UNr RSITY OF
ILLlK^iS LIBRARY

NATURAL HIST. SURVEY

FIELDIANA • GEOLOGY

Published by
CHICAGO NATURAL HISTORY MUSEUM

Volume 10 March 30, 1955 No. 21

FAUNA OF THE VALE AND CHOZA: lo

Trimerorhachis: Including a Revision of Pre- Vale Species

Everett Claire Olson

Research Associate, Division of Paleontology
Professor of Vertebrate Paleontology, University op Chicago

INTRODUCTION

Specimens of the rhachitomous amphibian Trimerorhachis are
relatively abundant in beds of Vale and early Choza age in north
central Texas. Of the tetrapods only Dimetrodon, Diplocaulus, and
Lysorophus outnumber this genus in our collections. A large pro-
portion of material of Trimerorhachis, however, has come from two
sites, one in the middle Vale and the other in the lower Choza.
The Vale specimens are all fragmentary, but those from the Choza in-
clude skulls and partial skeletons in a reasonably good state of
preservation. The latter provide a basis for a taxonomic evalua-
tion at a level not practical for the Vale specimens.

Evaluation of the taxonomic position of the new materials has
required a restudy of collections from lower horizons, for, as has
been apparent for many years (see Romer, 1947), the taxonomy of
this genus is badly in need of revision. The collections of the
Museum of Comparative Zoology, the American Museum of Natural
History, the University of Michigan, and Chicago Natural History
Museum were made available for this study. I am grateful to
these institutions for the opportunity to study their materials and
to the many staff members who facilitated the work.

As yet no specimens of the genus Trimerorhachis have been
found in post-Clear Fork beds. One specimen, CNHM-UR 155,
from the San Angelo has been tentatively assigned to the family
Trimerorhachidae (Olson and Beerbower, 1953). It consists of the
posterior part of a small lower jaw. If this assignment is correct,
the range of the family as now known is from the Putnam Formation
of the Wichita group to the middle part of the San Angelo For-
mation of the Pease River group.

No. 745 225

226 FIELDIANA: GEOLOGY, VOLUME 10

STRATIGRAPHIC AND GEOGRAPHIC DISTRIBUTION

The genus T rimer orhachis is now known from eight of the nine
formations included in the Wichita and Clear Fork groups of Texas.
The specimens considered in this report comprise individuals from
all of these formations, but do not include those collected from the
Arroyo and Vale in Taylor and adjacent counties of Texas under
the direction of the University of Texas (Wilson, 1948). As yet
the genus has not been identified in the lowest formation of the
Wichita group, the Moran. The sequence of formations involved
is as follows:

Clear Fork Group Wichita Group

Choza Clyde

Vale Belle Plains

Arroyo Admiral

Lueders Putnam

Moran

The lower five formations, largely non-marine in the area under
consideration, and the marine Lueders are considered to belong to
the Wichita group in many reports on the Texas Permian. The
Clyde and Lueders, however, are assigned to the Clear Fork by
some authorities (Romer and Price, 1940, among others). For the
purposes of the present report, the assignment of formations to one
group or the other is not of particular importance, but it is my
opinion that the sedimentological and faunal characteristics of the
Clyde suggest affinities with the Wichita rather than the Clear
Fork formations. Non-marine phases of the formations of the
Wichita group are well exposed and fossiliferous in Archer clay,
Baylor and Wilbarger counties, Texas. Fossiliferous Clear Fork
beds are extensively exposed in Baylor, Wilbarger, Knox, and Foard
counties. A general northeast-southwest strike is maintained over
this area and there is a gentle dip that averages about fifty feet to
the mile. Most of the localities that have produced fossils, except
those in the Vale and Choza, are shown on the map published by
Romer (1928). Vale and Choza localities are given in my earlier
papers (Olson, 1948, 1951a-d).

Moran Formation. — There is no definite evidence of Trimeror-
hachis in this formation. Materials so identified in the field have
been found to belong to an unnamed genus, which is similar to
Chenoprosopus and characterized by an elongated snout and a

OLSON: FAUNA OF VALE AND CHOZA 227

pattern of dermal sculpture rather different from that of Trimeror-
hachis.

Putnam Formation. — Both Trimerorhachis and the unnamed
genus mentioned from the Moran occur in this formation. The
former has been positively identified on only one slab, MCZ 1381,
from Table Branch, Archer County, Texas. Associated with Tri-
merorhachis at this locality is the long-snouted, undescribed amphi-
bian. The material is fragmentary. On the Fall School land
(section lb). Archer County, Texas, some scraps that suggest the
presence of Trimerorhachis have been found. Most specimens from
the Putnam previously identified as Trimerorhachis actually belong
to the other genus.

Admiral Formation. — Trimerorhachis occurs at a number of lo-
calities in this formation and, for the first time, has assumed a place
of prominence in the fauna. The finest specimens have come from
Rattlesnake Canyon, in northwestern Archer County. A slab con-
taining a series of skulls and skeletal parts was obtained from this
locality by the University of Michigan and Harvard University
(see Case, 1935). The sample from the Admiral used in the present
study came in large part from this slab. Specimens from other
localities are generally fragmentary.

Belle Plains Formation. — Trimerorhachis is abundant in the Belle
Plains. So far as is known, the time represented by this formation
marked the numerical climax of the genus. The sample used in
the current study was assembled from a number of sites, including
deposits along the upper reaches of the North Fork of the Little
Wichita, below Fulda, at Tit Mountain and along Slippery Creek
(see Romer, 1928, and Romer and Price, 1940, for locations).
About half of the skulls used in the analyses that follow occur on a
single slab from the North Fork of the Little Wichita, AMNH
7116. At various places in the Belle Plains, fragmentary remains
of Trimerorhachis are concentrated in great abundance.

Clyde Formation. — Relatively few specimens of Trimerorhachis
have been found in the Clyde, although fragments have been ob-
tained in a number of places and one skull has been recovered. In
large part, the material is not suitable for analysis of its specific
affinities.

Arroyo Formation. — Remains of Trimerorhachis occur throughout
the Arroyo Formation, but are scattered, and, for the most part,
fragmentary. Reduction from the Belle Plains, both in absolute
and relative numbers of specimens, is striking. The greatest accu-

228 FIELDIANA: GEOLOGY, VOLUME 10

mulation of specimens known is in a locality near the mouth of
West Coffee Creek. All of the skulls, five in number, that are
suitable for study, have come, with one possible exception, from
beds that are high in the Arroyo section. Fragments are present
at all levels.

Vale Formation. — The contact of the Arroyo and Vale formations
is considered to be at the top of a series of beds of even red shales in
western Baylor County (Olson, 1951a). At many exposures it is diffi-
cult to separate the two formations on physical grounds alone. Where
the contact is clear, it has been found that certain species are
characteristic of the Vale and may be used as horizon markers with
some confidence. The most important from the standpoint of fre-
quency and ease of recognition are Labidosaurikos barkeri and
Gnathorhiza dikeloda. Trimerorhachis occurs near the top of the
Arroyo and in basal beds of the Vale. It is encountered occasionally
in both stream and pond deposits throughout the latter, but the
greatest concentration is in a small series of channel fills in the "fish
quarry," locality KF. A number of fragmentary specimens have
been found in a small pond deposit in locality KH. Both of these
sites appear to be middle Vale in age. At both, specimens consist
of disarticulated and broken parts of jaws, skulls, shoulder girdle
plates, vertebrae and limb bones.

Choza Formation. — Virtually all of the Choza specimens of Tri-
merorhachis well enough preserved for detailed study have come
from a single quarry at the "pipe site" in locality FA (Olson, 1951a).
This site lies approximately 150 feet above the base of the formation.
Scattered remains have been found as high as middle Choza.

REVISION OF PRE-VALE TRIMERORHACHIS

The morphology of the skull and skeleton of Trimerorhachis
has become well known through the studies of Cope (1878, 1896a,
1896b), Case (1911, 1935), and others. Only a few details can be
added to this knowledge in the current report. These are noted in
discussion and figures of particular specimens later in the text.
Seven species have been named, as follows: T. insignis Cope (1878),
T. bilobatis Cope (1883), T. mesops Cope (1896), T. conangulus
Cope (1896), T. leptorhynchus Case (1902), T. alleni Case (1910),
T. medius Broom (1913).

The characters that were used to distinguish the various named
species have not been reviewed systematically since the genus was
revised by Case (1911). T. leptorhynchus was discarded, properly,

OLSON: FAUNA OF VALE AND CHOZA 229

as indeterminate by Case at that time. Many additions have been
made to collections since 1911, so that studies based on fair-sized
samples are now possible. Although there is continuity of the genus
from the earliest to the latest occurrence, it has been found con-
venient to divide this study into two parts, one that deals with the
pre- Vale specimens and the other that considers the Vale and Choza
finds, new materials heretofore unreported in detail. The final
section is devoted to a synthesis of known data into a summary of
the evolution of the species.

REVIEW OF THE PRE-VALE SPECIES

Class AMPHIBIA
Subclass Apsidospondyli

Order Rhachitomi

Family Trimerorhachidae

Genus Trimerorhachis^

Trimerorhachis insignis Cope (1878)

Type. — AMNH 4565. Rather poor skull, lower jaws with front
part missing, 24 vertebrae, some with ribs, humerus, and scapula.
Included in the specimen listed as the type, when it was examined,
were bones belonging to other genera and fragments of several
individuals of Trimerorhachis. There is no certainty that all parts
listed above as the type belong to the same individual. All parts
that could reasonably be assumed to belong to the described speci-
men have been retained under AMNH 4565. The remainder has
been segregated for renumbering.

Specimen referred (by Case). — AMNH 4557. Skull and lower
jaws.

Horizon and locality. — Belle Plains Formation, Tit Mountain,
Archer County, Texas.

Revised description (Case, 1911). — Posterior angle of jaw com-
posed of a single tuberosity, an outer tuberosity indicated by a small
protuberance. Skull somewhat expanded posteriorly. Orbits in
anterior half of skull.

' Of the seven species, all but T. medius Broom were reviewed by Case in
1911. The history for each of these six species is given in his report and is not
repeated here except for the details necessary to an understanding of its taxonomic
position.

230 FIELDIANA: GEOLOGY, VOLUME 10

Trimerorhachis bilobatis Cope (1883)

Type.— AMNE. 4562. Angle of mandible.

Specimens referred (by Cope). — AMNH 4558, a jaw labeled as a
paratype. Cope refers to two other specimens in his original de-
scription, but the numbers are not given.

Horizon and locality. — Belle Plains Formation, Tit Mountain,
Archer County, Texas.

Revised description (Case, 1911). — Posterior angle of the jaw
formed by two subequal tuberosities separated by a deep groove.
Inner side of inner tuberosity vertical, with strong, horizontal keel.

Trimerorhachis mesops Cope (1896a)

Type. — AMNH 4568. Imperfect skull and vertebral column.

Horizon and locality. — Arroyo Formation, West Coffee Creek,
Baylor County, Texas. This locality places the specimen in the
upper third of the Arroyo Formation.

Revised description (Case, 1911). — Inner side of inner tuberosity
continued inward horizontally; no keel. Skull not expanded pos-
teriorly. Orbits almost exactly in middle of skull.

Comments. — The skull is in rather poor condition with the right,
posterior one-fourth missing. Crushing is not extensive. It is
probable that the posterior margin as preserved does not represent
the original margin. The preserved part is well forward of the otic
notch and the posterior end of the parasphenoid lies back of the end
of the dorsal platform. The posterior margin of the platform, how-
ever, is smooth, and only its abnormal relative position suggests
that part may be missing.

Trimerorhachis conangulus Cope (1896b)

Type.— AMNH ^569. Skull.

Horizon and locality. — Arroyo Formation. Two localities are
given (on the label), Pony Creek and Indian Creek, both in Baylor
County, Texas. It appears virtually certain from the type of
preservation that the Indian Creek locality, rather than Pony Creek,
is correct, and that the specimen comes from deposits of late Arroyo
age.

Revised description (Case, 1911). — Smallest of genus. Skull not
so fiat in postorbital region. Intertemporal present. Doubtfully a
Trimerorhachis.

OLSON: FAUNA OF VALE AND CHOZA 231

Comments. — This is an excellent small skull. It is certainly
immature, for there is poor ossification in the occipital region, except
for the exoccipitals. This has made generic placement somewhat
difficult. The pattern of the dermal elements of the skull appears
to be comparable to that of other specimens of Trimerorhachis.
There is no postero-lateral (sensory) groove on the lower jaw, but
this may be an age character. This specimen clearly cannot be
assigned to any known genus other than Trimerorhachis and, since
its fundamental characters are definitely those of Trimerorhachis,
there seems little doubt that this is the correct assignment.

Trimerorhachis alleni Case (1910)

Type. — AMNH 4577. Four dorsal vertebrae.

Horizon and locality. — Admiral Formation. East side of Rattle-
snake Canyon, Archer County, Texas.

Description (Case, 1911). — Larger than in any others. Neural
spines show little indication of being divided. Pleurocentra propor-
tionately small. Intercentra with deep pits to either side of median
keel. Posterior edges of upper ends of intercentra reflected forward,
forming a concise facet.

Trimerorhachis miedius Broom (1913)

T^/pe.— AMNH 4895. Skull.

Horizon and locality. — There is some uncertainty concerning the
horizon and locality. The specimen comes from the north side of
the Little Wichita River, but there is conflict in the records with
respect to the exact locality. The specimen appears to be either
from the early Belle Plains or the late Admiral of Archer County,
Texas.

Summary of Broom's description. — Differences from T. insignis,
according to Broom, are as follows: frontal relatively wider, jugal
approaching the orbit more closely, suprasquamosal relatively rather
large, parietal smaller, squamosal relatively smaller; the eyes slightly
farther back, the whole skull relatively narrow.

CONSIDERATION OF THE "SPECIFIC" CHARACTERS

The six species were named by three individuals over a period
of 35 years. New materials were collected during this period and
concepts changed. The bases for the various species were characters
shown by one or two individuals in most cases (four in one instance) .

232 FIELDIANA: GEOLOGY, VOLUME 10

Comparisons were made for the most part with only one or two other
species. Adequate samples for studies of distributions of characters
did not become available until the last decade of the period. Little
attempt was made to coordinate the various observations. Even in
Case's revision (1911) no comprehensive comparison was made, but
it must be realized that he was undertaking the almost impossible
task of bringing together and unscrambling a series of studies that
had accumulated since about 1875.

Inevitably a rather lengthy list of more or less unrelated "dif-
ferences" was assembled during the 35-year period. Confusion was
increased by the fact that structures that showed supposed specific
differences were not present in the types of specimens of all the
species. The characters that were used may be summarized as
follows:

Skull Characters:

1. Size.

2. Proportions, especially width compared to length, either in
an all-over sense or involving posterior width only.

3. Curvature of the dorsal surface.

4. Position of the orbits with relation to the anterior and
posterior ends of the skull.

5. Proportions of various skull elements.

6. Elements present in the dermal roof of the skull.

Lower Jaw Characters:

1. Posterior angle of the jaw, whether it carries one or two
tuberosities.

2. Presence or absence of a groove on the posterior and ventral
surface of the jaw. This associated with the presence or
absence respectively of the second tuberosity.

3. The existence and orientation of a keel on the postero- ventral
surface of the jaw, also associated with the condition of
tuberosities.

Vertebral Characters:

1. Size.

2. Neural spine divided or single.

3. Relative size of pleurocentra.

OLSON: FAUNA OF VALE AND CHOZA 233

4. Presence or absence of lateral pits and keel on intercentra.

5. Distinctness of anterior facet on inflected dorsal parts of
intercentra.

All available material has been restudied, in an effort to de-
termine the significance of the various characters in this array.
A total sample of 47 skulls from the Admiral, Belle Plains, and Arroyo
formations was assembled. In addition, fairly large samples of
lower jaws, vertebrae, and limb bones from various localities were
used. Analysis of this material and its bearing on the taxonomy of
Trimerorhachis from pre- Vale deposits are considered in the following
paragraphs.

ANALYSIS OF TAXONOMY OF FRE-VALE TRIMERORHACHIS

Skull Characters

T. conangulus, T. mesops, T. insignis, and T. medius have been
separated, in part, on the basis of characters of the skull. T. bilobatis
was based on characters of the lower jaw and T. alleni on vertebral
characters, and cannot be considered in this section. Size, propor-
tions of major skull dimensions, position of orbits, and proportions
of skull elements are all suitable for quantitative analyses based on
linear measures. Frequency distributions, bivariate scatter dia-
grams, and regression lines have been used in the present study for
this purpose. If distinctions of specific importance exist, they should
be revealed, as the various dimensions are studied by these means.
To effect such a study all of the Admiral, Belle Plains, and Arroyo
skulls have been treated together, including the type skulls of T.
mesops, T. conangulus, and T. medius and the paratype of T. insignis,
for which the type skull is not well enough preserved to provide
suitable measurements. Fourteen measurements were taken on the
skulls. These include the dimensions that were used earlier to dif-
ferentiate species, so far as this was possible, as well as others.
The measurements used are shown graphically in figure 91. Values
for most of the Admiral and Belle Plains specimens were published
earlier (Olson, 1953). The previously unpublished values for speci-
mens from these formations and those for Arroyo specimens are
given in Table 1.

Studies were carried out in two general ways, one based on
frequency distributions of linear dimensions that provide information
on the factor of size, the other, on scatter diagrams and regression
lines for analyses of proportions. These studies cover items 1, 2, 4,
and 5 in the list of skull characters (p. 232) but do not have a bearing

CO lo • 00

•^

Oi IX> O^ (M CO

•S

CO

. • U5 rH •

iO y-l

• 1— (

U3

t~ CO ko • •

c
o

CO

o

CO

t>«00 «CI»0-(>q

oi n ■ y-t -^ 7-t • n

(MOO-O Tj<c^].U3

COOJ->— I U3i-H-C^

OOCXD-C^ CO(J5-<3i

(M CO • ■>* • r-l

CO

o

C3

^_i

05

S

■ T-H

•(<»

-«

►-I

Q

*-^

-si

CO

■^

• CO

T-H

1

6

CO

• C^J

H

g

►J

^^

s

• o

. ?D •

M

E^

o

• .— 1

<3

««-i

H

o

"5

o

• T-H

^

m

"o

►5

«o

c^

■ O U5

^-^

"-I

• '-'

• <M (M

S

S

"c

1

c

Q-

CO CO •

a> ^ Tj< ,-H

CO

iSJSiSJSiSJS s. s. s. s. §. §.

tr" T^ T"" T^ 'T* Z"^ P^ 1^ lO rO »0 (^

PhAhPhP^PhAh OOOOOO
%~t %-t %^ %^ %-t %^

== 1=::=:^ s:=5 <! <J <J <! <tj <J

pqpqpQpqpQpq

O) 00
CO «o
lO lO lO
T}< ■>* ;d

ILi ■^ "^ "^ "^ t^ "^

0) O O O O 1-H O

J2 ^ --I --- r-H ^ — I

3 UOOOOU

'Z ti t:) ;:2 & 1:2 ID

M ffi .-I t~ 00 OS
tt; C ^, CO ixi 00

Z Z CSI CO CO CO

c^ c4 C3 c^ c4
N N N N H

o o o o o

^ j3 -C -C J3

ooooo

00 t>

oi CO -^ CO CO 00

CO t- t- i-H T-H t-

05 tf P5 tf CtJ Oh
ID |D|D|D|Dt2

234

OLSON: FAUNA OF VALE AND CHOZA

286

on dorsal curvature or differences in elements present. Crushing
tends to alter curvature materially in most skulls of Trimerorhachis
and renders it an unreliable character for recognition of small
differences such as those that may have existed between members

,(-INA^-»|

Fig. 9L Diagram of skull of Trimerorhachis in dorsal aspect with key to
position and abbreviations of all measures used except Skd, the depth of the skull
at the level of the anterior termination of the otic notch.

of the sample. Examination of the sutures of T. conangulus, which
has been thought to have a pattern somewhat different from that
of others, indicates that the usual suite of bones is, in fact, present,
and that the relationships are normal.

1. The Factor of Size: The use of such terms as large and
small, or their comparatives and superlatives, in taxonomic descrip-
tions without specification of what is meant is always a source of
confusion. Presumably all-over size is implied, but even this in
itself is not a clear concept. In the use of linear dimensions as a
basis for a concept of size it must be assumed (unless there is a state-
ment to the contrary) that equal weight is given to each dimension.
For example, it is possible to think of the dimensions of a complex
object, such as a skull, in many ways, in linear, areal, or volumetric
terms. It would appear that most unqualified references to size

236 FIELDIANA: GEOLOGY, VOLUME 10

imply linear dimensions. There seems, as well, to be a subconscious
tendency to group measurements into a series of more or less loose
orders, one involving the maximum of such dimensions as skull
length, width, or depth, a second involving lesser dimensions of
major visible divisions of the skulls, and a third involving highly
varied arrays of small measures such as dimensions of individual
bones, apertures, etc. Impressions of size, as implied in the use of
comparative terms, generally seem to involve some combination
of the first order mentioned, influenced to some degree by measures
of the second order. If this is the case, it seems reasonable to use
some linear measure of the first order that is highly correlated with
other gross measures of this order, to test the distribution of size.
It is probably true, in many cases, that various dimensions that
contribute to a concept of size do not contribute with equal weight
to the impressions reported, but it is impossible to evaluate the
nature of any such bias in many descriptions. Fortunately, when
one or more dimensions have dominated the thinking of the describer,
this tends to be indicated by such statements as "proportionately
broader," "not so wide," or "long and narrow."

The statements about skull size of Trimerorhachis seem to have
been made in general accord with the assumptions and reasoning
of the last paragraph. If this is true, the factor of size may be
studied by the use of frequency distributions of skull length, since
this is a dimension of the first order that may be shown to be rather
highly correlated with other gross dimensions (see Olson, 1953).
This measure has been used in the following determinations.

The series of skulls of Trimerorhachis includes an array of growth
stages. A low value for skull length may express immaturity of a
specimen or it may indicate a small adult. Unless there is evidence
on the state of maturity, usually in the degree of ossification, a
decision may be difficult. The fact that frequency distributions of
a linear dimension on members of a growth series may be bimodal
or trimodal (see Olson and Miller, 1951, for examples) renders such
diagrams invalid as the sole evidence for a decision. It is unsafe
practice to erect a new species of amphibian on the basis of some
expression of gross size alone and any such determinations must be
viewed with skepticism in the absence of other information.

Figure 92 presents a single frequency distribution based on skull
length, as defined in this paper, for the skulls of Trimerorhachis
from the Admiral, Belle Plains, and Arroyo formations, and distri-
butions for samples from each of these formations separately.
Throughout this study, the hypothesis that there exists a strati-

oM

m

Wm

m

'M

35 45 55 65 75 85 95 105 115 125 135 145 155 165 175

A

SK ARROYO

W/.

^

W/

^

y/A

%

1

m

35 45 55 65 ^5 85 95 105 IIS 125 125 145 155 I6S 175
SK, ADMIRAL

35 45 35 65 75 85 95 105 115 125 135 145 155 185 175
Q SK, ADMIRAL + BELLE PLAINS + ARROYO

Fig. 92. Histograms showing frequency distribution of skull lengths of
Trimerorhachis from pre- Vale deposits; 1 = 10 mm. A, Arroyo skulls; B, Belle
Plains skulls; C, Admiral skulls; D, composite distribution of skulls from Admiral,
Belle Plains, and Arroyo formations.

287

238 FIELDIANA: GEOLOGY, VOLUME 10

graphic arrangement of species has been considered, and tests have
been made where appropriate. The composite frequency distribution
in figure 92, D, is of the type frequently encountered in growth
series of amphibians and reptiles where some gross linear dimension
is plotted. All specimens in this distribution could belong to the
same species. Most questionable, perhaps, is the inclusion of the
skull in the 170-179 class, but the large size is not prohibitive,
particularly in view of the fact that specimens of this general dimen-
sion and somewhat smaller are known from jaws, suggesting that
the sample based on skulls has a somewhat misleading bias. One
of the principal characters of T. conangulus, cited by Case, is size.
Although the type skull is the least in length of any in the sample,
there is no assurance that it might not be merely an immature
individual, for certainly juveniles could be expected to have skulls
less than 35 mm. in length at an early stage of ontogeny. The fact
that the skull is poorly ossified and obviously immature gives support
to its inclusion with the others.

2. Proportions: Skull proportions raise problems that are more
complex than those involving size alone, for two or more variables
are necessarily involved. A reasonably satisfactory method of pre-
liminary treatment of proportion is possible through the use of
bivariate scatter diagrams. These, plus a more refined treatment
of regression lines for pairs of characters that show some promise
for discrimination, have been used for the study of proportion in
this paper. Because growth series are involved, ratios cannot be
directly used with any hope of satisfactory results.

Relative posterior width and positions of the orbits relative to
the midpoint of the median longitudinal line of the skull roof appear
in the description of T. mesops. Frontal width, position of the
jugals relative to the orbit, size of the suprasquamosal (supra-
temporal of this paper), size of the squamosal, size of the parietal,
position of the orbits, and relative width of skulls are mentioned by
Broom as definitive of T. medius. All of the measurements and rela-
tionships, except those of the jugal and squamosal, are covered by
measures used in this paper as shown in the tabulation below. After
initial attempts to measure the jugal and squamosal had been made,
efforts to obtain quantitative data from them were abandoned,
since distortion and breakage were too great in most instances to
allow confidence in the measurements. These elements are thus
not covered, but the fact that no reliable information could be
obtained, suggests that their use as a basis for diagnosis is highly
questionable. The characters covered by the measures are as follows :

OLSON: FAUNA OF VALE AND CHOZA 239

Position of orbits: relationships of 0-Sl (orbito-snout length) to Ski (skull
length).

Posterior width of skull: Skw (posterior skull width).

Size of suprasquamosal= supra temporal: Stl (supra temporal length).

Size of parietal: Pal (parietal length), and Paw (parietal width).

The above measures and others, as shown in figure 91, have been
studied by use of bivariate scatter diagrams of all possible pairs of
measures. Skulls from each formation were entered in the original
diagrams in different colors. These diagrams then were analyzed
visually to separate them into four groups, as follows:

1. Highly correlated (r, sample correlation coefficient, estimated
to be >0.9).

2. Less well correlated (r estimated to be <0.9).

3. Showing some evidence of separation of groups from different
formations.

4. One or a few specimens appearing to fall out of the main
cloud of points. The diagrams in this category fall into one
of the other three if the rejected specimens are eliminated.

An example of each type is shown in figure 93. Relationships
that fell into groups 1 and 2 (above) were considered to be common
to the genus in the pre- Vale deposits. Those in group 1 show a
higher level of association of pairs in relative size than the pairs of
measures grouped under 2. Separation into groups in type 3 occurred
only in comparisons of Admiral and Belle Plains specimens, but
detection of differences between the Arroyo specimens and the earlier
ones would be possible, in view of the small size of the Arroyo
sample, only in the event that there was a very strong and uniform
divergence. The scatter diagrams that fell into group 4 did so in
all cases by virtue of the inclusion of certain Arroyo specimens.
With these specimens removed, all diagrams placed in group 4
could be included in one of the other three groups.

The 14 measurements provided a total of 91 pairs for study.
Sixty-eight fell in groups 1 and 2. Sixteen showed one or more
specimens well outside of the main cloud of points and were referred
to group 4. The scatter diagrams of eighteen pairs of measures
seemed to show some evidence of separation into two groups of
points. These were placed in group 3. Certain of these eighteen
were originally in group 4, because of the inclusion of Arroyo speci-
mens, and were allocated to group 3 upon the deletion of these
specimens. The relationships of the regression lines fitted to the
apparently separate groups of points, Admiral and Belle Plains

40
33
30
2S
PO, 20

•

c

»

••<

)

o

•

»

B

A

3 15 20 25

CLASS I

• PRE-ARROYO
(3 ARROYO
O CHOZA

30
25

20

0, 15

10

5

Q

(

©
> •

•

©•

•
•

•

•
•

B

10 15 20 25 30
CLASS 3

40

35

30

25

PO, 20

15

10

5

•
• •

e

•

^

'

©

•

o» (.

•

n

c

45

40

35

30

0-S|25

20

IS

10

5

D

15 20 25 30 35 40 45

0-5,
CLASS 2

(j

I

(

>•

©

©

o

.

•

•

►

20 25 30 35 40 45

FR,
CLASS4

Fig. 93. Sample bivariate scatter diagrams of four classes used in specific
determination of pre- Vale skulls. A, Class 1 : high correlation. B, Class 3: showing
some evidence of separation between regressions of specimens from Admiral and
Belle Plains formations. C, Class 2: moderate to low correlation. D, Class 4:
with one or more points appearing to fall out of the main cloud (s) of points.

240

OLSON: FAUNA OF VALE AND CHOZA 241

respectively, were studied by analysis of variance, following the
method outlined by Olson and Miller (1951). The following four
pairs of measurements of the total eighteen were found to have
differences considered to be significant on the basis of the probability
levels shown below:

1. OZ on low (test of hypothesis that regression lines were the
same):

F=9.3 (nl = 2, n2=18), P <0.01

test of hypothesis that hyx (slope) is the same:

F=0.5 (nl = l, n2=18), P >0.1

The difference lies in the position of the lines, ayx.

2. 01 on Pol (test of hypothesis that regression lines were the

same) :

F=6.85 (nl=2, n2=17), P <0.01

test of hypothesis that hyx is the same:

F=6.63 (nl = l, n2=17), 0.05 > P >0.01

The difference probably lies in the slope of the lines.

3. Ow on 0-Sl (test of hypothesis that hyx is the same) :

F=8.25 (wl = l, n2=22), P <0.01
The difference is in the slope of the lines.

4. Ow on Frl (test of hypothesis that the lines are the same) :

F=5.75 (nl = 2, n2=13), 0.05 > P >0.01
test of hypothesis that hyx is the same:

F=2.86 (nl = l, n2=13), P< 0.05

If lines are considered different on basis of probability in first test,
then the difference lies in position of lines.

All of the other relationships tested showed F ratios with proba-
bilities (that the lines were the same) >0.05. The type of T. medius
was included in the 18 scatter diagrams tested. This species was
separated from T. insignis on the basis of several skull proportions
by Broom (see p. 231). Five of these differences were tested by the
study of pairs of measures. In no case did the type of T. medius
fall either out of the main cloud of points or out of one of the main
clouds where there was a suggestion of two. Unless the four dif-
ferences that have been noted are considered of specific significance,
it must be concluded that T. medius is synonymous with T. insignis.
All of the differences involved either orbital width or orbital length.

242 FIELDIANA: GEOLOGY, VOLUME 10

That the sample from the Admiral and that from the Belle Plains
differ in proportional features of the orbit is evident. But differences
of this magnitude are common, almost to the point of being charac-
teristic, between local populations of species of amphibians at the
present time; witness, for example, the differences between the
samples of Rana pipiens from the northern and southern parts of
its geographic range. It seems to be entirely unwarranted to suggest
a separation into two species, even though there is a stratigraphic
difference, on the basis of the minor differences that have been
found to exist.

The series of relationships that fall into class 4 of the scatter
diagrams pose problems of analysis for which there appears to be
no simple statistical solution. Each specimen that departs from the
cloud of points must be considered a sample of one, since the rela-
tionships between them are not known. In the larger sample, that
of T. insignis, with which comparisons are made, there is a change
of proportion between the pairs of measures, as x, considered the
independent variable, increases in size. At any given value of x,
the independent variable y has a variance. This might be used to
estimate the probability that the value of y of the single specimen
belongs in the distribution. Even this test, which involves somewhat
dubious assumptions concerning the independent variable, is not
possible, for the larger sample does not have sufficient values of y at
any given value or sufficiently small range of x to provide a meaning-
ful estimate of population variance.

In view of this difficulty, a solution by estimation based on
inspection, without estimation of probabilities, has been made. A
chart including all measures that could be obtained on specimens in
question was compiled as follows: By inspection of the bivariate
scatter diagrams, points for the individuals were accepted as belong-
ing to or rejected from the cloud of points. If the point lay well
outside the cloud, it was rejected. A was entered into the chart
for accepted points and R for those rejected. For rejections in
which the point lay "above" the cloud of points (i.e., the value of
y was disproportionately high for the value of a:) a + was entered,
and for those where the point lay "below" the cloud, a — was entered.
From these charts, the diagram shown in figure 94 was compiled,
including all rejections encountered. Nineteen of 55 pairs studied
showed one or more rejections. The rejections occurred for three
Arroyo specimens, as follows: AMNH 4568, type of T. mesops;
AMNH 4569, type of T. conangulus; and CNHM-UC 688.

Vd

IS

M

01

Ol

'Od - '^OJ

'V<S - 'Vd

M

- aj

'tij

01 - O-ld

'tiJ

o-ld

MS - 'o-ld

o-ld

M
MS

I

M,

VN

M

VN

dl - "VN

M,.

Vd

M,

VN

M

yj - VN
*MS -'^VN

O-ld -'s-0
-•s-0

-'S-0

's-0

'wj -'s-0
'V)S -'s-0

<dl

Vd

<

4

O

<

+

o

<

+

o

1

<

o

-f

-H

o

+

<

o

1

1

o

1

+

o

1

<

o

4-

<

o

+

+

+

H-

<

o

+

<

o

+

<

o

+

+

+

+

+

o

+

<

+

-f

<

o

+

<

o

+

H-

o

+

•+-

H-

T ^

I I

Z Z

< <

«n

1-

V)

7

H

Z

O

Q.

o

U.

a.

O

u.
O

o

D

o

O

3

_)

O

o

-1

>

<

o

O

(0

1

<

(Q

n

m

in

z

z

o

o

;

«0

«n

z

z

H

»-

->

D

(O

CO

cc

oc -

\-

(-

in

<0

—

h-

Q

o

Z

z

tf)

<

o .
oc ■

ll.

UJ

2
u

UJ

f-

H

<

UJ

2

Q.

o

U

UJ -

a

O

<

a:

z

S

m

.So

o <v
.2 o

U CO

< +1 O -o

o o

I

243

244 FIELDIANA: GEOLOGY, VOLUME 10

Four measures show a high incidence of rejection: Inaw, 0-Sl,
Pi-Ol, and Frl. Relatively few measures were possible on UC 688,
but the pattern was similar to that in the other two so far as it could
be determined. Since each of the skulls shows a similar pattern of
rejections, a relationship is suggested. More important than the
mere fact of rejection, however, is the direction of deviation. The
general consistency for the three specimens is shown in figure 94.
Of the eight rejections common to AMNH 4568 and 4569, six are
of common sign and two, Pi-Ol-Frl and low-Frl, are different.
Ten rejections of AMNH 4568 are present in 4569, and four
rejections (all involving low) for AMNH 4569 are not found in
4568. The skulls of these two specimens are very different in length,
37 and 97 mm., respectively, for 4569 and 4568. The number of
rejections with a common sign, in view of the probable changes in
proportions to be expected in growth, is high. It seems unlikely
that it would be encountered unless there was a close relationship
between the specimens involved. On the basis of the deviations
from the condition in T. insignis, used as a standard of comparison,
it is concluded that the two specimens are specifically different
from T. insignis. On the basis of the similarities of the deviations,
it is concluded that they pertain to the same species. UC 688 is
tentatively referred to this species as well, on the basis of the rela-
tively meager evidence available. These three specimens are con-
sidered to belong to the species T. mesops (Cope, 1896a); T.
conangulus (Cope, 1896b) is thus a synonym for T. mesops.

From studies of the skulls, two species are recognized : T. insignis,
with a range from the Admiral Formation through the Arroyo, and
T. mesops, from the upper part of the Arroyo Formation.

Jaw Characters

T. bilobatis Cope (1883) was established as a species because of
the presence of both a medial and a lateral tuberosity on the angle
of the lower jaw. At the time this species was named, only one
other species, T. insignis, had been recognized. Both types came
from the same locality. Tit Mountain, in Archer County. The
difference between the two lies primarily in the presence of a well-
developed lateral process or tuberosity on the angle of the jaw in
T. bilobatis and a smaller, and less distinct comparable process in T.
insignis. The structure of the angle of the jaw is mentioned by
Cope (1896b) in his description of T. conangulus. In the description
of T. mesops. Cope (1896a) noted that the inner side of the medial

OLSON: FAUNA OF VALE AND CHOZA 245

tuberosity is continued horizontally rather than vertically as in the
case of T. insignis. The ventral keel, present in T. bilobatis, and the
external process are poorly developed in T. mesops of Cope. Finally,
some attention, especially by Case (1911), has been given the
presence or absence and degree of development of a longitudinal
groove, which appears to have housed a latero-sensory canal, that
passes forward along the ventral margin of the jaw from a position
lateral to the medial tuberosity and medial to the lateral one.
Different specimens compared one with another with respect to
these features may present markedly different appearances.

The problem faced in a revision of taxonomy of the genus is
whether these differences are discrete and of taxonomic value, or
whether specimens that appear different can be considered end
members of a continuous distribution. The types of differences
between T. conangulus and T. insignis and T. bilobatis might merely
be a matter of difference in maturity. The distinctive features of
T. mesops cannot be explained in this way, although T. conangulus
and T. mesops again could be members of the same species, differing
because of age (the interpretation that is made on the basis of skull
features described earlier).

Our principal concern is the validity of the separation of T.
insignis and T. bilobatis, as made by Cope. There is no difficulty
in distinguishing the two type specimens on the basis of development
of the tuberosities (see fig. 95) . There now exist, however, excellent
samples that show the nature of the distribution of the development
of the processes, and give a basis for evaluating the importance of
the difference. The largest, N=43, is from the Belle Plains Forma-
tion at Rendham, in Archer County. Frequency distributions in
smaller samples, including one, where N=15, from Tit Mountain,
appear to be similar to that of the larger sample, so that only the
results based on the latter will be considered in detail.

Measures that are considered appropriate for a study of the
relative development of the processes are figured (see fig. 95).
Measurements for the Rendham sample are given in Table 2. The
inner process, roughly equivalent to the retroarcticular process of
reptiles, is present in all specimens, although it is highly varied in
relative length. The lateral process ranges from virtual absence
to essential equality in length with the medial process. Size fre-
quency distributions for both processes are shown (fig. 96, A, B).
They are roughly bell-shaped and do not show discontinuities or
bimodality that might suggest the presence of two species, separable

pq

0)

,

"a

•s"

g

e

oS

^

3

c

E-I

3

o

"o

s

Oi

4->

P

+->

s

<i'

s

M

.1

m
OS

tj"

"rt

1

o

03

U

O

_g

S

e8

«0

m

s

Cs

m

"S

J3

s

13

•^*i

C

0)

05

-«

O

e

S

-c

p

^

(h

o

«*-«

s~

OT

OS

^

)i

1

O

Q

^ (11 ^

>-i CO

rt

O 0) T3

o S ^
ago

S S S
J3 2 o

SO

0> • !*
.f-1.2.

246

.5 1.5 2.5 35 45 55 6.5
MEASURE NO. 2

i5 4.3 3.5 6.5 7.5 8.5 9.5 10.5 11.5

MEASURE NO. i
B

16
14
12
10
F 8
6
4

^

^

v//

i

y//

fl

^A

M

vyy/

i

^

'M

§

i

1

^

'M

^

^

.5 1.5 25 35 4.5 3.5 6.5 75 85

RATIO MEASURES N OS. \- 2

Fig. 96. Histogram showing frequency distributions of measures 1 and 2
and ratio of 1:2 for jaw sample from Rendham.

247

248

FIELDIANA: GEOLOGY, VOLUME 10

by development of the processes. The frequency distribution (fig.
96, C) for the ratio of the lateral to the medial process is strongly
skewed, with about 65 per cent of the values showing ratios between
1:1 and 1:3. Almost complete absence of the lateral tuberosity
occurs in two specimens, producing a ratio of about 1:10. For
specimens with a ratio approaching 1:3.5, the lateral tuberosity is
not well defined. It is believed that the ratios most clearly depict
the difference noted by Cope for the two species. The type of T.
bilobatis falls in the modal class of the frequency distribution of
ratios, whereas the type of T. insignis is in the 3.00-3.95 class.
Neither "species" represents the extreme condition that exists.
Clearly, in this frequency distribution of ratios there is no indication
of a separation of the specimens studied into two groups. It must
be concluded that the variability of the processes, while considerable,
is continuous, and that, in spite of strong departures from the norm
in the frequency distribution of ratios, there is no basis for recognition
of the distinct groups.

TABLE 2

Measurements (in mm.) of Posterior Region of Jaws of T rimer orhachis
T. insignis, type; T. bilobatis, type and paratype; and sample from Rendham

1, length of internal tuberosity; 2, length of external tuberosity;
5, width of articular facet.

Types and Paratypes

12 3

T. insignis, AMNH 4565 (type) 7.0 2.0

T. bilobatis, AMNH 4562 (type) 7.0 3.0

T. bilobatis, AMNH 4558 (paratype) 7.5 5.0

Sample from Rendham
12 3

9.0

2.0

13.0

8.0

1.5

15.0

8.0

5.0

12.0

5.5

6.0

15.0

7.5

5.0

17.0

6.0

3.0

12.0

4.5

2.0

11.0

10.0

0.0

17.0

6.0

3.0

12.0

6.5

1.0

21.0

7.0

3.5

13.0

7.0

1.0

14.0

6.0

4.0

9.5

7.0

3.0

15.0

9.0

6.0

18.0

9.0

2.0

18.0

8.0

3.0

12.0

3.0

0.1

25.0

7.0

1.0

15.0

8.0

4.0

12.0

7.0

4.0

18.0

5.0

0.5

13.0

8.0

5.0

15.0

7.0

4.0

22.0

7.0

1.0

17.5

6.0

1.0

13.0

5.0

0.5

17.0

5.0

3.0

13.0

11.0

5.5

12.0

9.0

3.0

13.0

5.0

3.0

17.0

7.5

5.0

17.0

6.0

6.0

19.0

9.0

5.0

20.0

6.0

1.0

11.0

7.0

2.0

16.0

9.0

4.5

14.0

8.0

3.0

15.0

9.0

5.0

14.0

8.0

4.0

11.0

5.0

4.0

12.0

6.5

3.0

16.0

7.5

3.6

15.0

OLSON: FAUNA OF VALE AND CHOZA 249

The development of keels and the ventral grooves on the jaws is
positively associated with the degree of development of the lateral
process. Thus the same arguments used in the study apply in the
consideration of these characters as a basis for specific differentiation.

It appeared possible, during the examination of the specimens,
that part of the apparent difference might relate to differences
in sizes of the jaws. To test this, it was necessary to select some
measure to express the gross size of the jaw. The width of the
articular facet correlates rather highly (r=0.8 to 0.9 in samples of
N=10 to 20) with jaw length, so this measure was used as an ex-
pression of size. Correlations carried out for the measures of each
of the processes and width of the articular surface proved not to
be significantly different from zero at P=0.1 (see Table 3). The
degree of development thus is not dependent upon size. In this
table, as well, the correlation of the development of the two processes
is shown to be 0.4, low but significantly >0, where P=0.01.

From the study of the jaws the following conclusions may be
drawn:

1. Each tuberosity on the angle of the lower jaw has a frequency
distribution that is unimodal and approaches normality.
There is no evidence of a separation into two or more groups.

2. The ratios of the lateral to the medial process are highly
varied but form a continuous, skewed distribution.

3. T. insignis and T. hilohatis cannot be distinguished on the
basis of the characters used to separate them in the definition
of T. hilohatis and, therefore, T. hilohatis, in the absence of
any other basis for separation, must be considered synony-
mous with T. insignis.

Vertebral Characters

T. alleni was based on four dorsal vertebrae. It is not possible
to be certain precisely what part of the column these four occupied,
so that exact comparisons are difficult. Size, undivided neural
spines, proportionately smaller pleurocentra, pitted and keeled inter-
centra, and anteriorly reflected upper ends of the intercentra, with
the presence of concise facets, are the characters noted by Case
(1911) as distinguishing these vertebrae from those of all other
species.

No comprehensive study of the vertebrae of Trimerorhachis has
been possible, because of the paucity of complete or even partially

o o
V A

CO CO rH

©

o

o

-H

-H

-H

(M

(M

T}<

(M

C^

CO

a>

CO

«o

1—1

g.

W

t-

00

t-

»— 1

i— 1

1-H

a

o

o

o

JS

H

b

-H

■n

-H

X!

OJ

"•J"

(M

S

CD

CO

CO

0)

«

"H

rH

i-t

.g

eg

as

OS

CO

•^

■*

<M

o

o

o

CO

c3

l»

-{\

^^

-H

S

(M

eg

U3

^

U3

T-H

U5

T-H

CO

m

Clj

<

•-S

Eh

"o

o
O

o

(M (M kO

-H ^^ -H

(M kO C<I

CO CO C^

m 03 CO

(V Oi 0)

s s s

s s s

1-1 (N (N

0) 0) (V

s § ;§

250

OLSON: FAUNA OF VALE AND CHOZA 251

complete columns. It is, thus, impossible to be certain that some
of the characters cited have any validity. The work that has been
possible, however, suggests that there is considerable danger in the
use of any of the characters, together or singly, as a basis for dif-
ferentiation. Each of the features is analyzed as far as possible in
the following paragraphs:

1. Size: The concept of size is subject to the same difficulties
as those outlined in consideration of the skull. The vertebrae of
T. alleni look larger than those of most of the other specimens. The
nature of comparative materials, however, has limited quantitative
studies to the intercentra. Greatest width and greatest length are
the characters used. They give a fair concept of the gross dimensions
of the intercentra. They do not, however, necessarily serve as
representative of the total network of measures involved in the
visual estimate of size, since there appear to be differential rates of
ossification, and hence bone growth, in different parts of the verte-
brae. It seems reasonable, in spite of this difficulty, that these two
measures will serve as a rough check of the validity of the factor of
size in the definition of T. alleni.

Vertebral measurements for several samples are given in Table
4. In figure 97, histograms based on size frequency distributions,
including the measurements of T. alleni, are shown. It may be
noted from the figures in Table 4 (for example, T. insignis from Tit
Mountain, AMNH 4565) that in the individual, length of inter-
centrum tends to decrease from the anterior to the posterior, whereas
width tends to increase. There is a significant negative corre-
lation, with r= —0.47. In samples of randomly collected inter-
centra, on the contrary, there is a positive correlation of width and
length (Table 5) . Evidently a complex sampling problem is involved
in these results, for, while it is evident that vertebral width increases
posteriorly while length decreases, it is also apparent that, when the
factor of growth is introduced, the tendency for width to increase
with length, in a random sample, overshadows the first effect. The
intricacies of this problem are not especially pertinent to the problem
at hand except to indicate that the position of points in the scatter
diagrams depends in part upon the position of the column from which
the measurement was drawn and in part on growth.

Because of the complexity of joint treatment of measures, not
subject to full solution with the materials available, our studies of
the size of vertebrae are based on separate studies of length and
width. In the histogram (fig, 97), depicting length, T. alleni falls
within the distribution of the other specimens. The case is less

TABLE 4

Length and Width of Intercentra (in mm.) of Trimerorhachis from Admiral,
Belle Plains and Arroyo Formations

Single Specimens

T. insignis, AMNH 4565 (type)
(Belle Plains)

T. alleni AMNH 4577 (type)
(Admiral)

L

10

11

11

11

11

10

11

9

9

10

9

W

16 Anterior

17

17

18

18

17

18

19

19

19

19 Posterior

L

W

16

29 Anterior

15

28

15

28

14

28

14

28 Posterior

Composite Samples

Hackberry Creek (Belle Plains)

Tit Mountain (Belle Plains)

L

W

L

W

L

W

L

W

L

W

L

W

L

W

7

17

9

17

10

18

9

18

11

21

12

22

14

22

8

18

9

19

10

19

9

16

11

21

13

22

15

24

8

16

9

19

10

18

10

21

11

20

13

23

8

16

9

20

10

22

10

19

11

19

13

23

8

18

9

17

11

21

10

21

12

22

13

23

8

15

9

17

11

20

10

20

12

23

13

23

9

18

9

17

11

21

10

20

12

21

13

20

9

17

9

18

12

19

10

20

12

21

13

21

9

19

10

18

12

31

11

19

12

24

13

20

9

20

10

18

12

19

11

21

12

21

13

23

9

18

10

17

12

20

11

21

12

24

14

22

9

20

10

19

12

21

11

20

12

22

14

25

9

18

10

20

12
12

21
18

11

21

12

22

14

26

Below Fulda (Belle Plains)

AMNH 4574 (A]

rroyo)

L

W

L

W

L

W

9

19

11

17

12

21

9

17

11

18

12

21

10

17

12

20

12

22

10

19

12

21

12

21

11

19

14

21

12

22

11

17

15

24

12

23

11

19

11

20

TABLE

5

Correlation Coefficients for Vertebral Measurements
Intercentral length (x) and intercentral width (y)

N X y aX ay r

Hackberry Creek 40 9.7dz0.21 18.6±0.26 1.33±0.15 1.62±0.18 0.58
Tit Mountain .. 41 11.9zh23 21.4±0.30 1.47±0.16 1.90±0.21 0.72

252

OLSON: FAUNA OF VALE AND CHOZA 253

clear when width is considered, for there is a slight discontinuity.
Unless this difference is supported by other evidence, it seems an
insufficient basis for specific differentiation. If it were accepted,
we would have to conclude that in all the material measured there
is but a single specimen of the species, in spite of the fact that ma-
terials are all more or less contemporary.

2. Other Characters: None of the other characters used in the
specific description are subject to quantitative analysis with the
materials available. In some specimens of Trimerorhachis neural
spines are divided anteriorly, while in others they are not. There
appears to be a gradation from deeply grooved vertebrae through
stages in which grooves are poorly developed, to T. alleni, in which
grooves are nearly absent. Neural spines of the more posterior
vertebrae appear to tend to be better ossified than those in the
anterior part of the column in those specimens in which serial ob-
servations can be made. The differences along the column reflect
a difference in the degree of ossification. Differences between partial
columns of various individuals seem to reflect a tendency toward
greater ossification in larger vertebrae, suggesting that the degree
of groove development is in part an age factor. The vertebrae of
T. alleni may well represent a series from the posterior part of the
column of a large, particularly fully ossified individual.

The relative size of vertebral pleurocentra varies throughout
the column, but there is not sufficient material to establish any trend.
The relative size also is a function of maturity, reflected in the degree
of ossification. No sound comparisons can be made, even with
samples more adequate than those available when T. alleni was
described, and the character must be considered indeterminate.

Keels and lateral pitting are highly varied on intercentra. Some
intercentra are keeled but lack pits, while others are pitted and
virtually unkeeled. Various intermediate conditions are encountered
in the various samples. The degree of pitting and the distinctness
of keels vary widely even within a single column, and there appears
to be some tendency toward variation between geographic localities.
In view of this degree of variability, it seems unsafe to consider
even the rather extreme condition in T. alleni as taxonomically
important. The strong development of anterior inflection and
facets on the vertebrae of T. alleni seems to be primarily a function
of large size and unusually complete ossification. Facets are poorly
developed to absent on small intercentra, but attain some distinctness
in large specimens.

8 5 10.5 125 14.5

LtNGTH : INTERCENTRA

T. INSIGNIS //^
T.ALLENI

185 20.5 22.5 24 5

WIDTH; INTERCENTRA

B

Fig. 97. Histograms showing frequency distributions of length and width of
intercentra of composite sample of T. insignis from Admiral and Belle Plains.

254

OLSON: FAUNA OF VALE AND CHOZA 265

Essentially all of the characters involved in the definition of
T. alleni seem to be associated with its large size and the presumed
position of the vertebrae rather far back in the column. Since it
has been indicated that the size alone is insufficient for specific
distinction, the other characters, related to size, cannot be considered
important. T. alleni, in my opinion, cannot be considered distinct
from T. insignis. It is thus dropped as a species on the basis that
it is indeterminate.

SUMMARY, CONCLUSION, AND TAXONOMY

The analyses that have been performed lead to the conclusion
that all determinable specimens of the genus Trimerorhachis from
the Admiral and Belle Plains formations must be referred to the
first-named species of the genus, T. insignis. The three other
species named from these formations, T. bilobatis, T. medius, and
T. alleni, have been shown to be indistinguishable from T. insignis
on the basis of the characters that were used to define the various
species.

Samples from the Belle Plains and Admiral have been compared
by the use of regression lines for 18 pairs of characters. Four of
the 18, compared by analysis of variance, show high probabilities
of a difference. Two of these regressions involve 01 and two, Ow.
These indicate the existence of differences in proportions of the
orbital measures as compared to certain other measures of the
skull. That only four relationships, of a total of 91 tested, are
involved, and that these four involve either 01 or Ow suggests that
the differences are minor. They could have been produced by slight
genetic differences or, perhaps, by the effects of environmental
differences upon ontogenetic processes. It is important in this
regard to recall the source of the samples. The sample from the
Admiral was drawn from a single slab and the probability that
genetic similarity of the individuals exceeded that of the total
population of the species is certainly high. About one half of the
Belle Plains specimens likewise were from a single slab, which also may
have introduced a lower than average variability of genetic structure
into the sample. Morphological differences between the two samples
are no greater than those often encountered today between local
populations of the same species. Therefore, it seems unwise, in
spite of the fact that there is a stratigraphic, thus temporal, difference
in the samples, to consider that they were drawn from two species
populations.

256 FIELDIANA: GEOLOGY, VOLUME 10

The specimens from the Putnam and Clyde formations do not
lend themselves to the types of study necessary to determine their
specific identity with any certainty. In neither case is there any-
thing known that would exclude them from T. insignis. They are
considered, tentatively, as belonging in the species.

The situation in the Arroyo is somewhat more complex. Dif-
ficulties are severely increased by the paucity of materials suitable
for analysis. Of the five available skulls, two, CNHM-UC 667 and
MCZ 1165, show no rejections from the clouds of points on bivariate
scatter diagrams based on relationships of pairs of skull measures
taken on specimens from the Admiral and Belle Plains. There is
no basis for separating these two specimens from T. insignis, and
they are thus assigned to it. Three specimens, AMNH 4568 and
4569 and CNHM-UC 688 have been shown to differ from T. insignis
on a number of counts. In spite of the discrepancies in the size of
these three specimens, which suggests age differences, there is a
marked consistency in the ways that they differ from T. insignis.
Two of the three named are type specimens, AMNH 4569 being
the type of T. conangulus, and AMNH 4568, the type specimen of
T. mesops. These two, plus UC 688, are referred to the first-named
species, T. mesops. Of the characters given for T. mesops by Cope,
the position of the orbits, more posterior on the skull with reference
to the midpoint of skull length, is valid for separating this species
from T. insignis as recognized in this paper, and thereby can be
considered a valid character for the species. None of the characters
given in Case's revised description of T. conangulus is diagnostic.
A new revised description is necessary.

Trimerorhachis mesops Cope (1896a)

r^/pe.— AMNH 4568. Skull.

Horizon and locality. — Arroyo Formation, probably from Indian
Creek, Baylor County, Texas.

Referred specimens. — AMNH 4569, Arroyo Formation, West
Coffee Creek, Baylor County, Texas. CNHM-UC 688, Arroyo
Formation, West Coffee Creek, Baylor County, Texas.

Revised description. — Significant differences in skull proportions
from T. insignis as follows:

1. 0-Sl, greater in proportion to Frl and Pi-Ol (adults and
juveniles).

OLSON: FAUNA OF VALE AND CHOZA 257

2. Inaw, greater in proportion to Ski and Skw (adults and ju-
veniles).'

3. Pi-Ol, lesser in proportion to Frl and low (adults and
juveniles).

4. 0-Sl, greater in proportion to Ski, Pal, Ipl, Skw (adults).

5. Inaw, greater in proportion to Frl, Pal, Ipl, Pi-Ol (adults).'

6. Pi-Ol, lesser in proportion to Ski (adults).

7. low, greater in proportion to Pol, Stl, Paw (juveniles, adults
unknown).'

8. 0-Sl, greater in proportion to Pol (juveniles, adults un-
known).'

Jaws and vertebrae are of little value at present for differentiating
this species from T. insignis. The character of the inner tuberosity
of the angle of the jaw in AMNH 4568 seems distinctive, but no
study of distribution of the character is possible. The characters
of the angle of the jaw in AMNH 4569 show no positive development,
presumably because the specimen is immature. Vertebrae from
the Arroyo appear to be less commonly pitted than those from the
Admiral and Belle Plains, but samples are too small for valid
comparisons, and reference of Arroyo specimens to a particular
species, which requires Icnowledge of the skull, is usually impossible.
Presumably the unpitted and unkeeled columns would be associated
with T. mesops, for it is known that the vertebrae associated with
some skulls and jaws of T. insignis do have pits and keels. In view
of the lack of actual association and the small number of columns
of more than two or three vertebrae, this suggestion must remain
conjectural.

VALE AND CHOZA TRIMERORHACHIS

Vale

Materials, Distribution, and Habitat: Specimens of Trimero-
rhachis occur throughout the Vale, from the base to beds transitional
to the Choza at the top, but have been found in concentration only
at two local sites. Most materials are fragmentary, so that specific
determination cannot be made with any certainty. A few specimens
have been found in fine conglomeratic deposits, formed in streams,
in the lowest Vale beds of western Baylor County, Texas. Fragments
representing at least a score of individuals have come from a small

1 This condition not clearly shown in UC 688.

258 FIELDIANA: GEOLOGY, VOLUME 10

pond deposit in the middle Vale and a single jaw is known from a
second, nearby pond deposit. At the latter site, Trimerorhachis is
associated with Xenacanthus, Eryops, and Dimetrodon. Two speci-
mens, a jaw (UR 49) and part of the back of a skull (UR 69, fig.
98, A), are from the uppermost Vale, the jaw from a stream deposit
and the skull from a bed whose genesis is uncertain.

The preponderance of Vale specimens comes from the small,
fine-grained channel deposits of the "fish quarry" in locality KF,
Knox County, Texas. Here occur hundreds of skeletal parts,
mostly fragmentary. Associated, in the deposits, are abundant
remains of the lungfish, Gnathorhiza dikeloda, and fragments of
Waggoneria, Dimetrodon, and undetermined captorhinomorphs.
Fragmentation of the specimens occurred before and during depo-
sition, probably as the materials were transported by the waters of
the streams. This group of animals cannot be considered to represent
a life assemblage.

It is clear that Trimerorhachis inhabited standing water, ponds
and small lakes during the Vale and lived in association with the
other pond dwellers of the time. There is no clear evidence that
the streams were a natural habitat. All materials found in stream
deposits are broken and often water worn. This does not, of course,
preclude the possibility that the streams actually provided a suitable
habitat, but there is a striking discrepancy between the condition
of preservation of Diplocaulus recurvatus, definitely a stream dweller,
and Trimerorhachis. The rather fragile skulls of Diplocaulus are
often well preserved, even in coarse clay-pebble conglomerates.

Taxonomy: The numbered specimens of Trimerorhachis from
the "fish quarry" are listed in Table 6. In addition, there are
several hundred fragments in blocks of siltstone taken from this
quarry and much uncollected material. Even with this abundance
of individuals, and that from other sites, as well, it is difficult to
make trustworthy specific assignment because diagnosis requires
fairly complete skulls. The lower jaws (fig. 103, A), have the charac-
teristics of T. insignis rather than those of T. mesops. The structure
of the posterior angle differs from that seen in the Choza specimens,
described below, in that two distinct processes are present, rather
than a single process that is a continuation of the outer jaw margin.
The best partial skull is that from the upper beds of the Vale (fig.
98, A), but crushing is such that it is impossible to be sure of the
original depth in the region of the otic notch, which provides an
important difference between T. insignis and the Choza specimens.

OLSON: FAUNA OF VALE AND CHOZA

259

Fig. 98. A, dorsal view of partial skull from upper Vale, CNHM-UR 69;
B, part of CNHM-UR 71, from Choza, in lateral aspect, showing swimming type
of tail and indicating general scale arrangement. Both X M.

Thus the only characters that can be used for assignment are those
of the posterior end of the lower jaw.

The high variability of this region in T. insignis has been noted.
The Vale specimens all fall within the limits of variability of T.
insignis, as far as relative sizes of medial and lateral processes are
concerned. The available sample, however, is small, consisting of
only four specimens that show this part of the jaw, so that the
amount of variability in Vale specimens is uncertain. On the basis
of the meager evidence available, the Vale specimens appear to

260 FIELDIANA: GEOLOGY, VOLUME 10

TABLE 6

Specimens of Trimerorhachis from the "Fish Quarry," Vale

CNHM CNHM

UR 67, partial maxillary UR 62, part of lower jaw

UR 46, lower jaw UR 52, half of large interclavicle
UR 45, partial lower jaw and part of lower jaw

UR 40, jaws and skeletal UR 42, partial maxillary

parts in block UR 48, part of lower jaw

UR 55, interclavicle UR 44, part of lower jaw

UR 50, part of lower jaw UR 64, part of lower jaw,
UR 47, part of lower jaw interclavicle, clavicle

belong to T. insignis. Clearly they cannot be assigned to T. mesops
or the Choza species. It is not out of the question, of course, that
they may represent a species intermediate between T. insignis and
the Choza species, or even that two or more species are actually
present in the Vale.

Choza

Materials, Distribution, and Habitat: Except for a few specimens,
all known materials of Trimerorhachis from the Choza have come
from the "pipe site" locality FA, in the early Choza. The collections
include 18 numbered specimens, and fragments of about an equal
number have been observed but not collected. The site is by no
means fully excavated, and it is quite certain that the preserved
specimens number in the hundreds.

The deposits at the "pipe site" are extremely interesting and
unlike those encountered anywhere else in the Wichita or Clear
Fork sections. The bed from which the fossils come is a deep red
shale, exposed on a series of hills over an area about an eighth of a
mile on each side. This bed is capped in many places by a thin, dense
siltstone. Overlying the siltstone is an orange-red shale that carries
gypsum biscuits. Underlying the producing layer, which averages
about four feet in thickness, is a brownish-red shale that is devoid
of fossils. The most distinctive feature of the bone-bearing shale,
outside of the fact that it has fossils, is the occurrence of two types
of vertically oriented ovoid structures. One type consists of mottled
red and green, hard nodules that range from one to four inches in
maximum diameter and up to eighteen inches in length. They are
rich in vertebrate remains. They appear to have formed around
the bones after deposition, presumably as a result of the reducing
action of organic materials. Their vertical orientation is somewhat
puzzling. The contained vertebrate remains are usually oriented

ii«^- ■

, ^w V

v-^yTiimr-^it^f^^

Fig. 99. Two views of deposits at pipe site of locality FA, Choza.

261

262 FIELDIANA: GEOLOGY, VOLUME 10

with their long axes vertical, and it seems quite certain that this
position was attained after the formation of the nodules. Specimens
of vertebrates free in the shale rather than in nodules are orientated
with their long axes horizontal and have been subjected to con-
siderable distortion and breakage. The most reasonable suggestion
seems to be that the vertical orientation of the nodules was developed
during compaction of the shales, which has been extensive, with
the long axes shifting to become normal to the compacting forces.

The other vertical structures (fig. 99) have been termed pipes
and have given the name to the site. They pass from the base to
the top of the fossiliferous bed and, at the top, are continuous with
the siltstone layer that caps the bed. Their composition is similar
to the capping siltstone. The pipes are oval in cross section with
maximum diameters from about four to twelve inches. Orientation
of the long diameters is random, even in the "clumps" in which the
pipes tend to occur. These structures do not contain vertebrates,
although specimens are found "plastered" to their sides and oc-
casionally penetrating the pipe for a short distance.

The origin of the pipes is by no means certain. It is thought
that they were formed in open holes by the introduction of the
sediment that formed the siltstone layer. Superficially, they re-
semble lungfish burrows of the Arroyo and Vale (Romer and Olson,
1954), but the shape, size, composition, weathering, and fossil content
differ. The most probable explanation is that these openings were
formed by the root-like shafts of some type of plant, possibly some
Equisetales, that rotted to leave openings that were later filled
by sediment. There is no question that the red shales surrounding
the pipes were deposited in a shallow pond, and such a pond might
well have provided suitable habitat for the type of vegetation
envisaged.

Trimerorhachis occurs free in the red shales, in the nodules and
in the overlying siltstone layer. Skulls ranging in length from 42
to about 150 mm. have been found, indicating the presence of a
growth series. Some well-preserved, nearly complete, vertebral
columns with ribs in place have been found. The site clearly
represents a life habitat of the genus. Very abundant remains of
the aquatic, burrowing amphibian Lysorophus occur in the shales
and the nodules. Vertebral lengths, based on centra, range from
1 to 12 mm. and thus represent the full known ontogenetic size
range of this genus. Specimens are not coiled, as is usual for skeletons
of Lysorophus preserved during the aestivating phase of their

OLSON: FAUNA OF VALE AND CHOZA 263

existence. It would appear that Lysorophus at this site existed in
the free-swimming phase of its life. Fragments of Dimetrodon and
some well-preserved specimens of Labidosaurikos harkeri and Cap-
torhinikos chozaensis occur in the bed with Trimerorhachis and
Lysorophus. No traces of Eryops, Diplocaulus, Gnathorhiza, or
Xenacanthus have been found in this deposit, although all were in
existence in the area at this time. Interesting, principally because
of the rarity of invertebrates in the Clear Forks beds, is the presence
of small, plain fresh-water ostracods in the nodules.

Other specimens of Trimerorhachis have come from sites about
a quarter of a mile south and about a half mile northwest of the
pipe site. Traces of the genus have been seen as high as the middle
Choza. At the site to the south, Gnathorhiza dikeloda occurs with
Trimerorhachis. They are in a stream deposit and the meaning of
the association is uncertain. In the deposit to the northwest, there
is association with Lysorophus and Labidosaurikos barkeri in a pond
deposit. The association is similar to that at the pipe site and the
general environment of deposition probably was much the same.

Taxonomy: The materials of Trimerorhachis from the Choza
provide an excellent basis for analysis of their taxonomic position.
It has been determined that they represent a heretofore undescribed
species, as follows:

Trimerorhachis rogersi^ sp. nov.

Type.— CNHM-UR 138. Partial skull and lower jaws, forelimb,
femur, and part of scapula (fig. 100).

Horizon and locality. — Pipe site, locality FA, early Choza, Foard
County, Texas.

Referred specimens. — All from pipe site.

UR 74, skull and jaws (fig. 101, C)

UR 76, skull and jaws (fig. 101, A, B)

UR 141, part of skull and jaws

UR 135, part of skull and jaws

UR 75, part of skull and jaws

UR 72, vertebral column, part of shoulder girdle, skull fragment

UR 71, much of skeleton and fragment of skull (fig. 98, B)

UR 73, lower jaw (fig. 104, A, C)

UR 136, lower jaw (fig. 104, B, D)

UR 137, part of skull, lower jaws, and scrap (fig. 101, D)

UR 133, lower jaw

UR 70, part of skull

' Named for Mr. Chester Rogers, of Knox County, Texas, who has materially
aided our work in the exploration of the late Vale and the Choza.

264

FIELDIANA: GEOLOGY, VOLUME 10

Fig. 100. Type specimens of T. rogersi sp. nov. A, skull in lateral aspect;
B, inner surface of jaw; C, left humerus, radius, ulna, and metacarpals; D, part
of skull roof, ventral view of inner surface of bone; E, right femur and part of
tibia.

Diagnosis. — Skull proportionately deeper in relation to length
and width than for any other species of Trimerorhachis; ratio of
depth, at otic notch, to length 1:4 or greater, in size-range of skull
length 70 to 145 mm. Ratio of skull width to skull length less than
in T. insignis or T. mesops. Lower jaw with single tuberosity on
angle; this tuberosity a direct continuation of lateral surface of
jaw, not set off by well-defined notch.

OLSON: FAUNA OF VALE AND CHOZA

265

Description and discussion. — The two features by which the
type and referred skulls can be differentiated from other species are
the relatively greater depth and proportionate narrowness of the
skull. The first is immediately evident to one who has had ex-

FlG. 101. Skulls of T. rogersi sp. nov., from Choza. A and B, CNHM-UR
76; X 1. C, CNHM-UR 74; X :^. D, CNHM-UR 137; X 3^.

perience with series of skulls of the species of Trimerorhachis, but it
is somewhat difficult to treat in a definitive quantitative way because
of the small number of skulls on which essential measurements can
be taken. A second difficulty arises from the fact that the ratio of
depth to length changes as these two dimensions increase during
growth. To show this relationship graphically, ratios for five speci-
mens of T. rogersi, three Arroyo specimens, and one from the Belle
Plains are plotted against skull length in figure 102. The difference
indicated is evident in this graph, but it will be seen that in both
T. rogersi and the earlier specimens there is an increase in the ratio

266

FIELDIANA: GEOLOGY, VOLUME 10

with increase in skull length so that a direct comparison of ratios,
by use of means, for example, would be misleading. Samples are
not large enough to produce meaningful results by analysis of
variance of regression lines.

The character of relationship of skull width to skull length, for
which a larger series of measures is possible for pre-Choza specimens,
is shown in the histogram of the ratios of Skw to Ski in figure 105.

S'^d'^N .15

-J

•

o

! ^2

o

20 AO 60 80 100 120 140 160 180 200

» CHOZA

O PRE-VALE

Fig. 102. Ratio of Skd.Skl plotted against skull length for five Choza and
four measurable pre- Vale skulls.

Fig. 103. Ventral aspect of posterior part of lower jaws of T rimer orhachis.
A, CNHM-UR 46, left jaw, Vale, T. cf. insignis; B, CNHM-UR 137, left jaw,
Choza, T. rogersi sp. nov.; C, CNHM-UR 136, left jaw, Choza, T. rogersi sp.
nov.; D, CNHM-UR 73, right jaw, Choza, T. rogersi sp. nov.; E, CNHM-UC
699, left jaw, Admiral, T. insignis; F, CNHM-UR 722, left jaw, Arroyo, T.
insignia; G, CNHM-UC 646, right jaw, Belle Plains, T. insignis. All X 3^.

OLSON: FAUNA OF VALE AND CHOZA 267

Here, also, the factor of change of ratio with size enters in, but the
change is less radical and does not affect the results severely.

Comparative illustrations of the jaw structure are shown in
figure 103. The differences in structure are evident in spite of the
wide variation of the jaw structure in T. insignis. This is probably
a valid character, although its use could be more firmly established
were the sample from the Choza larger.

The post-cranial elements do not provide a basis for differentia-
tion of the species. This is true for the pattern of the dermal surface
of the skull as well. With much larger samples, differences in pro-
portions might be detected, but results of all tests on available
materials have been negative.

An interesting matter arises in connection with the specimens
from the Choza. There are two distinct types of jaw present, one
which is short and relatively broad, and the other, longer and rela-

TABLE 7
Ratios, SkUSkw in T rimer orhachis rogersi sp. nov.

Number

Ratio

UR 138
UR 76

0.94

0.97 J

Relatively
short, broad
skulls

UR 74
UR 137
UR 78

0.841
0.81 \
0.80

Relatively
longer, narrower
skulls

tively narrow (figure 104). This difference is reflected also in the
relative length and width of the various skulls. Ratios for the five
skulls that can be measured with some degree of confidence are
shown in Table 7.

From the differences between the two types of jaws and the
suggestion of a related difference in the ratio of Skw to Ski, it is
evident that two forms of Trimerorhachis were present in the pond
at the pipe site. There are, of course, the possibilities that two
species were present or that there was a single dimorphic species.
There are no absolute criteria for deciding between the two. Some
aid comes from the situation in the earlier species, T. insignis.
Both types of jaws are encountered in the Admiral, Belle Plains,
and Arroyo. There is relatively high variance of the ratio of Skw:Skl
in T. insignis (see fig. 105) but not a distinct differentiation of two
types of skulls. Here, as well as in the Choza specimens, there is,
of course, the possibility that two species exist, but the studies

268

FIELDIANA: GEOLOGY, VOLUME 10

reported in the earlier parts of this paper argue strongly to the
contrary. The fact that the Vale specimens of Trimerorhachis are
close to T. insignis, and probably belong to this species, suggests
that the later specimens, from the Choza, were derived from T.

Fig. 104. Two types of jaws of T. rogersi sp. nov., pipe site, Choza. A and
C, CNHM-UR 73, lateral and dorsal aspects respectively of long, narrow type
of jaw; B and D, CNHM-UR 136, lateral and dorsal aspects respectively of
short, broad type of jaw.

insignis. This is strengthened by the fact that the Choza specimens
show no evidence of the definitive characters of T. mesops, the other
known possible source.

OLSON: FAUNA OF VALE AND CHOZA

269

If it be assumed that the Choza specimens arose from T. insignis,
as defined in this paper, there are several possible courses of evolution
that might have resulted in the two Choza forms. If it be assumed
that the specimens that we have referred to T. insignis do actually
represent two species, then it is possible that the two Choza forms
arose independently, one from each of these species. Beyond the

T INMCNIS

'^-

Fig. 105. Histogram showing frequency distribution of ratio SkUSkw of five
skulls of T. rogersi sp. nov., Choza, and one measurable specimen of T. insignis,
pre-Vale.

fact that it seems most improbable that the initial assumption can
be correct, we are faced with the fact that independent origin requires
very similar changes in skull length and skull depth and in the struc-
tures of the angle of the jaw. These changes are probably adaptive.
While it is not impossible for such a series of independent changes
to have occurred in parallel fashion, it seems unlikely.

It is possible that two Choza species arose from a single earlier
species, T. insignis, the one emphasizing the short, broad jaw and
the other, the longer and narrower jaw. So far as the available
evidence goes, however, this would appear to require sympatric
speciation, which is improbable, particularly in view of the fact
that both "species" show similar, and presumably adaptive, changes.

A third explanation seems more in accordance with the facts —
that the Choza specimens represent a single species in which there is
a rather well-expressed dimorphism. Most probably this would be
sexual dimorphism, although this is not subject to proof. As noted,
there is apparently a tendency in this direction in T. insignis from
pre-Vale beds, and a slight increase in separation is all that is re-
quired to produce the degree of difference seen in the Choza speci-
mens. Sexual dimorphism is, of course, well known in various
extant amphibians, although I know of no case in which it is ex-
pressed in differences in proportions of the skulls and jaws as in
Trimerorhachis. This explanation has been adopted as most probable
in referring all known Choza specimens to the single species T. rogersi.

270 FIELDIANA: GEOLOGY, VOLUME 10

SUMMARY OF EVOLUTION OF TRIMERORHACHIS

The genus Trimerorhachis was relatively conservative in evolution
during the Wichita and Clear Fork times. The environment
throughout this span of time was consistently deltaic, but there
were marked modifications in precipitation, topography, and biota.
Trimerorhachis appears to have been predominantly, if not entirely,
a swamp and pond dweller during the whole period. In many
respects, this was the most persistent of the environmental subzones.
It appears to have continued in existence even during the very late
Arroyo, when the other known ecological subzones were temporarily
eliminated.

T. insignis persisted from near the beginning of Admiral time,
or possibly from the Putnam, through the Arroyo. It is known from
pond and swamp deposits as late as the very earliest Arroyo, and
from pond deposits thereafter. The various occurrences in stream
deposits may indicate that it also inhabited running waters, or per-
haps lived in more quiet backwaters, but there is no certain evidence
that the remains found in such situations were not washed in from
the standing water habitations. The best information available
indicates that this T. insignis persisted through the Vale, but it is
far from certain that the Vale specimens, in part or in toto, pertain
to T. insignis.

The first evidence of a species other than T. insignis comes
from specimens found in the upper third of the Arroyo Formation,
T. mesops. It is perhaps significant that the appearance of this
species coincides with the beginning of the marked changes that
occurred near the close of the Arroyo. At this time, the surface of
the delta was lowered, as the sea apparently transgressed from the
south; divides were low and apparently unpopulated, and the
streams appear to have been subject to invasion of brackish tidal
waters. The differences that can be noted between T. insignis and
T. mesops, however, have no evident adaptive significance. It may
be merely that increasingly restrictive conditions provided a geo-
graphic fragmentation and consequent speciation that was unrelated
to adaptation to changing circumstances. So far as is known,
T. mesops was short-lived and did not persist into the Vale.

The situation in the Vale is problematical. The few jaws, the
only preserved parts of the fragmentary material that can be ex-
pected to show specific characters, are similar to those of some speci-
mens of T. insignis, but they do not show the degree of variability
that is found in local populations of the species in the Admiral and

OLSON: FAUNA OF VALE AND CHOZA 271

Belle Plains. The only known part of the skull that is complete
enough for study is similar to the skull in T. insignis but lacks the
definitive characters. It shows, however, no evidence of the rela-
tively great depth that characterizes the Choza specimens.

Throughout this time, from the Putnam to the end of the Vale,
T. insignis, assuming that this species did persist through the
Vale, lived in a changing situation so far as its vertebrate associates
were concerned. As a swimming, carnivorous amphibian, which
must have spent at least a major part of its life in the water, it
presumably was in intimate ecological association with the other
pond dwellers and found its prey among them. Early, in the
Putnam and Admiral, a common associate was a long-nosed amphi-
bian that resembled Chenoprosopus and may have been closely
, related to or even identical with this genus, known from New
Mexico. During this time, and through the Belle Plains, the lung-
fish Sagenodus was a prominent inhabitant of the ponds. Eryops,
as well, appears to have lived in and about the bodies of quiet water.
Presumably there was association with paleoniscoid fish, crossop-
terygians and xenacanth sharks. The embolomere Archeria appears
to have lived in a similar habitat. The fact that Dimetrodon,
Ophiacodon, and Edaphosaurus are preserved in pond deposits sug-
gests that they may have inhabited the margins of the ponds. The
latter two were partially aquatic.

The associations were somewhat modified during the Arroyo.
Sagenodus, Archeria, and Ophiacodon were no longer present. Only
the small lungfish, Gnalhorhiza serrata, is known from pond deposits,
and it appears to have been represented by few individuals. In
contrast, Diplocaulus and Lysorophus were abundant, and Labido-
saurus and Seymouria are found in deposits with Trimerorhachis,
along with Trematops and Broiliellus, which are encountered less
frequently. Paleoniscoids still were present, but the xenacanth
sharks seem to have been absent from typical Arroyo ponds until
near the end of Arroyo times. Dimetrodon and Edaphosaurus,
specifically different from members of the respective genera in the
Admiral and Belle Plains, continue to be found in pond deposits,
with the latter, however, steadily decreasing in numbers until near
the end of the Arroyo, after which it is no longer found.

In spite of these marked changes in associates, T. insignis
persisted. There is no evidence that the development of the new
species, T. mesops, in the late Arroyo was related to the change in
biota, but, of course, relationships not evident in the materials may
be involved.

272 FIELDIANA: GEOLOGY, VOLUME 10

There was a marked change in the associations of Trimerorhachis
after the beginning of the Vale. Perhaps the most striking was the
invasion of the ponds by Xenacanthus, a predaceous shark. Con-
currently, Diplocaulus, after a brief period of association with
Xenacanthus in the ponds, disappeared from this environment. On
the other hand, a new lungfish, Gnathorhiza dikeloda, appeared in
the environment and persisted in association with Trimerorhachis
and Xenacanthus. Eryops continued in reduced numbers, and the
presence of paleoniscoids is evidenced by the presence of their scales
in coprolites of sharks and other predators. Dimetrodon appears to
have continued to prey upon the animals in the ponds, for its
remains are commonly found in the marginal deposits. Labidosaurus
is no longer encountered, but the remains of Lahidosaurikos and
Captorhinikos, both herbivores, appear during the Vale. In spite
of these modifications, as well as changes in the climate and topog-
raphy, there is no evidence of any noteworthy modification of
Trimerorhachis.

The most striking modification of the genus came very late in
its known history, after the end of the Vale. The Choza species,
T. rogersi, developed a deeper skull, in reversal of the usual trend
of evolution in most groups of amphibians. That this change was
adaptive is difficult to demonstrate, although it probably was. So
far as the evidence goes the only important ecological changes
affecting the ponds were a reduction in the number of animal types
with which Trimerorhachis was associated and a lessening of the
frequency of ponds at any given time. Xenacanthus has not been
found in the ponds with Trimerorhachis in the Choza. There was a
marked increase in the abundance of Lysorophus. Gnathorhiza
dikeloda appears to have been absent from the ponds and there is
no evidence of Eryops. It is difficult to associate the changes evident
in T. rogersi either with the modifications of the fauna or with what-
ever minor changes in the physical environment there may have
been. It is probable that the deepening of the skull and changes in
the angle of the jaw were associated with a modification of jaw
musculature and thereby related to changes in predatory habits.
Presumably such changes could be related to a change in prey,
shown in loss of certain elements from the fauna, but this is highly
speculative.

The most important information on evolution that comes from
the study of Trimerorhachis bears upon the persistence of morphology
over a rather long period of time in which there were marked changes
in both the physical and biological environments. Most probably

OLSON: FAUNA OF VALE AND CHOZA 273

this is related to two factors, one that the genus was carnivorous
and probably not particularly selective in its prey, and the other
that it inhabited an environmental subzone that was persistent in
general characteristics over the period of time involved. Eryops,
and possibly Lysorophns, show a somewhat similar persistence of
species. The only non-aquatic genera that have comparable con-
tinuity of form, and low rate of speciation, are Dimetrodon, a ter-
restrial carnivore in which the species are not markedly different
one from the other, and Captorhinus. One species of Dimetrodon,
D. gigashomogenes, persisted from the lower Arroyo, or perhaps
from the Clyde, through the middle Choza without change. Cap-
torhinus aguti ranges from the Admiral through the Vale and
probably into the Choza.

REFERENCES

Broom, R.

1913. Studies on the Permian temnospondylous stegocephalians of North
America. Bull. Amer. Mus. Nat. Hist., 32, p. 577, 1 fig.

Case, E. C.

1902. On some vertebrate fossils from the Permian of Oklahoma. 2nd Ann,
Kept. Dept. Geol. and Nat. Hist. Terr. Oklahoma, p. 64.

1910. New or little known amphibians from the Permian(?) of Texas. Bull.
Amer. Mus. Nat. Hist., 28, p. 18, 1 fig.

1911. Revision of the Amphibia and Pisces of the Permian of North America.
Carnegie Inst. Wash., publ. 146, pp. 38-47, 106-112, 10 figs, of Trimerorhachis.

1935. Description of a collection of associated skeletons of Trimerorhachis.
Mus. Paleo. Univ. Michigan, 4, pp. 227-274, 29 figs., 11 pis.

Cope, E. D.

1878. Descriptions of extinct Batrachia and Reptilia from the Permian For-
mation of Texas. Proc. Amer. Phil. Soc, 7, p. 524.

1883. Fourth contribution to the history of the Permian Formation of Texas.
Proc. Amer. Phil. Soc, 20, p. 629.

1896a. The reptilian order Cotylosauria. Proc. Amer. Phil. Soc, 34, p. 454.

1896b. Second contribution to the history of the Cotylosauria. Proc. Amer.
Phil. Soc, 35, p. 137.

Olson, E. C.

1948. A preliminary report on vertebrates from the Permian Vale Formation.

Jour. Geol., 56, pp. 186-198, 2 figs., 2 pis., 1 table.
1951a. Vertebrates from the Choza Formation, Permian of Texas. Jour.

Geol., 59, pp. 178-181, 1 fig.
1951b. Fauna of the Upper Vale and Choza: 1-5. Fieldiana, Geol., 10,

pp. 89-128, 15 figs., 3 tables.
1951c. Fauna of the Upper Vale and Choza: 6. Fieldiana, Geol., 10, pp.

147-166, 7 figs., 6 tables.
1951d. The evolution of a Permian vertebrate chronofauna. Evolution,

6, pp. 181-196, 5 figs.
1953. Integrating factors in amphibian skulls. Jour. Geol., 61, pp. 557-568,

3 figs., 3 tables.

274 FIELDIANA: GEOLOGY, VOLUME 10

Olson, E. C. and Beerbower, J. R.

1953. The San Angelo Formation, Permian of Texas, and its vertebrates.
Jour. Geol., 61, pp. 389-423, 10 figs., 7 tables.

Olson, E. C. and Miller, R. L.

1951. Relative growth in paleontological studies. Jour. Paleo., 25, pp. 212-223,
2 figs., 1 table.

Romer, a. S.

1928. Vertebrate faunal horizons in the Texas Permo-Carboniferous red beds.
Univ. Texas, Bull. 2801, pp. 67-108, 7 figs.

1947. Review of the Labyrinthodontia. Bull. Mus. Comp. Zool., 99, pp.
3-352, 48 figs.

Romer, A. S. and Olson, E. C.

1954. Aestivation in a Permian lungfish. Breviora, Mus. Comp. Zool., no.
30, 8 pp., 1 pi.

Romer, A. S. and Price, L. I.

1940. Review of the Pelycosauria. Geol. Soc. Amer., Special Papers, 28,
538 pp., 71 figs., 46 pis., 7 tables.

Wilson, J. A.

1948. Vertebrate fauna of a Permian locality in Taylor County, Texas. Bull.
Geol. Soc. Amer., 59, p. 1363 (abs.).