behind the articular facet
in the retroarticular process which might have participated in forming the middle ear
chamber. The smaller temporal openings, the quadrate and the quadrate fossa in
retroarticular process are synapomorphies that define pachypleurosaurids as a
monophyletic group (Rieppel, personal communication).
Fossils of pachypleurosaurids are found mainly in the European Alps, ranging
• 3 from late Early to late Middle Triassic, along the north rim of the Tethys ocean. By • far the largest number ofspecimens have been found in Mont San Giorgio, Switzerland. Four species of pachypleurosaurids are recognized there: Serpianosaurus mirigiolensis
(Rieppel, 1989), Neusticosaurus pusil1us, Neusticosaurus peyeri (Sander, 1989) and
Pachypleurosaurus edwardsii (Carroll and Gaskill, 1985). Sander (1989) suggested that
the last species be recombined as Neusticosaurus edwardsii, based on the result of
cladistical analysis that shows that it is positioned within the monophyletic group
known as Neusticosaurus. Sander's suggestion is accepted here.
Study of European pachypleurosaurid fossils goes back to the 1850's. A
summary of the early studies of the group can be found in Rieppel (1987).
In 1957, a field team from the Museum of Geology, Chinese Ministry of
Geology, led by C.C. Hu, collected a number of primitive sauropterygian fossils in the • Middle Triassic deposits of the province of Guizhou, China. These sediments belong to the same geological realm as the European Alps, the Tethys. The specimens were
sent to the Institute of Vertebrate Paleontology, Chinese Academy of Sciences to be
studied. A preliminary description was made by C.C. Young (1958) ofthese and other
specimens collected by T.T. Ts'ao of Guizhou Museum on behalf ofthe Institute. They
were named Keichousaurus hui. Because the specimens were not adequately prepared
at the time and the illustrations were crude, later workers have had great difficulty in
comparing them with the European genera.
During the past 30 years, more materia1 was collected from the same 10ca1ity.
Most of the specimens can he included in Keichousaurus hui based on genera1 body • 4 proportions and the configurations of the individual bones. Those new specimens • provide with us a more adequate basis to study the anatomy and ecology of this animal and its phylogenetic affiliation with other nothosaurs.
This paper is based on sorne ofthe new specimens collected by Mr. Huiyang Cai
of Guizhou Museum and by the author in 1984, as weil as the specimens studied by
C.C. Young (1958) now in the collection ofthe Institute ofVertebrate Paleontology and
Paleoanthropology (IVPP), Chinese Academy, Beijing. The purpose of this paper is to
give a full description of the osteology of Keichousaurus hui, so that a detailed
comparison with other pachypleurosaurids and nothosaurids is possible. This paper will
also discuss the functional morphology of the animal and its phylogenetic position
within sauropterygians. •
• 5 • SYSTEMATIC PALEONTOLOGY
DIAPSIDA Osborn, 1903
NEODIAPSIDA Benton, 1985
LEPIDOSAUROMORPHA Benton, 1985
SAUROPTERYGIA Owen, 1860
PACHYPLEUROSAURlA Sanz, 1980
Keichousaurus Young, 1958
• Type species Keichousaurus hui Young, 1958.
Diagnosis Small to medium-sized pachypleurosaur. There are 25-26 cervical and 18
19 dorsal vertebrae. The cervical region is longer than the trunk region. The rostrum
is short and blunt. The upper temporal openings are elongated, only slightly shorter
than the orbit. The parietal opening is anteriorly positioned, and tends to close up in
maturity. The humerus is stronger and longer than the femur. The ulna is broad. Two
or three sacral vertebrae. Slight hyperphalangy.
Distribution Middle Triassic. Southwestern China.
• 6 • Keichousaurus hui Young, 1958
Type V952 in IVPP.
Diagnosis Same as for genus.
Distribution Middle Triassic, Ladinian. Mount Langmu, Dazhai Village, Dingxiao,
Xingyi County, Guizhou Province, China. The locality is about 200 km
south-southwest of Guiyang, the capitol city of Guizhou Province (Figure 1).
Referred material The specimens whose numbers have the prefix GXD are from the • collection of Guizhou Province Museum, those with prefix V are from the Institute of Vertebrate Paleontology and Palaeoanthropology, Beijing, and the one with BPV is
from the Beijing Natural (History) Museum.
(i). V952. The type specimen of Keichousaurus hui designated by Young(l958); an
adult skull in dorsal view with 21 cervical vertebrae.
(ii). GXD7601. An extremely weil preserved skull in dorsal view with 14 cervical
vertebrae. Judged by its size and suture, this specimen represents a somewhat earlier
development stage than V952.
(iii). GXD7613. An adult individual in dorsal view. The skull and the first six or
seven cervicals are missing, as is the tip of the tail. The remainder part of the skeleton • 7 •
Figure 1 Location of Keichousaurus hui.
• 8 • •••
l------=~-+_---_l_--r_I 26"
GUIZHOU PROVo
l------.I--+------1---ri 26"
• lOS" '10~- including the rear portion of the right lower jaw, is extremely weil preserved. • (iv). GXD7621. A weil preserved individual in dorsal view. The skull, anterior cervicals, left epipodial of forelimb, left hindlimb and most part of tail are missing.
(v). GXD7603. This specimen is smaller, presumably YC:llIger, than the last two. It
is exposed in dorsal view. Left hindlimb and tail are missing. The skull is not weil
preserved.
(vi). GXD7602. A young individual exposed in dorsal view. Left hand, left hindlimb
and tail are missing. The right hindlimb is very weil preserved.
(vii). GXD838028. A fully grown individual in ventral view. The rear portion of the
skull is preserved. The size of the skull is comparable to that of V952.
(viii). V953. A young specimen in ventral view. Only the right limbs are weil
preserved. • (ix). V7919. A young specimen in ventral view. Only part of the tail is missing. (The pectoral girdle was originally buried in very hard matrix. Dental drill and
compressed-air-powered vibrating tools were used to remove this matrix. The work
was rewarded by uncovering the clavicles that are superficial to the scapulae, a reverse
of the relationship of the two bones in other sauropterygians.)
(x). GXD835002. An embryo in dorsal view. This specimen has a relatively large
skull, small limbs and a short tail (this is the only specimen with a complete tail).
(xi). V7917. A very young specimen is dorsal view. This specimen is vertually
complete from the snout to the tip of the tai!.
• 9 The specimens are embedded in thin layers of grey pelitic limestone. Most of • the specimens were preserved dorsoventrally flattened. The limbs usually lie close to the body. The forelimbs are kept more or less straight, while the epipodials of the
hindlimbs bend towards the base of tai!. Post-mortem disturbance appears to be
minimal. Only the manus and pes are affected.
Most ofthe specimens were prepared under a Wild M7 microscope with pin vice
and fine needles.
•
• 10 • DESCRIPTION
Because all ofthe specimens were flattened during preservation, only one plane
could be examined of any given individual. The reconstructions were based on all
available material (Figure 2, Figure 3).
Axial skeleton • Skull Description of the dorsal aspect of the skull of Keichousaurlls is based mainly
on GXD7601 and V952 (Figure 4). There is only one specimen (GXD838028) in
which the rear part of the palatal elements can be discerned (Figure 5).
Among known pachypleurosaurids, the general proportions of the skull of
Keichousaurlls are closest to those of Dactylosaurus (Sues and Carroll, 1985). The
snout is short and broad. The orbits are large, situated at the middle one-third of the
skull. The external nares are situated at the middle one-third of the antorbital region
of the skull. The skull is widest at the mid-point of the lower rim of the orbit. As in
Dactylosaurus, the upper temporal openings are long and narrow, extending to the rear • 11 •
Figure 2 Ske1etal reconstruction of Keichousaurus hui.
scale xO.85
• 12 • 1 il ,;,,'1 l' \ "l' ,\ r; 1 l' , ,,;..i 1" 1't' 1\\ 'V"'i \ 1 1\ l" ;\~l'" j"1~
l . •
.' •
Figure 3 Reconstruction of the skull of Keichousaurus hui.
3. dorsal view b. palatal view c. lateral view d. occipital view
scale x4
• 13 pm •
J epl pl po
pl po
c
po q
d q • •
Figure 4 Skull of Keichousaurus hui, dorsal view.
a. GXD7601 b. V952
scale x4
• 14 •
E Co
ltl
• end of the skull table, while in other pachypleurosaurids they are round to "key hole" • shaped. The cheek emargination, resuiting from the loss of the lower temporal bar as in ail other sauropterygians, is more pronounced than other pachypleurosaurids except
Dactylosaurus. The skull is quite shallow, less than 30% of the skull width, with the
highest point at the rear end of the skull table.
The premaxilla has a long medial process that extends back to make contact with
the frontal. Unlike other pachypleurosaurids, the anterior part of the premaxilla is short
and broad, so that the snout anterior to the external nares is wider than long. The
lateral process of the premaxilla meet3 the maxilla at the mid-point of the outer rim of
the external nares; the sigmoid suture l'uns anteroventrally. The surface of the bone
is irregularly pitted. Where the number ofteeth can be discerned, there are five conical
teeth in each premaxilla. • The very short and triangular nasals are situated in shallow grooves in the prefrontaJ. They are separated by the posterior process of the premaxilla so that they
do not touch their counterpart on the other side. The posterior end of the nasal does
not extend beyond that ofthe premaxilla. In Neusticosaurus edwardsi, Serpianosaurus
mirigiolensis and Neusticosaurus peyeri, the nasals are large, extending weil back ofthe
anterior rim ofthe orbits (Carroll and Gaskill, 1985; Rieppel, 1989; Sander, 1989). The
skull configuration of Neuslicosaurus pusillus varies considerably. Nevertheless, the
nasals are always longer than those in Keichousaurus (Sander, 1989). Whether
plesiosaurs have nasals is not certain due to the pOOl' preservation of the snout region.
In Pistosaurus, which shows a mosaic of plesiosaur and nothosaur characters (Sues, • 15 1987), the nasa1s are reduced to a splint. They lie far back of the external nares. • In ail the adult specimens, the frontals are fused along the midline without any trace of suture. The anterior half of the bone is very narrow and forros the septum
between the orbits. It is gradually broadened posteriorly. The anterior end of the
frontal is partially covered by the prefrontals on both sides. The cross-section of the
bone between the orbits is trhmgular. The posterior wings of the frontal are grooved
dorsolaterally to accommodate the postfronta1s. The suture with the parietal curves
forward at the midlineso that the parietal reached the level of the posterior rim of the
orbits.
The parietals are also fused at the midline. The skull table is flat. The lateral
flanges of the parietal extend down the medial side of the upper temporal opening,
presumably to provide more space for the attachment ofthe jaw adductor musculature. • This feature is seen in Youngina, Claudiosaurus and other advanced diapsids (Carroll, 1981), but not in primitive diapsids or in Serpianosaurus and Neusticosaurus (Carroll
and Gaskill, 1985; Rieppel, 1989; Sander, 1989). Thus, although the sku1l table of
Keichousaurus hui is narrower than that of European pachyp1eurosaurids, the parietal
itself is not necessarily so. The parietal opening is smaller and situated farther forward
than in ail other pachypleurosaurids, right between the posterior tips of the frontal. A
fon>vard situated parietal opening is thought to be a typica1 plesiosaur character. In
Pistosaurus, the parietal opening is also anteriorly placed (Sues, 1987). On the other
hand, the parietal opening in typica1 nothosaurs and other pachypleurosaurids a1ways
lies in the posterior half of the parietal. In the eosuchian Youngina and the primitive
• 16 sauropterygian Claudiosaurus, the parietal opening is large, and situated at the middle • of the parietal (Carroll, 1981). At the back ofthe skull table, the midline ofthe parietal extends a short distance
iJeyond the shal'p ridge that separates the skull table and the occipital surface, forming
a wedge-shaped protrusion. This can be seen in every specimen of Keichousaurus
where this region is preserved. This phenomenon is also present in some modern
lizards. On both side ofthe wedge, the parietal turns downward and somewhat inward,
forming Iwo notches for the attachment of the epaxial muscles.
On the occipital surface, just medial to the squamosals, there are a pair of
co1urnnar bones (?supratemporals). The upper end of the bone turns sharply inward
toward the midline, forming a beak-shaped projection. The points of the beaks meet
in the midline, separating the occipital portions of the parietal and supraoccipital • (Figure 4). Bones occupying this position have never been reported in other nothosaurs. In primitive diapsids, one might expect to find the postparietal, tabular or
supratemporal. Claudiosaurus retains a supratemporal (Carroll, 1981), but because the
configuration of the occipital surface is so different in Keichousaurus, it is not certain
whether this is the same element. Since this structure can be observed in ail the skulls
that are exposed dorsal1y, it is unlikely that it is a fragment of other, adjacent bone.
The squamosals are large bones mainly confined to the occipital surface. The
long anterior ramus of the bone extends forward underneath the postorbital and forms
the upper temporal bar in conjunction with the latter bone. The quadrate ramus of the
squamosal is slender, extending posteroventrolaterally. In the occipital region, the body • 17 " of the squamosal expands considerably medially as in other pachypleurosaurids. • Medially, it makes contact with the parietal, ?supratemporal, supraoccipital and opisthotic successively, forming a continuous occipital wall. The post-temporal
fenestra, seen in Claudiosaurus and primitive diapsids, is either very small or 10s1.
Whether there is a quadratojugal is not certain. Ifthere is one, it must be very
small.
The postorbital of Keichousaurus is a triradiate bone, while in Serpianosaurus
and Neusticosaurus, this bone is triangular rather than triradiate. The posterior ramus
extends back to the occipital margin above the postorbital ramus of squamosal. The
two bones interdigitate with each other to form the upper temporal bar which separates
the upper temporal opening and the lower temporal embayment. The upper temporal
bar is slender as in Anarosaurus (Nopcsa, 1928) and Dactylosaurus, and differs • considerably from that ofSerpianosaurus and Neusticosaurus where the cheek is broad and the emargination is less eviden1. The other two rami of the postorbital are short.
The superior ramus fits into a groove in the pqstfrontal; the inferior ramus points
anteroventrally, making contact with the small jugal where they form the posterolateral
rim of the orbit.
The jugal is reduced to a slender bar. The bone is partially covered laterally by
the maxilla. The posterior end thickens slightly and makes contact with the postorbital
and the ectopterygoid. In GXD7601, there seems to be a small downward extension
at the posterior end of the jugal which might fit into a small notch in the coronoid of
the lower jaw. This relationship is reversed from that in most nothosaurs, in which the • 18 coronoid fits into a notch in the upper jaw. • The postfrontal is triangular. Ils anterior edge butts onto the frontal and the latera! apex has a groove to accept the postorbital. The anterior edge fooos the
posteromedial rim of the orbit. The posterior edge participates in the formation of the
upper temporal fenestra.
The prefrontal is very large. Il occupies the region between the external nares
and the orbits. The orbital margin of the prefrontal is considerably thickened to form
a ridge, and turns steeply down the anterior margin the orbit, reaching the level of the
palate.
As in other sauropterygians, there is no trace of the lacrimal.
The maxilla is a large element. Il has an elongated triangular shape. Anteriorly,
it reaches the premaxilla and forms the posterolateral margin of the externat nares. • Dorsally, the maxilla has a finger-like median protrusion fitting into a shallow groove on the prefrontal. The posterior extension of the maxilla forms the lateral rim of orbit,
extending posteriorly beneath the cheek region.
The supraoccipital is a large element. A ridge runs along the midline, in
confluence with that of the parietal. Underneath the questionable supratemporal, the
supraoccipital makes contact with the squamosal. Two shallow depressions are apparent
in the area just above the rim ofthe foramen magnum, presumably to accommodate the
epaxial musculature of the neck region.
The suture between the exoccipital and opisthotic can be discerned in GXD7601
in dorsal view and in GXD838028 in ventral view. The exoccipital is thin and lies • 19 superlicial to the opisthotic. The opisthotic extends laterally and makes contact with • the supraoccipital and squamosai. The paroccipital process extends further beyond the contact with the squarnosal. The distal eEd of il bears a slightly expanded articulating
facet. !t probably articulated with the quadrate. The body of the paroccipital process
is broad, unlike that seen in eosuchians. The space between the lower margin of the
paroccipital process and the posterior margin ofthe pterygoid is closed by an extension
of the paroccipital process.
The quadrate is short and broad. !ts body is essentially horizontal. An
ascending process abuts in a shallow groove on the anterior surface ofthe squamosal's
quadrate rarnus. In the posterior aspect ofthe quadrate, there is a conspicuous quadrate
conch which is often associated with the tympanum. The mesial surface ofthe quadrate
is closely integrated with the quadrate rarnus of the pterygoid and paroccipital process. • No streptostyly was possible. The middle ear charnber is formed by the mesial part of the quadrate, the
squarnosal, paroccipital process, and possibly the quadrate fossa in the retroarticular
process of the lower jaw. No stapes is observed, though one is reported in
Neusticosaurus edwardsi (Carroll & Gaskill, 1985). The fact that the stapes couId not
be observed either indicates that there was not a stapes in the lirst place, or that the
stapes was too smalI and fragile to be preserved. Since other structures, such as the
quadrate conch and the middle ear charnber, indicate the existence of a impedance
matching middle ear, it is likely that stapes was present.
The whole occipital surface is well consolidated. This phenomenon can be • 20 found in other secondarily aquatic reptiles such as crocodiles, turtles and placodonts. • Il provides with more attaching area for axial muscles and indicates that it may be more important to stabilize the head region than in a terrestrial environment. In
sauropterygians, both pachypleurosaurids and nothosaurids have a closed occiput
(Schultze, 1970) while plesiosaurs retain a wide open occiput. Since it is a complex
modification involving more than one bone to change shape, the closed occiput could
weIl be an important synapomorphy uniting pachypleurosaurids and nothosaurids.
The palate is not weil preserved (Figure 5). Through the left orbit of V952 we
can see that the palatine extends backward to the level of the jugal, and the suborbital
fenestra remains as a narrow slit. In Alpine pachypleurosaurids the suborbital fenestra
is confluent with the subtemporal fenestra (Carroll and Gaskill, 1985; Rieppel, 1989; • Sander, 1989). The condition in Keichousaurus hui is more primitive. Related to thc separated suborbita1 fenestra is the retainment of the ectopterygoid in Keichousaurus
hui. The ectopterygoid can be seen in specimens GXD7601, V952 and GXD838028,
though the suture is not a1ways clear. Il makes contact with the jugal. The
ectopterygoid was reported in Neusticosaurus edwardsi (Carroll and Gaskill, 1985), but
not in Neusticosaurus pusil/us, N. peyeri and Serpianosaurus (Rieppel, 1989; Sander,
1989). As in ail other nothosaurs, the interpterygoid vacuity is comp1ete1y closed and
the pterygoid extends back to the level of basioccipital. The basisphenoid appears to
have been completely covered by the pterygoid. The quadrate ramus of the pterygoid
extends directly laterally rather than posterolaterally as in eosuchians. As in ail • 21 •
Figure 5 Skull of Keichousaurus hui, palatal view, GXD838028.
scale x6
• 22 • f
po
a pa
q : pt _-++1U::t:: \op--....:\fi~ \ ~~i ! ar---~'Y
eo ic2
• sauropterygians, the transverse flange of the pterygoid has disappeared. A marked • longitudinal ridge separates the quadrate ramus from the medial portion of the pterygoid, as in other pachypleurosaurids.
The occipital condyle is formed solely by the basioccipita1. The base of the
condyle constricts to form a neck.
The vomer and the palatine, and their boundaries, cannot be observed in any
specimen. It is also impossible to say anything about epipterygoid or sphenethmoid.
Lower Jaw
The anterior part of the lower jaw is not weil exposed. From what can seen
through the orbits, the dentary extends to the level of the back of the orbit. The teeth • are conical in shape. The coronoid is not observable. A thin splenial can be discemed along the inner
wall of the dentary. How far back it reached is not certain. The surangular is
immediately behind the dentary. It is very thick. Whether there is a mandibular fossa
is not known. There is a shallow groove on the lateral surface of the surangular to
accommodate the wedge-shaped angular. The remainder of the suture between the
surangular and the articular could not be discemed. The articular expands dramatically
inward to form the articulating surface. Posterior to the jaw joint is a prominent
retroarticular process. An medial expansion extends back two-third ofthe length ofthe
retroarticular process, forming a large quadrate fossa (Figure 4). A similar structure • 23 was found in European pachypleurosaurids, and it was believed to be part of the • articulating surface that received the posterior extension of the quadrate when the jaws are open (Sander, 1989; Rieppel, personal communication). In Keichol/sal/rl/s hl/i,
however, this fossa is considerably larger and deeper than in Alpine pachypleurosaurids.
In addition to accommodating the quadrate, the function of this quadrate fossa might
be to form part of the middle ear chamber and support the tympanum. Behind the
quadrate fossa, the retroarticular process tapers to a robust bony bar for the depressor
mandibulae.
Tooth Implantation
In the lower jaw of Keichol/sal/rus hui, the teeth are set in individual sockets • along a longitudinal groove as in typical primitive diapsids, a pattern termed subthecodont (Romer, 1956). However, Keichousaurus hui (as weIl as Neusticosaurl/s
pusillus, see Fig. 13 in Sander, 1989) is peculiar in that, in contrast to the subthecodont
condition of primitive diapsids where the inner wall of the lower jaw is thinner and
lower than the outer wall, the inner wall of the lower jaw in Keichollsaurus hui and
Neusticosaurus pusillus is thicker and higher than the outer wall ( see GXD7601). This
might be termed the reversed subthecodont condition. This condition could be derived
from the thecodont condition by lowering the outer wall. Since the upper jaw is not
exposed in ventral view in any of the specimens, the exact nature of tooth implantation
in the maxilla and premaxilla is not clear. • 24 Vertebrae and Ribs • (Figure 6, Figure 7) The vertebral count, especially the nurnber of cervical vs. dorsal vertebrae, is
considered important in the classification ofpachypleurosaurids. However, an accurate
count is hard to establish because the distinction between the cervical and dorsal regions
is not clearly established in reptiles. This problem has been addressed by Hoffstetter
and Gasc (1969). Carroll and Gaskill (1985) and Rieppel (1989) used the transition
from double-headed cervical ribs to the single-headed dorsal ribs as a mark to
differentiate the two regions in Neusticosaurus edwardsi and Serpianosaurus,
respectively. Sues and Carroll (1985) used the change in the rib heads as well as the
modification of the zygapophyses to distinguish the cervical and dorsal vertebrae of
Dactylosaurus. Among the specimens under study, none shows the nature of the rib • articulation in the cervical-dorsal transition area. The neural arch configuration, however, can be observed in all the specimens that are exposed dorsally. In the cervical
region, the transverse process does not project beyond the zygapophysis. The neural
arch is constricted in the middle. In the dorsal region, the pachyostosis (thickening of
the ribs and vertebrae) is apparent and the transverse processes extend beyond the
zygapophyses. It can also be observed that the last cervical ribs are much shorter and
stouter than the first dorsal ribs. This coincides perfectly with the change in the neural
arches.
There is, however, variability in the count from specimen to specimen. Among
the specimen studied, the count of presacral vertebrae ranges from 43 to 45. Of these, • 25 •
Figure 6 Dorsal aspect of Keichousaurus hui, GXD7613.
scale x1.2
• 26 . '.
1 • 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 .' 1 • 1 " . " 1
)j 1 tl r. / 1 1 1 1 1 1 •
Figure 7 Ventral aspect of Keichousaurus hui, GXD838028.
scale xl.5
.", .
• 27 ----_._- ..,
• , " "
"'.' , • ~.
•
" 25-26 are cervical and 18-19 are dorsal vertebrae. The ratios of cervical/dorsal • vertebrae of European species are: Neusticosaurlls edwardsi 17/19-20 (Carroll and Gaskill, 1985), Serpianosallrlls 15-18/20-23 (Rieppel, 1989), Neusticosallrus peyeri
15-16/19-20, Neusticosaurus pusil/us 18-20/22-24 (Sander, 1989). Dactylosaurlls has
the same count as Nellsticosaurlls edwardsi (Sues & Carroll, 1987). Since the length
of vertebrae does not vary much l'rom cervical to dorsal region, a higher count of
cervical vertebrae translates into a longer neck for Keichousaurlls hui than for other
pachypleurosaurids.
The number ofsacral vertebrae varies l'rom two to three. In GXD7621 two pairs
of sacral ribs thickened drastically, clearly diff-:rentiating them l'rom those immediately
anterior and posterior. In GXD7613 the last pair of dorsal ribs also thickens distally,
a1though not so much as in the sacral ribs proper. In GXD7602 the ribs in the sacral • region do not thicken noticeab1y but there are three pairs of ribs converging toward the ilium. In GXD7603, the distal end of three pairs of ribs in the sacral region are
expanded, but the proximal ends are not. Compared with other pachypleurosaurids in
which there are three or four sacral vertebrae, the condition in Keichousaurus is closer
to that of primitive diapsids.
No specimen except the smallest one (GXD835002) preserved the whole length
of the tai!. The count of caudal vertebrae is at least 25 (GXD7613) and might weil be
more than 30. The count of caudal vertebrae for various European pachypleurosaurids
are: Neusticosallrus pusil/us 51-58, N. peyeri 40-48 (Sander, 1989), Neusticosaurus
edwardsi 42 (Carroll and Gaskill, 1985), Serpianosaurus at least 47 (Rieppel, 1989). • 28 1t has been shown that the counts of the caudal vertebrae are correlated positively with • the ontogenetic stages (Sander, 1989). The elements of the atlas-axis complex (Figure 8) remain discrete. As in other
primitive sauropterygians, no proat!as was found. An oblong proatlas was reported for
Claudiosaurus (Carroll, 1981). In GXD838028, the first and second intercentra contact
on the ventral surface so that the first centrum cannot be seen in ventral view. It can
only be observed in dorsal view in GXD7601 (Figure 4). The first pair ofneural arches
can be seen in GXD7602 (Figure 8). They are paired triangular elements. The front
edge of the atlas arch is straight. Whether the arches articulated with the occiput and
the nature of their articulation with each other is not clear. The postzygapophyses
extend across the second neural arch. The neural spine was not developed on the atlas
arch. The axis is larger than the third cervical as in most reptiles. As in other primitive • reptiles, the axis intercentrum and centrum are not fused. The intercentrum is about half the length of the centrum. Both are constricted laterally in the middle. The
parapophysis, which is jointly formed by the intercentrum and centrum, is close to the
ventral surface of the column. A sharp keel is present on the ventral midline of both
intercentrum and centrum. The axis neural arch has essentially the same shape as that
of the rest of the cervical vertebrae, only the neural spine is thicker. and longer.
The vertebral centra of the cervical region are longer than broad. The middle
of the centra is constricted laterally. The length of the cervical vertebrae does not
increase much from the third to the last, but approaching the dorsal region, they become
broader. In the dorsal region, the diameter of the centra is subequal to the length, and
• 29 •
Figure 8 Atlas-axis complex of Keichousaurus hui.
a. dorsal view, GXD7602 b. ventral view, GXD838028.
scale xl1.4
• 30 • N lJ C.
" ."
• the mid-portion is expanded both ventrally and laterally instead of being constricted, • giving them a barrel-like appearance. In the anterior caudal region, the centra constrict once more.
In ail the specimens, the vertebral column is weil articulated. In specimens
GXD7603 and GXD838028 a slight displacement of the vertebral column occutTed
around the sacral region. From these two specimens we can see that the vertebral
centra are shallowly amphicoelous, as in other pachypleurosaurids.
The outline of the neural arch in the cervical region, when viewed from above,
is bell-shaped. The articulating surfaces between the zygapophyses slant
ventromedially, indicating that lateral and rotational movements ofthe neck are coupled
(to be discussed later). Pachyostosis of the neural arch begins at the 15th or 16th
cervical. • In the dorsal region, the outline of the neural arch is pentagonal as in Dactylosaurus. The transverse processes are more inflated in Keichousaurus and
Dactylosaurus than in Neusticosaurus and Serpianosaurus where the outline of the
neural arch is rectangular. The articulating facet of the transverse processes of the
dorsal vertebrae is elongated dorsoventrally and has a very shallow saddle-shaped
contour. This is reflected in the capitulum of the dorsal ribs. Thus, the dorsal ribs
seem to be quite free to move. In the sacral region and the tirst few caudal vertebrae
where the ribs are still prominent, the articulating facet is round and concave. This
indicates a more rigid connection of vertebrae and ribs in the sacral and anterior caudal
regions. In ventral view, the transverse process is anteriorly positioned on each • 31 vertebral segment. • The plane of the articulation surface of the prezygapophyses slopes ventromedially in the dorsal, sacral and anterior caudal regions. AIso, to a lesser
degree, they tilt up anteriorly. Thus, the successive vertebrae are interlocked together,
severely restricted lateral bending as in Dactylosaurus (Sues and Carroll, 1985).
An accessory articulation complex can be seen between the successive neural
arches in the cervical region (Figure 9) as observed in ail other pachypleurosaurids
(Kuhn-Schnyder, 1959, cÎted by Rieppel(1989); Carroll and Gaskill, 1985; Rieppel,
1989; Sander, 1989). At the rear end of the neural arch, a medial extension of the
postzygapophysis forms a shelf at the midline at a deeper level than the primary
articulating surface. At the base of the following neural spine, a tongue-like process
fits onto this shelf. The front edge of the lower part of the neural spine fits into a • vertical groove at the rear end of the more anterior neural spine. This articulation further restricts the rotational movement of the vertebral colurnn.
The neural spines ofthe cervical region are narrow and very low, except for the
second cervical. In the dorsal region, they become significantly thickened. The height
i~ also increased, but does not exceed the length. The tip of the dorsal neural spine is
usually slightly expanded lateraUy and typically unfinished at the top, suggesting that
there might have been a portion that remained unossified. The dorsal and the fust five
or six caudal neural spines are very long, fitting tightly with their neighbours.
Dorsoventral flexion was probably impossible in this area. The thickly packed rib cage
in the dorsal region and the long and laterally pointing proximal caudal ribs make • 32 lateral movement equally difficult. The highest neural spines along the vertebral • column are in the sacral region. From the 6th to the 1Dth caudal vertebrae, the front edge of the neural spine tilts back, forming a triangular gap between the successive
spines (Figure 9). This may indicate an increase in dorsoventral mobility.
The haemal arches can only be observed on GXD7613. The first haemal arch
is associated with the 6th and 7th caudal vertebrae. The 5th caudal bears an area of
unfinished bone on the ventral surface, but whether it supported a haemal arch is not
certain. About 8 haemal spines are present. In contrast with their counterparts in
Neusticosaurus and Serpianosaurus, there is no distinct pedicel on the haemal spines
of Keichousaurus. No cross piece is observed. The last few haemal spines diminish
rapidly in size, and transform to a small chevron pointing backward.
In GXD838D28, only one articular facet with the rib can be observed for each • cervical vertebra. The parapophyses are prominent, extending laterally from the centrum. From the 3rd to the 7th cervical, the parapophyses is in the anterior half of
the centrum. From the 8th cervical, the parapophyses starts to shift backward and
upward, and the neural arch participates in the formation of the facet, but there is not
transverse process per se. In the last few cervical vertebrae of GXD838D28, the suture
between the centra and neural arches runs thrûugh the articulating facet, and the
transverse process becomes prominent. The same condition is also reported for
Neusticosaurus edwardsi (Carroll and Gaskill, 1985). Sander CI 989) reported double
headed cervical ribs in both Neusticosaurus pusillus and N. peyeri. In Serpianosaurus,
the cervical ribs are also double-headed (Rieppel, 1989). In the dorsal region, the • 33 •
Figure 9
a. accessory articulation of neural arches, GXD7601
b. frrst caudal neural spine, GXD7613. • scale x6
• 34 •
a
sa2---.k~
b • prominent, dorsoventrally elongated transverse process is the sole articulating structure • for the rib, located high up on the neural arch. The suture between the centra and neural arches is still visible even in the largest specimen, GXD838028. In the tail
region, the articulating facet starts to migrate down again; in GXD7613, the suture runs
through the articulating facet of caudal vertebrae 6, 7 and 8.
In GXD838028 it can be seen that the first pair of cervical ribs is single-headed,
horn-shaped and articulates with the atlas intercentrum (Figure 8). In V952, the anterior
cervical ribs up to the 13th show the double process noted in Serpianosaurus by
Rieppel (1989). No trace ofanterior process can be seen on the last eight or so cervical
ribs of GXD7613. These ribs are straight and pointed. They show a graduai increase
in length caudally, with successive ribs about 20% longer than those more anterior. An
abrupt increase in length appears at the transition between cervical and dorsal regions. • The first dorsal rib is more than twice the length of the last cervical rib. It is also thinner than the last cervical rib. The typical dorsal ribs are five to six time the length
of the corresponding centrum. They bend backwards, extending to the lever of 4th or
5th vertebra behind. When restored, these ribs may point ventroposteriorly. The distal
end of the typical rib expands slightly, suggesting that there might be cartilagous link
between the rib and the gastralia, but no trace of it could be seen. The last two or three
ribs before the sacral region differ significantly from the typical dorsal ribs. They are
shorter and almost straight.
The sacral ribs never co-ossify with the vertebrae or with one another, but there
is sorne rugosity on their proximal surface, indicating strong muscle attachment. One
• 35 pair of sacral ribs (the first pair of the two, or the second pair of the three) bears a • recess at the distal end, into which would fit a small projection ofthe ilium (Figure 13). Eight or nine pairs of caudal ribs are present. The first five pairs are almost as
long as the sacral ribs, and Iike the sacral ribs, they also point laterally. The last four
pairs shorten rapidly and they are slightly bent, pointing anterolaterally. They never
fuse with the vertebrae.
Gastralia
Gastralia are present from the posterior end of the pectoral girdle to the anterior
end of the pelvic girdle. In ventrally exposed specimens the gastralia are either
completely eroded or oruy have sorne broken fragments left. They can only be • observed through the gap between the ribs in the specimens that are dorsally exposed.
The gastralia are thin and packed densely. There are more than one row of
gastralia to each dorsal segment, but the exact number is not certain. There are one
central piece and two lateral element on each side of it in each row of gastralia. The
central piece is "V" shaped as in ail early neodiapsids. The lateral elements partially
overlap the central piece, and adhere to the anterior side of it. Neuslicosaurus has three
elements per gastralia segment: one central and one lateral on each side (Carroll and
Gaskill, 1985; Sander, 1989), whereas Serpianosaurus, Iike Keichousaurus, has five
elements. • 36 • Appendicular Skeleton
Pectoral Girdle
(Figure 10)
The configuration of the pectoral girdle of Keichousaurus hui conforms to the
general pattern seen in other pachypleurosaurids. It is in the form ofa ventrally placed
circle with short dorsal extension of the scapular blade. The elements involved in
forming the pectoral girdle are: a single interclavicle, paired clavicles, scapulae and
coracoids.
The interclavicle is a small, thin, T-shaped element. 1t fits in a groove at the
posterior edge of the ventral surface of the clavicles. The posterior edge of the bones • are confluent. The transverse bar of the interclavicle is half to two thirds the length of the total length of the clavicles. There is a short stem of the interclavicle, about half
the length ofthe transverse bar, pointing posterioriy. The interclavicle and clavicles are
always solidly attached to one another.
The clavicles form the anterior bar of the pectoral girdle. This bar is basically
straight, unlike that of European pachypleurosaurids where there is a lateral shank
pointing posteriorly (Sander, 1989). The two clavicles are suturally iuticulated at the
midline in a zig-zag fashion. The length of the clavicular bar is about one-half to two
thirds of the distance between the glenoids. There are three low projections on the
anterior edge of the bar. One is in the middle where the clavicles meet, and the other • 37 •
Figure 10 Pectoral girdle of Keichousaurus hui, ventral view.
a. GXD838028 b. V7919 • scale x3
• 38 •
.-(J • two are situated near the corner of the c1avic1e-scapula articulation. • In sauropterygians, the normal relationship of the c1avicle and the base of the scapula is reversed from the typical terrestrial tetrapod condition. Instead of being
superficial to the scapula, the lateral portion of the clavicles lies deep to the anterior
portion of the scapula, suggesting development of the clavicles in deeper tissues than
in most primitive tetrapods. This is demonstrated in all the specimens where this
articulation can be observed except V7919. In V7919, the clavic1e and scapula show
the primitive relationship in which the scapula is deeper than the c1avicle (Figure 10).
The evidence of postmortal disturbances is minimum. Either the c1avicle bar might
have been accidentally dislocated in life, or the peculiar relationship may have been
caused by a developmental abnormality. It is not c1ear what was the cause of this
reversaI. Nor, indeed, is there an obvious reason for the pattern in sauropterygians. It • has been proposed that the peculiar configuration in sauropterygians might have resulted from an extreme reduction of the extent of the shoulder girdle early in the evolution of
the sauropterygians, followed by re-elaboration, with a reversai of the relationships
(Carroll and Gasldll, 1985).
As in all other sauropterygians, the scapula and coracoid ossified separately.
As is in other nothosaurs, the dorsal blade of the scapula is short and not at all
blade shaped. It is a low, horn-like structure arising from the base of the scapula and
pointing posteriorly and almost horizontally. The tip ofthis horn-like structure extends
behind the level of the glenoid. The ventral portion of the bone expands horizontally
to form a ventral blade. Normally, the dorsal side of the anteromedial portion of the • 39 ventral portion bears a depression for articulation with the lateral process ofthe clavicle. • The long axis of the ventral portion of the scapula fonns a 30-45 degree angle posteriorly with the longitudinal axis of the body.
The coracoid is the largest element of the pectoral girdle. The part contacting
the scapula is broad; the middle part of the bone contracts slightly. Where the
coracoids meet along the midline, they expand again, both anteriorly and posteriorly,
to form a long symphysis, as is typical of nothosaurs.
A coracoid foramen (or supracoracoid foramen) can be seen on the Hne of
contact between the scapula and coracoid in most specimens. Its margins are fonned
jointly by the two bones. The position of the coracoid foramen is two-fifths to one
third of the length of the line of contact between the scapula and coracoid. In the
European pachypleurosaurids, the coracoid foramen is usually situated at the medial end • of the scapula-coracoid suture line. In sorne cases, the two notches on the scapula and coracoid do not even meet to form a foramen. Instead, they both open into the large
space surrounded by the pectoral girdle.
The glenoid cavity is fonned by the scapula and coracoid at the lateral extremity
of the shoulder girdle, facing mainly laterally. Both scapula and coracoid thicken
considerably in this area The free ends of the two bones are rough and fonn an angle
of about 100 degrees. There must have been a largely cartilageous joint capsule.
• 40 Humerus • (Figure 11) The humerus is the longest bone of adult individuals, being 28.8 mm long in
GXD838028, about 31 % of the trunk length. The proximal end expands slightly and
has a rounded triangular outline in cross section. The expansion of the distal end is
more prominent, and it is flattened dorsoventrally. The distal end ofthe humerus bends
slightly inward so that the anterior edge is straight. In contrast with ail other
pachypleurosaurids, but as in plesiosaurs, there is no trace of the entepicondylar
foramen.
As in European species, two morphotypes can be recognized in Keichousaurus
hui based mainly on the structure of the humerus. This is interpreted as sexual
dimorphism (Sander, 1989), and the two types are assigned sex x and sex y (Figure II). • In sex x, the humerus is smooth and featureless whereas in sex y muscle scars, processes and crests are more prominent. The expansion of the distal end of the bone
is more pronounced in sex x. The relative length of the humerus of sex x is shorter
than in sex y. The overall appearance of the humerus of sex x retains the pattern of the
juvenile stage.
Ulna and radius
(Figure 12)
The most conspicuous feature of Keichousaurus hui is the shape of its ulna. In
contrast with other pachypleurosaurids, the ulna of Keichousaurus hui is a broad, • 41 blade-like bone. It is much wider proximally than distally. The cross section of the • ulna is wedge shaped, the mesial edge facing the radius is thicker. Dnly the nothosaurid Lariosaurus has a comparable configuration of the ulna (Kuhn-Schnyder,
1987; Mazin, 1985). There i~ no trace of olecranon process. The most obvious
functional advantage of the broad ulna is that it dramatically increases the surface area
ofthe forelimb, tuming the forelimb into a more effective oar. Probably, the broad ulna
evolved independently in Keichousaurus and Lariosaurus.
The radius is a little longer than the ulna. It is slim as is usually the case in
pachypleurosaurids. The proximal end is wider than the distal end. The radius curves
out slightly, and there is a wide space between the radius and the ulna at the distal end.
Carpus and Manus • (Figure 12) The number ofossified carpal e1ements is apparently age-dependent and follows
the pattern of primitive eosuchians (Carroll, 1985; Caldwell, in press). In adult
specimens, there are 5 elements ossified. The intermedium is large and round, situated
between the distal ends of the ulna and radius. This is a typical nothosaur character.
In pachypleurosaurids the intermedium is typically more elongated and situated to the
end of the ulna, beside the radius (see Fig. 17, Carroll and Gaskill, 1985). The ulnare
is also round. The diameter ofthe ulnare is only half that of the intermedium. Three
tiny round dements, which can be found in mature specimens, situated at the proximal
end of the second, third and fourth metacarpal, are their corresponding distal carpals.
• 42 •
Figure 11 Humerus of Keichousaurus hui.
a. left humerus of GXD838028, ventral view
b. right humerus of GXD838028, ventral view
c. right humerus of GXD7613, dorsal view
d. right humerus cf V953, ventral view
a, b, c are of sex y, d is of sex x.
scale x4
• 43 • "
(J
.c
• •
Figure 12 Lower arm of Keichousaurus hui.
a. GXD7613, left, dorsal
b. GXD7613, right, dorsal
c. V953, right, ventral
seale x4
• 44 1
1
1
, 1
1
1
1
1
1
1
1
1 "S..
1
1
1
1
1 • 1 1 In European pachypleurosaurids, Neusticosaurus edwardsi and Dactylosaurus have three • carpal elements: the intermedium, the ulnare and the fourth distal carpal (Carroll and Gaskill, 1985; Sues and Carroll, 1985). Neusticosaurus peyeri and N. pusil/us have two
elements: the intermedium and the ulnare (Sander, 1989). Serpianosaurus also has only
two ossified carpal elements. The number of carpal elements of Keichol/saurus hui is
the highest among weil known pachypleurosaurids. This is apparently a retention of
the primitive condition. Nothosaurids usually have a higher count of ossified carpals.
Lariosaurus has at least five ossified carpal elements as does Keichousaurus. A
juvenile specimen of Lariosaurus was observed to have a radial in addition to the
elements present in Keichousaurus (Carroll, 1985). Paranothosaurus has the
intermedium, ulnare, the third and the fourth distal carpals ossified (Peyer, 1939).
Ceresiosaurus has six ossified elements in the carpus (Kuhn-Schnyder, 1963; see • Carroll, 1985). In nothosaurids, the number ofthe ossified distal carpals may be related to their larger size.
The metacarpals and phalanges are reasonably weil preserved in GXD7603 and
GXD7613. As in Alpine pachypleurosaurids, the third metacarpal ofKeichousaurus hui
is the longest. The second metacarpal is only slightly shorter (in sorne case they are of
the same length). In fully grown individuals, the fourth metacarpal is shorter than the
second, and the first and the fifth are about the same length. In Neusticosaurus
pusil/us, the relative length of the third and the fourth metacarpal show continuous
spectrum from the third longer than the fourth, to the fourth longer than the third
(Sander, 1989). As in Alpine pachypleurosaurids, both ends of the metacarpals, • 45 especial1y the proximal end, expand, and they show a slight twist relative to one
1 another. Unlike the condition in Alpine pachypleurosaurids, however, the proximal end
of the metacarpals do not overiap, resulting in greater breadth of the hand, confluent
with the contour ofthe lower arm. In undisturbed specimens, the first four metacarpals
bow slightly toward the ulnar side, and the fifth bows toward the opposite direction.
The phalangeal formula is 3,4,4(?5),4,3. The third digit is usually the longest.
ln other pachypleurosaurids, it is always the fourth digit that is the longest (Carroll and
Gaskil1, 1985; Rieppel, 1989; Sander, 1989), as in primitive diapsids and lizards.
Therefore, the hand of Keichousaurus is more symmetrical.
Pelvic Girdle
(Figure 13) • The pelvic girdle of Keichousaurus hui can be observed in GXD838028, V953 and V7919 in ventral view, and GXD7602, GXD7603, GXD7613 and GXD7621 in
dorsal view.
The ilium is small, as in other nothosaurs. The iliac blade is merely a slender
vertical bar. On the middle of the medial surface there is a small process that fit into
the recess at the distal end of one of the sacral ribs. The base of the ilium is in the
shape of a rounded triangle in ventral view. A triradiate keel divides the ventral surface
into three areas. The two medial areas are fiat. The one in the front, which is smaller,
articul.ates with pubis; and the larger one at the back articulates with ischium. The
lateral partition is slightly concave. It participates in the formation of the acetabulum.
• 46 Figure 13 Pelvic girdle of Keichousaurus hui.
a. reconstruction of pelvic girdle in lateral view
scale x4.5
b. left ilium of GXD838028, medial view
c. same as above, lateral view
d. same as above, ventral view
scale x6
• 47 • o
oC
• Above the acetabulum, there is usually a scar on the ilium, suggesting the presence of • the base of a cartilagous supracetabular buttress. The pubis has an acetabular region, a narrow neck region and a broad media!
region. The acetabular region is the thickest part of the bone. Three articulating facets
are clearly distinguishable at the lateral end of the pubis: a posterior facet articulates
with the ischium; a dorsal facet articulates with the ilium; and a lateral facet forms part
of the acetabular. The pubis thins rapidly towards the neck region, where it also
constricts posteriorly. It then widens into a broad and thin medial region. The media!
region of the pubis forms a broad curve which is in contact with its counterpart of the
other side. No pubic rim or process could be observed which would have been used
to anchor the abdominal musculature or tendons (Romer, 1956). Rieppe1 (1989)
reported such a structure in smaller specimens of Serpianosaurus, but he noted that it • is not present in the larger specimens. As in Dactylosaurus, in V953 and V7917, the obturator foramen is enclosed in the pubis near the articulation with the ischium. In the
largest specimen, GXD838028, the obturator foramen has complete!y disappeared. In
Alpine pachypleurosaurids, the obturator foramen is a sHt open to the large puboischiac
fenestra. The obturator foramen of nothosaurs is not found in plesiosaurs (Fig. II in
Robinson, 1977). The plesiosau~ obturator foramen is probably homologous to the
puboischiac fenestra of nothosaurs.
The acetabular region ofthe ischium ofKeichousaurus hui is structurally similar
to that of the pubis. It is the thickest part of the bone with three articulating facets: an
anterior facet articulating with the pubis, a superior facet articulating with the ilium, and • 48 a lateral facet fonning the posterior part of the acetabular. After a short, s!ightly • narrowed neck region, the ischiurn expands posteromedially to fonn a f1attened medial region. The posterior angle, for attachment of the caudal musculature, is not as thick
as in Neusticosaurus pusillus (Sander, 1989, Fig. 16) and Neusticosaurus edwardsi
(Carroll and Gaskill, 1985, Fig. 18). The medial extremity of the ischium fonns a
more or less straight !ine in front, but curves s!ightly laterally at the rear end.
The posterior margin of the pubis and the anterior margin of the ischium
encloses a large space, the thyroid fenestra. The curved nature of the extremity of the
pubis indicates that the symphysis is probably cartilageous. Otherwise, it is hard to
imagine how the two pubes relate to each other.
The acetabular facet is a shallow socket facing ventrolaterally. Movements of • the femur are restricted below the plane of the acetabulum. Femur
(Figure 14)
The femur of Keichousaurus hui is not as weil developed as the hurnerus in the
adult; this is typical among pachypleurosaurids. It is shorter, more slender and the
surface of the bone is smoother than the hurnerus. The proximal end of the femur is
as broad as that of the hurnerus. The distal end, however, is oruy about half as broad
as the proximal end. The posterior side of the shaft is concave, whereas the anterior
side is straight.
The articulation surface at both ends are convex in contrast with Neuslicosaurus • 49 •
Figure 14 Hindlimb of Keichousaurus hui.
a. left femur of GXD7613, dorsal view
b. right femur of V953, ventral view
c. proximal end of left femur of GXD838028, ventral view
d. same as above, end view
e. right lower leg of V953, ventral view
f. left lower leg of V7919, ventral view
scale x4
• 50 •
: ,:~
. '.:.~
a b d
.l~'; . ; l, l'. ~.. ' \\, ti .. ~Q.
dt4
\l
• 'e f edwardsi, in which they are concave (Carroll and Gaskill, 1985). In adult specimens, • there is a very prominent horizontal posterior extension of the head. There is also a less weil developed ventral extension of the head. Between these two extensions is a
shallow recess which might be comparable to the intertrochanteric fossa (Romer, 1956;
see discussion on locomotion and musculature reconstruction later). The distal ·~nd of
the femur is round and featureless, indicating a weak articulation with the epipodials.
Tibia and fibula
(Figure 14)
The tibia and fibula are short in Keichousaurus hui. They are about 30·40% of
the length of the femur and about 60-80% of the length of the ulna and radius. Of the
two bones, the fibula is a little stronger. The tibia is quite straight and without • significant expansion at either ends. The two ends, especially the distal end, of the fibula expand towards the tibia. There is an interosseal space between the two bones,
which is comparable to the space between the ulna and radius.
Tarsus and Pes
(Figure 14)
Ty;o elements ofthe tarsus, the calcaneum and the astragalusare, are consistently
ossified in Keichousaurus hui. The calcaneum is smaller and is situated at the distal
end of the fibula. The large astragalus is situated at the distal end of the interosseal
space, and articulates with both the tibia and the fibula. The astragalus and the • 51 calcaneum may have also articulated with each other, though the articulating facets are • not weil defined. A third element of the tarsus - the fourth distal tarsal - is also present in sorne
adult specimens, including GXD7613, GXD7621 and V79l9.
The metatarsals of Keichousaurus hui follow the pattern of the European
pachypleurosaurids. The first metatarsal is the shortest, its length being about 60% the
length of the third metatarsal. The lebgth of the second metat:n:sal is about 80% that
of the third. The third and the fourth metatarsals are about the same length, and the
fifth is only slightly shorter.
The phalangeal formula for the pes of Keichousaurus hui is 2-3-4-4-4. The last
segments are usually broad and pointed. The elements of the pes are more robust than • their counterparts in the manus.
• 52 • MEASUREMENTS AND PROPORTIONS
The measurements ofKeichousaurus hui were made either under the microscope
when the part being measured could be enclosed in the field of view ofthe microscope,
using the grid in the ocular lens to calibrate, or else using a calliper for larger elements.
A total of 59 different measurements were taken from the specimens (Table 1).
When measuring the paired parts, such as the limb bones, the average of the individual
measurements is used. The length is measured along the line parallel to the longitudinal
axis, and the longest distance is used. The width is measured along a line perpendicular
to the long axis. • Measurements that may be variously interpreted are explained: The skull length is the distance between the tip of the snout and the posterior
edge of the parietal.
The lower jaw length is measured from the symphysis to the posterior end ofthe
retroarticular process.
The width of the skull is measured at the widest point across the skull, at about
two thirds of the length of the lateral dm of the orbit.
The neck length is measured from the base of the skull to the anterior margin
of the pectoral girdle.
The length of the trunk is the distance from the anterior margin of the pectoral
• 53 • •
GXDB35002 V7917 V7918 GXD7602 BPV601 V953 GXD7603 V7919 GXD7601 V952 GXD7613 GXD7621 GXD83802B Max/Min 1 '.0 10.9 - 15.5 16.5 18.0 19.1 17.5 24.4 2 '.5 IB.8 23.0 23.7 22.6 27.8 3 4.7 6.4 B.l - 9.3 10.0 ,.4 12.7 2.70 4 3.2 4.2 4.' 6.6 6.0 6.0 6.5 5 11.3 30.5 56.5 65.9 69.3 60.7 83.2 85.2 7.54 6 13.9 29.6 43.7 52.9 61.2 - 67.4 66.4 77.1 81.6 90.5 6.51 7 13.5 36.9 B '.4 23.3 33.7 41.9 53.4 56.4 51.2 62.1 64.5 71.2 7.57 • 34.2 71.0 124.9 143.6 154.7 146.2 10 3.7 7.1 '.7 13 .3 20.8 17.4 18.4 - 26.0 27.2 27.7 7.49 11 5.0 '.5 12.9 17.4 24.8 24.3 23.3 26.8 30.8 34.8 6.96 12 0.5 2.3 2.' 3.3 3.5 3.B 7.6 13 1.3 - 3.5 -- - 3.2 2.46 14 0.5 2.5 3.0 - 4.0 3.5 4.5 • 15 1.3 3.' -- 3.5 2.69 16 0.7 - 3.0 3.2 - 4.3 3.B 4.B 6.86 17 1.3 - - 4.2 3.23 lB 0.6 - 2.B - 4.0 4.0 4.B B 1.3 - 4.5 3.46 20" - - 3.5 - - 4.0 4.1 4.5 21 - 2.4 22 2.6 7.2 10.5 11.0 14.3 13.0 13.9 - 16.2 16.5 19.2 7.38 23 2.0 S •• '.3 10.1 14.5 16.3 15.0 20.0 - 25.0 25.6 28.8 14 .4 24 0.6 1.0 2.0 2.B 3.4 3.2 4.2 5.1 5.2 5.4 9 25 0.5 O., 1.1 2.0 2.B 2.B 2.B 3.4 3.0 3.7 7.4 26 0.6 1.1 2.2 2.5 - 3.B 3.' 5.0 - 6.4 6.5 B.O 13.33 27 1.0 2.' 4.5 5.0 - 7.5 7.0 9.2 - 11.5 :l1.S 13.1 13.1 28 0.6 1.5 2.1 2.6 - 4.1 3.0 5.3 -- 5.7 5.5 7.4 12.33 2' 1.2 3.0 5.0 4.7 - 7.5 7.1 10.0 12.1 12.9 14.2 Il.83 30 - - - O., 1.0 1.2 - 1.6 2.0 2.5 Table 1 Measurements of Keichousaurus hui • •
GXD835002 V7917 V7918 GXD7602 BPV601. V953 GXD7603 V7919 GXD7601 V952 GXD7613 GXD7621 GXDB38028 8028/5002
31 - - 1.2 - 2.0 2.0 2.0 - 3.0 3.1 4.0 32 0.4 --- 2.5 2.3 3.3 - 4.0 4.1 4.0 10 33 0.5 1.1 - 3.2 3.1 3.7 5.2 5.0 5.3 10.6 34 1.5 -- 4.0 3.3 4.0 4.5 5.2 5.5 35 - 1.4 -- 3.1 3.1 3.7 - 4.4 4.7 5.2 36 1.1 --- 2.9 2.7 3.1 - 3.9 4.0 4.8 37 - 5.5 - 4.7 38 2.0 - 6.0 7.0 7.0 39 3.1 5.7 7.2 7.5 40 3.5 5.9 6.0 7.3 41 - 2.5 4.2 4.5 42 2.5 6.2 9.5 11.3 13.0 15.5 14.1 16.5 - 18.8 20.9 21.6 8.64 43 0.5 1.6 2.3 3.5 - 3.5 3.9 4.1 5.1 6.0 6.9 13.8 44 0.2 0.6 1.5 1.6 - 1.2 2.0 1.5 - 2.5 2.5 3.0 15 45 0.7 0.8 2.0 1.9 2.1 3.1 2.2 2.8 3.5 4.8 6.86 46 0.8 2.2 3.4 4.0 - 5.1 5.0 6.0 7.5 7.5 8.0 10 47 1.1 2.5 4.2 4.7 - 5.5 6.0 7.0 - 8.0 8.2 9.9 9 48 - 1.0 1.1 1.5 1.5 1.9 1.7 49 1.7 1.8 2.3 2.3 - 2.5 2.8 50 0.4 1.0 - 2.0 3.0 2.3 2.5 3.3 3.3 51 0.6 1.7 2.5 - 4.5 4.1 4.5 5.3 5.5 52 0.7 2.0 - 3.2 - 5.0 4.6 5.0 5.9 6.5 53 0.7 2.0 /!.2 5.0 4.8 5.0 5.9 6.5 54 0.7 1.5 3.4 4.3 4.2 4.2 - 5.2 6.0 55 2.5 3.6 3.5 3.0 3.8 4.0 56 3.1 4.6 - 5.8 4.7 5.0 -- 6.8 57 - 4.0 6.5 7.5 6.0 8.5 58 6.5 - 7.5 8.8 8.0 8.5 59 - 5.0 6.5 7.2 - 9.0 Table 1 Continued List of measurements
1 skull length at midline 47 length of fibula • 2 length of lower iaw 48 length of calcaneum 3 width of skull 49 length of astragulus 4 length of orbit 50 length of 1st metatarsal 5 neck length 51 length of 2nd metatarsal 6 trunk length 52 length of 3rd metatarsal 7 taillength 53 length of 4th metatarsal 8 glenoid-acetabulum distance 54 length of 5th metatarsal 9 ~nout-vent length 55 length of Ist toe JO width between glenoids 56 length of 2nd toe II width between acetabulae 57 length of 3rd toe 12 length ai post~rior cel vical 58 length of 4th toe vertebrae 59 length of 5th toe 13 width of posterior cervical vertebrae 14 length of anterior dorsal vertebrae 15 width of anterior dorsal vertebrae 16 length of posterior dorsal vertebrae 17 width of posterior dorsal vertebrae 18 length of sacral vertebrae 19 width of sacral vertebrae 20 length of anterior caudal vertebrae 21 width of anterior caudal vertebrae 22 standard length 23 length of humerus 24 width of humerus at the proximal end 25 width of humerus at the middle of the shaft 26 width of humerus at the distal end 27 length of ulna 28 width of ulna 29 length of radius 30 length of ulnare 31 length of intermedium 32 length of 1th metacarpal 33 length of 2nd metacarpal 34 length of 3rd metacarpal 35 length of 4th metacarpal 36 length of 5th metacarpal 37 length of Ist digit 38 length of 2nd digit 39 length of 3rd digit 40 length of 4th digit 41 length of 5th digit 42 length of femur 43 width of femur at proximal end 44 width of femur at the middle of shaft 45 width of femur at distal end • 46 length of tibia girdle to the posterior margin of the pelvic girdle. • The snout-vent length is the sum of the skull length, the neck length and the trunk length.
The standard length is the length ofthe last four dorsal vertebrae (Sander, 1989).
This measurement was used as a substitute for the length of the trunk or the
glenoid-acetabular distance to compare the relative size ofdifferent individuals, because
the other two measurements depend on the position of the pectoral girdle, which is not
anchored to the vertebral column.
The proportions of different parts of the body, which appear in various parts of
the discussion, are listed in Table II. •
• 55 • •
GKù835002 V/917 V/918 =602 B?J601 V953 =603 V/919 GXD7601 V952 =617. =621 GXD838028
trunk/g-a 1.48 - 1.26 1.20 1.30 1.24 1.27 1.27 trunk/std 5.35 - 4.81 - 4.92 4.78 - 4.76 4.95 4.71 skull/tnmk 0.65 0.37 - 0.29 0.27 - 0.24 0.29 skull/std 3.46 - 1.41 1.17 1.37 skull/g-a 0.96 -- 0.37 -- 0.28 0.37 neck/tnmk 0.81 1.03 - 1.07 1.08 - 1.03 0.91 - 1.08 0.94 neck/g-a 1.20 -- 1.35 1.23 1.19 - 1.34 1.20 ned<:/std 4.35 -- 5.14 -- 5.06 4.37 -- 5.14 - 4.44 wid/tnmk(g) 0.27 -- 0.25 -- 0.26 0.28 - 0.34 0.33 0.31 widitnmk(a) 0.36 0.33 - 0.36 0.35 - 0.35 0.38 0.38 hun/tnmk 0.14 0.20 0.21 0.19 0.24 - 0.22 0.30 -- 0.32 0.31 0.32 bun/g-a 0.21 - 0.24 - 0.31 0.27 0.39 - 0.40 0.40 0.40 bun/std 0.77 - 0.92 - 1..14 1.09 1.44 - 1.54 1.55 1.50 wid/len(hun) .0.25 -- 0.20 - 0.17 0.19 0.14 - 0.14 0.12 0.13 p=>
Table II Proportions of measurements of Keichousaurus hui • SEXUAL DIMüRPHI8M
Sexual dimorphism - the morphological and morphometrical differences between
the members of the two sexes - is a common phenomenon among modern reptiles. It
manifests itself in size, shape, ornamentation and coloration differences between the
sexes. There is no reason to believe that the same was not true in the pas!. However,
because of the scarcity of specimens in mast fossil species and the fact thaf. only hard
parts of animais preserve as fossils, sexual dimorphism is not weil documented in
extincl taxa. One of the rare exceptions is the Monte San Georgio pachypleurosaurids
studied by Sander (\989). Because of the large number of specimens available (over • 200 prepared specimens), he was able to perform statistical analysis as weil as morphological study of sexual dimorphism.
Sexual dimorphism of Alpine pachypleurosaurids is demonstrated mainly in the
morphology and relative length of the humerus. The morphometric dimorphism is
shown in the ratios of measurements within the humerus, and that of the humerus vs.
other parts of the body. Sander (1989) used the following quantities: the ratio of the
distal width:minimal diameter of the humerus shaft; the ratio of length of
humerus:femur; the ratio of length of humerus:trunk and the ratio of humerus:standard
length.
Because it is impossible to determine the actual sex of the two morphotypes,
• 57 they were denoted sex x and sex y. • In sex x, the humerus is about the sarne length as the femur. In the case of small pachypleurosaurids (i.e., the Iwo specie.; of Neusticosaurus and Serpianosaurus), the
mean of the humerus:femur ratio is about 0.95; the surface of the bone in smooth and
featureless. The expansion of both ends of the humerus is not pronounced. On the
other hand, the humerus of sex y is highly developed. The humerus is signiflcantly
longer than the femur; the mean of the humerus:femur ratio is 1.11 in the small
pachypleurosaurids. The muscle scars and crests are 0 bvious. Both ends of the
humerus, especially the distal end, expand prominently. Neusticosaurus edwardsi is
separated from other species not orny by ils larger size, but also the body proportions.
The mean humerus:femur ratio for sex x is 1.54, and that for sex y is 1.71. This was
attributed to the shortness of the femur (Sander, 1989). However, the fact that the • humerus is very long is also demonstrated by the humerus:standard lengili ratio - \.23 for sex x and 1.32 for sex y (they are 0.86-\.09 and I.l0- 1.28 in small
pachypleurosaurids).
Although there are only data for 13 specimens in Keichousaurus hui, sexual
dimorphis:n is nevertheless apparent. To facilitate the comparison with the European
species, the sarne set of ratios is calculated for Keichousaurus hui.
To sex the specimens, we have to use the individuais that are mature. The
growth ofKeichousaurus hui will be discussed later. For now, we will use the relative
length of humerus:femur as an indicator of maturity. Among the specimens,
GXD835002 is obviously an embryo or a hatchling, and ils sex could not be • 58 determined. Specimens V7917, V7918 and GXD7602 have humeri shorter than the • femora, and may be considered subadults. The remaining specimens have humeri longer than the femora. Specimens V953, GXD7603 and BPV601 have a
humerus:femur ratio of 1.05 to 1.16; the humerus:standard length ratios are 1.14 for
V953 and 1.09 for GXD7603. While this ratio for BPV601 is not known, the
humerus:trunk ratio (0.24) is comparable to that of GXD7603 (0.22). These three
specimens represent sex x. Specimens V7919, GXD76I3, GXD7621 and GXD838028
are apparently of sex y with humerus:femur ratios between 1.21 to 1.33
(humerus:standard 1.43 to 1.55). The humerus:trunk ratio is 0.30 to 0.32. One may
notice that sex y specimens are ail larger than sex x specimens, and thus suspect that
the morphological difference of the humerus was, instead, age related. However, the
studies done by Rieppel (I989) and Sander (I989) shows that sexual dimorphism is a • common phenomenon in pachypleurosaurids. It is more possible than not that sexual dimorphism also occurs in Keichousaurus hui.
Sex x of Keichousaurus hui, like those of small European species, has slender
and smooth humeri. The expansion of both ends is not pronounced. The ratio of the
width at the proximal end versus that at the mid-point of the humerus is 1.21 for V953
and 1.14 for GXD7603. The ratio of the width of the distal end versus the midpoint
is 1.36 and 1.39, respectively. In sex y, the humerus is highly developed. The ratio of
proximaI:midpoint is 1.45 to 1.73; that for distal:midpoint is 1.78 to 2.16. The relative
length - the ratio of the elements vs. the standard length - of other elements of the
forelimb also shows dimorphism. The relative length of the ulna is about 0.52 for sex • 59 x and 0.66 to 0.71 for sex y. The relative length for radius is 0.52 for sex x and 0.75 • for sex y. Apparently, the forelimb as a whole is longer and stronger in sex y than in sex x.
The femur of Keichousaurus hui also shows sorne sexual dimorphism, though
it is not as strongly demonstrated as the humerus. This is also the case in
Neusticosaurus (Sander, 1989).
In living reptiles, sexual dimorphism is usually attributed to selective pressure
related to reproduction, though sometimes it is diet related. The male land tortoises are
often larger than the female, whereas the male aquatic turtles are generally smaller than
the female. Sphenodon has no noticeable size dimorphism between sexes. Male lizards
are often larger but there are many exceptions (Spellerberg, 1982). The actual sex of
sex x and y of Keichousaurlls hui is, therefore, difficult to establish. If Keichousaurlls • was highly territorial, then sex y probably corresponds to male. The larger overall size and stronger arms are definitely advantageous in conquering and defending territory and
impressing potential mates. On the other hand, as argued by Sander (1989), if the
female had to go ashore to lay eggs, a pair of strong and more differentiated arms
would be necessary, and a larger body could store more energy. The differentiated
humerus might also be interpreted as being diet-related. It is not unreasonable to
assume tllat sex y preyed on more active, faster swimming prey than sex x.
• 60 • ONTOGENY
Different ontogenetic stages are presented in the sarnple as demonstrated by the
wide range of body size. The standard length of the largest specimen (GXD838028)
is more than seven times that of the smallest one (GXD835002).
Description of GXD835002
Specimen GXD835002 (Figure 15) is a tiny individual that may be at a late
embryonic stage. The length from the snout to the tip of tail is 48 mm. The only other • possible nothosaur embryo is that of Neusticosaurus (Sander, 1988 and 1989), with a body size of 51 mm. Unlike the Neusticosaurus embryo, which is preserved in a
curled-up posture in lateral view, GXD835002 is exposed in dorsal view similar to the
adults.
The skull is relatively large, about 25G% longer than the standard length,
whereas in adults, the skull is about 50% longer than the standard length. The Icngth
of the orbits is two-fifths of the length of the skull. The preorbital portion of the skull
is short. The frontal is single even at this stage. The pineal opening is huge. The
diarneter of it is about 50% of the length of the temporal openings. A huge
supraoccipital can be seen in the occipital region, between the squarnosals. • 61 •
Figure 15 An embryo of Keichousaurus hui, GXD835002.
scale x6
• 62 •
• The vertebrae are short. The neural arches are only 50% of their width. The • neural spines are not developed. The neural arches of the two sides are not coossitied yet, and are usually separated.
The morphology of the scapula is similar to that of adults. Both ends of the
long bones are concave, indicating cartilageous epiphyses. The length of the humerus
is only 76% of the standard length, and 80% the length of the femur. The ulna is
slightly shorter than the radius, and twice as broad. The carpals are not ossified yet.
Only two metacarpals (1 and II) are preserved.
Three elements of the left pelvis can be seen. The pubis and the ischium arc
broad, but the extremities are not ossified so that the sutural articulation of the three
elements seen in the adult stage is not yet developed. The symphysis between the two
sides must also have been cartilageous. The ilium does not differ much l'rom that of • the adults. The femur is slender with both ends expanded slightly. Ils length is 96% of thc
standard length. The tibia and the fibula are similar tp those of the adult. The tarsal
are not ossified in this stage.
The five metatarsals of the right foot are weil preserved. As in the adults, the
first metatarsal is the shortest, only hall' the length of the second metatarsal and one
third ofthe length ofthe fourth, which is the longest. The fourth metatarsal is also the
thickest of the five.
The digits on the feet seemed better ossified than those on the hands. On the
third, fourth and fifth digits, at least two segments were ossified when the animal died.
• 63 The ends of the long bones are not fini shed. This is expected for such an early • developmental stage. The proximal end of the sacral ribs are slightly thicker than that of the dorsal
ribs. The distal end is blunt. The first pair of caudal ribs also converge toward the
ilium, but the distal end is pointed.
Ali the dorsal ribs bear a longitudinal groove at their dorsal side. This is not
seen in the adults. The ribs are thinner compared to those of Neusticosaurus embryo
(cf. Sander, 1989, Fig. 33, p.640) and pachyostosis is not apparent. About six caudal
ribs are preserved at the left side. The count of the caudal vertebrae is 34, including
7 impressions.
The gastralia are already present in this stage of development.
• Tite developmental stage of GXD83S002
The tiny specimen ofNeusticosaurus peyeri, similar in size to GXD835002, was
identified as an embryo by Sander (1988, 1989) based on its posture and size.
The posture of GXD835002 does not differ much from the other specimens of
Keichousaurus hui. Although the skull and the tail curl to the same side of the body,
the specimen is preservedin dorsoventral view, whereas the Neusticosaurus embryo
was preserved in lateral view.
In reptiles the size of the hatchlings is closely related to that of the adults
(Andrews, 1982; Currie and Carroll, 1984). If the adult size is known, the hatchling • 64 size can be predicted using a power function in the form of
• • p (1) • sizehatchling-lX XSlzeadult
Because the total length of most of the specimens of Keichollsallrlls hui is
unknown, the formula derived by Currie and Carroll (1984) could not be used since it
relates the total length of hatchlings and adults. Andrews' (1982) formula uses the
snout-vent length instead of the total length. Andrews calculated the power and the
intercept of the growth !ine for lizards, snakes, crocodiles and turtles separately. From
Andrews' result (Andrews, p.281) it is clear that lepidosauromorphs (lizards and snakes)
have a very different hatchling-adult size ratio across the size range than do turtles and
crocodiles. The term fI is 0.74 and 0.76 for lizards and snakes, respectively, but 0.20
for turtles and 0.24 for crocodilians. Because sauropterygians are probably more • closely related to lepidosauromorphs than to crocodiles (not to mention turtles), it is assumed that the formula for the lizards and snakes would be more appropriate in
determining the hatchling size of Keichousaurus hui from their adult size. Of the two
lepidosauromorphs, the !izards are closer to the general proportions of the
sauropterygians, so the formula for the !izards may be the best choice in predicting the
hatch!ing size.
Here the adult size is defined as the mean snout-vent length of sexually mature
individuals. Sexual maturity is achieved in modem reptiles when there arc oviducal
eggs found in females or motile spermatozoa found in testes or the efferent ducts of
males. Since eggs and spermatozoa are never preserved in fossils, only the secondary • 65 sexual characters can be used to determine the sexual maturity of palaeonto!ogical • material. In the case of the Alpine pachypleurosaurids, the first appelllance of sex y is used as evidence to indicate the onset ofsexual maturity (Sander, 1989). However, as
mentioned in the previous section, this is not suitable in our situation. In the present
study, the proportion of the humems vs. the femur is used as the indicator of maturity.
Ifthis turns out to be a misinterpretation, then we are more likely to underestimate than
overestimate the snout-vent length of mature animals. Another factor that leads to
underestimation is that there are only t..lrree adult specimens in which the snout-vent
length can be measured and they are ail of sex x, the smaller of the two sexes. As we
shall see, however, these factors, even without correction, will not aff'~ct our conclusion.
The mean of snout-vent length of the specimens with the maturity indicator
greater than one is used to estimate the mean adult size of Keichousaurus hui. From • this value (147.5 mm), the expected hatchling size is calculated using Andrews' formula. Comparing our probably underestimated value (46.7 mm) with the snout-vent
length of GXD835002 (34.2 mm), it is apparent the specimen GXD835002 is too small
to be a hatchling.
The fact that the specimens of different developmental stages were found at the
sarne locality in the sarne type of sediments suggests that Keichousaurus hui was
probably ovoviviparous; otherwise the egg would have been laid in a terrestrial
environment, and in that case, the embryo would almost certainly not have been
preserved. However, there is no direct evidence to support ovoviviparity, as is the case
in ichthyosaurs, where embryos were found inside the abdominal cavity of the adults.
• 66 • The ontogenetic stage of V7917 The second smallest specimen in this sample is V7917 (Figure 16). lt has a
snout-vent length of 71 mm, a little more than twice the size of GXD835002, and about
152% of expected hatchling size. Most small reptiles in the modem fauna double their
body size during the first year of their life (Currie and Carroll, 1984). lt is reasonable
to assume that V7917 was less than one year old when it died.
At this stage of development, most of the bony elements are already present.
The ulnare, however, was not yet ossified.
The proportions of the skull are similar to those of the embryo. The orbits are
huge. The preorbital part of the skull is short. There is a big gap between the parietal
and the postfrontal. Il seems that the ossification of the parietal is not complete. Even • though this gap may be attributed to the postmortem damage, it still indicate that this is a poody ossified area. The skull table of adult specimens is usually complete.
The skull of V7917 is 10 mm long, only 21% longer than that of the embryo.
The other parts of body, however, grew faster compared to the skull. The ratio of
standard length in V7917 vs. the embryo is 277%. That of the neck is 270%. The tail
of V7917 is 273% that of GXD835002.
The humerus is shorter than the femur. The whole length of the forelimb is
about 70% ofthe glenoid-acetabular distance. The elements of the foot are longer and
much stronger than those of the hand.
• 67 •
Figure 16 A juvenile Keichousaurus hui, V'I917.
sca1e x3
• 68 ••
Il (j gIl ~ i • \ • Subadult specimens The skull of specimen V7918 is not preserved, hence the snout-vent length is
not known. However, an estimate can be made. The ratio ofthe length ofthe neck vs.
the length of the trunk is 0.9 to l.l in other specimens with both neck and trunk
preserved. Assume it is 1.0 for V7918, then the length of the trunk plus the neck
would be over 87 mm. If the skull length is at least the same as V7917, then the
snout-vent length of V7918 would be at least 98 mm, which is more than double the
expected hatchling size. This suggests that V7918 was a little more than one year old
at the time when it died.
Specimen GXD7602 is slightly larger than V7918. The snout-vent length is
about 125 mm. The skull is 15.5 mm long, about 141% of standard length. • In both ofthese two specimens, the ulnare is present. However, the shape ofthe bone is irregular, which means that ossification is not complete yet. The distal
metacarpals are not ossified at this stage.
The forelimbs are still weaker than the hindlimbs at this stage. Their lengths are
70% the glenoid-acetabular distance, as in V7917.
The adults
In the adults, the parietal opening tends to close up. In specimen GXD7601 the
parietal opening is reduced to a tiny opening, only 0.3 mm in diameter. In the slightly • 69 larger specimen, V952, the parietal opening is completely closed. • The snout of the adult Keichousaurus hui is relatively longer than that of the juveniles. The size of the orbits does not change. The ratio of the length of the
skull:standard length is s!ightly lower than in the juveniles.
Three size landmarks on the growth of a reptile are recognized: the hatching
size, the size at sexual maturity and the maximum size (Andrews, 1982). Thc
relationship ofthe first!wo landmarks has been discussed ear!ier. From the data of the
modern !izards (Andrews, 1982, Appendix 11), we see that the ratio of size at sexual
maturity vs. maximum size varies between 1.16 and 2.00. The standard length of the
adult specimens ofKeichousaurus ranges from 13 mm to 19 mm, which means that the
largest specimen in the collection under study is 50% larger than the smallest "adult"
specimen. This is in accordance with the data of modern !izards. • The humerus is not only longer, but also stronger than the femur, especially in sex y.
The fourth distal carpal and tarsal are present in ail the sex y specimens, except
V7919. In the largest specimens, such as GXD7613 and GXD838028, there are two
more distal carpal elements: the second and the third distal carpals. The sequence of
carpal ossification is fully in accordance with that of the primitive eosuchian (Carroll,
1985; Caldwell, in press).
• 70 • Allometric growth ofKeichousaurus hui The last column of Table 1 lists the ratios of different measurements in
GXD838028 and GXD835002. The most striking phenomenon i~ the extent of the
diversity among different parts of the body. During growth, the width of the skull
increases I.7 times. The length of the neck increases 6.5 time during the same period.
The length of the humerus of GXD838028 is 14.4 times that of the embryo. It is
apparent that the rates of growth of different parts of the body are different. This
phenomenon is termed allometrie growth.
The word allometry was tirst adopted by Huxley and Teissier (1936, see
Blackstone, 1987) to describe the phencmenon in which functionally similar structures
may have different proportions due to differences in size. One of the familiar example • of allometry concerns the legs of a mouse and the legs of an elephant. The function ofa particular part of the body might be inferred through a1lometric analysis.
Allometric analysis can be performed either in an interspecific basis, in which the same
part of the body of the same age group (usually adults) of different but related species
are studied, or in an intraspecitic basis, where different age groups of the same species
are .tudied. An example of the former approach is the function of the 'sail' of
pelycosaurs (Romer, 1940). In the present study, the second approach is used.
In the study of allometry, the relationship of different parts of the body x and
y is usually expressed using a power function
• 71 (2)
• where b is called the allometric coefficient. The value of the allometric coefficient depicts the relation~hip of proportions to the size. When b is greater than 1, then the
proportion y:x is larger in larger animais and allometry is said to be positive. If b is
less than 1, then y:x is smaller in larger animais; this is called negative allometry.
Isometry is the situation in which b is equal to 1 and the proportion y:x is the same
regardless the size.
It is clear that allometry is relative: when part y shows positive allometry relative
to part x, then part x shows negative allometry relative to part y. To decide the growth
rate of different part of the body, a measurement has to be seleeted to represent the
standard body size. The possible candidates could be: snout-vent length, trunk length,
glenoid-acetabulurn length and standard length (the length of the last four dorsal
• vertebrae). The shortcoming of the snout-vent length is that skull length, which is
negatively allometric to almost every other body measurement, is included. It would
be unreliable as a measure of the growth of the animal. Also, not ail of the specimens
include the skull. The trunk length and the glenoid-acetabulum distance have the
problem ofdepending on the position ofthe pectoral girdle. Because the pectoral girdle
is not anchored on the vertebral column and Hable to postmortem disturbance, we
cannot rely of the trunk length or the glenoid-acetabulum distance as an accurate
indicator of the body size. Although the standard length was arbitrarily defined as the
length of the last four (instead of three or five or any other number) dorsal vertebrae
(Sander, 1989), it has the advantage ofbeing a quite accurate indicator ofthe body size • 72 and it is relatively easy to measure. • The allometric coefficients of various parts of the body relative to the standard length are calculated using the least squares linear regression method on
log-transformed data (Table II). Actually, the ca:culations were carried out twice, once
including and once excluding the data of GXD835002, the embryo. The significance
tests were performed about the hypothesis Ho:b=l (the growth is isometric), and
Hl :b=1.5 (the underlying meaning of which will be discussed later). The significance
level is set at 95%, double sided.
The growth of the skull is strongly negatively a1lometric. The allometric
coefficient is about 0.5 when the embryo is included, and 0.85 when it is not. The
adjusted ? is the unbiased coefficient of determination of the sample when the sample
is small. Notice that the adjusted ? is greater when the embryo is excluded than when • it is included, indicating that the allometric coefficient may not be constant during development. Initially, the growth of the skull must show high positive allometry,
outgrowing ail other parts of the body, as a result of the size constraints of the
functioning nervous system and related organs. Subsequently, the growth of other body
parts would catch up, turning the allometric coefficient of the skull highly negative.
The immediately postnatal growth rate of the skull is slightly higher than the last
embryonic stage, but it is still negative. The a1lometric coefficient in the first case
(embryo included) can be thought of as an average over the whole life span of the
animal, whereas in the latter case, it is specific for postnatal growth. In both cases, the
allometric coefficients are statistically significantly lower than 1. • 73 • •
# of (hg. lldjusted r- a b Tfor 'l'for HO:b=~ m:b=~.5
Skull ~~ GXD835002 incl. 5 0.<1430 5.468 0.444 5.937 - GXD835002 e>
Table m AIIometric coefficients of various body parts vs. the standard length When relative growth rate is more nearly constant, the coefficient of • determination would be proportional to the sample size, which is the case for the data ofthe neck length. Although the estimates of b are very similar and close to 1, we can
reject the hypothesis that the true value of b was 1.5 when the embryo is included in
the analysis, but we cannot do so when the embryo is excluded. The same also applies
to the trunk length and the snout-vent length.
The allO;1.1etric coefficients for neck length, trunk length, and glenoid-acetabulum
distance are all very close to 1, especially when the data for the embryo are included.
The body width at the glenoid and at the acetabulum show the same pattern as
the skull: the overall allometric coefficients are very close to 1, but they are much
higher when the data of the embryo are excluded from the calculation. The adjusted
,:z are greater in the latter cases. • The allometric coefficients for the forelimb elements are much higher than those for their counterpart in the hindIimb, whether the data for the embryo are included or
not. The humerus, the ulna and the third metacarpal have allometric coefficients
significantly higher than 1 -- closer to 1.5. The allometric coefficient for the third (the
longest) digit of the hand (and the foot) is much lower than the other Iimb elements.
This may refiect the true situation, or it may be due to the loss of the last segment
during preservation.
• 75 • LOCOMOTION
It is weil established that sauropterygians were aquatic animais. The aquatic
locomotion of the more advanced group of sauropterygians, the plesiosaurs, and their
possible muscle arrangement have been pondered upon since the tirst published
description of a plesiosaur in the 1820's. For historical reviews, see Robinson (1975),
Godfrey (1984) and Storrs (1993).
It is agreed that the primary swimming apparatus of plesiosaurs were their two
pairs of limbs. Judged from the reinforcement of the limb girdles along the midline,
they must have used symmetrical strokes to propel the rigid body in the water. • Recent years have seen an increased interest in the study of the functional morphology of the less specialized groups of the sauropterygians. Carroll and Gaskill
studied Neusticosaurus edwardsi (1985), Rieppel studied Serpianosaurus (1989), Sander
studied Neusticosaurus peyeri and N. pusillus (1991), Storrs studied Corosaurus (1991).
Schmidt (1986,1989) and Storrs (1993) discussed the functional anatomy of primitive
sauropterygians in general.
The pectoral and pelvic girdles of plesiosaurs and nothosaurs are functionally
similar in that they favour symmetrical movement of the limbs. The sternum that
existed in the primitive diapsids has disappeared. (For a different opinion see Nicholls
and Russell, 1991). In modern lizards the function of the sternum is to facilitate the
• 76 alternative movement of the forelimbs (Jenkins and Goslow, 1983). Preswnably, the • same structure would have had a similar function in primitive diapsids, and the lack thereof in the sauropterygians indicates that the alternative movement of the forelimbs
was not so structurally facilitated. AIso, the scapular "blade" ofthe pectoral girdle and
ilium of the pelvic girdle were reduced to rod-like structures. Meanwhile, the ventral
portion of the girdles was elaborated.
The degree of specialization of the girdles and limbs was different in the two
groups, however. In plesiosaurs the scapulocoracoids are massive. They meet in the
ventral midline of the body and extend posteriorly a great distance. The limbs of
plesiosaurs are highly specialized structures. The distinctions between fore- and
hindlimbs have almost disappeared. The propodials are massive elements. The
elements of the epipodials and beyond have ail become roundish bony disks. The • manus and pes exhibit hyperphalangy and the digits are tightly packed. The outline was wing-shaped with a broad base and a tapering end when the flesh was in place
(Robinson, 1975).
In primitive sauropterygians, the ventral expansion of scapula is minimal, and
the coracoids are not elongated anteroposteriorly. The ventral view of the pectoral
girdle is more or less a circle with a large fenestration in the rniddle, which might at
least be partially covered by the cartilage in sorne taxa when the animal was alive
(Schmidt, 1987). Another difference between nothosaurs and plesiosaurs in the
configuration of the pectoral girdle is that the scapula "horn" ofthe nothosaurs is placed
right above the glenoid, whereas in plesiosaurs, the scapula horn is situated weil
• 77 forward relative to the glenoid. Since it has been emphasised (Robinson, 1975) that the • latter is an adaptation to raise the arms, it follows that the placement of the scapula horn in nothosaurs would gready limit the extent that the hurnerus could be lifted.
Therefore, underwater flying was impossible for nothosaurs. (Godfrey (1984) argued
that underwater flying in sea turtle fashion as proposed by Robinson (1975) is not
feasible even for plesiosaurs.) The same argument can be applied to the pelvic girdles
as well. The outline of the nothosaur limbs is more fan-shaped than hydrofoil-shaped.
The propodials of nothosaurs are slim compared to those of plesiosaurs and the
proportions among the limb segments are not far from the primitive diapsid condition.
The fingers and toes usually spread out in fossils.
The general similarities of the pectoral and pelvic girdles of plesiosaurs and
nothosaurs indicates the similar muscle distribution in both groups, i. e. reduced dorsal • components and elaborated ventral elements. The different degrees of elaboration of the ventral elernents in the two groups may be partly due to different functions of the
limbs.
Function 01 the lorelimbs ofKeichousaurus hui
Among the nothosaurs, pachypleurosaurids and nothosaurids can be
differentiated by their body size and degree of specialization.
The hurneri of nothosaurids are stouter than those of pachypleurosaurids, the
epipodials are wide because of the large interosseous space between ulna and radius, • 78 giving the forelimbs the outline of an oar. It is apparent that the symmetrical stroke of • the forelimbs was an important component in the locomotion of nothosaurids. Pachypleurosaurids have slimmer humeri and, in general, the epipodials are not
expanded. They usually have a long tail with high neural spines. They are usually
considered lateral undulatory swimmers (Carroll and Gaskill, 1985; Rieppel, 1989;
Sander, 1989; but see Storrs, 1993)
Modern diapsids (iguanas, crocodiles and snakes) tend to employ lateral
undulation when they swim, which is quite natural because of their locomotion mode
on land. This is also exemplified by primitive aquatic eosuchian Hovasaurus (Currie,
1981).
Keichousaurus hui differs from European pachypleurosaurids in having a
stronger humerus, very broad ulna and slight hyperphalangy ofthe manus. The profile • of the forelimbs is more paddle shaped than in other genera. These differences indicate a more important role of the forelimbs in locomotion of Keichousaurus hui than in
other pachypleurosaurids.
Allometric analysis of the growth series supports this conjecture. Assume that
the locomotor pattern did not change during the life time of Keichousaurus hui (or of
other pachypleurosaurids), then the functions ofthe forelimbs would have remained the
same during growth. Ifthe forelimbs were not heavily used in locomotion, their growth
should be close to isometric, if not negative allometric, relative to the body length, as
exemplified by the Neusticosaurus edwardsi (Table 13 in Sander, 1989). Ifthe limbs
were heavily used, positive allometric growth would be expected. The expected • 79 allometric coefficient can be approximated as follows: • Since the forelimbs of Keichousaurus hui could not be used as underwater wings because ofthe structure of the shoulder girdle, it is only reasonable to assume that they
were used as paddles. If Keichousaurus did use their forelimbs as paddles in
locomotion, then when they swam in a constant speed, the drag force produced by
paddling should be equal to the drag force acting on the body surface. The drag force
acting on the body surface is proportional to the product of the swimming speed
squared and a characteristic length of the body squared, the second term, in this case,
is the standard length. This can be expressed as:
(3)
where D is the drag force, V is the speed and L is the characteristic length. • Since swimming speed cannot be estimated from the fossi! record, we have to turn to data of modern aquatic vertebrates to estimate swimming speed of
Keichousaurus. Even for modern aquatic vertebrates, however, the data on swimming
speed are sparse. But from what we have (Figure I7), it is observed that the active
swimming speed of aquatic vertebrates is proportional to the square root of the body
length:
(4)
• 80 •
Figure 17 Swimming speed vs. size of modern aquatic vertebrates.
(from McMahon & Bonner, 1983)
• 81 el i, •
Speed plotted against length for swimmers of intermediate and large size. The slope of the line marked Active is 0.5, showing that Barracuda the ·speed of sustained swimming is propor .,. Bluewhale tional ta the square raot of length. The 10 El Dolphin ,.1> slope of the braken line marked Burst is Burst; slope = 0.88 e Porpoise " O.BB. Because the slopes of the two Iines " -'" are very different, the larger fishes in this , " " series are able ta increase their swimming TroU[ • " speed by a considerably larger factor than ~ --" ~ ~;'~" the smaller ones cano The vertical bars " "" ... marked with letiers correspond ta: C. cod; E " ..... -:; .; Active; slope = 0.5 ~ F, flounder; G, goldfish; H, herring; P, ~ pout; R, redfish; S. sculpin. ~ 1~
FT.. , J.rllIILarval anchovy
0.1 9l ' 1 1 1 1 1 1 III, 1 1) 1 Il! 1 fIl 1 ! 1il!1 1 0.01 0.1 1 10 Length (m) This figure encompasses a body size from a few centimetres to over 20 metres, • and from ectothermic fish to endothermic whales. The body size of Keichousaurus in weil within the range on the figure, and it is reasonable to believe that their metabolic
rate is a:so within the range. Therefore, the speed relaticûship depicted above also
applies to them. Subslitute (4) into (3), we have:
(5)
i. e., the drag force acting on the body surface is proportional to the body mass.
The propulsion force produced by the paddle is proportional to the product of
the characteristic length of the paddle squared and the speed of the paddle relative to
the water squared: • (6) Hydrodynamics diclates that, for the paddle to be efficient, the speed of the
paddle relative to the waler should be as small as possible (Alexander, 1968). The
paddle can be thought of as a lever, and the tip of it is the pivot. The less il slips, the
more the body goes forward. This applies to the adults as weil as the juveniles. If we
assume that its swimming speed is close to constant, then the propulsion force produced
by the paddle is proportional to a characteristic length ofthe paddle squared. Since the
propulsion force produced by the paddle should balance the drag force acting on the
body surface:
Subslitute both sides, we gel: • 82 • (7) (8)
or
l.5 (9) :.Lpaddle ocLbody
In other words, the allometric coefficient of the functional paddles vs. body
length is expected to be 1.5. Recall that the allometric coefficients for most of the
forelimb elements ofour sample ofKeichousaurus are statistically significantly different
from 1, but the hypothesis of them being 1.5, on the other hand, could not be rejected
• (Table II). Il should be noted that the small European pachypleurosaurids (Serpianosaurus, Neusticosaurus pusil/us and N. peyeri) also exhibit positive allometry
in their forelimbs, though not as high as in Keichousaurus. Il is suggested that their
forelimbs are used for terrestrial locomotion (Sander, 1989), but this could not explain
why the type of pectoral girdle of sauropterygians, which is adapted to resist the
symmetrical pressure from the humeri, evolved.
• 83 • Reconstruction of the muscular system ofthe shoulder girdle The structure of the pectoral girdle of sauropterygians differs greatly from that
of terrestrial diapsids, and there is no living forrn with a similar configuration. Any
attempt to reconstruct the muscular system of group can only be an approximation.
Fortunately, study of living tetrapods shows that the muscular system tends to be
conservative, and we can infer from the rugosities on the bony elements the position
of the muscles with reasonable certainty. Among living terrestrial vertebrates, lizards,
with a weil studied musculature system, are most closely related to sauropterygians.
Therefore, the attempted muscle reconstruction is based on lizards. Major references
are Romer (1922) and Jenkins and Goslow (1983).
In terrestrial tetrapods, the thorax is suspended on the supporting structure • forrned by forelimbs and pectoral girdle via the serratus, trapezium and levator scapulae. These muscles originate from the scapular blade and occupy the major part
of the inner surface and the anterior and posterior edges of the blade. In an aquatic
environment, because of the buoyant of the medium, the supportive function of the
girdle and the limb is not as great, and the importance of these muscles is diminished.
As a result, the size of the scapular blade, as weil as the two muscles, are greatly
reduced. Furtherrnore, a less solidly attached pectoral girdle can even increase the
effective length of the power stroke.
The muscles of the forelimbs proper can be divided into two groups, a. dorsal
(levator) group and a ventral (depressor) group, according to their embryological origin • 84 (Romer, 1922). The former includes the latissimus dorsi, subcoracoscapularis, • deltoideus, scapulohumeralis, triceps and extensors of the lower limb. The latter includes the pectoralis, supracoracoideus, brachialis, biceps, coracobrachialis IOllgus
and c. brevis, and the flexors of the lower lirnb. However, since Keichousaurus (and
aIl other nothosaurs) is dorsoventraIly compressed, and the muscle attachrnent surfaces
on the pectoral girdle are more horizontaIly distributed, it might be more appropriate
to group these muscles into abductors and adductors, depending on their position of
origination relative to the glenoid and insertion on the humerus.
There is no evidence as to the size of the latissimus dorsi, because this muscle
originates from the dorsal fascia. The insertion of the latissimus dorsi is at the
convergence of the "Y"-~haped crest at the dorsal side of the humerus. The latissimus
dorsi is the only muscl,~ in the pectoral girdle ofpachypleurosaurids that originates weIl • above the level ofthe gbnoid and would have sorne mechanical advantage when acting as a "levator" of the humerus. In lizards, the origin of this muscle is at and behind the
level of the glenoid, and functions as an adductor of the humerus. This may also be
the case in Keichousaurus.
The subcoracoscapularis originates from the medial surface of scapula and
coracoid, and passes behind the scapular blade to insert between the two upper arms of
the "Y"-shaped crest at the dorsal side of the proximal head of the hurnerus. In sorne
reptiles this muscle can be separated to two slips, a subscapularis and a subcoracoideus.
Since in Keichousaurus and other nothosaurs the scapular blade is smaIl and the
coracoid large, the subcoracoideus would be the larger muscle of the two. Because the • 85 major part of the scapulocoracoid is horizontal, this muscle would work most • effectively when the humerus was in a horizontal or near horizontal plane. Also, since the major part of the scapulocoracoid is behind the glenoid in Keichousaurus, the
function of the subcoracoscapularis would be to pull the limb back to the side of the
body. Therefore, it should belong to the adductor group.
Primitively, the deltoideus originates from the upper part of the scapular blade
(deltoideus scapularis) and the clavicle (deltoideus clavicularis). The insertion is at the
anterodorsal part ofthe humerus head. In Keichousaurus the deltoideus scapularis was
either very small or lost due to the small size of the scapular blade. Because of the
peculiar relation between tiie clavicle and scapular, the anterior portion of the ventral
part of the scapular may share the origin of the deltoid clavicularis with the clavicle.
Since the deltoids originate anterior to the glenoid, they function as abductor of the • humerus. Inferred from !izards, the scapulohumeralis should originate from the lower part
of the scapular blade and insert on the anterior aspect of the proximal end of the
humerus, and function as a levator and abductor of the humerus. However, its origin
and insertion are not apparent in Keichousaurus. If it did not disappear, then it must
have been quite small and insignificant.
The triceps is the major extensor of the lower arm. In terrestrial tetrapods, the
insertion of triceps is on the olecranon of the ulna. In Keichousaurus the lower arm
is always in a extended position and moves in unison with the humerus. The triceps
is therefore basically a static muscle and is not as important as in terrestrial forrns. As • 86 a consequence, there is no muscle scars on the humerus and pectoral girdle where we • might expect to find the origin of the muscle, and the olecranon of the ulna is not conspicuous.
In living lizards, the pee/oralis is a broad sheet of muscle originating from the
clavicle, the interclavicle, along the midline ofthe pectoral girdle including sternum and
sternocostale, and from the midline along the linea alba and from thoracoabdominal
fascia (Jenkins and Goslow, 1983). In Keiehousaurus and other nothosaurs, the sternum
has disappeared and the coracoids meet in the midline. The pee/oralis is very likely to
have shifted part of its origin to the coracoids. The densely packed gastralia may
provide reinforced anchorage for the abdominal muscles and the thoracoabdominal
fascia, and hence stronger support for the posterior part ofthe pee/oralis. The insertion
of the muscle is on the deltopectoral crest, which is a major landmark on the humerus. • The direction ofthe force would be posteromedially, towards the ventral midline of the body. Il is basically an adductor, and when contracting in concert with other adductor
muscles, it would contribute significantly to the power stroke, although the anterior part
of the muscle could act as an abductor.
The supraeoraeoideus is an important muscle in terrestrial reptiles for
maintaining posture by stabilizing the glenohumerus joint The origin of the
supraeoraeoideus in lizards is at the front part of the coracoid, anterior to the glenoid.
The insertion is at the proximal margin ofthe deltopectorai crest of the humerus. Since
the coracoid does not extend anterior to the glenoid in Keiehousaurus, the orientation
of this muscle might shift to a lateral one. The supracoracoid foramen can be used as • 87 an indication of the position of the supracoracoideus. In Keichousaurus the • supracoracoid foramen is situated at the middle of the border between scapula and coracoid, and there should the supracoracoideus originate. If this is the case, then the
supracoracoideus would be much smaller and short:::r than in terrestrial forrns.
Considering that Keichousaurus !ived in an aquatic environment, this is to be expected.
The coracobrachialis longus and c. brevis are the major muscles that function
as the adductor of the humerus. The origin of these muscles occupies the major part
of the external (ventral/lateral) surface of the coracoid in !izards and Sphenodon. The
insertion of these muscles is on the flexor surface of the humerus with the c. brevis
more proximal and occupying more insertion surface. In Keichousaurus the scar for
the c. brevis is prominent on the ventral-medial surface of the humerus. Judging from
the size of the origination and insertion area, the c. brevis could be very strong. • The brachialis and biceps brachii are very closely related to coracobrachialis muscles developmentally, though they are basically forearm flexors. In sorne reptiles
the originations ofthe biceps can hardly be separated from the coracobrachialis muscles
(Holmes, 1977). In !izards, where the sternum is present, the biceps arises along a
narrow area adjacent to the coracosternal joint (Jenkins and Goslow, 1983). In
Keichousaurus, however, the sternum is absent. Therefore, the origin ofthe biceps may
have moved to the mid!ine of the body. In terrestrial forrns, the proximal belly of the
biceps is usually tendinous (Holmes, 1977), and this might be related to the reduction
of the coracoid. 8ince the coracoid of Keichousaurus is enlarged instead of reduced,
the biceps may have had a fleshy proximal belly and contributed to the power stroke. • 88 The distal belly of the biceps and the brachialis, on the other hand, might not be as • important as in terrestrial forms where the precise control of the forelimb is more critical in locomotion and posture maintenance. The function ofthe brachialis and the
distal belly of the biceps is to stabilize the elbow joint.
The muscles of the lower arms are not discussed here because there is no
evidence of their specific distribution. However, it is logical to assume that
Keichousaurus was able to spread out its fingers to increase the surface area during the
power stroke, and pull them together in the recovery phase to reduce drag.
The power stroke starts with the forelimbs forward and outward, horizontal with
the main axis of the body, palm facing downward. The contraction of the adductor
muscles would bring the arms backward with a shallow down curve. At the same time
the forearms would pronate (rotate inward) so that the palms face backward. The • fingers would spread out to maximize the drag surface (with the web between the fingers). In the recovery phase, the levator and the abductor muscles would bring a
supinated arms forward, with palms facing down and fingers collected to reduce the
drag.
Hindlimb and the pelvic girdle
Il is very common for secondarily aquatic vertebrates to hold their hindlimbs
close to the lail to reduce drag and not utilize them for propulsion. This arrangement
can be seen in crocodiles and alligators (Manter, 1940), sea lions (English, 1976; • 89 Godfrey, 1985) and penguins (Clark and Bernis, 1979) arnong living animals, and might • be the case in sorne pachypleurosaurids (Carroll and Gaskill, 1985; Rieppel, 1989; Sander, 1989). Therefore, the pelvic girdle and hindlimbs of secondarily aquatic
vertebrates are usually not weil developed, and the attachment between the pelvis and
the vertebral column tend to loosen. Ifthe hindlimbs were not used for other purposes,
the selective pressure might eventually eliminate them, as is the case in cetaceans. In
plesiosaurs and sea turtles, however, the hindlimbs are strong and as highly developed
as the forelimbs.
As in other pachypleurosaurids, the pelvic girdle and the hindlimbs of
Keichousaurus hui are not as weil developed as are the pectoral girdle and forelimbs.
The femur is slim, but the lower limb is expanded anteroposteriously. The epipodial
is only about 50% the length ofthe femur. The growth ofthe hindlimbs shows a lesser • degree of allometry than the forelimbs, indicating a lesser l'ole in propulsion. The iliac blade is reduced to a small rod, comparable to the reduction of the
scapular blade. However, two lines of evidence indicate that the connection of the
pelvis and the vertebral colurnn is still quite strong. First, the sacral ribs are visibly
pachyostotic, and second, there is a recess at the end of one of the sacral ribs (the
middle, or the "real" sacral rib), which accepts a process in the mid-shaft of the ilium.
The advantage of this unusual arrangement might be to achieve sorne mobility while
maintaining the stability ofthe pelvis. It may have been possible for the pelvis to rotate
around the functional sacral rib to a small degree when needed, such as when the
animal used the hindlimb for propulsion. However, it may have been easy to "lock" • 90 the pelvis into place by axial muscles when the manoeuvre required stability of the • pelvis, such as when the animal was doing a sharp turn and using the hindlimbs as the steering apparatus, or when it needed to stop suddenly. The additional "sacral ribs" in
front and behind the functional one may have functioned as support or reinforcement,
and spread the bracing force along a larger part of the vertebral column. If this is the
case, then those genera with more sacral ribs might be more agile and had more
manoeuvrability.
Muscle reconstruction ofpelvic girdle and hindlimbs
As pointed out earlier, in Keichousaurus the most solid part of attachment ofthe
pelvic girdle to the sacrum is at the middle of the iliac blade. The lateral part of the • dorsal muscles would have been interrupted as in ail vertebrates with a weil developed pelvic girdle. The muscles involved are the longissimus dorsi and iliocostalis in the
trunk, and extensor caudae lateralis and/or abductor caudae externus in the tail (Romer,
1922). The ilio-pubic, pubo-ischiadic and ilio-ischiadic ligaments of terrestrial forms
must also be present in Keichousaurus, though the exact positions of attachment to the
girdle are not certain. The ventral muscles involved are oblique, transversalis and
rectus abdominus in front, and ischio-caudalis in the tail. Since the pelvic girdle may
have heen capable of sorne mobility in Keichousaurus, these muscles may have
controlled the motion of the girdle.
The muscles that originate from the ilium and insert on the femur or the lower • 91 limb are: the iliofemoralis, iliofibularis and iliotibialis. Since the ilium is very small, • the muscles that originate from it may also have been small and weak, and their contribution to the movement of the hindlimb may not have been very significant.
The major muscles acting on the femur originate from the two surfaces of the
ventral plate of the pelvic girdle and associated ligaments. They include the
puboischiofemoralis internus and p. externus, ambiens, adductor and
ischiotrochantericus. Although the caudifemoralis longus and c. brevis originate from
the vertebral column instead of from the pelvic girdle, developmentally they belong to
·t!:Je ventral limb muscle mass (Romer, 1942).
The puboischiofemoralis internus originates from the broad inner (dorsal) surface
ofthe puboischium plate. It runs outward in front ofthe ilium-pubis margin and inserts
on the proximal dorsal surface of the femur. The function of this muscle is to pull the • femur upward and draw it close (0 the side of the body. When acting with the triceps, they would cause the hindlimb to form a large fan-shaped area facing the direction of
the current, thus braking the forward thrust of the animal or change its direction.
The ambiens, together with the iliotibialis and femorotibialis, form the pelvic
triceps. It originates from the antero-Iateral edge ofthe pubis and merges with the other
two heads to insert on the proximal end of the tibia Judging from the preservation
posture of the Keichousaurus specimens where the hindlirnbs are always in a flexed
position, these muscles were not very strong.
The puboischiojèmoralis externus occupies the entire ventral (extemal) surface
ofthe pubo-ischiadic plate. The insertion ofthis muscle is in the intertrochanteric fossa
• 92 at the ventral proximal end of the femur. This muscle is the major depressor of the • femur when working with the adductor, and also contributes to keeping the hindlimb close to the side of the tail when working with the caudifemoralis muscles.
The caudifemoralis longus and c. brevis originate from the underside of the
anterior caudal vertebrae and their associated ribs. In lizards the muscle fibres converge
into a strong tendon passing under the femur and insert into the anterior side of the
proximal end of it. A thinner tendon branches out from the stronger one and runs along
the femur and attaches to the ligaments on the knee joint. The situation would not be
much different for Keichousaurus except that these muscles might be stronger and
function in stabilizing the limb.
The ischiotrochantericus is a small muscle originating from the rear corner of
the ischium and inserting via a tendon on the proximal head of the femur. Ils function • is to stabilize the acetabular joint.
The tait
The European pachypleurosaurids are generally considered lateral undulatory
swimmers with their tail as the major source of power (Carroll and Gaskill, 1985; Sues,
1986; Rieppel, 1989; Sander, 1989; but see Storrs, 1993b). However, there are several
factors that would preclude Keichousaurus from being an effective lateral undulatory
swimmer. The first few caudal vertebrates of Keichousaurus bear long caudal ribs;
their neural spines are low. Therefore, the cross-section at the base of the tail would • 93 show a dorsoventrally compressed outline. It is apparent that such a tail could not • produce much thrust if undulated laterally. If Keichousaurus used its Iimbs in locomotion in a symmetrical fashion as hinted by the structure of the pectoral girdle,
their tail could not be used in a laterally undulating fashion as proposed in other
pachypleurosaurids, because the two patterns of locomotion are incompatible. In lateral
undulatory swimming, undulatory waves travel from head backwards to tail, producing
the driving force. For each cycle of the body wave, there are two phases. The force
produced by the two phases direct alternatively to rear-Ieft and rear-right. The posterior
l:0mponent drives the animal forward. The lateral component is compensated by the
force that points to the opposite direction, which is the lateral component of the force
produced by the other half-phase ofthe undulation. Therefore, the pressure field along
the body alternate constantIy (Manter, 1940). On the other hand, the symmetrical • strokes of the limbs produce a symmetrical pressure field along the two sides of the body. For the symmetrical strokes to be effective, there should exist a symmetrical
pressure field along the two sides of the body. Ifthe pressures at the two sides of the
body were different, the limbs would have to exert different amount of force, thus
destroy the symmetry. In lateral undulatory swimming, the direction of the body is
constantIy changing, and the force produced by the symmetrical strokes could not direct
just posteriorly, but posteriorly and to one side. This, in turn, would destroy the
r"gularly alternating pressure field produced by the undulatory movement, and require
compensation ofthe body. There are two possibilities: the taïl was either trailed behind
passively, or was actively propulsing, but moved dorsoventrally. There is no modern
• 94 reptile that could undulate its tail dorsoventra!ly (Rieppel, persona! communication), and • the innervation of the tail muscles are not weil studied. Therefore, this point is hard to corroborate or disprove. Structurally, the tail ofKeichousaurus is capable ofmoving
vertically because there are gaps between successive neural spines at the base of the
tail. On the other hand, there is no known animal that combines bilaterally symmelrical
propulsion and lateral undulation in locomotion.
The neck
The intervertebral articulations of Keichousaurus hui constitute a complex
functiona! unit The prezygapophysis faces inward and rearward as weil as upward, and
the postzygapophysis faces outward and forward as weil as downward, so the • successive vertebrae are interlocked with each other. In the dorsal region, the postzygapophyses are swollen and fits tightly with the prezygapophyss ofthe following
vertebra, so that there could be very little inter-vertebral movement, and the body is
kept very stiff. In the neck region, however, the zygapophyses are not as swollen, so
that the neck can bend more freely. As shown by Holmes (1989), when zygapophyseal
surfaces of the two sides of the vertebrae are not in the same plane and form an angle
with the longitudinal axis of the vertebral column, coupling of movements tends to
occur. In the case ofKeichousaurus hui, when the neck flexes sideways as the animal
is turning, the cervical vertebrae also tend to bend upwards relative to the vertebra
following it and rotate in such a manner that the neural spine leans towards the centre • 95 of the curve. This leads the body to rotate in such a way that the broad area of the • abdomen would face the outside of the curve. The effect is that the centrifugaI force, which tends to keep the animal moving in a straight line, is now better balanced by the
increased drag force due to increased surface area that faces the outside of the curve.
When birds make a turn, they have to do exactly the sanle thing: rotate so that the belly
and the large area ofthe wings face outward. The difference is that, for Keichousaurus
the mechanism was buili into the skeleton, whereas in birds it is more a learned
behaviour.
The swimming pattern ofKeichousaurus
Subaqueous locomotion of secondary aquatic vertebrates falls into two broad • categories: axial undulatory and appendage oscillatory. Members ofthe first class, such as ichthyosaurs and whales, employ the tail and/or part of the body as the main source
of propulsion. Members of the second class use the paired fins or limbs. The
appendage oscillatory swimmers are further divided into drag based (rowers) and lift
based (fliers) (Webb and Blake, 1985). In rea1ity, however, the situation is not as
simple and clear eut. Sorne swimmers may mix the locomotion patterns. Example of
the mixture of drag based and lift based swimming patterns are sea lion and possibly
plesiosaurs (Godfrey, 1984). Axial and appendage propellers can also be used
simultaneously. One example is human swimmers using the butterfly stroke. In this
style ofswimming, the arms are used in a symmetrical rowing pattern, while the whole • 96 body and legs undulate dorsoventrally. • The locomotion of Keichousaurus hui might also be a mixture of different patterns. The power stroke of the forelimbs was probably drag based. The recovery
stroke might have elements of lift based propulsion. Furthermore, the tail might also
have contributed by performing dorsoventral undulation.
The function of the hindlimbs of Keichousaurus might be similar to that of sea
lions (Godfrey, 1985). When swimming in straight line, sea lions hold their hindlimbs
in a inverted V position. In this position, the hindlimbs act as stabilizers. During turns,
the body of sea lions rotates so that the abdomen faces the outside ofthe curve, and the
hindlimbs fans out posteriorly so that the plantar surface of the hindlimbs would also
face the outside of the curve, preventing the skidding of the posterior portion of the
body. As in sea lions, the hindlimbs of Keichousaurus are also fan-shaped. When • swirnming in straight line, the hindlimbs of Keichousaurus would trail passively along the side ofthe tail. In this position, they would increase the ventral surface at the base
of the tail. When Keichousaurus turned, the bending of the neck would have caused
the body to rotate so that the abdomen and the broad area of the base of the tail faced
the outside. At the same time, the hindlimbs would spread out posterolaterally, adding
more surface to prevent skidding. The hindlimbs of Keichousaurus could also be used
as paddles in concert with the forelimbs when precise manoeuvring was called upon,
such as backing up from a tight place where turning was difficult.
• 97 • PHYLOGENETIC ANALYSIS
When the tirst description of Keichousaurus was published (Young, 1958), he
assigned it to the family Pachypleurosauridae, based on the small body size, the "very
small" upper temporal openings which were described as "smaller than the nasal
openings" and the posterior position of the parietal ol'ening on the skull table. Kuhn
Schnyder (1959) pointed out that the temporal openings were not as small as described,
but large and elongated (which was correct), and thought that the narrowness of the
temporal openings was caused by the lateral compression during preservation (which • was not correct). Based on this and the broad ulna, he put Keichousaurus in the Lariosauridae. Based on the proportions of the fore and hindlimbs, von Huene (1959)
placed Keichousaurus in the Simosauridae. Young (1965), when describing sorne new
materials of Keichousaurus, acknowledged that the temporal openings were elongated,
but asserted that the narrowness of the temporal region was natural. Furtherrnore, he
suggested erecti!,1g a new family, Keichousauridae, within thè suborder
Pachypleurosauroidea.
Recently, several authors have investigated the phylogeny of nothosaurs (Sues,
1987; Storrs, 1991, 1993; Rieppel, 1993a; Sander, 1991; Schmidt, 1987). Among these,
Storrs' 1991 analysis was ideal to serve as the basis for further studies. It included ail
• 98 the weil known (and sorne not so weil known) genera of pachypleurosaurids and • nothosaurids. Eighty-four characters were originally analyzed. It has been used as a frame of reference for further studies of nothosaur phylogeny when new information
was gather~d (Storrs, 1993; Rieppel, 1993). Storrs' study agreed with Young (1958)
in placing Keichousaurus in Pachypleurosauria which, according to that analysis, is
monophyletic. Since many character states ofKeichousaurus were unknown at the time
of Storrs' analysis, and sorne others were incorrectly interpreted, it is necessary to
reinvestigate the phylogenetic position of Keichousaurus in sauropterygians.
The amended data matrix of Storrs (1993), which includes 88 characters of 21
taxa, was used in the current study as the base of the phylogenetic analysis. Since the
computer program package PHYLIP (Felsenstein, 1993) is used in this analysis, the
multistate characters are recoded as two-state characters, resulting in a total of 99 • character transformations. Corrections and modifications mainly concern Keichousaurus.
1. The overall body size. O-small, l-Iarge.
2. The neck length. Storrs coded this character with three states. Since the
third state (neck much longer than the body length) only occurs in plesiosaurs,
collapsing this state with the second state (neck and body length are about equal)
would not affect the resolution of the part of the c1adogram in which we are
interested. O-Iess than body length, l-subequal to or much longer than body
length. • 99 3. The cranial kinesis. There is no kinesis in the Keichousaurus skul1. The • basioccipital and the pterygoid are sutural1y articulated. Neither is cranial kinesis reported in Neusticosaurus or Serpianosaurus (Carrol1 and Gaskil1, 1985;
Rieppel, 1989; Sander, 1989). O-kinetic, l-akinetic.
4. The lower temporal fenestra. O-present, l-absent.
5. The upper temporal fenestra. O-smaller than orbit, l-Iarger than orbit.
6. Temporal emargination. O-present, l-absent.
7. The quadratojugal. This bone is not found in Keichousaurus. O-present, 1
absent.
8. A quadrate notch is present in Keichousaurus, Neusticosaurus and
Serpianosaurus, but absent in Claudiosaurus (Carrol1, 1981). It is also present
at least in Nothosaurus and Simosaurus, judging from the configuration of the • occipital region (Schultze, 1970). O-absent, I-present. 9. The suborbital fenestra is probably present in Keichousaurus, though it may
be quite small. O-present, l-absent.
10. The palatines. The palatines were not preserved in our specimens of
Keichousaurus. O-separated by pterygoids, I-meet in the midline.
11. Interpterygoid vacuity. O-present, l-absent.
12. Transverse flange of pterygoid. O-present, l-absent.
13. Premaxilla separates nasals. In Keichousaurus, Neusticosaurus,
Lariosaurus, Simosaurus and Cymatosaurus the dorsal-posterior process of the
premaxillae makes direct contact with the prefrontal, preventing the nasals from • 100 touching each other. O-no, 1-yeso • 14. Nasals contact prefrontals. O-yes, 1-no. 15. The supratemporal bones. O-present, 1-absent.
16. The shape of the upper temporal openings. The upper temporal openings
could be round or elongate. Neusticosaurus, Serpianosaurus, Anarosaurus and
Corosaurus have round temporal openings. Dactylosaurus, Keichousaurus,
Lariosaurus, Cymatosaurus, Ceresiosaurus and Nothosaurus have elongated
upper temporal openings. The temporal openings of Simosaurus are not as
elongated as the latter group, nor are they as short as the former. The outgroup
have round upper temporal openings. O-round, 1-elongated.
17. The postorbital. O-contacts supratemporal fenestra, 1-does not.
18. As other primitive sauropterygians, Keichousaurus has a triangular • postfrontal. O-triangular, l-subrectangular. 19. Contact of postfrontal with the temporal opening. Primitively, the
postfrontal does not enter into the supratemporal openings, as in Petrolacosaurus
(Reisz, 1981). However, in Youngina and Claudiosaurus, as weil as most ofthe
sauropterygians, the postfrontal does enter into the rim of the supratemporal
opening. Therefore, the primitive condition for sauropterygians should be
postfrontal contact with the temporal opening. Some specimens of Lariosaurus
lack the contact (Mazin, 1985). O-yes, 1-no.
20. The postparietal. O-present, 1-absent.
21. The tabular. O-present, 1-absent. • 101 22. The lacrimal. O-present, l-absent. • 23. The lacrimal-narial contact. O-present, l-absent. 24. The position of the jugal. O-meets the orbit, l-excluded from the margin
of the orbit.
25. The jugal contacts the quadratojugal. O-yes, I-no.
Character 26 describes the ,shape and direction of the long axis of jugal. This
character is recoded here as two characters.
26. The shape of jugal. The forms with a lower temporal bar have a triradiate
shaped jugal, whereas those that lack the lower temporal bar have a crescent
shaped jugal. O-triradiate, 1- crescent-shaped.
27. The orientation of the jugal. l-along orbit margin, 2-away from orbit.
28. Parietals fused. The outgroups and most sauropterygians have paired • parietals. Keichousaurus, like Lariosaurus and Nothosaurus, has fused parietals. O-paired, l-fused.
29. Position of the parietal foramen. In the outgroups, the parietal foramen is
situated at middle of the parietal. In Lariosaurus, Ceresiosaurus and
Nothosaurus the parietal foramen shifted rearward, whereas in Keichousaurus
it is situated anterior to the midline of the parietal, which is coded separately in
next character. O-middle, I-posteriorly positioned.
30. Parietal foramen anteriorly positioned. O-no, l-yes.
31. Frontals fused. This occurs in Keichousaurus, Lariosaurus and
Nothosaurus, partially in Neusticosaurus, Serpianosaurus, and probably also in • 102 Anarosaurus. O-paired, I-fused. • 32. Posterolateral process ofthe frontal. In forms with small temporal opening such as Keichousaurus, Neusticosaurus, Dacty/osaurus, Serpianosallrlls and
Anarosaurus there is a posterolateral process on the frontal. The suture between
the frontal and the parietal is more or less W-shaped. In forms with large
temporal opening (except Corosaurus), on the other hand, the suture between the
frontal and the parietal is straighter, and ouly has second degree zig-zags.
Corosaurus shows the W-shaped suture as forms with small temporal opening.
The outgroups exhibit the W-shape suture. O-absent, I-present.
34. Relative size of the prefrontals and the postfrontals. Primitively, the
prefrontals are larger than the postfrontals, as seen in C/audiosallrlls. Sorne of
the forms with small temporal opening, such as Neusticosaurus, Anarosaurus, • and Keichousaurlls, retained this condition. In Serpianosaurus, Dacty/osaurus, Corosaurus and possibly Ceresiosaurus the prefrontals and postfrontals are
about equal-sized. In typical "nothosaurs", the postfrontals are much larger than
the prefrontals. O-prefrontals larger than postfrontals, I-no.
35. O-prefrontals and postfrontals are about the same size, I-prefrontals much
smaller than postfrontals.
36. Palatal dentition. O-present, I-absent.
37. The rostrum. O-unconstricted, I-distinctly constricted.
38. The nasals. O-present, I-absent.
39. The external nares. O-terminal, I-retracted. • 103 40. The nasal contacts the extemal naris. O-yes, l-no. • 41. The diastema O-absent, l-present. 42. The tooth orientation. O-vertical, l-procumbent.
43. The dentition. O-anisodont, l-isodont.
44. The teeth. O-conical, l-squat.
45. The maxillary dentition. O-caniniform, l-not caniniform.
46. The premaxillary dentition. O-not caniniform, l-caniniform.
47. The mandibular symphysis. O-weak, l-robust.
48. The retroarticular process. O-absent, l-present.
49. The trunk intercentra. O-present, l-absent.
50. The type of vertebral centra. O-amphicoelous, l-platicoelous.
51. The foramina subcentralia. O-absent, l-present. • 52. The zygapophyses. O-wider than centrum, l-narrower than centrum. 53. The presence or absence of zygosphene/zygantrum articulations. 54. The
hyposphene/hypantrum. O-absent, l-present.
55. The number of sacral vertebrae. O-two, l-more than two.
56. The number of sacral vertebrae. O-three, l-four or more.
Although we know that having two sacral vertebrae is primitive, it is hard to
determine whether the increase of the sacral count is Iinear, i.e. 2->3->4->5, or
bushy, i.e. 2->3, 2->4, 3->4, 3->5, and 4->5. The sacral ribs could be added
from front, from rear, and from both.
57. The shape of sacral ribs. O-distally expanded, I-distally reduced. • 104 58. Pachyostotic and thickened ribs. O-absent, l-present. • 59. Number of gastralia elements. Keichousaurus has five elements per segment. Only Neusticosaurus has three elements per segment. AlI other taxa
have five elements per segment. O-five, I-three.
60. Dermal armour. O-absent, l-present.
61. The sternum. O-present, 2-absent
62. The ossification of the sternum. O-ossified, I-unossified.
63. The cleithrum. O-present, I-absent.
64. The posterior process of the interclavicula. O-present, I-absent.
65. Interclavicular process. O-elongate, l-short.
66. Clavicular corner. O-absent, I-present.
67. Scapulocoracoid. O-united, 1-divided. • 68. Position of the scapula relative to the c1avicle. One of the synapomorphies of the sauropterygians is the superficial position of the scapula relative to the
c1avicle. There is one exception in one of the Keichousaurus specimens
(V7919). Il has to be considered abnormal before more specimens with the
same condition were found and studied. O-medial to c1avicle, I-superficial to
clavicle.
69. The posterior extension of the coracoid. O-absent, l-present.
70. The supracoracoid (or coracoid) foramen. O-present, 1-absent.
71. The position ofthe supracoracoid foramen. In the primitive outgroups, this
foramen was completely surrounded by the scapulocoracoid. Il was situated at • lOS the border of the separated coracoid and scapular in nothosaurs. The position • of the supracoracoid foramen differs in different taxa: in Keichousaurus, Lariosaurus, Ceresiosaurus and Nothosaurus it lies in the middle ofthe border
line between the coracoid and the scapular. In Neusticosaurus and
Serpianosaurus it is at the medial margin of the scapulocoracoid and may be
open to the pectoral fenestration. In Dactylosaurus and Corosaurus the
supracoracoid foramen is completely lost, which is coded in character 70. 0
supracoracoid foramen in the middle ofthe scapulo-coracoid suture line, 1-open
to the pectoral fenestration.
72. The pectoral fenestration. O-absent, I-present.
73. The pectoral bar. O-absent, I-present.
74. The posterior ramus of the ilium. O-present, I-absent. • 75. The size of the posterior ramus of the ilium, when it exists. O-prominent, l-small.
76. The ilium-pubis contact. O-present, l-absent.
77. The anterior corner of the pubis. O-unexpanded, 1-expanded.
78. The obturator foramen. In Keichousaurus, the obturator foramen tends to
close up when the animal gets older. Therefore, the state is "varied". O-present,
l-absent.
79. The thyroid fenestration. O-absent, I-present.
80. The forelimb. O-as robust as the hindlimb, O-more robust than the
hindlimb.
• 106 81. The shape of the humerus. O-straight, I-strongly curved. • 82. The robustness of the humerus. O-slender, I-robust. 83. The humeraI epicondyles. O-prominent, I-reduced or absent.
84. Ectepicondylar foramen. This foramen is not present in Keichollsallrlls.
O-absent, I-present, 2-notch.
85. Entepicondylar foramen. O-present, I-absent.
86. Spatium interosseum. This is the space between the ulna and the radius.
In plesiosaurs, because of the transformation of the ulna and the radius, the
spatium interosseum disappeared. O-present, I-absent.
87. The distal end of the spatium interosseum. Primitively, the distal end as
weil as the proximal end of the two bones are close to each other, and the
spatium interosseum is narrow and closed. In Keichousaurus, Dactylosaurus • and ail the large "nothosaurs", the distal ends of the ulna and the radius were separated by the intermedium, and the spatium interosseum is wide and open at
the distal end. O-closed, I-open.
88. The radius and ulna length. Since there is no apparent olecranon in
sauropterygian ulna, we should discount the length of the olecranon in the
outgroups when doing the comparison. In the outgroups, the radius and the ulna
(minus the olecranon) are of about the same length. This is also the case in
most of the sauropterygians except Neusticosaurus and Serpianosaurus where
the radius is much longer than the ulna. O-equal, I-radius much longer.
89. The midlimb joint. O-present, I-absent. • 107 90. Hyperphalangy. O-absent, I-present. • 91. When hyperphalangy is present, the degree of hyperphalangy. O-slight, 2- extreme.
92. Carpus/tarsus. O-well ossified, I-poorly ossified.
93. The hindlimbs. O-Ionger than forelimbs, l-shorter than forelimbs.
94. The ulna. O-narrow, I-broad.
95. The pisiform. O-present, l-absent.
96. The length of the supratemporal fenestra. O-shorter than snout, l-Ionger
than snout.
97. Impedance-matching middle ear. This character is inferred from the shape
of the quadrate. Like Neusticosaurus and Serpianosaurus, the quadrate of
Keichousaurus also has the posterior embayment, indicating the similar middle • ear structure. O-absent, I-present. 98. The cranial bone omamentation. O-absent, I-present.
99. Presence or absence of the ectopterygoid. O-present, I-absent.
The data matrix is shown in Table IV. The data were anaIyzed using Mix
program of PHYLIP program package, version 3.5c (Felsenstein, 1993). Ali characters
are equally weighted, and no direction of transformation was assurned for characters.
A total of 57 most parsimonious trees are found with 214 required steps. The
consensus tree is shown in Figure 18. The configuration of the tree is similar to that
of Storrs' (1991). A few differences are Iisted below.
• 108 • •
10 20 30 40 50 60 70 80 90 G1ptoJ:hin:>ids OOO???OO?O OOOOO??O?O OOOOOO?OO? OOO?OOOOOO OOOO?OOOOO OOOO?OO?O? ?OOOOOO?O? O?OOOOOOOO OOO?OOOOOO ?OOOO?OO? RotroIacnsaurus OOOOO?OOOO 0000000010 OOOOOO?OO? OOO?OOOOOO OOOO?OOOOO OOOO?OO?O? ?OOOOOO?O? O?OOOOOOOO OOO?OOOOOO ?O?OOOOOO
Y~cmœs 1)0000?0100 0000000000 001000?00? 010?000000 0101?00000 0100?00000 OPOOOOO?O? O?OOOOOOOO 0010000000 ?00000100 Clau:iiasaurus 0001000000 0100000000 001011000? 010?000000 0101000000 0100?00000 1100000?0? O?OOOOOOOO 0111000000 ?OOOOOOOO
l'ëJsticnsaurus 0111000110 llPOlPlO01 1??011000? PJ.O?lOOlOO 110"l001100 0101Pl1101 ?101P11001 1001000000 0111000100 ?lPOIOlll Sezpi.anœaurus 0111000110 1100101001 11?011000? Pll0100100 1101001100 010lPlOO01 ?lPlll10n.l 1001000000 0111000100 ?lP010111
ArBrœamus OlllQO??10 1100100001 11?0?1000? ?10?100100 11010011?0 0?01011001 ?1?111101? 101100?000 0111000?01 010070?0?
IBctylasaurus 011100??10 1100110001 11?0?10010 0110100100 11010011?0 O?OlOlPOOl ?1??1.1101? 1001000000 0111001001 OllOPO?O? Keic!xlusaurus 011100110? 1110110001 11?0?10111 110?100100 1101001100 010P?PlO01 ?101P11000 100100P001 110?101001 010110100 Plaaxi:ntids 1011110101 1000100001 11?001100? 000?l00101 1P?1011100 OOllO?OOP1 ?lOU11?OO 1000000001 ll11POOOO? ?1?000000 sint:saurus 1111100110 1111110101 ll.?OlllOO? 0010100100 lll10?111Q 0101100?01. ?100111000 1000000001 1111001?0? ?110?0000
Cl:ItœatJrus 111110101? ?000100001 1l?1?1100? 0110100100 1001011100 0101000001 ?10111101? 1000010001 111100100? ?l0010000 a.resiasaurus lll1101Dl.? 1100UO??1 11??11?00? ?010100110 1?0??111.?0 0?01111001 ?11??11000 1001000011 1111001001 011100000 Iariasaurus 0111101010 llP01100Pl 1l.?1.???P1.0 POlllOOlOO 1000111100 O?OllllOOl ?11?111000 10010?0001 1111001001 010110000
c;}<1atœaurus 1111101010 1111110101 11?0???00? 0011110lPO 100001???? ?????????? ?????????? ?????????? ?????????? ?????OOOO Silvest:=saunts 0111101010 1101110001 11.???1110? IDllllOIOO 1000111100 010??00001 ?1??1ll01? 1001000101 l1l110100? ??OO?OOOO
lbtirsaurus lllllO1110 1101110101 l1?l?llllO 1011110100 1000111110 010lPOO?01 ?l01111000 1001000101 1111001001 010011000 RiIallotirsaurus lllllO1010 1101110111 ???1?11110 1011110100 1000111110 0101100?01 ?11?111000 1001000101 111100100? ?l0011000 l'.istJ: Table IV Characler states of sauropterygians. • • Figure 18 A consensus cladogram of sauropterygians. • 110 '-':; .... • • In the Pachypleurosauria clade, Anarosaurus is the primitive sister-group of • «Neusticosaurus, Serpianosaurus), (Dactylosaurus, Keichousaurus)) assembly. However, this might be due to incompleteness of our knowledge of Anarosaurus. In Eusauropterygia clade, Simosaurus is the primitive sister-group of the rest of the taxa. Corosaurus is the primitive sister-group of the next level. Below this level is the dichotomy of Plesiosauria and conventional Nothosauria minus Simosaurus. Silvestrosaurus is the primitive sister-group of (Nothosaurus, Paranothosaurus). Both this and Storrs' cladogram (Storrs, 1991) also requires 214 steps. • • 111 • CONCLUSIONS The primitive quadrupedal tetrapods are (or were) sprawlers. The feet are placed laterally to the body, as apposed to beneath the body as dinosaurs and marnmals. Their locomotion mode can be termed a "walking trot", which is defined as "diagonally opposite legs swing in unison" (Hildebrand, 1985). The vertebral column is usually flexed laterally to increase the length of steps. It is natural for these animais to use lateral undulation when they swim. Many semi-aquatic and aquatic reptiles are (or were) lateral undulatory swimmers. The Alpine pachypleurosaurids are also believed to have employed a lateral undulating swimming pattern. However, the more advanced • sauropterygians, the plesiosaurs, were definitely not lateral undulatory swimmers. Both the fore- and the hindlimbs ofplesiosaurs are equally highly developed and are believed to have been used for propulsion (either as wings or paddles.) Judging from their stout forelimbs, nothosaurids were probably not undulatory swimmers either. However, only the forelimbs were probably used in locomotion. The elaboration and inclusion of hindlimbs into swimming apparatus may have occurred after the paraxial stroke pattern had been established. Ifthe phylogenetic analysis ofStorrs (1993) and presented in this paper is accurate, the dichotomy of swimming patterns between lateral undulatory and paraxial propulsion probably coincided with the splitting of pachypleurosaurid and nothosauriformes in the early stage of evolution of the sauropterygians. • 112 Keichousaurus was the only known pachypleurosaurid that employed a swimming pattern similar to that of the nothosaurids. It is not clear what was the cause (or causes) of this pattern switch in primitive nothosaurids and in Keichousaurus. The mode of reproduction of Keichousaurus hui is tentatively hypothesized as ovoviviparous. The limb structures of Keichousaurus hui are specialized for aquatic locomotion. The mid-arm joint was greatly sirnplified and the olecranon process ofthe ulna disappeared completely. Crawling up the beach to lay eggs would be a very awkward business. The fact that an embryo was discovered in the same sedirnentary environment also favours an ovoviviparity hypothesis. At birth, the forelimbs of Keichousaurus hui were shorter than the hindlimbs. However, since the growth of the forelimbs relative to the body was highly positively allometric while the growth of the hindlimbs was isometric to slightly allometric, the • adult Keichousaurus hui had longer and stronger arms than legs. It is believed that the forelimbs were the primary locomotion apparatus of Keichousaurus hui. Sexual dimorphism is apparent in Keichousaurus hui, and manifested primarily in the configuration and length ofthe hurnerus. However, the true sex ofthe two forms could not be determined. • 113 • REFERENCES Alexander, R.M. 1968 Animal Mechanics. Sidgwich & Jackson, London. Andrews, R.M. 1982 Patterns of growth in reptiles. In Biology ofthe Reptilia, vol. 13 (Physiology D)(ed. C. Gans & F.H. Pough), pp. 273-320. London: Academie Press. Arthaber, G.v. 1924 Die Phylogenie der Nothosaurier. Acta Zool. 5, pp. 439·516. • Blackstone, N.W. 1987 Allometry and relative growth: pattern and process in evolutionary studies. Syst. Zool. 36(1), pp. 76-78. Caldwell, M.W. 1994 Developmental constraints and limb evolution in Permian and modern lepidosauromorph diapsids. in press. Carroll, R.L. 1981 Plesiosaur ancestors from the Upper Permian of Madagascar. Phil. Trans. R. Soc. Land. B 293, pp. 315·383. Carroll, R.L. 1985 Evolutionary constraints in aquatic diapsid reptiles. Special • 114 • Papers in Palaeontology. 33, pp. 145-155. Carrol~ R.L., Gaskill, P. 1985 The nothosaur Pachypleurosaurus and the origin of plesiosaurs. Phil. Trans. R. Soc. Lond. B 309, pp. 343-393. Clark, B.D., Bemis, W. 1979 Kinematics ofswimming ofpenguins at the Detroit Zoo. J Zool., Lond. 188, pp. 411-428. Currie, P.J. 1981 Hovasaurus boulei, an aquatic eosuchian from the Upper Permian of Madagascar. Palaeont. afro 24, pp. 99-168. Currie, P.J., Carroll, R.L. 1984 Ontogenetic changes in the eosuchian reptile • Thadeosaurus. J Vert. Paleont. 4 (1), pp. 68·84. English, A.W. 1976 Limb movements and locomotor function in the Califomia sea lion (Zalophus californianus). J Zool., Lond 178, pp. 341·364. Felsenstein, J. 1993 PHYLIP (Phylogenetic Inference Package) version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle. Gaffney, E.S. An introduction to the logic of phylogenetic reconstruction. In • • 115 Phylogenetic Analysis and Paleontology 1979 (ed. J. Cracraft and N. Eldredge), • pp. 79-111, Columbia University Press, New York. Godfrey, St.J. 1984 Plesiosaur subaqueous locomotion, a reappraisal. N. .lb. Geol. Palaont. Mh. 1984 (11), pp. 661-672, Stuttgart. Godfrey, St.J. 1985 Additional observation of subaqueous locomotion in the California sea lion (Zalophus californianus). Aquatic Mammals 11.2, pp. 53-57. Gould, S,Jo 1971 Geometrie similarity in allometric growth: a contribution to the problem of scaling in the evolution of size. The American NaturaUst 105 (942), • pp. 113-136. Hecht, M.K. 1976 Phylogenetic Inference and methodology as applied (0 the vertebrate record. Evolutionary Biology Vol. 9, pp. 335-363. Hildebrand, M., 1985 Walking and running. In Functional Vertebrate Morphology. (ed. Hildebrand,· Bramble, Liem, Wake), pp. 38-57. The Belknap Press of Harvard University Press, Cambridge, Massachusetts and London, England. Hoffstetter, R., Gasc, J.P. 1969 Vertebrae and ribs of modern reptiles. ln Biology of the Reptilia, vol. 1. (Morphology A) (ed. C. Gans, A. d'A. Bellairs & T.S. • 116 • Parsons), pp. 201-310. New York & London: Academie Press. Holmes R. 1977 The osteology and musculature of the pectoral Iimb of small captorhinids. J. Morph. 152, pp. 101-140. Holmes R. 1989 Functional interpretations of the vertebral structure in palaeozoic labyrinthodont amphibians. Historical Biology 2, pp. 111·124. Huene, F.R.von 1959 Simosaurus Guilielmi aus dem Unteren Mittelkeuper von Obersontheim. Palaeont. Abt. A113, pp. 180-184. Jenkins, F.A., Goslow, G.E. 1983 The functional anatomy of the shoulder of the • Savannah Monitor Iizard (Varanus exanthematicus). J. Morph. l75,pp. 195-216. Kubn-Scbnyder, E. 1959 Ein neuer Pachypleurosaurier von der Stulseralp bei Bergün (Kt. Graubünden, Schweiz). Ec!. Geol. Helv. 52, pp. 639-658. Kubn-Scbnyder, E. 1987 Die Triasfauna der Tessiner Kalkalpen. XXVI. Lariosaurus lavizzarii n. sp. (Reptilia, Sauropterygia). Schweizerische Paliiont. Abh. 110, pp. 1-24. Manter, J.T. 1940 The mechanics of swimming in the alligator. The Journal of • 117 • Experimental Zoology 83(3), pp. 345-358. Mazin, J.M. 1985 A specimen of Lariosaurus balsami Curioni 1847, from the eastern Pyrenees (France). Palaeontographica Abt.A. 189, pp. 159-169. McMahon, T.A., Bonner, J.T. 1983 On Size and Life. Scientific American Books, Inc. New York. Nicholls, E.L., Russell, A.P. 1991 The p1esiosaur pectoral girdle: the case for a sternum. N. Jb. Geol. Palaont., Abh. 182, pp. 161-185; Stuttgart. Peyer, B. 1939 Die Triasfauna der Tessiner Kalka1pen. 14. Paranothosaurus amsleri • nov. gen. nov. spec. Abh. Schweiz. palaont. Ges. 62(4), pp. 1-87. Rieppel, O. 1987 The Pachypleurosauridae: an annotated bibliography. With comments on sorne lariosaurs. Eciogae geol. Helv. 86 (3), pp. 1105-1118. Rieppel, O. 1989 A new pachypleurosaur (Reptilia: Sauropterygia) from the Middle Triassic of Monte San Giorgio, Swithzerland. Phil. Trans. R. Soc. Lond. B 323, pp. 1-73. Rieppe1, O. 1993a Euryapsid relationships: A preliminary analysis. N. Jb. Geol. • 118 • Paitiont. Abh. 188 (2), pp. 241-264, Stuttgart. Rieppel, O. 1993b Status of the pachypleurosauroid Psilotracheiosaurus toeplitschi Nopcsa (Reptilia, Sauropterygia), from the Middle Triassic of Austria. Fieidiana: Geoiogy, N.S., No. 27, pp. 1·17. Rieppel, O. 1994 The braincases of Simosaurus and Nothosaurus: monophyly of the Nothosauridae (Reptilia: Sauropterygia). J. Vert. Paieont. 14 (1), pp. 9-23. Robinson, J.A. 1975 The locomotion of plesiosaurs. N. J.h. Geoi. Paitiont. Abh. 149 (3), pp. 286-332, Stuttgart. • Robinson, J.A. 1977 Intracorporal force transmission in plesiosaurs. N. J.h. Geoi. Paitiont. Abh. 153 (1), pp. 86-128, Stuttgart. Romer, A.S. 1922 The locomotor apparatus of certain primitive and mammal-like reptiles. Bull. Am. Mus. nat. Hist. XLVI, pp. 517-606. Roiner, A.S. 1942 The development of tetrapod limb musculature--the thigh of Lacerta. J. Morph. 71, pp. 251-298. Romer, A.S. 1956 Osteoiogy of the Reptiles. The University of Chicago Press, • 119 • Chicago & London. Sander, P.M. 1988 A fossil reptile embryo from the Middle Triassic of the Alps. Science. 239, pp. 780-783. Sander, P.M. 1989 The pachypleurosaurids (Reptilia: Nothosauria) from the Middle Triassic of Monte San Giorgio (Switzerland) with the description of a new species. Phil. Trans. R. Soc. Land. B 325, pp. 561-670. Schmidt, S. 1987 Phylogenie der Sauropterygier (Diapsida; Trias - Kreide). [Phylogeny of the Sauropterygians (Diapsida; Triassic - Cretaceous)]. N. Jb. • Geai. Palliant. Abh. 173 (3), pp. 339-375, Stuttgart. Schultze. H.-P. 1970 Über Nothosaurus. Neubeschreibung eines Schiidels aus dem Keuper. Senckenbergiana lethaea. 51, pp. 211-237. Snyder, R.C. 1954 The anatomy and function of the pelvic girdle and hind!imb in !izard locomotion. Am. J. Anatomy. 95 (1) pp. 1-45. Spellerberg, I.F. 1982 Biology ofreptiles. Blackie & Son Ltd., Glasgow & London. Storrs, G.W. 1991 Anatomy and relationships of Coro~'aurus alcovensis (Diapsida: • 120 Sauropterygia) and Triassic Alcova Limestone ofWyoming. Bull. Peabody Mus. • nal. Hisl. 44, pp. 1-151. Storrs, G.W. 1993a The systematic position of Silveslrosaurus and a classification of Triassic sauropterygians (Neodiapsida). Pallionl. Z. 67 (1/2), pp. 177-191, Stuttgart. Storrs, G.W. 1993b Function and phylogeny in sauropterygian (Diapsida) evolution. American Journal ofScience, 293(A), pp. 63-90. Sues, H.-D. 1986 Postcranial skeleton of Pislosaurus and interrelationships of the • Sauropterygia (Diapsida). Zool. J. Linn. Soc. 90, pp. 109-131. Sues, H.-D., Carroll, R.L. 1985 The pachypleurosaurid Dactylosaurus schoederi Diapsida: Sauropterygia). Cano J. Earlh Sci. 22, pp. 1602-1608. Young, C.C. 1965 On the new nothosaurs from Hupeh and Kweichou, China. Verlebrala Palasialica 9(4), pp. 315-356. Young, C.C. 1958 On the new Pachypleurosauroidea from Keichow, South-West China. Verlebrala Palasialica 2(2-3), pp. 69-81. • 121 • ABBREVIATIONS a angular ar articular ast astragalus bo basioccipital c centrum ca caudal cal calcaneum cb ceratobranchial • cl c1avicle cor coracoid . d dentary de distal carpal dt distal tarsal eo exoccipital ept ectopterygoid f frontal fi fibula ic intercentrum • 122 ici interclavicle • il ilium im intermedium isc ischium j jugal m maxilla n nasal na neural arch op opisthotic pa parietal pc centrum pf postfrontal • pm premaxilla po postorbital prf prefrontal pt pterygoid pub pubis q quadrate ra radius sa sacral sc scapula so supraoccipitai • 123 sq squamosal • st supratemporal ti tibia ul ulna ulr ulnare • • 124