Title: Phylogenetic relationships of the soft-shelled turtles (family trionychidae)
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00099338/00001
 Material Information
Title: Phylogenetic relationships of the soft-shelled turtles (family trionychidae)
Physical Description: vi, 200 leaves : ill. ; 28 cm.
Language: English
Creator: Meylan, Peter A ( Peter Andre )
Copyright Date: 1985
Subject: Soft-shelled turtles   ( lcsh )
Zoology thesis Ph. D
Dissertations, Academic -- Zoology -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Statement of Responsibility: by Peter Andre Meylan.
Thesis: Thesis (Ph. D.)--University of Florida, 1985.
Bibliography: Bibliography: leaves 192-199.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00099338
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000870301
notis - AEG7378
oclc - 014443860


This item has the following downloads:

phylogeneticrela00meyl ( PDF )

Full Text








The progress of this study has relied on a great number of people

and to all of them I extend my thanks. My Ph.D. committee, Walter

Auffenberg, David Webb, Jon Reiskind and Doug Jones has been supportive

and encouraging throughout the course of this study. I thank them for

their guidance and assistance in the preparation of this thesis.

I have benefitted from numerous discussions of turtle systematics

and phylogeny with my colleagues, Dennis Bramble, Chuck Crumly, Jim

Dobie, Gene Gaffney, Ren Hirayama, Howard Hutchison, John Iverson and

Peter Pritchard.

For loans and/or access to museum collections I am grateful to the

following curators and collection managers: Franz Tiedemann and Heinz

Grillitsch, Naturhistorisches Museum, Vienna; Nick Arnold, Colin

McCarthy, and Barry Clarke, British Museum of Natural History; D. F. E.

Thys van den Audenaerde, Mus6e Royal de l'Afrique Central, Tervuren; E.

R. Brygoo, Roger Bour, and J. P. Gasc, Mus6um National d'Histoire

Naturelle, Paris; M. S. Hoogmoed, Rijksmuseum van Natuurlijke Historie;

K. Klemmer, Natur-Museum Senkenberg, Frankfurt; Ulrich Gruber and Dieter

Fuchs, Zoologisches Sammlung der Bayerisches Staates, Munich; U. Rahm,

Naturhistorisches Museum, Basel; J. L. Perret, Museum d'Histoire

Naturelle, Geneva; J.P. Gosse, Institut Royal des Sciences Naturelles de

Belgique, Brussels; Don Broadley, National Museum of Zimbabwe; Pere

Alberch and Jose Rosado, Museum of Comparative Zoology, Harvard; Harold

Voris, Field Museum of Natural History; Arnold Kluge and Dennis Harris,

Museum of Zoology, University of Michigan; Jim Dobie, Auburn University

Museum of Paleontology; and George Zug, United States National Museum.

For access to specimens in their private collections I thank Richard

Etheridge, Ren Hirayama, John Iverson, Ed Moll and Peter Pritchard.

Vince DeMarco made the histological preparations. Bill Maples

provided the photographic equipment used to prepare Figures 1 and 5. Ian

Brehney prepared Figure 6.

This study has been supported by a grant from the Leakey Foundation,

by Colonel Barkau, by the Department of Zoology and by the Florida State

Museum. To the staff of the Florida State Museum I give special thanks

for support of many kinds.

And for Anne there are never thanks enough.




ACKNOWLEDGEMENTS............................................... ii

ABSTRACT ........................................................ v

INTRODUCTION................................... ............. .... 1

METHODS ......................................................... 8

The Phylogenetic Method..................................... 8
Terminology............................................... 15

RESULTS.......................................................... 17

Variation in Shell Morphology................................ 17
Variation in Skull Morphology................................ 59
Variation in the Mandible and Nonshell Postcrania............ 96

DISCUSSION....................................................... 118

The Higher Relationships of the Trionychidae................. 118
Relationships among the Recent Trionychidae.................. 132
Trends and Mechanisms in Soft-Shelled Turtle Evolution....... 152

REFERENCES CITED............................................... 192

BIOGRAPHICAL SKETCH............................................. 200

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy



Peter Andre Meylan

December, 1985

Chairman: Walter Auffenberg
Major Department: Zoology

Phylogenetic analysis of nearly one hundred characters of the

osteology of trionychid turtles provides abundant data on the phyletic

relationships of this family to other turtles and on the

interrelationships of its members. These data suggest that the family

Trionychidae shares a unique common ancestor with the Dermatemydidae,

Kinosternidae and Carettochelyidae, and that the Kinosternidae shares a

unique common ancestor with the Trionychidae and Carettochelyidae.

Furthermore, it is the staurotypine kinosternids that are most closely

related to the Trionychidae and Carettochelyidae. Carettochelyids and

trionychids share numerous unique features and constitute a monophyletic


Within the Trionychidae, the subfamilies Cyclanorbinae and

Trionychinae are recognized as monophyletic clades. Recognition of three

cyclanorbine genera, Cycloderma, Cyclanorbis and Lissemys, is warranted.

Within the Trionychinae, four distinct clades are recognized. The

Trionyx cartilagineus group includes Chitra indica and Pelochelys bibroni

on the basis of the unique location of the foramen posterior canalis

carotici interni, and features of the trigeminal region. The North

American group, which includes T. triunguis, T. euphraticus, T. swinhoei,

T. ferox, T. spiniferus and T. muticus, can be recognized by the presence

of eight or fewer neurals (first and second are fused), deeply emarginate

prefrontals and a large contribution by the parietal to the processus

trochlearis oticum. The Indian group includes four species: T.

gangeticus, T. hurum, T. leithii and T. nigricans. All exhibit a free

first neural, five plastral callosities and intermediately extended

epiplastra. Lastly, the T. steindachneri group is diagnosed by a

descending spine of the opisthotic which divides the fenestra postotica

in most specimens.

Two equally parsimonious arrangements of the Trionychinae differ in

the placement of the North American clade. In one, this clade is the

sister group of the T. cartilagineus clade; in the other it is the sister

group of the T. steindachneri clade. In both, the Indian group is

paraphyletic and gives rise to the T. steindachneri clade.

A tentative revision of the classification of the family

Trionychidae is provided. In it, five generic names currently in the

synonymy of Trionyx are resurrected to denote unique clades of

trionychine turtles.


Within recent years a fundamental revision of the systematic

relationships of turtles has begun. This revision was precipitated by

Gaffney (1975), who presented a reorganization of the Testudines using

the phylogenetic method (as outlined in Gaffney, 1979a; Wiley, 1981).

Other authors have followed Gaffney's lead in applying this method to

problems in chelonian systematics, resulting in a much clearer

understanding of the phylogenetic relationships among turtle taxa.

Concise hypotheses of the relationship within most families are now

available (Proganochelidae, Gaffney and Meeker, 1983; Chelidae, Gaffney,

1977; Baenidae, Gaffney, 1972; Meiolaniidae, Gaffney, 1983; Chelonoidea,

Gaffney, 1976; Kinosternidae, Hutchison and Bramble, 1981; Emydidae,

Hirayama, 1985; Testudinidae, Crumly, 1982, 1985) and are summarized by

Gaffney (1984). The largest remaining family for which no such

hypothesis exists is that comprising the soft-shelled turtles,


This gap is significant, considering the large size, abundance and

great age of the family. The Trionychidae includes more than 250 species

(ca. 230 fossil and 22 extant) and occurs on every continent except

Antarctica. It is a very ancient family, with definite representatives

from the Cretaceous (Romer, 1956) and a probable representative from the

Jurassic (Young and Chow, 1953). Representation of this family in the

fossil record is considerable, although difficult to document because few


authors treat the fossils of this troublesome group. The best evidence

of its ubiquity is reported by Hutchison (1982), who shows that the

Trionychidae has the most continuous record of 11 reptile families

examined from the Cenozoic of western North America.

Although authors do not agree on the relationships of trionychids to

other turtles, I have never seen a single reference doubting the

monophyly of the family. It is so distinctive that some authors have

placed the family in a separate suborder equivalent to the Cryptodira and

Pleurodira (Boulenger, 1889; Siebenrock, 1909; Bergounioux, 1932, 1955),

an arrangement for which Loveridge and Williams (1957) found some

support. Modern morphologists argue that this family is a group of

aberrant cryptodires allied to the Carettochelyidae, Dermatemydidae and

Kinosternidae (McDowell, 1961; Albrecht, 1967; Zug, 1971; Gaffney, 1975,

1984). But others disagree, arguing that on the basis of karyology the

Trionychidae, along with the Carettochelyidae, is the sister group of all

other cryptodires (Bickham, Bull and Legler, 1983). On the basis of

serological tests, Frair (1983) supported the placement of the family in

its own suborder.

Among those workers willing to place the Trionychidae among the

Cryptodira, there is difference of opinion about which species are the

closest relatives of soft-shelled turtles. Since its discovery,

Carettochelys (Ramsay, 1886) has been considered to be closely related to

the Trionychidae, although some authors were confused by false reports of

mesoplastra in this genus (Boulenger, 1889; Pritchard, 1967). Many

authors have recognized close relationship between Carettochelys and the

Trionychidae (Boulenger, 1889; Baur, 1890, 1891b; Waite, 1905;

Siebenrock, 1902, 1913) and some have recommended that superfamilial

status be recognized (Trionychia, Hummel, 1929; Trionychoidea, Walther,


Several morphologists (Baur, 1891a; McDowell, 1961; Albrecht, 1967;

Zug, 1971; and Gaffney, 1975, 1984) have allied the Trionychidae and

Carettochelyidae with the Dermatemydidae and the Kinosternidae. Gaffney

(1975) applied the name Trionychoidea to this group. This enlarged

concept of the Trionychoidea is in clear conflict with the frequent

association of the Kinosternidae with the Chelydridae and the inclusion

of these two families in a clade with the Emydidae and Testudinidae. The

latter arrangement was proposed by Williams (1950) based on the

morphology of the cervical vertebrae and other osteological characters

and includes the Kinosterninae and Staurotypinae within the Chelydridae.

The Williams arrangement has been followed by Romer (1956), Pritchard

(1967, 1979a, 1979b), Mlynarski (1976) and others. Karyotypic data have

recently been cited which partially support this arrangement (Bickham and

Carr, 1983). It is obvious from these various arguments that the

phylogenetic position of the family within the Testudines is still in


A narrower but equally urgent problem concerns the

interrelationships within the family Trionychidae. The lack of resolution

of relationships within the family is indicated by the current placement

of nearly all species (ca. 235) in a single "wastebasket" genus, Trionyx.

For more than fifty years since the major revision by Hummel (1929),

there has been a strong tendency to synonymize trionychine genera (except

Chitra and Pelochelys) with Trionyx (Bergounioux, 1955; Romer, 1956;

Huene, 1956), with the result that about 40 generic names are now

considered synonyms (Smith and Smith, 1979). The apparent reason for

this is not uniformity of morphology, but rather an absence of a complete

and systematic interpretation of the characters. The large number of

taxa and high variability of the characters makes any study of trionychid

relationships using a phenetic method extremely difficult. The most

important recent studies are those of Loveridge and Williams (1957) and

De Broin (1977). On the basis of osteological characters, color pattern

and geography, Loveridge and Williams arrived at the arrangement redrawn

as Fig. 23. The De Broin (1977) arrangement is based largely on

characters of the shell and skull (especially the palate), but is given

in insufficient detail to allow construction of a branching diagram.

Both the Loveridge and Williams (1957) and De Broin (1977) arrangements

contain features which appear in a cladistic analysis of the family based

on shell morphology (Meylan, 1985).

Phylogenetic analysis provides a methodological breakthrough that

may elucidate trionychid relationships. This method results in an

arrangement of taxa in a hierarchy of interested natural groups.

Because uniquely derived character states are used only from that point

in the hierarchy beyond which they are shared by all taxa, these states

automatically form diagnoses. Recognition of the diagnostic features of

monophyletic groups produces a strong hypothesis for the proper position

of the Trionychidae among the Testudines and clarifies the

interrelationships of its living species. In this work I have developed

a hypothesis of evolutionary relationships for the 22 Recent species of

the family Trionychidae. A complete revision of the entire family

(fossil and extant species) lies beyond the scope of this dissertation

and will take many years to complete.


The species of living trionychid turtles recognized for this study

are essentially those listed by Wermuth and Mertens (1961). The only

differences are 1) the use of the generic name of Trionyx for Dogania

subplanus, following Loveridge and Williams (1957); and 2) the use of the

name Trionyx swinhoei for the large and colorful Chinese species which

these authors had relegated to the synonymy of Trionyx sinensis (De

Broin, 1977; Meylan, in prep.); and 3) the relegation of Trionyx ater to

subspecific level within T. spiniferus (Smith and Smith, 1979). The

twenty-two species used are the same as those employed in Loveridge and

Williams (1957).

Two species recognized since Wermuth and Mertens (1961) have been

deemed insufficiently distinct to be used in the current study. On the

basis of the absence of integradation between Lissemys scutata and L.

punctata, Webb (1982) proposed that the former be considered a full

species rather than a subspecies of L. punctata (Annandale, 1912;

Deraniyagala, 1939; Wermuth and Mertens, 1961). The primary

morphological differences between the two are the configuration of the

peripherals and the well developed plastral callosities at a small size.

All superficial dermal callosities are highly variable within trionychid

species, and thus additional, less variable features should be found to

corroborate the validity of L. scutata before it is used as a distinct

species. If valid, there is little doubt that L. punctata is its closest


The name Trionyx nakornsrithammarajensis was proposed for a "rare

softshell" from Thailand (Wirot, 1979). Judging from the color pattern

of the specimen in the figure included with the description, this name

applies to Trionyx cartilagineus.

One of the most laudable aspects of phylogenetic analysis, which is

absent from phenetic methods, is that it requires an observer to look

beyond the taxa of immediate interest. Decisions about the direction of

evolutionary change of characters of the ingroup (the Trionychidae)

requires information from related forms. Therefore this study of the

relationships of the members of a single family includes an investigation

of interfamilial relationships and consequently has evolved into a study

of representatives of the entire order. With its scope expanded by the

requirements of phylogenetic methodology, this study has produced

significant data on the distribution of character states among all

turtles. These data are valuable in assessing the interfamilial

relationships of trionychids.

The methodology employed also provides a means for identifying those

characters which have states that appear to have been gained or lost

independently, or which may have undergone reversal. All of these events

are termed homoplasy and are the single most confounding feature in

systematics. When systematic evaluations must be made from limited data

sets, as in paleontology, it is important that characters subject to

homoplasy are identified. Because most fossil Trionychidae have been

described from shell material, an analysis of homoplasy in shell

characters is critical to future work on the systematics of fossil forms.

The descriptive portions of this study focus entirely on characters

significant in producing a phylogenetic arrangement for the ingroup.

They are not meant as an exhaustive survey of the osteology of the

Trionychidae (see Ogushi, 1911).

The primary objectives of this project are to fill the largest

remaining gap in our understanding of testudine phylogeny by 1)


determining the best placement of the Trionychidae within the Testudines;

and 2) resolving relationships among the 22 extant trionychid species. In

addition, it is hoped that this study will help to provide a basis for

future analysis of the relationships among the ca. 230 species known only

from fossil material.


The Phylogenetic Method

Systematics is not only a means of providing names for organisms and

groups of organisms, but also a method by which we can infer and express

the historical data of descent. Biologists agree that all organisms have

evolved by a true phylogenetic progression. The actual pedigree of taxa

represents a succession of shared ancestries. Analysis of common ancestry

can be a powerful explanatory tool for the co-occurrence of traits of

morphology (Lauder, 1982), ecology (Stearns, 1984), physiology (McNab,

1978) and behavior (Meylan and Auffenberg, in press). But the

possibility that any features are a result of phylogenetic propinquity

cannot be explored unless classification reflects the correct history of

descent. Consequently, it is critical for systematists to propose

classifications which best reflect the descent of organisms. The

phylogenetic or cladistic method is explicit in its reliance on shared

derived characters, which are a function of the descent of species.

The Method

The phylogenetic method relies on the identification and use of

shared derived characters to identify recency of common ancestry. Given

that parallelism is the exception rather than the rule, any two taxa are

more likely to have a shared trait because it was present in their common

ancestor rather than because it appeared independently on two occasions.

Thus the distribution of shared derived characters among taxa can be used

to build a hierarchical ranking of recency of common ancestry.

Developing this hierarchical ranking requires 1) identification of

characters with appropriately distributed character states; 2)

identification of primitive versus derived states for the characters; and

3) a system for the formulation of hypotheses of hierarchical

relationship using the shared derived character data in the most

efficient (parsimonious) manner.

Characters and Character States

The systematics of soft-shelled turtles has been based almost

exclusively on skeletal morphology (see for example Baur, 1893;

Siebenrock, 1902; Hummel, 1929; Loveridge and Williams, 1957; De Broin,

1977). Characters of the external soft anatomy are apparently of little

use, and few studies have employed them. The exceptions are the use of

color pattern (Loveridge and Williams, 1957; Webb, 1962) and the presence

of femoral valves (most studies). For this reason, the character survey

in the present study was restricted to skeletal morphology. A secondary

advantage of this emphasis is its future direct application to the

interpretation of the relationships of fossil trionychid species.

Characters of two types were sought: those uniform within the family

but varying among higher taxa outside the family; and those varying

among different groups of trionychid species. The former (interfamilial

characters) provide a data set for hypotheses on the placement of

trionychids within the Testudines. The latter (intrafamilial characters)

provide a basis for developing phylogenetic hypotheses for species within

the family Trionychidae.


Variation in a given character is treated as states of that

character. Many of the characters used have only two states, such as

presence or absence of a given bone, structure, or contact. Other

characters include three or more discrete states or even continuous

variation. Multistate and continuous characters pose two methodological

problems. First, for purposes of analysis it is necessary to divide a

continuously varying character into a number of discrete states.

Secondly, it has been proposed (Gaffney, 1979a, pers. comm.) that

recognition of intermediate states requires ad hoc hypotheses that

evolution has occurred in certain ways, and therefore multistate

characters should be avoided.

Information contained in multistate characters, or morphoclines, is

extremely useful for understanding the history of descent of any group

(Maslin, 1952) and has been critical in formulating a hypothesis of

relationships for trionychids based on the shell alone (Meylan, 1985).

The multistate characters that have been used in the present study are of

three different types: 1) continuous characters of shape, size, relative

position, etc., for which states have been determined by the occurrence

of natural breaks along a continuum; 2) discrete characters of a meristic

nature for which more than two possible states exist; and 3) two-state

characters in which both states frequently occur in the same species,

requiring the recognition of that third intermediate condition. I submit

that in all of these cases, as for two-state characters, only a

hypothesis of character polarity is necessary. By invoking the principle

of parsimony we can suggest that the amount of evolutionary change

required to arrive at a given state should be minimized, just as the

number of postulated evolutionary changes in a clade are ordinarily

minimized by phylogenetic systematics. Why should we propose that

turtles with 18 or fewer peripherals arrived at that condition directly

from the primitive number, 22, when the shortest evolutionary pathway

that could be used to explain this condition would be a change from a

form with 20 peripherals?

Multistate discrete characters (e.g., number of peripherals or

neurals) present little problem for the recognition of character states.

Continuous characters of relative size must be divided into states by

some artificial but objective means. As in other studies (Marx and Rabb,

1972; Drewes, 1984), I have divided continuous characters by plotting the

average values for terminal taxa along a continuum, and employing natural

breaks in distribution as evidence of various character states (see Fig.

9, for example). If no natural breaks in distribution occurred, the

character was discarded.

The characters employed in this study are given equal weight.

Certain authors, most notably Hecht and Edwards (1977), have argued that

some types of characters, for example those involving loss, should be

given little weight. In this study characters are weighted only in the

sense that they have been included or discarded, depending on the

distribution of variation. I disagree with the concept of character

weighting in general, and in particular, I do not accept the supposition

of Hecht and Edwards that loss is simple and subject to homoplasy, and

therefore should be given low weight. The loss of a major structure such

as the peripheral bones in turtles or the neural spine in snakes (a

character of lowest value in the Hecht and Edwards' scheme) can occur

only when a complex structural alternative (a character of highest value

in the Hecht and Edwards' scheme) is available. The losses mentioned


above require the development of strong and deeply sutured rib heads in

the case of certain trionychoids, and relocation of numerous muscles that

originate or insert on the neural spine in snakes. This may explain why

the loss of each of these features has apparently occurred only once. In

both instances, loss is the immediately apparent result of a complex

evolutionary event, and therefore should not be discounted.

There are two reasons to include a maximum number of characters in

this analysis. First, such inclusiveness is necessary to provide results

that will be of greatest value to paleontology. Paleontologists are

often faced with solving systematic problems on the basis of incomplete

material. By increasing the number of characters, there is an increased

likelihood that characters present in any given fossil have been studied.

The second reason is to provide redundancy as a test of problematic

groups or possibly homoplastic character states. In this project I use

about 100 characters taken from all parts of the skeletal system.

Character codes are used to assist in the identification of

characters. Because there are about 100 characters discussed in this

paper, some means of assisting the reader is required. The simplest

solution was an alpha-numeric code. Characters of the shell are given a

number only, following the code introduced in Meylan (1984) with some

additions. Characters of the skull are preceded by an S if they are

qualitative, except for those of the trigeminal region which are preceded

by T. Quantitative (mensurative) characters of the skull are preceded by

M. Characters of the lower jaw are preceded by the letter L; those of

the pectoral and pelvic girdles by P; and those of the cervical and

posterior body vertebrae by C; those of the extremities, both fore and

hind limbs, by E; and those of the hyoid by H.

Character Polarity

Once the states of a given character have been recognized, it is

essential to identify the primitive and derived extremes, or character

polarity. Numerous criteria for determining the polarity of character

transformations have been offered in the literature. The most often

treated are outgroup comparison, commonality, evidence from the fossil

record, evidence from embryology, and correlation of character states

(Kluge and Farris, 1969; Marx and Rabb, 1972; Wiley, 1981). I follow

Gaffney (1979a), Watrous and Wheeler (1981) and Wiley (1981) in relying

on outgroup comparison as the best criterion for character polarity

decisions. This criterion has been discussed at some length in recent

systematic literature and methods have been outlined for making the most

efficient use of outgroups when they are well established (global

parsimony, Maddison, Donoghue and Maddison, 1984), or when a number of

outgroups could be the sister taxon to the ingroup (outgroup

substitution, Donoghue and Cantino, 1984).

In this study I have employed data from all families of turtles and

the arrangement of Gaffney (1984; Fig. 10) to make use of the concept of

global parsimony. That is to say, the outgroup for the Trionychidae is

all other turtles. Decisions concerning polarity of characters within

this family are most directly affected by the distribution of states

within the Trionychoidea. The concept of the Trionychoidea is based on

characters polarized at a higher level of universality.

Formulation of Phylogenetic Hypotheses

In my provisional arrangement of the Recent species of the

Trionychidae based on 16 characters of shell morphology (Meylan, 1985), I

conducted the search for the most parsimonious cladogram (that requiring

the fewest evolutionary steps) by hand. As additional data have been

assembled for this study I have partitioned them into three sets (shell,

skull, and lower jaw and nonshell postcrania). Nonetheless, as each of

these data sets became very large, it has become necessary to employ a

computer program to generate cladograms. I have used "PAUP" by D.

Swofford, which is available through the Northeast Regional Data Center

at the University of Florida.

The PAUP program emphasizes simple unrestricted parsimony procedures

(Swofford, 1984). Its author finds that there is close correspondence

between results obtained by hand and those generated via PAUP. One

advantage in addition to the time-saving capabilities of PAUP is the

MULPARS option. This option results in a listing of all "most

parsimonious" trees. It seems certain that when working by hand one is

unlikely to discover all such trees. The ability of the program to

handle missing values improves its utility for use in the current


I have employed PAUP to formulate the most parsimonious hypothesis

of relationship for the species within the family Trionychidae that can

be derived from each of the three independent sets of osteological data.

These are 1) 22 characters of the shell (an expanded version of Meylan,

1985); 2) 23 characters of the skull; and 3) 13 characters of the lower

jaw and postcrania (exclusive of the shell). Additionally, an analysis

of the three data sets combined was performed.

Comparison of Fundamental Hypotheses and Formulation of a General

Following the development of cladograms from the three separate

data sets, it was desirable to formulate a single general cladogram from


them and to compare the utility of various characters, especially those

of the shell, in the formulation of this general hypothesis. Two

methods, analysis of the three data sets in combination and a stepwise

consideration of compatible characters, have been used for this

procedure. Neither the Nelson (1979) method or the similar Adams (1972)

method produced a single, well-resolved cladogram of trionychid




The names used for the Recent Trionychidae are, with minor exception

(see introduction), those employed by Wermuth and Mertens (1961).

Specific epithets are often used without a generic name. Because the

generic name Trionyx is currently used with about three-fourths of the

species, little information is conveyed by the use of that name.

Certain collective terms are used provisionally for groups of

trionychid species throughout the text. They are used for groups which

have been suggested to be monophyletic by more than one author. The

Cyclanorbinae (Cyclanorbidae of Deraniyagala, 1939; or Lissemydinae, of

Williams, 1950) includes Cyclanorbis elegans, Cyclanorbis senegalensis,

Cycloderma aubryi, Cycloderma frenatum and Lissemys punctata. These

species are considered to constitute a natural group in treatments by

Deraniyagala (1939), Loveridge and Williams (1957), and Meylan (1985).

The sister group of the Cyclanorbinae is the Trionychinae, which includes

all non-cyclanorbine members of the family. There is good evidence that

this, too, is a monophyletic group (Meylan, 1985). It has also been

recognized as such by Deraniyagala (1939), and Loveridge and Williams

(1957). Within the Trionychinae two species groups have been treated as



natural groups in all recent accounts: the four species of the Indian

subcontinent (Trionyx gangeticus, T. leithii, T. hurum and T. nigricans);

and the three North American forms (T. ferox, T. muticus and T.

spiniferus) (Loveridge and Williams, 1957; De Broin, 1977; Meylan, 1985).

Names of familial and higher taxa of the Testudines follow Gaffney

(1984). Monophyly of these taxa is not reexamined except for the

superfamily Trionychoidea and its member families. The suffixes -oidea

for superfamilies, -idae for families and -inae for subfamilies are used

consistently throughout the Testudines.


Terminology for elements of the carapace and plastron follows

Loveridge and Williams (1957). The concepts of Williams and McDowell

(1952) concerning the anterior lobe of the plastron are rejected. These

authors suggest that the anterior midline element in trionychids is not

the entoplastron, but rather a fused pair of epiplastra, and that the

anteriormost paired elements are neomorphs which they term preplastra.

Bramble and Carr (n.d.) have shown that, on the basis of the sites of

origin and insertion of the anterior trunk musculature, this is incorrect

and that the anterior plastral elements in trionychids correspond to

those of other turtles. The midline element is termed the entoplastron,

and the anteriormost pair are referred to as epiplastra.

For skull and lower jaw terminology, I follow Gaffney (1972, 1979b),

who has developed his glossary of skull morphology in part from Parsons

and Williams (1961). A variety of sources is used for the nonshell

postcrania: Williams (1950) for cervical vertebrae; Baur (1891a) and Zug

(1971) for the pelvic girdle; and Annandale (1912) for the hyoid.


Variation in Shell Morphology


Twenty-seven characters of the carapace and plastron have been

determined to be useful for establishing inter- and/or intrafamilial

relationships of trionychid turtles (Table 1). They pertain to total

shell size and shape and five areas of the shell: the nuchal region, the

neural series, the periphery, posterior end of the carapace, and the

plastron. Because of the unique nature of the shell of trionychids few

of these characters are useful in testing proposed interfamilial


All character polarities discussed in this section are based on

outgroup comparisons. It is therefore important that doubts about the

homology of the shell of trionychids to that of other turtles be

considered. Zangerl (1969) contends that the external bony layer in the

Trionychidae and Dermochelyidae is composed of epithecal ossifications of

more superficial origin than the dermal ossifications considered to form

the shell in other turtles. This implies that the superficial layer of

the shells of members of these two families are not homologous to the

same layer in other turtles. The existence of a nonhomologous

superficial layer seems quite possible for Dermochelys in which there is

total independence of the superficial bone and the deeper dermal elements

of the shell (i.e., the ribs and neural spines of vertebrae). In cross

sections these "epithecal bones" which make up the superficial bony

mosaic, lack dense layers on the external and internal surfaces (Fig. 1).

Thus they do not fit Zangerl's (1969) description of turtle shell bone of

typical dermal origin. The case is less clear for the most superficial

bony layer in the Trionychidae. In members of this family, as in other

turtles, there is complete correspondence between superficial bony

elements and underlying deep dermal elements of the carapace.

Furthermore, cross-sections of either carapacial or plastral elements of

trionychids reveal the presence of a spongy middle region with compact

lamellar layers on either side (Fig. 1). This agrees with Zangerl's own

description of typical dermal shell bone and fits Suzuki's (1963)

description of the results of development of dermal shell bone in

Pseudemys script. Zangerl's (1939) original argument for an epithecal

origin of the superficial bone in trionychids is based on its delayed

development rather than on its site of origin. The late development of

the superficial layer does not have any clear bearing on the homology of

its origin, and must yield to the physical evidence that in cross section

trionychid shell elements do not differ significantly from other

sectioned chelonian shells which are considered to be of normal dermal

origin. Thus, unless other evidence can be provided, the superficial

elements of trionychid shells may be regarded as homologous to those of

other turtles and presumably of equivalent origin.

Shell Size and Shape

Even the smallest fragment of trionychid shell is immediately

recognizable by its characteristic sculpturing. This sculpturing is

never divided by scute sulci because the sulci and the epidermal scutes

they delineate, which are present on the shells of most other turtles,


are always absent in trionychids. The only other living turtle which has

a sculptured shell and lacks epidermal scutes is Carettochelys (character

30, Table 8).

Recent trionychids are, for the most part, large turtles and many

species approach one meter in total carapace length. The carapace

consists of a bony disc with cartilaginous margins. In discussion of

osteological material, including this one, it is the bony disc length

rather than total carapace length which is used. The largest species of

trionychids have bony discs over 500 mm in length; most reach disc

lengths of 300 mm (Table 2). The exceptions are few, and these are

usually 200 mm or less in disc length.

Six species of Trionyx are small: Trionyx muticus, T. spiniferus,

T. steindachneri, T. sinensis and T. subplanus. All of the carapacial

discs of T. subplanus measured during the course of this study are under

180 mm, but one unusually large skull, BMNH (figured as T.

cartilagineus, Dalrymple, 1977), could have come from a specimen with a

disc as large as 250 mm. Awaiting complete analysis of the relationship

of head to shell size in this megacephalic form, T. subplanus is

tentatively included among the smaller species. This list agrees in part

with a list of diminutive forms assembled by De Broin (1977) based on

skull size. Her inclusion of T. leithii and T. ferox as small forms,

however, was clearly an artifact of small sample size.

Among other trionychoids, small size is common only in the

Kinosternidae. Most known species of the Dermatemydidae and

Carettochelyidae reach bony carapace lengths of 400-500 mm. Among the

Kinosternidae the genus Staurotypus reaches adult sizes close to those of

Dermatemys and Carettochelys; whereas Claudius, Kinosternon and

Sternotherus are smaller, usually under 200 mm. It seems likely that

reduction in total size is a derived condition common to the

Kinosterninae and that similar diminution occurred independently in one or

more groups within the Trionychidae. Thus, small carapace size is

considered to be a derived condition among trionychids (character 23,

Table 3).

Sexual dimorphism in total size is well known for turtles. In

certain forms the male is larger and in others the female is larger. The

latter occurs most frequently among aquatic emydids but also occurs in

some trionychids. Webb (1962) provides data which indicate that all

three North American forms are sexually dimorphic in size. This has not

been shown for any Old World forms with the possible exception of Chitra

indica (Wirot, 1979). Due to its apparent absence among other

trionychoids sexual dimorphism, in which the female is larger, can be

considered a derived feature within the Trionychidae (character 28, Table


The carapace of trionychids is unique among the Testudines in having

a cartilaginous margin. This margin varies in extent and thus in

flexibility. In one species (Lissemys punctata) it makes up less than

10% of the total carapace length and has in it bony elements which are

most likely homologous to the peripherals of other turtles. In other

forms the cartilaginous margin makes up almost one-half of the carapace

length and the bony disc is thus quite reduced.

There can be little doubt that reduction of the bony disc relative

to the total carapace is a derived condition, as it occurs only within

this family. However, variation in this condition among trionychid

species shows no natural breaks and I have not been able to convert this

continuous variable into a discrete one. It should be pointed out,

however, that cyclanorbines consistently have relatively larger discs

than trionychines and in this respect they represent the more primitive


Elsewhere (Meylan, 1985) I have suggested that the shell outline of

Cycloderma frenatum is unique in having a sharply tapering rear half of

the carapace with straight to concave posteriolateral edges. After

examination of numerous carapaces of Cycloderma aubryi and Lissemys

punctata, it is apparent that these species share the unique carapacial

outline noted above. Other trionychids, like most other testudines, have

round-to-oval shells that are convex posteriolaterally (character 25,

Table 3).

Nuchal Region

Dalrymple (1979) provided an excellent discussion of the role of

the cervico-dorsal joint in trionychids in allowing the retraction of a

long neck into a small space. In order to accommodate such modification

of this joint, the entire anterior portion of the trionychid carapace

must have been extensively remodeled. In most cryptodires, the first

body vertebra is directly ventral to the first neural bone of the

carapace and is firmly sutured to it. It is loosely jointed and usually

more anteriorly located in trionychids. In Lissemys and Cycloderma, the

first body vertebra lies directly below the "preneural" to which it is

weakly sutured, suggesting that the "preneural" is actually a first

neural (see also Baur, 1893; Hay, 1908; Carpenter, 1981). The nuchals of

Lissemys and Cycloderma are also the longest (relative to their width)

among the trionychids (Fig. 2). Separate anterior and posterior


costiform processes can be recognized. Grooves for the postzygapophyses

of the eighth cervical vertebra are present on either side of the midline

at the base of the posterior costiform process. This combination places

the well-fixed first body vertebra well back from the edge of the

carapace (Fig. 2B). Among trionychids the condition in Lissemys and

Cycloderma most closely approaches that seen in other cryptodires.

Further derived conditions apparently arose as contact between the first

body vertebra and the first neural was reduced, and the nuchal came to

lie above the first body vertebra.

An advanced condition of the nuchal region appears in Cyclanorbis

senegalensis, in which the length of the nuchal bone is reduced, bringing

the first body vertebra closer to the anterior edge of the carapace. At

this stage the anterior and posterior costiform processes of the nuchal

are not clearly separate (Fig. 2C), but the first neural (preneural) is

still distinct from the second. A similar condition is found in

Cyclanorbis elegans and in Trionyx gangeticus, T. leithii, T. nigricans,

and T. hurum.

Presumably, fusion of the first neural to the second occurred only

after the first body vertebra, through reduction of its neural arch, had

become free of the first neural. This evidently left the overlying

neural element available for further modification. Fusion of the first

and second neurals occurs in all Chitra, Pelochelys, and Recent Trionyx

except T. hurum, T. leithii, T. nigricans, and T. gangeticus (except for

one BMNH specimen that is kyphotic). Up to about 10% of certain Trionyx

species (T. ferox, T. formosus, T. triunguis) show separate first and

second neurals.

The extreme of development in this suite of characters is found in

Chitra (Fig. 2D). In C. indica, prezygapophyses of the first body

vertebra are immediately adjacent to the anterior rim of the carapace,

and the nuchal is reduced to a narrow sliver of bone. A single costiform

process occurs on the anterior margin, and depressions which allow

passage of the postzygapophyses of the eighth cervical are present just

inside the rim of the carapace. In Chitra there is also a new pair of

processes at the posterior edge of the nuchal.

Variation in the nuchal region has been analyzed through the use of

four characters (Tables 1, 3). The primitive condition for nuchal shape

(character 1) is that most similar to that of other turtles, i.e., as

wide or nearly as wide as long. Costiform processes (character 2) are

not present in adult Carettochelys, but in some juveniles of the related

genus Anosteira, there are two pairs (Bramble, pers. comm.). In other

trionychoids (Kinosternidae and Dermatemydidae) there is one pair in

adults. But in newly hatched Dermatemys (BMNH 1984.1291) there are, in

fact, two pairs. A cleared and stained hatchling Sternotherus minor in

the UF collection also has two pairs of costiform processes. Thus it

seems likely that two pairs are present early in the ontogeny of all

trionychoids. In the Dermatemydidae and Kinosternidae the anterior of

the two pairs disappears with age while in the Trionychidae the two pairs

occur separately in some forms (Lissemys punctata and both species of

Cycloderma) and appear to be united in all others. Because it is the

condition common to all trionychoids the possession of two pairs of

costiform processes is considered primitive for trionychids.

Most cryptodires have the first body vertebrae at the posterior edge

of the nuchal. Through evolutionary foreshortening, the anterior edge of


the nuchal apparently moves toward the first body vertebra. Proximity of

the anteriormost body vertebra to the margin of the carapace is

considered derived (character 3, Table 3).

The trionychid "preneural" is here considered to be the first neural

(see also Hasan, 1941). As suggested by Webb (1962) and Gaffney (1979c),

fusion of the first neural to the second neural must be a derived

character state (character 4). No non-trionychid member of the

Trionychoidea has two neurals between the first pleurals, but there are

two body vertebrae between the first pleurals of all trionychids. In T.

ferox two neurals form (one on each of the first two body vertebrae) and

then they fuse into a single element (Carpenter, 1981; present study).

The carapace of adult turtles is ordinarily a solid bony structure

without openings or fontanels. Peripheral fontanels are not uncommon:

they occur in juveniles of all cryptodires and are retained in some adult

chelydrids, cheloniids, and trionychids. In trionychids, peripheral

fontanels are difficult to visualize because the peripheral bones are

lacking. Fontanels on the midline are much less common. They occur

above the ilia in very old individuals of some testudinoids (e.g.,

Terrapene, Cuora, Gopherus, Homopus) and above the scapulae

(=suprascapular fontanels) in certain trionychids and at least one

testudinid, Homopus.

Suprascapular fontanels are probably present early in the

development of all trionychids. They are closed at hatching in some

forms (Lissemys punctata) but remain open throughout life in others

(Trionyx subplanus, T. spiniferus [except some old males], T. muticus and

T. steindachneri). In most trionychids suprascapular fontanels close up

at some point between hatching and adult size (Table 4). Insufficient

data on the timing of closure in most species prevents the use of this

character. Early loss of the fontanels is likely the primitive condition

and life-long retention derived.

The Neural Series

The above argument shows that in the Trionychidae the

"preneural" of many authors is probably the first neural. Thus the most

complete neural series in trionychids includes nine elements between the

nuchal and eighth pleurals. The normal pattern in cryptodires is a

continuous series of neurals from the nuchal to the suprapygal, usually

with some uniform orientation. All trionychids lack a suprapygal, and

the eighth pleurals meet at the midline (except in Trionyx subplanus).

The most complete series of nine neurals, with all or the majority

(numbers 2-7) hexagonal and uniformly facing posteriorly, is likely to be

the most primitive condition among living trionychids (Fig. 3A). It is

also the condition present in the Jurassic species Sinaspideretes wimani

Young and Chow (1953), which appears to be the oldest known trionychid.

Modification of the presumed primitive condition results from four

apparently independent changes: (1) the fusion of the first and second

neural (treated above, character 4); (2) interruption of the neural

series by pleurals meeting at the midline (character 15); (3) variation

in the number of neurals expressed on the dorsal surface of the carapace

(character 13, Table 5); and (4) variation in the location at which

orientation of the neurals reverses (character 16, Table 6). There are

also interspecific differences in the amount of variability in the point

of neural reversal (character 14). That is to say, in some species the

location of reversal is always the same neural; in others, reversal

occurs only at either of two adjacent neurals; and in still others,

it may occur anywhere along the neural series (Table 7).

Interruption of the neural series by pleurals meeting at the midline

is not common among cryptodires. Most species have a neural series which

is uninterrupted from the nuchal to the suprapygals. In dermatemydids

and kinosternids, posterior pleurals may meet on the midline but in this

case the posteriormost neurals do not appear so they can not be isolated

from the anterior portion of the series. In Carettochelys pleurals often

meet along the midline, isolating sections of the neural series. The

neurals of Carettochelys are quite narrow and thus appear to be less

generalized than those of trionychids. Relying on global parsimony in

establishing polarity in this case, the absence of pleural interruption

of the neural series must be considered primitive for the Trionychidae.

Actually, interruption of the neural series is rare in trionychids.

The last neural is isolated from the rest of the neural series in

occasional specimens of Lissemys punctata (2 of 19), Trionyx ferox (5 of

31), T. gangeticus (1 of 7) and T. hurum (1 of 5). More frequent neural

isolation occurs only in the two species of Cyclanorbis. Siebenrock

(1902) discussed the marked variability of the neural series in these two

species in his paper which establishes the existence of the two forms on

osteological grounds. Both Cyclanorbis species can have long continuous

rows of neurals or many isolated neurals. Although C. senegalensis tends

to have more isolated neurals than C. elegans, the most reliable

diagnostic features of these two species are found in the plastron. C.

senegalensis is unique among living trionychids in possessing gular

callosities. C. elegans is unique among cyclanorbines in having


callosities of the fused hyo-hypoplastra that are flat or concave along

their anterior edge.

The number of neurals appearing on the surface of the carapace in

trionychids varies from three to ten. The occurrence of a tenth neural

is very rare (3 of 242 specimens, two Trionyx subplanus and one T.

cartilagineus) and seems to be anomalous. Thus, nine neurals form the

most complete series, and the possession of nine neurals is considered to

be the fundamental condition for trionychids. This is not supported by

evidence from the outgroups. The entire superfamily seems to have lost,

or be in the process of losing, posterior neurals which makes arriving at

a primitive number based on the trionychoids quite difficult.

Looking outside of the Trionychoidea, one finds nine neurals

commonly in the Chelydridae, where they are packed closely together

posteriorly. In the Cheloniidae, Chelonia mydas and E. imbricata

frequently have two neurals between the first pair of pleurals, as is

proposed to be primitive for trionychids (see for example Fig. 85 in

Deraniyagala, 1939). Other sea turtles have higher numbers of neurals

but this is due to division of neural elements (Zangerl and Turnbull,

1955); nine neurals may actually be the primitive number for these

species as well.

Variation in the number of neurals among living trionychids is given

in Table 5. The number of neurals (character 13, Tables 1, 3) is treated

as five character states, with nine neurals considered most primitive and

seven or fewer neurals most derived.

Nearly all neurals of trionychids are six-sided. Anterior and

posterior ends of each neural contact adjacent neurals, the four lateral

sides contact adjacent pleurals. The lateral sides consist of two


unequal pairs (Fig. 4). In the anterior part of the neural series the

short lateral sides face posteriorly, but in the posterior part of the

series (in most species) they face anteriorly. Thus, there is usually a

reversal in orientation of these anterio-posteriorly assymetrical

elements in every neural series.

Reversal of orientation occurs in two ways (Fig. 4). More commonly

it occurs via a four-sided neural (= a diaphragmaticc" neural of Hummel,

1929). This four-sided neural and the adjacent pleurals contact the

three posterior-facing short sides of the next anterior neural, and the

three anterior-facing short sides of the next posterior neural (Fig. 4A).

The second and less common reversal occurs via two successive

assymetrical pentagonal neurals (Fig. 4B). The anterior of the pair

contacts an anterior short side of one of the next posterior pair of

pleurals, while the posterior neural contacts a short posterior side of

the preceding pleural on the opposite side.

In the presumed primitive neural arrangement, reversal of neural

orientation, if it occurs at all, is posteriorly located. But in many

forms, reversal occurs anteriorly and this is considered to be derived.

Such reversal usually accompanies other changes from the primitive neural

configuration. Reversals can occur from neural one through eight and

multiple reversals are common in some species (Tables 6, 7). Where

multiple reversals occur the location of the most posterior one is

thought to indicate the degree of anterior migration of neural reversal.

Data on location of neural reversal are treated as five states of

character 16 (Tables 1, 3), with most anterior being most derived. Data

on the amount of intraspecific variability in the location of the last

neural reversal are treated via three states of character 14 (Tables 1,

3), with the most variable being considered most derived.

Shell Periphery

With the exception of the Trionychidae, the margins of all

testudine carapaces are rigid. This is due to the presence of peripheral

bones that form a complete ring around the carapace. In nearly all

turtles this ring is composed of 22 peripherals, a nuchal and a pygal.

Only in the Trionychoidea is there reduction and complete loss of any of

these elements. In all kinosternids and Carettochelys there is one fewer

peripheral on each side (total of 20). The peripherals of Carettochelys

are not sutured to the pleurals, which is also true for the only

trionychid which retains bones in the periphery, Lissemys punctata

(character 5A, Table 3). The homologies of the bones in the shell

periphery of Lissemys have been questioned by many authors. Boulenger

(1889), Loveridge and Williams (1957), Zangerl (1969) and others have

considered these bones to be neomorphic structures. Walther (1922), Webb

(1982) and Meylan (1984) have treated the peripherals of Lissemys as

homologs of the peripherals of other turtles. Although these elements

are found only in the carapace posterior to the bridge and they lack one-

to-one correspondence with the pleurals, there is other evidence which

suggests that they are degenerated peripherals and not neomorphs. In

cross-section the peripheral ossifications of Lissemys are like those of

other turtles in that they consist of two laminar layers of bone which

converge distally (Fig. 5). Between these two layers is cancellous bone.

Lissemys peripherals differ from those of other turtles principally in

the absence of the proximal portion. Unless some developmental

constraint that results in the formation of V-shaped elements in the

periphery of all turtle shells can be identified, it may be best to

consider these details of morphology as evidence of homology.

Peripherals are found in the carapace of Lissemys only posterior to the

bridge and are usually about 14 in number (Deraniyagala, 1939).

Peripherals are absent in all other trionychids. The reduction and loss

of bones in the periphery is clearly derived (character 5, Tables 1, 3,


Although the rib heads of each pleural bone normally reach the

centrum of the corresponding neural, the contact is not always a strong

one. Only in trionychids and Carettochelys among the Cryptodira have I

found strong, interlocking sutures (character 28, Tables 1, 8). Dennis

Bramble (pers. comm.) has suggested that the peripheral bones of most

turtles form a locking ring around the shell that keeps it from expanding

laterally when a dorso-ventral force is applied and that these

strengthened contacts between the rib-heads and centra may be an

alternate means of countering such forces. Thus the carapace of

Carettochelys may be "preadapted" for the loss of peripherals.

Both Lissemys punctata and Cyclanorbis senegalensis possess a

prenuchal that is an isolated element that lies above the neck, just

anterior to the nuchal (character 6, Tables 1, 3). The prenuchal is a

neomorph not found in any other cryptodire, and its appearance is a

derived condition.

Posterior End of Carapace

In nearly all turtles, the eighth and last pair of pleurals forms as

significant a portion of the carapace as those which precede it.

Although the eighth pleurals of trionychids develop allometrically, being

relatively larger in adult turtles than in juveniles, it is still

possible to detect a difference in their size among species. In some

forms they are large, in others somewhat reduced, and in yet others they

are absent. The presence of large eighth pleurals provides a complete

complement of pleural bones. The reduction of this complete complement

is considered to be derived. Large eighth pleurals are present in all

cyclanorbines as well as all Old World trionychines, except Trionyx

euphraticus. There is a trend toward the loss of the eighth pleurals in

New World forms (character 8, Tables 1, 3).

The ilia of cyclanorbines, except Cyclanorbis elegans, articulate

with the eighth pleurals, as they do in other cryptodires. In all

trionychines and in C. elegans the ilia articulate with the tough

connective tissue just posterior to the end of the shell. The presence

of depressions for the ilia on the 8th pleurals is considered primitive,

their absence derived (character 20, Table 3).


The plastron of most cryptodires includes nine elements (one pair

each of epi-, hyo-, hypo- and xiphiplastra and a single entoplastron).

These nine elements are well sutured to one another and form a solid bony

structure. The same nine elements are present in all trionychids

(Bramble and Carr, n.d.) but they are relatively incomplete; they are

often not sutured to one another and do not normally result in a single

solid structure. Where plastral sutures are present in trionychids they

occur between the superficial dermal callosities with minor contributions

from underlying elements. The presence of sutures, and thus of the

callosities that allow them to occur, is interpreted as a primitive



A suture is found between the hyo- and hypoplastra of all

trionychids, and in many fusion occurs along this suture. The

xiphiplastral callosities make contact at the midline in large Lissemys,

Cycloderma, and Trionyx, but only in Lissemys punctata and Cycloderma

aubryi does a sutured contact occur. This suture fuses in very old

individuals of these two species. Sutures are absent between epi- and

entoplastron, entoplastron and hyoplastra, hypo- and xiphiplastra, and

along the midline (except for the xiphiplastra of the two species noted

above) in all Recent trionychids.

The number of plastral callosities in all trionychids increases with

age but is stable in large adults (character 9). Callosities are present

on all nine plastral elements in certain species and this is proposed as

the primitive condition. The callosities covering the hyo- and

hypoplastron on each side are here considered to be a single structure

making seven the primitive number. Seven callosities are found in

Lissemys, Cycloderma and some Trionyx. Derived conditions include both

an increase and a decrease in the number of callosities (character 9,

Tables 1, 3). Only Cyclanorbis senegalensis has increased the number of

callosities by the addition of a gular pair. The cyclanorbine

Cyclanorbis elegans parallels the trend in the Trionychinae in having

marked reduction in the number of callosities to two.

Although the fusion of two plastral elements is certainly derived,

it can occur only when the primitive condition, a suture between two

elements, is present. Thus the xiphiplastral suture in Lissemys punctata

and Cycloderma aubryii suggests that they are primitive. However,

xiphiplastral fusion is unique to these forms among trionychids and is

considered a shared derived state (character 11, Table 3).


Hyo-hypoplastral sutures occur at some stage in the ontogeny of all

extant trionychids. Recent cyclanorbines share the common character

state of hyo-hypoplastral fusion at a very small size (as small as 62 mm

disc length). Fusion of the hyoplastra to the hypoplastra occurs in all

adult Trionyx ferox and in adults of some populations of T. triunguis.

The presence of hyo-hypoplastral fusion is considered to be derived and

to occur independently in cyclanorbines and trionychines (characters 10A,

10B, Tables 1, 3).

The xiphi-hypoplastral union in trionychids is of two types. In all

trionychines the two anterior xiphiplastral processes lie on either side

of the most lateral of the three posterior processes of the hypoplastron.

In cyclanorbines the two anterior processes of the xiphiplastron lie on

either side of the middle of the three posterior processes of the

hypoplastron. The trionychine condition occurs in cheloniids,

Carettochelys and among kinosternids (Kinosternon, Sternotherus, and

Staurotypus), suggesting that it is the primitive condition. Thus the

presence of the hypoplastron lateral to the xiphiplastron at their

junction is considered to be a derived condition unique to cyclanorbine

trionychids (character 12, Table 3).

Relative to the epiplastra of other Testudines those of trionychids

are quite reduced in basic structure. The deep element, which may or may

not be covered by a callosity, is I- or J-shaped. The J-shaped elements

have a long ramus that is oblique to the midline and has a long contact

with the entoplastron. They also have an anterior projection of variable

length that roughly parallels the midline. I-shaped elements consist of

only the anterior portion and have minimal contact to the entoplastron.

J-shaped epiplastra are found in all trionychids except Lissemys

punctata, Cycloderma aubryi, and Cycloderma frenatum, which have the

alternate I-shape.

Long medial contact between the epiplastra and the entoplastron

occurs in all turtles in which these elements are present. The posterior

contact of the J-shaped epiplastra to the entoplastron maintains this

contact and thus the J-shape is considered primitive, the I-shaped

derived (character 18, Table 3).

The anterior extension of J-shaped epiplastra varies in length among

the species in which it is found. The extension beyond the entoplastron

varies from 0.165 to 0.48 times the width of the hypoplastron of the

right side (Table 9). It is difficult to be certain which length of the

extension is primitive for trionychids but it seems clear that the marked

extension of Trionyx cartilagineus, T. subplanus, T. sinensis and T.

steindachneri is derived. As suggested by De Broin (1977), the species of

the Indian subcontinent have epiplastra of intermediate length relative

to the most elongate forms and other trionychids. Variation in this

feature is treated as three states of character 19 with the longest

extension considered to be most derived (character 19, Table 3).

The boomerang shape of the entoplastron of trionychids is unique

among turtles (character 21, Table 8). Zangerl (1939) has implied that a

T-shaped entoplastron is primitive for reptiles. The entoplastron in

trionychids apparently arises from a proliferation and bending of the

transverse portion of the "T", combined with suppression of development of

the longitudinal portion. The amount of bending of the transverse bar

varies among trionychids and results in an angle of 62 to 122 degrees

between the two posteriolaterally directed rami. Variation within each

species spans about 15 degrees. Variation among species is quite

continuous, with no natural breaks. Establishing a polarity for this

character has not been possible because no other members of the

Trionychoidea have similar entoplastron morphology. This fact, combined

with problems of variability, has made it impossible to include angle of

the entoplastron as a character in intrafamilial analysis.

Plastral reduction in trionychids includes a marked reduction in the

length of the bridge. Bridge length was compared to hypoplastron width

as an index of this reduction. Bridge length varies from more than

three-quarters of hypoplastron width (Cycloderma aubryi) to about one-

eighth hypoplastron width (Trionyx subplanus). But variation falls into

two discrete groups: those species in which the bridge is well over one-

half hypoplastron width, and those species with a bridge less than one-

half hypoplastron width. The former group includes all cyclanorbines

except Cyclanorbis elegans; the latter includes all trionychines plus

Cyclanobis elegans.

Long plastral bridges occur in Dermatemys and Carettochelys but not

in kinosternids. They are also long in testudinoids, with the exception

of the most kinetic forms. Thus a long bridge is considered to be

primitive, a short bridge derived (character 22, Tables 1, 3, 8).

In addition to being short, the bridges of trionychid turtles lack

ascending buttresses and sutured contacts to the elements of the

carapace. Ascending processes cross the peripherals in both the axillary

and inguinal regions in pleurodires and testudinoids except for those

taxa with well develop plastral kinesis. In Dermatemys only the axillary

buttress reaches the pleurals. In all other living families the

buttresses are quite reduced and do not cross the peripherals (character

26, Tables 1, 8). The distribution of the states of this character can


be explained about as parsimoniously by loss or by gain of buttresses if

only Recent forms are examined. Buttresses occur in such extinct

families as the Baenidae, Plesiochelyidae, and Meiolaniidae, however,

suggesting that their presence is in fact the primitive condition.

In a few taxa that lack large plastral buttresses, the bridge is

further weakened by the absence of strong sutures between the carapace

and plastron. This occurs in chelydrids, cheloniids, Claudius,

Carettochelys and trionychids. The absence of sutures at the bridge is

considered derived (character 27, Tables 1, 8).

Figure 1. Cross sections of the shell of three cryptodiran turtles.
Top, Trionyx ferox, UF 54212, X 10; middle, Chrysemys picta, UF 37557,
X 5; bottom, Dermochelys coriacea, UF 37557, X 20.

Figure 2. Internal view of the nuchal region of four trionychoid
turtles. A, Carettochelys insculpta, UF 43823; B, Cycloderma frenatum,
UF 52704; C, Trionyx ferox, UF 53383; D, Chitra indica, PCHP unnumbered.

Figure 3. Variation in the neural series in the Trionychidae. Arrows
indicate proposed progression of change in the trionychid neural series.
A, Cycloderma frenatum, TM unnumbered; B, hypothesized intermediate
condition; C, Trionyx formosus, BMNH 1947.3.6.9; D, Trionyx ferox, PCHP
1171; E, Lissemys punctata, BMNH; F, Cyclanorbis senegalensis,
after Villiers, 1955.

Figure 4. Two types of reversal in the neural orientation of
trionychids. A, reversal via a four-sided neural; B, reversal via two
assymetrical pentagonal neurals.

Figure 5. Cross sections of the peripherals of two cryptodires. Top,
Lissemys punctata, UF 56017; bottom, Chrysemys picta, UF 40615;
both X 20.

Table 1. Shell characters and character states used for resolving
phylogenetic relationships of Recent trionychid turtles. For each
character the most primitive state is number 1.


1) width/length of nuchal bone

Character States

1) less than 2
2) greater than 2
3) greater than 3
4) greater than 4

2) anterior and posterior costiform
processes of nuchal bone united

3) position of anterior edge of first
body vertebra relative to nuchal

4) first and second neurals fused

5) total number of peripherals

5A) peripherals sutured to pleurals

6) prenuchal bone

8) size of eighth pleurals

9) number of plastral callosities

10A) hyoplastra and hypoplastra
fuse just after hatching

10B) hyoplastra and hypoplastra fuse
in adults

11) fusion of xiphiplastra

1) no
2) yes

posterior edge of nuchal
middle of nuchal
anterior edge of nuchal

1) no
2) yes


1) yes
2) no

1) absent
2) present

1) large
2) reduced or absent





Table 1--continued.

Character States

12) hypo-xiphiplastral union

13) number of neurals (fused 1 and
2 counted as 2)

14) variability in position of
neural reversal

15) pleurals which meet at the

16) point of reversal of orienta-
tion of neurals

18) epiplastron shape

19) length epilastra anterior to
entoplastron contact

20) depressions on eighth pleurals
for contact of ilia

21) shape of entoplastron

xiphiplastra lateral to
hypoplastra lateral to

eight or nine
seven or eight
seven or fewer

always at same neural
always at adjacent neurals
highly variable

eighth only
seventh and eighth or
eighth only
sixth, seventh and eighth
or seventh and eighth
more than sixth, seventh
and eighth

at neural eight
at neural seven
at neural six or seven
at neural six
at neural four, five or six

1) J-shaped
2) I-shaped


1) present
2) absent

anterio-posteriorly elongate
or round

22) bridge length


1) long

Table 1--continued.

Character States

23) largest adult size 200 mm or
less (disc length)

25) carapace straight or concave

26) plastral buttresses reach across
peripherals to contact pleurals

27) carapace sutured to plastron all
across bridge

28) rib heads strongly sutured to
vertebral centra

29) sexual dimorphism in disc

30) shell sculptured and lacking
epidermal scutes

1) no
2) yes

1) no
2) yes

1) both axillary and inguinal
2) axillary only
3) neither

1) yes
2) no

1) no
2) yes

1) no
2) yes

1) no
2) yes


Table 2. Maximum known size of Recent trionychids (character 23).

Length of
Species Specimen Bony disc (mm)























BMNH 61.7.29


ZSM 832/1920

NMW 1437

cited in Siebenrock, 1913

UF 45341*

cited in Annandale, 1912

BMNH (Type of Aspidochelys livingstoni)

cited in Annandale, 1912

cited in Annandale, 1912

MNHNP 1880-182

EOM 2819

UMMZ 128086

cited in Annandale, 1912

cited in Deraniyagala, 1939

BMNH 1949.1.3.51

ZSM 429/1911

UF 37228

MNHNP unnumbered

calculated from skull BMNH

calculated from Fig. 1A, Heude, 1880


*Allen (1982) reports a larger specimen of Trionyx ferox but a disc
length is not available.























Table 3. Modal character states for shell characters of the Recent
Trionychidae that are used in the analysis of intrafamilial
relationships. For descriptions of the characters and character states
see Table 1. Periods indicate missing values.


Species 1 2 3 4 5 6

8 9 10A 11























2 1 2 1 4 1

3 2 2 2 4 1

3 2 2 2 4 1

2 2 2 1 4 1

3 2 2 2 4 1

3 2 2 2 4 1

2 2 2 2 4 1

2 1 1 1 4 1

3 2 2 1 4 1

3 2 2 1 4 1

3 2 3 2 4 1

3 2 2 1 4 1

4 2 2 2 4 1

3 2 2 1 4 1

2 1 1 1 3 2

3 2 2 1 4 2

4 2 2 2 4 1

3 2 2 2 4 1

2 2 2 2 4 1

4 2 2 2 4 1

2 4 1

3 2 2 2 4 1

1 1 2

1 3 1

1 2 1

1 4 2

2 4 1

2 3 1

1 3 1

1 1 2

1 2 1

1 2 1

1 3 1

1 2 1

2 1 1

1 2 1

1 1 2

1 0 2

1 1 1

2 1 1

1 3 1

1 3 1

2 4 1

1 3 1

Table 3--continued.


13 14 15 16 17 18 19 20 22 23 25 29























2 1 2 1 1 2 1 1 1 1

2 1 1 3 2 1 1 2 2 1

1 1 1 2 2 1 3 2 2 1

2 1 2 1 1 1 1 2 2 1

3 1 2 3 2 1 1 2 2 1

3 3 2 4 2 1 1 2 2 1

1 1 1 2 1 1 1 2 2 1

2 1 2 2 1 2 1 1 1 1

2 2 2 3 1 1 2 2 2 1

1 2 2 3 1 1 2 2 2 1

1 1 1 3 1 1 1 2 2 1

2 1 1 3 1 1 2 2 2 1

2 3 2 4 3 1 1 2 2 2

1 1 2 3 1 2 2 2 1

4 1 2 2 1 2 1 1 1 1

5 4 1 1 1 1 1 1

2 3 2 4 2 1 2 2 2 2

3 3 2 4 3 1 1 2 2 2

2 1 2 2 1 1 3 2 2 2

1 2 0 2 3 1 3 2 2 2

1 1 2 2 1

3 1 2 3 2 1 1 2 2 1


2 1

1 1

1 1


1 1

1 2

1 1

2 1

1 1


1 2?

I 1

1 2

1 1

2 1

1 1

1 1

1 2

1 1

1 1

1 1

1 1

Table 4. Occurrence of suprascapular fontanels (character 17) in the
carapace of Recent trionychids. Disc length for the largest specimen
with fontanels (A), smallest specimen without fontanels (B), and largest
specimen examined for fontanels (C) are given for each species.

Species A B C





































males 89.5
females 186.5












































Table 5. Number of neurals in Recent trionychid turtles. Values
represent the frequency of occurrence for the sample. A fused first and
second neural is counted as two elements. Trionyx formosus, T. nigricans
and T. swinhoei are excluded due to insufficient sample size.


N 3 4 5 6 7 8 9 10




















0.06 0.06 0.18 0.47



0.10 0.20


0.21 0.43


0.06 0.88

0.20 0.20





0.21 0.74

0.18 0.06


0.06 0.88




















Table 6. Location of reversal in neural orientation in Recent
trionychids. Location of the most posterior reversal is given as a
frequency of occurrence at or between neurals. Values which do not sum
to 1.0 are due to individuals with no neural reversal (see Table 7).
Trionyx formosus, T. nigricans and T. swinhoei are excluded due to
insufficient sample size; in T. senegalensis the neural series is too
fragmented to allow the detection of reversals.

4/5 or
N anterior 5

5/6 6 6/7

7 7/8 8

0.12 0.06

0.10 0.30 0.50 0.10



0.28 0.17 0.44 0.06 0.11


euphraticus 6


0.17 0.33 0.33 0.17

31 0.16 0.13 0.19 0.32 0.19


gangeticus 7








0.40 0.20 0.40

0.08 0.92

3 0.33


0.14 0.14 0.57 0.14

25 0.08

0.08 0.12 0.28

spiniferus 18 0.28 0.11 0.33 0.06

steindachneri 3

subplanus 8

triunguis 14

0.17 0.06

0.33 0.33

0.25 0.63

0.08 0.69 0.08 0.15




0.20 0.40




0.32 0.11

0.40 0.04



Table 7. Number of reversals of orientation in the neural series of
Recent trionychids. Number of reversals is given as a frequency.

Species N 0 1 2 3

aubryi 17 0.82 0.18

bibroni 10 1.00

cartilagineus 18 1.00

elegans 14 0.57 0.43

euphraticus 6 1.00

ferox 31 0.66 0.31 0.03

frenatum 5 0.20 0.80

gangeticus 7 1.00

hurum 5 1.00

indica 13 1.00

leithii 3 1.00

muticus 7 0.86 0.14

punctata 19 0.58 0.42

sinensis 25 0.80 0.16 0.04

spiniferus 18 0.44 0.28 0.28

steindachneri 3 0.33 0.67

subplanus 8 1.00

triunguis 14 1.00

Table 8. Modal character states for shell characters of the Recent
Trionychidae that are used in the analysis of interfamilial
relationships. For descriptions of characters and character states see
Table 1.


5 5A 21 26



28 30













1/4 2 2 3

2 2 1 3

2 1 1 3

2 1 1 3

2 1 1 3

1 1 1 2

1 1 1 3

1 1 1 3

1 2 1 3

1 I

1 1

1 1

1 1

1 1

1 1

* except in kinetic forms

1 1

1 1

1 I

1 1

1 1

1 1

1 1

1 1








Table 9. Extension of the right epiplastron beyond the entoplastron
relative to total hypoplastron width of the right side (character 19).
Sample size, average and one standard deviation are given for each
species. Species with I-shaped epiplastra and T. nigricans are not

Species N X + 1 S.D.

bibroni 3 0.165 0.042

cartilagineus 4 0.482 0.022

elegans 2 0.356 0.019

euphraticus 2 0.263 0.025

ferox 13 0.228 0.019

formosus 1 0.287

gangeticus 2 0.314 0.038

hurum 2 0.358 0.010

indica 5 0.230 0.011

leithii 3 0.312 0.021

muticus 3 0.183 0.017

senegalensis 3 0.280 0.015

sinensis 13 0.423 0.036

spiniferus 10 0.248 0.024

steindachneri 2 0.418 0.014

subplanus 5 0.479 0.039

swinhoei 1 0.221

triunguis 3 0.228 0.023

Variation in Skull Morphology


The value of the trionychid skull in systematics has been recognized

by numerous authors (Gray, 1864, 1869, 1873a, 1873b; Boulenger, 1889;

Hummel, 1929; Loveridge and Williams, 1957; De Broin, 1977). As pointed

out by Loveridge and Williams (1957) there has been too much emphasis on

the size and form of the jaws and too little on details of morphology and

contacts of skull elements. Numerous authors have expressed concern

about the validity of characters of the size and shape of the jaws

(Boulenger, 1889; Villiers, 1958; Barghusen and Parsons, 1966; Eiselt,

1976; De Broin, 1977). But only Dalrymple's (1977) account of variation

in the skull of Trionyx ferox treats the correlation of skull size and

shape to environmental factors in a detailed and systematic fashion.

Dalrymple has found that the most variable features of size and shape of

the skull of T. ferox are those which relate to feeding. Those

structures which provide sites of origin or passage for jaw musculature

increase allometrically with age, and the amount of relative increase is

highly variable. Furthermore, the development of features related to

feeding can occur independently of one another. This high degree of

variability in characters of the feeding apparatus indicates that they

are not useful systematic features, as had been suspected.

In this study quantitative characters of the jaws and associated

structures palatall groove, supraoccipital spine) are avoided. Treatment

of the skull concentrates on contacts between elements and between

elements and features of external morphology. Because complete

interspecific comparison is the goal of this study, data from sectioned

skulls (8 of 22 trionychid species available) will not be treated. This

is the first study of trionychid systematics for which at least one skull

of every currently recognized Recent species was available.

The skull characters and character states which are treated in this

section are summarized in Table 10. The details of distribution of the

states of characters important for resolving relationships within the

Trionychidae are given in Table 11. The states for characters important

for resolving interfamilial relationships are given in Table 12.

Character states which are autapomorphic for a living trionychid species

are listed in Table 13. Discussion of these characters is arranged by

region of the skull beginning anteriorly and proceeding posteriorly, with

the dorsal surface treated first.

Nasal Region

The premaxillae of cryptodires are usually paired elements that make

up the ventral edge of the apertura narium externum. Among trionychoids

this is true only for dermatemydids and kinosternids. In Carettochelys,

as well as all trionychids, these normally paired elements are fused to

one another (character S15, Tables 10, 12; Fig. 6 A, B, D). In

trionychids this fused premaxillary differs further from those of the

outgroups in being excluded from the apertura narium exturnum by the

maxillae which meet dorsally to it (character S16, Table 12; Fig. 6 A,


In three trionychids the premaxillary is either often absent (Chitra

indica, 4 of 10), or nearly always absent (Cycloderma frenatum, 4 of 5;

Pelochelys bibroni; 6 of 7) (character S59, Table 11). The loss of this

element is clearly derived.


Because nasals are absent in all trionychids, as they are in all

living cryptodires (Gaffney, 1979b), the prefrontals are the anteriormost

paired elements on the dorsal surface of the skull. Thus, the

prefrontals form the dorsal border of the apertura narium exturnum.

Laterally these elements contact the maxillae and border the anterior

portion of each orbit between the maxilla and frontal. In most

cryptodires the descending processes of the prefrontals contact the vomer

and palatines. There is significant variation among trionychids in these

contacts. There is also useful variation in the degree of emargination

of the prefrontals at the dorsal edge of the apertura narium externum and

in the degree of separation of the maxillae and frontals along the

anterior margin of the orbit.

Through reduction of the prefrontals, vomer and palatines,

contact between the prefrontals and palatal elements in trionychids is

greatly reduced, or lost. The prefrontal-palatine contact found in most

cryptodires is lost in all trionychids (Gaffney, 1979b) and this loss can

be considered a shared derived character for the family (character S9,

Table 12). Contact between the vomer and prefrontals is the common

condition among trionychids, as it is for all testudines. It is absent

only in Cycloderma aubryi, Cycloderma frenatum, Cyclanorbis senegalensis

and Chitra indica, and is clearly a derived condition (character S7,

Table 11).

With the exception of two very primitive forms, Proganochelys and

Kallokibotion, testudines have an unpaired apertura narium externum with

a nearly straight to somewhat anteriorly convex dorsal margin that is

usually formed by the prefrontals (Fig. 6 C, D). This is true for the

outgroups and for some living species of trionychids. The remaining


trionychids show some degree of emargination of the prefrontals and thus

alteration of this primitive shape of the external narial opening. With

one exception emargination occurs laterally and is either shallow or

quite deep (character S13A, Table 11; Fig. 6 A, B). Only in Cyclanorbis

elegans does emargination occur medially (character S13B, Table 13).

The condition in C. elegans is considered to occur independently from

that in other emarginate forms. Weak lateral emargination is considered

to be intermediate between the strongly emarginate and non-emarginate


It is the prefrontal that normally separates the maxilla from the

frontal at the anterior edge of the orbit in turtles. In a single

trionychid, Trionyx subplanus, the maxillae contact the frontals lateral

to the prefrontals in about one-half of the specimens examined. In the

others, these elements are quite close and their proximity can be

considered a unique feature of this species (character S49, Table 13).

Orbital Region

A frequently used character in trionychid systematics is the

relationship between the width of the postorbital bar and orbit diameter.

The postorbital bar varies in width among the species of this family from

two times wider than the orbit to one-sixth of orbit width. Variation in

the width of the postorbital bar relative to the width of the orbit is

not continuous but constitutes four separate sets of species.

The outgroups vary in width of postorbital bar between state two

(equal to or wider than orbit) and state three (one-half to one-third

width of orbit). Only Claudius, with a very narrow postorbital bar

(state 4), and Platysternon and the chelonioids, which lack temporal

emargination (state 1), show the extreme conditions. In the current

context it seems most appropriate to consider most divergent postorbital

bar widths to be derived relative to the combined intermediate groups.

Skull Emargination

The advanced cryptodires (Chelomacryptodira of Gaffney, 1984), the

Trionychoidea and Testudinoidea, have highly developed temporal

emargination. But these two superfamilies differ greatly in the degree

of cheek emargination that they exhibit.

As reviewed by Gaffney (1979b) there has always been a problem

identifying landmarks suitable for making comparisons between taxa. The

use of exposed elements seems to be most appropriate, but use of exposure

of the postorbital as an index of temporal emargination in trionychids is

problematical. All trionychids have very deep temporal emargination that

leaves the processus trochlearis oticum fully exposed, and the

communication of the fossa temporalis dorsalis with the fossa temporalis

ventralis is visible over a significant distance. With this degree of

temporal emargination the postorbital bone, which makes up a significant

portion of the portorbital bar, is usually exposed. This is true for all

outgroup trionychoids and testudinoids examined. The postorbital in

trionychids is one of several skull elements which has undergone extreme

reduction. This reduction is so extreme that contact between the jugal

and parietal occurs below the skull surface in all trionychids (character

S6, Table 12) and these two elements make up much of the postorbital bar.

In some trionychids jugal-parietal contact is so strong that it is

present on the skull surface and the postorbital is isolated from

temporal emargination. Isolation of the postorbital from the temporal

emargination might seem quite primitive and it certainly is if isolation


is via parietal-squamosal or parietal-squamosal-quadratojugal contact.

But isolation via jugal-parietal contact is a derived feature found only

among trionychids (character S5, Table 12).

Jugal-parietal contact on the skull surface can vary within a single

trionychid species. This variable condition is considered to be

intermediate between the primitive absence of jugal-parietal contact on

the skull surface and its presence which is certainly derived (character

S5, Table 11).

Lateral to the temporal emargination in trionychids is a very narrow

bar formed by the jugal and quadratojugal. The trionychids parallel the

condition seen in some emydids of extreme quadratojugal reduction. But

unlike the case in emydids this element is never lost. In all

trionychids the quadratojugal does not contact the maxilla and

postorbital but only the jugal. Posteriorly it sutures to the quadrate

and squamosal. In other living trionychoids the contact of the

quadratojugal to the postorbital is maintained and the quadrato-jugal

maxillary contact is maintained except in some Dermatemys (UF 29168; Fig.

172 in Gaffney, 1979b). Reduced contacts of the quadratojugal is

considered derived within the Trionychoidea (characters S1,S4, Table 12).

Because of the reduced size of the quadratojugal, the jugal and

squamosal lie quite close to one another in all trionychids. In six

species they are occasionally in contact. This is considered to be the

derived state for character S2 (Table 11).

Strong cheek emargination, which accompanies temporal emargination

in testudinoids, is not found among living trionychoids. Although cheek

emargination is visible in Dermatemys, Carettochelys and kinosternids, it

does not extend above an imaginary line extending horizontally from the


lower edge of the orbit (character S10, Table 12). In testudinoids, on

the contrary, cheek emargination is quite well developed and extends well

dorsal to such a line (except in Malayemys). In all testudinoids and

trionychoids except for the Trionychidae, cheek emargination is limited

anteriorly by the maxillary. In the Trionychidae, cheek emargination

occurs within the jugal when it is present (character S12, Table 12).

Because of flexure of the snout in trionychids, ventral emargination of

the jugal does reach above the lower rim of the orbit in a few cases.

But emargination occurs only within the jugal and is the site of origin

of the M. zygomatico-mandibularis (Dalrymple, 1977), a muscle which is

unique to trionychids. Therefore, it is likely that cheek emargination

in trionychids is not homologous to that of other turtles and that

restriction of true cheek emargination ventral to the lower rim of the

orbit can be considered a derived feature of the Trionychoidea (character

S10, Table 12).

Stapedial Foramen

The most significant differences between testudinoid and trionychoid

turtles is in the pattern of blood flow to the head (McDowell, 1961;

Albrecht, 1967; Gaffney, 1975, 1979b). This is reflected in variation of

the size of the stapedial foramen and in the morphology of the prootic

and parietal adjacent to this foramen. In testudinoids the majority of

anterior blood flow is via the stapedial artery. Therefore the foramen

stapedio-temporale is large and there is often a groove in the prootic

and parietal for the large stapedial artery. In trionychoids, the

stapedial artery is reduced because most of the anterior blood flow is

via the internal carotid artery. In this superfamily the foramen

stapediotemporale tends to be reduced or absent and rarely is there

evidence of a groove for the stapedial artery on the prootic or parietal

(character S43, Table 12). These feature are important at the family

level, there is little variation within the Trionychidae.

The Processus Trochlearis Oticum and the Quadrate

The processus trochlearis oticum is a distinctive feature of the

Cryptodira. It is over this structure that the majority of the jaw

adductor musculature lies. This is in contrast to the condition in

Pleurodira in which the lower jaw adductors operate over a process of the

pterygoid. In most cryptodires the majority of the processus is formed

by the quadrate.

In trionychids the processus trochlearis oticum can be quite

large and it always involves the quadrate, prootic and parietal (Table

14). In thirteen species the quadratojugal is included in at least some

individuals. Within the Trionychidae, three useful patterns of variation

are noted: the inclusion of the quadratojugal into the processus

trochlearis oticum, reduction in the contribution made by the quadrate,

and increase in the contribution made by the parietal. The first occurs

when the quadratojugal sends a medial process across the anterior edge of

the quadrate. It results in reduction of the quadrate contribution and

is absent from the processus in all outgroups. It is thus considered to

be derived within the Trionychidae (character M16, Table 11). In

trionychids unlike other cryptodires the quadrate makes up less than one-

third of this structure (character M17, Table 12).

There is additional variation among trionychids in the amount of

parietal contribution. In the majority the parietal contribution is

small, always less than one sixth of the total (Table 14). In the North


American forms, and also Cyclanorbis elegans, Trionyx euphraticus, T.

nigricans, T. swinhoei and T. triunguis the parietal contribution is

slightly larger, about one-fourth or more of the processus trochlearis

oticum (character M19, Table 11). The contribution of the parietal to

this structure in other cryptodires is quite limited or absent. Thus the

large contribution in trionychids is clearly derived.

In very few chelonians does the quadrate completely surround the

collumela (Gaffney, 1979b). This occurs in the Trionychidae,

Carettochelys, Chelydridae and Testudinidae (Tables 10, 12).

The Trigeminal Region

The trigeminal foramen lies lateral to the braincase and below the

processus trochlearis oticum of cryptodires. In trionychids it is a

large opening providing an exit for the maxillary and mandibular branches

of the trigeminal nerve as well as the mandibular artery (Gaffney,

1979b). In trionychids the parietal, prootic, quadrate, pterygoid and

epipterygoid may contact this foramen but there is significant inter- and

intraspecific variation in the degree and form of contact of each element

(Fig. 7).

An epipterygoid is present in all trionychid species but tends to

fuse to the pterygoid in larger individuals (Table 15). Fusion occurs

less frequently (perhaps later in life) in trionychines than in

cyclanorbines. Variation in the length of retention of a distinct

epipterygoid is treated via three states of character T7 (Tables 10, 11).

This element usually fuses to the pterygoid in older adults of most

cryptodires. Long-term retention of the epipterygoid is therefore

considered to be a derived and possibly paedomorphic feature.


Because the epipterygoid is an important landmark in describing

variation in the morphology of the trigeminal region of trionychids,

descriptions of this region are based on individuals in which this

element is not yet fused to the pterygoid. Complication of these

descriptions arises because the epipterygoid is a superficial element of

variable shape and size that can cover certain contacts in some

individuals of a given species but not in others. This results in the

ungainly appearance of the three states of character T2B (Table 10) in

which all states include the possibility of no pterygoid-trigeminal

contact (the case when the epipterygoid is large), but show different

forms of pterygoid-trigeminal contact if the epipterygoid is not

enlarged. When the pterygoid does contact the foramen nervi trigemini

the contact may occur posteriorly between the prootic and epipterygoid

(state 1), ventrally between the epipterygoid and quadrate (state 0),

anteriorly between the parietal and epipterygoid (state 2), or in no

individuals at all (character T2A, state 2). See Table 11 for

distribution of these character states.

Contact of the pterygoid to the foramen nervi trigemini between the

prootic and epipterygoid (state 1, character T2) ocurs in Trionyx

formosus, T. gangeticus, T. hurum, T. nigricans and Lissemys punctata and

results in the isolation of the quadrate from the foramen nervi trigemini

(Fig. 7 D, F). In both Cyclanorbis species and both Cycloderma species

the quadrate is also isolated from the foramen nervi trigemini. But in

this case it is the epipterygoid that meets the prootic posteriorly and

thus intervenes (character T4, state 2; Fig 7 E). When the epipterygoid

fuses to the pterygoid, the two groups mentioned above (those with state

1 of character T2 and those with state 2 of character T4) look identical

but they arrive at this condition via very different pathways.

Significant variation in this region among the outgroups makes

assigning polarities to characters of contact of the epipterygoid to the

foramen nervi trigemini difficult. Identification of polarity for other

contacts in the trigeminal region seems clear. In no other trionychoid

does the epipterygoid contact the prootic posteriorly (Fig. 7 A, B, D) as

it does in Cyclanorbis and Cycloderma (character T4, Table 11; Fig. 7 E)

or anteriorly as it does in some or all members of certain trionychine

species (character T3, Table 11; Fig. 7 H). Similarly, all trionychoid

outgroups have contact between the epipterygoid and palatine (Fig. 7 A,

B, C) and the absence of this contact in some or all members of a species

is considered derived (character TI, Tables 10, 11).

An important feature of the Trionychoidea (sensu Gaffney, 1979b) is

participation of the palatine in the formation of the lateral wall of the

braincase. This occurs in all trionychoids examined and can be seen just

anterior to the foramen nervi trigemeni (Fig. 7 A-H). In trionychoids

the pterygoid is excluded from the interorbital fenestra by the expanded

palatines. In testudinoids and in other turtles the pterygoid either

reaches the interorbital fenestra or is immediately adjacent to it

(character 14, Table 12).

The Occipital Region

There are numerous systematically useful characters visible on the

skull in posterior view. One of these is a reflection of the importance

of the internal carotid artery (Albrecht, 1967; McDowell, 1961; Gaffney,

1975, 1979b). The large diameter of the canalis carotici interni and the

straight path that it follows in trionychoids can be observed even in

articulated skulls. A stiff wire, slightly narrower than the canal, will

pass into the foramen posterius canalis carotid interni and out of the

foramen anterius canalis carotici interni with ease (character S31, Table

12). In large trionychids the latter opening is clearly visible through

the former. This is in contrast to the case in other cryptodires in

which this canal makes an S-shaped curve or a high angle bend (see Figs.

25-29 in Gaffney, 1979b). It seems likely that this straight, wide path

facilitates blood flow through the internal carotid in trionychoids.

The location of the foramen posterius canalis carotici interni in

the Trionychidae is also of some interest. In all species of this family

it is completely surrounded by the pterygoid. The same is true for

Carettochelys, but in kinosternids it can be open dorsally to the

fenestra postotica (Staurotypus) or be bordered dorsally by the prootic

(Kinosternon). In Dermatemys and in most testudinoids it is open

dorsally to the fenestra postotica (character S30, Table 12).

In some trionychids the foramen posterius canalis carotici interni

is quite ventrally located and is reminiscent of the condition in the

Paracryptodira. However, in all other trionychoids and other

Eucryptodira it is posteriorly located. Thus the presence of these

foramina on the ventral surface of the skull is considered derived.

Variation in the location of the foramen posterius canalis carotici

interni within the Trionychidae is best described in relation to a crest

of bone which is a lateral extension of the tuberculum basioccipitales.

In no member of this family is this foramen located above such a crest,

but in Pelochelys bibroni, Chitra indica, Trionyx cartilagineus and T.

nigricans (only one specimen available) it is found within the crest.


The latter condition is considered to be primitive relative to the

ventral position found in all other species (character S34, Table 11).

The foramen jugularis posterius is located lateral to the foramen

magnum in turtles and is visible in posterior view. In most cryptodires

it is surrounded by the exoccipital or exoccipital and opisthotic (Fig. 8

A, B, C). In some cheloniids, some trionychids, some Claudius and

Platysternon, this opening is continuous with the fenestra postotica

(Fig. 8 E, F). Isolation of the foramen jugularis posterius from the

fenestra postotica when present in the Trionychidae occurs in a unique

manner, that is by contact of the pterygoid to the opisthotic (Fig. 8, D,

F). In all cyclanorbines the pterygoid arches dorsally to meet the

opisthotic (Fig. 8 D). In the trionychines infrequent isolation occurs

via the descent of a narrow process of the opisthotic across an otherwise

open fenestra postotica (8 F). These two types of isolation of the

foramen jugulare posterius appear to be independent evolutionary events

(Loveridge and Williams, 1957) and are treated as such in the analysis of

intrafamilial relationship (characters S32A, S32B, Table 11).

In nearly all trionychids, as in most other trionychoids and in

chelydrids (including Platysternon) and chelonioids, there is contact

between the exoccipital and pterygoid. Only in Trionyx triunguis does

the basioccipital intervene between these elements, separating them as it

does in most testudinoids. In the current context this is a unique

feature most useful for the recognition of T. triunguis (character S57,

Table 13). Separation of the pterygoid from the exoccipital may be a

shared derived feature of the Testudinoidea.

The crista dorsalis occipitalis is a small tubercle on the dorsal

surface of the basioccipital found within the braincase. When present

this tubercle is visible (under correct lighting) through the foramen

magnum. Gaffney (1979b) reports that it is variably developed in most

turtles but that it is missing in Trionyx ferox. I find this structure

to be absent in all trionychoids and testudinids examined, but clearly

visible in cheloniids, dermochelyids, chelydrids, and emydids (except

Rhinoclemys pulcherimma). This is therefore a useful character at the

interfamilial level (character S29, Table 12).


The most striking differences between the palates of trionychids and

those of other cryptodires is the presence of a median foramen anterior

to the apertura narium internum and the presence of unconstricted

pterygoids. This midline opening is usually of large size and is called

the foramen intermaxillaris. It varies in size in the Trionychidae (see

below, character M4) but it always separates the vomer from the

premaxilla. The same structure appears to be present in Carettochelys

where it is continuous with the apertura narium internum. In

Carettochelys the vomer and maxillae do not meet anterior to the apertura

narium internum and the posterior limits of the foramen intermaxillaris

remain undefined.

A structure that appears to be homologous to the foramen

intermaxillaris is present in mature individuals of all three living

staurotypine kinosternids and in Xenochelys formosus (Oligocene of South

Dakota, Williams, 1952). The deep pit in the premaxillae, which

accommodates the symphyseal projection of the lower jaw in all

kinosternids, opens dorsally in large individuals of both species of

Staurotypus, in Claudius and in Xenochelys. This does not occur in large

individuals of any other living forms with strongly hooked lower jaws


such as chelydrids (including Platysternon). This trait, which appears

late in life in staurotypines, may have come to arise earlier in

carettochelyids, which also have sharply hooked lower jaws. Once

present, this novelty apparently remained in trionychids in spite of the

fact that they have unhooked lower jaws (character S19, Table 12).

Variation in the size of the intermaxillary foramen among

trionychids has been utilized by several authors (Loveridge and Williams,

1957; De Broin, 1977). Comparison of the length of the intermaxillary

foramen relative to the total skull length is not satisfactory; the

distribution of this character for trionychids is quite continuous (Fig.

9). It should be noted, however, that all of the members of a proposed

monophyletic group (Meylan, 1985), the North American forms plus Trionyx

swinhoei, T. euphraticus and T. triunguis have the highest values for the

ratio of intermaxillary foramen length to total skull length.

This character can be utilized if examined in terms of its size

relative to the primary palate. Variation in the ratio of length of the

intermaxillary foramen to length primary palate among trionychids falls

into five distinct groups (Fig. 9, Table 10). Identification of a

character polarity for the states of this character is difficult. The

intermaxillary foramen in other trionychoids is highly specialized in one

case (Carettochelys) and incompletely developed in the other

(Staurotypinae). It appears prudent to assume that the medium size

classes together approximate the primitive state and that the most

divergent conditions (states 0 and 2) are derived within the Trionychidae

(character M4, Tables 11, 13).

The vomer is one of several elements which is reduced in the

Trionychidae. In most turtles it lies between the paired maxillae and


palatines. Anteriorly it reaches the premaxillae and posteriorly it

often contacts the paired pterygoids. When an intermaxillary foramen is

present, premaxillary contact is prevented. In some trionychids the

vomer divides the maxillae completely and reaches the intermaxillary

foramen between them. This most closely approximates the condition in

the outgroups in which no intermaxillary foramen is present and is

therefore considered to be the primitive condition for the relationship

of the vomer to the maxillae (character S20, Table 11).

In other trionychids the maxillae meet on the midline of the palate

ventral to the vomer. Depending on the degree of reduction of the vomer

and the length of this intermaxillary suture, the vomer may still enter

the intermaxillary foramen by reaching it dorsally over the united

maxillae. Reduction of the vomer to the extent that it does not reach

anteriorly to the intermaxillary foramen is interpreted as the derived

state of character 21 (Table 11).

Posteriorly, the vomer of most cryptodires reaches between the

paired palatines as far as the pterygoids. This is true of Dermatemys

and kinosternids but not Carettochelys or any trionychids. The failure

of the vomer to reach as far posterior as the palatines is considered a

derived condition (character S22, Table 12).

In most chelonians the vomer is the only unpaired midline element

reaching the transverse pterygoid-palatine suture. In Carettochelys and

all trionychids (except T. euphraticus), only an enlarged basisphenoid

does so. This unique contact of palatal elements has been recognized as

evidence of unique common ancestry of these two taxa (Baur 1891b; Meylan,

1985). It is treated as such here (character S18, Table 12). The


absence of contact of palatines and basisphenoid in most specimens of T.

euphraticus appears to be a unique reversal (character S18, Table 13).

The vomer of turtles does not normally contact the basisphenoid, but

with the anterior elongation of the latter in trionychids comes a greater

possibility that such contact might occur. Siebenrock (1897) reports

vomer-basisphenoid contact on the dorsal surface of the palate in

Pelochelys. In two of the seven Pelochelys skulls examined during this

study (USNM 231523 and NMW 1857) contact between these elements is

present on the palate. This condition is unique to Pelochelys among the

Trionychidae (character S23, Table 13). In Cycloderma frenatum the vomer

is absent. This is a unique condition among trionychids (character S60,

Table 13).

At or near the palatine-pterygoid suture in all chelonians is

located a pair of ventral openings in the palate, the foramena palatinum

posterius. These openings are never large in trionychids. They may be

entire, divided into two openings, or divided into numerous small

openings not larger than the nutritive foramina of the palate. Small

size of these openings may be shared by all trionychoids as well as some

testudinoids, but the variation in the division of this opening is useful

within the Trionychidae, and in particular among the Cyclanorbinae

(character S26, Tables 11). Division of the foramen palatinum posterius

is considered derived.

The contacts of the foramina palatinum posterius also vary among the

living species of the Trionychidae. In most forms, as in most

chelonians, these foramina contact the palatine and the pterygoid and/or

maxillary. In a limited number of trionychids this opening is restricted

to the palatine, which is considered to be a derived condition (character

S27, Table 11).

The processus pterygoideus externus of cryptodires usually takes the

form of a moderate to short posterior or posteriolateral projection from

the anteriolateral edge of the pterygoid just anterior to some degree of

medial constriction. They are found in nearly all cryptodires and they

vary considerably in degree of development. In trionychids there is no

medial constriction of the pterygoids and no free projection of this

process. In Carettochelys, the pterygoids are only slightly constricted

and the processus pterygoideus externus projects very weakly or not at

all. In other trionychoids these processes may be present (Kinosternon,

Staurotypus, some Dermatemys, some Claudius) or absent (some Claudius,

some Dermatemys, Xenochelys), but they are never as large and posteriorly

projecting as in the Chelydridae or some of the Emydidae. Reduction of

this projecting quality could be a shared derived feature of the

Trionychoidea. It occurs elsewhere among the Cheloniidae (Chelonia),

Emydidae (Malayemys) and Testudinidae (several genera). The absence of a

projecting processus is certainly derived for the Trionychidae and

possibly for the Trionychidae plus Carettochelys (character S25, Table


The elongate basisphenoid of trionychids varies in shape. In most

species, as in the outgroups, it has a subtriangular shape although

somewhat more elongate. In a few forms medial constriction of the

basisphenoid occurs either occasionally or frequently. The presence of

an hour-glass shaped basisphenoid is considered derived within the

Trionychidae (character S58, Table 11).

e x


C El
a) ) -C 0

4-4 0-4 14
S0 *00

Uo Q)

"in 0 4 C

ed w
- 4Io C

c 0 Qc
0 0 00

M ILf,
0 r- 0-

4C 4J

,n4 a) e C

-4 "t C C "C

0 a 0 0
0 C e 0 w

v *-0 0C
4d< C- C

-t ZC E4a
w c o C '4 CC

0 0 a

C 0 4 NO

*HCO 4 a P
. pa mI oH
44CCu a) r-
-io *i- 0
PL, 0 t *>< >

a s ~ >- i
b0 0) ) fl

Figure 7. The trigeminal region of eight trionychoid turtles showing
contacts of the skull elements around the foramen nervi trigemini and
participation by the palatine in the lateral wall of the brain case. The
foramen interorbitale is crosshatched, the foramen nervi trigemini is
stippled. Abbreviations: e, epipterygoid; pal, palatine; par,
parietal; pr, prootic; pt, pterygoid; q, quadrate. A, Dermatemys mawii,
BMNH 1911.1.28.1; B, Staurotypus salvinii, BMNH 1879.1.7.5; C,
Carettochelys insculpta, BMNH 1903.4.10.1; D, Lissemys punctata, UF
56017; E, Cyclanorbis elegans, BMNH 1954.1.14.3; F, Trionyx hurum, BMNH; G, Trionyx triunguis, BMNH; H, Chitra indica, ISNB

Figure 8. Diagramatic representations of contact between the foramen
jugulare posterius (stippled) and the fenestra postotica (crosshatched) in
six trionychoid turtles. Abbreviations: ba, basioccipital; ex,
exoccipital; op, opisthotic; pt, pterygoid. A, Staurotypus triporcatus,
UF 13482; 8, Dermatemys mawii, UF 29168; C, Carettochelys insculpta, UF
43888; D, Cycloderma frenatum, NMZB 1245, E, Pelochelys bibroni, USNM
231523; F, Trionyx subplanus, UF 56317.

0 *

o -e

uu a
oe w
0 *0 L
*1 C- 0

C 0
0 (1) CO c

CO *) 0

0 c

C bO
C 0 Q- C

0O 0

C 4C -
CO1 0 C4
C Cm

C -4J

. H m c
. C> I C

3e CO
CO m
o0) CO 4

C)0 0i 0)

CO m C -

*1 0)

CO C tl 41

C0 0 )0

0- *'4 C C
CO CO -l -rI

* C CO -.4



snOIlnw -
>- snlI.VuIHdn3 -
<{: xoud-
snlUdINIdS -




z SnOI.L3ONvo -
W VIVl.ONrld-
r.WVN3Uz -

>, IIHH113-1 iuanv.


Z wnunH-
- SNV0313-




0 co



o 0>




N 0

- sno3nI.
- snollVhIHdn3

130HNIMS -
sinoNnijI. IW
-GsnNlVdns D






fnunfH I
-IA uanv

- IflJVN3U,' U


- VI30NI



Table 10. Systematic characters and character states of the trionychid
skull. All skull characters are preceded by the letter S except those of
the trigeminal region (T) and those which are based on measurements (M).

Characters Character States

Si) quadratojugal contacts maxillary 1) yes
2) occasionally
3) no

S2) jugal contacts squamosal 1) no
2) in one-half of sample

54) quadratojugal contacts 1) yes
postorbital 2) no

S5) jugal contacts parietal on 1) no
skull surface 2) in one-half of sample
3) yes

S6) jugal contacts parietal within 1) no
fossa temporalis 2) yes

S7) vomer contacts prefrontal 1) yes
2) no

S8) incisura collumella auris closed 1) no
2) yes

S9) palatines contact prefrontals I) yes
lateral to vomer 2) no

S10) cheek emargination extends 1) yes
above lower edge of orbit 2) no

S12) anterior limit of cheek 1) maxilla
emargination formed by 2) jugal

Sl3a) dorsal edge of apertura narium 1) no
externum laterally emarginate 2) weakly
3) strongly

S13b) dorsal edge of apertura narium 1) no
externum medially emarginate 2) yes

S14) palatine forms a significant 1) no
part of the lateral wall of 2) yes
the brain case

S15) premaxillae fused into single

1) no
2) yes

Table 10--continued.

Character States

S16) premaxillae enter apertura
narium exturnum

S18) basisphenoid contacts palatines

S19) foramen intermaxillaris

S20) vomer divides maxillae

S21) vomer reaches intermaxillary

S22) vomer contacts pterygoid

S23) vomer contacts basisphenoid

525) processus pterygoideus exter-
nus projects from pterygoid

S26) size of foramen palatinum

S27) foramen palatinum posterius
forms in

S29) crista dorsalis basiocci-
pitalis present

S30) foramen posterius canalis
carotici interni completely
within pterygoid

S31) canalis carotici interni
straight and wide

S32a) foramen jugulare posterius
excluded from fenestra postoti-
ca by pterygoid arching to
contact opisthotic

1) yes
2) no

1) no
2) yes

1) absent
2) present

1) yes
2) no

1) yes
2) no


1) no
2) occasionally

1) yes
2) no

small and divided
many small openings

1) palatine and pterygoid
and/or maxilla
2) palatine only

1) yes
2) no

1) no
2) yes


Table 10--continued.


Character States

S32b) foramen jugulare posterius ex-
cluded from fenestra postotica
by descending process of opis-
thotic which reaches pterygoid

S34) foramen posterius canalis caroti-
ci interni relative to lateral
crest of basioccipital tubercle

1) no
2) yes

in it

S43) groove for some portion of sta- 1) yes
pedial artery visible on prootic 2) no
or descending process of parietal

S49) maxilla contacts frontal in front 1) no
of orbit 2) yes

S57) exoccipital contacts pterygoid

S58) basisphenoid shape

S59) premaxilla absent

S60) vomer lost

Tl) epipterygoid, if present,
contacts the palatine

T2A) contact between pterygoid and
foramen nervi trigemini occurs
when epipterygoid is present

T2B) when epipterygoid is present
pterygoid contacts foramen
nervi trigemini

T3) epipterygoid contacts prootic
anterior to foramen nervi

1) no
2) yes

1) not medially constricted
2) occasionally medially
3) medially constricted

1) no
2) occasionally
3) usually

1) no
2) yes

in ca. 50%

1) yes
2) no

between epipterygoid and
quadrate or not at all
between prootic and epiptery-
goid or not at all
between epipterygoid and
parietal or not at all

in ca. 50%


Table 10--continued.


Character States

T4) epipterygoid
posterior to

contacts prootic
foramen nervi

T7) epipterygoid fuses to pterygoid

M4) average ratio of intermaxillary
foramen length to length primary

M8) postorbital bar relative to
orbit diameter

M16) quadratojugal participates in
processus trochlearis oticum

M17) quadrate make up of the
processus trochlearis oticum

M19) proportion of processus
trochlearis oticum made up by

1) in subadults
2) in adults only
3) never

0) 0.07
1) about 0.20 to 0.40
2) about 0.60

0) about 2 times orbit
1) about equal to orbit
to 1/3 of orbit
2) less than 1/5 orbit

1) no
2) yes

1) greater than 50%
2) 33 to 50%
3) less than 33%

1) 15.6% or less
2) 22.1% or more

1) no
2) yes

Table 11. Character states for characters of the trionychid skull that
have been found to be useful in assessing intrafamilial relationships.
Numbers refer to character states outlined in Table 10. Periods indicate
missing values.


S2 S5 S7 S13A S20 S21 S26 S27 S32A S32B S34
























2 2 2 2

1 1 1 1

1 2 2 2

1 1 2 2

1 2 1 1

1 3 1 1

1 2 2 2

2 1 2 2

1 2 2 2

1 3 2 2

2 1 1 1

1 2 2 2

1 3 1 2

1 2 2 2

1 1 2 1

2 2 1 1

1 3 2 2

1 3 1 1

1 3 2 2

1 3 2 2

1 2 1 1

1 3 2 2

4 2

2 1

2 1

3 2

2 1

2 1

2 1

4 2

2 1

2 1

2 1

2 1

2 1

2 1

2 2

3 2

2 2

2 1

2 1

2 1

2 2

2 1

2 1 3

1 1 2

1 1 2

2 1 3

1 1 3

1 1 3

1 1 3

2 1 3

1 1 3

1 1 3

1 1 2

1 1 3

1 1 3

1 1 2

2 1 3

2 1 3

1 2 3

1 1 3

1 2 3

1 2 3

1 1 3

1 1 3

Table 11--continued.


S58 S59 Tl T2A T2B T3 T4

T7 M4 M8 M16 M19

aubryi 1 1

bibroni 1 3

cartilagineus 2 1

elegans 1 1

euphraticus 2 1

ferox 1 1

formosus 3 1

frenatum 1 2

gangeticus 3 1

hurum 3 1

indica 1 2

leithii 3 1

muticus 1 1

nigricans 3 1

punctata 1 1

senegalensis 1 1

sinensis 3 1

spiniferus 2 1

steindachneri 3 1

subplanus 2 1

swinhoei 1 1

triunguis 1 I

2 1 2

1 0 2 1

1 0 2 1

2 1 2

1 0 1 1

1 0 1 1

1 1 1 1

2 1 2

1 1 1 1

1 1 2 1

1 0 3 1

1 0 1 1

2 I I

1 1 1 1

1 1 1 1

2 1 2

1 2 1 1

1 1 1

2 3 1

1 1 1

1 0 1 1

1 0 1 1


1 1 1

2 1 1

3 1 1

2 1 1

3 2 1

3 2 1

3 1 1

2 1 1

2 1 1

3 1 1

3 0 0

3 1 1

2 2 2

3 1 1

3 1 1

3 2 2

3 1 2

2 1 2

3 2 1

2 1 1

Table 12. States of skull characters important in interfamilial
analyses. Numbers represent the character states listed in Table 10.


Taxa S1 S4 S6 S8 S9

S10 S12 S14 S15 S16 S18













3 2 2 2 2

1 1 1 2 1

1 1 1 1 1

3 1 1 1 1

1 1 1 1 1

2 1 1 1 1

3 1 1 1 1

3 1 1 2 1

1 1 1 2 1

2 1 1 1 1

2 1 1 2 1

3 1 1 1 1

2 2 2 2

2 1 2 2

2 1 2 1

2 1 2 1

2 1 2 1

2 1 2 1

2 2 1 1

1 1 1 1

2 1 1 1

1 1 1 1

1 1 1 1

1 3 1 1

2 2

1 2

1 I

1 1

1 1

1 1
1 1

1 1

1 I

1 1

1 1


Table 12--continued.


Taxa S19 S22 S25 S29

S30 S31 S43 M17













2 2

2 2

1 1

1 I

I 1

1 1

1 1 2

1 1 2

1 3 1

2 2 2 2 3

2 2 2 2 2

2 1 2 2 2

2 1 1 2

Table 13. Autapomorphic skull features of trionychid turtles.


Autapomorphic State

S13B Cyclanorbis elegans

S18 Trionyx euphraticus

S23 Pelochelys bibroni

S49 Trionyx subplanus

S57 Trionyx triunguis

S60 Cycloderma frenatum

S61 Cycloderma aubryi

M4 Chitra indica

M8 Trionyx subplanus

apertura narium externum
medially emarginate

basisphenoid fails to contact

vomer contacts basisphenoid

maxillae contact frontals in

pterygoid isolated from
exoccipital by basioccipital

vomer is absent

jugal excluded from orbit

intermaxillary foramen quite

postorbital bar one-ninth of
orbit diameter


Table 14. Average contribution of quadratojugal, quadrate, prootic and
parietal to the processus trochlearis oticum of Recent trionychid

Species N Quadrato- Quadrate Prootic Parietal

aubryi 8 0.000 0.227 0.607 0.166

bibroni 7 0.000 0.207 0.602 0.117

cartilagineus 7 0.007 0.239 0.655 0.139

elegans 5 0.011 0.236 0.565 0.221

euphraticus 9 0.020 0.166 0.557 0.266

ferox 11 0.032 0.290 0.396 0.260

formosus 4 0.000 0.294 0.635 0.071

frenatum 4 0.026 0.130 0.734 0.136

gangeticus 7 0.007 0.137 0.720 0.144

hurum 6 0.000 0.213 0.744 0.054

indica 8 0.000 0.312 0.626 0.062

leithit 3 0.000 0.249 0.684 0.091

muticus 5 0.027 0.072 0.581 0.320

nigricans 1 0.000 0.200 0.500 0.292

punctata 6 0.000 0.192 0.717 0.094

senegalensis 6 0.000 0.177 0.671 0.152

sinensis 9 0.005 0.154 0.728 0.122

spiniferus 8 0.019 0.262 0.527 0.225

steindachneri 1 0.000 0.180 0.819 0.000

subplanus 6 0.088 0.180 0.625 0.112

swinhoei 1 0.033 0.100 0.500 0.300

triunguis 10 0.004 0.189 0.584 0.223

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs