*

o

LI B RAHY

OF THE

UNIVERSITY

Of ILLINOIS

550.5
FI

GfcUlOUX

UNIVERSITY OF

ILLINOIS LIBRARY

AT URBANA-CHAMPA1GN

GEaOGY

Return this book on or before the
Latest Date stamped below.

GEOLOGY LIBRARY

University of Illinois Library

dec 30 tiw :juni:

Iff tt9te "ft^5

FES 3 1367

&NW

<S

"s

^

#

X

MAY

i*n&

JUL j ;

rtftY"2"2

MS

1972

Hi"1

HWJ,«

",

982

•S

198

V

•«5
<5>

172004

L161— H41

LATE DEVONIAN FRESH-WATER FISHES

FROM THE

WESTERN UNITED STATES

ROBERT H. DENISON

Curator of Fossil Fishes

FIELDIANA: GEOLOGY

VOLUME 11, NUMBER 5

Published by

CHICAGO NATURAL HISTORY MUSEUM

DECEMBER 28, 1951

""-IP LIBRARY OF THE

JAN 3 0 1S52

UrtlVtRSiTY OF ILLINOIS

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

SSO^S

INTRODUCTION

During the geological exploration of the western United States,
occasional finds of fossil fishes have been made in Late Devonian
rocks. Some of these have received only casual mention in geological
reports. Others, though identified by vertebrate paleontologists,
have not been adequately studied, probably because the remains
are fragmentary and thus in many cases identification even to
family is difficult. In the summer of 1949 I visited most of the
sites at which Late Devonian fishes have been reported in Arizona,
Colorado, Wyoming, and Utah. Although most of the specimens
obtained were poorly preserved, a study of the collections revealed
the inadequacy of the earlier work. For a clearer picture of the
Devonian fishes of this period, and the conditions under which they
were preserved, a revision seemed desirable. This has been made
as complete as possible, although a few of the early collections have
not been seen.

The earliest discovery of Late Devonian fishes in the western
United States was made in 1874 in southwestern Colorado by
Endlich (1876, p. 212) of the Hayden Survey. Endlich's material
and more collected later by Cross in the same rocks, the Elbert
formation, were described by Eastman (1904, pp. 253-258), who
referred them to Bothriolepis and Holoptychius. In 1879, Walcott
(1880, p. 225) discovered some "placoganoid fishes" in a Devonian
sandstone in Kanab Canyon, Arizona. These fragmentary remains
rom what came to be called the Temple Butte formation, were
dentified by Gidley (Schuchert, 1918, p. 361) as Bothriolepis and
Holoptychius. Another early discovery was made in 1896 by Tower,
vho found fish remains in the Late Devonian Parting Quartzite
lear Aspen, Colorado. The teeth in this collection were referred to
• rossopterygians by Girty (Spurr, 1898, p. 21), and other fragments
vere referred doubtfully to lung fishes and arthrodires by Eastman
1904, pp. 258-259). In 1912, Worcester collected some fossils
; t Fossil Ridge, Gunnison County, Colorado, in strata supposed to
be of Ordovician age. The fishes in this collection were briefly
] oted by Cockerell (1913, pp. 246-247), who was able to identify
1 hem as Coccosteus, Rhizodus and Diplacanthus, and still maintain

221

222 FIELDIANA: GEOLOGY, VOLUME 11

that they were Ordovician. As will be shown below, Cockerell was
mistaken both in his identification of the fossils and in their age.

In more recent years, a number of occurrences of Late Devonian
fishes have been reported. Several of them are in the Parting
member of the Chaffee formation of central Colorado (Behre and
Johnson, 1933; Bryant and Johnson, 1936; Johnson, 1944). Fishes
have also been reported from the Darby formation of Wyoming
(Branson, 1929; Branson and Mehl, 1930; Baker, 1946). Descrip-
tions of the Darby fishes have never been published, but as will be
shown below, they are of Late Devonian fresh-water types. This
is also true of the fishes from what was identified as the Leigh
member of the Bighorn formation (Foster, 1947). Since the
Leigh is Late Ordovician in age, it is clear that part of the Darby
formation has been misidentified here.

Typically marine Late Devonian fishes will not be considered in
this work. These have been reported in the Sly Gap formation of
western New Mexico (Colbert and others, 1950), and in the Ouray,
Picacho de Calera, and Martin formations of eastern Arizona and
southwestern Colorado (Hay, 1902; Stoyanow, 1936, 1942). The
Jerome formation of central Arizona (Hussakof, 1942; Stoyanow,
1926, 1936), which includes a near shore marine facies with a mixed
marine and fresh-water fish fauna, will be considered at another time.
The Jefferson formation is marine, but the lung fish teeth that have
been found in it in Montana and Utah will be discussed because of
the possibility that they were of fresh- water origin. The reported
occurrence of fish remains in the lower part of the Jefferson forma-
tion of south central Idaho (Ross, 1934, 1937) is misleading. The
fish are of Early Devonian types, and so the rocks in which they occur
must represent a formation distinct from the Jefferson.

This study is based in large part on collections I made in 1949
and 1950 in Colorado, Wyoming, and Utah, for Chicago Natural
History Museum. In addition, Dr. David H. Dunkle of the United
States National Museum has generously made it possible for me to
study many of the early collections of members of the United States
Geological Survey. A small collection from the Darby formation
of Teton Pass, Wyoming, has been made available by Dr. Claude W.
Hibbard of the Museum of Paleontology, University of Michigan.
Except as otherwise noted, the specimens described are in the col-
lections of Chicago Natural History Museum. The drawings were
made by Mr. John Conrad Hansen, Artist in the Department of
Geology, and me, mostly from my original sketches.

DENISON: DEVONIAN FRESH-WATER FISHES 223

ANTIARCHI

The peculiar armored placoderm, Bothriolepis, is a characteristic
element of Late Devonian fresh-water faunas throughout much of
the world. It is common in the western United States, but its
remains are usually fragmentary or poorly preserved, so that identi-
fication is difficult. Two species are to be distinguished in this area.

Bothriolepis coloradensis Eastman: The original description of
this species by Eastman in 1904 was based on material collected by
Cross in southwestern Colorado. Stensio (1931, p. 11; 1948, p.
224) concluded from Eastman's figures with some justification that
it did not belong to Bothriolepis, but to the Remigolepidae. East-
man's figures, however, are quite inadequate to characterize this
form, and a study of the type, as well as new material collected for
Chicago Natural History Museum, clearly shows that this is Bothrio-
lepis and possibly a valid species.

Stensio referred this species to the Remigolepidae because East-
man's figures (1904, figs. 1, 4) suggest that the pectoral fin was not
divided into two articulated segments. The type (U.S.N.M. no.
16852) is a ventral trunk shield exhibiting the inner surface, with
fragments of both pectoral appendages in place. The more com-
plete appendage, that of the left side, does not preserve the articula-
tion, but shows quite convincingly that one was present between
proximal and distal segments. The ventral central plate 2 and the
lateral marginal plate 2 are recognizable and show the beginnings
of the taper towards the articulatory area, exactly as in Bothriolepis.
A second specimen in the United States National Museum, originally
on the same slab of rock as the type, preserves a distal segment of
a pectoral fin of the Bothriolepis type. Thus it is certain that this
form does not belong to a remigolepid, as Stensio suspected.

Since Cross's collections were made, no additional specimens have
been found at the type locality, one mile south of Rockwood, La
Plata County, Colorado. The Elbert formation does not outcrop
here, and the original finds were made in talus below its presumed
occurrence. A number of specimens of Bothriolepis have been found
both by Cross and by me 113^ miles to the east on Endlich Mesa,
where the Elbert formation is well exposed. Eastman identified
in incomplete proximal fin segment from this locality as B. colora-
iensis because of its large size. Other fragments he referred to
5. "leidyi" (=B. nitida) largely because of their smaller size and
^resumed differences in ornamentation. Size, of course, is not a
•eliable specific character, and, as will be shown below, Eastman

224 FIELDIANA: GEOLOGY, VOLUME 11

probably misinterpreted the ornamentation of the Rockwood speci-
mens. In 1949, I found fragmentary remains of B. coloradensis
eight miles north of Rockwood in a limestone in the Elbert formation.
This is the locality described by Cross (1904, p. 250) on Little
Cascade Creek, about one-half mile south of Columbine Lake. It
is probable that all of the Bothriolepis remains from this region in
southwestern Colorado belong to B. coloradensis, and it is possible
with the additional material now available to characterize the species
somewhat better, though not completely.

Size: Eastman distinguished B. coloradensis in part by its large
size. It is true that this species does attain a moderately large
size for the genus, but measurements of a few individuals have
little significance. For example, the antero-median-dorsal plates
(AMD) of B. canadensis in the Dartmouth College Museum have
lengths ranging between 14.3 and 74 mm., depending on the age
of the individual. The two antero-median-dorsal plates of B. colora-
densis in Chicago Natural History Museum are 58 mm. and 62 mm.
long, corresponding in size to the large mature B. canadensis. The
width of the ventral armor of B. coloradensis, measured at mid-length,
is 85 mm. (Eastman, 1904, p. 256), but this may be exceeded by
very large B. canadensis. It is clear then that these two species
cannot be distinguished by their size.

Ornamentation: Eastman described the superficial ornamenta-
tion of the specimens from Rockwood, Colorado, as clearly tubercu-
late, with the tubercles more or less confluent, but not forming
vermiculating ridges. The type, however, shows mainly the inner
side of the ventral armor and is entirely inadequate to indicate the
character of the ornamentation. This feature is shown much better
in material from Endlich Mesa and Little Cascade Creek (fig. 40,
D, E). These specimens show that, compared to B. canadensis of
equal size, the ornamentation is not coarser, as Eastman said, but
about the same. It does differ, however, from B. canadensis, and
resembles B. cellulosa in being reticular, with very slight or no
tubercular elevations at the anastomoses of ridges. In parts of
certain plates, as, for example, on the medial part of the lateral plate
(fig. 40, D), the even reticular arrangement gives way to a more ridged
type of ornamentation, with the ridges straight or sinuous. Reticular
ornament is characteristic of juvenile individuals of most species of
Bothriolepis, as Stensio has pointed out (1948, p. 212), but it is
clear that in the known specimens of B. coloradensis we are dealing
with parts of fully adult individuals.

DENISON: DEVONIAN FRESH-WATER FISHES

225

Fig. 38. Bothriolepis coloradensis Eastman (natural size). A, Centro-nuchal
l>late, restored from PF 591, Chaffee formation, Gunnison County, Colorado.
15, Antero-median-dorsal plate, restored from PF 360, Elbert formation, La Plata
< 'ounty, Colorado. C, Right ventral centrah plate, PF 593, Chaffee formation,
( runnison County, Colorado, aim, anterior division of lateral margin; ar, articular
area; avp, anterior ventral pit; crpl, post-levator crest; If, levator fossa; mc, lateral
C >rner between anterior and posterior divisions of lateral margin; npp, post-pineal
rutch; pc, postero-lateral corner; plm, posterior division of lateral margin; prpm,
p )stero-median process; sadl, area overlapping antero-median-dorsal plate; scvi,
s iture for second ventral central plate; sml, suture for lateral marginal plate.

Plates of the carapace: The only known plate of the head shield
i^ the lateral plate (PF 364, fig. 40, D), which agrees with the same
p ate of B. canadensis and B. cellulosa. The antero-median-dorsal
p ate (AMD; fig. 38, B) resembles in most respects that of B. cana-
rf< nsis and B. cellulosa. It differs from these species in that the

226

FIELDIANA: GEOLOGY, VOLUME 11

spmd-V

Fig. 39. Bothriolepis coloradensis Eastman, Elbert formation, Endlich Mesa,
La Plata County, Colorado (X %). A, Left antero-ventro-lateral plate and
proximal segment of pectoral fin, restored from PF 355. B, Right mixilateral plate,
PF 357. al, antero-lateral corner; cvi, CV2, first and second ventral central plates;
dl, dorsal lamina; dlr, ridge between dorsal and lateral laminae; 11, lateral lamina;
ml, lateral marginal plate; ndc, notch for dorsal corner of postero-ventro-lateral
plate; sadl, area overlapped by antero-dorso-lateral plate; spmd, area overlapped
by postero-median-dorsal plate; spvl, area overlapped by postero-ventro-lateral
plate.

post-levator crests nearly meet in front of the anterior ventral pit,
resulting in a shorter and relatively broader levator fossa (fig. 38,
B, If). The anterior border of the antero-median-dorsal plate is
relatively short, even shorter than the posterior border, which differs
from the situation in B. canadensis and B. cellulosa. The postero-
median-dorsal plate agrees well with that of B. canadensis. The
mixilateral plate (MXL; fig. 39, B) indicates that the lateral walls

DENISON: DEVONIAN FRESH-WATER FISHES 227

of the trunk shield were relatively high. The length /height of its
lateral lamina is 2.4, about the same as in B. canadensis and B.
cellulosa.1

The dorsal lamina is very broad, its length /width ratio being 1.4,
compared to a corresponding ratio of 1.5 to 2.0 in B. canadensis
and B. cellulosa. This broad dorsal lamina may signify a slightly
broader trunk armor than in the latter species, or a greater angle
between the dorsal and lateral walls of the trunk armor; the angle
could not be measured because of crushing. The anterior border of
the mixilateral plate, that is, the ornamented part that meets but is
not overlapped by the antero-dorso-lateral plate (ADL), is nearly
straight, not concave as in B. canadensis and B. cellulosa. The notch
in the posterior part of the ventral border (fig. 39, B, ndc) that
received the dorsal corner of the postero-ventro-lateral plate (PVL)
is rounded and agrees closely with that of B. canadensis. The
antero-ventro-lateral plate (AVL; fig. 39, A) resembles that of
B. canadensis. Judging from Eastman's description and figure
(1904, pp. 255-256, fig. 1), the postero-ventro-lateral plates are
of a peculiar type. However, it should be noted that Eastman had
only the ventral laminae, and that their margins were incomplete
posteriorly. In two incomplete specimens from Endlich Mesa
(PF 354, PF 361) the postero-ventro-laterals agree closely with
those of B. canadensis.

Pectoral appendages: As has been pointed out above, the type
)f B. coloradensis undoubtedly possessed a jointed pectoral fin,
iivided into proximal and distal segments. The specimen figured
)y Eastman from Endlich Mesa (1904, fig. 4; U.S.N.M. no. 16853)
is part of the proximal segment of a very large individual, the pre-
served part measuring 98 mm. in length. Its narrow appearance is
due to the fact that the margins are missing. A specimen that I
( ollected (PF 355, fig. 39, A) shows a crushed proximal segment,
with the proximal articulation preserved but the distal articulation
i lissing; its estimated length was 78 mm. One of the specimens in
ihe United States National Museum collected at Rockwood shows
t le distal segment of the appendage, but it is too poorly preserved
t ) furnish any distinctive characters.

Comparison with other species: The study of the new material
aid restudy of the old collections from southwestern Colorado
snow that Bothriolepis coloradensis is similar in many respects to
L . canadensis and B. cellulosa. It differs from both in the shorter,

1 Measurements as in Stensio, 1948, pp. 11-16.

228 FIELDIANA: GEOLOGY, VOLUME 11

broader levator fossa of the antero-median-dorsal plate, with the
post-levator crests approaching in front of the anterior ventral pit,
in the narrower anterior border of the antero-median-dorsal plate,
and in the broader dorsal lamina and straighter anterior ornamented
border of the mixilateral plate. It resembles B. cellulosa but may be
distinguished from B. canadensis in its reticular and ridged type of
adult ornamentation. B. canadensis, known only from the Late
Devonian of Quebec, is clearly distinct from B. coloradensis. Better
material of B. coloradensis will probably show additional differences
from B. cellulosa; it is hardly likely that the two are cospecific, since
B. cellulosa occurs in the Baltic States, northwestern Russia and
Siberia (Gross, 1941a; Stensio, 1948).

Other species of Bothriolepis reported from North America in-
clude B. traquairi Bryant, B. minor Newberry, and B. nitida (Leidy).
B. traquairi from Quebec is so distinct in its ventral trunk armor that
it may not belong to this genus (Bryant, 1924, p. 55; Stensio, 1948,
p. 400). B. minor, from the Late Devonian of New York and
Pennsylvania, is inadequately described and defined. Its small
size is not in itself distinctive, especially as Eastman (1899, p. 326)
has referred to this species plates as big as those of B. nitida. Since
the original description (Newberry, 1889, p. 112), it has been reported
from the entire extent of the Late Devonian, and in two cases (East-
man, 1899; Willard, 1932, p. 29) from the same horizon and locality
as B. nitida, so its geologic age cannot distinguish it from the latter.
Its ornamentation, described by Newberry (1889, p. 112) as consist-
ing of "fine, closely crowded vermicular furrows," only doubtfully
distinguishes it from B. nitida. B. nitida has never been adequately
described, although it occurs abundantly in the Late Devonian rocks
of northern Pennsylvania and southeastern New York. The type
is a distal segment of a pectoral fin from Tioga County, Pennsylvania,
originally described (Leidy, 1856a, p. 11) as an "ichthyodorulite."
Additional material has been described since that time by New-
berry (1889, p. Ill, pi. 18, fig. 2; pi. 20, figs. 1-5) as B. "leidyi,"
and by Eastman (1907, pp. 50-51, pi. 3, fig. 4; pi. 7, fig. 1) as B.
nitida, but their accounts do not allow a satisfactory diagnosis
to be made. I have briefly examined material of this species from
Tioga County, Pennsylvania, collected by Andrew Sherwood and
R. H. Lacoe, and now in the United States National Museum. The
material does not permit a complete diagnosis to be made, but
clears up important points regarding the structure of B. nitida.

The size attained is quite considerable; one antero-median-dorsal
plate in the United States National Museum collection measures

DENISON: DEVONIAN FRESH-WATER FISHES 229

62.5 mm. in length, comparable to a large B. canadensis. The
ornamentation, even in large, fully mature plates, retains some
reticular character, although in places the reticulations are broken
into tubercles or sinuous ridges. The ornamentation of the distal
segment of the pectoral fin (U.S.N.M. no. 1982) is not strictly linear
as shown by Leidy (1856b, pi. 16, figs. 7, 8), Cope (1891, pi. 30,
fig. 7) and Eastman (1907, pi. 3, fig. 4), but is elongate reticular.
The centro-nuchal plate is incompletely known, but is distinguished
from B. canadensis by an important character. The posterior oblique
cephalic pit lines come together rather far forward, and at their
point of approach are separated by a wide ornamented area from the
smooth nuchal area forming the posterior edge of the plate. This
same feature is found in B. cellulosa, B. pander i, and B. minor, as
well as in Bothriolepis from Colorado and Wyoming, as will be seen
below. The antero-median-dorsal plate agrees in general with that
of B. canadensis, but, as in the case of B. color adensis, the post-
levator crests approach each other in front of the anterior ventral
pit, resulting in a relatively short, broad levator fossa. No characters
have been found that would distinguish the postero-median-dorsal
plate from that of B. canadensis. The antero-dorso-lateral has a
relatively wide dorsal lamina, the ratio of its length/width being
2.1. The antero-ventro-lateral plate also has a rather broad ventral
lamina, the ratio of its length to its width being 2.1; it is not as broad
as Stensio (1948, p. 396) estimated from Newberry's figure of an
incomplete plate. The postero-ventro-lateral plate indicates that
;his species had relatively high lateral walls in its trunk shield; the
ength/height ratio is 2.1, about as in B. canadensis. The postero-
/entro-lateral also has a rounded dorsal corner as in the latter species,
ind probably in B. coloradensis. The pectoral fin is more slender
han in B. canadensis of equal size. The ratio of length/width of
he proximal segment in the large individuals measured is 5.2,
while in B. canadensis of comparable size it is 3.3 to 3.6. In the distal
segment the length /width ratio is 5.2 to 5.4 in large B. nitida,
4.0 to 4.8 in large B. canadensis (ratios computed from measurements
( f figures in Stensio, 1948).

It is interesting to note in reviewing these characters of B. nitida
tiat it cannot be clearly distinguished from B. coloradensis. It is
I ossible that these two species will be found to be identical, in which
cise B. nitida must have priority. Before the question can be
s ;ttled, more specimens of B. coloradensis are needed, and a restudy
o ' all the Bothriolepis material from New York and Pennsylvania is
d isirable. In the meantime, B. coloradensis is retained provisionally.

230 FIELDIANA: GEOLOGY, VOLUME 11

Other occurrences of Bothriolepis color adensis: Fragments of
Bothriolepis have been collected by Walcott (1880, p. 225) and Noble
(1922, p. 52) from the base of the Temple Butte formation in Lower
Kanab Canyon and Grand Canyon, Arizona. The remains are not
sufficiently well preserved to permit specific identification, but the
reticular, non-tubercular type of ornamentation on plates of rather
large size suggests that they may belong to B. color adensis. In
light of the comparisons made above, it is interesting to note that
these were identified by Gidley (Schuchert, 1918, p. 361) as B.
nitidens (=B. nitida), and by Eastman (United States National
Museum collection) as B. leidyi (=B. nitida).

A few fragments in the United States National Museum, collected
near Glenwood Springs, Garfield County, Colorado, also resemble
B. coloradensis in their ornamentation. According to their label
they came from the formation overlying the Cambrian shales and
are presumably from the Parting member of the Chaffee formation.

In 1913, Cockerell reported fossil fishes from the cirque above
Boulder Lake at the north end of Fossil Ridge, Gunnison County,
Colorado. Since Cockerell did not give specific localities for the
different fishes he described, it is not possible to compare his identi-
fications with those made later. Kirk (1930, 1931) showed that
Cockerell's specimens came from the Devonian, not the Ordovician,
and Bryant and Johnson (1936) reported crossopterygians from the
Parting member of the Chaffee formation at this locality. In the
summer of 1949, I visited this site and collected from the Parting
member not only crossopterygian remains, but fragments of a large
Bothriolepis. The few plates obtained do not allow an entirely satis-
factory comparison, but it is probable that they belong to B. colora-
densis. The ornamentation of young individuals is apparently
reticular, as is usual in this genus. On large mature individuals
much of this reticular character is retained but with development of
sinuous ridges in certain regions, exactly as in B. coloradensis. One
unidentified fragment of a very large individual (PF 587) has the
ridged character developed over part of its extent to a very high
degree (fig. 40, C). The fossil "with vermiform ridges" ascribed by
Cockerell to Rhizodus is probably such a fragment of Bothriolepis.
The size attained by these central Colorado Bothriolepis was perhaps
even greater than that recorded for B. canadensis and the typical
B. coloradensis. The total length of a centro-nuchal plate (measured
as in Stensio, 1948, fig. 1) is approximately 37 mm. and its para-
median length is 34 mm. (PF 591). The postero-median-dorsal and

DENISON: DEVONIAN FRESH- WATER FISHES

231

Fig. 40. Bothriolepis, dermal bone ornament (X 3A). A, B. darbiensis, sp.
nov., left half of centro-nuchal plate, Univ. Mich. 27547, Teton Pass, Wyoming.
B, B. darbiensis, sp. nov., undetermined plate of large individual, PF 378, near
Fort Washakie, Wyoming. C, B. coloradensis(l) , undetermined plate of large
individual, PF 587, Fossil Ridge, Gunnison County, Colorado. D, B. coloradensis
Eastman, cast of impression of left lateral plate, PF 364, Endlich Mesa, La Plata
County, Colorado. E, B. coloradensis, part of postero-median-dorsal plate, PF
362, near Columbine Lake, La Plata County, Colorado.

postero-median-ventral plates agree with those of B. coloradensis.
A ventral central plate (Cvi), measuring 81 mm. in length and thus
of a large individual, indicates a pectoral fin of about the same
proportions as in B. canadensis; the ratio of length/width is 2.8
(fig. 38, C). The centro-nuchal, unknown in B. coloradensis from
southwestern Colorado, is represented by three, more or less complete,

IfllTlj- -

smxl

Fig. 41. Bothriolepis darbiensis, sp. nov. (B, X 2; A, C-G, natural size).
A, Distal segment of right pectoral fin, dorsal side, Univ. Mich. 27546. B, Type,
centro-nuchal and incomplete right paranuchal and lateral plates, PF 396. C,
Left half of centro-nuchal plate, Univ. Mich. 27547. D, Right ventral centrah
plate, based on PF 366 and PF 365. E, Left postero-ventro-lateral plate, inner
side, based largely on PF 396. F, Antero-median-dorsal plate, anterior half,
inner side, PF 407. G, Right antero-ventro-lateral plate, ventral lamina, PF 381.
A and C from Darby formation, Teton Pass, Wyoming; B, D-G, from Darby
formation, near Fort Washakie, Wyoming, af, articular fovea; al, antero-lateral
corner; ar, articular area; avp, anterior ventral pit; cd&.&, dorsal central plates;
crpl, post-levator crest; dc, dorsal corner of postero-ventro-lateral plate; eavl,

(Continued on page 233)

232

DENISON: DEVONIAN FRESH-WATER FISHES 233

large specimens, and shows some interesting features (fig. 38, A).
The postero-median process is strong, and the postero-lateral corners
are very long and pointed, comparable to the situation in B. maxima
(Gross in Stensio, 1948, figs. 236, 237). This may not be too reliable
a specific character, since it probably is correlated to some extent
with the large size of these individuals; in B. canadensis, these proc-
esses and corners become more pronounced with increasing age
(Stensio, 1948, fig. 93). A characteristic of greater taxonomic value
is the relatively very short anterior part of the lateral margin; the
ratio of the anterior to the posterior part of this margin is 0.39
compared to 0.70-0.85 in B. canadensis. Of significance also is the
fact that the posterior oblique cephalic pit lines meet relatively far
forward, and are separated by a wide ornamented strip from the
nuchal area, exactly as in B. nitida.

Bothriolepis darbiensis,1 sp. nov.

Type.— C.N.H.M -PF 396, a centro-nuchal plate with part of
the right paranuchal and right lateral plates attached, belonging to
a small individual (fig. 41, B).

Type locality. — Late Devonian, Darby formation, SW. x/i,
sec. 18, T. 1 S., R. 2 W., on South Fork of Little Wind River, near
Fort Washakie, Wind River Indian Reservation, Wyoming.

Diagnosis. — A species of Bothriolepis attaining about the same
size as B. canadensis and B. coloradensis. Ornament of small indi-
viduals reticular, with slight tubercular elevations at junctions of
ridges; ornament of large individuals consisting of tubercles and
nodose, sometimes sinuous, ridges, often with a somewhat concentric
disposition near the borders of trunk plates (fig. 40, A, B). Posterior
oblique cephalic pit lines coming together about half way between
the post-pineal notch and the posterior border of the centro-nuchal

edge overlapped by antero-ventro-lateral plate; fap, articular fossa for pectoral
fin; fed, external opening of endolymphatic duct; ifcb, posterior oblique cephalic
pit line; ifcbi, anterior oblique cephalic pit line; If, levator fossa; Ip, lateral plate;
mc, lateral corner between anterior and posterior divisions of lateral margin;
w?3-6, lateral marginal plates; mmz-i, medial marginal plates; nm, unornamented
nuchal area; npp, post-pineal notch; pc, postero-lateral corner of centro-nuchal
plate; pn, paranuchal plate; pre, post-branchial crest; sadl, area overlapping
antero-dorso-lateral plate; sevt, suture for second ventral central plate; sml,
suture for lateral marginal plate; smxl, area overlapping mixilateral plate; spvl,
irea overlapping right postero-ventro-lateral plate; t, terminal plate; vlr, ventro-
ateral ridge.

1 Named after the Darby formation of western Wyoming.

LIBRARY

UNIVERSITY OF ILLINOIS

234 FIELDIANA: GEOLOGY, VOLUME 11

plate, with a wide ornamented border posterior to their junction
(fig. 41, B, ifcb). Anterior part of lateral border of centro-nuchal
plate about two-thirds as long as posterior part. The post-levator
crests of the antero-median-dorsal plate approaching each other in
front of the anterior ventral pit, resulting in a relatively short, broad
levator fossa (fig. 41, F, If). The lateral walls of the trunk shield
relatively high, the ratio of length /height of the lateral lamina of the
antero-dorso-lateral plate being 2.0-2.4, of the mixilateral 2.9-3.0.
The angle between the dorsal and lateral walls of the trunk shield
probably relatively great (120°-125°) resulting in a high trunk
shield. The antero-dorso-lateral plate with a weak post-nuchal
corner, and the antero-ventro-lateral plate with a weak antero-
lateral corner (fig. 41, G, al). The lateral lamina of the postero-
ventro-lateral plate with a pronounced pointed dorsal corner (fig. 41,
E, dc) . Pectoral fin more slender than in B. canadensis of comparable
size; the ratio of length /width of ventral central plate (Cvi) is
approximately 3.7 in small and 3.2 in large individuals (fig. 41, D).

Discussion. — Fish remains were first discovered on the South
Fork of the Little Wind River by N. H. Brown of Lander, Wyoming.
These fragmentary fossils from the Darby formation were first identi-
fied as Dinichthys and Acanthaspis (Branson, 1929), then as "a new
genus of fishes closely allied to Pterichthys" as well as some fishes of
Early Devonian affinities (Branson and Mehl, 1930), and finally as
"ostracoderms" (Branson and Branson, 1941). In 1949, I was able
to collect at this site and obtained, with the exception of a few
remains of crossopterygians, only plates and fragments of Bothrio-
lepis darbiensis. There can be no question of their Late Devonian
age.

Most of the material collected in 1949 consists of isolated plates
of small, presumably juvenile individuals. Remains of larger in-
dividuals are common, but are difficult to collect because of the nature
of the matrix and poor preservation of the bone. Because of the
individual age difference of the majority of the available specimens,
comparison with B. coloradensis is difficult, but it is apparent from
the diagnosis above that the two species are similar in many respects
and presumably closely related. B. darbiensis may be distinguished
from B. coloradensis and B. nitida by its more tubercular type of
adult ornamentation (fig. 40, A, B), by the longer anterior part of
the lateral border of the centro-nuchal plate (fig. 41, B), by the
weak development of the antero-lateral process of the antero-ventro-
lateral plate (fig. 41, G), and by the pronounced pointed dorsal corner

DENISON: DEVONIAN FRESH-WATER FISHES 235

of the postero-ventro-lateral plate (fig. 41, E, dc). More material
would undoubtedly reveal additional differences. From B. cana-
densis, which it resembles in ornamentation, it may be distinguished
by the more anterior meeting of the posterior oblique cephalic pit
lines on the centro-nuchal plate (fig. 41, B, ifcb), by the relatively
short, broad levator fossa on the antero-median-dorsal (fig. 41, F,
If), by the weaker post-nuchal corner of the antero-dorso-lateral,
and the weaker antero-lateral corner of the antero-ventro-lateral
(fig. 41, G, al), by the probable higher angle between the dorsal
and lateral walls of the trunk shield, by the pronounced dorsal corner
of the postero-ventro-lateral (fig. 41, E, dc), and by the more slender
pectoral fins (fig. 41, D).

In 1947, Foster reported the discovery of fish remains in the
Leigh member of the Bighorn formation on Glory Mountain, near
Teton Pass, Teton County, Wyoming. J. T. Gregory considered
the fish to be arthrodire and probably Devonian in age. Through
the kindness of Dr. Claude W. Hibbard of the University of Michigan,
I have been able to examine this material, and find that it consists
almost entirely of fragmentary remains ascribable to Bothriolepis
darbiensis. Since the Leigh member is Late Ordovician in age, it
is apparent that the formation has been misidentified, and the thin-
bedded sandy limestones that contain the fish should be referred to
the lower part of the Darby formation of Late Devonian age.

The available identifiable plates from Teton Pass are few, and
are all of very large individuals, exceeding in size any individuals of
B. canadensis, B. color adensis, or B. darbiensis that I have measured.
A distal segment of the pectoral fin (Univ. Mich. 27546) measures
75 mm. in length, excluding the anconeal processes. A centro-nuchal
(Univ. Mich. 27547) has an estimated paramedian length of 45 mm.
The ornamentation is distinctly tuberculate, although the tubercles
are in places united more or less into short ridges (fig. 40, A). The
centro-nuchal (Univ. Mich. 27547) agrees with that of B. darbiensis
in the proportions of the anterior and posterior parts of the lateral
margin (figs. 40, A; 41, C) and in the position of the posterior oblique
cephalic pit lines; its postero-lateral corners are moderately large,
but blunt (fig. 41, C, pc), and not produced into sharp points as in
large B. coloradensis. The antero-ventro-lateral (Univ. Mich. 27548)
agrees with that of B. darbiensis in having a reduced antero-lateral
orocess. The pectoral fin is relatively slender, the distal segment of

:he very large individual having a ratio of length/width of 5.0

fig. 41, A).

236 FIELDIANA: GEOLOGY, VOLUME 11

ARTHRODIRA

Arthrodires have been reported from the Late Devonian in many-
places in the western United States. They are definitely present in
large numbers in the near shore facies of the Jerome formation of
Arizona, but most of the other records in the Rocky Mountain
region are open to question. Eastman (1904, p. 258) mentioned,
but did not figure, "a finely tuberculated plate . . . which appears
to be a posterior ventral of some Arthrodire, of about twice the size
of the type species of Coccosteus." This plate was collected from the
Elbert formation on Little Cascade Creek, about one-half mile south
of Columbine Lake, La Plata County, Colorado. In 1949, at this
locality, I collected crossopterygian remains showing a fine tubercula-
tion, and I venture to suggest that Eastman's specimen may belong
to the same form. Cockerell (1913, p. 246) reported a plate re-
sembling Coccosteus, having a structure nearly identical with As-
traspis, from what is presumably the Parting member of the Chaffee
formation in Gunnison County, Colorado. I have not seen this
specimen, nor is it possible from Cockerell's description to determine
the exact locality, but it is probable that he misidentified the plate;
no arthrodires have been reported from this region by subsequent
collectors.

Eastman (1904, pp. 258-259) rather doubtfully assigned to
arthrodires some finely tuberculated plates from the Parting Quart-
zite, near Aspen, Pitkin County, Colorado. The only determinable
specimens in my own collection from Aspen belong to a crossop-
terygian; its plates and scales are often finely tuberculated, and it
is possible that this is the form that Eastman identified as arthrodiran.
Bryant (Bryant and Johnson, 1936, p. 658) recognized two types
of arthrodires, resembling Aspidichthys and Dinichthys, in the Part-
ing member of the Chaffee formation on Gribbles Creek, Fremont
County, Colorado. Only fragments were available to Bryant, but
his identification was supported by thin sections and may be con-
sidered as probably correct. The "Dinichthys" and "Acanthaspis"
that Branson reported (1929) from the Darby formation near Lander,
Wyoming, are undoubtedly specimens of Bothriolepis darbiensis.
According to Newton Brown of Lander (personal communication),
they were collected from the same locality as the type of the latter
species.

I have been unable to obtain for study the "arthrodire jaw"
reported by Baker (1946, p. 575) from Sheep Mountain on the
southwest side of the Wind River Mountains.

DENISON: DEVONIAN FRESH-WATER FISHES 237

DIPNOI

Lung fishes are not uncommon in Late Devonian formations of
the western United States, but, with the exception of a few ques-
tionable scales, only dental plates have been recognized. Unfor-
tunately, it is becoming increasingly evident that a single tooth or
a few teeth are insufficient to diagnose a species. There are numerous
cases where several species, described from a small collection of
dental plates, have been shown to be variations of a single species
when a large series of plates has become available. Thus, Miall
(1878, p. 27), with 350 Ceratodus teeth to study, was able to reduce
to only two the many species described by Agassiz from the Rhaetic
of Aust Cliff, near Bristol, England. Peyer (Stromer and Peyer,
1917, p. 61) concluded that many of the species previously described
from the Triassic of Europe were synonyms. The variation of dental
plates within a single species can be great, as is well shown by Peyer
(1924, pi. 1) in Ceratodus africanus Haug, and is due to a number of
factors. Because lung fish dental plates can grow, even while func-
tional (Stromer and Peyer, 1917, pp. 46-48), there may be considerable
difference in their size in juvenile and old individuals; within a single
species, Ceratodus africanus, the length of the dental plate may range
between 13 and 77 mm. (measurements from Peyer, 1924, pi. 1).
Palatal and splenial tooth plates differ, and an isolated plate is often
difficult to place. Wear may greatly change the appearance of teeth
in old individuals, and finally post-mortem erosion, even possibly
by digestive fluids of predators (Eastman, 1908, p. 217), must be
considered. Taking all these facts into consideration, minor differ-
ences exhibited by a single specimen or a small sample are not
necessarily valid criteria for the erection of a new species.

Among the species of Dipterus described from the Devonian of

North America, two from the Middle Devonian resemble each other

n having only a few rows of compressed denticles (each denticle

)resumably representing a tooth ontogenetically) fused only at the

)ase. The unique tooth of D. eastmani Stauffer from Ohio was

distinguished from D. uddeni Eastman of Iowa because it had one

less major ridge of denticles; but Stauffer's figures suggest that the

inner ridge in his specimen may have been reduced by wear. D.

iherwoodi, from the "Catskill group" of Pennsylvania is of a similar

t ype. Eastman (1908, p. 214) came to the conclusion that a number

( f species from the Late Devonian of New York and Pennsylvania

\ ere variants of D. nelsoni Newberry; he showed that D. flabelli-

j wmis represented palatal tooth plates, D. levis a worn tooth plate,

238 FIELDIANA: GEOLOGY, VOLUME 11

and D. quadratics and D. minutus juvenile specimens. All are
characterized by numerous (9 or 10) ridges with the denticles fused
except at the tips. Another group of four species from the Late
Devonian is characterized by having from six to eight ridges com-
posed of denticles that are rounded in cross section (or perhaps some-
what compressed on the medial side of the plate) and fused only at
the base. Two of these, D. mordax and D. pectinatus, from the /
State Quarry beds of Iowa, were recognized as being closely similar
by their describer (Eastman, 1908, p. 223). Other species having
this type of dental plate are D. gemmatus Hussakof and Bryant
from the Genesee group of New York State and D. johnsoni Bryant
from the Chaffee formation of Colorado. It is possible that D.
fleischeri Newberry, from the "Catskill group" of New York and
Pennsylvania, is related to this group. At present there are no
entirely satisfactory characters to distinguish these species. A
number of the Late Devonian teeth from the Rocky Mountain region
belong to this type and are discussed below:

Dipterus mordax Eastman: The first definite record of Late De-
vonian lung fish in the western United States was a small collection
of dental plates described by Eastman (1915, p. 282) from the
Elbert formation, La Plata County, Colorado. One of these
(op. cit., fig. 3) was identified as a juvenile D. mordax, another (op.
cit., fig. 4) as D. pectinatus. In my opinion the former, with its
relatively narrow proportions and convex functional surface, may
represent a splenial tooth, and the latter, with its broader proportions
(although not as restored), may be a palatal tooth of the same
species, D. mordax. An incomplete tooth collected in the same re-
gion in 1949 (PF 583) is probably also referable to D. mordax.

In 1936, Bryant (Bryant and Johnson, 1936, pp. 658-659)
described D. johnsoni from the Parting member of the Chaffee
formation in two localities in central Colorado. This species agrees
very closely with D. mordax, to which it probably belongs.

Species based on worn dental plates : A number of North American
species of Dipterus have been based on what are presumably worn
dental plates. It is possible that they belong to some of the species
mentioned above, but without a series including intergrading forms, ,
their exact relationship must remain uncertain. D. contraversus from
the "Catskill group" of Pennsylvania is such a type. Another is
D. calvini, which does not seem to be related to D. uddeni, associated
with it in the Cedar Valley limestone of Iowa. D. costatus, from
the State Quarry beds of Iowa, appears to have more compressed

DENISON: DEVONIAN FRESH-WATER FISHES 239

denticles than the D. mordax-D. pectinatus group in the same
formation. The teeth from the State Quarry beds referred to D.
digitatus have much of the functional surface smooth, with a tumid
ridge on the inner margin, features suggestive of worn tooth plates.
Some of the tooth plates of this species figured by Eastman (1908,
pi. 7, figs. 16-25) have the distal part of the ridges only slightly
worn, with the individual denticles clearly distinct; others have the
ridges more completely worn, shorter, and with the denticles hardly
discernible. There is no great gap between the latter and the dental
plates figured by Eastman (op. cit., pi. 8) as Conchodus variabilis.
Considering the differences between young and worn teeth demon-
strated in Ceratodus parvus (Stromer and Peyer, 1917, pp. 26-32),
and C. africanus (Peyer, 1924, pi. 1), it seems very probable that
Conchodus variabilis Eastman represents dental plates of Dipterus
digitatus showing even greater wear. Many of the dental plates
of D. digitatus figured by Eastman (op. cit., pi. 7, figs. 20, 22, 24)
are suggestive of D. mordax in having discrete denticles, roundish
in section. Their relationship to the latter species, however, cannot
be demonstrated from the figures.

Dipterus digitatus has been recorded in the Early Mississippian,

Bushberg sandstone of Missouri and is known from the Late Devonian

of the Rocky Mountain region as well. A greatly worn tooth from

the Elbert formation of southwestern Colorado (Eastman, 1915,

fig. 2) was identified as D. digitatus; it could conceivably belong to

an old individual of D. mordax. An incomplete and highly worn tooth

of this type in the University of Michigan collection (no. 27549)

was collected in the Darby formation near Teton Pass, Teton County,

Wyoming. The functional surface is flat or slightly concave, and the

ridges are short and low, with only a suggestion of the original

denticles. Another tooth showing a superficial resemblance to

D. digitatus was collected in 1950 by Bryan Patterson, 108 feet above

the base of the Jefferson formation, in Blacksmith Fork Canyon,

Bear River Range, Utah (PF 495). It resembles D. digitatus in

laving nearly two-thirds of the functional area worn smooth, and

n having an inner tumid ridge. From its narrow proportions and

•on vex functional surface it is probably a splenial tooth plate. Its

ength is only 9 mm. and a worn tooth is hardly to be expected in

uch a small individual of D. mordax or D. digitatus. This may well

represent a distinct species, but is hardly definable from a single

vorn specimen.

Other records of Dipnoi: Undescribed and unidentified Dipterus
« ental plates have been reported from the Jefferson formation near

240 FIELDIANA: GEOLOGY, VOLUME 11

Logan, Montana (Eastman, 1917, p. 246) and from the Jerome
formation, near Flagstaff, Arizona (Hussakof, 1942, p. 9).

A new species of Conchodus, C. parvulus, was described by Bryant
(Bryant and Johnson, 1936, p. 659) from the Parting member of
the Chaffee formation at Cross Mountain, Gunnison County,
Colorado. This is the third species referred to this genus in North
America. The first, C. plicatus Dawson, from the Coal Measures
of Nova Scotia, was believed by Woodward (1891, p. 255) to repre-
sent an "abraded dental plate of Ctenodus." The second, C. vari-
abilis Eastman, from the Late Devonian of Iowa, intergrades with
Dipterus teeth, as Eastman himself recognized (1898, p. 113); as
was suggested above, it probably is founded on worn teeth of Dipterus.
Bryant's species, C. parvulus, may also represent worn teeth of
Dipterus johnsoni (=D. mordax), which is found in the same forma-
tions at the same locality.

The genus Synthetodus was considered by Eastman (1898, pp.
112-113; 1908, p. 231) to represent dental plates of Dipnoi, but
this was questioned by Hussakof and Bryant (1919, p. 151), who
provisionally referred it to the Cochliodontidae. Whatever the
relationships of Synthetodus, the partial plate from the Elbert
formation of southwestern Colorado, figured by Eastman (1915,
fig. 1), shows little resemblance to the type material from the Late
Devonian of Iowa. Probably it is the pavement tooth of one of
the Bradyodonti.

CROSSOPTERYGII

HOLOPTYCHIIDAE

Holoptychius cf. giganteus Agassiz: Eastman (1904, pp. 257-258)
reported two species of Holoptychius from the Elbert formation of
Endlich Mesa, La Plata County, Colorado. His identification was
based upon scales alone, those having an ornamentation of irregular,
tortuous ridges being referred to H. giganteus Agassiz, and those
ornamented with tubercles to H. tuberculatus Newberry. Scales
from the same horizon and locality in Chicago Natural History
Museum resemble those described by Eastman. They demonstrate
that the ridged and tuberculated types of scale form an intergrading
series, and that both fall within the limits of H. giganteus as defined
by Agassiz (1845, pi. 24, figs. 3-10). There is little reason to believe
that two species are present in the same bed at this restricted locality,
especially when they are distinguished only by scales, which are
hardly sufficient to define a species of this genus satisfactorily. No

DENISON: DEVONIAN FRESH-WATER FISHES 241

character is known by which the Elbert formation scales may be
distinguished from those of H. giganteus, although if more complete
remains were available, the Colorado Holoptychius would probably
prove to be specifically distinct.

Other specimens obtained from the same locality and horizon
include a natural cast of a lower jaw fragment, the impression of
the inner side of the anterior part of the skull roof, as well as some
impressions of deeply ornamented dermal bone fragments. Pre-
sumably all belong to Holoptychius.

Holoptychius sp.: There are two published records of crossop-
terygians from Fossil Ridge, Gunnison County, Colorado. Cockerell
(1913, p. 246) mentioned a fragment covered with vermiform ridges
that he compared to Rhizodus, but his description suggests the
ornament of Bothriolepis coloradensis. Bryant (Bryant and Johnson,
1936, p. 657) compared a crossopterygian from this locality to
Glyptopomus sayrei. My collection from Fossil Ridge includes
crossopterygian fragments, but only of Holoptychius. The scales
resemble some of those from Endlich Mesa referred to H. cf . giganteus;
the exposed part is ornamented with wavy, branching ridges, but
no tubercles. The teeth clearly belong to the Holoptychiidae. The
fang teeth are slightly curved, flattened with two crests in the crown,
and round in section without crests, but with many grooves in the
lower part. The enamel is smooth on the crown, but has a number
of sharp ridges between the grooves near the base. Thin sections
are not satisfactory because of abundant burrows attributed to fungi,
2alled Mycelites ossifragus (Peyer, 1945). It is possible, however,
to determine that there are complex, branched dentine folds in the
ower part of the teeth, and that enamel and cement do not enter
:he folds. The structure resembles that of Holoptychius and Glypto-
epis as described by Bystrow (1939).

Scales from the Temple Butte formation in the Grand Canyon

of Arizona were identified as belonging to Holoptychius by Gidley

Schuchert, 1918, p. 361). They were not described, nor have I

been able to find them in the collections of the United States National

Museum.

RHIZODONTIDAE

Litoptychus1 gen. nov.

Type species. — Litoptychus bryanti sp. nov.

1 Xitoj, simple+*-Ti>x°s> fold; in reference to the manner in which the dentine
o the teeth is infolded.

242

FIELDIANA: GEOLOGY, VOLUME 11

Diagnosis. — Scales cycloidal, very thin, ornamented in exposed
portion with wavy anastomosing ridges, and with a fan-shaped
arrangement of rows of fine tubercles just anterior to the ridges.
Fang teeth robust, oval in section, lacking trenchant edges, with
furrows and grooves only in the lower half; dentine infolded mostly
in very simple fashion at the base, with occasional simply branched

Fig. 42. Litoptychus bryanti, gen. and sp. nov. A, Inner side of scale, PF 619
(X 2). B, Fragment of exposed area of scale, PF 617 (X 6).

infoldings. Dermal bones mostly ornamented with tubercles united
at least at the base by ridges. Dentary with internal expansion at
anterior end of jaw overlying the symphysis and the prearticular;
coronoids bearing large fang teeth, the dentary a small one. Verte-
brae ring-shaped.

Litoptychus bryanti,1 sp. nov.

Type.— C.N.H.M.-PF 610 (fig. 46), incomplete left lower jaw
showing inner side, and preserving most of the dentary, intercoronoid
and precoronoid, as well as parts of the prearticular and infradentalia.

Horizon. — Late Devonian. Lower member of the Chaffee forma-
tion (bed 29 of Johnson, 1944, pp. 362-363).

Locality.— Deadman's Creek, NW. %, sec. 23, T. 14 S., R. 84 W.,
Gunnison County, Colorado.

Diagnosis. — As for the genus. Scales as much as 30 mm. in
diameter. Lower jaw attaining an estimated length of 210 mm.

1 Named after the late W. L. Bryant, who gave the first description of Devonian
fishes from central Colorado.

DENISON: DEVONIAN FRESH-WATER FISHES 243

Discussion. — In 1936 Bryant (Bryant and Johnson, 1936, p. 657)
reported the presence of a large crossopterygian in the lower member
of the Chaffee formation on Deadman's Creek, Gunnison County,
Colorado. A small collection from this locality in Chicago Natural
History Museum consists mainly of fragments of a large fish that is
surely the same one mentioned by Bryant. Apparently it belongs
to the Rhizodontidae, and thus cannot be related closely to Glypto-
pomus sayrei, to which Bryant compared it. Nor can it be referred
at present to any of the genera of Rhizodontidae hitherto described.
It should be remarked, however, that many of the genera of this
family are unsatisfactorily defined on the basis of tooth and scale
characteristics, and until such time as a revision is undertaken the
affinities of many members of this family must remain uncertain.

The scales of Litoptychus bryanti are large and cycloidal; those
in the collection measure about 30 mm. in the antero-posterior
diameter, and about 25 mm. in the shorter diameter. They are
extremely thin, only about 0.25 mm. in thickness in the overlapped
portion. The inner surface (fig. 42, A) shows a few faint concentric
growth lines, especially near the periphery, and is marked with
numerous pores; there is no central boss or ridge. The outer surface
was not exposed on any of the specimens collected, and preparation
was difficult because of their extreme thinness. The overlapped area
of the scale is probably large and is marked with a very fine-meshed,
elongated reticulation. Thin sections reveal that the surface here
is composed of the middle or cancellar layer. Only one small frag-
ment (PF 617) reveals the nature of the ornamentation of the
exposed area of the scale (fig. 42, B). This consists of a fan-shaped
arrangement of rows of very fine tubercles directed anteriorly, and
a number of low ridges directed towards the posterior border; the
ridges are wavy and anastomosing. In this ornament, they show
some resemblance to Strepsodus (Woodward, 1891, pi. 16, fig. 1),
to Glyptolepis, and perhaps to Sauripterus (Eastman, 1907, pi. 7,
fig. 9). The tubercles and ridges of the scales are presumably com-
Dosed of bone, not dentine, although thin sections through this part
)f the scale were not obtained. Most of the thickness of the scale
s composed of the basal laminar layer, with the younger and deeper
aminae obliquely overlapping the older, more superficial laminae
near the periphery.

The fang teeth are robust, rapidly tapering, straight, and oval
i a section throughout their height. The upper half of the tooth has
j smooth surface, while the lower half is marked with approximately

244

FIELDIANA: GEOLOGY, VOLUME 11

Fig. 43. Litoptychus bryanti, gen. and sp. nov. (X 25). A, Transverse section
through base of fang tooth, PF 619. B, Obliquely transverse section through
base of palatal fang, PF 616. bif, branched infolding of dentine; d, dentine;
en, enamel; p, pulp cavity; pc, pulp canals; sif, simple infolding of dentine.

twenty deep furrows; the ridges between these furrows are occasion-
ally marked by shorter, shallower grooves. These teeth show no
fine ridges or trenchant edges. Cross sections of fang teeth indicate
that the structure is relatively simple. One tooth (PF 619), approxi-
mately 7 mm. in maximum diameter, shows at the base only simple,
open unbranched infoldings of the dentine; there is no indication of
cement in the infoldings and enamel is restricted to the outer surface
of the ridges (fig. 43, A). Another tooth (PF 616), 5.2 mm. in
maximum diameter, has a number of large unbranched folds at the
base, but on one side (the section is oblique, and this is presumably
lower) there are simply branched infoldings (fig. 43, B). At mid-
height there is no trace of any infolding, and the dentine tubules
have a simple radial arrangement. The fang-tooth structure is
much simpler than that described in Eusthenopteron and Panderich-
thys (Polyplocodus) (Bystrow, 1939).

The dermal bones of the skull and jaw are ornamented with rather
small tubercles united near their bases, and sometimes to mid-height

DENISON: DEVONIAN FRESH-WATER FISHES

245

or to the top (fig. 45). Superficially, the ornament resembles that
of Eusthenopteron. An undetermined fragment (PF 618) has an
elongate reticular ornament, similar to that figured by Gross (1936,
fig. 1) on the pterygoids of Porolepis and Panderichthys (Polyplo-
codus). In thin sections (fig. 44), the structure of the tubercles is
not clear, presumably because of some peculiarity of preservation,
but there is probably no dentine or enamel. The tubercles appear
to be formed of bone of a lamellar type, as in Eusthenopteron (By-
strow, 1939, fig. 18, A). An expansion of the dentary near the
symphysis bears true denticles composed of dentine and possibly
enamel. Whether the prearticular bore similar denticles, as is the
case in many crossopterygians, is not known. The dermal bones
have a well-developed middle or cancellar layer and a thick basal
laminated layer.

Three specimens of lower jaws are available, one (PF 613, fig. 45)
showing a nearly complete outer surface, another (PF 610, fig. 46)
the anterior two-thirds of the inner side, and the third (PF 619)
the inner side of the external bones of the posterior part. PF 613
indicates a relatively deep jaw, the ratio of the depth at the inter-
coronoid to the total length being 0.22. Its proportions agree quite
well with those of the Holoptychiidae. The known Devonian
Rhizodontidae have a more slender jaw, although some Carboni-
ferous genera show a deepening. The considerable depth of the

Fig. 44. Litoptychus bryanti, gen. and sp. nov. (X 25). Vertical section
hrough dermal bone showing a thick basal laminated layer, a middle cancellar
ayer, and an upper layer consisting probably of laminated bone.

.aw is due to the deepening of the dentary anteriorly, as well as to
; n expansion of the infradentaries, especially the splenial.

The dentary shows a progressive feature; it is expanded hori-
: ontally in the symphysial region, where it comes to overlie the
symphysis and the anterior end of the prearticular (fig. 46, dn{).

246 FIELDIANA: GEOLOGY, VOLUME 11

In addition to the three large fang teeth borne by the coronoids, there
is a small fang, 5 mm. in maximum diameter, in the dentary at the
anterior end of the jaw (fig. 46, df) . Panderichthys is the only other
Devonian crossopterygian with dentary fangs (Gross, 1941b, figs.
18, 19) ; they are probably present in all the Late Paleozoic genera,
certainly in Rhizodus and Rhizodopsis (Traquair, 1877, pp. 302-303)
among the Rhizodontidae, and in Ectosteorhachis (=Megalichthys

Fig. 45. Litoptychns bryanti, gen. and sp. nov. (X 34)- Right lower jaw,
PF 613, outer side.

Ag. 1844) among the Osteolepidae. The medial part of the dentary
expansion that overlies the prearticular is covered with denticles,
presumably composed of dentine and enamel.

Three coronoid bones are present, although the most posterior
is not well shown in the available specimens. The first two, incom-
pletely preserved in PF 610, bear larger fang teeth than does the
dentary; the fang of the precoronoid is 11 mm. in diameter at the
base, that of the intercoronoid 9 mm.

On the external surface (fig. 45) the sutures are poorly shown,
only that between the dentary and infradentaries being partly
discernible. At the level of the posterior fang tooth the dentary
starts to deepen and in the anterior third of the jaw is very deep.
This differs from the situation in Holoptychius and Glyptolepis,
where the considerable depth of the anterior part of the jaw is due
more to the expansion of the splenial, and the dentary remains
relatively shallow.

A number of other fragmentary remains of Litoptychus bryanti
are available. An operculum (PF 612), exposed on the inner side,
agrees in general shape with that of Eusthenopteron. A parasphenoid
(PF 614) is of the long narrow type characteristic of the Osteolepidae
and Rhizodontidae. Fragments of the endocranium and of the
dermal bones of the skull are present in the collection, but are not

DENISON: DEVONIAN FRESH-WATER FISHES

247

sufficiently complete to offer any diagnostic characters. The verte-
brae (PF 611, 619) have oval, ring-shaped centra with large noto-
chordal foramina. There is no evidence of the paired lateral processes
(?parapophyses) such as are present in Strepsodus, and no neural
or haemal arches have been recognized. Scattered fin supports are
common.

The relationship of Litoptychus to the Rhizodontidae is indicated
by the simple manner in which the dentine of the fang teeth is
infolded, and by the thin cycloidal scales. From Eusthenopteron,
it may be distinguished by the presence of an anterior, fang-bearing
expansion of the dentary, the greater depth of the lower jaw, the
structure of the fang teeth, the ring vertebrae, and the scale orna-
mentation. Tristichopterus is closely related to Eusthenopteron and
probably differs from Litoptychus in similar features. Panderichthys
possesses a dentary fang, but has a very peculiar and distinctive
tooth structure, and rhombic scales set with tubercles. Sauripterus
is insufficiently known. It may resemble Litoptychus in its scales
and dermal bone sculpture, but it differs in having a shallow jaw
and fang teeth with trenchant edges. Rhizodus is characterized
particularly by its fang teeth, which possess one or two sharp
trenchant edges. The dentary fang is much larger than the coronoid
fangs in the jaw figured by Woodward (1891, pi. 12, fig. 1). In
Rhizodopsis, the teeth lack the cutting edges and the scales are very

anq

Fig. 46. Litoptychus bryanti, gen. and sp. nov. (X %). Type, PF 610,
ncomplete left lower jaw, inner side, ang, angular; cfi, cfo, coronoid fangs; coru
precoronoid; cor2, intercoronoid ; df, dentary fang; dn, dentary; dnu symphysial
< xpansion of dentary; drii, denticulate expansion of dentary overlying prearticular;
j'art, prearticular; sy, symphysis.

1 hin as in Litoptychus, but the scales are typically of a narrow ovoid
i hape and the dentary fang is apparently larger than those borne by
1 he coronoids, as in Rhizodus. Strepsodus is incompletely known and

248

FIELDIANA: GEOLOGY, VOLUME 11

is characterized at present chiefly by its slenderly tapering teeth
with a peculiar curvature.

Litoptychus cannot be assigned to any of the previously described
rhizodontid genera as they are now defined. It is clearly a rather
progressive genus for the Devonian, showing many features in com-
mon with the Carboniferous members of the family.

OSTEOLEPIDAE

In addition to the common, large rhizodontid in the Chaffee
formation at Deadman's Creek, Gunnison County, Colorado, there

Fig. 47. Osteolepidae indet., from Parting member of Chaffee formation,
Colorado. A, Tuberculate scale, PF 603 (X 4), Spar Gulch, near Aspen. B,
Smooth scale, PF 599 (X 9). C, Scale with superficial layer partly covering
exposed area, PF 598 (X 4). B and C from Deadman's Creek, Gunnison County,
Colorado.

are occasional scales and dermal plate fragments of a small fish
belonging to the Osteolepidae. The scales have a rhombic exposed
area, 4.0-7.5 mm. in length, covered with a smooth superficial
layer of dentine and enamel (fig. 47, B). The overlapped area is
relatively narrow and is separated from the exposed area by a narrow
groove. The inner side has a central longitudinal ridge or boss. A
thin section shows the typical osteolepid structure, with a well-
developed laminated basal layer, a middle cancellar layer that forms
the surface of the overlapped area, and in the exposed area a thin
superficial layer with the usual flat-topped denticles of dentine and
enamel. Many details of the structure are obscure because of the
abundance of borings attributed to fungi (Mycelites ossifragus).
The dermal plate fragments are usually completely covered with
dentine and enamel, and frequently show "Westoll lines" between

DENISON: DEVONIAN FRESH-WATER FISHES 249

different zones of dentine deposition. One scale (PF 598, fig. 47, C)
is interesting in lacking the superficial layer on the anterior part
of the exposed area. This part of the scale is ornamented with bony
tubercles, probably largely composed of the upper part of the
cancellar layer. The more posterior part of the exposed area is
covered with a layer of dentine and enamel, thin enough to show
in its preserved state some suggestion of the underlying structure
of the cancellar layer. Since this scale was prepared under a micro-
scope, it is certain that the absence of the superficial layer in the
anterior part of the exposed area was not due to loss in collecting
or preparation; it may have been resorbed, or more probably not
yet deposited in this region.

As Jarvik (1948, p. 283) has pointed out, there are "hardly any
differences . . . between scales from different Osteolepid genera and
species which are of sufficient importance to warrant their use in
taxonomy." Thus it is not worth while to attempt to identify
these fragmentary remains, and until better material is available
they will be referred to as Osteolepidae indet.

As was mentioned above, fragmentary fish remains were col-
lected in 1896 by Tower from the Late Devonian Parting Quartzite
near Aspen, Pitkin County, Colorado. A few teeth in this collection
were attributed to crossopterygians by Girty (Spurr, 1898, p. 21)
and compared to Rhizodus and Eusthenopteron. This collection was
also studied by Eastman (1904, p. 258; 1915, p. 281). In addition
to the crossopterygian teeth, he mentioned some finely tuberculated
plates that he attributed to arthrodires, and some smooth, perforated
scales that he believed to belong to dipnoans. After seeing some of
Tower's collection (although not the teeth) in the United States
National Museum, and collecting near the base of the Parting
Quartzite at this locality, I have come to the conclusion that all
of the remains belong to an osteolepid crossopterygian. My own
collections were obtained in 1949 both from the point of East Aspen
Mountain, overlooking Aspen, and from the bed of Spar Gulch
nearby, but the description below is based on the better preserved
Spar Gulch material.

The material consists of scales and plates only. The dermal
I lates and most of the scales are covered with a fine tubercular
crnament, with the tubercles joined more or less by ridges. Pre-
simably these are the plates referred by Eastman to arthrodires.
The scales (fig. 47, A) have a rhombic exposed area on which the
t ibercles are high and quite independent anteriorly, lower and

250

FIELDIANA: GEOLOGY, VOLUME 11
-d

Fig. 48. Osteolepidae indet., from Aspen, Colorado (X 25). A, Vertical sec-
tion through dermal bone, PF 606. B, Vertical section through part of a scale,
PF 604. bl, basal laminated layer; d, dentine; den, dentine-covered denticle;
ea, exposed area of scale; en, enamel; ml, middle cancellar layer; oa, overlapped
area of scale; re, area of resorption.

joined nearly to the top by ridges posteriorly. The overlapped
area on two sides of the rhomb is relatively narrow. A thin section
(fig. 48, A) reveals that there is a thick basal laminar layer and a mid-
dle cancellar layer; the tubercles are largely bone but are capped by
a thin layer of dentine. A few of the scales are not tuberculate, but
are covered with a smooth superficial layer in the exposed area.
Presumably these are the scales that Eastman attributed to dipnoans.
A thin section (fig. 48, B) of one of these scales reveals a thin, super-
ficial layer overlying dentine-capped tubercles. The superficial
layer is typically osteolepid in structure, consisting of closely spaced
flat-topped denticles of dentine and enamel or vitro-dentine. It is
not necessary to assume that the smooth and tuberculate types of
scales belonged to different fishes. Jarvik (1948, 1950) cites a
number of cases where scales are ornamented in the Osteolepidae.
For example, Gyroptychius groenlandicus usually has a thick, shiny
superficial layer on its scales, yet in one associated group of scales
a few lack the superficial layer and are tuberculate (Jarvik, 1950,
pi. 22, figs. 2-4) ; the latter resemble scales from Aspen, except that
their tuberculation is finer. The scales from Aspen possessing a
superficial layer of dentine overlying dentine-capped tubercles are
particularly interesting (fig. 48, B). It is probable that the tubercles

DENISON: DEVONIAN FRESH-WATER FISHES 251

were originally superficial (as they still are in most of the scales and
dermal plates). The deposition of a smooth layer of dentine (and
?enamel) over earlier dentine must have involved a spreading super-
ficially of ectodermal and mesodermal tissue. This could only have
been accomplished from the regions between the tubercles where
dentine had not been developed.

The reference of the remains from Aspen to the Osteolepidae is
based largely on the rhombic shape of the scales, and the small over-
lapped area; the occasional presence of a smooth superficial layer
supports this. As is the case in the osteolepids from Deadman's
Creek, generic identification is not possible.

UNDETERMINED CROSSOPTERYGII

A fragment of a lower jaw (PF 609) from the Darby formation
on the South Fork of the Little Wind River, Wyoming, belongs to
a crossopterygian but it is not more exactly determinable. The
outer face is ornamented with tubercles, united at their bases, and
the marginal teeth of the dentary are relatively large, striated, and
probably widely spaced.

Another lower jaw of a crossopterygian (PF 602) from a limestone
in the Elbert formation on Little Cascade Creek south of Columbine
Lake, La Plata County, Colorado, is also undeterminable. Nearly
the entire jaw is present but is poorly preserved. The external face
is ornamented with tubercles, and the ramus is rather shallow.
Probably this does not belong to Holoptychius, which is abundant
elsewhere in the Elbert formation.

Bryant (Bryant and Johnson, 1936, p. 657) reported a cross-
opterygian from the Parting member of the Chaffee formation at
dribbles Creek, Fremont County, Colorado. Along with other
(•rossopterygians from central Colorado, he compared it with Glyp-
lopomus sayrei, but this identification cannot be accepted without
confirmation.

GEOLOGICAL OCCURRENCE

Detailed geological work has been done on only small parts of
t le area covered in this study. However, sufficient information is
a mailable so that it is now practical to attempt a reconstruction of
tie conditions under which fishes were preserved in the Late
I evonian of the western states. Such an attempt is useful not only
a a basis for future field work but also is essential in assessing the

252 FIELDIANA: GEOLOGY, VOLUME 11

general zoological as well as the geological significance of the fish
faunas.

The geological history of this region, as it may now be recon-
structed from published reports is as follows: At the beginning of
the Devonian (Helderberg), the seas did not cover any of this area.
Presumably relief was low or moderate, since erosion without any
major orogenies had been at work since the Cambrian or Ordovician.
In Oriskany times, the seas advanced in the Cordilleran geosyncline,
but did not overlap to any extent on to the shelf lands beyond. In
Onondaga times, this sea withdrew again, and deposits of this age
are found only in the Cordilleran trough in Nevada, Utah, and
California.

At the beginning of the Late Devonian (Genesee) a much greater
transgression began. In the Cordilleran trough of Nevada and
California the Stromatopora zone of the Devil's Gate formation and
equivalent formations were being deposited. Farther north and east,
the deposition of the Jefferson formation may have started at this
time in northern Utah and southeastern Idaho. The Canutillo
formation of the Franklin Mountains of Texas may have heralded
a similar advance of the sea in the Sonoran trough.

By Naples time, the sea had advanced farther, although it was
still restricted to the geosynclines. In Nevada, the Spirifer argen-
tarius zone of the Devil's Gate formation is of this age, in western
Utah possibly part of the Guilmette formation, while the Jefferson
formation was being deposited farther north in northern Utah,
southeastern Idaho, northwestern Wyoming, and southwestern
Montana. In the Sonoran trough, marine rocks of this age have
been reported in northern Sonora, and the seas may have reached
southern Arizona to deposit the Picacho de Calera formation.

Fig. 49. Paleogeographic map showing the maximum extent of the Late
Devonian sea and the total thickness of the Devonian sediments. The encircled
numbers indicate the localities at which the Late Devonian fishes discussed in this
paper have been found. (1) Darby formation, Teton Pass, Wyoming; (2) Darby
formation, southwestern side of Wind River Mountains, Wyoming; (3) Darby
formation, South Fork of Little Wind River, near Fort Washakie, Wyoming;
(4) Chaffee(?) formation, Glen wood Springs, Colorado; (5) Parting Quartzite,
Aspen, Colorado; (6) Parting member, Deadman's Creek, Colorado; (7) Parting
member, Fossil Ridge, Colorado; (8) Parting member, Gribbles Creek, Colorado;
(9) Elbert formation, Rockwood and Little Cascade Creek, Colorado; (10) Elbert
formation, Endlich Mesa, Colorado; (11) Temple Butte formation, Grand Canyon,
Arizona; (12) Jerome formation, near Flagstaff, Arizona; (13) Jerome formation,
near Payson, Arizona.

.EGEND :

0-500'

500 - 1000

1000'- 2000' 2000'- 3000' 3000'- 5000'

253

254 FIELDIANA: GEOLOGY, VOLUME 11

In the Chemung stage, the seas spread out of the geosynclines
on to the bordering shelf lands and inundated land that had not
been under water since Ordovician, Cambrian or pre-Cambrian
times. Far from shore in the Cordilleran geosyncline, the Philips-
astraea zone of the Devil's Gate formation and parts of the Jefferson
formation and Grand View dolomite were being deposited. On the
bordering shelf the sea formed shallow embayments in Wyoming,
Colorado, and Arizona. On the Wyoming shelf, deposits of this
age are known as the Darby formation, in the Central Colorado
basin they are called the Parting member of the Chaffee formation
(farther east the Williams Canyon formation may be an equivalent)
and in the Southwest Colorado basin they are the Elbert formation.
At the same time, the seas spread southeastward into central Arizona
where the Temple Butte and Jerome formations were deposited,
and also northward out of the Sonoran trough to form the Sly Gap
formation in southern New Mexico and the Martin limestone in
the Southeast Arizona basin. An attempt has been made in figure 49
to reconstruct the paleogeography at this time of maximum extent
of the Devonian seas.

In Chautauquan times, seas remained extensive. In the Cordill-
eran syncline, the Cyrtospirifer zone of the Devil's Gate formation
represents this period in Nevada, while the upper Grand View
dolomite was laid down in parts of Idaho. In shallower waters, or
in receding seas, the Three Forks shale succeeded the Jefferson
formation in Montana, Idaho, and Wyoming. The upper part of
the Darby formation of Wyoming, the Dyer dolomite member of
the Chaffee formation of central Colorado, and the Ouray limestone
of southwestern Colorado may be in part of this age. In the North-
western Arizona basin, part of the Jerome formation and the Island
Mesa beds are possibly Chautauquan. To the south, on the shelf
bordering the Sonoran trough, the Ouray limestone is recognized
in southeastern Arizona, and the Percha shale of southwestern New
Mexico may be of this age (although it is considered by some to be
younger) .

In the latest Devonian (Conewango stage) the seas withdrew
not only from the shelf but also from the geosynclines. The so-called
"Three Forks with Syringoihyris" which is attributed by some to
this age, may belong to the Mississippian.

As far as I know, no fishes have been discovered in the Late
Devonian rocks of the Cordilleran or Sonoran geosynclines. The
limestones deposited in the open waters of the bordering shelves

DENISON: DEVONIAN FRESH-WATER FISHES 255

contain remains of typical marine fishes such as cladoselachian
sharks, hybodonts, bradyodonts, and ptyctodonts; sometimes teeth
of lung fishes are discovered in such limestones. In the great em-
bayments of the Chemung-Chautauquan sea remains of different
types of fishes are common in the Darby, Chaffee (Parting member),
Elbert, Temple Butte, and Jerome formations (fig. 49). The first
three of these formations are very similar lithologically and are
distinguished mainly by their geographical position; the Darby
formation is restricted to western Wyoming and adjacent Idaho,
the Chaffee formation is found only in west-central Colorado, and
the Elbert formation outcrops only in southwestern Colorado. The
three formations consist of a variable series of sandstones, con-
glomerates, shales, and limestones that are usually sandy, and often
dolomitic. Cross bedding, ripple marks, mud cracks, and salt
crystal casts indicate very shallow water deposition with occasional
sub-aerial desiccation. Under these circumstances the large amount
of elastics suggests that deposition was near to shore. Presumably
the beds were deposited on beaches, mud flats, lagoons, estuaries,
or river mouths on a shore of low relief; some beds may even have
been deposited in fresh-water stream channels and flood plains.
These formations represent a near-shore facies of the transgressing
Late Devonian sea, and as such cannot be of exactly the same age
wherever they occur. After the shoreline had advanced, usually
more limy beds were deposited farther from shore. Thus in Wyoming
the upper part of the Darby formation contains less clastic material
and more limestone, with occasional crinoids. In central Colorado,
the Parting member is succeeded by the Dyer dolomite of the Chaffee
formation, with a marine invertebrate fauna. In southwestern
Colorado, the Ouray limestone overlies the Elbert formation and
has a rich marine invertebrate fauna.

Most of the fishes are found in the lower near-shore transgressive
facies and are about the only fossils known from it. (The absence
of invertebrates is probably due to non-preservation, for they surely
existed in the near-shore environments.) As far as the geological
evidence goes, all could be littoral marine fishes, but such is surely
not the case. The most common genus, Bothriolepis, with a world-
wide distribution, is everywhere characteristic of the continental
"Old Red Sandstone" type of deposits. In fact, the whole order
of Antiarchi is believed to have been restricted to fresh water with
the exception of two highly specialized marine genera, Grossapis
(Ceraspis) and Lepadolepis (Ceratolepis) . Holoptychius is commonly

256 FIELDIANA: GEOLOGY, VOLUME 11

associated with Bothriolepis and is likewise a characteristic member
of the Late Devonian, Old Red Sandstone faunas. Rhizodontidae
and Osteolepidae are elsewhere found in continental deposits in the
Devonian, although some of the Mississippian and Pennsylvanian
genera moved into the seas. The occurrence of lung fishes, on the
other hand, does not allow the assumption that they were exclusively
a fresh-water group in the Devonian. Dipterus does occur com-
monly in continental deposits both in Europe and in North America,
but it is found also in marine limestones, associated with marine
invertebrate faunas, in some cases far from the nearest shore line.
An outstanding example of a marine occurrence of Dipterus in this
country is the Late Devonian State Quarry limestone of Iowa,
which contains teeth of this genus in abundance, along with ptyc-
todonts and marine invertebrates. This association surely suggests
that some Dipterus inhabited the seas, even if their teeth had been
dropped in this area by a predator that had eaten them, as Eastman
(1908, pp. 216-217) suggested. The two Dipterus teeth from the
Jefferson formation, one found near Logan, Montana, and the other
found on Blacksmith Fork, near Logan, Utah, occur in marine
limestones, presumably deposited at some distance from the shore.
In the Late Devonian, arthrodires were predominantly a marine
group, although a few lingered on in rivers. Their rarity in the
faunas of this region may be taken as supporting the view that the
fishes came mostly from a fresh-water environment.

It would thus appear that most of the fishes of the Darby and
Elbert formations, and of the Parting member of the Chaffee forma-
tion, were inhabitants of fresh-water streams and were washed after
death into the edges of the advancing Chemung sea. The nature
of the preservation supports the view that there was considerable
post-mortem transportation of the fish remains in most cases. Gen-
erally the fossils are isolated plates, scales or fragments, obviously
without any association. "Bone beds" consisting largely of worn
and sorted fish fragments occur near the base of the Elbert and
Darby formations; probably the beds represent concentrations by
current action on the sea bottom, or by wave action on the shore.
Better preservation of fishes is most unusual in these Late Devonian
sediments. At certain horizons in the Parting member on Dead-
man's Creek, Gunnison County, Colorado, there are abundant
remains of Litoptychus bryanti, including many entire bones as
well as jaws and large parts of skulls. It is possible that these
remains represent elements of one or a few individuals, although
association has not been demonstrated. If, as seems probable,

DENISON: DEVONIAN FRESH-WATER FISHES 257

there was less transportation in this case, the thin-bedded sandy
limestones in which the fish occur were probably deposited near
shore, perhaps near a river's mouth. The only articulated remains
that have been found are intact, though poorly preserved, shields
of Bothriolepis color adensis. They were collected by Cross near
Rockwood, and I also collected them ten feet above the base of the
Elbert formation, two miles southeast of Devon Point, Endlich Mesa.
Both localities are in La Plata County, Colorado. The sandstone
in which they occur may not be marine, but a deposit in the stream in
which the Bothriolepis lived.

From one locality in the Central Colorado basin has been reported
a fish fauna containing definite marine elements. It occurs twenty
feet above the base of the Parting member of the Chaffee formation,
on Gribbles Creek, Fremont County, Colorado (Bryant and Johnson,
1936). Remains of an undetermined crossopterygian and a tooth
of Dipterus may be of fresh-water origin, but the arthrodire plate
fragments more probably belonged to marine fishes, and this is
surely the case with the cladoselachian shark, Ctenacanthus, and
the bradyodont, Sandalodus. These fragments are the only clear
record of marine fishes in the Colorado and Wyoming embayments.

Little information is available on the Devonian geography of
the Grand Canyon region of Arizona. The Temple Butte formation
was deposited on a rugged surface with as much as 100 feet of relief.
Fishes have been found for the most part in its basal sandstones and
conglomerates. Considerable limestone occurs, although it is often
sandy, and in it a few casts of marine invertebrates have been dis-
covered. It seems probable that this is a near-shore facies, deposited
in the advancing Chemung sea. Then the only fishes reported,
Bothriolepis and Holoptychius, must have been transported from a
nearby land. Since the Paleozoic rocks of northern Arizona and
southern Utah are nearly completely buried below later rocks of
the Colorado Plateau, little can be said about the position and con-
figuration of this land that must have existed nearby. According
to McKee (1951, pi. 1, fig. B) a large promontory projected from
the Defiance-Zuni positive area, northwestward between the South-
west Colorado basin and the Northwest Arizona basin (fig. 49).

The fauna of the Jerome formation of central Arizona is related
itratigraphically and faunally to those faunas described in this
Daper. Deposition in this region occurred near a steeper, rockier
;hore, and probably for this reason there is a larger admixture of
narine fishes. This fauna will be discussed in a later paper.

258 FIELDIANA: GEOLOGY, VOLUME 11

REFERENCES

Agassiz, L.

1845. Monographie des poissons fossiles du vieux gres rouge, ou systeme
Devonien (Old Red Sandstone) des Isles Britanniques et de Russie. Neu-
chatel. pp. i-xxxvi, 1-171, pis. A-I, 1-33.

Baker, C. L.

1946. Geology of the northwestern Wind River Mountains, Wyoming. Bull.
Geol. Soc. Amer., 57, pp. 565-596, fig. 1, pis. 1-11.

Behre, C. H., and Johnson, J. H.

1933. Ordovician and Devonian fish horizons in Colorado. Amer. Jour. Sci.,
(5), 25, pp. 477-486.

Branson, E. B.

1929. New localities for Devonian fishes. Bull. Geol. Soc. Amer., 40, p. 245.

Branson, E. B., and Branson, C. C.

1941. Geology of Wind River Mountains, Wyoming. Bull. Amer. Assoc.
Petrol. Geol., 25, pp. 120-151, figs. 1-8.

Branson, E. B., and Mehl, M. G.

1930. Primitive fishes from the Devonian of Utah and Wyoming. Bull. Geol.
Soc. Amer., 41, p. 180.

Bryant, W. L.

1924. A new species of Bothriolepis from the Upper Devonian of Canada. Bull.
Buffalo Soc. Nat. Sci., 13, pp. 54-55, pis. 1, 2.

Bryant, W. L., and Johnson, J. H.

1936. Upper Devonian fish from Colorado. Jour. Pal., 10, pp. 656-659, figs.
1-5.

Bystrow, A. P.

1939. Zahnstruktur der Crossopterygier. Acta Zool., 20, pp. 283-338, figs. 1-32.

COCKERELL, T. D. A.

1913. Ordovician(?) fish remains in Colorado. Amer. Nat., 47, pp. 246-247.

Colbert, E. H., Northrop, S. A., Romer, A. S., and Simpson, G. G.

1950. Guidebook for the Fourth Field Conference of the Society of Vertebrate
Paleontology in northwestern New Mexico, pp. 1-91.

Cope, E. D.

1891. On the characters of some Paleozoic fishes. Proc. U. S. Nat. Mus., 14,
pp. 447-463, pis. 28-33.

Cross, W.

1904. A new Devonian formation in Colorado. Amer. Jour. Sci., (4), 18,
pp. 245-252.

Eastman, C. R.

1898. On the occurrence of fossil fishes in the Devonian of Iowa. Iowa Geol.
Surv., 7, pp. 108-116, 1 fig., pi. 4.

1899. Upper Devonian fish fauna of Delaware County, N. Y. 17th Ann.
Rept. N. Y. State Geol., 1897, pp. 317-327, 6 figs.

1904. Upper Devonian fish remains from Colorado. Amer. Jour. Sci., (4),

18, pp. 253-260, figs. 1-6.
1907. Devonic fishes of the New York formations. Mem. N. Y. State Mus.,

10, pp. 1-235, figs. 1-34, pis. 1-15.

DENISON: DEVONIAN FRESH-WATER FISHES 259

1908. Devonian fishes of Iowa. Iowa Geol. Surv., 18, pp. 29-386, figs. 1-41,

pis. 1-16.
1915. Dipterns remains from the Upper Devonian of Colorado. Ann. Carnegie

Mus., 9, pp. 279-283, figs. 1-4.
1917. Fossil fishes in the collection of the United States National Museum.

Bull. U. S. Nat. Mus., 52, pp. 235-304, figs. 1-9, pis. 1-23.

Endlich, F. M.

1876. [On the geology of the San Juan mining district.] Ann. Rept. U. S. Geol.
and Geog. Surv. Terr., 1874, pp. 181-240, sections I-VIII, figs. 1-3.

Foster, H. L.

1947. Paleozoic and Mesozoic stratigraphy of northern Gros Ventre Mountains
and Mount Leidy Highlands, Teton County, Wyoming. Bull. Amer. Assoc.
Petrol. Geol., 31, pp. 1537-1593, figs. 1-9.

Gross, W.

1936. Beitrage zur Osteologie baltischer und rheinischer Devon-Crossopterygier.

Pal. Zeitschr., 18, pp. 129-155, figs. 1-10, pis. 7, 8.
1941a. Die Bothriolepis-Arten der Cellulosa-Mergel Lettlands. K. Sven.

Vetens.-Akad. Handl., (3), 19, no. 5, pp. 1-79, figs. 1-45, pis. 1-29.
1941b. Ueber den Unterkiefer einiger devonischer Crossopterygier. Abhandl.

Preuss. Akad. Wiss., 1941, Math.-Nat. Kl., Nr. 7, pp. 1-51, figs. 1-27.

Hay, O. P.

1902. Description of a new species of Cladodus (C. formosus) from the Devonian
of Colorado. Amer. Geol., 30, pp. 373-374, 1 fig.

Hussakof, L.

1942. Fishes from the Devonian of Arizona. Amer. Mus. Nat. Hist., Nov.,
no. 1186, pp. 1-9, figs. 1-16.

Hussakof, L., and Bryant, W. L.

1919. Catalogue of the fossil fishes in the museum of the Buffalo Society of
Natural Sciences. Bull. Buffalo Soc. Nat. Sci., 12, pp. 1-346, figs. 1-64,
pis. 1-70.

Jarvik, E.

1948. On the morphology and taxonomy of the Middle Devonian osteolepid
fishes of Scotland. K. Sven. Vetens.-Akad. Handl., (3), 25, no. 1, pp. 1-301,
figs. 1-85, pis. 1-37.

1950. Middle Devonian vertebrates from Canning Land and Wegeners Halvo
(East Greenland). Part II. Crossopterygii. Medd. Gr0nland, 96, Nr. 4,
pp. 1-132, figs. 1-37, pis. 1-24.

Johnson, J. H.

1944. Paleozoic stratigraphy of the Sawatch Range, Colorado. Bull. Geol.
Soc. Amer., 55, pp. 303-378.

Kirk, E.

1930. The Harding sandstone of Colorado. Amer. Jour. Sci., (5), 20, pp. 456-
465.

1931. The Devonian of Colorado. Op. cit., 22, pp. 222-240.

Leidy, J.

1856a. Descriptions of two ichthyodorulites. Proc. Acad. Nat. Sci. Phila-
delphia, 8, pp. 11-12.
1856b. Descriptions of some remains of fishes from the Carboniferous and
Devonian formations of the United States. Jour. Acad. Nat. Sci. Phila-
delphia, (2), 3, pp. 159-165, 3 pis.

260 FIELDIANA: GEOLOGY, VOLUME 11

McKee, E. D.

1951. Sedimentary basins of Arizona and adjoining areas. Bull. Geol. Soc.
Amer., 62, pp. 481-506, figs. 1-3, pis. 1-3.

Miall, L. C.

1878. Monograph of the sirenoid and crossopterygian ganoids. Palaeontogr.
Soc. London, 32, pp. 1-32, 4 pis.

Newberry, J. S.

1889. The Paleozoic fishes of North America. U. S. Geol. Surv., Monograph
16, pp. 1-340, figs. 1-3, pis. 1-53.

Noble, L. F.

1922. A section of the Paleozoic formations of the Grand Canyon at the
Bass trail. U. S. Geol. Surv., Prof. Paper, 131B, pp. 49-53.

Peyer, B.

1924. Ergebnisse der Forschungsreisen Prof. E. Stromers in den wiisten

Aegyptens. II. Wirbeltier-Reste der Baharije-Stufe (unterstes Cenoman).

6. Die Ceratodus-Funde. Abhandl. Bayer. Akad. Wiss., Math.-Nat. Abteil.,

30, Abhandl. 5, pp. 1-23, pis. 1, 2.
1945. Ueber Algen und Pilze in tierischen Hartsubstanzen. Arch. Julius

Klaus-Stiftung; Erganzungsbd. zu Bd. 20; Festschr. Prof. Alfred Ernst,

pp. 496-546, figs. 1-48.

Ross, C. P.

1934. Correlation and interpretation of Paleozoic stratigraphy in south-
central Idaho. Bull. Geol. Soc. Amer., 45, pp. 937-1000, 2 figs.

1937. Geology and ore deposits of the Bayhorse region, Custer County, Idaho.
Bull. U. S. Geol. Surv., 877, pp. i-vii, 1-161, figs. 1-14, pis. 1-18.

SCHUCHERT, C.

1918. On the Carboniferous of the Grand Canyon of Arizona. Amer. Jour.
Sci., (4), 45, pp. 347-370, figs. 1-5.

Spurr, J. E.

1898. Geology of the Aspen mining district, Colorado. U. S. Geol. Surv.,
Monograph 31, pp. i-xxxv, 1-250.

Stensio, E. A.

1931. Upper Devonian vertebrates from East Greenland. Collected by the

Danish Greenland expeditions in 1929 and 1930. Medd. Gr0nland, 86,

pp. 1-212, figs. 1-95, pis. 1-36.
1948. On the Placodermi of the Upper Devonian of East Greenland. II. Anti-

archi: subfamily Bothriolepinae. Medd. Gr0nland, 139, pp. 1-622, figs.

1-308, pis. 1-75.

Stoyanow, A. A.

1926. Notes on recent stratigraphic work in Arizona. Amer. Jour. Sci., (5),

12, pp. 311-324, 1 fig.
1936. Correlation of Arizona Paleozoic formations. Bull. Geol. Soc. Amer.,

47, pp. 459-540, figs. 1-5, 1 pi.
1942. Paleozoic paleogeography of Arizona. Op. cit., 53, pp. 1255-1282, figs.

1-3, 5 pis.

Stromer, E., and Peyer, B.

1917. Ueber rezente und triassische Gebisse von Ceratodontidae. Zeitsch.
Deut. Geol. Ges., 69, pp. 1-80, figs. 1-6, pis. 1-4.

DENISON: DEVONIAN FRESH-WATER FISHES 261

Traquair, R. H.

1877. On the structure of the lower jaw in Rhizodopsis and Rhizodus. Ann.
Mag. Nat. Hist., (4), 19, pp. 299-305.

Walcott, C. D.

1880. The Permian and other Paleozoic groups of the Kanab Valley, Arizona.
Amer. Jour. Sci., (3), 20, pp. 221-225.

WlLLARD, B.

1932. Devonian faunas in Pennsylvania. Penna. Geol. Surv., (4), G4, pp. 5-43.

Woodward, A. S.

1891. Catalogue of the fossil fishes in the British Museum (Natural History).
Part II, pp. i-xliv, 1-567, figs. 1-58, pis. 1-16.