THE VERTEBRATE FAUNA OF THE
SELMA FORMATION OF ALABAMA

PART IV
THE TURTLES OF THE FAMILY TOXOCHELYIDAE

RAINER ZANGERL

CURATOR OF FOSSIL REPTILES

FIELDIANA: GEOLOGY MEMOIRS

VOLUME 3, NUMBER 4

Published by

CHICAGO NATURAL HISTORY MUSEUM

MAY 12, 1953

PRINTED IN THE UNITED STATES OF AMERICA
BY CHICAGO NATURAL HISTORY MUSEUM PRESS

PREFACE

Tui'tles belonging to the family Toxochelyidae constitute by far the largest
part of the material collected in the Mooreville Chalk of the Selma Formation.
Not only is the number of specimens considerable, but the materials belong to a
relatively large number of genera and species. This is true not only in the
Mooreville Chalk, but likewise in the Niobrara Chalk of Kansas. The Toxo-
chelyidae, rather than the Protostegidae, formed the most conspicuous elements
of the turtle fauna.

The study of toxochelyid turtles is beset with a gi'eat many difficulties,
paramount among which are the relatively poor state of preservation of most
specimens and the fact that entire shells are very rare finds. Most specimens
consist of only a very small portion of the skeleton and, more often than not,
the remains are badly crushed and often distorted as well as flattened.

The question exists whether materials of this kind furnish enough evidence
to permit significant conclusions as to the morphology of the original animals.
The answer is, I believe, favorable in this case, because of the relatively large
number of available specimens and because of the fact that the turtle skeleton
generally conforms to a very rigid, basic pattern of organization. In a single
family, such as the Toxochelyidae, this basic pattern of organization to which
all of the genera conform is so stable that unknown parts can be predicted to a
considerable extent on the basis of the known portions of the skeleton. This is
especially true in the case of the shell, where the elements show the same funda-
mental relationships to one another in all species of the family. It is, for instance,
not necessary to have all twenty-two peripheral plates in a given specimen in
order to be certain that normal individuals of the species possessed that many;
if only one mid-dorsal neural plate is preserved, and this is keeled, it is entirely
safe to predict a carapace with a sagittal carina. Fragmentary specimens, if
available in sufficiently large numbers, permit essentially accurate reconstruction
of the shell.

The major difficulty in the study of the toxochelyid turtles is not so much
the fragmentary state of the remains as the varying degree to which the materials
are deformed. There is evidence among the materials from the Mooreville
Chalk of Alabama that compression may have reduced shell plates to as little
as 20 per cent of their original thickness. Crushing results in gi'eater or lesser
alterations in the shape of the bones, and these alterations present a major
obstacle in the comparison of specimens. Familiarity with the materials can
partially overcome this difficulty, but it is often impossible to distinguish indi-
vidual variations in form from individual differences in preservation.

137

138 PREFACE

In view of the difficulties mentioned, the systematic treatment of the mate-
rial must be conservative. It is possible that fewer species are here recognized
than actually existed.

It was evident from the beginning that the toxochelyid materials from
Alabama could not be satisfactorily studied by themselves. A review of all of
the materials of toxochelyid turtles was inevitable and to this end I visited all
major collections in the United States that preserve specimens of this group.

I wish to express my sincere gratitude to the authorities of the American Mu-
seum of Natural History, the United States National Museum, the Peabody
Museiim of Yale University, the Museum of Natural History of the University
of Kansas, the Academy of Natural Sciences of Philadelphia, the Natural History
Museum at South Dakota School of Mines, and the University of Tennessee for
the privilege of studying their materials. I feel greatly indebted to my colleagues
in charge of these collections for their kind help during my visits and for the loan
of much of the important Niobrara material. I was thus enabled to assemble for
direct comparison most of the toxochelyid materials known at present.

My very best thanks are due, furthermore, to Mr. C. M. Barber, of Flint,
Michigan, who collected much of the Alabama material, and to Mr. William D.
Turnbull, of Chicago Natural History Museum, who collected some of the speci-
mens and took a great interest in the patient and skillful preparation of the entire
Mooreville Chalk collection.

The drawings were made by the late John Conrad Hansen (figs. 61, 91,
100, 112 in part), Mr. Douglas E. Tibbitts (figs. 105, 117), Miss Maidi Wiebe
(figs. 63, 70, 90, 122-124, pi. 20), and myself. The photogi-aphs were made by
Mr. John Bayalis (pis. 10, c, 11-13, 17-19, 21-25) and myself. A number of
photographs (pi. 15, figs. 3, 4, pi. 16, figs. 2-12) were kindly given to me by
the American Museum of Natural History.

Rainer Zangerl

CONTENTS

PAGE

List of Illustrations 141

Introduction 145

General Morphology of the Toxochelyidae 147

Skull 148

Vertebral Column 153

Carapace and Plastron 158

Girdles and Limbs 160

Revision of the Family Toxochelyidae 173

Family Toxochelyidae 173

Subfamily Toxochelyinae 173

Genus Toxochelys 174

Toxochelys latiremis 178

Toxochelys moorevillensis sp. nov 186

Toxochelys barberi 193

Toxochelys weeksi 194

Toxochelys atlantica sp. nov 196

Toxochelys browni 197

Toxochehjs sp 198

Genus Thinochelys gen. nov 199

Thinochelys lapisossea sp. nov 200

Genus Porthochelys 202

Porthochelys laticeps 204

Subfamily Osteopyginae 205

Genus Osteopygis 206

Osteopygis emarginatus 208

Incertae sedis 215

Rhetechelys platyops 215

Subfamily Lophochelyinae nov 216

Genus Lophochelys gen. nov 217

Lophochelys natatrix sp. nov 218

Lophochehjs niobrarae sp. nov 220

Lophochelys venatrix sp. nov 224

139

140 CONTENTS

PAGE

Genus Ctenochelys gen. nov 227

Ctenochelys tenuitesta sp. nov 230

Ctenochelys stenopora 237

Ctenochelys acris sp. nov 242

Ctenochelys procax 247

Genus Prionochelys gen. nov 248

Prionochebjs naiita sp. nov 249

Prionochelys matutina sp. nov 254

Prionochelys galeotergum sp. nov 258

Forms Questionably Referred to the Toxochelyidae 260

Cijnocercus incisivus 260

Toxochelys gigantea 260

Sineniys lens 261

The Paleoecology of the Toxochelyidae 262

Horizontal and Vertical Distribution 262

Habitat 263

Food 264

Locomotion 264

Parasitic Infestation 264

Burial Conditions 265

The Relationships of the Toxochelyidae to Other Families .... 265

Comparison of the Thalassemyid with the Toxochelyid Pattern . . . 266

Comparison of the Cheloniid with the Toxochelyid Pattern 267

Morphological and Phylogenetic Relationships Among

THE Toxochelyidae 269

Summary and Conclusions 272

References 275

Index 277

LIST OF ILLUSTRATIONS

PLATES

9. Comparison of braincases of Chelonia nujdas, Toxochelys moorevUlensis, and
CJielydra serpentina in dorsal view.

10. Skulls of Toxochelys latiremis and T. browni.

11. Dorsal and ventral views of skull of Toxochelys latiremis.

12. Dorsal and ventral views of skull and mandible of Toxochelys latiremis.

13. Dorsal and ventral views of skulls and mandibles of Toxochelys latiremis.

14. Skulls and mandibles of Toxochelys moorevillensis.

15. Dorsal and medial views of lower jaw tentatively referred to Toxochelys

atlantica; dorsal view of type specimen of Toxochelys latiremis; jaw
fragment of Osteopygis emarginatiis.

16. Dorsal views of mandibles of Osteopygis emarginatus; fragments of skull and

mandible of Toxochelys latiremis (type specimen of Toxochelys serrifer) .

17. Ctenochelys stenopora (type specimen of Toxochelys elkader), skull and man-

dible in dorsal and ventral views.

18. Ctenochelys acris, skull fragments and mandible in dorsal and ventral views.

19. Ctenochelys procax, skull and mandible in dorsal and ventral views.

20. Vertebrae of to.xochelyid turtles.

21. Juvenile specimen of Toxochelys latiremis.

22. Carapace and plastron of Toxochelys moorevUlensis.

23. Carapace of Toxochelys moorevUlensis.

24. Carapace and plastron of Toxochelys barberi (type specimen).

25. Carapace of Thinochelys lapisossea.

26. Carapace and plastron of Osteopygis emarginatus.

27. Prionochelys nauta, neural carina, costal plates, and portions of peripheral

plates.

28. Parts of carapace and plastron of Prionochelys nauta.

29. Peripheral plates of Prionochelys nauta.

141

142 LIST OF ILLUSTRATIONS

TEXT FIGURES

PAGE

60. Reconstruction of skull of Toxochelys latiremis 149

61. Palatal views of skulls and dorsal views of mandibles of Toxochelys,

Ctenochelys, and Osteopygis 151

62. Diagram showing relative size relationships in cervical vertebrae of

Chelydra serpentina, Toxochelys latiremis, and Chelonia mydas . 156

63. Comparison of centra of first and second shell vertebrae with their ribs

in Chelydra serpentina, Ctenochelys teniiitesta, and Eretmochelys
imbricata 159

64. Proportional differences in shoulder girdles of chelydrid, toxochelyid

and cheloniid turtles 159

65. Comparison of pelves of Chelydra, Toxochelys, and Eretmochelys . . . 163

66. Indices of pubic-ischial surface areas in Chelydra, Toxochelys moore-

villensis, and Eretmochelys 165

67. Limb proportions in Chelydra, Toxochelys, and cheloniid turtles . . . 165

68. Comparison of angle a in humeri of Chelydra, Lepidochelys, and a

toxochelyid 165

69. Comparison of angle /5 in humeri of Chelydra, Lepidochelys, and a

toxochelyid 165

70. Forelimb of Toxochelys latiremis and hind limb of Toxochelys moore-

villensis; comparison of right femora of Chelydra, Toxochelys, and
Lepidochelys 167

71. Shell, shoulder girdle, and humerus, illustrating functional significance

of angles a and (3 171

72. Outline drawing of important parts of Toxochelys latiremis, C.N.H.M.

PR123 183

73. Reconstruction of carapace and plastron of juvenile specimen of

Toxochelys latiremis 185

74. Carapace of Toxochelys latiremis 187

75. Carapace of Toxochelys moorevillensis 189

76. Plastra of Toxochelys ynoorevillensis 189

77. Carapace and plastron of Toxochelys barberi 192

78. Plastron of Toxochelys weeksi 195

79. Carapace of Toxochelys atlantica 197

80. Side view of snout region of skull of Toxochelys browni 197

81. Toxochelys sp. from Taylor Marl of Texas 199

82. Carapace and plastron of Thinochelys lapisossea 201

LIST OF ILLUSTRATIONS 143

PAGE

83. Skull and mandible of Porthochelys laticeps 203

84. Carapace and plastron of Porthochelys laticeps 203

85. Internal anatomy of mandible of Osteopygis emarginatus 207

86-88. Osteopygis emarginatus (annotated sketches made from origi-
nals) 209-211

89. Carapace and plastron of Osteopygis emarginatus 213

90. Osteopygis emarginatus, humerus and left ulna 215

91. Carapace, plastron, girdle, and limb-bones of Lophochelys natatrix

(type specimen) 219

92. Reconstruction of carapace and plastron of Lophochelys natatrix . . . 221

93. Peripherals and neural of Lop/joc/ie??/s watoinx 222

94. Carapace and ?plastron of Lophochelys niobrarae 223

95. Type specimen of Lophochelys venatrix 224

96. Specimens of Lophochelys venatrix 225

97. Reconstruction of Lophochelys venatrix 226

98. Comparison of proportions of lower jaw in four species of Ctenochelys . 227

99. Proportional differences in snout region of skull in Ctenochelys acris

and C. tenuitesta 227

100. Fragment of braincase of Ctenochelys tenuitesta 231

101. Type specimen of Ctenochelys tenuitesta and plastron of another speci-

men 282

102, 103. Referred specimens of Ctenochelys tenuitesta 233, 235

104, 105. Reconstruction of carapace of Ctenochelys tenuitesta .... 236, 237

106. Girdle and limb elements of Ctenochelys tenuitesta 237

107. Plastron of Ctenochelys stenopora 239

108. Carapace (fragments) and plastron of Ctenochelys stenopora (juvenile) . 240
109, 110. Tracings of parts of slab specimen of Ctenochelys stenopora . 241, 242

111. Ctenochelys stenopora. Most significant parts of skeleton associated

with skull (Toxochelys elkader) 243

112. Ctenochelys acris. Type specimen and referred materials 244

113. Ctenochelys acris. Reconstruction of carapace from all available

materials 245

114. Comparison of morphology of keel in Prionochelys nauta, P. matutina,

and P. galeotergum 250

115. Prionochelys nauta, skull fragments and portions of anterior carapace . 251
116, 117. Prionochelys nauta, reconstruction of carapace 252, 253

144 LIST OF ILLUSTRATIONS

PAGE

118. Prionochelys matutina (type specimen) 255

119. Comparison of relative sizes of third neural plates and xiphiplastra in

Prionochelys nauta and P. matutina 257

120. Mode of ossification (growth) in pygal and eleventh peripheral plates

of Prionochelys matutina as suggested by two different ontogenetic
stages 257

121. Prionochelys galeotergum (.type specimen) 259

122. Neural fragment of Ctenochelys tenuitesta in lateral view, showing

severe parasitic lesions 259

123. Comparison of the plastral differentiation in Toxochelyidae and

Cheloniidae 268

124. Diagrammatic representation of state of specialization in seven genera

of toxochelyid turtles and morphological relationship of these
forms to morphotypic condition of family 271

Turtles of the Family Toxochelyidae

INTRODUCTION

In his revision of the fossil turtles of North America, Hay (1908) included
in the family Toxochelyidae only the genera Toxochelys Cope, Porthochelys
Williston, and the little-known Cyuocercus Cope. Several other forms were
placed among the Thalassemyidae, namely, Osteopygis Cope, Catapleura Cope,
Lytoloma Cope, Erquelinnesia Dollo, and Rhetechelys Hay.

Hay's reasons for placing the latter genera in the Old World family Thalasse-
myidae are not known; he characterized the family after having included the
above genera in it. Apparently, morphological criteria were not the reason.
Thei'e is a far gi'eater degi'ee of similarity between the skulls and mandibles of
the genera in question and those of the cheloniid turtles than there is between
any of these foiTns and the thalassemyids. Fm'thermore, there is a striking
similarity between the plastra of Osteopygis and Toxochelys or Porthochelys. Thus,
the morphological affinities of the four genera in question are with the Cheloniidae
and the Toxochelyidae, not with the Thalassemyidae. As will be shown below,
the basic organization of the Thalassemyidae is very different from that of any
of the genera listed above.

I have come to the conclusion that there is a convergence in the develop-
ment of secondary palates and wide mandibular triturating surfaces in the genera
Osteopygis and Rhetechelys on the one hand and the cheloniid turtles on the other
and that the forms mentioned belong to the family Toxochelyidae.

The materials referred to the genus Lytoloma from the Greensand deposits
of New Jersey have long been a source of confusion. The type material on which
the type species, L. angusta Cope, is based consists of some peripheral plates
(A.M.N.H. 1133), one of which was figured by Cope,'^ as has been elaborately
demonstrated by Hay (1908, p. 155, fig. 192). A mandible bearing the same cat-
alogue number (A.M.N.H. 1133) was evidently not associated with the periph-
eral bone and was not used by Cope to establish either the genus or the species.
The mandible bears a label in Cope's handwriting: "?Lytoloma angusta Cope,
Birmingham, N. J." (see pi. 15, fig. 4). Hay argues correctly that Cope would
hardly have doubted the identification of this specimen had he regarded it as
part of his type material.

' There seems to be some doubt as to the association of these bones.

145

146 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

With the association of these bones in doubt, the one peripheral plate figured
by Cope becomes the sole name-bearer. The latter is, however, not characteristic
of any specific group of turtles. It might belong to a cheloniid, to Osteopygis,
Catapleura, or Toxochelys. The mandible, poorly preserved, is indistinguishable
from that of Osteopygis. In view of the fact that Cope did not make the mandible
the name-bearer of the genus and species and since the peripheral is unidentifi-
able, I suggest that the name Lytoloma ougusta be considered a nomen vanum.
The mandible (A.M.N.H. 1133) is here referred to the genus Osteopygis and like-
wise a mandible (Y.P.M. 913) and an anterior skull fragment (Y.P.M. 913a)
that had been referred to Lytoloma angusta by Wieland (1904b). This lower jaw
(Y.P.M. 913) was made the type of Hay's Lytoloma wielandi (Hay, 1908), and
the skull fragment was assigned to this species along with a fairly complete
carapace (Y.P.M. 625). The latter is here recognized as a member of the genus
Toxochelys and is designated as the type of T. atlantica sp. nov. (p. 196).

The unfigured type material, consisting of a questionable nuchal and a
first peripheral, of Lytoloma jeanesi Cope is lost. The referred bones (A.M.N.H.
1473), figured by Hay (1908, p. 157, figs. 193-195), belong very probably to a
mixed specimen. The first peripheral might belong to Toxochelys and the sixth
peripheral to Osteopygis. L. jeanesi Cope must also be regarded as a nomen
vanum.

The various species from the London Clay, referred to Lytoloma by Lydekker
(1889), will have to be reassigned; but to do this from the literature would, I
fear, merely result in a further complication of their already confused taxonomic
status. A restudy of the European material seems necessary to determine the
generic allocation of these species.

Until now, it was understood that the genus Toxochelys (even though some
species were known from skulls only) included those members of the family
Toxochelyidae whose carapaces have a strong sagittal carina extending from the
first neui'al plate to the suprapygal area. This was a proper deduction, since
none of the previously studied specimens that were associated with parts of the
carapace lacked a mid-dorsal keel.

Recently, Chicago Natural History Museum received a partial skeleton,
without skull, of a turtle from the Niobrara deposits of Kansas (C.N.H.M.
PR123). The carapace difi^ered from any known member of the genus Toxo-
chelys by the absence of a carina, yet it could not be identified with Porthochelys
laticeps, the only previously known flat-neuraled toxochelyid from the Niobrara.
On the other hand, C.N.H.M. PR123 agreed very closely with the most common,
unkeeled toxochelyid from the Mooreville Chalk of Alabama. The suspicion
arose that the genus Toxochelys might include flat-neuraled as well as carinated
forms, but this could not be proved until an unpublished, unkeeled specimen with
the skull associated was discovered in the collection of Yale Peabody Museum
(Y.P.M. 3602). The skull leaves no doubt as to the species to which this indi-
vidual belongs — Toxochelys latiremis Cope, the type species. Thus, unfor-
tunately, the familiar picture of the carinated chelonian from the Niobrara must

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 147

be divorced from the equally familiar name Toxochelys, which name now applies
to a far less spectacular gi'oup of tui'tles.

The genus Catapleura Cope appears to be a primitive cheloniid that con-
nects the families Cheloniidae and Toxochelyidae (see p. 267).

Phyllemys Schmidt belongs to Toxochelys. The isolated mandible from New
Jersey on which Hay based his, Erquelinnesio molaria (Hay, 1908) very probably
belongs to Osteopygis. Rhetechelys Hay is tentatively included as an osteopygine
toxochelyid, but it may possibly be a cheloniid.

The family Toxochelyidae includes, under the present interpretation, the
following genera:

Carapace unkeeled: Toxochelys Cope, Porthochehjs Williston, Thiuochelys
gen. nov., Osteopygis Cope.

Carapace keeled: Lophochelys gen. nov., Ctenochelys gen. nov., Prionochelys
gen. nov.

Incertae sedis: Cynocercus Cope, Rhetechelys Hay.

As will be shown, our knowledge of the toxochelyid turtles has increased
considerably since Hay's revision. In addition to the then known occurrences
in the Niobrara Chalk of Kansas, the Pierre Shale of South Dakota, Wyoming,
and Kansas, and the Greensand deposits of New Jersey, toxochelyid turtles have
since been collected from the Mooreville Chalk of the Selma Formation of
Alabama, the Taylor Marl of Texas, the Marlbrook Marl of Arkansas and the
Coon Creek Tongue of the Ripley Formation of Tennessee. Furthermore, notable
additional collections have accumulated from the Niobrara and the Pierre.

Abbreviations of Institutions

A.M. N. H. = American Museum of Natural K.U. (V. P.) = University of Kansas Museum
History of Natural History

A.N.S.P.= Academy of Natural Sciences of S.D.S.M. = South Dakota School of Mines

Philadelphia U.S.N.M. = United States National Museum

C.N.H.M. = Chicago Natural History U.T. = University of Tennessee

Museum Y.P.M. = Yale Peabody Museum

See figure 29, Part III of this volume, for the names of localities used in
designations of the Alabama specimens.

GENERAL MORPHOLOGY OF THE TOXOCHELYIDAE

The Toxochelyidae are a group of marine tui'tles whose degree of aquatic
specialization ranges from that of near-shore dwellers to that of very highly
advanced, efficient sea-faring types. The outstanding aquatic specializations
appear in the differentiation of the skull, the shell, and the forelimbs but are by
no means typical for this family. Both carapace and plastron are relatively light
in construction, and in nearly all of the forms fontanelles occur between the costal
plates and the peripherals in the carapace, and central and lateral fontanelles

148 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

occui' in the plastron. The more generaUzed representatives have an unkeeled
shell of oblong, or, more commonly, circular outline, whereas the advanced
species are provided with a mid-dorsal carina and the outline of the carapace
tends to be cordiform.

Skull

An unusually large number of toxochelyid skulls and skull fragments have
accumulated in various collections. Most of the material belongs to Toxochelys
latiremis, so that the following account is based primarily, though not exclusively,
on this species. The vast majority of the skulls have been found isolated, but
the few specimens associated with shell material have made possible the proper
allocation of all skull and jaw specimens to species also known from shells. The
state of preservation must be discussed here because it tends to bias the com-
parison with such Recent forms as sea turtles and snapping turtles. Quite gen-
erally, the skulls are crushed dorso-ventrally to a much greater extent than is
apparent at casual examination. Taking a moderate amount of flattening into
account, one is led to compare the over-all shape of the toxochelyid skull with
that of the common snapping turtle (Chelydra serpentina). More careful analysis
of the crushing and some rare, uncrushed fragments reveal, however, that the
toxochelyid skull is much higher than the chelydrid skull, possibly as high as
that of the Recent Caretta.

The toxochelyid skull is more or less triangular in dorsal outline, with
broadly rounded sides and a fairly blunt snout region. Its height is difficult to
determine, but it probably approximates that of the modern sea turtles; it is
most certainly higher than the skull of Chelydra. The palate is a primary one,
except in the Osteopyginae, which possess a solid secondary palate much as
the Recent Caretta. The orbits face outward and upward and the external nasal
opening faces forward and upward. The posterior excavation of the roof of the
skull is intermediate in extent between that of living sea turtles and snapping
turtles. Regarded as a whole, the toxochelyid skull combines features observed
either in cheloniid or chelydrid turtles; only rarely are they intermediate between
the compared conditions.

The general arrangement of the bones of the skull roof resembles that of
Recent sea turtles. In the Toxochelyinae and Lophochelyinae, the frontals form
part of the dorsal rims of the orbits (fig. 60) ; in the Osteopyginae (e.g. in Rhete-
chelys) the frontal bones do not reach the orbital rims. The postorbital bones
extend much farther backward than in cheloniid or snapping turtles, namely,
to the very tips of the squamosal processes. The squamosal bones are unusually

Fig. 60. Reconstruction of skull of Toxochelys latiremis. a, angular; art, articular;
b, basisphenoid ; bo, basioccipital ; cor, coronoid; d, dentary; eo, exoccipital; /, frontal;
j, jugular; mx, maxilla; no, nasal; o, opisthotic; p, parietal; pal, palatine; pf, prefrontal;
pmx, premaxilla; po, postorbital; pro, preangular; pt, pterygoid; q, quadrate; qj, quadrato-
jugal; s, squamosum; sa, surangular; so, supraoccipital ; v, vomer.

Toxochelys latiremis

pmx

pmx

pmx

dorsal

149

150 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

small, forming only the ventro-lateral (but not the dorsal or medial) faces of the
squamosal processes (fig. 60). The squamosals are thus excluded from contact
with the parietal bones, a condition not due merely to the posterior excavation
of the skull roof, as in the snapping turtles and many other forms. Small nasal
bones have definitely been observed in Toxochelys and Porthochelys.

The bones of the ventral and posterior faces of the skull present a picture
of considerable morphological interest. In the Toxochelyinae, the palate is a
primary one; the Osteopyginae have a secondary palate; the palate of the Lopho-
chelyinae presents a condition intermediate between the two extremes. The
palates in these three subfamilies, arranged in a morphological series (fig. 61),
illustrate the mode in which the secondary palatal condition arose from the
primary one within this family.

The toxochelyid skull (especially that of Toxochelys latiremis) has been
compared repeatedly (Hay, 1908; Wieland, 1902; Case, 1898) with that of the
Recent sea turtles and the snapping turtles. The observers agree that the
palatal and pterygoidal areas of Toxochelys resemble the corresponding regions
in Chelydra, whereas the basisphenoid, occipital, and quadratal areas compare
much more closely with those of the cheloniids. In view of the fact that, ob-
viously, two primary palates (e.g. those of Toxochelys latiremis and Chelydra
serpe7itina) resemble each other more closely than do a primary and a secondary
palate (e.g. those of Toxochelys and Chelonia), one might be justified in suspecting
that the noted similarity between Toxochelys and Chelydra is merely a superficial
one. The skull of Toxochelys might be cheloniid in structure throughout, ex-
hibiting, however, a primitive palate.

The fortunate find of a beautifully preserved braincase of Toxochelys moore-
villensis (C.N.H.M. PR219) furnishes additional evidence in support of the view
that the toxochelyid skull combines cheloniid with chelydrid features (pi. 9), as
stated by earlier students. The region of the floor of the braincase anterior to
the dorsum sellae is chelydrid in its basic construction, whereas the area posterior
to the dorsum sellae closely resembles that of Chelonia or Caretta.

In Toxochelys, the basioccipital is visible from the dorsal aspect back to the
occipital condyle, a condition found in Chelonia only in juvenile forms. In adult
individuals, the exoccipitals encroach upon the basioccipital and form a sagittal
suture above it (pi. 9). In Toxochelys and Chelonia, the basioccipital, partic-
ularly in its anterior half, forms a sagittal crest (crista basioccipitalis, pi. 9).
According to Nick (1912), a cartilaginous knob, the tuberculum basale, rests
on this elevation. Anteriorly, the basis tuberculi basalis is formed by a small,
acute elevation on the posterior margin of the basisphenoid (pi. 9). The basis
tuberculi basalis is tiny in Chelydra but very pronounced in Chelonia and Toxo-
chelys. In the latter, it is more slender and more conspicuous because it is flanked
by deep sulci leading to the foramen jugulare anterius (pi. 9). The crista basi-
occipitalis divides posteriorly into a number of sharp crests that extend, in a
postero-lateral direction, to the exoccipitals. In Chelonia, similar ridges are but
faintly indicated or entirely absent.

a o

ex

>> c

(33 _>

P. i-

'So

». m
o •—
^ a;

s -^

^ a;

.2 c

CS o

I/l

+3

3
M

-2

to

cS

o

'c3

M

p.

&

h

O o

151

152 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The basisphenoid' can, for descriptive purposes, be divided into two por-
tions. The posterior part represents a wide, dorsally concave plate and forms
anteriorly the dorsum sellae, with the processus clinoidei on either side of the
latter. The anterior portion, the rostrum basisphenoidale, is an ossification
replacing the trabeculae baseos cranii and an intertrabecula. In most turtles,
the fossa hypophyseos lies at the base of the relatively thin and wide rostrum
basisphenoidale, in which case a true dorsum sellae occurs. In cheloniid tur-
tles, the rostrum basisphenoidale is high and narrow and its base is drawn
out lengthwise so that there is a notable distance between the fossa hypophyseos
and the transverse ridge that is homologous to the dorsum sellae of other forms.
In Toxochelys the rostnmi basisphenoidale resembles that of Chelydra, but in
Ctenochelys the entire basisphenoid compares very closely with that of Chehnia
(except for a slight difference in the position of the carotid canals; see below).
In all cases, including Chelonia, a short canal for the passage of the nervus abdu-
cens runs on either side, horizontally through the basisphenoid from the base of
the rostrum basisphenoidale underneath the processus clinoidei, to the main part
of the bone (pi. 9).

The canal for the internal carotid enters on both sides of the skull either by
a separate foramen or through an incisura at the ventro-medial corner of the
fenestra postotica (fig. 60). In Chelonia and Toxochelys there is a separate fora-
men, entirely embedded in the pterygoid. The course of the internal carotid
varies considerably among different forms (Siebenrock, 1897; Nick, 1912). In
Chelydra, Ctenochelys, and most probably in Toxochelys, the canal runs forward
and inward between the pterygoid, the pro-otic and the basisphenoid to the area
of the basisphenoid rostrimi, but does not join the sulcus cavern osus. There,
the internal carotid divides into the arteria cerebralis, which pierces the rostrum
laterally and enters the fossa hypophyseos, and the arteria palatino-nasalis, which
follows the lateral margin of the basisphenoid rostrum forward. In Chelonia, the
internal carotid canal is entirely embedded in the pterygoid posteriorly. Ante-
riorly, it runs between the pterygoid and the basisphenoid and finally joins the
sulcus cavernosus, where the carotid artery divides into the branches mentioned
above.

Close to the anterior opening of the abducens canal, there is, in Chelydra, a
second foramen that, according to Siebenrock (1897), serves for the passage of
the nervus vidiani. In Toxochelys, this foramen lies farther forward at the
bottom of the sulcus cavernosus, and in Chelonia it opens into the passage-
way of the arteria cerebralis through the lateral wall of the basisphenoid ros-
trum.

' It is now understood that the chelonian basisphenoid is a complex element, represent-
ing in many, if not most, turtles a fusion of the basisphenoid proper with a more or less well-
developed parasphenoid. In the adult skull, the two components are so intimately united
that they appear as one element. In Chelonia, only vestiges of a parasphenoid are reported
and it is virtually certain that none of the major parts of the fully developed basisphenoid
are formed by this dermal element. In Chelydra, on the other hand, the floor of the fossa
hypophyseos is said to be formed by the parasphenoid. For further detail, see Nick (1912).

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 153

The posterior, concave portion of the basisphenoid is divided into lateral
halves by a sharp sagittal crest, in Chelonia, Toxochelys, and Ctenochelys. In
Chelydra, this crest is but faintly indicated (pi. 9, crista basisphenoidalis) .

The lower jaws (figs. 60, 61) in the three subfamilies of Toxochehadae reflect
the palatal condition of the skulls. In the Toxochelyinae, the mandible is essen-
tially as in Chelydra. The masticatory surface is much narrower at the s>Tnphysis
than the ventral s\inphyseal shelf. The masticatory surface is much wider in
the Lophochelyinae. At the symphysis, it is as wide as the ventral shelf. In the
Osteopyginae, the mandible resembles that of the Recent Caretta. The sjmphysis
is very long, probably more than a third of the length of the mandibular ramus
(figs. 61 and 85).

Vertebral Column

Wieland (1902) described the cervical vertebrae of Toxochelys from a speci-
men in the Yale collection, Y.P.M. Accession 2491. He came to the conclusion
that these vertebrae are "inteiTnediate in character between those of Chelydra
and the Cheloniidae, being most like the former." In the meantime, another
series of cervical vertebrae, complete from the second to the eighth and in
rather good state of preservation (pi. 20, fig. A), has become available for study
(C.N.H.M. PR123, Toxochelys latiremis), as well as a fair number of isolated
cer\ncal vertebrae of Toxochelys moorevillensis, Ctenochelys tenuitesta, and C. acris.
Furthermore, Williams 1 1950) has published an extensive paper on the cervical
central articulations in living tui'tles, in which the results of careful observa-
tions on a large number of species and individuals are conveniently tabulated.
With this excellent ground work at hand, it is now possible to evaluate the
characteristics of the cer\ncal vertebrae of Toxochelys with accuracy much
gi-eater than Wieland could achieve at the time. The available material per-
mits, without doubt, the detemiination of the pattern of central articulations
among Recent tuitles to which the Toxochelyidae conform. Using Williams'
articulation diagram, the formula for T. latiremis (C.N.H.M. PR123) is as follows:

(2(, (3(, (4), )5), )6), )?), )8).

In all of the other specimens in which this observation can be made, there is
no doubling of the joint sui-faces between centra 7 and 8 (eight observations).
In none of the specimens is there a plane joint between centra 6 and 7 (seven
observations). This would tend to suggest that the most common foimula is:

(2(, (3(, (4), )5), )6), )7), )8) (see Table 1),

with doubling of the joint surfaces between centra 7 and 8 as a variant. The
above formula agi'ees, in detail, with Williams' fomiula PI, gi'oup 2, Derma-
temyidae, Chelydridae (Williams, 1950, p. 552). P2, the second most common
pattern for the same group, shows double joints between centra 6 and 7 and
between 7 and 8. Doubling of only the posterior one of these joints (as in
C.N.H.M. PR123) is apparently not known in Recent chelydrids.

Table 1.— PATTERN OF CENTRAL ARTICULATION IN
CERVICAL VERTEBRAE OF TOXOCHELYIDAE

Numbers of Vertebrae

Species
Toxochelys latiremis

PR123

Toxochelys moorevillensis

P27391

PR28

P27550

2

(2(,

3

(3(,

4 5

(4), )5),

6

)6),

7

)7),
)7),

8

J8)*

)8)*t

)8)*

)8)

Lophochelys natatrix

PR220

)8)*

Ctenochelys tenuitesta

P27351

P27548

P27361

PR248

')6)'
)6)

)8)*
)8)*

Ctenochelys acris

PR97

PR137

P27354

)5),

)6)
■)6)'

)8)t
)8)t

* Centrum of 8 short, j Centrum of 8 long.

X Slight indication of doubling between 7 and 8; ribs on 8.

Table 2.— MEASUREMENTS AND INDICES OF CERVICAL VERTEBRAE IN

CHELYDRA, TOXOCHELYS AND CHELONIA

{Indices in italics)

Cervical vertebrae
2 3 4 5 6 7 8

Species Length of centra

Chelydra serpentina* 24.5 30.0 39.5 33.0 32.0 32.0 22.0

100. 122. 161. 135. 131. 131. 90.

Toxochelys latiremis"^ 21.5 23.0 31.0 22.5 27.0 27.5 19.0

100. 107. 144. 105. 126. 128. 88.

Chelonia mydast 40.0 46.0 68.0 46.5 46.0 62.0 30.0

100. 115. 170. 116. 115. 155. 75.

Length of neurapophyses

Chelydra serpentina* 26.5 19.0 21.5 21.0 15.0 14.5 18.5

108. 78. 88. 86. 61. 59. 76.

Toxochelys latiremis^ 22.5 14.011 16.5 16.011 18.5 22.511 22.5

105. 65. 77. 74. 86. 105. 105.

Chelonia mydast 49.5 29.5 25.0 19.0 21.0 33.0 37.0

124. 74. 63. 48. 53. 83. 93.

Width of postzygapophyses

Chelydra serpentina* 20.0 21.0 21.0 22.0 19.0 22.5 27.0

82. 86. 86. 90. 78. 92. 110.

Toxochelys latiremis] 26.5 26.0 30.0 30.5 26.5 21.0 23.0

123. 121. 140. 142. 123. 98. 107.

Chelonia mydast 35.0 42.0 49.5 51.0|| 37.0 42.0 48.0

88. 105. 124. 128. 93. 105. 120.

* Specimen in private collection (No. 21). f C.N.H.M. PR123.
I C.N.H.M. 22066. H Approximate.

All indices calculated as follows: value X 100/ length of centrum of second vertebra.

154

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 155

Rarely, a formula identical with PI occurs in Recent cheloniid turtles
(Williams, N3, gi-oup 3, Cheloniidae, Dermochelyidae). Normally, however,
there is a plane joint between centra 6 and 7, and the two vertebrae may even
fuse entirely (e.g. in Chelonia mydas, C.N.H.M. 22066). The joint surfaces
between centra 7 and 8 are usually double in cheloniids.

There can be little doubt that double articulation surfaces and plane joints
are specializations tending to reduce the movability of the neck at specific points.
The fact that plane joints may even disappear by fusion of the adjoining centra
would certainly indicate that they are not to be considered as morphologically
intermediate conditions between amphicoelous and pro- or opisthocoelous joint
surfaces. In view of the vast amount of evidence gathered by Williams (1950),
it would seem reasonable to assume that, depending on whether the intercentral
cartilage mass becomes attached to the preceding or the succeeding of two
amphicoelous centra, pro- or opisthocoelous vertebrae develop directly, not via
an intermediate plane-jointed condition. If we thus interpret double articula-
tions and plane joints as specializations, the following formulae and their relative
frequency of occurrence present an interesting picture:

J 6 J, J7I, j8)
Chelydridae . less common

)6), J7/ i8),
Toxochelyidae rare

)6I. /7J, i8),
Che lor) iidae : comm on

)6I_ 17), )8)
Cheloniidae: less common

(2(. C3C, (4X )5). )6X )7X )8)

Chelydridae common — Toxochelyidae common — Cheloniidae:rare

The PI foiTiiula, here interpreted as primitive in the three families, occurs
in all three groups. In the Cheloniidae, it is rare and the gi'eatest number of
individual variants fall within the specialized patterns. In the Chelydridae and
Toxochelyidae, PI is the most common individual variant; the much less fre-
quent, specialized variants differ, however, possibly indicating trends in different
directions, since the less specialized toxochelyid variant does not seem to occur
in the chelydrids. Even though the number of observations on toxochelyid
turtles is still limited, the facts are at least suggestive.

If the relative lengths of the cervical centra, the neurapophyses, and the
widths of the latter across the posterior zygapophyses are compared with the cor-
responding dimensions in Chehjdra and Chelonia (fig. 62) it would seem that
Toxochelys exhibits a type of cervical differentiation peculiar to itself (see Table 2).
A noteworthy featui'e is the difference in the relative length of the centrum of the
eighth cervical vertebra in various species of the Toxochelyidae (see Table 3).

Of the thoracic vertebrae, individual centra and detached neurapophyses
are associated with a number of specimens, and nemapophyses occasionally
remain attached to neural plates. The centra are wider than high and, as usual.

^

TtT

TTT

WWW

p. c3

? c

4^

-5 -^

CS II

.S -2

^ 00

^ rt

2 ^

c3 II

Q-^

156

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 157

are notably narrower at mid-length than at the ends. The centra are hollowed
out lengthwise to form the ventral half of the neural canal, as in Chelydra. The
ninth, or first shell vertebra, is represented by a number of specimens. The
centrum is short and broad, ventrally keeled. On either side, there is a small
notch separating the articulation facet of the reduced first rib from that of the
large second rib (fig. 63). In Chelydra, these facets are very close together, and
in some specimens the ribs are proximally and distally in contact. In Recent
cheloniid turtles, the two described facets are spaced much farther apart (fig. 63).

The neurapophyses of the typical shell vertebrae are thin and probably
similar to those in Chelydra, though they are invariably distorted and difficult
to compare. A few well-preserved vertebrae from the area directly preceding
the sacral vertebrae are available in Toxochelys moorevillensis C.N.H.M. P27391.
There is a close degree of similarity between the centra of these elements and
those of Chelydra, but a sharp difference in the size of the neurapophyses. The
latter are relatively stout and strong. They are free from the carapace and the
most anterior of the five available elements forms strong diapophyses for the
attachment of the ribs. Closer to the sacral region, the transverse processes,
as in Chelydra, are synapophyses, formed jointly by the neural arches and the
centra. Some of the siirfaces with which these centra articulate are strongly
concave and convex, rather than plane. This, coupled with the fact that the
neurapophyses are free from the carapace, would indicate that there was a
certain amount of movability of the vertebral column in front of the pelvis,
which is an unusual situation.

The sacral region of Toxochelys consists of three vertebrae instead of two,
as in Chelydra. These are free from the carapace and bear strong, distally
expanded ribs. The size of these ribs increases notably from the first to the third.

Table 3.— MEASUREMENTS AND INDICES OF EIGHTH CERVICAL CENTRUM

IN TOXOCHELYIDAE

Eighth cervical centrum
Species Length* Width t Index

Toxochelys latiremis ™™- "^"^•

PR123 18.5 12.0 64. 8t

Toxochelys moorevillensis

P27550 19.0 11.0 57.9

P27391 15.5 12.5 80.6

PR28 11.0 10.0 90.9

Lophochelys natatrix

PR220 6.0 5.0 82.2

Ctenochelys tenuitesta

P27351 18.0 13.0 72.2

P27548 21.5 15.0 69.7

Ctenochelys acris

PR137 27.0 14.0 51.8

PR97 36.0 +19.0 52.8

* Length : distance from depth of anterior facet to peak of posterior facet.

t Width: measured at narrowest point of centrum.

X Low index due to double anterior joint surface in this individual.

158 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

In contrast to the presacral vertebrae, the centra of the sacral elements are
ventrally keeled, exactly as in Chelydra. The centrum of the (?)second sacral
vertebra of T. moorevillensis (C.N.H.M. P27391) is plane in front and strongly
convex behind. The transverse processes are synapophyses with the larger
portion of the joint surfaces formed by the neural arches.

The caudal region of the vertebral column is not yet adequately known and
the available evidence is somewhat confusing. In general shape, the tail verte-
brae correspond with those of Chelydra except that they all are procoelous
throughout and comparison reveals a number of details in which these vertebrae
resemble those of Chelouia. An entire caudal series is not known in any toxo-
chelyid, but in one specimen of T. moorevillensis (C.N.H.M. P27391), there are
three vertebrae; in another (C.N.H.M. PR136), there are two; in Ctenochelys
tenuitesta (C.N.H.M. P27351), there are two; and inC. acris (C.N.H.M. P27352),
there are five. All of these vertebrae, except those of C.N.H.M. P27391, can
be compared rather closely with the posterior caudals of Chelydra (save for being
procoelous) or Chelonia. They suggest a rather long tail much as in Chelydra
and probably longer than in Chelonia. The vertebrae of C.N.H.M. P27391,
however, do not seem to agree with this picture. The three vertebrae are small,
yet, compared to the tail series of Chelydra, they appear to belong to the anterior
third rather than to the posterior region of the tail. These vertebrae had dis-
tinct ribs attached chiefly to their neurapophyses and the size of the centra is
not greatly different from that of the sacral vertebra in this specimen (see above).
All this would tend to indicate that the three vertebrae here discussed are ante-
rior tail elements. But they seem much too small to have been followed by
vertebrae the size of those of C.N.H.M. PR136, an individual whose shell is
scarcely larger than that of C.N.H.M. P27391. In Chelydra, there is a slight
increase in the length of the centra from the first caudal to about the sixth
(in C.N.H.M. 22056, from 13.5 mm. to 19.0 mm.) ; thereafter the centra gradually
decrease in length. If a similar situation should prevail in Toxochelys, we should
have to assume an increase of centrum length from about 12 mm. to 21 mm.,
judging from the materials of C.N.H.M. P27391 and PR136. While these
figures do not seem inconceivable, they do not present the over-all size relation-
ship between the vertebrae, and direct comparison of the elements certainly
gives the impression that they could not possibly have occurred in the same
caudal series. The most plausible explanation is the assumption that a marked
sexual dimorphism in the tail length of Toxochelys existed, a condition not with-
out parallel in Recent turtles (e.g. in the Trionychidae).

Carapace and Plastron

As usual, the shell is the most frequently preserved part of the skeleton and
is consequently the best known part. The state of preservation varies greatly
even among individuals buried in the same formation and locality. Comparison
between well and poorly preserved shell remains is often possible only if large
collections are at hand.

"2

:^

5
5

DIUOISLID ,^

~o

, s-

S/^|aL(D0pid31 3

o

In
en

SjSuaiiiASJOouj X g

sAiaMOOL|do-i "=

15
c

SX|a^30J3DlM -S,

5
a

ojpXiaqo 2

^ 3

■? ^ S 5 i

C o - ^ ^
O Oj .= S -.^

:n :n Oj ^ -^

rt

-n

-I-I

^.

r;

^

«

P

cs

o

, ,

i?

O

rn

f,<

-r-

as

y-:

t.

^

■V

CO

"o

o
^

C>

o

S-.

>

A

x: c. -^ 0-i ^

159

160 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The morphology of the toxochelyid shell shows, as does the rest of the skel-
eton, similarities both to the chelydrids and cheloniids, and there is no single
feature that would distinguish it at once from these groups.

The carapace of the Toxochelyidae is circular, oval, or broadly cordiform.
It is either smooth (not broadly tuberculated as in the Chelydridae), or it is
provided with a sharp mid-dorsal carina, formed by the normal number of neural
plates (8-10), usually in conjunction with epithecal ossicles. The neural series
is always normal; there is no tendency toward reduction of these plates as
in the Thalassemyidae, or toward fragmentation into many supernumerary
elements as in cheloniid tui'tles. Successive neurals are hexagonal, rather narrow
and of similar outline, never expanded transversely or differentiated into alter-
nating octagonal and rectangular plates as in the Recent Chelydridae. A pre-
neural element occurs in some species. The nuchal plate lacks lateral processes
such as those in the Chelydridae, and the ventral boss of the cheloniid, denno-
chelyid and protostegid nuchal is small, insignificant or entirely wanting. The
peripheral plates are nearly always well developed; they may or may not be
suturally attached to the costal plates. The rib of the first costal plate always
forms a pit in the third peripheral, never in the fourth, as in some cheloniid
turtles. The peripheral edge of the shell may be even or serrated, but these two
conditions are always correlated with the differentiation of the neural series;
flat, uncarinated forms have an even, unserrated carapacial edge, whereas all
carinated genera have a serrated peripheral edge. Typically, there are two large
suprapygals; the lower one, however, may be greatly reduced.

The plastron is cruciform, essentially chelydrid in shape and construction,
but the axillary and inguinal notches are not as large as in Chelydra. The width
of the plastral bridge (hyo- and hypoplastron) at the narrowest point amounts
to about 17 per cent of half the width of the plastron in Chelydra. The com-
parable value ranges, in the Toxochelyidae, from 35 per cent, in forms with
circular to oval shells, to 60 per cent in elongated species. In the Cheloniidae,
it ranges from 55 per cent in forms with more or less circular outline to 110 per
cent in species with greatly elongated, cordiform carapaces (see Table 4). In
all known toxochelyid species, the epiplastra are small, relatively insignificant
elements in contrast to those in the Cheloniidae, which are as well developed
as the xiphiplastra. Hyo-, hypo-, and xiphiplastra form, parasagittally, two
more or less distinct keel ridges to which keel bosses are added in the more
specialized genera.

Girdles and Limbs

The toxochelyid shoulder girdle is cheloniid in shape, resembling most
closely that of the Recent Lepidochelys. The scapula has a definite neck area
between the glenoidal and coracoidal facets and the base of the bifurcation.
In chelydrid turtles, this neck area is wanting. The coracoid is distally expanded
as in Lepidochelys and generally resembles this form quite closely. The length
relationships between the neck, the medial and dorsal processes of the scapula

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 161

Table 4. COMPARISON OF THE RANGE OF INDICES IN THE PLASTRON OF

THALASSEMYIDAE, CHELONIIDAE AND TOXOCHELYIDAE

Index = axillo-inguinal distance xlOO half width of plastron

Thalassemyidae

Idiochelys fitzingeri (Lortet, 1892) 93

Thalttssemys marina (from Fraas, after Abel, 1919) 93

Cheloniidae

Lepidochelys kempi,C.N.liM.SlSS4: 55

Chelonia mydas,C.l<i.}iM. 22066 97.5

Eretmochelys imbrkata,C."Si. ¥1.1^1.222^2 100

Catapleura arkansaw, C.N.H.M. P27045 72

"Lytoloma" longiceps (Owen 1849-84, vol. 2, pi. 13) 82

Argillochelys subcristata (Owen 1849-84, vol. 2, pi. 10) 86

Argillochelys convexa (Owen 1849-84, vol. 2, pi. 14) 98

"L2jtoloma" planimentum (Owen 1849-84, vol. 2, pi. 16) 98

Protosphargis veronetisis (from Capellini, after Abel, 1919) 107

Toxochelyidae

Toxochelys barberi, C.N.H.M. P27047 37

Toxoc/ie/j/s 7wooreri7Zens?s, C.N.H.M. P27330 39

C.N.H.M. P27348 45

Toxochelys laiiremis, K.U.(V.P.) 1244 40

Toxochelys weeksi, U.T. K20 43

Thinochelys lapisossea, C.N.H.M. P27453 42

Osteopygis emarginatiis, Y.P.M. 783 47

Porthochelyslaticeps,K.\].(Y. P.) 1204 53

Lophochelysnatatrix, C.N. }i.M.PR220 57

Ctenochelys stenopora (type material) 57

U'S.N.M. 6013 58

A.M.N.H. 6137 47

Ctenochelys temiitesta, C.N.H.M. P27551 59

and the coracoid are illustrated in figure 64. The dorsal process of the scapula
is much longer than the coracoid in the chelydrid turtles. In the Toxochelyidae
(except in juvenile forms), it is shorter. In the Cheloniidae, the coracoid is
enormously elongated (see Table 5) .

In contrast to the shoulder girdle, the pelvis is chelydrid in shape (fig. 65).
The most obvious difference between the chelydrid and cheloniid pelves lies in
the relative sizes of the pubes and ischia. In the cheloniid tiu'tles, the ischia are
diminutive compared to the pubes and the strong posterior processes of the
chelydrid ischia are either absent (e.g. in Chelonia mydas, C.N.H.M. 22066) or
exceedingly small (fig. 65). The toxochelyid ischium is quite large (figs. 65 and
66 and Table 6) and the posterior process is even more pronounced than in
Chelydra. In Toxochelys mooreviUensis, the shape of the pubis is highly variable
with regard to the size and shape of the anterior notch and the width of the
antero-lateral process (fig. 65), which is wider than in Chelydra but not much
wider than in Macrochelys. The pubes of C.N.H.M. PR166 (fig. 65) show,
furthermore, that the antero-lateral process pointed ventrad as in Chelydridae;
in cheloniid tui'tles, it lies in the same plane as the sjTnphyseal portion of the

162 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

pubis. The anterior notch is deeper than in the Cheloniidae. The ilia of Toxo-
chelys also compare rather more closely with those of Macrochelys than with any
cheloniid. The dorsal portion, attached to the sacral ribs, is long and very
rugose medially. In cheloniid species, this dorsal process of the ilium is short
and not notably rugose medially. Instead, there is a dorso-lateral rugosity that
contacts the carapace. Nothing comparable is seen in Toxochelys.

Limbs that are articulated or even partially articulated are unknown in
toxochelyid turtles. A forelimb of Toxochelys latiremis was described and figured
by Wieland (1902), but it was not articulated and there are some doubts as to
the correctness of the reconstruction. Isolated limb-bones are associated with
many specimens, but only rarely is there enough material to permit identification
and to justify detailed description and comparison.

Table 5.— MEASUREMENTS AND INDICES OF SHOULDER GIRDLE IN

CHELYDRIDAE, TOXOCHELYIDAE AND CHELONIIDAE
Species Value Index Value Index Value Index Value Index

Chelydridae abed

Chelydra serpentina* 80.5 100 17.0 21.1 95.0 118.0 73.0 90.7

Macrochelys temmincki

C.N.H.M. 22060 108.0 100 30.0 27.7 150.0 138.8 108.0 100.0

Toxochelyidae
Toxochelys latiremis

Y.P.M. 3602 26.0 100 13.5 51.9 34.5 132.7

C.N.H.M. PR123 94.0 100 48.0 51.0 122.0 129.7 137.0 145.7

Toxochelys moorevillensis

C.N.H.M. PR136 ±87.0 100 42.5 48.8 100.0 114.9 + 125.0 143.6

Lophochelys natatrix

C.N.H.M. PR220 (juv.) 34.3 100 13.5 39.4 39.4 114.8 37.0 107.8
Ctenochelys stenopora

A.M. N.H. 6137 74.0 100 34.5 46.6 88.0 118.9

Cheloniidae
Lepidochelys kenipi

C.N.H.M. 31334 76.0 100 35.5 46.7 90.0 118.4 118.0 155.2

Eretmochelys imbricata

C.N.H.M. 31009 66.0 100 31.0 46.9 78.0 118.1 119.0 180.3

Chelonia mydas

C.N.H.M. 22066 153.0 100 89.0 58.1 183.0 119.6 304.0 198.7

Value a = length of ventral prong of scapular fork, measured from tip of process to edge

across neck of scapula.

Value b = length of scapular neck, measured from base of fork to ridge dividing glenoidal

facet from coracoidal suture face.

Value c = length of dorsal prong of scapular fork, measured from tip of process to edge across

neck of scapula.

Value d = maximal length of coracoid.

All indices are calculated as follows: value X 100/ value a

* Private collection (No. 21).

On the basis of a comparative study of the turtle humerus (1900) and the
forelimb (1902) of Toxochelys, Wieland concluded that the limbs of the Toxo-
chelyidae were primitive sea-turtle flippers, but that they indicated, with more
or less distinctness, a Chelydra-Mke ancestry. Comparison of various features

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

163

Chelydra serpentina

Toxochelys moorevillensts cnhm P2739I

^C^

Eretmochelys imbncata

Fig. 65. Comparison of pelves of Chelydra, Toxochelys, and Eretmochelys. C.N.H.M.
PR166, pubis of another specimen of T. moorevillensis.

of the forelimb reveals a definite mixture of resemblances to the sea turtles on
the one side and the chelydrids and trionychids on the other. To this extent,
Wieland's analysis was correct. There is, however, a profound difference in the
type of locomotion between sea turtles and fresh-water turtles. If the Toxo-
chelyidae present an intermediate condition, as the morphology seems to suggest,
a more detailed analysis of the problem is required.

Morphology and Function of the Limbs in Fresh- Water Turtles:
All fresh-water turtles propel their bodies by means of lateral strokes of the
limbs. In some cases, all foiir limbs are actively used (Chelydridae, Kinosterni-
dae, Trionychidae) ; in others, the hind limbs alone act as paddles and the fore-
limbs are held tightly against the body during swift locomotion (Emyidae,
particularly Graptemys, Pseudemys, etc.). In all instances, the right and left

164 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Table 6.— MEASUREMENTS* AND INDICES OF SURFACE AREAS OF ISCHIUM
AND PUBIS IN CHELYDRA, ERETMOCHELYS AND TOXOCHELYS

Ckelydra serpentina Pubis, mm.-' Index Ischium, mm.'^ Index

Private specimen (No. 21) 2174 100 1322 60.8

Eretmochelys imbricata

C.N. H.M. 22352 2217 100 331 14.9

Toxoch elys moorevillensis

C.N.H.M. P27391 2308 100 1069 46.3

* The mm.'- count was made on outline drawings of the bones, obtained by photographic
means and enlarged to about the same over-all size. Outline drawings illustrated in figure 66.

hind limbs and /or forelimbs are used in alternate fashion, never (to my knowl-
edge) simultaneously. The paddle stirfaces of the hind limbs are larger than
those of the forelimbs, and the greater size tends to indicate that they are the
more important organs of propulsion. Both fore- and hind limbs are retractable,
that is, they can be folded up. In the retracted forelimb, the humerus points
forward and inward; the remainder of the limb turns backward sharply at the
elbow joint and lies nearly parallel to the humerus. The hind limb can be folded
up at two points, the knee and the ankle. In the retracted position, the femur
and the pes point forward and outward, the zeugopodium backward and inward.

The type of locomotion and the range of movability of the limbs of fresh-
water turtles are reflected in the structure of the limb bones. The hind limbs are
usually somewhat longer than the forelimbs (fig. 67). The humerus is an "S"-
curved bone, with the head standing nearly at a right angle to the main axis of
the shaft (fig. 68). The curvature in the distal two-thirds of the bone is not as
pronounced as near the proximal end and varies notably in different genera.
The distal joint sui'face extends in a marked elevation onto the ventral' face, thus
permitting extreme bending of the limb at the elbow joint. The principal move-
ment of the humerus is in a horizontal plane that forms a very small angle
(about 10° in Chelydra, fig. 69) with the main axis of the humeral shaft. As
Wieland (1900) has pointed out, the radial process lies at the level of the caput
humeri. Radius and ulna are located, side by side, in the plane of the manus and
form a typical spatium interosseum. The radius is slightly shorter than the ulna.
The carpus consists of numerous, nearly spherical bones that permit a great deal
of movement in this region. The manus differs somewhat among genera. As a
rule, digits 1, 2, and 3 are the most robust and bear strong claws. The remaining
digits may bear claws or lack them as in trionychid turtles, where even a moderate
polyphalangy may occur on digits 4 and 5 (for example, in Trionyx cartilagineus,
C.N.H.M. 11088, the phalangeal formula is 2, 3, 3, 4, 4). Digits 3 {Chelydra)
or 4 {Triovyx) may be the longest.

The femur of fresh-water turtles is usually longer than the humerus (see
Table 7). Tibial and fibular trochanters are separated by a deep "V"-shaped
depression. The proximal ends of these processes have cartilage-covered surfaces

' The morphologically ventral side of the humerus of fresh-water turtles faces postero-
laterad in the animal.

/oo. . . ,

o

1

I

Chelydra

Lepidochelys

Fig. 66. Indices of pubic-ischial surface
areas in Chelydra, Toxochelys moorevillensis,
and Eretmochelys.

Fig. 68. Comparison of angle a in humeri
of Chelydra, Lepidochelys, and a toxochelyid.

Chelydro

Fig. 67. Limb proportions in Chelydra,
Toxochelys, and cheloniid turtles. Shaded
columns = forelimbs (H, humerus; U, ulna);
black columns = hind limbs (F, femur; T,
tibia).

Fig. 69. Comparison of angle (3 in humeri
of Chelydra, Lepidochelys, and a toxochelyid.

165

166 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

that are connected with the head of the femui' (fig. 70). Tibia and fibula are
very long (fig. 67). In Chelydra, the fibula has distally a thin, laterally projecting
ridge. The tarsus is provided with intertarsal joint surfaces extending to the
dorsal faces of the tarsal elements, thus providing the required range of move-
ment for the retraction of the pes. The latter, in retracted position, is bent for-
ward against the zeugopodium. The foot of fresh-water turtles is specialized
on the fibular side. Phalanges 1, 2, and 3 are large and bear claws. Metatarsal
V is enlarged and stands at an angle of about 55° to the main axis of the pes and
is movable against distal tarsal 3+4. Metatarsal IV rests against metatarsal V.
The mechanism pennits the fourth and fifth toes to be spread away from the
third and to tighten the large webs between these toes. The fifth digit has four
phalanges in Chelydra.

The limbs of the fresh-water turtles are, of course, not exclusively swimming
organs. They also serve for walking on land and for digging, and the forelimbs
are often used to aid in feeding. The fresh-water type of aquatic locomotion is
highly developed in the Trionychidae and amazingly efficient as regards speed.

Morphology and Function of the Limbs in Sea Turtles: The type of
locomotion in sea turtles differs greatly from that of fresh-water turtles. In all

Table 7.— HUMERUS FEMUR PROPORTIONS IN FRESH-WATER TURTLES,
SEA TURTLES AND TOXOCHELYID TURTLES*
Fresh-water turtles Femur

Gbjptopsplicatulus, AM.N.}1.S36 114.3

Baewo npona, A.M.N. H. 5977 101.1

Chelydra serpenti7ia, private specimen (No. 21) 109.1

Chehjdra serpentina, C.N. UM. 22058 103.2

Sternotherusodoratus, C.N. E..M. 31218 105.8

rnow2/x/e?-ox,C.N.H.M. 31014 103.0

Tn"oKi/x mM/icMS, C.N.H.M. 22065 109.4

Psewdem^/s^onV/om, C.N. H.M. 22473 112.1

Sea turtles

Lepidochelyskempi,C.N. a. M.SlSSi 92.1

Eretmochelysimbricata, C.N. li.M.S1009 71.8

Protostega gigas (from Wieland, 1906) 79.8

Dermochelys coriacea (from Wieland, 1906) 72.8

Toxochelyid turtles

Toxochelys latiremis, Y. P.M. 3602 91.9

Osteopygis emarginatus, Y.F.M. 183 105.2

* Humerus = 100.

cheloniid, dermochelyid and protostegid turtles, the forelimbs are the principal
organs of locomotion; the hind limbs are used only for steering diiring straight
propulsion. The forelimbs, specialized flippers, are longer than the hind limbs
and propel the body by simultaneous, vertical strokes. A sea turtle "flies" through
the water. Neither the flippers nor the hind limbs are retractable; that is, they
cannot be folded up as in fresh-water turtles. In contrast to the forelimbs of
fresh-water turtles, the sea-turtle flippers are almost exclusively organs of pro-
pulsion. The propulsion speed is slow compared to that of trionychid turtles.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

167

Toxochelys lotiremis
CNHM PRI23

Fig. 70. Forelimb of Toxochelys latiremis (C.N.H.M. PR123; elements of third digit
may not be correctly placed); hind limb of Toxochelys moorevillensis (C.N.H.M. PR136);
comparison of right femora of Chelydra, Toxochelys, and Lepidochelys.

but the limbs of the latter are put to intermittent use, whereas those of the sea
turtle are in continuous, slow motion during most of the waking houi's of the
individual.

The peculiarities of the sea-turtle type of locomotion are clearly reflected
in the limb skeleton. The flipper is conspicuously longer than the hind limb.
The humerus, longer than the femur (see Table 7), moves in a vertical plane and
is almost straight. The angle between the axis of the caput humeri and the shaft
is about 127° in Lepidochelys kempi (fig. 68). The surface of the distal joint is
restricted, thus permitting only a moderate amount of movement at the elbow.

168 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The angle between the axis of the shaft and the principal plane of movement of
the humerus in Lepidochelys is 68° (fig. 69), as against 11° in Chelydra. As
Wieland (1900) stated, the radial process is located on the shaft more distally
than in fresh-water turtles.

In the zeugopodium, the radius is much the longer of the two bones (see
Table 8). Radius and ulna do not lie side by side in the plane of the manus;
instead, the ulna lies above the radius in a plane more or less rectangular to
that of the manus. Near its distal end, the ulna is attached to the radius by
means of a contact rugosity. The carpus is characteristically modified. Inter-
medium and ulnare of the proximal carpal row are greatly elongated and dorso-
ventrally flattened, as are all of the carpal elements except the radiale and the
distal carpal I. The joint surfaces of all of the carpal elements are flat and sharply
restricted in the proximo-distal and transversal contact areas of adjoining
elements. In old individuals of Chelonia, the distal carpals on the ulnar side
fuse to form a solid, thin, transverse plate. Very little movement is thus possible
in this area.

Metacarpals and phalanges are much elongated. All of the fingers, including
the first, which bears a strong, curved claw, are firmly included in the flipper and
therefore not capable of separate movement. The manus is stifi'ened by tight
dorsal and ventral ligaments and by tough skin that forms large, horny scales
along the edges. The third is the longest digit in the flipper.

The hind limb, specialized as a steering organ, is characterized by the
relatively great width of the tarsus and the pes and by a peculiarity at the
proximal end of the femur. In fresh-water turtles, the femoral trochanters are
separated by a depression. In sea turtles, there is no trace of a depression and
the cartilage-covered proximal surfaces of the trochanters form a continuous
ridge opposite the head of the femur (fig. 70). Tibia and fibula are straighter
than in fresh-water turtles, thus leaving a much narrower spatium interosseum
between them. The fibula is thick distally, lacking a thin lateral ridge as in
Chelydra. In old individuals of Chelonia, the tarsal bones undergo extensive
fusion, so that only three compound elements remain. In a near-adult specimen
of Lepidochelys, there are six tarsal elements; namely, a tibiale + intermedium +
centrale, a fibulare, and four distal tarsals. The fourth is gi'eatly enlarged and
represents, probably, a fusion of distal carpals 4 + 5. In the specimen of
Chelonia mentioned above, the two proximal elements of the Lepidochelys tarsus
have fused into one large ossicle. Distal tarsals 1 and 2 are fused and distal
tarsal 3 is fused to the enlarged fourth element. Metatarsi I and V are much
enlarged, giving the pes its great width. Digits 2, 3, and 4 are moderately
elongated, the third being the longest. The end phalanx of the first digit bears
a large claw. Nearly all elements of the tarsus and the pes are dorso-ventrally
flattened.

The Morphology of the Limbs and the Probable Type of Locomotion
IN THE Toxochelyidae : The morphology of the limbs of Toxochelys, compared
to that of the Chelydridae and Cheloniidae, presents a very interesting picture.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 169

Briefly, it may be stated that the forelimb is a primitive sea-tiu'tle flipper, whereas
the hind hmb is chelydrid throughout its known parts.

Even though the over-all length of the fore- and hind limbs cannot yet be
determined in any specimen, there is considerable indirect evidence suggesting
that the forelimbs are longer than the hind limbs, but they are probably not as
different in length as in Recent sea turtles.

The humerus is longer than the femur (fig. 67). Its shaft is slightly S-shaped
in side view, and the axis of the caput humeri stands at an angle of about 140°
to that of the shaft (fig. 68), about the same value obtained in Eretmochelys.
The angle between the longitudinal axis of the shaft and the plane of movement
(fig. 69) is about 29°, intermediate between those of Chelydra ( ± 10°) and Lepi-
dochelijs (±68°). These two angles (figs. 68, 69) express most clearly the typical
differences between humeri that are moved in a horizontal plane (as in fresh-
water tm-tles) and those that move in a vertical plane (as in sea turtles). The
transition from the former to the latter type involves (1) enlarging of angle a
(fig. 68), which straightens the caput humeri in relation to the shaft, (2) an axial
rotation of the entire humerus to the extent of something less than 90°, and
(3) increase in the angle i3 (fig. 69), which produces a more anterior position of
the distal end of the humerus relative to the caput humeri and therefore permits
greater freedom of movement within the limits imposed by the shell. While
angles a and (3 can be measured on any well-preserved humerus, the amount of
axial rotation of this element with regard to the body as a whole can only be
determined on Recent material. Angle a (fig. 68) of + 90° is primarily correlated
with the retractability of the forelimb. Since the bridge of the shell in most
fresh-water turtles, though not in chelydrids and trionychids, sets a limit to the
extent to which the humerus can be pulled backward, angle a of + 90° extends
the range of horizontal movability forward to the point where the elbow touches
the neck (fig. 71). If angle a were significantly greater than 90° it would impair
the retractability of the limb and would extend the range of movement farther
outward and backward. An extended position such as is theoretically indicated
in figure 71, right, is not seen, to my knowledge, in any fresh-water turtle,
including those with reduced plastra where the axillary notch of the shell is more
than adequate to permit that position.

In fresh-water turtles, angle I3 (fig. 69) determines the position of the elbow
with regard to the plane of the plastron; it is above this plane if the angle is
small ( + 10° in Chelydra). A significantly greater angle would lower the elbow
below the plastral plane, a theoretical position that could scarcely be considered
functional (fig. 71).

In view of these functional considerations, we may be justified in regarding
the values of angles a and (3 as indicative of the type of movement of the humerus
in a fossil form even though the axial position of this element with regard to
the body is not determinable. The values of a and i3 in the toxochelyid humerus
suggest that it was moved in a vertical plane as in sea turtles and that the fore-
limb was not retractable. With regard to angle (3, it is less specialized than in any

170 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

of the Recent cheloniids. This conclusion is supported by the position of the
radial tuberosity distal to the caput humeri. In the Toxochelyidae, the radial
tuberosity is a single, large process; in the Cheloniidae, it actually consists of
two processes whose cartilage-covered ends are proximally connected and form
an inverted "V." The distal joint surface of the toxochelyid humerus may have
permitted greater mobility at the elbow than is possible in Cheloniidae, but much
less than in the Chelydridae. According to Wieland's (1902) account of the
forelimb of Toxochelys, the zeugopodial bones are of nearly equal length (the
radius slightly longer than the ulna) and they lie side by side in the plane of
the manus. My own observations on other materials confirm Wieland's account,
except that one specimen (C.N.H.M. PR136, T. moorevillensis) indicates a some-
what longer radius, to judge from the length of the proximal and distal pieces
of the left radius that do not fit together yet are together as long as the left ulna.
Two well-preserved ulnae are available and neither shows a distal rugosity, indi-
cating a contact with the radius. In this respect, the forelimb is chelydrid, not
cheloniid. The few carpal elements available in the specimens at hand differ
notably from those illustrated by Wieland. One of these elements (C.N.H.M.
PR123, T. latiremis, fig. 70) compares well with the central carpal of Chelonia,
and the other, very probably the product of fusion of two or three distal carpals,
resembles somewhat the fused distal carpal element 3+4+5 in Chelonia. Since
PR123 is a larger specimen than that illustrated by Wieland, the noted dis-
crepancy probably reflects a difference in age of the two individuals. The carpus
of Recent sea turtles attains its final state of ossification only in old individuals.
According to Wieland, ulnare and intermedium are not elongated as in cheloniids.
The manus is partially specialized in the direction of a sea-turtle flipper. The
first and second digits are chelydrid and have large, curved, claw-bearing end
phalanges (fig. 70). The remaining digits, particularly the third and fourth, are
elongated. The elements illustrated in the position of the third digit in figure
70 are not necessarily correctly determined as such, but there is no reasonable
doubt that they belong to the manus.

In summary, it may be said that comparative morphological and functional
analysis suggests that the toxochelyid forelimb was a flipper moved in a vertical
plane as in the cheloniids, but it was not as highly specialized as in that group.

The hind limb of Toxochelys, in contrast to the forelimb, is chelydrid through-
out, with the possible exception of the tarsal elements, whose joint surfaces are
somewhat more restricted than in Chelydra (fig. 70).

In Toxochelys, the major organs of propulsion of the sea turtles are thus
combined with those of the fresh-water turtles. Functionally, this could mean
that the Toxochelyidae could employ either the chelydrid or the cheloniid type
of locomotion, using either the forelimbs or the hind limbs for steering. The
chelydrid type of locomotion might have been used for quick escape, the cheloniid
type for cruising. Phylogenetically, the Toxochelyidae represent a stage in the
transition from a chelydrid to a cheloniid type of locomotion, in which the hind
limbs are scarcely modified.

Fig. 71. A. Dorsal view of shell, shoulder girdle, and humerus to illustrate functional
significance of angle a. B. Cross section view of shell, shoulder girdle, and humerus to
illustrate functional significance of angle /3. See page 169.

171

172

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

COMPARISON OF MORPHOLOGICAL AND FUNCTIONAL CHARACTERISTICS
OF LIMBS IN FRESH-WATER TURTLES, SEA TURTLES, AND TOXOCHELYS

Fresh-water turtles

Allfo2ir legs used in swimming;
hind limbs major organs of pro-
pidsion

1. Hind limbs longer than
forelimbs

2. Femur longer than
humerus

Forelimbs move in horizontal
plane in swimming; retractable;
also used for walking and digging

3. Humerus "S"-shaped

4. Angle of head to shaft
±90°

5. Axis of shaft ±10° off
plane of movement

6. Radial process at level of
head

7. Distal joint surface
strongly convex; extends
to ventral side

8. Radius and ulna lie in
plane of manus

9. Large spatiuminterosseum

10. Radius shorter than ulna

11. No contact rugosity be-
tween radius and ulna

12. Carpal elements almost
spherical

13. Digits 2, 3, and 4 normal
in length

14. Claws on digits 1, 2, and 3

Hind limbs major organs of
propulsion; retractable; toes 4
and 5 can be spread to increase
paddle surface

15. Femoral trochanters sepa-
rated by depression

16. Distal joint surface of
femur strongly convex

17. Fibula with distal lateral
ridge

18. Tarsal elements with joint
surfaces reaching to dorsal
face of tarsus

19. Claws on digits 1, 2, and 3

20. Metatarsus V movable in
lateral direction

21. Four phalanges on fifth
digit

Toxochelys

Forelimbs probably longer

than hind limbs

Humerus longer than femur

Humerus slightly "S"-shaped

Angle of head to shaft

±140°

Axis of shaft ± 29° off plane

of movement

Radial process distal to caput

humeri

Distal joint surface convex;

extends to ventral face

Radius and ulna lie in plane
of manus

Large spatium interosseum

Radius slightly longer than

ulna

No contact rugosity between

radius and ulna

Carpal elements flattened ;

joint surfaces restricted

Digits 3 and 4 elongated
Claws on digits 1 and 2

Femoral trochanters sepa-
rated by depression
Distal joint surface of femur
strongly convex
Fibula with distal lateral
ridge

Tarsal elements intermediate
between Chelydridae and
Cheloniidae
?

Metatarsus V movable

Sea turtles

Forelimbs major organs of pro-
pulsion; hind limbs used for
steering

Forelimbs much longer than

hind limbs

Humerus longer than femur

Forelimbs move in vertical
plane; not retractable; cannot
be used for ivalking or digging

Humerus nearly straight
Angle of head to shaft
± 127° or more
Axis of shaft ± 68° off plane
of movement

Radial process distal to caput
humeri

Distal joint surface much re-
stricted and relatively flat

Radius and ulna lie in a plane

approximately rectangular to

that of manus

Small spatium interosseum

Radius much longer than

ulna

Contact rugosity between

radius and ulna

Carpal elements flattened ;

joint surfaces restricted; ul-

nare and intermedium

elongated

Digits 2, 3, and 4 greatly

elongated

A claw on first digit only

Hind limbs steering organs;
not retractable

Femoral trochanters con-
nected

Distal joint surface of femur
relatively restricted
Fibula without distal lateral
ridge

Tarsal elements with flat
joint surfaces

A claw on first digit only

Metatarsus V scarcely

movable

Three phalanges on fifth digit

ZANGERL: VERTEBRATE FAUXA OF SELMA FORMATION 173

REVISION OF THE FAMILY TOXOCHELYIDAE
Family TOXOCHELYIDAE

Characterization. — Extinct sea turtles. Skull with primary or secondary
palate; temporal region partially roofed. Small nasal bones present or absent.
Cervical vertebrae short, as in Cheloniidae. Forelimbs developed as primitive
flippers; hind limbs chelydrid throughout. Carapace circular, oval, or, more
rarely, cordiform. Neural plates flat or keeled, in continuous series. Epineurals
only in keeled fonns. Almost always fontanelles of varying size between costal
and peripheral plates. A pair of small fontanelles at posterior end of nuchal
plate, or conspicuous thinness of shell at that point. Two suprapygal plates.
Vertebral shields in adult individuals always nan'ower than pleural shields.
Plastron of cruciform shape with distance between axial and inguinal notches
60 per cent or less than half the width of the plastron. Late Cretaceous, North
America.

The toxochelyid turtles readily fall into three groups: generalized, unkeeled
forms with primary palates; unkeeled forms with secondary palates; and keeled
forms with palates intermediate between the two types. These tliree groups
will be recognized as subfamilies below.

Subfamily Toxochelyinae

Characterization. — Relatively generalized toxochelyid turtles. Skull with
primary palate and narrow triturating surfaces anteriorly. Anterior vomer area
wide and smooth. Triturating shelf of mandible at symphysis narrower than
lower shelf. Small nasal bones. Carapace circular, rarely cordiform. Neuralia
flat. Peripheral edge of carapace without serrations (see below). Moderate
fontanelles between costal and marginal plates. Nuchal fontanelles in Toxo-
chelys only. Plastron with small central and lateral fontanelles and, consequently,
long hyo-hypoplastral sutiire. Xiphiplastra widest at mid-length, posteriorly
tapering to a point. No elevations, or only faint ones, on hyo- and hypoplastra.

Toxochelys, Porthochelys, Thinochelys

Horizon and locality. — Late Cretaceous, North America.

Discussion. — Skulls are known only in the genera Toxochelys and Portho-
chelys. In both genera, the basic designs of the skull and mandible are very
similar, and the diffei'ences are mostly of a quantitative nature. The skull of
Porthochelys is much more robust and much wider than that of Toxochelys.
Both strength of construction and over-all width are correlated with the size of
the masticatory surface areas on upper and lower jaws.

The masticatory surfaces of upper and lower jaws in Porthochelys are widest
near their posterior ends and involve notable portions of the palatines. An
isolated mandibular ramus, here tentatively referred to T. atlantica (p. 196), is

174 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Toxochelys latire7nis Porthochelys laticeps, type
Y.P.M. 3604 K.U. (V.P.) 1204

a. Length of skull 115 mm. Ill mm.

b. Width of skull at posterior end of triturating

surface 80 mm. 109 mm.

c. Posterior width of triturating surface 12 mm. 24 mm.

6XlOO/a = 69.6 98.2

cXlOO/a = 10.4 21.6

specialized as in Porthochelys. Small nasal bones are present in both genera.
The vertebral column, the girdles and the limb-bones are discussed above.

The shells in the three genera are rather flat, circular, or broadly cordiform
in outline and there are always at least small lateral fontanelles. There is no
pit in the second peripherals for the antero-lateral prong of the hyoplastron.
The carapace is unkeeled and the peripheral edge is even except in Portfwchelys,
where the peripherals are slightly serrated behind the marginal scale furrows,
thus differing from the condition in lophochelyine forms where the serrations
are always in front of the scale furrows.

The plastra in the three genera differ notably with regard to the relative
degree of excavation of the axillary and inguinal notches and the length of the
hy o-hypoplastral sutures :

Axillary-inguinal Hyo-hypoplastral Index
distance suture

mm. mm.

Toxochelys weeksi 46.5 70 66 . 4

U.T. K20
Toxochelys barberi 35.0 65 53 . 8

C.N.H.M. P27047
Toxochelys latiremis (juv.) 33 . 5 44 76 . 1

K.U. (V.P.) 1244
Toxochelys latiremis 24 . 44 54 . 5

Y.P.M. 3602
Toxochelys nioorevillensis 42.0 +76 55.2

C.N.H.M. P27330
Toxochelys moorevillensis 46.0 68 67 . 6

C.N.H.M. P27348
Porthochelys laticeps 53.0 +59 89.8

K.U. (V.P.) 1204
Thinochelys lapisossea 41.5 86 48 . 2

C.N.H.M. P27453

Toxochelys Cope

Characterization. — Skull with narrow masticatory surfaces, not at all or
only slightly involving palatine bones. Small nasal bones. Carapace oval or
more or less circular, with moderate to extensive costo-peripheral fontanelles.
A pair of post-nuchal fontanelles. Sutures between adjoining neural plates
variable, but most often rounded as in Thinochelys. Greatest width of mid-dorsal
netu'als exceeds half their length. Greatest width of foiu'th neural contained
fourteen to eighteen times in width of carapace. Postero-lateral peripherals (in

ZAXGERL: VERTEBRATE FAUNA OF SELMA FORMATION 175

specimens with small fontanelles) about as wide as long. Suprapygal plates
more or less equal in maximum width. Posterior width of second vertebral
shield much more than half its length. Medial sulci of marginal shields not
visible on peripheral plates, except sometimes on posterolateral elements.
Posterior edge of pygal plate not distinctly notched. Plastron with moderate
central and lateral fontanelles. Width of hypoplastron at narrowest point slightly
greater than corresponding dimension of hyoplastron. Medial edges of hyo-
and hypoplastron sen-ated. Shortest axillo-inguinal distance greater than half
the length of the hyo-hypoplastral suture (observed percentages 54 to 76).
Greatest width of xiphiplastron at mid-length, or, more often, in posterior half
of element.

Type of genus. — Toxochelys latiremis Cope.

Horizon and locality. — Niobrara Formation, Kansas.

Recognized additional species. — T. moorevillensis sp. nov.; Mooreville
Chalk, Alabama. T. barberi Schmidt; Marlbrook Marl, Arkansas. T. atlantica
sp. nov.; Greensand, New Jersey. T. weeksi Collins; Coon Creek Tongue,
Ripley Formation, Tennessee. T. browni Hay; Pierre Shale, South Dakota and
Wyoming. T. sp.; Taylor Marl, Texas.

Discussion. — T. hrachyrhina Case (pi. 11) is based on a badly crushed skull
of T. latiremis.^ The type specimen of T. serrifer Cope (pi. 16) consists of some
skull fragments, a lower jaw and two postero-lateral peripheral plates, the latter
not now with the specimen. Considering the fact that the skull fragments and
particularly the mandible are uncrushed, these parts are virtually indistinguish-
able from T. latiremis, but careful comparison revealed that the skull fragments
belong to a skull of the size of Y.P.M. 3602, whereas the mandible belongs to a
much smaller individual, probably about the size of C.N.H.M. UR4. There
can be no reasonable doubt that the lower jaw and the skull parts belong
to two individuals. Since the peripheral plates are serrated, and since there is
no other member of the genus Toxochelys that possesses seiTated peripheral
bones, the suspicion arises that a third individual (probably a lophochelyine
turtle) became mixed with this lot of bones before it was handed to Professor
Cope for description. Of the remaining species of Toxochelys recognized by Hay
(1908), only T. latiremis belongs to the flat-neuraled forms. All other species are
keeled and referable to the new genus Ctenochelys (p. 227).

Once the nature of T. latiremis and its morphological relationship to the
Mooreville material were recognized, it became evident that a number of un-
described specimens and several species described under different generic names
from various formations belong to Toxochelys. Among the described species,
Phyllemys barberi Schmidt (pi. 24) from the Marlbrook Marl of Arkansas,
Porthochelys browni Hay (pi. 10) from the Pierre Shale of South Dakota, and the
carapace (Y.P.M. 625) referred to Lytoloma angusta by Wieland (1904b) and to

' Dr. Edwin C. Galbreath, of the University of Kansas, has studied the type specimen
independently and has arrived at the same conclusion.

176 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Lytoloma wielandi by Hay (1908) from the Greensand of New Jersey (fig. 79)
clearly belong to Toxochelys. The latter specimen is the type of T. atlantica sp.
nov. (p. 196). Among undescribed specimens, the United States National
Museum possesses from the Taylor Marl of Texas a partial shell that is referable
to Toxochelys. A superbly preserved plastron, associated with three posterior
peripheral plates, was recently discovered by Dr. R. Lee Collins, of the University
of Tennessee, in the Coon Creek Tongue of the Ripley Formation in Tennessee.
This specimen also belongs to Toxochelys.

Skull and mandible. — There are a surprising number of Toxochelys skulls
available for study. Most of these belong to T. latiremis and a smaller number to
T. hrowni; two partial skulls and a mandible belong to T. moorevillensis and an
isolated mandibular ramus was tentatively referred to T. atlantica. The remaining
species of the genus are known from shells only. The comparison of the skulls
of the above-mentioned species involves careful consideration of the mode of pres-
ervation, in order to avoid, if possible, the description of "Gesetzmassigkeiten"
in the post-mortem changes rather than actual morphological differences between
the species.

T. latiremis and T. hrowni are very similar and can easily be distinguished
from T. moorevillensis by the shape of the mandible. In the latter species, the
triturating surface of the lower jaw forms a sharp sagittal crest along the dorsal
shelf of the symphysis. This ridge divides posteriorly and continues as a blunt
elevation along the inner margins of the triturating surface. The latter forms a
pronounced valley. In T. latiremis and T. hrowni, the sagittal ridge is faint or
entirely absent. The lingual border of the triturating surface is lower than the
alveolar ridge, often not clearly delimited, and the masticatory surface is much
less concave in medio-lateral direction. The skulls of T. latiremis and T. hrowni
are more difficult to distinguish, perhaps because of the generally poor state of
preservation of the materials from the Pierre Shale. The skulls from this forma-
tion range in size from about 74 mm. to about 170 mm. in snout to occipital
condyle length. The latter value was never observed in T. latiremis, where a skull
length of about 120 mm. appears to be adult size. In some skulls, fui'ther-
more, the premaxillary region of T. hrowni is more pointed — beak-like — than in
T. latiremis. There seems to be good reason to consider T. broivni as a distinct
species, even though the state of preservation renders detailed comparison
difficult. The jaw fragment tentatively assigned to T. atlantica differs from
all other known species of the genus by the relatively great posterior width
of the triturating sui'face. It is similar to Porthochelys in the direction of jaw
specialization.

Carapace and plastron. — The shell of Toxochelys is easily distinguished from
that of any other known genus of the subfamily and it is the best known. In the
Mooreville Chalk of Alabama, shell remains of Toxochelys are numerically the
most common finds, and this may also be true in the Niobrara, where the great
majority of the turtle skull materials belong to this genus and where unbiased
collecting would probably produce more shell material than is now available.

f^

i— 1

<

5 c«

Q

??'s'5

S °^

s

J «

H

a

X

o

c

X

o

J=

H

C O-C

p— 1

Ol

Eiq

iz:

o

T5

pq

C

in t-

O 00

PQ

J

H

cob

0) cb

Q
2

Q

hJ =-

K

Z

q

Q

c

Z

5

^^

<:

ffi

C O^

M

o

g

Q

^ ^

J
H

Z

X

Q

0/
C

pL4

-c «

O

5

MV-.2

Q

c o-o

^ 2

Q

ffi

X

z

O

tr

c

P

Wi

:z,

■5 =

<

Leng

of

humei

TJI

;z;

«

§

H

JD

02

<

H

g (m'

H

O O

in CO

03

-a

C CO
O; in

(M

to •

ri; "^

gffi
.gz
go

a; bo

rt ^

0) 01
03 03

O O

H E-H

^ ^

u o

U fcj O

177

178 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The shell of Toxochelys shows a high intraspecific variability, a matter of
importance if isolated finds are to be properly identified. The wide range of
variation of the shell of T. moorevillensis (pi. 23) probably exists also in other
species of which only one individual is known at present. Additional specimens
that might be found in the Greensand of New Jersey or the Marlbrook Marl of
Arkansas may thus be expected to differ notably from the type specimens of the
respective species.

Presently available materials show clearly that the plastra of the different
species of Toxochelys are very similar but can be separated on the basis of differ-
ences in proportion, as indicated in Table 9.

The vertebral column, the girdles and limb-bones of Toxochelys were dis-
cussed at length above. Since these are unknown or poorly known in other
genera of the family, valid generic comparisons are not yet possible.

Table 9.— PLASTRAL INDICES IN DIFFERENT SPECIES OF TOXOCHELYS

Species A

Toxochelys moorevillensis, C.N.H.M. PR136 . . . +75.0

Toxochelys moorevillensis, C.N.H.M. P27330 ... 73 .

Toxochelys moorevillensis, C.N.H.M. P27348. . . 69.7

Toxochelys weeksi, v. T.K20 64.8 17.7 21.9 16.8 65.4 87.3

Toxochelys barberi, C.N. E..M.P27047 67.7 16.5 22.0 16.5 64.2 102.4

Toxochelys latiremis, K.U. (V.P.) 1244 (juv.) ... 55.0

Toxochelys latiremis, Y.P.M. 3602 59.2

A = length of hyo-hypoplastral suture

B = width of umbilical fontanelle

C = hypoplastron at narrowest point

D = width of lateral fontanelle

E = length of xiphiplastron

F = lateral expansion of hyoplastron

All indices except F were calculated as follows: value XlOO, total width of hyoplastron

F was calculated as follows: value XlOO lateral expansion of hypoplastron

Toxochelys latiremis Cope

Toxochelys latiremis Cope, Proc. Acad. Nat. Sci. Phila., p. 10, 1873; Rept. U. S. Geol.
Surv. Terr., 2, pp. 98, 260, pi. 8, figs. 1, 2, 1875; Proc. Amer. Phil. Soc, 17, p. 176,
1877; Hay, Field Columbian Mus., Zool., 1, p. 101, pis. 14, 15, 1896; U. S. Geol.
Surv. Bull., 179, p. 442, 1902; Bull. Amer. Mus. Nat. Hist., 21, p. 177, 1905;
Carnegie Inst. Wash. Publ., 75, p. 168, figs. 200, 202-206, 1908; Case, Univ. Kansas
Geol. Surv., 4, p. 371, pi. 79, pi. 80, figs. 1, 2, pi. 81, figs. 1-8, 10-13, pi. 82, figs.
1-3, pi. 83, figs. 2-4, 1898; Wieland, Amer. Jour. Sci., (4), 14, p. 95, fig. 1, 1902.

Toxochelys brachyrhina Case, Univ. Kansas Geol. Surv., 4, p. 378, pi. 84, figs. 1, 2, 1898;
Hay, U. S. Geol. Surv. Bull., 179, p. 442, 1902; Bull. Amer. Mus. Nat. Hist., 21,
p. 177, 1905; Carnegie Inst. Wash. Publ., 75, p. 171, pi. 31, fig. 1, 1908.

Toxochelys serrifer Cope, Rept. U. S. Geol. Surv. Terr., 2, p. 299, 1875; Hay, U. S. Geol.
Surv. Bull., 179, p. 442, 1902; Bull. Amer. Mus. Nat. Hist., 21, p. 178, figs. 1-7,
1905; Carnegie Inst. Wash. Publ., 75, p. 170, figs. 207-213, 1908.

Cynocercus incisivus Leidy, Rept. U. S. Geol. Surv. Terr., 1, p. 269, pi. 36, figs. 17-21,
1873.

B

C

D

E

21.9

±78.3

15.4

22.1

9.6

74.5

16.5

24.3

13.4

17.7

21.9

16.8

65.4

16.5

22.0

16.5

64.2

18.7

20.6

23.7

66.9

21.4

18.2

19.5

64.9

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 179

r^pe.— A.M.N.H. 2362, jaw fragment (pi. 15) and coracoid.

Horizon and locality. — Niobrara Formation. Near fork of Smoky Hill
River, Kansas.

Referred specimens. — Two skulls (A.M.N.H. 1496 and 1497) were described
and referred to this species by Cope (1877) and were figured by Hay (1908).
Other skulls were described and figured by Hay and Case as cited. The follow-
ing, mostly unpublished, skulls and mandibles were studied and identified as
T. latiremis: Y.P.M. 3602, skull and partial shell; Y.P.M. 3609, skull and man-
dible; Y.P.M. 3604, skull and mandible; Y.P.M. 3611, skull, vertebrae, girdle-
bones, epiplastron, and limb-bones; Y.P.M. 1386, partial skull; Y.P.M. 3610,
mandible; Y.P.M. 1387, mandible; A.M.N.H. 5118, skull and mandible; A.M.
N.H. 1497, skull and mandible (figured by Hay, 1908); A.M.N.H. 11875, partial
skull; A.M.N.H. 1042, partial skull; A.M.N.H. 5119, mandible; A.M.N.H. 1498
mandible; A.M.N.H. 181, mandible; A.M.N.H. 5999, mandible; A.M.N.H
1835, skull fragments and mandible (2 individuals) (=type of T. serrifer)
U.S.N.M. 11560, skull; U.S.N.M. 11559, skull; U.S.N.M. 11639, partial skull
U.S.N.M. 1496, partial skull and mandible; U.S.N.M. 11558, partial skull
U.S.N.M. 12011, mandible; U.S.N.M. 12010, mandible; K.U. (V.P.) 1212, skull
and mandible (=type of T. brachyrhina); K.U. (V.P.) 8352, mandible; K.U.
(V.P.) 1210, mandible; K.U. (V.P.) 8256, mandible; C.N.H.M. PR125, frag-
mentary skull (juv.); C.N.H.M. UR9, fragmentary skull (juv.); C.N.H.M. UR4,
mandible, vertebrae (juv.); C.N.H.M. UR3, mandible; C.N.H.M. PR124,
mandible.

The shell of this species is represented by only three specimens, the most
important of which is Y.P.M. 3602, because of its association vvath a good skull.
The other shells are: C.N.H.M. PR123, partial carapace and plastron, girdle-
bones, limb-bones, and vertebrae; K.U. (V.P.) 1244, partial carapace, good
plastron, anterior part of skull, vertebrae, girdle-bones, and limb-bones (juv.).

Questionably referred specimens: Y.P.M. 3605, nuchal plate, first peripheral,
epiplastron, humeri; K.U. (V.P.) 1291, isolated neural plate; U.S.N.M. 12007,
epiplastron, first peripheral, other shell fragments, humerus, scapula; C.N.H.M.
UR93, poorly preserved assemblage of miscellaneous bones on a slab.

Amended diagnosis. — Skull similar to that of T. moorevillensis, but mandible
without sagittal ridge on triturating surface. Shell thin, circular in outline,
with wide lateral fontanelles extending from the nuchal plate probably to the
suprapygal area. Width of lateral fontanelles much greater than that of adjoining
peripherals. Antero-lateral and postero-lateral peripherals not greatly different
in width. Nuchal plate with small, pronounced ventral boss. Second and third
vertebral shields much wider than long. Plastron considerably reduced medially
and laterally (see Table 9).

Discussion. — Considering the nimiber of specimens, the abundance of isolat-
ed skulls and mandibles is difficult to understand. It is true that the shell of
T. latiremis is thin and delicate, but certainly no more so than the skull. While

180 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

I am not familiar with the circumstances of burial and preservation in the
Niobrara deposits, I am hesitant to accept the above proportion of skulls to
shells as the ratio at which these skeletal parts actually occur in the deposit.
The suspicion arises that a bias in collecting was involved, since partial shells,
broken into countless pieces, may well have been considered worthless specimens,
whereas skulls, even when broken, were collected.

There is, of course, no way of knowing how many of the skulls were accom-
panied by elements of the post-cranial skeleton in the field. What evidence there
is in this and other species from the Niobrara Chalk seems to indicate that skulls
and shells usually occur dissociated, as is the rule in most other formations.

Skull and mandible. — At first sight, the skull material listed above presents
a picture of apparently great variety. Not only are there broad and narrow
skulls with pointed or blunt snouts, but there are enough qualitative differences
between them to permit distinguishing each skull from all the others (pis. 10-13).
Almost all materials show obvious signs of post-mortem distortion, particularly
dorso-ventral crushing to various degrees of intensity, as evidenced by numerous
breaks through the bones. This form of distortion produces differences in pro-
portion, and, because it is plainly visible, causes few difficulties. But there are
cases where the dorsal part of a skull, for example, is badly fractured and crushed,
whereas the palate shows few and very minor fissures (pi. 12). Such apparently
unchanged areas demand the exercise of caution. The differences between skulls
in such areas may be real individual variations, but there is rarely proof of this,
since bone, under formational pressure, may be altered in shape without external
evidence of such change. It is thus impossible to separate entirely true morpho-
logical features from those brought about by the hazards of preservation. It is
with the utmost caution that similarities and differences are treated in the fol-
lowing account.

The skull of T. latiremis is, on the whole, delicately constructed. It is
much longer than wide, possibly of the general proportions of that of Chelonia.
In their present condition, the width of all skulls is greatly exaggerated. It is
even more difficult to arrive at a conclusion as to the original height of the skull
of this species. Earlier students thought it had been much like that of Chelydra,
but this view underestimates the severity of the flattening in the posterior half
of the skull. The snout region may have been of the general relative height of
that of Chelydra, but the posterior half was higher, perhaps as high as in Recent
sea turtles. The cranial roof extends backward to approximately the junction of
the pro-otic and opisthotic bones. It is thus farther excavated than in Chelonia
or Caretta and much less so than in Chelydra. The orbits were slightly longer
than high and faced outward and upward, to judge from the relative narrowness
of the interorbital bridge. The shape of the external nasal opening was a char-
acter used to distinguish the species (Hay, 1908), but this is certainly not a
reliable feature, since it is always injured by compression. It appears safe to
state, however, that the opening was wider than high in life. In front of the

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 181

nasal opening, the premaxilla forms a triangular, flattened area, with the apex
of the triangle facing the tip of the snout and the base forming a transversal
crest that delimits the nasal opening anteriorly. The apex of this triangular
area is separated from the alveolar edge of the premaxillae by a very short
distance, about 2 mm. in the largest skulls. The typical pattern of the dorsal
skull bones is illustrated in figure 60. Distinct, small nasal bones are definitely
determinable in U.S.N.M. 1496.

The ventral aspect shows a number of differences between the skulls that
may, possibly, be interpreted as individual variations. In Y.P.M. 3604, the
anterior vomer area is concave and laterally confluent with the triturating sur-
faces of the maxillae. In A.M.N.H. 5118, the vomer area is separated from the
triturating surfaces by a pronounced, sharp crest that continues backward along
the palato-maxillary suture. Intermediate conditions with either blunt crests
or sharp crests not reaching forward to the vomer area are seen in Y.P.M. 3602,
8609, and others. In Y.P.M. 3602, the ventral surface of the pterygoids is sub-
divided into three areas by two curved, longitudinal ridges that follow approxi-
mately the lateral edges of the pterygoids. In most of the specimens, these
ridges are less pronounced and in Y.P.M. 3604 they are but faintly visible. It
may also be noted that there is considerable variation in the size of the palatine
vacuities.

The mandibles show many recognizable differences, but, again, distortion
may in part be responsible. Most mandibles are somewhat flattened in dorso-
ventral direction, resulting in similar over-all preservation but in differences in
proportion between the masticatory shelf and the chin-shelf at the symphysis.
The length of the chin-shelf appears to be nearly always exaggerated. In the
apparently undistorted mandible associated with the type specimen of T. serrifer
(A.M.N.H. 1835), the chin-shelf extends downward and not as far backward
as in most specimens. C.N.H.M. UR4, of about the same size as A.M.N.H.
1835, is so severely crushed that the chin-shelf lies almost at the level of the
triturating surface and mostly behind it. Notable differences in the antero-
posterior curvature of the mandibular rami may also be due to differences in
preservation. In the better-preserved specimens, the mandible is rather acutely
pointed. There is a faint sagittal ridge on the triturating surface. The suture
pattern of the mandible is illustrated in figure 60.

Vertebral coluymi. — Save for the atlas, the cervical region of the vertebral
column is documented by a complete series of relatively well-preserved vertebrae
in PR123 (pi. 20, A). As has been noted in the general discussion above, the
cervical region in Toxochelys latiremis is shorter than in Chelonia and Chelydra
(see fig. 62). The centra of vertebrae 2, 3, 4, and 5 are approximately circular
in cross section (disregarding the large ventral keel projection) ; those of numbers
6 and 7 are flat and wide and the centrum of number 8 is wide anteriorly, more
or less circular behind. The pattern of central articulation is

(2(, (3(, (4), )5), )6), )7j, JS).

182 - FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The double facet between centra 7 and 8 may be an individual variant, rather
than the typical condition of this species (see general discussion). The neura-
pophyses are connected by well-developed pre- and postzygapophyses and a
spinal process occurs only on the second vertebra. Neurapophyses 2-5 have
pronounced, paired muscle pits on the dorso-medial sides of the postzyga-
pophyses; neither in Ckebjdra nor in Chelonia are there pits of equal depth and
sharpness of definition, and on the second vertebra such pits are absent. The
postzygapophyses of vertebrae 6-8 are sharply crested dorsally. In vertebrae
6 and 7, these crests converge anteriorly and meet near the anterior ends of the
neurapophyses. In the eighth cervical, the crests ascend to form the surface
resting against the ventral knob of the nuchal plate. The transverse processes
are similar to those of Chelonia, and much more pronounced than in Chelydra.
Associated with PR123 is the first shell vertebra, the centrum of which is very
strongly keeled ventrally. The spinal process of its neui'apophysis appears to
have been free from the carapace; it is much expanded at its end. The centrum
of the third or fourth shell vertebra is without ventral keel. The three sacral
vertebrae, with their ribs in place, are preserved in Y.P.M. 3602. The neura-
pophyses show clearly that this region was not fused with the shell. The sacral
ribs increase in size from the first to the third and they are distally in firm con-
tact.

Carapace and plastron. — Entire shells of this species are not yet known and
our knowledge of the carapace, in particular, still leaves much to be desired.
The available materials suggest that the carapace is flat and is circular or broadly
oval in outline. Continuous lateral fontanelles extend from the nuchal to the
suprapygal or even the pygal. The fontanelles are about as wide as the lateral
dimensions of the adjoining costal plates. The peripherals are notably longer
than wide. There is a pair of large post-nuchal fontanelles.

The nuchal plate (fig. 73) is wider than long and moderately excavated along
its cervical border. In C.N.H.M. PR123 (fig. 72), it is straight in front, but this
is due to crushing. On the ventral side of the nuchal, there is a conspicuous
boss, no doubt homologous to the similar but much larger protuberance in
cheloniid, dermochelyid, and protostegid turtles. In Y.P.M. 3605 and C.N.H.M.
UR93, tentatively referred to this species, the nuchal appears to have a different
shape. Y.P.M. 3605 was figured by Wieland (1905, p. 337). Examination of
the specimen left some doubt as to the outline of the postero-lateral edges of the
plate; these may have been trimmed somewhat during preparation. C.N.H.M.
UR93 has a very severely crushed nuchal that resembles Y.P.M. 3605 in some
respects, but its postero-lateral and lateral edges are not satisfactorily preserved.
Severe crushing may account for the apparent differences in shape between
these nuchals and those in other specimens, but there is one difference that cannot
be explained by post-mortem changes, namely, the size of the ventral boss.
In C.N.H.M. UR93, it is very much smaller than in C.N.H.M. PR123.

There is no complete neural series in any specimen. The preserved elements
indicate that the series was essentially as in T. moorevillensis. The first* costal

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION
Toxochelys latiremis cnhm pri23

Cp/p/3s/ra

183

Fig. 72. Outline drawing of some important parts of specimen of Toxochelys latiremis
(C.N.H.M. PR123).

plate is much wider in antero-posterior direction than the following costals
(fig. 73). The lower suprapygal and the pygal are unknown.

In the posterior part of the carapace, the width of the peripherals amounts to
about 60-70 per cent of their length in Y.P.M. 3602. In the juvenile K.U. (V.P.)
1244, as expected, the peripherals are much narrower, the width being about
30-40 per cent of their length. The marginal edges of the peripherals are faintly
notched at the shield furrows. The vertebral shields are much wider than long,
more so than in T. atlantica and in' notable contrast to those of T. moorevillensis
and T. barberi, where the third vertebral shield, for instance, is as long as wide.

The plastron of T. latiremis is better known than the carapace. In K.U.
(V.P.) 1244, the plastron is preserved largely in situ (fig. 73 and pi. 21) and is
complete except for the entoplastron. The juvenile proportions of the plastral

184 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

parts differ notably from those of Y.P.M. 3602. The central and lateral fonta-
nelles are larger in this species than in any of the others and the plates are thin
(see Table 9). Furthermore, the hyo- and hypoplastra are shorter across the
narrowest points and the xiphiplastra are more slender than in the other species
of the genus. The epiplastra (fig. 72) have an enlarged anterior end with a sharp
process on the dorsal side. Behind this point, the elements are constricted. A
thin, bony flange faces the entoplastron (fig. 72). In K.U. (V.P.) 1244, this
flange had probably not yet developed. Along the posterior half of the me-
dial border, the epiplastra adhere to the hyoplastron. The area of contact is
clearly marked on both elements by rugosities. An epiplastron associated with
the skull of Y.P.M. 3611 resembles those of C.N.H.M. PR123 described above,
except for the lack of the medial flange. Epiplastra are, furthermore, associated
with Y.P.M. 3605 (figured by Wieland, 1905) and with U.S.N.M. 12007, both
specimens doubtfully referred to this species. There is probably a good deal of
variation in detail among the epiplastra of different individuals and those
described here are probably not characteristic of this species.

In C.N.H.M. PR123, most of the entoplastron is preserved (fig. 72). It is
a thin, more or less triangular plate with a sagittal ridge on the posterior half of
the dorsal surface. This ridge becomes gradually broader and blunter and ceases
to be an elevation at the anterior, broken end of the plate. The ventral face of
the entoplastron is flat.

Shoulder girdle and limbs. — The right scapula and left coracoid (C.N.H.M.
PR123) are complete but somewhat crushed (fig. 72). The glenoidal region of
the scapula is slightly heavier than in Lepidochelys and the angle formed by the
glenoidal and coracoidal facets is but little larger than 90°, as in Chelydra;
in Lepidochelys and other cheloniids, it is in the vicinity of 150°. The shape of the
coracoid is very much like that of Lepidochelys. The proportions of the different
shoulder girdle elements are discussed on page 162.

A fair portion of the pelvis, both ilia, the right ischium, and a very incom-
plete pubis are associated with Y.P.M. 3602. I can see no differences between
these bones and their homologues in T. moorevillensis.

Besides the limb-bones described by Wieland (1902), only one other speci-
men with associated limb-bones has come to my attention. This is C.N.H.M.
PR123. Both humeri and ulnae, a proximal fragment of a radius, some carpal
bones, metacarpi and digits I and II, and other elements of II and/or IV are
preserved (fig. 70). None of the bones of the hind limbs are present. Most of
these limb-bones are well preserved and not unduly crushed. A general descrip-
tion of these bones was given in the discussion of the toxochelyid limb. In addi-
tion to the elements of the manus mentioned and figured (fig. 70) above, there
are nine partial bones of the hand, none of which can be pieced together. Their
position in the hand cannot be determined. In so far as comparison with the
forelimb of T. moorevillensis is possible, I cannot detect any differences other
than those of preservation.

p.
-a

0^

185

186 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Of the hind Hmb of T. latiremis, only the femur is definitely known (Y.P.M.
3602), and it compares, in every respect, very closely with that of T. moorevillensis.
Y.P.M. 3602 is the only specimen of the genus in which the relative length of
humerus to femur can be determined.

Toxochelys moorevillensis sp. nov.

Type. — C.N.H.M. P27330, a nearly complete shell, vertebral fragments and
girdle bones.

Horizon and locality. — Mooreville Chalk, Selma Formation, Late Creta-
ceous. Moore Brothers farm, Harrell Station area, Dallas County, Alabama.

Referred specimens. — Eutaw area, Greene County, Alabama: C.N.H.M.
PR168, costal plates; C.N.H.M. P27434, fragmentary specimen; C.N.H.M.
PR32, fragmentary specimen; C.N.H.M. P27348, partial carapace and plastron;
C.N.H.M. P27438, peripherals; C.N.H.M. P27444, peripherals.

Burkville area, Lowndes County, Alabama: C.N.H.M. PR136, nearly
complete shell, girdle bones, some limb-bones, and some vertebrae (locality:
SW. U SW. U NW. M sec. 33, T. 16 N., R. 16 E.); C.N.H.M. PR218, mandible
(locality: NE. ^ NE. }4 NE. M sec. 5, T. 15 N., R. 16 E.).

Clinton area, Greene County, Alabama: C.N.H.M. PR113, fragment.

Harrell Station area, Dallas County, Alabama: C.N.H.M. P27391, a nearly
complete carapace, fragments of plastron, vertebrae, and a good pelvis; C.N.H.M.
PRllO, fragmentary specimen; C.N.H.M. PR224, shell fragments; C.N.H.M.
PR188, fragmentary specimen; C.N.H.M. P27367, fragmentary plastron;
C.N.H.M. PR191, fragmentary plastron; C.N.H.M. PR189, fragmentary
plastron; C.N.H.M. PR199, anterior part of carapace; C.N.H.M. PR109,
fragmentary specimen; C.N.H.M. PR264, peripherals; C.N.H.M. PR265, costal
plate; C.N.H.M. P27358, peripherals; C.N.H.M. P27549, fragmentary specimen;
C.N.H.M. P27347, fragmentary plastron; C.N.H.M. P27554, fragmentary
specimen; C.N.H.M. P27562, costal plate; C.N.H.M. P27556, fragmentary spe-
cimen; C.N.H.M. P27555, fragmentary specimen; C.N.H.M. PR266, periph-
erals; C.N.H.M. P27349, fragmentary specimen; C.N.H.M. P27544, costal
plate; C.N.H.M. P27436, peripheral fragments; C.N.H.M. P27539, peripheral
fragments; C.N.H.M. P27345, costal plate; C.N.H.M. PR194, neural and costal
plates; C.N.H.M. P27550, partial skeleton; C.N.H.M. PR140, peripherals;
C.N.H.M. PR144, partial carapace; C.N.H.M. PR219, skull (good braincase);
C.N.H.M. P27338, posterior portion of skull.

Cedarville area, Hale County, Alabama: C.N.H.M. P27346, fragmentary
specimen (locality: T. 11, west of Highway 13).

West Greene area, Greene County, Alabama: C.N.H.M. PR167, nearly
complete shell and pelvis (locality: 1.6 miles north and N mile west of West
Greene); C.N.H.M. PR155, carapace fragments; C.N.H.M. PR154, fragmentary
specimen; C.N.H.M. PR253, peripherals; C.N.H.M. PR166, partial carapace.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 187

Toxochelys lotiremis ypm3602

Fig. 74. Carapace of Toxochelys latiremis (Y.P.M. 3602).
is'the skull illustrated on plate 10.

Associated with this shell

Mt. Hebron area, Greene County, Alabama: C.N.H.M. PR28, nearly com-
plete carapace and several vertebrae (locality: I9 mile north of Mt. Hebron);
C.N.H.M. PR26, peripherals.

Boligee area, Greene County Alabama: C.N.H.M. PRlll, fragmentary spe-
cimen; C.N.H.M. PR112, fragmentary specimen; C.N.H.M. PR60, costal plates;
C.N.H.M. PR33, fragmentary specimen; C.N.H.M. PR59, anterior peripherals.

Diagnosis. — Skull as far as known similar to that of T. latiremis. Mandible
with sharp sagittal crest on triturating surface, dividing the latter into right and
left halves. Tritui'ating sm-face concave on either side of ridge. Carapace
circular, with continuous lateral fontanelles of moderate size and variable extent,
usually from second to eleventh peripherals. Smallest observed fontanelles
extending from posterior end of third to anterior end of ninth peripherals. Fonta-
nelle opposite eleventh peripheral (if present) always small, not significantly
affecting suprapygal plates. Postero-lateral fontanelles always naiTower than
adjoining peripherals. Antero-lateral peripherals much narrower than postero-
lateral peripherals. Plastron relatively primitive with small timbilical and lateral
fontanelles, and long xiphiplastra (see Table 9).

188 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Table 10.— MEASUREMENTS AND INDICES IN THE PELVIS OF

TOXOCHELYS MOOREVILLENSIS

Pubis

ABC
Specimen number mm. Index mm. Index mm. Index

C.N.H.M. P27391 49.0 100 30.0 61.2 18.0 36.7

C.N.H.M. PR167 34.0 100 26.0 76.4 18.0 52.9

C.N.H.M. P27330 43.0 100 35.0 81.4 19.5 45.3

C.N.H.M. PR166 36.5 100 23.5 64.4 21.5 58.9

Ischium

C.N.H.M. P27391 49.0 100 47.0 95.5 14.0 28.5

C.N.H.M. P27550 40.0 100 11.0 27.5

Ilium

C.N.H.M. P27391 58.0 100 47.0 81.0 16.0 27.5

C.N.H.M. PR167 44.0 100 34.0 77.3 11.0 25.0

C.N.H.M. P27330 50.0 100 49.5 99.0 14.5 29.0

C.N.H.M. P27550 52.5 100 48.0 91.4 14.0 26.6

C.N.H.M. P27554 53.0 100 46.0 86.7 15.0 28.3

C.N.H.M. PR136 58.0 100 46.0 79.3 14.5 25.0

Pubis A = width of plate from middle of symphyseal edge to lateral edge.

B = width of symphyseal process of pubis at its base (narrowest point).
C = width of antero-lateral process of pubis.

Ischium A = maximal transversal extent of bone.

B = length of posterior process measured from its end to anterior edge of ischium.
C = width of ischium at narrowest point lateral to posterior process.

Ilium A = maximal dorso-ventral height of ilium.

B = length of dorsal process, measured from its posterior end to antero-dorsal face of shaft.
C = antero-posterior diameter of shaft at narrowest point (about mid-height).

Description. — Among about sixty specimens collected, only three skulls
and one mandible were secured. In view of the fact that, as a matter of
policy, all fossil remains were picked up, this ratio reflects the actual occurrence
of the skulls and shells in the field and probably in the formation. These figures
are particularly interesting if compared with the number of shells and skulls of
T. latiremis preserved in the collections (see p. 179).

The two partial skulls (C.N.H.M. P27338 and C.N.H.M. PR219) and the
mandible (C.N.H.M. PR218) were found isolated (pi. 14). Although there are
some differences between these materials and those of T. latiremis, the similarity
leaves no doubt that they belong to the Mooreville species of the genus Toxo-
chelys. The skulls are smaller than the largest skulls of T. latiremis. Specimen
P27338 corresponds in size approximately to T. latiremis Y.P.M. 3602; PR219
and the mandible are slightly larger and compare with T. latiremis Y.P.M. 3609
(pi. 14).

Unfortunately, the skulls are too incompletely preserved to permit close
comparison with T. latireynis, but in so far as comparison is possible, no significant
differences can be observed. The braincase of PR219 is excellent and is described
and compared with Chelonia and Chelydra (see p. 150 and pi. 9).

The mandible {PR218), on the other hand, is provided with a sharp ridge
along the symphyseal line of the triturating surface, the latter being thus divided

CNHM P2739I

Toxochelys moorevillensis , lOOmm

Fig. 75. Carapace of Toxochelys moore-
villensis.

Toxochelys moorevillensis

Fig 76. V\a?,tr^ oi Toxochelys moorevillensis. The plastron of P27330 is the
tjTJical of the two.

189

190 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

into right and left portions. The sagittal ridge divides posteriorly and is con-
tinued as a rounded swelling along the lingual margins of the triturating surfaces
(pi. 14). The latter are thus clearly delimited medially and are rather deeply
concave transversely. The mandible of T. moorevillensis is easily distinguishable
in all of the described features from that of T. latiremis, where there is none or
only a faint symphyseal crest, where there is no sharp medial delimitation of the
triturating surfaces, and where the latter are only slightly concave in medio-
lateral direction.

Vertebral column. — Considering the large number of specimens, only a few
vertebrae are available. The best are associated with P27391 (pi. 20). Of the
cervicals, the centrum of the seventh lacks an unmistakable double joint surface
posteriorly. The neural canal is much wider, even between the attachment
scars of the neural arches, than in the sixth centrum of Ctenochelys tenuitesta and
there is but a faint sagittal ridge at the base of the neural canal. The eighth
cervical vertebra lacks most of the dorsal extent of the neurapophysis and the
right prezygapophysis. This is a most remarkable vertebra, since it bears
perfectly discrete, short, conical ribs, the apices of which face backward (pi. 20).
The centrum is very short and broad (Table 3). A single centrum of an anterior
shell vertebra is definitely unkeeled ventrally. The few shell vertebrae directly
preceding the sacral series, the (?)second sacral, and the anterior caudal verte-
brae of this specimen have been discussed (see p. 157 and pi. 20). The only
other vertebrae of T. moorevillensis are two caudals (PR136) belonging probably
to the middle section of the tail. These vertebrae correspond rather closely with
those of C. tenuitesta, except for the apparent lack of haemapophyses. In place
of such elements, there are paired ventral processes of the centrum (pi. 20).

Carapace arid plastron. — The shell of T. moorevillensis is rather well known.
A number of specimens are complete enough to permit reconstruction of the
shells (pis. 22 and 23).

The carapace is quite fiat and almost circular in outline, at least in the
fossil condition. The bones are thin compared to the size of adult individuals;
anterior costal plates of such specimens are about 3-4 mm. thick in uncrushed
condition. In crushed condition, the thickness may be reduced to half. The
pattern of the carapace plates consists of a nuchal, nine neurals, two suprapygals,
a pygal, and on either side eight costals and eleven peripherals. Occasionally,
there is a ninth pair of costal plates (as in P27330, pi. 22) or there are nine costals
on one side and eight on the other (as in P27391, pi. 23). The relative frequency
of occurrence of this irregularity must be high in this species, since two out of
three shells (P27391, P27330, PR136) in which this area of the shell is known
show the abnormality. The eighth costal is not preserved in any other specimen,
so that the relative frequency of occurrence cannot yet be determined. A com-
parable abnormality is seen in Chelydra serpentina, where it was seen once in
eight individuals inspected. As a rule, lateral fontanelles extend from the third
(in PR28 from the first) peripheral to the eleventh, but occasionally they are
small and extend only from the fourth to the anterior half of the eighth (pis. 22

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 191

and 23), in which case costals 6 to 8 are sutui-ally attached to peripherals 8 to 11.
The degi'ee of fontanelHzation is apparently independent of size, since PR136
and P27330 are of nearly equal size. Individuals PR167 and PR28 are much
smaller than P27330 and P27391, yet their fontanelles are not much larger than
in the fully grown specimens. In juvenile individuals, however, a notable differ-
ence should be expected, since the costal disk gi'ows laterad in all turtles. There
is a considerable difference in the width between the antero-lateral and postero-
lateral peripherals (pis. 22 and 23). This is not the case in species with larger
fontanelles (e.g. T. barberi or T. latiremis). The process of fontanellization
affects not only the distal ends of the costal plates, but also the medial edges of
the peripherals. This is evident if individuals PR136 and PR167 are compared.
In the former, the ninth peripheral is as wide as long, in the latter it is much
longer than wide (pi. 23). The medial margins of the dorsal faces of the periph-
erals are notably scalloped in T. moorevillensis. The vertebral shields, par-
ticularly the second, third, and fourth, are approximately equal in length and
width, but there is a specimen (P27330, pi. 22) where they are much wider than
long, as in T. atlantica and in T. latiremis. The plastron varies in detail indi-
vidually, as does the carapace. On the whole, the posterior lobe is narrow and
sharply pointed (pi. 22), but P27348 (fig. 76) has a plastron resembling that of
T. barberi (fig. 77). The epiplastra are represented by fragments only; they are
very narrow, delicate bones of the approximate shape indicated in figure 76.
The entoplastron is absent in all specimens. Hyo- and hypoplastra have slight
ventral elevations at the corresponding places where keel bosses are present in
the lophochelyine members of the family. The general proportions of the
plastron are given in Table 9.

Girdles and limbs. — Associated with PR136 is a nearly complete left half
of the shoulder girdle with the scapula and coracoid preserved in situ. The
dorsal process of the scapula is relatively shorter than in T. latiremis (Table 5).
The coracoid is much more expanded posteriorly than in T. latiremis, where the
narrowest part of the coracoid is contained 3.6 times in the posterior width;
in T. moorevillensis (PR136), the comparable value is 4.8.

Pelvic elements are relatively well represented in the present material.
Save for the left ilium, a well-preserved pelvis (fig. 65) is associated with P27391.
Partial pelves are available in PR167, PR166 (fig. 65), PR136, P27550, P27554,
P27330 and P27349. As pointed out above (p. 161), there is a notable amount of
individual variation in the dimensions of the pubes and ilia (see Table 10). The
pubes of PR167 fit together perfectly along their symphyseal borders, indicating
an original bone-to-bone contact as in the Chelydridae, rather than a partial
separation by a cartilage sheet as in the Cheloniidae. The symphyseal border
of the pubis, gradually increasing in thickness, swings outward anteriorly where
it probably supported a prepubic cartilage.

The forelimb of this species is documented by rather meager materials.
There is not a single complete humerus preserved with any of the numerous shell
specimens, but there is at least one isolated humerus that probably belongs to

192 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

T. moorevillensis (e.g. P27403, illustrated in fig. 69). Associated with PR 136 are
the distal halves of both humeri, the left ulna, the proximal two-thirds of the
left radius and the distal ends of both radii. The foot-bones preserved with this
specimen very probably belong to the pes. A fragment of a proximal humerus is
associated with P27554. All of these elements show very little crushing and have

Toxochelys borberi

Fig. 77. Carapace and plastron of Toxochelys barberi.

good surface detail. The radius of this species may be relatively longer than in
T. latiremis, but the evidence is not entirely conclusive. The two radial frag-
ments mentioned are together about as long as the ulna, but the cross sections at
their broken ends are very different; a portion of about 4-5 mm. may be missing.

The hind limb of T. moorevillensis is better documented than the flipper.
Proximal and distal femur fragments are associated with PR136, PRllO and
P27349 (fig. 70) . The head of the femm- is displaced far to the tibial side (ventral
in the articulated position of the femur to the pelvis) and the tibial trochanter
is directly behind it (fig. 70). The fibular trochanter is relatively much more
massive than in Chelydra. The positional relationship of the trochanters to the
head is illustrated in figure 70. The distal end of the femur is very strongly
convex; the joint surface reaches from the dorsal side of the bone to the ventral
side, and there can be no doubt but that there was great movability at the knee
joint.

In PR 136, both tibiae and the right fibula are complete; the left fibula is
represented by proximal and distal fragments (fig. 70). There is a notable
similarity between these elements and their homologues in Chelydra. The fibula,
in particular, has a flaring, thin ridge at its distal, lateral edge, essentially as in
Chelydra. The distal end of the tibia bears two joint facets, separated by a

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 193

saddle-shaped depression. In Chelydra, there is an indication of such a separation,
but in Chelonia there is none. The large proximal tarsal element (a product of
the fusion of tibiale + intermedium + fibulare + centrale) is very well pre-
served (fig. 70). It belongs to the left side and has clearly marked joint facets
facing the zeugopodial bones and another for the enlarged distal tarsal 4 + 5.
The tibial joint sui'face consists of two facets that are separated by a "V"-
shaped depression. The fibular joint surface is transversely elongated (fig. 70).
The joint surfaces between this large proximal element and the distal tarsal
4 + 5 are much more clearly defined than in Chelydra and in general resemble
the condition in an old specimen of Chelonia, except that the joint permits a
somewhat greater range of movement than in the latter form.

The proximal end of metatarsus V has a rather extensive joint surface
corresponding to a complementary surface on distal tarsal 4 + 5. There is, on
metatarsus V, a small additional joint facet for at least a partial articulation
of metatarsus IV, much as in Chelydra and Chelonia. No doubt the fifth meta-
tarsal was movable, as in Chelydra, and both digits IV and V could be spread as
in fresh-water turtles (fig. 70).

There are two additional tarsal elements whose position I cannot determine
with certainty. Most of the phalanges are broken and their position in the pes
is doubtful. One of these, an entire element, is believed to be the ground phalanx
of digit V. The other elements are tentatively identified as metatarsal frag-
ments, and if this is correct, have distinctly chelydrid character.

Toxochelys barberi (Schmidt)

Phyllemys barberi Schmidt, Field Mus. Nat. Hist., Geol. Ser., 8, p. 65, figs. 20-22, 1944.

Type. — C.N.H.M. P27047, a mounted shell with part of the posterior costal
area restored.

Horizon and locality. — Marlbrook Marl, Late Cretaceous. Cox farm, one
mile northeast of junction of Hollywood-Okolona road to Arkadelphia, Clark
County, Arkansas.

Amended diagnosis. — Carapace circular, with continuous lateral fontanelles
from the second to the eleventh peripherals. Lateral areas of suprapygal plates
reduced. Postero-lateral fontanelles about as wide as the adjoining peripherals.
Degree of fontanellization intermediate between T. moor evillen sis and T. latiremis.
Vertebral shields 2 and 3 about as long as wide. Plastron similar to that of
T. weeksi, but with lateral expansion of hyoplastron greater than that of hypo-
plastron and with medial expansion of hyoplastron smaller in area than in
T. weeksi.

Discussion. — The type is the only known specimen. When it was re-exam-
ined a pair of post-nuchal fontanelles were found, though these were filled with
plaster in the mount. The peculiarities of the shell pattern at the proximal ends
of the eighth costal plates, as reconstructed, were not necessarily tliere in life;
the area is too incompletely preserved, however, to permit a definite conclusion.
At any rate, the matter is of little concern, since irregularities of this sort are

194 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

known in Recent turtles. In the plastron, the xiphiplastra, in particular, had
to be remounted.

The gi'eat similarity of this specimen to members of the genus Toxochelys,
as here understood, is unmistakable.

The carapace of this species (pi. 24) shows clearly a more advanced stage of
aquatic specialization than that of T. moorevillensis (fig. 75). The lateral fonta-
nelles are larger, particularly in the posterior part of the shell where the suprapy-
gals, reduced laterally, adjoin the fontanelles for a notable distance (fig. 77),
and where the peripherals are considerably longer than wide. In the degree of
fontanellization, T. barberi is probably intermediate between T. moorevillensis
and T. latiremis, though more material of T. barberi is necessary to determine
the amount and the trend of variation in this form. It cannot be satisfactorily
determined whether eight or nine neurals are present. As in the typical specimens
of T. ynoorevillensis (fig. 75), vertebral shields 2 and 3 are about as long as wide.

The plastron (pi. 24; fig. 77) is not as acutely pointed posteriorly as in T.
latiremis and T. moorevillensis, and the xiphiplastra are relatively wider at mid-
length than in the compared species. The lateral expansions of hyo- and hypo-
plastra are of nearly equal width, thus differing from the similar plastron of
T. weeksi, where the hypoplastron is much wider laterally than the hyoplastron.
For the general proportions of the plastron, see Table 9 and the discussion of
T. weeksi below.

Toxochelys weeksi Collins

Toxochelys weeksi Collins, Journ. Tenn. Acad. Sci., 26, (4), pp. 262-269, 2 pis., 1951.

Type. — U.T. K20, Collins collection, a beautifully preserved partial plastron
and three posterolateral peripherals.

Horizon and locality. — Coon Creek Tongue of Ripley Formation, Late Cre-
taceous. Dave Weeks place. Coon Creek, McNairy County, Tennessee.

Dr. R. Lee Collins, of the University of Tennessee, who collected the speci-
men, kindly furnished me with excellent casts of the bones.

Diagnosis.^ — Peripheral plates 8, 9, and 10 suggest a carapace with lateral
fontanelles. Plastron similar to that of T. barberi, but with relatively greater
surface area of medial expansion of hyoplastron and with lateral expansion of
hyoplastron smaller than that of hypoplastron. Umbilical fontanelle relatively
larger than in T. barberi.

Discussion. — The specimen (fig. 78) resembles the plastron of T. barberi
more closely than that of any other species of the genus. There are a number
of differences in proportion between the two forms, the most obvious of them
being the lateral expansion of the hyo- and hypoplastra (see Table 9). The sur-
face area of the medial expansion of the hyoplastron is considerably larger in T.
weeksi than in T. barberi. The following figures may illustrate some of these pro-

1 This species was described and placed in the genus Toxochelys by Dr. Colhns on my
advice. The task of diagnosing the species was left for the present revision.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

195

portional differences. If the hyoplastron of T. barberi (which is preserved in a
flattened condition but is not otherwise crushed) is shghtly bent into its original
shape, its width is the same as that of the undistorted hyoplastron of T. weeksi.
Both measure 185 mm. between the tips of the lateral and medial prongs. The

Toxochelys weeksi

Fig. 78. Plastron of Toxochelys weeksi . Drawn from cast of type specimen.

width of the hyoplastron at the naiTOwest point is 39 mm. in T. weeksi, 35 mm. in
T. barberi. The distance between the medial end of the hyo-hypoplastral suture
and the posterior end of the epiplastral suture scar (on the antero-lateral edge
of the hyoplastron) is 88 mm. in T. weeksi, but only 78 mm. in T. barberi. The
degree of fontanellization of the plastron of T. weeksi is greater than in T. barberi,
as is evident by comparison of the length of the hyo-hypoplastral suture; it is
115 mm. in T. weeksi, but 138 mm. in T. barberi. These values indicate two modes
of plastral reduction : in T. barberi, the plastral bones are reduced in size so that
the plastron as a whole covers a smaller area of the ventral surface of the animal
than in T. weeksi. In the latter, the plastral reduction manifests itself in the
large size of the umbilical fontanelle.

196 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Toxochelys atlantica sp. nov.

Lytoloma angusta Wieland, not Cope (in part), Amer. Journ. Sci., 18, p. 187, fig. 3,

pis. 6-8, 1904.
Lytoloma wielandi Hay (in part), Carnegie Inst. Wash. Publ., 75, p. 157, text figs. 196-

197, 1908.

Type.—Y.VM. 625, carapace.

Horizon and locality. — Greensand. Bai'nesboro, Gloucester County, New
Jersey.

Tentatively referred specimen. — Y.P.M. 728, fragment of left side of man-
dible.

Diagnosis. — Mandible with masticatory surface wider posteriorly than
anteriorly, as in Porthochelys laticeps. Carapace oval and greatly fontanellized.
Pygal-suprapygal connection nearly severed by right and left fontanelles.
Anteriorly, fontanelles extend to nuchal plate. Anterolateral peripherals very
narrow. Pygal longer than wide. Vertebral shields much wider than long.

Discussion. — The mandible (Y.P.M. 728) is an isolated find, provisionally
referred to this species because of its similarity to toxochelyine jaws. The speci-
men lacks most of the triturating surface near the symphysis, but the chin-
shelf shows the symphyseal suture. The fragment of this left ramus extends
back to about the posterior end of the triturating surface. The postero-lateral
face of the bone is severely injured, showing a pick mark.

This specimen, like much of the material from the Greensand of New Jersey,
is entirely uncrushed and presents excellent surface detail. The general con-
figuration of the masticatory surface relative to the chin-shelf resembles strikingly
that of PortJwchelys laticeps. As in the latter species, the triturating surface is
much broader posteriorly than anteriorly. It is concave in medio-lateral direc-
tion, except near the symphysis, and its medial boundary is clearly delimited as
in Porthochehjs and in T. moorevillensis. In the anterior half, the medial masti-
catory ridge is smooth, not rough as in Porthochelys, and in the posterior half it
is sharp. Seen from the medial side, the inner masticatory ridge is peculiarly
flat and rather sharply delimited against the mandibular sulcus (sulcus carti-
laginis meckelii) (pi. 15, figs. 1 and 2).

To date, there is no evidence from bones of the shell that Porthochelys occurs
anywhere but in the Niobrara beds of Kansas, and it would be unwise to postulate
its occurrence in New Jersey on the basis of this fragment, since increase in the
width of the masticatory surface is a feature of turtle organization that has
evolved independently many times in the history of the group. Until proved
otherwise, it would seem much more reasonable to assume that T. atlantica
possessed a skull and jaws specialized similarly to those of Porthochelys laticeps.

The carapace is oval and characterized by a degree of fontanellization inter-
mediate between T. barberi and T. latiremis. The lower suprapygal is posteriorly
reduced to such an extent that its sutural contact with the pygal is nearly broken.
Anteriorly, the fontanelles probably reached the nuchal plate; the nuchal and
the anterior peripherals are not known. The pygal plate, in contrast to all other

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

197

species of the genus, is longer than wide. Peripherals 4 and 5 are very narrow
compared to the postero-lateral peripherals, much more so than in T. latiremis
or T. barberi.

The vertebral shields (fig. 79) are considerably wider than long, but less so
than in T. latiremis. The plastron is unknown.

Fig. 79. Carapace of Toxochelys allantica.
After illustrations by Wieland.

YPM625

Toxochelys otlantico ■ 'OO'"'"

ipmx

V e o I r

Toxochelys brown!

Fig. 80. Side view of snout region of skull
of Toxochelys browni. mx, maxilla; pmx,
premaxilla.

Toxochelys browni (Hay)

Toxochelys latireviis Wagner, Kansas Univ. Quart., 7A, p. 201, fig. 1, 1898.
Porthochelys browni Hay, Bull. Amer. Mus. Nat. Hist., 21, p. 183, figs. 15-16, 1905;
Carnegie Inst. Wash. Publ., 75, p. 182, figs. 235-237, 1908.

Ti/pe.— A.M.N.H. 6080, large skull and mandible.

Horizon and locality. — Pierre Shale. Twenty miles southeast of Edgemont,
South Dakota.

Referred specimens. — A.M.N.H. 5843, large skull and attached mandible;
A.M.N.H. 1773, snout of medium-sized individual; A.M.N.H. 1774, small skull
and mandible; A.M.N.H., unnumbered mandible; C.N.H.M. PR214, snout of
medium-sized individual; C.N.H.M. PR213, skull and mandible of medium-
sized individual; S.D.S.M. 425, large skull and mandible; S.D.S.M. 4610, large
skull; K.U. (V.P.) 1221, partial skull; K.U. (V.P.) 1222, partial mandible.

Amended diagnosis. — Skull similar to that in T. latiremis, but premaxillary
area more pointed and set off from maxillary region. Species considerably
larger than T. latiremis. Shell unknown.

198 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Discussion. — Nearly all of the material is badly preserved and not well
suited for detailed description and comparison. Disregarding the size of the
largest skulls, there is little ground for separating this form from T. latiremis.
However, it may well be that the poor state of preservation obscured actual
differences between the two forms. There are at least indications to this effect.
Several of the skulls, including the type, seem to have a more slender premaxillary
region, more or less distinctly set off from the maxillary region. This is seen in
the large as well as in the small skulls, particularly along the alveolar ridge,
where the latter appears slightly and broadly notched in the area directly behind
the premaxillary-maxillary suture (pi. 10; fig. 80). This feature is most con-
spicuous in S.D.S.M. 4610, less so in S.D.S.M. 425. The squamosal has no
forward extension as figured by Hay (1908, fig. 235, right side); as nearly as can
be determined, it is exactly as in T. latiremis. The ventral aspect of the type skull
appears to be devoid of ridges along the medial margins of the triturating surfaces
and on the pterygoids, possibly because of bad preservation, since the bone
surface is generally injured by adhering minerals. The mandible is indistinguish-
able from that of T. latiremis, except for the size range.

There can be no doubt that this form belongs to Toxochelys rather than to
Porthochelys, and its closest known relative is T. latiremis. The fact that the size
range greatly exceeds that of T. latiremis, together with indications of other
differences, would seem to justify the retention of T. browni as a species distinct
from T. latiremis.

Measurements: Length from Tip of Snout to Occipital Condyle

T. latiremis
T. browni mm. (largest skull) mm.

A.M.N.H. 6080 ±160 A.M.N.H. 5118 120

A.M.N.H. 5843 ±160

S.D.S.M. 425 174

S.D.S.M. 4610 173

C.N.H.M. PR213 ± 95

A.M.N.H. 1774 ± 75

Toxochelys sp.

The United States National Museum preserves a fragmentary turtle speci-
men (U.S.N.M. 11797) that was collected by C. H. Kinzey in 1928, near Farmers-
ville, Collin County, Texas, from the Taylor Marl, probably near the upper
contact of this formation.

The specimen consists of a nuchal plate, three neurals, six partial costals,
the first and third right peripherals, a mid-lateral peripheral, some fragments
of the plastron, a proximal half of a femur, and some girdle fragments.

These materials unquestionably belong to a species of Toxochelys, but there
is not quite enough evidence to determine whether it belongs to one of the
described species or represents a new one. Under these circumstances, it would
seem reasonable to refrain from making a decision in either direction at this time.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 199

Toxochelys sp.

USNM 11797

I 50 mm I

Fig. 81. Toxochelys sp. U.S.N.M. 11797, from Taylor Marl of Texas.

The nuchal plate resembles closely that of T. barber i (fig. 81). The anterior
furrow of the first vertebral shield traverses the plate just posterior to its lateral
corners. On the ventral side, there is a slight medial elevation as in T. moore-
villensis, in place of a distinct knob as developed in T. latiremis. On the right
side of the plate, the sharp edge bordering the post-nuchal fontanelle is preserved
(fig. 81). The lateral fontanelles reach forward almost to the nuchal plate. The
inner margin of the first peripheral is rather acute and without suture, except for
a distance of about 5 mm. adjoining the nuchal. The degree of fontanellization
anteriorly is thus greater than in T. barberi but not as great as in T. latiremis.
The right third peripheral is of typical Toxochelys type, with the rib-pit at the
extreme anterior end and with a very short ventral face. The neural plates are
very slightly arched from side to side as is usual in Toxochelys, and the suture
line between adjoining neui-als is cui'ved as in all species of Toxochelys. The
second vertebral shield is about as wide as long.^ The plastral fragments and
that of the femur are clearly toxochelyid in form.

Thinochelys gen. nov.

Characterization. — Skull unknown. Carapace ovate, widest at level of
second costal plates, tapering toward pygal. Costo-peripheral fontanelles very

' This statement depends upon the correctness of my interpretation of the costal plates
with regard to their position in the carapace.

200 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

small, discontinuous. No post-nuchal fontanelles. Adjoining neurals meet with
sharply rounded sutures. Mid-dorsal neural plates about half as wide as long.
Greatest width of fourth neui'al contained 21 times in width of carapace (measured
over curvature). Postero-lateral peripherals longer than wide. Upper supra-
pygal much narrower than lower suprapygal. Posterior width of second vertebral
shield half its length. Medial sulci of marginal shields visible on all peripheral
plates. Posterior edge of pygal plate notched. Plastron with small central and
lateral fontanelles. Width of hypoplastron at narrowest point about the same
as corresponding dimension in hyoplastron. Medial edges of hyo- and hypo-
plastra only slightly serrated. Shortest axillo-inguinal distance half the length
of hyo-hypoplastral suture (48 per cent). Greatest width of xiphiplastron in
anterior half of element.

Type of genus. — T. lapisossea.

Horizon and locality. — Mooreville Chalk, Selma Formation. Alabama.

Thinochelys lapisossea sp. nov.

Type. — C.N.H.M. P27453, well-preserved, nearly complete carapace, partial
plastron and some vertebrae and girdle fragments.

Horizon and locality. — Mooreville Chalk, Selma Formation, Late Cretaceous.
Harrell Station area, one mile east of Harrell Station, south of railroad tracks,
Dallas County, Alabama.

Referred specimens. — Harrell Station Area, Dallas County, Alabama:
C.N.H.M. PR27332, a few peripheral bones and distal half of humerus; C.N.H.M.
PR263, fragmentary carapace of large individual; C.N.H.M. PR201, both
xiphiplastra.

Diagnosis. — As for the genus.

Description. — The bones of all three of the large specimens of this species
have a bluish, opaline appearance and appear to be relatively heavier than those
of most other turtles from the Mooreville Chalk of Alabama. Microscopic
examination revealed that the bones are impregnated with calcite, as is the rule
in bones from the Mooreville Chalk, and the bone cavities are entirely uncrushed
in the microscopic section. This may be the reason why these bones are relatively
heavy for their size.

Thinochehjs lapisossea is a large, flat-neuraled toxochelyid. It has a heart-
shaped carapacial outline and is rather highly arched. The degree of fontanelli-
zation of the carapace is nearly identical with that of Porthochelys laticeps; only
small, irregularly shaped fontanelles exist between costal and peripheral plates
from the second to the eighth costal. Between the fontanelles, the costal and
peripheral plates are suturally united (pi. 25 and fig. 82).

The shortness of the nuchal plate in the photograph (pi. 25) is due to the
curvature of the shell; actually, it is very similar in shape and proportion to that
of Toxochelys moorevillensis. As in the latter form, the cervical edge of the nuchal
is moderately excavated, more so than in figure 82. On the ventral side, the

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION
Thinochelys lapisossea

201

Fig. 82. Carapace and plastron of Thinochelys lapisossea.

nuchal shows a sagittal crest in its posterior half and there may or may not have
been an anchor knob for the neural spine of the eighth cervical vertebra. The
neural plates, nine in number as in Toxochelys, are narrow and elongated. The
upper suprapygal is much smaller than the lower suprapygal. In Thinochelys and
Porthochelys, the pygal is conspicuously notched at the shield furrow. The
anterior two peripherals are very narrow. In peripherals 3 to 8 the dorsal face
is very much longer than the ventral face. The bridge portion of the plastron
was loosely connected to peripherals 4 to 8. In the type specimen, there is a
pit, anterior to the rib-pit, in the eighth peripheral, which received the posterior-
most lateral prong of the hypoplastron. A similar pit for the foremost lateral
prong of the hyoplastron could not be found, but the width of the bridge area
of the plastron suggests that it was near the anterior end of the fourth peripheral.
In P27453, there are no smaller pits along the ventral edges of the bridge periph-
erals corresponding to the smaller lateral prongs of the plastron, but such pits
are present in the larger PR263. The posterior peripherals are much longer
than wide, in contrast to those in Porthochelys, which are much wider than long.

The scale pattern of the carapace is typically toxochelyid. The nuchal
scale is short and wide, much as in the genus Toxochelys. The vertebral shields
2, 3, and 4 are much longer than wide. The inner shield furrows of all the
marginal shields are visible on the peripheral plates, as in Porthochelys.

The shape of the plastron, except for the epi- and entoplastron, can be
determined satisfactorily. Hyo- and hypoplastra are united by a long suture, so
that neither the central nor the lateral f ontanelles are as large as in Toxochelys or

202 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Porthochelys. Compared to the last-named genus, the axillo-inguinal distance
is relatively shorter, and the xiphiplastra are shorter and less pointed (fig. 82).
The shield pattern cannot be determined.

A few posterior shell vertebrae and an anterior caudal are present in P27453.
They are poorly preserved, however. Fragments of both shoulder girdle and
pelvis are associated with the type specimen. The glenoidal ends of the right
scapula and coracoid are intact. It is possible that the scapular contribution
to the surface of the glenoidal cavity compared to that of the coracoid is a little
gi'eater in this species than in Toxochelys. The right ilium is attached to the
acetabular ends of the pubis and ischium. It is very similar to that of Toxochelys
moorevillensis. The posterior process of the ischium is exactly like that of
T. moor evUlen sis.

Seen as a whole, there are a number of similarities in the shells of ThinocJielys
and Porthochelys and there are likewise differences. The two forms correspond
with each other in the similar relative degree of fontanellization of the carapace,
similar elongated shape of the anterior neural plates and the second vertebral
shield, in the position of the medial shield furrows of the marginal scutes on the
peripherals, and in the presence of a marked notch in the pygal at the scale sulcus.

The two forms differ notably in the general shape and curvature of the shells,
in the proportions of the peripherals, in the relative sizes of the two suprapygal
plates, and in the degree of specialization of the plastron.

It is difficult to evaluate this situation taxonomically. I have chosen to
regard the Alabama species as representing a genus different from Porthochelys
rather than belonging to it, because differences between species of the same
genus in other toxochelyid turtles are of a much lower order of magnitude than
they are between the forms under discussion. This is also true among species
of the same genus in Recent turtles.

Porthochelys Williston

Characterization. — Skull very broad, massive. Masticatory surfaces broader
posteriorly than anteriorly. Carapace circular. Costo-peripheral fontanelles
small, discontinuous; none lateral to the first costal plates. No post-nuchal
fontanelles. Adjoining neurals meet with straight transverse sutures. Neurals
long and narrow, but adjoining elements seem to differ greatly in length. Great-
est width of fourth neural (estimated) contained about 18 times in width of the
carapace. Postero-lateral peripherals wider than long. Suprapygals of about
equal width. Posterior width of second vertebral shield about half its length.
Medial sulci of marginal shields visible on all peripheral plates. Posterior edge
of pygal plate notched. Plastron with moderate central and lateral fontanelles.
Width of hypoplastron at narrowest point slightly smaller than corresponding
dimension of hyoplastron. Medial edges of hyo- and hypoplastra serrated.
Shortest axillo-inguinal distance about 90 per cent of length of hyo-hypoplastral
suture. Greatest width of xiphiplastron in anterior half of element.

203

204 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Type of genus. — P. laticeps Williston.

Horizon and locality. — Niobrara Formation. Kansas.

Porthochelys laticeps Williston

Portkochelys laticeps Williston, Trans. Kansas Acad. Sci., 17B, p. 195, pis. 19-22, 1901;
Hay, Bull. Amer. Mus. Nat. Hist., 21, p. 183, 1905; Carnegie Inst. Wash. Publ., 75,
p. 180, pi. 31, figs. 2-3, text figs. 231-234, 1908.

Type. — K.U. (V.P.) 1204, shell, skull, mandible, and humerus.

Horizon amd locality. — Niobrara Formation, Late Cretaceous. Kansas.

Amended diagnosis. — As for the genus.

Discussion. — Other than the type specimen, no additional material was dis-
covered in the various collections that I studied. The genus has not been identi-
fied from any other formation and is absent in the Mooreville Chalk of Alabama,
the fauna of which is otherwise similar to that of the Niobrara. All evidence
indicates that P. laticeps is a rare find. The form may not have lived in the
Niobrara sea, but rather in the adjacent coastal fresh water, and it may have
become buried only occasionally in the chalk.

The skull is relatively short and broad and of heavy construction. It is
dorso-ventrally crushed, however, and its width is somewhat exaggerated (fig.
83). The masticatory surfaces increase in width from front to back and their
inner margins are distinctly serrated. The choanal and palatine openings are
very large. The small nasal bones are easily visible in this specimen. The frontal
bones are connected with the prefrontals and parietals by means of deeply inter-
fingering sutures (fig. 83). The mandible is blunt in front. There is a slight
sagittal ridge on the masticatory surface. The inner border of the latter is
rugose anteriorly, smooth in the posterior two-thirds. In over-all appearance,
the mandible resembles very much an isolated fragment, here tentatively referred
to Toxochelys atlantica (pi. 15, figs. 1 and 2). There is no way of telling whether
the latter was blunt or pointed in front; if my interpretation of a convergence
in this case is correct, we may, on the basis of morphotypic predictability, assume
that it was pointed. There are minor differences between these jaws. The
triturating surface of T. atlantica is more deeply concave than that of Porthochelys
laticeps and its inner margin is smooth in front rather than rough as in P. laticeps.
There might well be other differences that could only be noted in side by side
comparison of the specimens. While the general similarity is striking, it is not
greater than that of the jaws of Osteopygis to those in the Recent genus Caretta.

The carapace is circular in outline, in preserved condition as wide as it is
long. The costal disk is fianked by an unusually wide series of peripherals that
are suturally connected with the costal plates, save for small fontanelles as in
Thinochelys. In Porthochelys there is, however, no fontanelle lateral to the first
costal. The anterior neural plates and the vertebral shields are similar in pro-
portions to those of Thinochelys. Both suprapygals are equally wide. The
medial shield furrows of the marginal scutes lie on the peripherals throughout,
and the pygal is notched at the shield furrow. The peripherals show a slightly

ZANGERL: VERTEBRATE FAUXA OF SELMA FORMATION

205

sen-ated marginal edge. The seirations are located behind the shield fuiTOws,
not in front of them as in the Lophochehinae. Nothing comparable is seen in
Thinochelijs.

The plastron has a relatively short hyo-hypoplastral suture and therefore
rather large umbilical and lateral fontanelles. The xiphiplastra are long and
relatively more pointed than in Thiyiochelys (fig. 84).

The humerus (Williston. 1901) should be difficult to distinguish from that
of Toxoclielys.

Subfamily Osteopyginae

Characterization. — Toxochelyid tmlles with relatively generalized shells, and
skulls provided with extensive secondary palates. SjTnphysis of mandible
probably more than one-third the length of the ramus. Probably no nasal bones.
Carapace oval, wnth or without small lateral fontanelles. Neuralia unkeeled.
Peripheral edge of carapace not, or only slightly, sen-ated. A deep pit in the
postero-medial face of the second peripheral plate for the reception of the antero-
lateral prong of the hyoplastron.

Osteopygis, (?)Rhetechelys

Discussion. — The type of specialization revealed by the fragmentary- skulls
and jaws in the Osteopyginae is very similar to that in the Cheloniidae, par-
ticularly the Recent genus Caretta, and some of the Early Cenozoic sea turtles of
western Europe. If it were not for the definite association of one of the lower
jaws with shell material of Osteopygis (A.M.N.H. 2216), there would be no
reason for doubting the cheloniid affinities of these cranial materials. Accord-
ingly, earlier students have placed various lots of these bones in different genera
and families. Comparison of all available specimens reveals a moderately wide
range of individual variation in the relative length of the mandibular sjmiphysis,
as well as some qualitative differences, such as a slight, blunt elevation on the
triturating surface along the sjTiiphysis, or lack of such, as follows:

A.M.X.H. 2216

Y.P.M. 913

Y.P.M.602

Y.P.M. 1001

Y.P.M. 490

Y.P.M. 3613a

Y.P.M. 3613b

Erquelinnesia molaria, tjTJe.

Width

Angle

between

between

mental

Index

Length of

canales

Elevation of

foramina

symphysis

alveolares

symphysis

mm.

mm.

87

72

±63

55°

none

61

70.5

±43

50°

slight

68

76

52 (est.)

58°

slight

74

78

58 (est.)

54°

slight

75

73

55

62°

medium

53

79

41 (est.)

53°

very slight
pronounced

74

65

+ 57'

206 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The differences noted are quite insignificant, if viewed in the light of the
unquestionably great over-all similarity of the specimens (pi. 16), and I propose,
therefore, to include all of the skull and jaw materials listed above in the genus
Osteopygis. The skull has a relatively broad and short snout region and an almost
fiat palate, thus differing from that of RhetecJielys, in which the snout is some-
what elongated and there is a pit behind the alveolar ridge of the premaxillae,
presumably for the reception of an upturned mandibular beak.

The shell of Osteopygis is thick and spongy; the carapacial outline is oval
and the degree of fontanellization varies greatly with the individual. Specimens
without fontanelles are known. The lateral expansion of the hyoplastron reached
forward to the second peripheral. The plastral shields are essentially as in
Toxochelys.

The shoulder girdle is not satisfactorily known. The pelvis was described
and illustrated by Wieland (1904a). Examination of these bones reveals rather
too great an amount of modeling clay used in the reconstruction of the bones.
Better material is necessary to determine the accurate shape of the pelvis.

Several humeri are known. The best one is associated with the type speci-
men of Osteopygis gibbi (Y.P.M. 783), and there is, with it, a left ulna. These
elements are unmistakably toxochelyid in character, and such minor differences
as seem to exist may be merely differences in preservation. This conclusion is
based on the comparison of the humerus (Y.P.M. 778) with the humeral fragment
of Toxochelys moorevillensis (P27554). These two specimens are entirely un-
crushed and have good surface detail. As nearly as I can determine, the corre-
sponding areas are practically identical. Comparison of the ulna of Y.P.M. 783
with that of Toxochelys latiremis (PR123) or T. moorevillensis (PR136) and
Chelydra reveals that it is more chelydrid than in Toxochelys. In both genera,
there is a sharp ridge along the outer edge of the ulna. Along the opposite edge,
facing the radius, there are pronounced tuberosities proximally and distally
in Osteopygis (fig. 90), indications of which are seen in Chelydra. In Toxochelys,
there is, instead, a continuous sharp ridge.

The femur (Y.P.M. 783) is very much like that of Toxochelys (see Wieland,
1904a, fig. 6; Hay, 1908, fig. 146).

Osteopygis Cope

Characterization. — Skull with broadly rounded snout, no deep pit behind
premaxillary beak. Mandible with fairly flat triturating surface. Carapace
oval, moderately arched and consisting of thick plates. Nuchal plate more
than half as long as wide. First peripheral not encroaching upon nuchal from
the side. Postero-lateral marginals wide. Upper suprapygal much wider than
lower one. Costo-peripheral fontanelles absent (rare) or small, but continuous
at least laterally. Medial borders of marginal shields lie on peripheral plates.
Second peripheral with conspicuous medial pit for reception of anteriormost prong
of hyoplastron. Pygal plate long and relatively narrow. Plastron with central

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

207

and lateral fontanelles. Hyo-, hypo- and xiphiplastra join sagittally in open
suture. Xiphiplastra broad, stout. Epiplastra very small and thin. Axillo-
inguinal distance about one-half the width of hyoplastron.

Tijpe of genus. — 0. emarginatus Cope.

Horizon and locality. — Greensand. New Jersey.

AMNH22I6

con. mentalis

for. mentale
con olveoloris
foss masseteriCQ
ifor. con. olveoloris

Osteopygis emarginatus

Fig. 85. Internal anatomy of mandible of Osteopygis emarginatus. Drawn from
X-ray plate.

Discussion. — Up to the time of Hay's revision of the fossil turtles of North
America (1908), six species referable to the genus Osteopygis had been described,
to which list Hay added another and Leidy's doubtful Chelone sopita. The latter
is based on unsatisfactory type material, consisting of four peripherals. One of
these is identified by Hay ( 1908, fig. 181) as the seventh of the right side. This
bone is far too long for its width to belong to the genus Osteopygis and has the
proportions of some of the cheloniid turtles. There is, furthermore, some doubt
as to the association of the four bones. In view of these circumstances, it would
seem justifiable to consider Chelone sopita a nomen vanum.

Thus, there remain seven species of Osteoptjgis, all of them from the Green-
sand deposits of New Jersey. With the exception of the shell Y.P.M. 783,
described as 0. gibbi by Wieland (1904a), all specimens are very fragmentary.
A survey of nearly all Osteopijgis materials in the major collections reveals that
there are only a few more specimens than described species and these cannot
definitely be identified with any of the seven species. The truth of the matter
is that no two specimens agi'ee closely with each other if one uses the criteria
that have been proposed to distinguish the numerous forms of Hay's revision.
These criteria are: presence or absence of lateral fontanelles in the carapace;
presence or absence of notches above and below the rib-pits in the peripherals;
thickness of the shell; emargination or lack of such in the posterior peripherals;
flattened or round rib-pits; proportions.

208 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

According to Hay (1908, p. 128), the eight species are supposed to fall into
two groups, namely, with and without lateral fontanelles in the carapace. In
the first group, lacking fontanelles, he placed 0. emarginatus, 0. gibbi, and 0.
robustus; yet the fifth peripheral of the type specimen of 0. emarginatus (A.M.N.H.
1485) shows no sign of a medial suture and certainly had lateral fontanelles
(fig. 87). 0. gibbi is distinguished from 0. robustus, among other features, by
peripherals notched by the rib-pits in the former and lack of such notches in
the latter. Re-examination of the type material of 0. robustus reveals that
peripheral 10 is unnotched but peripheral 9 is notched as in 0. gibbi (fig. 86).

Nearly all characters used to distinguish these species are related to aquatic
specialization and are highly variable not only in this genus but quite generally.
The degree of fontanellization, for instance, varies greatly in Toxochelys moore-
villensis (pi. 23), and some of the shells have notched peripherals, while others
do not.

After careful examination of nearly all available material, I have come to
the conclusion that there is only one species represented. The observable differ-
ences are individual variations, correlated, for the most part, with marine
adaptation. Some may be age differences.

Osteopygis emarginatus Cope

Osteopygis emarginatus Cope, Proc. Acad. Nat. Sci. Phila., p. 147, 1868 (nomen nudum);
Cook's Geol. N. J., p. 735, 1868 (1869) (nomen nudum); Amer. Nat., 3, p. 89,
1869; Trans. Amer. Phil. Soc, 14, (n.s.), pp. 135-136, 235, pi. 7, fig. 3, 1870 (1871);
Kept. U. S. Geol. Surv. Terr., 2, p. 259, 1875; Hay, U. S. Geol. Surv. Bull., 179,
p. 441, 1902; Carnegie Inst. Wash. Publ., 75, p. 129, figs. 134-141, 1908.

Osteopygis sopita Cope, Proc. Acad. Nat. Sci. Phila., p. 147, 1868; Rept. U. S. Geol.
Surv. Terr., 2, p. 258, 1875; Hay, U. S. Geol. Surv. Bull., 179, p. 441, 1902.

Osteopygis borealis Hay, Carnegie Inst. Wash. Publ., 75, p. 141, pi. 25, fig. 3, text figs.
163-171, 1908.

Osteopygis platylomus Cope, Amer. Nat., 3, p. 89, 1869; Cook's Geol. N. J., p. 735,
1868 (1869); Trans. Amer. Phil. Soc, 14, (n.s.), pp. 135, 137, figs. 38, 39, 1870
(1871); Rept. U. S. Geol. Surv. Terr., 2, p. 258, 1875; Hay, U. S. Geol. Surv.
Bull., 179, p. 441, 1902; Carnegie Inst. Wash. Publ., 75, p. 146, figs. 172-180, 1908.

Osteopygis gibbi Wieland, Amer. Jour. Sci., (4), 17, p. 118, pis. 5-8, text figs. 3-7, 1904;

Hay, Carnegie Inst. Wash. Publ., 75, p. 132, pi. 26, fig. 1, pi. 27, figs. 1, 2, text figs.

142-146, 1908.

Osteopygis robustus Hay, Carnegie Inst. Wash. Publ., 75, p. 134, figs. 147-151, 1908.

Osteopygis erosws Cope, Rept. U. S. Geol. Surv. Terr., 2, p. 258, 1875; Hay, U. S. Geol.
Surv. Bull., 179, p. 441, 1902; Carnegie Inst. Wash. Publ., 75, p. 138, pi. 26,
fig. 2, text figs. 155-162, 1908.

Osteopygis chelijdrinus Cope, Proc. Acad. Nat. Sci. Phila., p. 147, 1868 (nomen nudum);
Cook's Geol. N. J., appendix, p. 735, 1869 (nomen nudum); Amer. Nat., 3, p. 89,
1869; Trans. Amer. Phil. Soc, 14, (n.s.), pp. 135, 138, pi. 7, fig. 8, 1870 (1871);
Hay, U. S. Geol. Surv. Bull., 179, p. 441, 1902; Carnegie Inst. Wash. Publ., 75,
p. 136, pi. 23, figs. 4-7, pi. 28, figs. 1-4, text figs. 152-154, 1908.

Catapleura chelydrina Cope, Rept. U. S. Geol. Surv. Terr., 2, p. 259, 1875.

ZAXGERL: VERTEBRATE FAUNA OF SELMA FORMATION

209

Propleura sopita Cope, Cook's Geol. N. J., p. 735, 1868 (1869); Amer. Nat., 3, p. 88,
1869; Trans. Amer. Phil. Soc, 14, (n.s.), pp. 140, 235, pi. 7, figs. 4-7, text fig. 39,
1870 (1871).

Osteopygis emorginatus

//f/j ipecimen a mixed
//) AMAIM 1132

Fig. 86. Osteopygis emarginatus. Annotated sketches made from originals.

Propleura borealis Wieland, Amer. Jour. Sci., (4), 17, p. 129, pi. 9, 1904; Amer. Jour.

Sci., (4), 18, p. 190, fig. 4, 1904.
L/ytoloma angusta Cope, Trans. Amer. Phil. Soc, 14, (n.s.), p. 145, pi. 11, figs. 1, la,

1870 (1871) (part); Hay, Carnegie Inst. Wash. Publ., 75, p. 155, pi. 28, fig. 5,

1908 (part); Wieland, Amer. Jour. Sci., (4), 18, p. 183, figs. 1, 2, 1904.
Lytoloyna icielandi Hay, Carnegie Inst. Wash. Publ., 75, p. 157, pi. 28, figs. 7, 8, pi.

29, fig. 1, 1908 (in part).
Erquelinnesia molaria Hay, Carnegie Inst. Wash. Publ., 75, p. 160, figs. 198, 199, 1908.

d=i:

210

Osteopygis emarginatus

epiplastrai pit

indication o/
Suinuc/}al knoi

\

Fig. 88. Osteopygis emarginatus. Annotated sketches made from originals.

211

212 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Type. — A.M.N.H. 1485, type of Osteopijgis emarginalm (fig. 87).

Horizon and locality. — Greensand. Barnesboro, New Jersey.

Referred specimens. — A.N.S.P. 14676, type of 0. plahjlomus, Pemberton,
New Jersey (fig. 86); Y.P.M. 778, type of 0. borealis, Hornerstown, New
Jersey (pi. 26); A.M.N.H. 1132, partial rim of carapace, partial plastron, which
may not belong to the same individual, Hornerstown, New Jersey (fig. 88);
A.M.N.H. 1132, another specimen with the same number (fig. 86); A.M.N.H.
2351, fragmentary anterior portion of carapace, Birmingham, New Jersey
(fig. 88); A.M.N.H. 2360, type of 0. robustus, Birmingham, New Jersey (fig. 86);
A.M.N.H. 1130, type of 0. erosus, Barnesboro, New Jersey (fig. 87) ; Y.P.M.
783, type of 0. gibbi, Barnesboro, New Jersey (fig. 89); A.M.N.H. 1131, type of
0. chelydriuus, Barnesboro, New Jersey; A.M.N.H. 2216, jaw (pi. 16, fig. 1, b)
and parts of shell, New Jersey; Y.P.M. 1001, jaw, Hornerstown, New Jersey
(pi. 16, fig. 1, d); Y.P.M. 490, jaw, Hornerstown, New Jersey (pi. 16, fig. 1, c);
Y.P.M. 602, jaw, Birmingham, New Jersey; Y.P.M. 3613, several jaw fragments,
Hornerstown, New Jersey; Y.P.M. 913, jaw, Hornerstown, New Jersey (pi. 16,
fig. 1, a) ; Y.P.M. 913a, anterior region of skull, Hornerstown, New Jersey (fig. 61) ;
Y.P.M. 734, fragmentary specimen, New Jersey; Y.P.M. 991, fragmentary speci-
men, New Jersey; A.M.N.H. 2870, seven peripherals. New Jersey; Y.P.M. 1585,
nuchal, anterior peripheral and fragments, Mullica Hill, New Jersey; A.N.S.P.
9220, jaw, type of Erquelinnesia malaria, Birmingham, New Jersey; A.M.N.H.
1133, fragment of jaw, Birmingham, New Jersey.

The present procedure of including all materials listed above in one species
resulted from the realization that the fragmentary specimens do not permit the
recognition of more than one form. But the possibility of the existence of more
than one species in all of the Greensand deposits is undeniable. Additional mate-
rial, more carefully collected than in the past, is required to decide this matter.

Amended diagnosis. — As for the genus.

Discussion. — Y.P.M. 913a presents the snout region of the only known skull
referable to Osteopygis (fig. 61). It has been described and illustrated by Wieland
(1904a, fig. 2) and by Hay (1908, pi. 28, figs. 7, 8, pi. 29, fig. 1). The palatal
face of this fragment is remarkable because of the unusual relative flatness of the
triturating surface, outwardly bordered by a very low alveolar ridge. As in the
Recent Caretta and in contrast to Chelonia, the triturating surface does not
form sharp, masticatory ridges; instead, it has a low, broadly rounded elevation
that runs in a blunt arc from one side of the triturating surface to the other,
approximately parallel to the alveolar border, thus differing in detail also from
Caretta. Posteriorly, the triturating shelves of the maxillae are greatly expanded,
almost excluding the palatine bones from the formation of the secondary palate.
The ventral plate of the vomer is wider than the combined width of the pre-
maxillae and forms a broad sutural contact with them. The vomerine pillar,
separating the narial passages, is stout and extends to the posterior end of the
ventral plate (thus to the rim of the choanal opening), differing markedly from
Caretta, where the pillar ends a notable distance farther forward. The dorsal.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION
Osteopygis emarginatus

213

Fig. 89. Carapace and plastron oiOsteopygis emarginatus (Y.P.M. 783). After Wieland.

expanded portion of the vomer shows a pronounced sagittal sulcus that runs
backward to the broken edge of the bone. A similar longitudinal groove is absent
in Caretta caretta. Furthermore, the descending processes of the prefrontals join
the dorsal plate of the vomer in a posteriorly extended, horizontal suture. In
C. caretta, this suture area is smaller and does not reach as far backward.

On the whole, this skull fragment of Osteoptjgis resembles the corresponding
area of the skull of Caretta, but, in detail, there are numerous differences through-
out, which should be expected if my interpretation of a convergence with the
cheloniid turtles is correct.

In so far as the surface bone is well preserved, the mandibles (pi. 16) are
of fairly uniform appearance. In most specimens, the tritui'ating surface is
divided along the symphysis into two slightly concave areas by an obtuse eleva-
tion. The latter is more pronounced in front, where it forms a blunt ridge.
Posteriorly, it is merely a gradual increase in thickness of the triturating shelf
toward the symphysis. In A.M.N.H. 2216 and in A.N.S.P. 9220, the triturating
surfaces are nearly flat. Compared to Toxochelys, the medial face of the masti-
catory portion of the mandibular ramus is greatly shortened but it is provided
with a deep sulcus (fig. 85, sulcus cartilaginis meckelii) and a sagittal pit that

214 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

extends forward along the symphysis. Presumably, the meckehan cartilages
entered into this pit for an undetermined distance. Directly below the most
posterior point of the triturating surface in each ramus the canalis alveolaris
opens into the meckelian sulcus (fig. 85). On the lateral side of the ramus, at
the foremost point of the masseteric fossa, there is a large foramen mentale.
X-ray plates of all these jaws reveal that the canalis alveolaris and the canalis
mentalis meet below the center of the triturating surface in each ramus (fig. 85),
and a single canal of decreasing caliber runs forward to a point some distance
behind the tip of the jaw. The angle formed by the alveolar canals varies from
50° to 62° and the observed range of the length-width index of the triturating
surface is 70.5 to 79 (see p. 205) among the seven specimens at hand. On the
ventral face of some of the jaws, the posterior extent of the horny beak is marked
by a shallow impression. It begins laterally directly in front of the foramen
mentale, arches slightly forward and then backward and reaches the posterior
end of the sjonphysis.

Only one complete shell of Osteopygis is known (Y.P.M. 783), the remaining
specimens being rather fragmentary (see figs. 86-89). Since seven species have
been based on these materials, it was felt that new illustrations — based on my
own drawings of the specimens along with pertinent comments — should be ren-
dered for future consideration of the species question.

The carapace of Osteopygis is oval and rather highly arched. The thickness
of the shell plates differs, of course, in different areas of the shell, as usual, but,
on the whole, the plates are thick (see comments on drawings). The degree of
fontanellization varies from none (Y.P.M. 783) to fontanelles laterally in the
vicinity of peripheral 5 but not extending backward beyond peripheral 7
(A.M.N.H. 1485), to continuous fontanellization from opposite the third to
opposite the tenth peripheral. This is not actually observable in any one speci-
men, but A.M.N.H. 2870, A.M.N.H. 1130, and Y.P.M. 778 make the supposi-
tion probable. Peripherals 1, 2, and 11 are suturally connected with the costal
disk in all specimens. The situation observed in Porthochelys and Thinochelys
is not found among these specimens, but it should probably be expected to
occur. The peripherals of Osteopygis indicate clearly that the plastron was in
relatively intimate contact with the carapace. There is a deep pit on the visceral
side of the second peripheral, which received the large antero-lateral prong of
the hyoplastron (pi. 26). A similar pit for the reception of the postero-lateral
prong of the hypoplastron is located on the visceral side of the seventh peripheral
(A.M.N.H. 1132), or it may be so deep as to lie across the suture between the
seventh and eighth peripherals (A.M.N.H. 1130). Small pits along the ventro-
medial edges of some of the lateral peripherals, corresponding to the smaller
hyo- and hypoplastral prongs, are not seen in all specimens, reflecting differences
in the degree of ossification of the plastra. The medial furrows of the marginal
scutes lie on the peripherals, though very close to their inner edges. In speci-
mens with notable fontanelles, they overlie the vacuities. In the postero-lateral
peripherals, the ribs may enter regular pits dorsally and ventrally covered by

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

215

thin sheets of bone. In some cases, however, these pits may be more or less
exposed above or below or both (A.M.N.H. 1132). This feature is subject to
variation. Similar conditions are seen less frequently in Toxochelys moorevillensis.
The upper suprapygal is much larger than the lower suprapygal, the reverse from
what it is in Thinochelys. The pygal is longer than wide anteriorly.

Fig. 90. Humerus and left ulna of Osteopijgis emarginatus (Y.P.M. 783).

The plastron is characterized by wide lateral expansion of hyo- and hypo-
plastra. The suture between these plates is very long. The umbilical fontanelle
is more or less circular and there are notable lateral fontanelles. The pattern
of the plastral shields is indicated in figui-e 89; it is essentially like that of Toxo-
chelys (see T. weeksi, fig. 78). It is virtually certain that there was a series of
inframarginal scales, but these shields rarely leave imprints on plastra that are
partially reduced laterally (for example, in Recent Chelydridae, Dermatemyidae,
and Cheloniidae). The shield pattern at the anterior end of the plastron is
unknown. The limited girdle and limb material is discussed on page 206.

Incertae sedis

Rhetechelys platyops (Cope)

Euclastes platyops Cope, Proc. Acad. Xat. Sci. Phila., p. 41, 1867; Cook's Geo!. N. J.,
p. 735, 1868 (1869); Amer. Nat., 3, p. 89, 1869; Trans. Amer. Phil. Soc, 14,

216 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

(n.s.), p. 149, pis. 6, 7, fig. 9, 1870 (1871); Kept. U. S. Geol. Surv. Terr., 2, p. 259,
1875.

Lytoloma platijops Hay, U. S. Geol. Surv. Bull., 179, p. 442, 1902; Wieland, Amer.
Jour. Sci., (4), 18, p. 185, 1904.

Rhetechelys platyops Hay, Carnegie Inst. Wash. Publ., 75, p. 162, pi. 29, figs. 2, 3, 1908.

Type. — A.N.S.P. 10187, a large, partially damaged skull.

Horizon and locality. — Late Cretaceous. Hurffsville, Camden County, New
Jersey,

Discussion. — Hay (1902) proposed the generic name Rhetechelys for this
form and placed it near Erquelinnesia. Although I have seen the specimen,
side by side comparison with the snout fragment of Osteopygis (Y.P.M. 913a)
was not possible. In so far as the published illustrations permit comparison,
the differences between the specimens are as follows: There is a deep pit behind
the premaxillary beak in Rhetechelys that suggests a sharp upturned point on the
mandible, a feature not present in Osteopygis and not observed in typical develop-
ment in any toxochelyid turtle. The snout area of Rhetechelys, furthermore, is
distinctly more slender and more pointed than in Osteopygis. Rhetechelys is a
very large form. None of the mandibles of Osteopygis are large enough to fit
this animal. The length of the symphysis in Caretta caretta amounts to about
80 per cent of the sagittal extent of the secondary palate. The symphysis of
the largest mandible of Osteopygis (A.N.S.P. 9220) measures only 68 per cent
of the palatal length of Rhetechelys. The greater part of the braincase and
the squamosal region are missing, so that it is, at present, impossible to allocate
Rhetechelys to, or reject it from, the Osteopyginae with reasonable confidence.

Subfamily Lophochelyinae nov.

Characterization. — Specialized toxochelyid turtles. Skull with a technically
primary palate, but somewhat specialized in the direction of an undershelving
of the internal nasal openings. Triturating surfaces wider than in the Toxo-
chelyinae. Anterior vomer area narrow and rough. Small nasal bones. Upper
and lower shelves of mandible at symphysis of nearly equal width. Shell con-
sisting of thin plates. Carapace oval to cordiform in outline. Neuralia serrated,
forming mid-dorsal keel. Epineural elements in Ctenochelys and Prionochelys.
Peripheral edge of carapace moderately to strongly serrated. Costo-peripheral
fontanelles always large, extending forward at least to the second peripheral
plate. Connection between pygal and suprapygal very narrow or entirely
broken. A pair of post-nuchal fontanelles, at least in young forms. Plastron
with large central and lateral fontanelles. Hyo-, hypo- and xiphiplastra with
conspicuous ventral keel knobs.

Lophochelys, Ctenochelys, Prionochelys

Discussion. — Useful skull material is known in the genera Ctenochelys and
Prionochelys. An adequate characterization of the morphology of the skull

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 217

within the subfamily is, however, not yet possible. The skull is more robust
than that of Toxochelys and readily distinguished from the latter by the wide
triturating sui'faces that encroach significantly upon the lateral edges of the
palatines and the vomer. Thus, the development of a secondary palate is here
initiated. The basisphenoid (fig. 100) is much as in Chelonia (pi. 9), but the
course of the carotid canal is more nearly like that of Chelydra.

The carapace is characteristically possessed of a serrated mid-dorsal keel
and a serrated marginal edge. These two features are strictly correlated with
each other and with the partial undershelving of the palate in the family Toxo-
chelyidae. No form is known that has only one or two of these characters.
The dorsal keel may be formed simply by crested, thecal neural plates (as in
Lopkochelys) , or it may contain epithecal ossicles that produce conspicuous keel
elevations beneath the apices of the vertebral shields (as in Ctenochelys) , or,
finally, modified thecal neural plates, in conjunction with large epithecal ele-
ments, may form large keel elevations (as in Prion ochehjs). Lateral fontanelles
are always present and large, even in adult individuals. Furthermore, there are
post-nuchal fontanelles, except in large individuals of Prionochelys nauta, where
such fontanelles are absent (but probably will be found in smaller specimens)
and where the lateral fontanelles are relatively small, though continuous.

The plastron is not as characteristic as the carapace. Compared to the
Toxochelyinae, it exhibits larger central and lateral fontanelles in the adult.

Because of the unsatisfactory state of preservation of most of the material,
it is not yet possible to work out the basic morphology of the girdle and limb
skeleton within the subfamily.

Lophochelys gen. nov.

Characterization. — Most primitive genus of subfamily. Skull unknown.
Neui'al carina lacking epithecal plates, except possibly in most posterior part of
carapace. Peripheral serration of carapace moderate. No connection between
pygal and suprapygal in young; adult condition unknown. A pair of post-nuchal
fontanelles.

Tijj)e of genus. — Lophochelys natatrix.

Horizon and locality. — Niobrara Formation, Late Cretaceous. Kansas.

Discussion. — The genus Lophochelys is represented by three species, L.
natatrix and L. niobrarae from the Niobrara Chalk of Kansas, and L. venatrix
from the Mooreville Chalk of Alabama. The skull is not known at present. The
carapace has a mid-dorsal carina, serrated in side view, with the apices corre-
sponding to the apices of the vertebral shields. No epithecal ossicles are involved
in the construction of the carina, except possibly in the posterior area of the shell.
The evidence is subject to some doubt, and the observed "suture scars" might be
mere erosion surfaces. The marginal edge of the carapace is scalloped, as in
Ctenochelys stenopora and C. tenuitesta. There was apparently no bony connec-
tion between the pygal and the upper suprapygal plates. The anterior face of
the pygal and the posterior end of the suprapygal show indications that the

218 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

connection was by an especially tough strand of connective tissue in the juvenile;
the adult condition is unknown. This region of the shell is different, however,
from that in Ctenochelys, as a comparison of juvenile individuals of both genera
reveals (figs. 92, 108). In Ctenochelys, a bony lower suprapygal is indicated by
an anterior suture projection on the pygal.

The plastron resembles rather closely that of Ctenochelys.

The Niobrara species are based on juvenile specimens, which makes com-
parison with the subadult and adult materials from the Mooreville Chalk very
difficult. The most obvious features that appear to separate them are differ-
ences in the degree of ossification of the shell, and these are definitely correlated
with size and age.

Lophochelys natatrix sp. nov.

Type. — C.N.H.M. PR220, partial carapace and plastron, scapulae, coracoid,
limb-bones, vertebrae. Juvenile. Collected by Mr. G. F. Sternberg.

Horizon and locality. — Niobrara Formation, Late Cretaceous. One mile
north of the Pyramids, Logan County, Kansas.

Referred specimen. — Y.P.M. 3606, peripherals, neurals. Locality?, Niobrara
Formation, Late Cretaceous.

Diagnosis. — Carapace oval, pointed posteriorly. Nuchal plate less than
three times as wide as long; slightly excavated in front. No preneural. First
pair of costal plates embraces only anterior half of first neural. Neural plates
hexagonal, with antero-lateral and postero-lateral sides approximately equal in
length. Pygal plate at mid-line half as long as wide. Peripherals of normal
width. Plastron with pronounced keel bosses.

Description. — Both specimens referred to this species are juveniles. The
type specimen contains enough skeletal material to permit an accurate char-
acterization of the shell (figs. 91, 92). The carapace is oblong, moderately
pointed behind, gently arched and gi-eatly fontanellized. The latter is certainly
a feature of age. The edge of the carapace is moderately serrated, much as in
Ctenochelys tenuitesta and C. stenopora. The relatively long and narrow peripheral
bones are those of immature individuals, and proportionately wider elements
are to be expected in adults. The nuchal plate is slightly excavated in front and
notably arched from side to side. The neui^al series consists of hexagonal plates
whose antero-lateral and postero-lateral sides are of nearly equal length. The
first neural is sharply crested and forms a keel apex anterior to the transversal
shield furrow (fig. 91). A small, posterior neural, still suturally united to frag-
ments of the adjoining elements, has a longitudinal scar along its ridge line. It
may have borne an epithecal ossicle, or the scar may be an erosion surface. The
matter cannot be decided. The first pair of costal plates embraces only the
anterior half of the elongated first neural, namely, back to the level of its greatest
width. The costo-nuchal sutures slant forward and outward. The nuchal shield
is very short; its posterior furrow lies very close to the cervical margin of the
nuchal plate (figs. 91, 92).

o

219

220 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

In the type specimen and in Y.P.M. 3606, tlie pygal is of normal shape.
The anterior face is entirely smooth. If PR220 and Y.P.M. 3606 are compared
with the likewise juvenile type specimen of C. stenopora, K.U. (V.P.) 1205 —
about the same size as PR220 — the pygal plate of the latter is seen to possess a
marked sagittal projection facing the suprapygal region (fig. 108). The lack of
a similar process in L. natatrix suggests that the pygal-suprapygal connection
was effected by connective tissue only.

The beautiful plastron of PR220 is clearly that of a young individual with
larger fontanelles than should be expected in an adult. Hyo-, hypo- and xiphi-
plastra have pronounced ventral keel bosses. An isolated plastron of this
species should be difficult to distinguish from that of C stenopora; the xiphiplastra
with their acute postero-lateral angles differ, however, considerably from those
of all known specimens of C. stenopora.

The eighth cervical vertebra has a short centrum and a deep, single con-
cavity in front. The neurapophysis has a well-developed dorsal platform that
indicates the original presence of a ventral knob on the nuchal, which shows
merely a lesion in the surface bone where the knob has been.

The scapula, coracoid and femur are shown in figure 91. The distal expan-
sion of the coracoid is exaggerated by crushing.

Lophochelys niobrarae sp. nov.

Type. — C.N.H.M. URl, partial carapace. Juvenile.

Horizon and locality. — Niobrara Formation, Late Cretaceous. Kansas.

Referred specimens. — C.N.H.M. UR2, plastron of an individual the size of
URl; horizon and locality as for URl. C.N.H.M. UR7, partial plastron, pygal
plate, ilium; horizon and locality as for URl.

Diagnosis. — Carapace oval, not pointed posteriorly. Nuchal plate about
three times as wide as long; slightly excavated in front. No preneural. First
pair of costals embrace first neural entirely. First neural without keel ridge or
keel apex anterior to transverse shield furrow. Anterior neural plates much
longer than wide, hexagonal with very short antero-lateral and long postero-
lateral sides. Pygal plate at mid-line only one-fifth its width. Peripherals
probably very narrow. Plastron with indistinct keel bosses.

Description. — It is, indeed, unfortunate that there is no way of determining
satisfactorily whether UR2 is the plastron of URl. The type of preservation,
lack of crushing and perfect surface texture, light beige color of the bones and
the size strongly suggest it. The available records indicate that both specimens
were collected by Harold T. Martin and purchased in 1894, together with a
small number of other fragmentary turtle remains, among them UR7. The latter
is differently preserved and severely flattened; the siirface texture is poor, the
color is dark brown and the specimen is more than twice the size of URl and UR2.

The differences in size and preservation of UR2 and UR7 render comparison
between corresponding parts all but impossible. Additional collections will be

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

221

Lophochelys notatrix

Fig. 92.
plastron.

Lophochelys natatrix (C.N.H.M. PR220). Reconstruction of carapace and

required to detemiine whether the reference of UR2 to this species is correct.
If so, it probably belongs to the type carapace, URl.

The type, URl, is an unusually well-preserved carapace of a very juvenile
individual of about 175 mm. carapacial length. It lacks all of the peripherals,
but the short and wide pygal plate (fig. 94) is present. The shape of this bone is
that expected of an early ontogenetic stage of ossification, gi'owing, with age,
to the shape of any typically toxochelyid pygal. This interpretation would
undoubtedly be in order were it not for the fact that the pygal of UR7, a much
larger individual, shows no change in proportion in the expected direction.
It is very short and wide, as in URl. The pygal of C.N.H.M. PR220 (type of
L. natatrix), which is about half the size of UR7, has a normally proportioned
pygal (fig. 91); the same is true of Y.P.M. 3606 (fig. 93), an even smaller indi-
vidual than PR220. In view of this, it must be assumed that the pygal plate

222

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

(and of necessity the peripherals) of L. niobrarae show a retardation or even
arrestation of growth at an early stage of ossification. The neural plates of
URl are well documented either by the bones themselves (nos. 1, 3, 4, 5, 6, 9) or
by their outline, formed by the adjoining costal plates (nos. 2, 7, 8). In contrast
to L. natatrix, the dorsal keel ridge begins behind the first neural, which is an

Lophochelys notatrix
YPM 3606

I SOitim I

Fig. 93. Peripherals and neural of Lophochelys natatrix (Y.P.M. 3606).

almost flat plate. The lateral corners of all the neurals are sharply defined, which
is not the case in L. natatrix. The antero-lateral sides of the hexagonal neurals
are very short in neurals 2 to 5, shorter than the postero-lateral sides in neurals
6 and 7, and the two are of equal length in neural 8. The ninth of the series is
elongated and deeply inserted into the anterior margin of the suprapygal (fig. 94).
The nuchal is relatively shorter and wider than inL. natatrix (fig. 94).

The beautifully preserved plastron, UR2, is characterized by thick hyo-,
hypo-, and xiphiplastra, with keel bosses that are not particularly conspicuous —
far less so than in L. natatrix (PR220). The xiphiplastra are relatively stout and
rather wide and blunt posteriorly. The plastral fragments of UR7 are so crushed
that their original thickness cannot be determined. They agree with UR2 in
having only indications of keel bosses. The shape of the posterior end of the
xiphiplastron of UR7 differs notably from that of UR2; it resembles PR220,
L. natatrix.

(y;

C3

^

J.1

O

§

00

ffi

^

:z

1^

Q

o

►-J

og

d

223

224

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Lophochelys venatrix sp. nov.

Type. — C.N.H.M. P27355, fragmentary specimen.

Horizon and locality. — Mooreville Chalk, Selma Formation, Late Cretaceous.
Moore Brothers farm, Harrell Station area, Dallas County, Alabama.

Lophochelys venatrix

CNHM P27355

<^-

preneural

Fig. 95. Type specimen of Lophochelys venatrix (C.N.H.M. P27355).

Referred specimens. — Harrell Station area, Dallas County, Alabama:
C.N.H.M. P27426, peripherals and costal plates; C.N.H.M. P27350, nuchal,
peripherals. Mt. Hebron area, Greene County, Alabama: C.N.H.M. PR22,
three peripherals.

Diagnosis. — Carapace broadly oval, rounded posteriorly. Nuchal plate
one-third as long as wide, strongly excavated in front. A preneural. First pair
of costal plates embrace preneural and anterior half of first neural. First costal
forms sutm'al connection with second peripheral. Possibly an epithecal ossicle
in suprapygal region. Lateral fontanelles much narrower than lateral extent
of adjoining costals. Nuchal scale large; its posterior furrow at about mid-
length of nuchal plate.

Among the keeled turtles of the family Toxochelyidae, from the Mooreville
Chalk of Alabama, there is a form, poorly known at present, that does not possess

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

225

f>oss/o/c/ rros/on
Sur/^ces on post
/Tifd/a/ e/fmenfs

P 7c/eM CNHM P27350

Lophochelys venotrix

Fig. 96. Specimens oi Lophochelys venatrix.

epineural elements. As in L. natatrix, there is possible evidence in this form that
an epithecal element might have been located in the posterior region of the
carapace; but here, as there, the evidence is not conclusive, since the "suture
scars" on the keel elements might also be erosions of the surface bone.

The type specimen is of medium size and the bones are crushed, though not
entirely flattened. It appears certain that the peripherals are much wider now
than in life (figs. 95, 97). In side by side comparison of the peripherals with
corresponding plates of Ctenochelys tenuitesta, there is very little difference in
shape, but those of Lophochelys venatrix seem to consist of denser bone.

In the type specimen, there is a preneural, an element not observed in
Ctenochelys tenuitesta but present in C. acris, whose peripherals are characteris-
tically different from both C. tenuitesta and L. venatrix. The nuchal plate is
represented by a fragment of the left half in P27350. It differs from that of
C. tenuitesta by forming a notable segment of the first fontanelle (fig. 97), and
from C. acris by being much less excavated in front. The inner margin of the
first peripheral shows no suture. The second peripheral is peculiar because its
inner margin appears to have been suturally joined by the first costal (fig. 95).
The evidence is fairly good. The second peripheral is crushed more than the
first or the third. It is thus wider than the adjoining plates. Yet its inner
margin, only slightly eroded, is so thick that it probably did not face a fontanelle.
This featiu'e is shown in the reconstruction (fig. 97). The carapacial disk, formed
by the costal and neural plates, was very probably wider than in Ctenochelys.

Only fragments are known of the plastron ; these are shown in their relative
position in figure 95.

226

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The first vertebral shield did not extend as far forward on the nuchal plate
as in Lophochelys natatrix. Its anterior imprint lies approximately at mid-length
of the nuchal (fig. 96). The nuchal shield was thus large.

Lophochelys venatrix

CNHM P27355

Fig. 97. Reconstruction of Lophochelys venatrix (C.N.H.M. P27355). Dotted line
indicates probable original width of peripherals. The presence of an epithecal ossicle in the
suprapygal area is not certain.

Comparison of L. venatrix withL. natatrix andL. niobrarae is difficult because
of the size difference in the specimens. The degree of lateral fontanellization is
an age character as well as a feature of aquatic specialization. The degree of
shell reduction correlated with the aquatic environment can only be determined
on fully mature individuals.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

227

Ctenochelys gen. nov.

Characterization. — Most common genus of subfamily. Skull fairly massive.
Anterior portion of vomer narrow and rugose. Triturating surfaces involving
significant portions of palatine bones. Mandible massive; triturating shelf
widest at symphysis, gradually narrowing toward posterior end. Chin-shelf not
protruding posteriorly beyond masticatory shelf. Shell consisting of very thin,
spongy plates. Neural carina with epineural elements located below apices of
vertebral shields, dorsal to junction of neurals 2 and 3, 4 and 5, 6 and 7; some-

white; c precox, CNHM 0C6I
block: C ocris, cnhm P27366

white C cf stenoporo, ypm3607
block: C tenuitesta, cnhmp27402

Fig. 98. Comparison of proportions of lower
jaw in four species of Ctenochelys.

Fig. 99. Proportional differences in snout re-
gion of skull in Ctenochelys acris and C. tenuitesta.

C ocris

CNHMP2r'337

228 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

times at junction of pygal and suprapygal. Peripheral edge of carapace mod-
erately to deeply serrated posteriorly; connection between pygal and suprapygal
very narrow in adult (condition in young unknown). A pair of post-nuchal
fontanelles.

Type of genus. — Ctenochelys tenuitesta.

Horizon and locality.— MooreYiWe Chalk, Selma Formation, Late Creta-
ceous. Alabama.

Discussion. — Ctenochelys is known both from the Mooreville Chalk of Ala-
bama and from the Niobrara Chalk of Kansas. It is the only genus of the family
in which two species can be recognized from both Mooreville Chalk (C. tenuitesta,
C. acris) and Niobrara Chalk (C. stenopora, C. proeax).

C. tenuitesta resembles C. stenopora more closely than either C. acris or C.
proeax, and the latter two forms appear to be more closely related.

The Mooreville Chalk specimens of Ctenochelys, although far from perfect
in preservation, are represented in fairly large series. The species identification
of fragmentary specimens (depending largely on which parts are preserved) may
present difficulties or may be impossible. Specimens of which a fair portion
of the skeleton is preserved are readily identified.

The Niobrara Chalk material, on the other hand, poses practically insur-
mountable difficulties. The number of available specimens is still very limited.
Many specimens consist of isolated skulls and/or mandibles, or isolated shell
remains. Different age stages are represented and comparisons between them
are almost always inconclusive. Finally, the state of preservation ranges from
lack of distortion to severe crushing and flattening of the bones. Under these
circumstances, needless to say, the revision and species allocation of the mate-
rials were complicated and largely indirect procedures, and future discoveries
will be required to determine the merit of the present conclusions.

Skull. — Of the ten specimens of cranio-mandibular remains from the Moore-
ville Chalk, eight were found isolated. The remaining two, both mandibles, are
associated with identifiable shell plates of each of the two species. The identifica-
tion of the eight isolated finds presented thus relatively little difficulty, and the
differences between the skulls and mandibles of C. tenuitesta and C. acris can be
characterized as follows: In C. acris, both nostril and orbit are relatively much
closer to the tip of the snout than in C. tenuitesta (fig. 99). The mandibles of the
latter species are more slender in the areas of the triturating surfaces than in
C. acris (fig. 98). This difference is slight, however, and not reliable except in
uncrushed specimens.

Only six specimens of Ctenochelys consisting of or including cranio-mandib-
ular elements are known to me from the Niobrara deposits. A seventh, an
isolated mandible, K.U. (V.P.) 2050, figured by Hay (1908, p. 176) as C. proeax,
is too severely injured to permit identification. Of the six specimens, two are
isolated skulls and mandibles, namely, the type of C. proeax (A.M.N.H. 234)
and C.N.H.M. UC614, a smaller skull and jaw that closely resemble the type

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 229

specimen mentioned. Hay (1908) referred this skull to C procax and I am in
full agreement with this interpretation.

The remaining four specimens belong to two size classes, juvenile and
adult or near-adult. The small individuals are the type specimen of C. steuopora
and U.S.N.M. 6013, referred to C. stenopora by Hay (1909). The larger indi-
viduals are A.M.N.H. 6137, a severely crushed skull, a mandible, and a portion
of the post-cranial skeleton; the type specimen of Toxochelys elkader Hay; and
an undistorted, isolated mandible (Y.P.M. 3607).

In an attempt to compare the last four specimens, all of the difficulties
alluded to at the beginning of this chapter descend upon the investigator. I have
finally concluded, tentatively, that A.M.N.H. 6137 is an adult, or near-adult,
individual of C. stenopora (there is collateral evidence contained in the plastron)
and that Y.P.M. 3607 represents the undistorted mandible of the same species.^
If this theory should prove to be correct, the mandibles of C. stenopora and
C. procax could be distinguished by a more slender ramus and a distinctly pointed
symphyseal tip (fig. 98) in C. stenopora and a notable symphyseal ridge on the
triturating surface of C. procax.

Comparison of the mandibles of C. procax and C. acris reveals a wider jaw
angle in the former species (fig. 98). The lower jaws of C. tenuitesta and C.
stenopora differ in the relative width of the triturating surfaces. These are
narrower in C. tenuitesta (fig. 98). The described differences are rather slight
and quite probably subject to considerable individual variation, but the available
material, unfortunately, does not permit a more satisfactory analysis.

Carapace and plastron. — The carapace is elongated and was, probably,
rather narrow in life. There is no tendency toward reduction of the peripheral
plates. Lateral fontanelles, even in mature specimens, extend continuously
from the nuchal plate to the pygal. The neural keel is pronounced and mod-
erately high. The keel elevations are regularly formed by epithecal ossicles
whose position coincides with the highest points of vertebral shields 2, 3, and 4.
The epineurals are spaced two neural plates apart. An additional epithecal
ossicle may occui* behind the fourth vertebral shield to form an anal elevation.
A preneural element may be present or absent. Circumstantial evidence sug-
gests that where the preneural is apparently absent it is fused to the first neural.
Post-nuchal fontanelles are always present. In all cases, the peripheral edge of
the carapace is serrated, but the degree and the exact outline of the serrated
edge differ individually and between species.

The plastron is difficult to distinguish from that of Lophochelys. The hyo-
hypoplastral suture, as would be expected, increases in length with age, but

' Neither the skull nor the mandible is known in the genus Lophochelys. The possibility
that Y.P.M. 3607 might belong to this genus cannot be disregarded. There is, however,
indirect evidence in support of the view here expressed. Y.P.M. 3607 closely resembles
the mandible of C. tenuitesta. Furthermore, differences in the skull and jaw of members of
different genera within this family are on a higher order of magnitude (cf. Toxochelys and
Porthochelys) than those distinguishing Y.P.M. 3607 from C. tenuitesta and the other species
of Ctenochelys.

230 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

large lateral and central fontanelles persist. The two Mooreville Chalk species
can be distinguished by the shape of the xiphiplastra.

Although a fair amount of vertebrae, girdle, and limb-bones is associated
with A.M.N.H. 6137, these elements are so severely crushed that no useful
comparisons can be made.

Ctenochelys tenuitesta sp. nov.

Type. — C.N.H.M. P27361, partial carapace and plastron, girdle elements,
limb-bones, and vertebrae.

Horizon mid locality. — Mooreville Chalk, Selma Fonnation, Late Creta-
ceous. Moore Brothers farm, Harrell Station area, Dallas County, Alabama.

Referred specimens. — Eutaw area, Greene County, Alabama: C.N.H.M.
PR30, partial skull roof, braincase, part of snout.

Harrell Station area, Dallas County, Alabama: C.N.H.M. P27354, partial
shell; C.N.H.M. PR95, peripheral bones; C.N.H.M. P27404, pygal, peripherals;
C.N.H.M. P27362, partial shell, humerus, neurals; C.N.H.M. P27557, hypo-
plastron, xiphiplastron; C.N.H.M. P27551, hyoplastron; C.N.H.M. P27357,
peripheral; C.N.H.M. PR209, peripheral; C.N.H.M. P27481, peripherals;
C.N.H.M. P27360, peripherals; C.N.H.M. P27552, shell fragments; C.N.H.M.
P27425, peripherals, ulna; C.N.H.M. P27351, partial plastron, scapulae, verte-
brae; C.N.H.M. P27429, peripherals; C.N.H.M. P27563, pelvis, peripherals,
neurals; C.N.H.M. P27548, partial carapace, pelvis; C.N.H.M. PR248, costals,
neurals, peripherals, partial plastron, humerus, vertebrae; C.N.H.M. P27560,
peripherals, neurals; C.N.H.M. P27341, mandible; C.N.H.M. P27431, pe-
ripherals, coracoid; C.N.H.M. P27402, snout fragment, parietals, mandible;
C.N.H.M. P27339, snout fragment, mandible; C.N.H.M. P27558, peripherals,
neurals, vertebrae. Specimens tentatively referred to C. tenuitesta: C.N.H.M.
P27559, shell fragments; C.N.H.M. P27432, shell fragments; C.N.H.M. PR96,
shell fragments.

Cedarville area. Hale County, Alabama: C.N.H.M. P27316, partial carapace.

West Greene area, Greene County, Alabama: C.N.H.M. PR161, posterior
rim of carapace; C.N.H.M. PR252, xiphiplastra, peripherals, humerus, scapula;
C.N.H.M. PR258, mandible, skull fragments, peripherals, partial plastron.

Mt. Hebron area, Greene County, Alabama: C.N.H.M. PR180, peripheral;
C.N.H.M. PR24, costals, peripherals, suprapygal, pygal. Specimens tentatively
referred to C. tenuitesta: C.N.H.M. PR27, peripherals; C.N.H.M. PR57, periph-
erals; C.N.H.M. PR25, poorly preserved, partial carapace.

Diagnosis. — Skull with elongated snout area; distance between anterior
rim of orbit and tip of snout greater than half the width of skull at level of
anterior orbital rim. Mandible with slightly concave lateral jaw outline. Periph-
eral edge of carapace serrated, with individual serrations rounded. Nuchal very
wide, moderately concave in front; costo-nuchal sutures nearly transverse. No
preneural. Epineural ossicles about half the length of adjoining neurals. A large

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

231

upper and small lower suprapygal. Anal elevation formed by an epithecal
ossicle overlying lower suprapygal and anterior margin of pygal. Anterior
suture of eleventh peripheral not notably oblique. Plastron much longer than
wide. Xiphiplastron longer than distance from medial corner of lateral fontanelle
to medial, serrated margin of hypoplastron. Xiphiplastron widest at level of
keel knob; area behind keel knob relatively long and pointed.

ventral view

dorsal view

Fig. 100. Fragment of braincase of Ctenochelys tenuitesta.

Description. — Five lots of skull fragments are referred to this species.
C.N.H.M. P27339 contains a snout fragment of the right side in uncrushed
preservation. The ascending maxillary pillar that separates the nostril from the
orbit and joins the prefrontal is preserved in situ and indicates the relatively
great height of the skull in this area. The nostril was certainly higher than wide
and the angle between the plane of the nostril opening and that of the palate lies
in the vicinity of 60°. Compared with the snout of C. acris, that of C. tenuitesta
is relatively narrower and more pointed. Both nostril and orbit lie relatively
farther back (fig. 99). The distance from the tip of the premaxilla to the ante-
rior rim of the orbit measures 45 mm. The antero-posterior width of the max-
illary pillar at the narrowest point is 14 mm. The vertical height of the nostril
is about 24 mm., measured along the slant about 26 mm. The height of the orbit
was approximately 41 mm. The anterior rim of the choanal opening lies 38 mm.
back of the tip of the snout. The height of the maxilla below the orbit, at the
level of the posterior delimitation of the choanal opening, is about 17 mm.,
compared to 12 mm. in specimens of about equal size of C. acris (C.N.H.M.
PR251 and P27337). The mandible of P27339 has, in the horizontal plane, a
slightly concave outer masticatory edge. The latter is narrower relative to the

232

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

symphyseal length in C. temiitesta than in C. acris (fig. 98). Specimen PR30
includes an interesting braincase fragment (fig. 100) that shows the basisphenoid
developed in the direction of that of Recent sea turtles (pi. 9), fragments of both
maxillae, and a portion of the roof of the skull with one frontal, both parietals
and both post-orbitals intact. The dorsal pattern of this section of the skull

CNHM P2736I

Fig. 101. Type specimen of Ctenochelys tenuitesta (P27361) and plastron of another
specimen (P27551).

roof agrees very well with the corresponding area in Toxochelys latiremis. Lot
P27402, belonging to a subadult individual, consists of a mandible (fig. 98), a
snout fragment and a small portion of the anterior parietal area. The mandible
of this specimen is virtually identical in form and size with that of PR258,
identified on the basis of the shell remains. An isolated, right mandibular ramus,
P27341, agrees closely with the others.

Isolated elements of the vertebral column are associated with five specimens.
In P27361 and PR248, the centra of the sixth cervical element have strongly
convex joint surfaces posteriorly. The scars to which the neurapophyses are
attached reach almost to the mid-line (pi. 20). There is a sharp sagittal ridge at
the base of the neural canal, very pronounced in PR248, less so in P27361.
Nearly complete eighth cervicals are associated with specimens P27351 and
P27548. The centra of these elements are relatively short compared to their
width at the narrowest points (see Table 3) but not as short as in T. moore-
villensis (P27391). The apex of the neurapophysis forms a fairly flat tuberosity
that rests against the ventral knob on the nuchal plate; in this as well as in the
relative shortness of the centrum, C. tenuitesta resembles Toxochelys rather than

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

233

C. acris. Centra of the ninth (or first shell) vertebra (P27361, PR248 and
P27558) are quite massive and ventrally keeled. The neural canal is narrow
in front, but widens to more than twice that size posteriorly. In so far as compari-
son is possible, this element is similar to that in Toxochelys latiremis. All of the
centra of anterior shell vertebrae are ventrally keeled. One of the two caudal

Ctenochelys tenuitesta

CNHM P27362

Fig. 102. Referred specimens of Ctenochelys tenuitesta.

vertebrae of P27351 is an anterior one; its ribs, standing at right angles to the
longitudinal axis of the element, are fused to the transverse processes. The other
caudal vertebra belongs to the posterior half of the tail and shows no peculiarities.
The shell material, although represented by a fairly large number of speci-
mens, does not include a single entire carapace or plastron. Since the shell plates
are thin, the sutural connections between the different elements were probably
weak and easily separated in the process of maceration. Even though this situa-
tion is far from ideal, it was possible to work out the basic moi'phology of the
shell and to gain some insight into the range of individual variation, which.

234 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

apparently, is gi'eat in this species. Most conspicuous is tlie variable degree of
serration on the peripheral bones (figs. 101, 102, 103).

The nuchal is a wide plate, moderately concave in front and acutely pointed
laterally. There is a long suture contact with the first costal plates and the
sutures run more or less transversely (figs. 102, 104). The first peripheral is
without inner suture, which indicates that the nuchal bordered on the lateral
fontanelles. There is no preneural. The antero-lateral corners of the first
neural form small portions of the rims of the post-nuchal fontanelles. It is a
fairly flat plate with a slight keel ridge in its posterior half. All of the following
neurals are roof-shaped in cross section. There seem to be two suprapygals, as
usual in this family, but the upper one is much larger than the lower plate. The
latter, present (in a poor state of preservation) only in P27316 (fig. 103), is
probably very narrow in uncrushed condition, to judge from the well-defined
suture areas on the upper suprapygal and the pygal of PR24. An epithecal
ossicle, forming an anal keel elevation at the junction of the suprapygal and
the pygal, may have covered the small lower suprapygal entirely. Relatively
small epineural elements are located above the contacts of neurals 2 and 3,
4 and 5, 6 and 7 (figs. 101, 104). The peripheral plates are scalloped much as in
Lophockelys. The degree of serration varies greatly from very slight (P27316,
PR24) to very pronounced (P27404, PR161), with the extremes connected by
intermediate conditions. The edge of peripherals anterior to the serration
point is usually slightly convex, in contrast to C. acris in which it is slightly
concave. The first peripheral is almost triangular, the inner margin being short.
The second peripheral is much wider anteriorly than posteriorly.

The plastron is much longer than wide (fig. 101), thus conforming with the
general shape of the carapace. Specimen P27551, an isolated partial plastron,
belongs to an unusually large individual. Its xiphiplastron shows the character-
istic shape for this species except for the postero-lateral keel knob area, which
is severely crushed and presents a definitely exaggerated, angular appearance
of the lateral margin. The xiphiplastron of the type specimen (P27361) presents
the nonnal, undistorted outline (fig. 101). The bone is widest at about mid-
length, where the keel knob is located, gradually tapering to a point posteriorly.
Slight serrations are visible along the medial edge.

In the shoulder girdle, the scapula is all but indistinguishable from that of
Toxochelys, but entirely complete elements are not known. The two scapular
processes are subequal in length, the dorsal, as usual, being the longer. In
P27404 (fig. 102), the ?ventral process is abnormally short and there is a foramen
that pierces the neck of the bone in approximately an antero-posterior direction.
A foramen of this sort was not observed in any other specimen. This bone is so
unusual that it is difficult to determine whether it is a right or a left scapula.
The coracoid shows considerable distal expansion, probably not much exaggerated
in the specimen illustrated (fig. 106). The humerus (fig. 106) lacks the distal
end. Its ulnar process is crushed toward the main shaft of the bone. The prox-
imal end of another humerus (PR248) is somewhat flattened in the plane of the

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

235

ulnar process and consequently presents quite a different appearance. Because
of these difficulties, detailed descriptions of these bones are of doubtful value.

&>

Ctenochelys tenuitesta

CNHM P273I6

Fig. 103. Referred specimen of Ctenochelys tenuitesta.

Of the pelvis, the pubis is represented in a fragmentary condition only.
The right ilium and the left ischium (PR248) are shown in fig. 106. The angle
between the axes of the vertical shaft and the postero-dorsal process of the ilium
is about 110°. A well-preserved left ilium is associated with P27425, a specimen
with unusually weak peripheral serration. The dorsal half of the shaft of the
ilium is concave on the medial side, but the surface bone is not as rough as, for
example, in the Recent Macrochelys. On the lateral side of the bone, there is a

236 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Ctenochelys tenuitesta

Fig. 104. Ctenochelys tenuitesta. Reconstruction of carapace based on all available
materials.

corresponding convexity whose surface is sculptured with many parallel ridges.
The postero-dorsal process ends with a clearly delineated, slightly concave,
smooth area. The ventral end of the bone shows clearly the ischial and pubic
attachment surfaces and the concave ilial component of the acetabular cavity
(fig. 106).

The postero-lateral process of the ischium is short and stout, less than half
the medio-lateral width of the bone (fig. 106). Only the proximal end of the

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION
Ctenochelys tenultesta

237

Fig. 105. Ctenochelys tenuitesta. Reconstruction of carapace in side view.

Ctenochelys lenuitesta

CNHM P27S63

humerus sin cnhm pz73Si

ifiunj (/ext

CNHM P27425

femur sin

CNHM P27563

Fig. 106. Girdle and limb elements of Ctenochelys tenuitesta.

femur is known (P27563). In the present state of our knowledge, it is not
distinguishable from that of other toxochelyid turtles (fig. 106).

Ctenochelys stenopora (Hay)

Toxochelys serrifer Case (non Cope), Univ. Kansas Geol. Surv., 4, p. 379, pi. 80, figs. 3-9,
pi. 82, figs. 4, 5, pi. 83, fig. 1, 1898; Hay, Amer. Nat., 32, p. 935, figs. 1-3, 1898;
U. S. Geol. Surv. Bull, 179, p. 442 (in part), 1902; Williston, Trans. Kansas Acad.
Sci., 17, p. 198, 1901.

238 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

T. stenoporus Hay, Bull. Amer. Mus. Nat. Hist., 21, p. 180, figs. 8-12, 1905.

T. stenopora Hay, Carnegie Inst. Wash. Publ, 75, p. 172, figs. 214-220, 1908; Proc.

U. S. Nat. Mus., 36, p. 191, pi. 5, text fig., 1909.
T. bauri Wieland, Amer. Jour. Sci., 20, p. 325, pi. 10, figs. 1-8, 1905; Hay, Carnegie

Inst. Wash. Publ., 75, p. 178, figs. 229, 230, 1908.
T. procax Hay, Carnegie Inst. Wash. Publ, 75, p. 176, fig. 227, 1908.
T. elkader Hay, Carnegie Inst. Wash. Publ., 75, p. 174, figs. 221 223, 1908.
Toxochelys sp. Hay, Carnegie Inst. Wash. Publ., 75, p. 164, pi. 30, 1908.

Material. — See below.

Amended diagnosis. — Skull with slender snout region. Triturating sur-
faces narrower than distance between them near posterior end. Mandible
pointed in front; symphyseal ridge on triturating surface weak or absent. Cara-
pace oval with large lateral fontanelles, equaling or exceeding in width the
lateral extent of the adjoining costal plates. Peripheral edge serrated as in
C. tenuitesta. Nuchal moderately concave in front. Costo-nuchal sutures slant-
ing forward and outward. Epineural ossicles more than half the length of
adjoining neural plates. Position of epineural ossicles apparently somewhat
variable. No epithecal anal elevation. Large upper and small lower suprapygal.
Anterior suture of eleventh peripheral not notably oblique. Plastron wider than
long; length of xiphiplastron equal to distance between medial corner of lateral
fontanelle and medial, serrated edge of hypoplastron.^ Xiphiplastron widest at
level of keel knob; area behind keel knob short.

Discussion. — The type material of this species appears to have suffered from
repeated, somewhat inadequate, curatorial care in the past. At the time of
Hay's (1908) redescription of the specimen, some of the bones that were quite
probably associated when Case (1898) rendered the first description of it bore
different collection numbers, namely, 2060 for the skull, peripherals, and hyo-
plastron, 1270 for the hypo- and xiphiplastra, neurals, suprapygal, and costal
plates. Today, a specimen bearing the number K.U. (V.P.) 1205 (fig. 108)
consists of the same posterior carapace bones that appear on Case's plate 83
(op. cit.). The similarity is such that there is no doubt that it was the specimen
studied by Case. In fact, there is no good reason to suspect that different indi-
viduals or even species are contained among these materials. The size of the
skull fragment relative to the width of the hyoplastron agrees closely with the
comparable parts of specimen U.S.N.M. 6013, a juvenile of about the same size.
While there is little doubt there is likewise no ultimate proof that the skull frag-
ments were associated with the shell bones. Since Hay (1908, p. 173) designated
especially the skull as the name-bearer of the species, the question of the associa-
tion of the type material is, in so far as this case is concerned, of interest and
importance, because the skulls are much more difficult to identify than are the
shells. The relatively great number of skulls of Toxochelys latiremis gives an

' This is a peculiar correlation that applies to both juvenile and adult individuals
(4 specimens tested).

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

239

idea of the effect of post-mortem changes that alter the appearance of skulls to
such an extent that they may appear entirely dissimilar in very many respects.
Hay (1909) described the juvenile slab specimen U.S.N.M. 6013, of nearly
the same over-all size as the type, and referred it to this species. Another speci-
men (A.M.N.H. 6137), of which only the skull and mandible were prepared at

Ctenochelys stenopora
20mm

Fig. 107. Plastron of Ctenochelys stenopora.. After Case and Hay.

the time, was made the type of Toxochelys elkader by Hay (1908). The skull,
severely compressed dorso-ventrally, differs in a number of features from the
type of C. stenopora, particularly in the illustrations rendered by Hay. The
post-cranial skeleton, on the other hand, belongs to this species without doubt.
Since there is no doubt as to the association in A.M.N.H. 6137, one might suspect
that the skull of the type material of C. stenopora does not belong with the shell.
There is, however, no doubt as to the association of skull and shell in U.S.N.M.
6013, and the skull of the latter compares quite well with that of the type. The
differences between the skulls of A.M.N.H. 6137 on the one hand and the type
skull and U.S.N.M. 6013 on the other are thus probably due to differences in
preservation and in age, since A.M.N.H. 6137 is an adult or near-adult individual.
In addition to the three specimens mentioned, the fairly complete shell,
Y.P.M. 1786, described as Toxochelys bauri (Wieland, 1905), an undescribed

240

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

juvenile skeleton, U.S.N. M. 12009, from the Niobrara Chalk, 2 miles south
of Russell Springs, Kansas, an isolated mandible, Y.P.M. 3607, a specimen,
K.U. (V.P.) 1228, described and figured by Hay (1908, p. 164, pi. 30) but not
specifically identified, and a juvenile plastron, C.N.H.M. UR2, are here referred
to C. stenopora.

Ctenochelys stenoporo

-D

o

Fig. 108. Plastron and fragments of carapace of Ctenochelys stenopora (juvenile).
After sketches and photographs of originals.

The type specimen of Toxochelys bauri, Y.P.M. 1786, and K.U. (V.P.) 1228
(fig. 108) resemble each other fairly closely except in size. Both differ from
K.U. (V.P.) 1205 notably in the posterior neural series and the suprapygal
region. It is, indeed, doubtful that a difference of this magnitude can be assumed
to represent extremes in individual variation, although deviations from the
normal pattern of even greater scope can be observed rather frequently in Recent
emyid turtles. It might be suggested that these two shells belong to C. procax,
of which no associated shell material is known. This remains a possibility, of

ZANGERL: VERTEBRATE FAUNA OF SELAIA FORIMATION

241

coui'se. I have chosen to include the two specimens in question under C. stenopora
for a number of reasons, none of which are, however, conclusive. The species
C. procax appears to be rare, only two skulls being definitely known. In \aew of

Ctenochelys stenopora

USNM 6013

Fig. 109.
juvenile).

Tracing of part of slab specimen of Ctenochelys stenopora (U.S.X.M. 6013,

the similar situation in the Moore\ille Chalk, it appears unlikely that there are
more than two species of Ctenochelys represented in the Niobrara beds. The
Mooreville species, C. tenuitesta and C. acris, differ from each other throughout
their known anatomy, not merely in one area of the shell. In so far as can be
determined, C. tenuitesta and C. stenopora, on the one hand, and C. acris and
C. procax, on the other, are the most closely related species. No shell material
from the Niobrara was noticed that was similar to that of C. acris. All of this is
indirect evidence and in no way proves the case. Additional material is required
to clarify the matter.

The features of the skull and mandible of this species had to be discussed
in the general account of the genus. Comparison of the shell materials here
refeiTed to C. stenopora with those of C. tenuitesta reveals the gi'eat over-all
similarity between these fonns. The plastra are wider than long, however,
indicating that C. stenopora has either a more broadly oval shell or relatively

242

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

shorter anterior and posterior plastral lobes, leaving a relatively greater area of
the ventral surface unarmoured. The peripherals are indistinguishable from
those of C. tenuitesta. The nuchal plate is moderately excavated in front as in
C. teimitesta, but the costo-nuchal sutures slant forward and outward as in C. acris.

Ctenochelys stenopora

USNM 6013

Fig. 110. Tracing of part of slab specimen of Ctenochelys stenopora (U.S.N.M. 6013,
juvenile).

The neural keel elevations are as pronounced as those of C. acris. The plastron
of A.M.N.H. 6137 is complete except for the right hyoplastron (fig. 111). The
epiplastra are thin, narrow plates of typical toxochelyid shape. The entoplastron
is very thin and twice as long as wide anteriorly. The central fontanelle extends
forward to the epiplastra. The xiphiplastra are widest at the level of the keel
knob; the area back of this point is short and bluntly pointed. The shape of
this element is thus intermediate between that of C tenuitesta and C. acris.

The girdle and limb-bones of A.M.N.H. 6137 are entirely flattened. The
outlines of some of them retained recognizable features and are illustrated (fig.
111). The proportions of the scapula are almost identical with those oi Lepido-
chelys kempi (see Table 5).

Ctenochelys acris sp. nov.

Type. — C.N.H.M. P27354, partial carapace, plastral fragments.

Horizon and locality. — Mooreville Chalk, Selma Formation, Late Creta-
ceous. Moore Brothers farm, Harrell Station area, Dallas County, Alabama.

Ctenochelys stenopora amnh6I37

Fig 111 Ctenochelys stenopora (A.M.N.H. 6137). Most significant parts of skeleton
associated with skull described by Hay as Toxochelys elkader (see pi. 17).

243

244

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Referred specimens. — Harrell Station area, Dallas County, Alabama
C.N.H.M. P27437, partial carapace; C.N.H.M. P27344, costal, peripheral
C.N.H.M. P27337, partial skull, mandible; C.N.H.M. P27340, partial skull
C.N.H.M. PR137, eighth cervical vertebra, girdle bones; C.N.H.M. P27352
vertebrae and limb-bones.

Ctenochelys acris

Fig. 112. Type specimen (P27354) and referred materials of Ctenochelys acris.

Cedarville area, Hale County, Alabama: C.N.H.M. P27356, hyoplastron,
xiphiplastron, neural, peripheral, costal (juvenile) (Crawford farm); C.N.H.M.
P27366, mandible, fragments of limb-bones (Twp. 11, west of Alabama High-
way 13).

West Greene area, Greene County, Alabama: C.N.H.M. PR251, partial
skull; C.N.H.M. PR157, left mandibular ramus.

Mt. Hebron area, Greene County, Alabama: C.N.H.M. PR97, partial
shell, vertebrae, mandible; C.N.H.M. PR62, partial carapace.

Diagnosis. — Skull with short snout area; distance between anterior rim of
orbit and tip of snout less than half width of skull at level of anterior orbital
rim. Mandible with straight or slightly convex lateral jaw outline. Width of
eighth cervical centrum at narrowest point about 52 per cent of its length.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 245

Ctenochelys acris

Fig. 113. Ctenochelys acris. Reconstruction of carapace from all available materials.

Peripheral edge of carapace with small but acute individual serrations located
almost directly at shield furrow in posterior peripherals. Nuchal deeply con-
cave in front; costo-nuchal sutures slanting forward and outward. A preneural.
Epineural ossicles more than half as long as adjoining neurals. No anal elevation.
Anterior suture of eleventh peripheral notably oblique. Plastron insufficiently
known. Xiphiplastron equals distance between medial corner of lateral fontanelle
and medial serrated margin of hypoplastron. Xiphiplastron widest across pos-
terior third, blunt posteriorly.

246 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

Description. — This appears to be a relatively rare species. The materials
on which the description is based belong to very fragmentary specimens and much
information is still wanting. If it were not for the fact that the turtle shell
exhibits a fairly stable pattern in its arrangement of parts, it would surely not
be possible to give a reconstruction of this animal (fig. 113) at present. It may
be noted at the outset that while the two forms (figs. 104 and 113), seen as morph-
ological entities, clearly show distinct features, many single characters of this
species are also present in C. temiitesta. It is thus not always possible to identify
fragmentary specimens without access to the entire present collection.

Specimen P27337 represents the best skull material so far collected. It
consists of the braincase, the snout area, a portion of the roof, and the mandible,
lacking the posterior end on the right side. The snout fragment (pi. 18) con-
forms both in size and shape to PR251. Both specimens are characterized by
relatively short snouts as compared with C. tenuitesta (fig. 99). In both of these
individuals, furthermore, the maxilla is notably thinner in dorso-ventral direction
at the base of the ascending maxillary pillar than in C. tenuitesta. The braincase
fragment of P27837 resembles that of C. tenuitesta (PR30, fig. 100) in the sense
that the basisphenoid rostrum approximates the condition in Recent cheloniid
turtles. On the ventral side of the fragment, there is a deep depression in the
forward area of the basioccipital, framed by the basisphenoid ridges, which is
also seen in P27340. Severe crushing appears to have made these depressions
more conspicuous. The mandibles are broadly rounded in front, not pointed as
in C. tenuitesta (fig. 98). A well-preserved jaw (P27366) shows a notable symphy-
seal ridge, sharpest at mid-length.

An isolated, well-preserved, eighth cervical vertebra (PR137) shows two
characters of interest. The centrum is about twice as long as wide at its narrow-
est point and nearly the same proportion is seen in the fairly well-preserved
eighth cervical of PR97 (width 53 per cent of length). The high neurapophysis
does not form a flat tuberosity at its apex, which, in other toxochelyid turtles,
rests against the ventral knob on the nuchal plate. Instead, it is medially
divided by a deep sulcus. Specimen PR97 shows exactly the same situation.
The additional three cervical vertebrae of PR97 are so badly distorted that their
numerical position in the column cannot be definitely determined. The two
posterior sacral and the first caudal vertebrae (PR97) are preserved intact.
All three elements have ventrally keeled centra, but these ridges are blunt. The
last sacral rib is larger than the second, as in Toxochelys. A sixth cervical centrum
(P27354) is virtually indistinguishable from its homologue in C. tenuitesta. Two
centra of anterior shell vertebrae are acutely keeled ventrally. Five caudal
vertebrae (P27352) belong to different parts of the tail series. These vertebrae,
as those of Toxochelys, are precocious and relatively broader, not laterally com-
pressed as are those of Chelydra, but resemble them rather closely otherwise.
They may or may not have borne dermal ossicles. Small haemapophyses of
chelydrid shape are present on one of the vertebrae.

The carapace was probably long and narrow. The nuchal plate is deeply
concave anteriorly and forms but a short suture with the first costal, which slants

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 247

sharply fonvard and outward. There is no ventral anchor knob for the eighth
cer\'ical vertebra. In the type specimen (P27354) and in PR62, preneural
elements are developed. The first neural is thus traversed by the vertebral
shield furrow near its anterior end, not near the middle as in C. tenuitesta (figs.
112 and 113). In C. acris, the epineural ossicles are relatively larger than in
C. tenuitesta; their length exceeds half the length of the adjoining neural plates.
The suprapygal region is not known, nor is the pygal. But the approximate
shape of the pygal is largely determined by the peculiarities of the eleventh
periplieral (fig. 112). The anterior suture contact of the latter with the
tenth peripheral is unusually oblique, and determines its approximate position
relative to the shell as a whole. The marginal edge of all of the posterior periph-
erals exhibits a feature characteristic of this species, namely, a type of serration
in which the edge anterior to the seiration point of each peripheral is slightly
concave; the point itself is small but acute and located immediately in front of
the scale fuiTow or directly at the spot where the shield fuiTOw reaches the
margin (fig. 112). The marginal edge of the anterior peripherals (1-6) is not
distinguishable from that of C. temiitesto. The nuchal suture of the first periph-
eral is less oblique in this species than in C. tenuitesta and the second peripheral
is of nearly equal width anteriorly and posteriorly, whereas in C. tenuitesta there
is a considerable difference (figs. 104, 113).

The plastron is not satisfactorily known. The xiphiplastron (fig. 112),
however, is readily distinguishable from that of C. tenuitesta since it is widest
in its posterior third, rather than at mid-length.

Most of the girdle and the limb-bones are so incompletely preserved that
they are difficult to compare. Of the two scapulae, P27352 and PR137, the
foiTner has a more elongated neck region than that of any other known toxo-
cheKid scapula. Specimen PR137 is indistinguishable from C. tenuitesta. Neither
of these specimens has complete processes. The coracoid (PR137), lacking the
posterior end, cannot be distinguished from C. tenuitesta.

Ctenochelys procax (Hay)

Toxochelys procax Hay, Bull. Amer. Mus. Nat. Hist., 21, p. 181, figs. 13, 14, 1905;
Carnegie Inst. Wash. Publ., 75, p. 176, figs. 224-226, 1908.

Tijpe. — A.M.N.H. 234, large skull and partial mandible.

Horizon and locality. — Niobrara Chalk. Kansas.

Referred specimens. — See below.

Amended diagnosis. — Skull with wide, broadly rounded snout area. Width
of triturating surfaces gi'eater than distance between them near posterior end.
Mandible not pointed in front. A notable s^miphyseal ridge forming a broad
convexity near posterior end of symphysis. Width of triturating surface relative
to length of symphysis conspicuously gi'eater than in other species of the genus.
Post-cranial skeleton unknown.

Discussion. — The type of this species is a large skull and partial mandible,
A.M.N.H. 234. Another skull with mandible of somewhat smaller size ( C.N.H.M.

248 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

UC614, formerly UC572) resembles the type so closely in every respect but size
that the identification is in no doubt. The isolated jaw fragment K.U. (V.P.)
2050, referred to this species by Hay (1908), is badly crushed and cannot be
identified as to species. The mandibular fragment A.M.N.H. 220, also referred
to C. procax by Hay (1908), compares more closely with Y.P.M. 3607 and is
thus refeiTed to C. stenopora. There are no shell materials associated with the
type or with C.N.H.M. UC614. Concerning the possible reference of shell
specimens to this species, see discussion under C. stenopora (p. 240).

Prionochelys gen. nov.

Characterization. — Highly specialized genus. Skull with lophochelyine
palate and epidermal shield cover. Carapace cordiform, with deep peripheral
serrations and pronounced excavation of anterior nuchal rim. Neural carina,
consisting of ten elements, one preneural and nine neurals, very high; contains
epithecal elements located at junction of neurals 2 and 3, 5 and 6, 8 and 9.
Nuchal plate much wider than long. First peripheral plates encroach upon
nuchal laterally. Preneural much wider than long. No post-nuchal fontanelles
in adults. Plastron poorly known. Xiphiplastra similar to those of Ctenochelys.

Type of genus. — Prionochelys nauta.

Horizon and locality. — Marlbrook Marl, Late Cretaceous. Arkansas.

Discussion. — Prionochelys occurs in both the Mooreville Chalk of Alabama
and the Niobrara Chalk of Kansas. The best materials, however, come from
the Marlbrook Marl of Arkansas, which represents a notably higher stratigraphic
level. All of the Marlbrook specimens are large, but there is evidence that the
older Mooreville form was an equally large turtle. Prionochelys is apparently
very rare in both the Mooreville Chalk and the Niobrara, whereas in the Marl-
brook Marl it would seem to be the dominant form.

In so far as comparison is possible, it can be stated that Prionochelys pos-
sessed a typical toxochelyid skull. The palate was of lophochelyine character,
with a large portion of the palatines forming the medial edges of the masticatory
surfaces on either side of the choanae. The roof of the skull was covered with
epidermal shields. The articular surfaces of the quadrates consist of large medial
and small lateral facets.

The features of the carapace are outstanding. The neural elements (a
preneural and nine neurals) form a high keel with four or five dorsal eleva-
tions, namely, a cervical (above preneural), a pectoral, an abdominal and a
pelvic; an anal elevation may be located at the junction of the suprapygal and
pygal plates (fig. 114). The pectoral, abdominal and pelvic elevations may be
formed either by epithecal elements only (as in Ctenochelys) or by neural plates
in conjunction with epithecals. The latter are always spaced three neurals apart,
instead of two as in Ctenochelys. Among the three recognized species, there are
considerable differences in the detail of keel specialization (fig. 114).

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 249

The peripherals are deeply serrated, with the serration points at, or near,
the suture of adjoining elements. Lateral fontanelles are always present. Post-
nuchal fontanelles are seen in P. galeotergum, absent in P. nauta. Since the only
known specimen of the former is a young individual, whereas the latter is based
on large material only, the presence or absence of post-nuchal fontanelles in this
genus might well be a matter of age rather than a specific difference.

No suprapygal plates are known. It is possible that there is only one, the
lower one. The upper suprapygal of the generalized toxochelyid shell appears
to have become segmentally displaced into the position of neural 9 in this genus,
thus assuming the role and the shape of that element.'

The plastron is incompletely known. It is typically toxochelyid, however,
and the xiphiplastra are very similar to those of Ctenochelys.

Of the girdles, only the ilium and ischium are known. They are practically
indistinguishable from those of Toxochelys. The limbs are unknown.

Prion ochelys nauta sp. nov.

Type. — C.N.H.M. P26237, partial shell, neurals, peripheral fragments,
costals, and fragments of plastron.

Horizon and locality. — Marlbrook Marl, Late Cretaceous. Devil's Back-
bone, Saratoga, Howard County, Arkansas.

Referred specimens. — C.N.H.M. P26238, partial carapace and plastron
(Devil's Backbone, see above); C.N.H.M. PR271, fragment of plastron (same
locality as above); C.N.H.M. P27456, peripherals, plastral fragment (Nick
Gaiter farm, ca. 1 mile northeast of junction of Okolona-HoUywood road to
Arkadelphia, Clark County, Arkansas; Marlbrook Marl); C.N.H.M. P27461,
partial carapace, skull fragments (same locality as P27456); C.N.H.M. PR221,
nuchal and first peripheral (same locality as P27456); C.N.H.M. P27460, frag-
mentary specimen (same locality as P27456) ; C.N.H.M. P27454, partial carapace
and plastron (Cox farm, same data as P27456).

Diagnosis. — Spectacular turtles. Carapace with specialized neural keel and
high cervical, pectoral, abdominal, and pelvic elevations. Epithecal ossicles
large. Ridge line of neural 2 as long as distance between anterior and posterior
sutures. Neural 8 short, almost excluded from the ridge profile. Preneural
with antero-lateral wings; no post-nuchal fontanelles in adult. Shell thick.
Xiphiplastron widest at level of keel knob.

Description. — The only skull material, consisting of fragments that cannot
be fitted together, is associated with P27461. The identifiable pieces include a
portion of the right mandible, portions of both maxillae with parts of the palatine
bones suturally attached, the left ascending ramus of the maxilla with part of
the prefrontal in place, a portion of the roof of the skull in the area of the junc-
tion of the frontal, parietal, and post-orbital bones, a fragment of the left otic

' For comparable examples of segmental displacement of serial structures, see Kiihne,
1936; Schinz and Zangerl, 1937.

250

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

area, and the articular processes of both quadrates. The mandible is represented
by three disconnected sections from the masticatory region.

These fragments belong to a skull of the approximate size of that of a
large Caretta. The suture pattern of the bones of the skull roof resembles closely
that of other toxochelyid turtles; the frontal bones form the dorsal edges of the

Fig. 114. Comparison of morphology of keel in Prionochelys nauta, P. matutina, and
P. galeotergum.

orbits. The skull was covered with epidermal shields. The interorbital bridge
at the fronto-parietal junction is 76 mm. wide. Fragments from either side of the
choanal opening show the maxillo-palatine suture. The palatine bones formed a
considerable portion of the triturating surface; it is not possible, unfortunately,
to determine whether the condition is similar to that of Ctenochelys or further
advanced in the direction of secondary undershelving of the choanal passages.
The alveolar ridge is relatively low, about the same relative height as in Cteno-
chelys. The articular processes of the quadrates show articular joint surfaces
characterized by large medial and small lateral facets, separated from each other
by longitudinal elevations, thus differing from those of Toxochelys and Ctenochelys
in which the lateral facets are larger and are separated from the medial facets by
longitudinal depressions (fig. 115). The mandible seems very massive. The
fragments permit the statement that it is, in general organization, similar to
that of Ctenochelys. The triturating surface may be relatively wider than in the
latter genus, but it was certainly not expanded as in Osteopygis.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

251

The carapace (fig. 116) does not reach the degi-ee of ossification illustrated
in the reconstruction until the individual has attained the size of the largest
specimens now under study. In P26238, an individual somewhat smaller than
the rest, the lateral fontanelles extend forward to the nuchal plate and the first
peripheral is thus relatively long and narrow (pi. 28). In the largest specimens

Prionochelys nouta

ngfit <fuadrate. ar/icu/ar SUr/ace /moress,

CNHM P2746I

CNHM P2746I

, 50mm . CNHM P27454

Fig. 115. Skull fragments and portions of anterior carapace of Prionochelys nauia.

(P27461 and P27454, fig. 115), the lateral fontanelles extend forward approxi-
mately to the junction between peripherals 2 and 3. The first peripheral is, at
this stage, triangular in shape. Clearly, the rate of ossification progi-esses at a
faster pace than the general growth of the individual at these rather late onto-
genetic stages. There is no indication in any specimen that the lateral fontanelles
disappear entirely.

The nuchal plate, in the fully developed condition, is a wide plate, pointed
laterally and excluded from the rims of the lateral fontanelles (figs. 116, 117).

252 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

It is deeply excavated in front and fonns an unusual suture contact with the
preneural (pi. 28). It is probable that post-nuchal fontanelles are present in the
young, but this remains to be demonstrated in future discoveries.

Fig. 116. Prionochelys nauta. Reconstruction of carapace from all available materials.

The neural series of this form is the most advanced of the three known
species. Between the two specimens P26237 and P26238, the preneural and eight
thecal neurals are known; a ninth, bearing a transverse shield furrow and resem-
bling in type neurals 3 and 6, must have followed neural 8 (the ninth neural is
known in P. matutina and P. galeotergum) . The thecal neural elements form part
of the keel profile and have thus become greatly modified. Four dorsal keel
elevations in cervical, pectoral, abdominal, and pelvic positions are documented

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 253

in P. nauta. The cervical elevation is formed mostly by an epithecal ossicle
located at the junction of the nuchal plate and the preneural. Three neural
plates and one epithecal ossicle each contribute to the formation of the pectoral,
abdominal, and pelvic elevations. Neurals 1, 4, and 7, neurals 2, 5, and 8, and
neui-als 3, 6, and 9 are of essentially similar shape and are in comparable positions
relative to the apices of these elevations. There is, however, a progressive crowd-
ing of the neurals from the pectoral to the pelvic elevation, affecting primarily
neurals 2, 5, and 8. In the pectoral elevation, neural 2 forms the ascending
component of the apical profile. Neural 5 in the abdominal elevation has a short

Prionochelys nouto

Fig. 117. Prionochelys nauta. Lateral view of reconstructed carapace.

ridge line and forms the apex proper. Nem-al 8 is the central element of the
pelvic elevation; it is, however, short and all but excluded from the apical profile
(fig. 114; pi. 27).

The keels of P26237 and P26238 show minor differences probably correlated
with the size difference between the two specimens. In the smaller individual
{P26238), the ridge line of the keel is sharper and the apices are more acute.
The fourth neural in the abdominal elevation does not reach to the apex as it
does in the larger individual; it resembles the more primitive condition in P.
matutina (fig. 114).

The costal plates, of which numerous fragments but only a few nearly
entire elements are available, are rather thick ( + 10 mm. along the sutures of
the costals illustrated in pi. 27) and the ribs are not clearly distinguishable
from the overlying plates, as is always the case in the other toxochelyid genera.
A fair number of proximal costal fragments, on which shield furrows are deeply
impressed, permit a reasonably accurate reconstruction of the vertebral and
pleural shield pattern (figs. 116, 117).

The peripheral plates, like the rest of the carapace, are relatively massive.
Of P27456, a moderate-sized individual, only peripherals and a piece of plastron
were preserved. The lot contains peripherals 3, 8(?), 10, and 11 of the right side
(pi. 29) and 8(?) and 11 of the left side. The marginal edge is deeply notched
and sharp. The elements attain their greatest width directly behind or at the

254 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

sutural contacts with the preceding peripherals (pi. 29 and fig. 116). The anterior
suture plane of the eleventh peripheral stands at an oblique angle to the longi-
tudinal axis of the element, resembling the condition of Ctenochelys acris (fig. 113).
The medial faces of the peripherals are slightly concave and rather smooth.
Distinct but shallow rib-pits are present on the third and on the extreme posterior
end of the tenth peripherals. On the eighth(?), there is no trace of a pit, but the
dorsal half of the medial face, overlying the rib end, shows a slight impression
half way between the shield furrow and the posterior suture. The dorsal edge
of the medial face of the peripheral is thick and blunt throughout its length,
except at the place where it overlay the rib ; there it is much thinner and sharper.
The dorsal surfaces of these peripherals are slightly convex in medio-lateral
direction, a matter likewise true in the other species of Prionochelys, but not in
any other toxochelyid genera where the dorsal surface of the peripherals is more
or less concave.

The plastron remains inadequately known. The best fragments belong to
P26238 (pi. 28). They consist of a medial half of the left hyoplastron, the medial
end of the right hypoplastron and the greater portion of the left xiphiplastron.
These fragments show a typically toxochelyid condition, but do not permit a
more detailed comparison with the plastra of other genera. All three fragments
show pronounced keel knobs on their ventral surfaces, much as in Ctenochelys,
and the xiphiplastron resembles in outline that of C. acris.

Prionochelys matutina sp. nov.

Type. — C.N.H.M. P27561, partial carapace and plastron, elements of pelvis.

Horizon arid locality. — Mooreville Chalk, Selma Formation, Late Creta-
ceous. Moore Brothers farm, Harrell Station area, Dallas County, Alabama.

Referred specimens. — Eutaw area, Greene County, Alabama: C.N.H.M.
PR31, crushed peripherals.

Harrell Station area, Dallas County, Alabama: C.N.H.M. P27479, large
peripheral; C.N.H.M. PR222, nuchal and first peripheral of large individual.

West Greene area, Greene County, Alabama: C.N.H.M. PR185, periph-
erals, pygal, neural, and costal of juvenile specimen.

Diagnosis. — Species closely related to P. nauta, but with much less acute
keel profile. Ridge line of neural 2 only half the distance between anterior and
posterior sutures. Neural 8 not reduced in length as in P. nauta. No anal
elevation. Marginal edge of pygal sharp. Xiphiplastron widest in front of
keel knob.

Description. — The material on which this species is based consists of the
remains of very small, medium-sized, and large individuals. PR222, consisting
of the nuchal and left first peripheral only, indicates a turtle of about the same
size as P. nauta. The more adequate specimens are thus considered subadult
(P27561, type specimen) and juvenile (PR185). In view of the fact that a
number of differences, apparently brought about by differential growth rates of

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

255

the various carapacial elements, were noted in two large specimens, of unequal
size, of P. nauta, the morphology of these rather young individuals of P. matutina
should be evaluated with care.

In comparing the type specimen of this species with the smaller individual
of P. nauta (P26238), it may be noted that elements of the keel are about two-

Prionochelys matutina

CNHM P2756
50mm

Fig. 118. Type specimen of Prionochehjs matutina (C.N.H.M. P27561).

thirds the size of those of P26238, whereas the xiphiplastra of the two specimens
are of nearly equal size (fig. 119). Since the neural series determines, to a great
extent, the length of the carapace, I am inclined to consider the size relationship
between the neurals (rather than that between the xiphiplastra) as an approxi-
mate indication of the over-all size category of the two specimens. The morphol-
ogy of the neural keel in the type specimen of P. matutina may thus differ in
detail from that of a fully grown individual, and adult specimens of both species
might reveal a greater or lesser degree of similarity than do the specimens now
at hand.

I do not believe, however, that the keel differentiation as seen in the type
specimen of P. matutina merely represents an earlier ontogenetic stage in a
developmental series ending with the condition observed in P. nauta. It is a

256 FIELD lANA: GEOLOGY MEMOIRS, VOLUME 3

general rule, in turtles, that spines and ornamental processes on the shell are
far more pronounced and acute in juveniles than in adults. A difference in this
regard is seen even in the two unquestionably adult individuals of P. nauta
(P26237 and P26238). The keel of P. matutina is blunt by comparison with the
larger specimen of P. nauta, and its adult differentiation is likely to be less rather
than more acute. In view of these considerations, I am inclined to regard the
keel differentiation in P. matutina, even though materials of comparable size
are not available, as notably more primitive than that of P. nauta.

The neural elements of P27561 (fig. 118) are basically similar in structure
to those in P. nauta. The keel elevations are lower and less abrupt, and there
are differences in proportion among the elements, as is readily seen in figure
114. Neural 2 has a much shorter, and neurals 5 and 8 a much longer, ridge line
than their homologues in P. nauta. In the abdominal keel elevation, the epi-
neural ossicle is a rather insignificant bone, to judge from the suture scars on
neurals 5 and 6. The pygal plate is of ordinary shape, lacking a scar for an
epithecal ossicle (fig. 118). It is, on the whole, convex both dorsally and ventrally,
except for a shallow depression along the ventral mid-line. Its marginal border
is sharp. A small sagittal process on the anterior face of the pygal indicates that
the suprapygal tapered to a very narrow, rod-like plate posteriorly, and that the
lateral fontanelles extended very close to the mid-line. Only the posterior periph-
erals are known in the type specimen. They form a deeply serrated margin as in
P. nauta. The anterior suture of peripheral 11 is less oblique than in P. nauta,
indicating that the carapace was more evenly pointed posteriorly. Very little
is known of the costal plates. One posterior element, possibly the left eighth, is
thin, distally barely covering the underlying rib.

The plastron is represented only by the left xiphiplastron. It is a sturdy
bone with a strong keel knob on its ventral face near the lateral edge. The
shape of this plate differs notably from that of the xiphiplastron in P. nauta
(pi. 28). It is widest at mid-length and resembles in outline the xiphiplastron
of Ctenochelys tenuitesta.

Both ischia and the left ilium are present in P27561 (fig. 118). The ilium
shows typical toxochelyid shape but no specific peculiarities. The ischium is
wider laterally than medially and possesses a spur that faces postero-laterad.
This again appears to be typical for toxochelyid turtles in general, and side by
side comparison with the ischia of Toxochelys moorevillensis (P27391) shows a
remarkable degi'ee of similarity in this element among otherwise very different
animals.

A noteworthy specimen is the juvenile individual PR185. It consists of
two lateral and the tenth and eleventh peripherals, the pygal, the ninth neural,
and a small posterior costal. In linear measurement, it is less than half the size of
P27561 (length of pygal along mid-line: P27561, 46 mm.; PR185, 20 mm.).
The pygal, entirely uncrushed, is very thick for its size (maximal thickness
10 mm.) and strongly convex from side to side dorsally. The ventral face is
also convex, except for a depression along the mid-line as in the type specimen.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

257

On the anterior face, there is a relatively large area for the junction with the
suprapygal.

Comparison with the eleventh peripheral of P27561 shows an interesting
detail concerning the growth of this plate and that of the pygal. It may be
noted in figure 120 that the anterior (oblique) suture face of the eleventh periph-
eral of P27561 does not form a straight contact line with the preceding peripheral ;
the marginal third of this line forms an angle of about 20° with the inner two-
thirds. In PR185, this suture line is straight. Enlarging (photographically)
PR185, so that its anterior suture face corresponds in length to the inner two-
thirds of the suture in P27561 (fig. 120), brings the centers of ossification of the
peripherals and pygals into congruence. This indicates that up to the two-thirds
mark of the anterior suture in P27561 most of the growth in width of the periph-
eral took place medial to the center of ossification; then the direction of growth
was reversed and the element grew marginally outside of the center of ossifica-
tion. The pygal plate, as would be expected, underwent a similar two-phase
gi'owth, first forward, then marginal.

Prionochelys nouta

Prionochelys matutino

CNHM P26238

Fig. 119. Comparison of relative
sizes of third neural plates and xiphi-
plastra in Prionochelys nauta and P.
matutina.

:^'-~.

Prionochelys matutino

_CNHM PRI85 ^^

...CNHM P2756I ^"-'' "~-'

X center of ossification for both specimens

Fig. 120. Mode of ossification
(growth) in pygal and eleventh periph-
eral plates of Prionochelys matutina, as
suggested by two different ontogenetic
stages.

258 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

The ninth neural shows two interesting features. Anteriorly, at this juvenile
stage, the epithecal ossicle is small and has not yet formed a sutural contact
with neural 9. This is in agreement with observations reported earlier (Zangerl,
1939) to the effect that epithecal ossifications are acquired relatively late in
ontogeny, always after the anlage of the thecal plates. Judging from the suture
scar on the eighth neural in P27561 (fig. 118), there is no doubt that the epineural
made sutural contact with neural 9 in this older specimen, as shown in figure
114. On the posterior end of neural 9 of PR185, there is a long dorsal process
that must have been locked into the anterior margin of the suprapygal as illus-
trated in figure 114.

The largest specimen (PR222) consists of the badly weathered nuchal
fragment and the left first peripheral. These elements indicate great similarity
to the corresponding plates of the largest specimens of P. nauta. The nuchal,
however, is strongly arched from side to side near the mid-line and some distance
back of the cervical edge.

Prionochelys galeotergum sp. nov.

Type. — C.N.H.M. PR125, fragmentary carapace of subadult individual.
Collected by Mr. G. F. Sternberg, 1931.

Horizon and locality. — Niobrara Formation, Late Cretaceous. Cove County,
Kansas.

Diagnosis. — Most primitive member of the genus. Elevations of the neural
keel formed by large epithecal ossicles only. Epithecals anchored in deep twin
pits on neurals. Anal elevation present. Post-nuchal fontanelles in young
individual. A knob near mid-line on dorsal face of nuchal plate, close to cervical
edge. Marginal edge of pygal blunt.

Description. — The type is the only known specimen. It was purchased
from Mr. G. F. Sternberg along with a number of other fragmentary but highly
interesting turtle materials. The label indicates no specific locality.

All of the identifiable parts of this specimen are illustrated in figure 121.
PR125 is about two-thirds the size of the type specimen of P. matutina; it is thus
probably a rather young individual.

The shape and deep serration of the posterior peripherals are very similar
to P. matutina and the neural keel shows the epithecal ossicles spaced three
thecal neurals apart (fig. 114). The generic affiliation of this form is thus in no
doubt. With regard to the differentiation of the neural keel, however, P. galeo-
tergum is much more primitive than the other two species. As in Ctenochelys, the
epithecal ossicles alone form the keel elevations (figs. 114 and 121). The epi-
neural elements have a characteristic shape and are connected with the under-
lying neural plates by an unusual sutural contact. At the junction of neurals
2 and 3, there is a very deep sutural pit into which the anterior portion of the
epineural was locked. A second and more shallow pit lies behind the first and is
separated from it by a ridge. On the under side of the epineural, there is a

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION

259

depression that fits over the ridge between the pits on the neural plates (fig. 121).
In contrast to P. matutina, this form has an anal elevation, an epithecal ossicle
overlying the junction of the suprapygal and the pygal plates (fig. 121). The
pygal is thick and its marginal edge is blunt rather than sharp as in P. matutina.
It is far narrower anteriorly than in the compared species (fig. 121).

The nuchal plate is incomplete, but shows a number of important features.
Postero-laterally, the fragment shows a thin, natural edge indicating the pres-
ence of post-nuchal fontanelles. The anterior margin is moderately excavated.
A small knob is located on the mid-line between the cervical edge and the ante-
rior shield furrow of the first vertebral shield (fig. 121). The lateral extent of the
nuchal shield cannot be determined satisfactorily. The dorsal face of the nuchal
is gently arched from side to side.

The two anterior peripherals, numbers 3 right and 4 left (fig. 121), do not
show distinct rib-pits and neither does the tenth. There is an abnormality in the
posterior peripheral series on the right side. Instead of an eleventh peripheral,

Prionochelys galeotergum

CNHM PRI25

Fig. 121. Type speci-
men of Prionochelys galeo-
ferfifMm(C.N.H.M.PR125).
The right eleventh periph-
eral is abnormally divided.

Fig. 122. Neural frag-
ment of Ctenochelys tenui-
testa (C.N.H.M. P27425)
in lateral view (keel ridge
on left side), showing se-
vere parasitic lesions.

260 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

there are two plates, neither equivalent in size and shape to the eleventh element
on the left side.

In addition to the bones described, there are fragments belonging to the
costal plates, peripherals, and hyo- or hypoplastron.

Forms Questionably Referred to the Toxochelyidae

A number of forms that have been referred to the Toxochelyidae can not
readily be identified as belonging to any of the three subfamilies distinguished
above, and there is no convincing reason why any of these species should be
included in this family. It may be, of course, that the Toxochelyidae had a
much wider distribution and that forms from other continents may represent
additional subfamily patterns. On the basis of the present record this is not
a likely probability.

Cynocercus incisivus Cope

Cynocercus incisivus Cope, Proc. Amer. Phil. Soc, 12, p. 308, 1872; Proc. Acad. Nat.
Sci. Phila., p. 129, 1872; Fifth Ann. Kept. U. S. Geol. Surv. Montana, .... p. 335,
1871 (1872); Kept. U. S. Geol. Surv. Terr., 2, pp. 96, 260, pi. 8, figs. 3-5, 1875;
Williston, Univ. Kansas Geol. Surv., 4, p. 368, fig. 6, 1898; Hay, Field Columb.
Mus., Zool., 1, p. 106, 1896; U. S. Geol. Surv. Bull., 179, p. 442, 1902; Carnegie
Inst. Wash. Publ., 75, p. 180, 1908.

This species is based on a few caudal vertebrae and a metapodial bone,
probably from the Niobrara Chalk of Kansas. The specimens are described and
illustrated by Cope and Williston as cited above. The vertebrae differ in a
number of respects from those few caudal vertebrae known in Toxochelys and
Ctenochelys, even though elements of probably the same general region of the
tail are available for comparison. It is possible that Cynocercus is not a toxo-
chelyid, but the matter cannot be decided at present.

Toxochelys gigantea Oertel

Toxochelys gigantea Oertel, Jahresber. Niedersachs. Geol. Ver. Hannover, 7, pp. 91-106,
1 fig., 1914.

Under the above name Oertel described an isolated, very large skull ( + 250
mm. sagittal length) of a chelonian from the lower Aptian of Karstendamm,
near Hannover, Germany. The posterior half of the skull roof is missing and the
anterior part is crushed to the left side. The palatal aspect of the skull is fairly
complete. Of particular interest is the area of the vomer and the palatines. The
situation is essentially the same as in Ctenochelys in the sense that the roof of the
mouth cavity exhibits a condition intermediate between a primary and a sec-
ondary palate. The vomer and the lateral portions of the palatines are unusually
rugose and a partial undershelving of the choanal passages seems to have taken
place. In detail this area differs, however, considerably from that of Ctenochelys.

ZAXGERL; VERTEBRATE FAUNA OF SELMA F0R:\IATI0N 261

In attempting to decide whether or not this species represents a toxochelyid,
it should be remembered that the toxochelyid skull is in many ways similar to
that of the Cheloniidae and differs from it primarily in the region of the palate.
The secondary cheloniid palate is unquestionably derived from a primary palate
and inteiTnediate conditions are thus to be expected in the Cheloniidae as much
as in the Toxochelyidae. In my opinion the skull described by Oertel belonged
to a cheloniid rather than a toxochelyid turtle.

Sinemys lens Wiman

Sinemys lens Wiman, Palaeontogr. Sinica, ser. C, 6, (3), p. 7, pi. 1, figs. 1-4, pi. 2,
figs. 1-4, pi. 3, figs. 1-5, 1930; Xopcsa, Centralbl. Min. Geol. Pal., (B), p. 510, 1930.

The material on which Wiman's species is based consists of seven specimens
of partial shells of different size. Nopcsa (1930) reviewed Wiman's description
and illustrations and decided that Sinemys is a "definitely primitive cheloniid
belonging to the Lytolominae," the exponent representative of which group
Nopcsa considered to be Osteopygis; he actually compared Sinemys with this
genus. It seems certain that Nopcsa's familiarity with Osteopygis was from the
literature only; I doubt that he would have been so definite in his conclusion had
he actually seen specimens of Osteopygis.

Sinemys differs from all unquestionable toxochelyid turtles in so many
respects that its inclusion in the family would constitute an arbitrary procedure
at the present state of our knowledge of the group. The carapace of Sinemys,
a small turtle (approximate carapace length of largest specimen — an adult,
judging by the proportions of the vertebral shields -is 190 mm.), differs from
Toxochelyidae in the suprapygal area, in the mode in which the peripheral
plates are sutured to the costals, and in the number of neural plates. But the
most remarkable feature is the lack of lateral fontanelles even in the smallest
indi\idual (carapace length about 70 mm."). In all toxochelyids the juveniles
have larger fontanelles than the adults.

The plastron is preserved only in two very small individuals ( Wiman, 1930,
pi. 1, figs. 1, b, and 2, a). Again, it is most peculiar, since hyo-, hypo-, and
xiphiplastra are sutured along the mid-line, leaving a large umbilical fontanelle
and apparently no lateral fontanelles (loc. cit., pi. 1, fig. 2, a, left side of picture).
The axillaiy and inguinal notches are not as deeply excavated as in toxochelyid
turtles, the axillo-inguinal distance being 80 per cent of half the width of the
plastron, whereas in the most advanced toxochelyid forms it is only 60 per cent
or less and in the primitive genera around 35 per cent. The posterior lobe of the
plastron of Sinemys is much longer and narrower than in any toxochelyid.

Sinemys is stated to be of Early Cretaceous age; it is thus much older than
the toxochelyid turtles discussed above.

Sinemys could well be a turtle remotely related to the Cheloniidae and the
Toxochelyidae. There is no reason to deny, on the basis of present knowledge,
a likewise remote relationship to the Chelydridae and the Dermatemyidae.

262 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

It has, to my knowledge, not been pointed out that Sinemys might be an
advanced plesiochelyid. A comparison of the plastron with that of Plesiochelys
etalloui, and the carapace (with its short and wide first vertebral shield, and with
a similar suture pattern between peripherals and costals) with that of a number
of species of Plesiochelys is, I believe, rather suggestive. The relatively narrow
vertebral shields and the greater slenderness of the posterior lobe of the plastron
in Sinemys should not be regarded as a serious obstacle to this interpretation,
since there is a tendency in both these directions among various species of
Plesiochelys.

THE PALEOECOLOGY OF THE TOXOCHELYIDAE

Horizontal and Vertical Distribution

The geographical distribution of the toxochelyid turtles is, so far as is
definitely known, restricted to the coastal waters in the southern half of the
East Block of the Cretaceous North American Continent (Eastland; see W. A.
VerWiebe, 1933). On the west side of this land mass they inhabited the relatively
shallow, intracontinental Niobrara-Pierre Sea; in the south they populated the
Gulf shores from eastern Texas to Alabama and were apparently abundant
along the Cretaceous coast of New Jersey. It would seem reasonable to assume
that the shores between the coastal areas mentioned were likewise inhabited by
these turtles. The absence of records is due to a number of circumstances, such
as unfavorable physiographic conditions for fossil collecting, for instance in
Georgia; lack of determined collecting efforts in formations with meager fossil
occurrence; and application of soil conservation programs that prevent large
scale erosion in potentially fossiliferous areas.

While the distribution of these turtles, at the family level, presents a fairly
clear-cut picture, it must be kept in mind that they were seafaring animals and
may well have ranged farther south along the Cretaceous shores of Central
America. It is, of course, possible that the toxochelyid turtles occurred in
Europe, but the one record, a skull described as Toxochelys gigantea by Oertel
(1914), is in all probability a cheloniid (see p. 260).

On the generic level it is worth noting that Toxochelys, probably the most
common type of turtle in the coastal waters at the time, is represented in every
formation from which toxochelyid turtles have been recorded. There is no evi-
dence that Osteopygis, or even a closely related genus, occurs anywhere but on
the east coast. Thinochehjs and Porthochelys are rare in the formations in which
they occur and may be found elsewhere in the future. Lophochelys, Ctenochelys,
and Prionochelys were well represented along the Gulf coast and in the Niobrara
waters, but none of these genera ranged as far as the Atlantic coast.

Of the three subfamilies distinguished above, the Toxochelyinae extend
throughout the range of the family as a whole, but the Osteopyginae and Lopho-
chelyinae occupy mutually exclusive geographic areas.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 263

Little need be said about the species distribution, since the record is far
too incomplete to merit discussion. In the Mooreville Chalk of Alabama there
is no doubt but that all of the common species occur all along the strike of the
outcrop area of the marls from the western border of the state to the region
of Montgomery.

Most of the toxochelyid turtles are of Senonian age. If all of the labels
on the Greensand materials of New Jersey are to be trusted, Osteopygis may
extend well into the early Tertiary (Eocene). Toxochelys ranges vertically
from Coniacian to early Maestrichtian time, the lophochelyine genera from
Coniacian to late Campanian, and Osteopygis probably from early Maestrichtian
time on, possibly into the Tertiary.

Very little is known as to the vertical range of species. In the Mooreville
Chalk of Alabama there is no evidence to the effect that various species are
stratigraphically restricted. All common species appear to occur throughout
the Mooreville Chalk profile.

Habitat

There is no question but that all of the formations in which toxochelyid
turtles occur are either near-shore deposits, or beds laid down in shallow inland
waters (Niobrara-Pien-e Sea). However, these may not have been the habitat
of all species; the question of habitat and burial gi-ound should be discussed in
somewhat greater detail in the case of the Mooreville Chalk, where systematic,
unbiased collections of all vertebrate remains, regardless of how fragmentary,
were made over a period of years.

The relative frequency of occurrence of the different species is as follows:
Toxochelys moorevillensis, 52 (5 good shells); Thinochelys lapisossea, 4 (1 good
shell); Lophochelijs venatrix, 4 (fragmentary); Ctenochelys temiitesta, 38 (frag-
mentary); C. acris, 13 (fragmentary); Prionochelys matutina, 5 (fragmentary).

The two most abundant species are Toxochelys moorevillensis and Ctenochelys
temiitesta; the remaining species are rare by comparison. Since the sediments in
which these forms are buried are not in the immediate vicinity of the shore line,
and since the skeletons are not found in abundance in any one horizon and area,
there is no apparent correlation between the occurrence figures and egg-laying
behavior, that is, seasonal concentration of individuals near shore. The frag-
mentary nature of most of the specimens is not solely due to the erosion prior to
collection of the specimens (see below). If that were the case one should expect
a much higher percentage of nearly complete skeletons. The burial position of
the bones in such specimens as were found in place shows clearly that the skeletons
were severely macerated and often became scattered before burial. Entirely
articulated shells were never encountered. It would thus seem that the death
rate, from whatever causes, was fairly evenly distributed in time and over a
large area. Under these circumstances the figures of relative frequency of occur-
rence may indicate that the common species lived in the waters that produced
their burial ground (that is, the waters directly above).

264 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

In the case of the relatively rare species it is probable that their habitats
were more or less distant from the burial ground. Tfmiochelys, for example, is
a large, heavy-shelled form, and, by comparison, not as well adapted for a
marine existence as Ctenochelys tenuitesta. It is conceivable that Thinochelys
lapisossea lived in coastal swamps and large inlets and ventured away from the
shore only occasionally. Ctenochelys acris may have frequented the region of
the burial ground, but was essentially a pelagic form; the same, possibly, might
be true of Prionochelys matutina.

Food

The differentiation of the jaws is not a good indicator of food habits in
turtles. A secondary palate, for example, is not necessarily indicative of a
durophagous diet, for it occurs also in species that feed exclusively on plant
materials (for example, Chelonia viydas). On the other hand, a jaw apparatus of
the type of Recent snapping turtles such as Chelydra serpentina is likewise
unreliable as an indicator of exclusive fish or even exclusive animal diet. Chelydra
itself consumes large quantities of vegetable materials (Lagler, 1943). Inverte-
brates occur in the burial ground, but they are not particularly numerous.
Remains of a variety of fishes, including elasmobranchs, are relatively common
and so are poorly preserved plant remains. The burial ground itself thus con-
tains a considerable variety of potential sources of food, none of which is so
abundant as to dominate the specific character of the biotope.

There is, unfortunately, no acceptable evidence suggestive of the probable
food habits of the toxochelyid turtles, a matter of potential interest, in view of
the fact that the transition from a primary palate to a secondary palate is real-
ized within the group.

Locomotion

Functional analysis of the girdle and limb skeleton of Toxochelys (see p.
163) would seem to indicate a transitional type between the typical mode of
locomotion of fresh-water turtles and that of true sea turtles. There is, indeed,
good reason to believe that Toxochelys was capable of fast, short-distance pro-
pulsion in the general manner of trionychid turtles, and of cruising in the fashion
of cheloniid turtles, but it may be assumed that neither type of locomotion was
as efficient as it is in the compared groups. Ecologically, the toxochelyid type
of locomotion suggests a greater home range of the individuals than has been
determined for a number of Recent fresh-water forms (Cagle, 1944). In the
latter the home range appears to be remarkably limited.

Parasitic Infestation

Parasitic lesions in the shells of toxochelyid turtles are very common;
nearly every specimen shows at least minor parasitic erosions. As a rule, these
are shallow pits of varying size and usually more or less circular outline. In

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 265

some cases these lesions perforate the shell entirely. The most severely infested
specimens belong to Ctenochelys tenuitesta, where the keel-forming neurals, in
particular, but also the peripherals often contain numerous, large erosion cavities
(fig. 122). As a rule the lesions are larger on the outside of the shell, or, on the
peripherals, on the side on which the infestation began. In cases of perforation,
the inner opening is usually smaller in diameter than the outer opening. In other
cases both outer and inner openings are small but there is an enlarged cavity in
between. There is, at the moment, no way of determining whether these different
lesions were caused by the activity of one or of several kinds of parasites.

In microscopic section the parasitic lesions show dense bone of varying
thickness surrounding the cavities. Apparently there is an increased deposition
of bone in the vicinity of many erosions, often producing notable, local thickening
of the plates above and below the cavities.

Burial Conditions

Entirely articulated shells have not been found in the Mooreville Chalk.
A number of specimens of Toxochelys mooreviUensis and one specimen of Thino-
chelijs lapisossea comprise so large a portion of the shell and even some vertebrae,
girdle, and limb bones that we may assume that the carcasses were left undis-
turbed; but the bones were not, or were only partially articulated. It seems
certain in all cases that the carcasses became severely macerated prior to being
covered by sediment. Most of them, however, fell victim to scavengers and the
skeletons became scattered. By far the gi'eatest number of specimens consist
of only a small portion of the shell ; surface erosion prior to the collection of the
specimens, no doubt, rendered them even less complete than they were, but I
do not believe that this is the primary cause for the great abundance of partial
specimens. In a few instances we have direct evidence in field observation that
carcasses were scattered before burial. Specimens containing bones of the skull
or mandible are exceedingly rare, which is the general rule in fossil turtles.

The bui'ial conditions in the Niobrara Chalk may differ from those in the
Mooreville Chalk, to judge from the abundance of fine skulls and the rarity of
shells in the collections. In the Greensand of New Jersey the general situation
is profoundly different. The shells are, probably for the most part, preserved
in articulation.'

THE RELATIONSHIP OF THE TOXOCHELYIDAE
TO OTHER FAMILIES OF TURTLES

In the following an attempt will be made to outline the basic structui'al
plan of the Toxochelyidae in the light of the basic patterns of families such as
the Cheloniidae {sens, lat.) and the Thalassemyidae, in order to determine, more

' Two essentially complete shells of different turtles, not belonging to this family,
were recently collected with proper care and may be cited as evidence.

266

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

clearly than has hitherto been possible, the typical organization of the three
gi'oups.

Comparison of Thalassemyid with Toxochelyid Pattern

The Thalassemyidae are a group of late Jurassic, marine turtles, restricted
geogi-aphically (so far as is known) to Europe. The genera Acichelys, Thalas-
semys, Idiochelys, Chelonides, Hydropelta, and Pelobatochelys represent a group
of forms that undoubtedly are closely related to each other and to the Plesio-
chelyidae, as I had an opportunity to learn by examination of the famous Solo-
thurn materials and those from Cerin preserved in the Museum of Lyon, France.
The genera mentioned above are generally agreed upon as forming the family
Thalassemyidae, but some authors, as, for example, Lydekker (1889), include,
probably for convenience, the genus Tropidemys. This form is difficult to assign
to any presently recognized family, but it is almost certainly not a thalassemyid.

The above statement to the effect that the Thalassemyidae are closely re-
lated to the Plesiochelyidae is based on personal study of the species of both
Plesiochelys and Thalassemys from Solothurn.^

As far as the Solothurn species are concerned, the genera Plesiochelys and
Thalassemys can only be distinguished by characters of a mostly quantitative
nature, as follows:

Plesiochelys

No costo-peripheral fontanelles

Carapace highly arched

Anterior lobe of plastron normal

Medial suture between xiphiplastra

Plastron suturally attached to carapace

Plastron without lateral fontanelles

Vertebral shields relatively wider than in
Thalassemys

Thalassemys

Costo-peripheral fontanelles

Carapace fairly flat

Anterior lobe of plastron lacking epiplastra

and possibly entoplastron
Xiphiplastra free medially (except in Neu-

enburg specimen)
Plastron attached to carapace by dermal

connective tissue
Plastron in some specimens with lateral

fontanelles

Needless to say, differences of this kind can hardly be considered to justify
family distinction, but it must be pointed out that the species of Thalassemys
from the Solothurn quarries are among the most primitive of all thalassemyids.
The dilTerences stated are nearly all correlated with increased aquatic specializa-
tion in Thalassemys. There is little doubt as to the origin of the Thalassemyidae
from the Plesiochelyidae, but the details of this relationship appear to be compli-
cated and should be considered in future revisions of the pertaining groups.

For the comparison below it may be permissible to regard thalassemyid
turtles as a closely related group of marine forms of different degi'ees of aquatic

' Since Rutimeyer (1873) described the materials from Solothurn, much additional
material has accumulated, with the result that some of the problems that confronted Ruti-
meyer are now within sight of solution.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 267

specialization. The Thalassemyidae and the Toxochelyidae thus differ in the
following features of their organizations:

Thalassemyidae Toxochelyidae

Carapace

Vertebral shields always much wider than Vertebral shields always much narrower

pleural shields than pleural shields

Neural plates tend to be suppressed Neural series always normal

Usually one suprapygal Two suprapygals

Plastron

Distance between axillary and inguinal Distance between axillary and inguinal
notches 90 per cent or more of half the notches 60 per cent or less of half the

width of the plastron width of the plastron

The dimensional difference mentioned above for the plastron determines
the basic character of the ventral armor and is not affected by extreme aquatic
specialization in either family. Besides the differences in the shells of the two
families of turtles, there are important differences in the forelimb skeleton.
Idiochelys has the typical forelimb of a fresh-water turtle, whereas Toxochelys
possesses a flipper of essentially cheloniid structure and probable function.

Comparison of Cheloniid with Toxochelyid Pattern

Turtles of definitely cheloniid affinities are known from late Cretaceous time
to the present. The Cretaceous forms are, on the whole, more highly specialized
than the Recent representatives and some of those from Eocene deposits. Among
the latter some are more primitive, others more specialized than the living
species. The present evolutionary picture of this family is thus not one of
gradual increase in aquatic specialization from an original generalized group;
instead we seem to know primarily (but not exclusively) the advanced species
of branches that appear to have developed from generalized, central stock
repeatedly since Mid-Cretaceous time.

There are, beyond doubt, many similarities between the cheloniid turtles
(particularly Lepidochelys) and the Toxochelyidae, and the two groups are very
probably derivatives of generalized chelydrid stock. The morphology of the
Cretaceous genus Catapleura, as well as the living genus Lepidochelys is nearly
intermediate between the cheloniid and toxochelyid organizations; the trend of
specialization of the plastra of both these genera is, however, cheloniid, rather
than toxochelyid.

In view of this rather close relationship between the Cheloniidae and the
Toxochelyidae, generalized species may be difficult to assign. In terms of trends
of specialization, however, the two families diverge markedly. The most striking
difference lies in the shape of the plastron, which in Toxochelyidae remains
essentially cruciform, with a short axillo-inguinal distance and a relatively long
hyo-hypoplastral contact sutui'e. In the cheloniid plastron the axillo-inguinal
distance increases from primitive to advanced genera and the hyo-hypoplastral

268

FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

suture becomes progressively shorter (fig. 123). The epiplastra are always
very small in Toxochelyidae, always well developed in Cheloniidae.

The carapace of all cheloniid turtles is oval to elongated-cordiform. In
Toxochelyidae it is circular to oval, or broadly cordiform. Long and narrow
shells are not known. The carapace of all Toxochelyidae is fringed by well-

FiG. 123. Comparison of plastral differentiation in Toxochelyidae and Cheloniidae.
(a), primitive toxochelyid; (b), advanced toxochelyid; (c), primitive cheloniid; (d), advanced
cheloniid. Figures diagrammatic.

developed peripherals. In the Cheloniidae there is a trend toward reduction
of the peripherals to narrow bars. The neural series of the toxochelyid carapace
consists of 8-10 elements, either unkeeled or keeled; if keeled, it is serrated in
side view and may or may not have epithecal ossicles associated with the keel
elevations. Among the Cheloniidae keeled forms are exceptional (for example, in

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 269

AUopleurou and Peritresms) and if present, the keels are not serrated in side
view in such a way as to form keel elevations underneath the apices of the
vertebral shields (except in juvenile individuals of Lepidochelys) . There appears
to be a trend, furthermore, toward fragmentation of the neural plates, so that
their number is notably increased {Lepidochelys, Procolpochelys).

With the exception of Osteopygis, the toxochelyid skull has a technically
primary palate; that of the Cheloniidae, so far as definitely known, has a sec-
ondary palate (but see remarks concerning "Toxochelys" gigantea, p. 260). In
the toxochelyid skull small nasal bones are present, in the Cheloniidae these
are reduced.

At the present state of knowledge, the Toxochelyidae and Cheloniidae
differ with regard to the normal pattern of cervical central articulation. The
normal formula in the Toxochelyidae corresponds with that of the Recent
Chelydridae (see p. 155). In the cheloniid turtles the most common pattern
shows plane joint surfaces between centra 6 and 7 (Williams, 1950).

The above discussion thus leads to the conclusion that the Toxochelyidae
and the Cheloniidae are marine families of similar origin that existed together
during Cretaceous time. The direction of specialization in the marine habitat
differs, however, considerably in the two families.

MORPHOLOGICAL AND PHYLOGENETIC

RELATIONSHIPS AMONG THE

TOXOCHELYIDAE

If the over-all organization of the turtles of the family Toxochelyidae is
considered, it may be noted that the genera differ from each other in featui'es
that are, in one way or another, connected with aquatic specialization. The
thickness of the shell, the circular or oblong outline of the carapace, fontanelliza-
tion, the presence of a mid-dorsal carina, and the development of a secondary
palate are all features of aquatic adaptation. None of the genera known at
present are specialized in all of these features but, as expected, they combine
primitive with advanced characters.

If we combine the generalized features of all toxochelyid turtles we arrive
at an unkeeled form with heavy shell plates and oval carapace outline and with
all carapace plates joined by sutures as in some specimens of Osteopygis. The
plastron would have small lateral and umbilical fontanelles and be excavated
at the axillary and inguinal notches to the extent observed in Porthochelys ; its
lateral attachment to the peripherals, however, would be as in Osteopygis; the
skull would have a smooth, primary palate as in Toxochelys. A comparison of
the genera with this morphotypic form is shown in figure 124.

There is no evidence in this group that the trends of specialization of the
features mentioned wei'e changed or reversed. The outline of the carapace of
Toxochelys, for example, is nearly circular in the relatively primitive T. moore-

270 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

tillensis as well as in the much more specialized T. lotirernis and T. atlantica.
Figure 124 shows that only two genera (LopJwchelys and Ctenochelys) are similarly
specialized in such a way that the more advanced Ctenochelys could have been
derived from the more generalized condition, as in Lophochelys.

A faint trend toward keeled neurals is seen in young specimens of Toxo-
chelys, where they are very gently arched from side to side. It is possible that
the keeled genera originated from early Toxochelys stock. Since all of the genera,
with the exception of Osteopygis, occur in the same deposits, namely, the Moore-
ville Chalk of Alabama and the Niobrara Chalk of Kansas, and very probably
at the same stratigraphic levels within these formations, we have to assume that
their differentiation lies farther back in time.

The relationship between the species, occurring at different stratigraphic
levels, is of notable interest, since it is different from what might have been
expected. The stratigraphically oldest species of Toxochelys is T. latiremis from
the Niobrara, yet it is as highly specialized or more so than the youngest member
of the genus, T. atlantica from the Upper Greensand of New Jersey. The better
known species of Toxochelys, arranged according to their stratigraphic age and
their relative degrees of specialization, produce the following picture:

Order of decrease in stratigraphic age: T. latiremis, T. moorevillensis, T.
barberi, T. atlantica.

Order of increase in specialization: T. moorevillensis, T. barberi, T. atlantica,
T. latiremis.

T. moorevillensis, T. barberi, and T. atlantica are forms that occur along the
open waters of the Gulf and Atlantic coasts, whereas T. latiremis lived in a
semi-isolated inland basin. We may thus assume that T. latiremis underwent
specialization earlier and at a faster rate than its relatives along the Gulf coast, a
phenomenon possibly correlated with limited population size in the Niobrara Sea.

The three species of Prionochelys, galeotergum (Niobrara), matutina (Moore-
ville Chalk), and nauta (Marlbrook Marl), show a different picture. In this
series, the stratigraphically oldest species is also the most primitive, while the
youngest is the most advanced form. If P. matutina and P. nauta are compared,
there can be little doubt that they are very closely related; in every feature
P. matutina shows a somewhat more primitive condition than does P. nauta.
The situation strongly suggests a direct ancestor-descendant relationship between
these two species. The Niobrara form, P. galeotergum, although more primitive
than P. matutina, differs from the latter in particular details of keel specializa-
tion (see fig. 114) and represents a different line within the genus.

The evidence in the cases of Toxochelys and Prionochelys strongly suggests
a progressive isolation of the Niobrara Sea from the open Gulf waters, resulting
in the evolution of endemic species in the inland sea. If we consider, along with
the faunal picture of the Niobrara Sea, that of the later Pierre Sea, we may
note that Toxochelys browni can readily be interpreted as a direct descendant of
r. latirernis; if so, it appears to have increased in size considerably. Further-

271

272 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

more, T. broivni, represented in the record by a fair number of specimens (skulls),
appears to be the only toxochelyid turtle in the Pierre deposits; Porthochelys,
Lophochelys, Ctenochelys, and Prio7iochelys seem to have become extinct in this
inland sea. This may constitute an interesting parallel to the situation in
the protostegid turtles (Part III, this volume), where there is likewise a reduc-
tion in the number of species to one, and an increase in the size of the surviving
form {Archelon ischyros).

SUMMARY AND CONCLUSIONS

1. The present account represents a monographic study of the family
Toxochelyidae, including a restudy of nearly all specimens of toxochelyid turtles
known.

2. Since the last revision of the family a significant amount of new material
has accumulated, primarily from the Mooreville Chalk of Alabama. All of the
new materials are described.

3. The comparative osteology of the toxochelyid skeleton in general could
be worked out satisfactorily, save for minor details. The skull and the cervical
portion of the vertebral column are known in considerable detail. Small nasal
bones are definitely present in Toxochelys and Porthochelys. The normal pattern
of cervical central articulation is identical with that most commonly observed
in Chelydridae.

4. A functional analysis of the type of locomotion in Toxochelys compared
to that of fresh-water turtles (such as Chelydridae and Trionychidae) and sea
turtles (Cheloniidae, Protostegidae, and Dermochelyidae) led to the conclusion
that Toxochelys was capable of both types of propulsion, namely, of fast, short-
distance locomotion by means of rapid paddling with the hind limbs, and of
cruising by slow, simultaneous, vertical strokes of the forelimbs. The latter are
interpreted as flippers, but appear to have been somewhat less specialized than
in Recent cheloniids.

5. As here understood, the family Toxochelyidae contains three sub-
families with the following genera:

A. The Toxochelyinae include the forms without a mid-dorsal carina
on the carapace, and with a smooth primary palate: Toxochelys,
Thinochelys gen. nov., and Porthochelys.

B. The Osteopyginae include the forms with uncarinated shells, but
with extreme secondary palates: Osteopygis and ?Rhetechelys.

C. The Lophochelyinae include the forms with carinated shells, and
with palates having partially undershelved choanal passages:
Lophochelys gen. nov., Ctenochelys gen. nov., and Prionochelys
gen. nov.

6. Lytoloma angusta Cope, L. jeanesi Cope and Chelone sopita Leidy are
considered to be nomina vana.

ZANGERL: VERTEBRATE FAUNA OF SELMA FORMATION 273

7. PJujUenujs is synonymous with Toxochehjs. Catapleura is thought to
be a primitive cheloniid. Sinemys, regarded by Nopcsa as a relative of Osteopygis,
is not a toxochelyid, but might be an advanced plesiochelyid turtle. Cynocercus is
a very doubtful toxochelyid.

8. Toxochelys brachyrhina Case is interpreted as a badly flattened skull and
T. serrifer Cope (non Case) as a nearly uncrushed partial skull of Toxochelys
latireniis. Toxochehjs gigantea Oertel is probably a cheloniid turtle.

9. The shell of Lytoloma wielandi Hay was recognized as Toxochelys
atlantica sp. nov. The skull and mandible belong to Osteopygis.

10. Porthochelys brotvni Hay is a species of Toxochelys.

11. The eight species of Osteopygis recognized by Hay from the Greensand
deposits of New Jersey are interpreted as belonging to one species.

12. Toxochelys serrifer Case (non Cope), T. bauri Wieland, and T. elkader
Hay are tentatively regarded as synonymous with T. stenopora Hay; these are
carinated forms belonging to the genus Ctenochelys. Toxochelys procax Hay is
likewise a member of the genus Ctenochelys.

13. The geographical distribution of the toxochelyid turtles follows the
shores of the southern half of the East Block of the Cretaceous North American
Continent. In vertical direction the family ranges from Coniacian time to the
close of the Cretaceous and, possibly, into the early Tertiary.

14. There is evidence (circumstantial) that the different species of toxo-
chelyid turtles buried in the same formation occupied different habitats, some
possibly the region that produced the burial ground.

15. There is no satisfactory evidence as to the probable food habits of
toxochelyid turtles.

16. A possible ecological effect of the mode of locomotion of Toxochelys
(see point 4) may have been the widening of the home range of the individuals.

17. The shells of most adult specimens of toxochelyid turtles show more
or less severe parasitic lesions ; there are no clues to the nature of the parasite.

18. The families Toxochelyidae and Thalassemyidae are not closely
related; the two groups differ in the basic features of their organization. The
families Toxochelyidae and Cheloniidae, on the other hand, are rather closely
allied.

19. All of the genera (except Osteopygis) of the Toxochelyidae are co-
existent in the Mooreville Chalk of Alabama and the Niobrara Chalk of Kansas.
The morphological relationships between the genera are illustrated in figure
124, along with the morphotypic condition of the skeleton of the group as a whole.

20. The relationships between the species show a notable endemic differen-
tiation of the Niobrara species, indicative of a progressive isolation of the Nio-
brara Sea from the open waters of the Gulf. In the Pierre Sea the toxochelyid
fauna appears to be reduced to one species {Toxochelys browni) of large size — a

274 FIELDIANA: GEOLOGY MEMOIRS, VOLUME 3

possible parallel to the reduction in the number of species and increase in size
of the survivor among the Protostegidae (Protostega-Archelon).

21. Prionochelys nauta can be interpreted as a direct descendant of P.
matutina.

REFERENCES

The systematic-descriptive literature to the Toxochehidae is hsted under each species.
Only general references and studies cited in the section on comparative anatomy are listed
below.

Abel, Othenio

1919. Die Stamme der Wirbeltiere. Berlin and Leipzig, xviii+914 pp., 669 figs.

Case, E. C.

1898. Toxochehjs. Univ. Kansas Geol. Surv., Pal., 4, pp. 370-385, pis. 79-84.

Cagle, F. R.

1944. Home range, homing beha\ior and migration in turtles. Misc. Publ. Mus.
Zoo!. Univ. Mich., 61, pp. 1-34, 2 pis.

Capellini, G.

1884. II Chelonio Veronese Protosphargis reronensis. Mem. Reale Accad. dei Lincei,
Rome (sci. nat.), 36 pp., 7 pis.

Fraas, Eberhardt

1903. Thalasseynys marina E. Fraas aus dem oberen weissen Juj-a von Schneitheim
nebst Bemerkungen iiber die Stammesgeschichte der Schildkroten. Jahresh.
Ver. vaterl. Xaturk. Wurtt., 59, pp. 72-104, 3 figs., pis. 1-3.

Hay, 0. P.

1898. On Protostega, the systematic position of Dermochehjs, and the morphogeny of

the chelonian carapace and plastron. Amer. Nat., 32, p. 963, fig. 2.
1908. The fossil turtles of North America. Carnegie Inst. Wash. Publ., 75, iii+568

pp., 704 figs., 103 pis.

KtJHNE, K.

1936. Die Zwillingswirbelsaule (Eine erbgenetische Forschung). Zeitschr. Morph.
Anthrop., 35, pp. 1-376.

Lagler, K. F.

1943. Food habits and economic relations of the turtles of Michigan. Amer. Midi.
Nat., 29, pp. 257-312, 9 figs.

Leidy, Joseph

1865. Cretaceous reptiles of the United States. Smithsonian Contr. Knowl., 14,
pp. 104-105.

Lortet, L.

1892. Les reptiles fossiles du Bassin du Rhone. Arch. Mus. Hist. Nat. Lyon, 5,
pp. 5-28, pis. 1-3.

Lydekker, Richard

1889. Catalogue of the fossil Reptilia and Amphibia in the British Museum (Natural
History). III. Chelonia. London, xviii+235 pp., 53 figs.

Nick, Ludwig

1912. Das Kopfskelett von Dermochelys coriacea. Zool. Jahrb. Abt. Anat. Ontog.
Tiere, 33, pp. 1-238, 16 figs., 12 pis.

275

276 FIELD lANA: GEOLOGY MEMOIRS, VOLUME 3

Owen, Richard

1849-84. A history of British fossil reptiles. London. Vols. I and II. xv+657pp.,
figs. 1-6, 48 pis.

RtJTIMEYER, L.

1873. Die fo.ssilen Schildkroten von Solothurn. Neue Denkschr. Schweiz Ges. Natur-
wiss., 25, pp. 1-185, 17 pis.

ScHiNZ, H. R., and Zangerl, Rainer

1937. Beitrage zur Osteogenese des Knochensystems beim Haushuhn, bei der Haus-
taube und beim Haubensteissfuss. Denkschr. Schweiz Naturf. Ges., 72, (2),
pp. v+ 117-165, 28 figs., 4 pis.

Siebenrock, Friedrich

1897. Das Kopfskelett der Schildkroten. Sitzungsber. K. Akad. Wiss. (Math. Nat.),
106, (1), pp. 245-328, 6 pis.

VerWiebe, W. a.

1933. Present distribution and thickness of Mesozoic systems. Bull. Geol. Soc.
Amer., 44, pp. 827-864, 4 figs.

WiELAND, G. R.

1888. The terminology of vertebral centra. Amer. Jour. Sci., (4), 8, pp. 163-164.
1900. Some observations on certain well marked stages in the evolution of the testudi-

nate humerus. Amer. Jour. Sci., (4), 9, pp. 413-424.
1902. Notes on the Cretaceous turtles Toxochelys and Archeloyi, with a classification

of the marine Testudinata. Amer. Jour. Sci., (4), 14, pp. 95-108, 2 figs.
1904a. Structure of the Upper Cretaceous turtles of New Jersey: Adocus, Osteopygis

and Propleura. Amer. Jour. Sci., (4), 17, pp. 112-132, figs. 1-7, pis. 1-9.
1904b. Structure of the Upper Cretaceous turtles of New Jersey: Lytoloma. Amer.

Jour. Sci., (4), 18, pp. 183-196, figs. 1-6, pis. 5-8.

1905. A new Niobrara Toxochelys. Amer. Jour. Sci., (4), 20, art. 36, pp. 325-343,
pi. 10, figs. 1-8.

1906. The osteology of Protostega. Mem. Carnegie Mus., 2, (7), pp. 279-298, figs.
1-8, pis. 31-33.

Williams, Ernest E.

1950. Variation and selection in the cervical central articulations of living turtles.
Bull. Amer. Mus. Nat. Hist., 94, (9), pp. 505-562, 20 figs., 10 tables.

Zangerl, Rainer

1939. The homology of the shell elements in turtles. Jour. Morph., 65, pp. 386-406,

9 figs., 2 pis.
1948. The vertebrate fauna of the Selma Formation of Alabama. I. Introduction.

II. The pleurodiran turtles. Fieldiana, Geol. Mem., 3, (1 and 2), pp. 1-56,

16 figs., 4 pis.

INDEX

PAGE

Catapleura 147, 267

chelydrina 208

Chelone sopita 207

Cheloniidae 267

Ctenochelys 227

acris 242

procax 247

stenopora 237

tenuitesta 230

Cynocercus incisivus 178, 260

Erquelinnesia molaria 147, 209

Euclastes platyops 215

Lophochelyinae 216

Lophochelys 217

natatrix 218

niobrarae 220

venatrix 224

Lytoloma 145

angusta 145, 196, 209

jeanesi 146

platyops 216

wielandi 146, 196, 209

Osteopyginae 205

Osteopygis 206

borealis 208

chelydrinus 208

emarginatus 208

erosus 208

gibbi 208

platylomus 208

robustus 208

sopita 208

PAGE

Phyllemys 147

barberi 193

Porthochelys 202

browni 197

laticeps 204

Prionochelys 248

galeotergum 258

matutina 254

nauta 249

Propleura borealis 209

sopita 209

Rhetechelys 147

platyops 215

Sinemys lens 261

Thalassemyidae 266

Thinochelys 199

lapisossea 200

Toxochelyidae 173

Toxochelyinae 173

Toxochelys 174

atlantica 196

barberi 193

bauri 238

brachyrhina 178

browni 197

elkader 238

gigantea 260

latiremis 178, 197

moorevillensis 186

procax 238,247

serrifer 178, 237

stenopora 238

weeksi 194

sp. (Taylor Marl) 198

277

EXPLANATION OF PLATE 12

Dorsal and ventral views of skull and mandible of Toxochelys latiremis (A.M.N.H.
5118), the best-preserved skull known to date. One of hyoid elements lies across pterygoids.
Nasal bones not visible.

Fieldiana: Geology Memoirs, Volume 3

Plate 12

Fieldiana: Geology Memoirs, Volume 3

Plate i;

Skulls and mandibles of Toxochelys latiremis in dorsal and ventral view. Upper row, Y.P.M. 3604
middle row, A.M.N.H. 1497; lower row, Y.P.M. 3609.

Fieldiana: Geology Memoirs, Volume 3

Plate 14

CMHM P27338

CNHW PR2I3

:A :s-;<*'

Skulls and mandibles of Toxochelys moorevillensi

EXPLANATION OF PLATE 15

Figs. 1 and 2. Dorsal and medial views of lower jaw (Y.P.M. 728) tentatively referred
to Toxochelys atlantica.

Fig. 3. Dorsal view of type specimen of Toxochelys latiremis. Photograph by courtesy
of American Museum of Natural History.

Fig. 4. Poorly preserved jaw fragment of Osteopygis emarginatus (A.M.N.H. 1133).
This specimen is not the type of Lytoloma angusta; note question mark on label in Cope's
handwriting. Photograph by courtesy of American Museum of Natural History.

Fieldiana: Geology Memoirs, Volume 3

Plate 15

, 20min

EXPLANATION OF PLATE 16

Fig. 1. Dorsal views of mandibles of Osteopygis emarginatus: la, Y.P.M. 913; 16, A.M.
N.H. 2216; Ic, Y.P.M. 490; Id, Y.P.M. 1001.

Fig. 2. Fragments of mandible of Toxochelys latiremis (A.M. N.H. 1835, type of Toxo-
chelys serrifer): 2a and 26, dorsal and ventral views (uncrushed).

Figs. 3 and 8. Anterior and posterior views of quadrate of Toxochelys latiremis (A.M.
N.H. 1835).

Figs. 4 and 9. Ventral and dorsal views of maxilla of Toxochelys latiremis (A.M.
N.H. 1835).

Figs. 5 and 10. Ventral and dorsal views of base of braincase of Toxochelys latire7nis
(A.M.N.H. 1835).

Figs. 6 and 11. Ventral and dorsal views of parietal-frontal area of skull of Toxochelys
latiremis (A.M.N.H. 1835).

Figs. 7 and 12. Not definitely identifiable skull fragment.

Photograph by courtesy of American Museum of Natural History.

Fieldiana: Geology Memoirs, Volume 3

Plate 16

Fieldiana: Geology Memoirs, Volume 3

Plate 17

Ctenochelys stenopora (A.M.N.H. 6137, type of Toxochelys elkader). Skull and mandible in

dorsal and ventral views.

Fieldiana: Geology Memoirs, Volume 3

Plate 18

__go_

Ctenochelys acris (C.N.H.M. P27337). Skull fragments and mandible, dorsal and ventral

Fieldiana: Geology Memoirs, Volume 3

Plate 19

20

%.,00^\

mm

Ctenochelys procax (C.N.H.M. UC614). Skull and mandible in dorsal and ventral views.

Fieldiana: Geology Memoirs, Volume 3

Plate 20

i r^ 'i ~^W^

^

G a

^y% w^ -^ -v^^ % ^'m %s

Vertebrae of toxochelyid turtles. A, cervical vertebrae 2 to 8 of Toxochelys latiremis (C.N.
H.M. PR123): Aa, in dorsal view; Ab, in side view; Ac, in ventral view. B, cervical
vertebrae of Toxochelys moorevillensis (C.N. H.M. P27391): Ba, posterior, Bb, anterior. Be,
lateral, Bd, dorsal. Be, ventral views of eighth cervical vertebra (note the ribs) ; Bf, centrum
of seventh cervical vertebra in anterior \new. C, eighth cervical vertebra of Toxochelys
latiremis (C.N. H.M. PR123) in anterior view. Da and Db, antero-ventral and lateral views
of eighth cervical vertebra of Ctenochelys acris. Ea and Eb, dorsal and ventral views of cen-
trum of sixth cervical vertebra, and Ec and Ed, dorsal and ventral views of centrum of first
shell vertebra of Ctenochelys tenuitesta (C.N.H.M. P27361). F, Ctenochelys acris (C.N. H.M.
P27352), caudal vertebrae: Fa, caudal vertebra in dorsal view; Fb and Fc, dorsal and lateral
views of another caudal; Fd and Fe, dorsal and lateral views of a third, more posterior
caudal vertebra. G, Toxochelys moorevillensis (C.N.H.M. P27391): Ga and Gb, posterior
shell vertebra in dorsal and lateral view; Gc and Gd, ?second sacral vertebra in dorsal and

lateral aspect.

Fieldiana: Geology Memoirs, Volume 3

Plate 21

Juvenile specimen of Toxochelys latiremis, K.U.(V.P.) 1244.

o

o

Fieldiana: Geology Memoirs, Volume 3

Plate 23

^ ;?9q|#^'.'

Carapace of Toxochelys moorevillensis. Upper left C.N.H.M. P27391, upper right C.N.H.M.
PR167, lower left C.N.H.M. PR28, lower right C.N.H.M. PR136.

PL,

o

o

Fieldiana: Geology Memoirs, Volume 3

Plate 25

Carapace of ThinocJielys lapisossea (C.N.H.M. P27453).

Fieldiana: Geology Memoirs, Volume 3

Plate 26

V

Carapace and plastron of Osteopygis emarginatus (Y.P.M. 778). Carapace in ventral view.

Fieldiana: Geology Memoirs, Volume 3

Plate 27

Prionochelys nauta (C.N.H.M. P26237). Neural carina in dorsal and side views, two costal
plates, and portions of peripheral plates.

Fieldiana: Geology Memoirs, Volume 3

Plate 28

'**'«a•^

Parts of carapace and plastron of Prionochelys nauta (C.N.H.M. P26238).

Fieldiana: Geology Memoirs, Volume 3

Plate 29

^'i°%^

Peripheral plates of Prionochelys nauta (C.N.H.M. P27456).