FIELDIANA • GEOLOGY

Published by
CHICAGO NATURAL HISTORY MUSEUM

Volume 10 Sbftember 22, 1968 No. 88

NEW SALAMANDERS OF THE FAMILY

SIRENIDAE

FROM THE CRETACEOUS OF NORTH AMERICA

Coleman J. Coin

AND

Walter Auffenberg
Dbpabtmbnt op Biology, Univcrsity op Florida

Since we published our first report on fossil salamanders of the
family Sirenidae (Goin and Auffenberg, 1955) material has continued
to accumulate. In addition to the species described in that pap>er, we
have since described Siren dunni (Goin and Auffenberg, 1957) from
the Eocene of Wyoming. These records were all from the Cenozoic.
Thanks to the kindness of many friends who have let us work through
unsorted microfossil material, we now have located two genera of
Sirenidae from Mesozoic deposits.

The first of these genera was taken from the Trinity beds of the
Lower Cretaceous of Texas.

Prosiren, new genus

Type species. — Prosiren elinorae, new species.

A genus of amphicoelous salamanders referred tentatively to the
Sirenidae and characterized by moderately developed aliform proc-
esses; well-developed zygapophyseal ridges extending between the an-
terior and posterior zygapophyses; transverse processes, each formed
of two plate-like structures, of which the upper fuses with the zygapo-
physeal ridge just posterior to the anterior zygapophysis; anterior
glenoid cavity nearly round, just slightly higher than wide; a well-
developed subcentral median ridge that has a nearly straight margin.

Diagnosis. — This genus differs from the genera Siren and Pseudo-
hranchus in that it has a well-developed basapophyseal accessory

Library of Congress Calaloii Card Number: S8-1S1 79
No. 868 449

450 FIELDIANA: GEOLOGY, VOLUME 10

process that originates on the lower plate of the transverse process
slightly anterior to the subcentral foramen and extends downward and
anteriorly to meet the anterior glenoid cavity near its lower margin.
From the genus described below (p. 453), which also has basapophys-
eal processes, it differs in that the neural spine continues posteriorly
between the aliform processes and in that these aliform processes are
lower and thicker. Also, the dorsal margin of the aliform process
slopes downward and backward from its point of contact with the
neural spine rather than continuing posteriorly at about the same
level for some distance, then turning down abruptly.

Prosiren elinorae, new species. Figure 187.

Type. — Chicago Natural History Museum PR 391, a dorsal
vertebra.

Horizon and locality. — Turtle Gully, southwest portion of Green-
wood Canyon Gully system, Lower Cretaceous, 2}/^ miles southwest
of Forestburg, Montague County, Texas.

Referred material. — Two dorsal vertebrae, CNHM-PR 390, the
same locality as the type, and PR 392, from Triconodont Gully,
northeast portion of Greenwood Canyon Gully System, Lower Creta-
ceous, 23/^ miles southwest of Forestburg, Montague County, Texas.

Diagnosis. — A sirenid with the neural spine extending posteriorly
between the aliform processes and with well-developed basapophyseal
accessory processes on the antero-ventral portion of the centrum.

It can be distinguished from all other Sirenidae, including the
genus described below (p. 453), by the combination of well-developed
basapophyseal accessory processes and a neural spine that extends
between the arms of the aliform processes. It seems to differ from
other Sirenidae also in the manner in which the dorsal margin of the
aliform process slopes downward from the point at which it connects
with the neural spine, so that the process appears somewhat triangu-
lar as seen from the side. In the other sirenids these processes con-
tinue posteriorly at a nearly uniform height until well over the
postzygapophyses, at which point the dorsal margin turns downward
at nearly a 90 degree angle in Siren and the genus described below
and at least a 60 degree angle in Pseudobranchus. Thus, in these
genera the aliform process as seen in profile appears to be more nearly
rectangular than triangular.

Description of type. — Measurements (in mm.; measurements in
parentheses are thousandths of length of centrum) : Length of centrum
along midventral line, 2.1. Width of vertebra at narrowest point of
zygapophyseal ridges, 1.1 (0.524); height of vertebra from lower

COIN AND AUFFENBERG: NEW SALAMANDERS

461

mm

Fig. 187. Pro«ren ehnora«, type (CNHM-PR 391). A, Lateral view. B, Dor-
sal view. C, Anterior view. D, Ventral view.

margin of centrum to a line drawn between facets of postzygapo-
physes, 1.8 (0.857). Distance from tips of prezygapophyses to tips of
postzygapophyses, 2.8 (1.333). Angle between aliform processes, 55°.
Width of anterior glenoid cavity, 0.9 (0.429); height of anterior
glenoid cavity, 1.0 (0.476). Width of neural canal, 0.8 (0.381);
height of neural canal, 0.7 (0.333).

Centrum longer than high; glenoid cavity mildly ovate, slightly
higher than wide. Centrum provided with an elevated, ridge-like,
median ventral keel, on either side of which is found a moderate
subcentral foramen. Margin of ventral keel slightly concave.

Total length of neural arch greater than length of centrum and its
width at the narrowest portion of the zygapophyseal ridges slightly
greater than width of centrum. Neural canal an inverted crescent
anteriorly; about rounded posteriorly.

452 FIELDIANA: GEOLOGY, VOLUME 10

Articulating surface of prezygapophysis an elongate oval in shape,
much longer than wide, directed more anteriorly than laterally. Ar-
ticulating surface of postzygapophysis ovate. Zygapophyseal ridges
well developed, slightly concave as seen from- above, particularly
posteriorly. As seen from the side the zygapophyseal ridge forms a
very shallow V with the apex at the point where the dorsal portion
of the transverse process meets the zygapophyseal ridge. It then
continues posteriorly and somewhat dorsally to meet the posterior
zygapophysis at the base of the aliform process.

Aliform processes but moderately developed, vertical in position;
they are thicker and lower than in the genus Siren. As seen from
above they form an anteriorly pointing V. A floor with a broadly
crenate posterior margin is present between the aliform processes.
The neural spine continues between the aliform processes for a dis-
tance to form a median ridge on the floor between them.

Neural spine well developed, higher than aliform processes. Its
anterior margin broken so that its height anteriorly cannot be de-
termined.

Transverse processes broken off; only their bases present. As
in Siren, each is apparently composed of two plate-like portions of
which the ventral is larger than the dorsal. The ventral portion a
wing-like structure extending from close to the anterior margin of the
side of the centrum posteriorly nearly the entire length of the centrum.
The dorsal portion a flat plate extending from the zygapophyseal
ridge, somewhat behind the posterior margin of the prezygapophysis,
ventrally and posteriorly to meet the ventral portion to which it is
fused. A well-developed basapophyseal accessory process originates
on each side on the ventral surface of the lower portion of the trans-
verse process. This process originates about one-third of the way
back and extends downward and forward nearly to the point where
the subcentral keel meets the margin of the glenoid cavity. Laterally
a foramen is present in the angle between the dorsal and ventral por-
tions of the transverse process and another lies somewhat ventral
and posterior to the angle between the dorsal portion of the trans-
verse process and the zygapophyseal ridge.

Variation. — The two referred specimens are much more frag-
mented than is the type, but both seem to belong to this species.
In characters that can be determined in them they agree with the
type: in the slope of the posterior margin of the aliform processes; in
the size and formation of the zygapophyses; and in the formation of
the transverse processes. One of the specimens (CNHM-PR 392)

COIN AND AUFFENBERG: NEW SALAMANDERS 453

further agrees with the type in the formation of the basapophyseal
processes, although they are better developed and extend below the
lip of the glenoid cavity; in the form of the glenoid cavities of the
centrum; and, as far as can be determined in this uncleaned speci-
men, in the shape of the neural canal. These three characters cannot
be determined in the other specimen (CNHM-PR 390), but it agrees
with the type in the formation of the neural spine and in the slope of
the wings of the transverse process.

Both of these specimens have been excessively fragmented. In
PR 392 the matrix itself seems to be holding the vertebra together;
in PR 390 the matrix has separated and the specimen has fallen into
discrete fragments.

Associated fauna. — In addition to these salamanders, the Forest-
burg stratum contains fragments representing fishes, other nonde-
script salamanders (Salamandridae?), and at least two families of
frogs; there are also lizard jaws, and turtle and crocodile fragments,
along with the remains of dinosaurs and of a primitive mammalian
fauna. The mammals have been summarized by Patterson (1956).

The important Carnegie Museum collection of Upper Cretaceous
vertebrate material, taken by J. B. Hatcher in 1900 from the Lance
Formation in Niobrara County, Wyoming, has been examined for
salamander remains. Included among the salamander fossils are two
specimens that represent a large sirenid. These specimens differ from
both the genus Prosiren described herein from the Lower Cretaceous
and the Cenozoic genera Siren and Pseudobranchtis. We therefore
erect for them a new genus.

Adelphesiren, new genus

Type species. — Adelphesiren olives, new species.

A genus of amphicoelous salamanders referred to the Sirenidae
and characterized by well-developed aliform processes; well-devel-
oped zygapophyseal ridges extending between the anterior and pos-
terior zygapophyses; transverse processes formed of two plate-like
structures, the upper of which fuses with the zygapophyseal ridge
just posterior to the anterior zygapophysis; anterior glenoid cavity
nearly round, just slightly wider than high; a well-developed sub-
central median ridge that has a nearly straight margin.

Diagnosis. — This genus differs from the genera Siren and Psevdo-
branchus in that it has well-developed basapophyseal accessory proc-
esses that originate on each side of the centrum just anterior to the
subcentral foramen and extend anteriorly and slightly ventrally to

454

FIELDIANA: GEOLOGY, VOLUME 10

Fig. 188. Adelphesiren olivae, type (CM 6467). A, Lateral view. B, Dorsal
view. C, Anterior view. D, Ventral view.

meet the glenoid cavity near its lower margin. PYom Prosiren, which
has similar basapophyseal processes, it differs in that the aliform
processes are higher and thinner, as in the genus Siren, and in that
the neural spine terminates at the point of origin of the aliform proc-
esses rather than extending between them as it does in Prosiren.

Adelphesiren olivae, new species. Figures 188, 189.

Type. — Carnegie Museum no. 6467, a dorsal vertebra.

Horizon and locality. — Lance Formation, Upper Cretaceous, Nio-
brara County, Wyoming.

Referred material. — A single dorsal vertebra (CM 6448) with the
same data as the type.

Diagnosis. — A large sirenid with well-developed basapophyseal
accessory processes, with high, well-developed aliform processes that
do not have the neural spine extending between them, and with a
nearly round glenoid cavity.

COIN AND AUFFENBERG: NEW SALAMANDERS 466

It differs from the genus and species described above in its larger
size, in that there is no evidence of neural spine between the aliform
processes, and in that the dorsal margins of the aliform processes do
not slant downward posteriorly but rather are nearly horizontal.
From all the species of Siren and Pseudohranchus it differs in that it
has well-developed basapophyseal accessory processes on the antero-
ventral surface of the centrum and from Pseudohranchus and all
known species of Siren except lacertina in its larger size.

Description of type. — Measurements (in mm.): Length of centrum
along midventral line, 8.0. Height of vertebra from lower margin of
centrum to a line drawn between facets of postzygapophyses, 4.5
(0.563). Distance from tips of prezygapophyses to tips of postzyga-
pophyses, 10.8 (1.350). Angle between aliform processes, 36". Width
of anterior glenoid cavity, 3.0 (0.375); height of anterior glenoid
cavity, 2.8 (0.350). Width of neural canal, 3.1 (0.388); height of
neural canal, 1.6 (0.200).

Centrum longer than high; glenoid cavity nearly round, but
slightly wider than high. Centrum provided with a well-developed,
elevated, ridge-like median ventral keel, on either side of which is
found a relatively large, subcentral foramen. Margin of ventral keel
nearly straight.

Total length of neural arch greater than length of centrum and
its width at the narrowest portion of the zygapophyseal ridges greater
than width of centrum. Neural canal an inverted crescent anteri-
orly; broadly oblong posteriorly; provided with a weakly developed,
very low median epapophyseal ridge on the floor.

Articulating surfaces of prezygapophyses broadly oval in shaj)e,
only slightly longer than wide, and directed but slightly more later-
ally than anteriorly. Articulating surfaces of postzygapophyses
ovate. Zygapophyseal ridges well developed, markedly concave as
seen from above. As seen from the side the zygapophyseal ridge
forms a very shallow "step" at the point where the dorsal portion
of the transverse process meets the zygapophyseal ridge, with the
posterior portion about a millimeter higher than the anterior portion.

Aliform processes well developed, vertical in position, their dorsal
margins eroded so that neither their height nor their original profile
can be determined. As seen from above they form an anteriorly
pointing V. Floor between aliform processes present, with a broadly
rounded, apparently eroded, posterior margin.

Neural spine well developed ; dorsal margin eroded but still higher
than the aliform processes.

456 FIELDIANA: GEOLOGY, VOLUME 10

Transverse processes obviously well developed and composed of
two plate-like portions; the processes are broken near their bases.
The ventral portion, a wing-like structure, extends from close to the
anterior margin of the side of the centrum for about two-thirds of
the length of the centrum, then continues posteriorly to the end of
the centrum as a very low ridge. The dorsal portion extends from
the zygapophyseal ridge somewhat behind the posterior margin of the
prezygapophysis ventrally and posteriorly to the posterior margin of
the ventral portion, to which it is fused. Its lateral extent cannot be
determined because of fracture. What remains of the posterior mar-
gin of the transverse process is approximately perpendicular to the
axis of the centrum.

A well-developed basapophyseal accessory process originates on
each side of the centrum just anterior to the subcentral foramen and
extends anteriorly and slightly ventrally to the margin of the ante-
rior glenoid cavity. It becomes both wider and higher as it pro-
gresses anteriorly.

Laterally a foramen is present in the angle between the dorsal and
ventral portions of the transverse process.

Variation. — A single vertebra (CM 6448), with the same data as
the type, is somewhat larger (centrum length 9.5 mm.) but in the
characters that can be ascertained it is essentially a duplicate of the
type. Only one zygapophysis, the right posterior, remains, but it
indicates that the shape of the articulating surfaces and the flare of
the zygapophyses are essentially the same in the two specimens.
Likewise, the structure of the aliform processes and the floor between
them, the zygapophyseal ridges, the subcentral keel, the shape of the
neural canal, and the form of the basapophyseal processes indicate a
striking degree of similarity between the two specimens. In fact,
only in that the posterior glenoid cavity seems to be slightly higher
than wide does it appear to differ essentially from the type and even
here the margin of the cavity is broken in the larger specimen so that
its exact size and shape cannot be determined.

Associated fauna. — A genus of salamanders allied to the Salaman-
dridae and a single vertebra of a snake were in the same lot that con-
tained the specimens of Adelphesiren. The mammals of this collection
have been reported by Simpson (1929).

DISCUSSION

The discovery of these new sirenids from the Upper Mesozoic
greatly broadens our knowledge of the fossil history of the family

Fig. 189. Adelphesiren olivae (CM 6448). A, Lateral view. B, Dorsal view.

467

458 FIELDIANA: GEOLOGY, VOLUME 10

but, it must be admitted, does not entirely clarify our information
concerning the lineage of these salamanders.

The earliest known sirenid, Prosiren, which is here described, can
at this time be but doubtfully referred to the Sirenidae. We include
it here because it has rudimentary aliform processes and because the
transverse processes seem to be of the same basic type as in the other
sirenids, but it differs from all of them in that the aliform processes
are low and heavy and slope down strikingly toward the margin of
the postzygapophyses, whereas in all of the other sirenids the aliform
processes are high and thin, like the neural spine. Also, the whole
neural arch seems to stand higher and is more narrow than is typical
of the sirenids. These characters seem to us to permit two possible
interpretations. Either this is a very primitive sirenid, among the
earliest of the line, and hence may be expected to be somewhat differ-
ent from the later forms, or — a more likely interpretation, it seems
to us — it is off the main evolutionary line and, while basically a sire-
nid, it probably does not represent at all the true ancestor of the
modern sirenids.

Adelphesiren, on the other hand, seems to be quite typically sire-
nid. The high, thin, rectangular aliform processes are very Siren-like
and, in fact, if it were not for the presence of well-developed basapo-
physeal accessory processes on the antero-ventral surface of the cen-
trum, we would be inclined to include it in the genus Siren. The
presence of these well-developed processes, however, indicates to us
that it should be set off in a separate genus.

The early Eocene form, Siren dunni, is a typical Siren in every
sense of the word, whereas the Miocene form. Siren hesterna, seems
now not to fit in the general line of evolution, but to be somewhat of
a side branch. With its short, stubby centrum and wide-flaring ali-
form processes, we just cannot conceive that it would be in between
Siren dunni on the one hand and Siren simpsoni, or either of the
recent species, on the other.

It is rather difficult to determine exactly what size means and
how it should be interpreted in salamanders that have indeterminate
growth, but it perhaps is worth mentioning that Siren lacertina has
been, until this time, the only large sirenid, and it apparently came
into existence in the late Pleistocene. Now, the presence of Adelphe-
siren from the Upper Cretaceous gives us two large forms. We do
not mean to imply, of course, that they are necessarily closely related,
but it does seem strange that all of the others, from the Lower Cre-
taceous, Eocene, Miocene and Pliocene, and the Recent species, inter-
media, are small. This, as mentioned above, may not be significant,

COIN AND AUFFENBERG: NEW SALAMANDERS 459

but it may indicate that Adelphesiren itself is not on the line of sirenid
evolution leading to the present species.

Also of some possible significance is the fact that nowhere below
the Pliocene do we find two forms together. In the Pliocene, Siren
simpsoni and Pseudohranchus vetustus occur, but from the Miocene,
Eocene, Upper Cretaceous and Lower Cretaceous only single forms
are known. This does not necessarily indicate that only single species
were in existence at those times. It rather seems to us to indicate
the fragmentary state of our knowledge and the inadequacy of our
collections. If these were the only forms in existence at these times,
it would be expected that one would lead logically to the other and
that the sequence would be a smooth one from Lower Cretaceous to
Recent. But at least two forms, Prosiren of Lower Cretaceous and
Siren hesterna of the Miocene, seem to be definite sidelines, and the
same may well be true of Adelphesiren from the Upper Cretaceous.
It seems much more likely that we have just scratched the surface
when it comes to collecting, and that many more data must accumu-
late before we can begin to work out the true evolutionary history of
the family Sirenidae.

ACKNOWLEDGMENTS

We are indebted to Dr. M. Graham Netting and Dr. J. LeRoy
Kay for making travel to F*ittsburgh possible and for the privilege
of sorting through Carnegie Museum's microvertebrate material.
Dr. Rainer Zangerl has us in his debt for his kindness in sending us
representative material from the Chicago Natural History Museum's
microvertebrates from the Trinity beds of Texas so that we might
sort through it here in Gainesville. Mrs. Olive B. Goin has been of
material assistance in the preparation of this manuscript, typing it
through several of its stages. To all of these we express our thanks.

REFERENCES

GoiN, Coleman Jett, and Auffenberg, Walter

1955. The fossil salamanders of the family Sirenidae. Bull. Mus. Comp. Zool.,
113, (7), pp. 497-614, figs. 1-3.

1957. A new fossil salamander of the genus Siren from the Eocene of Wyoming.
Copeia, 1957, (2), pp. 83-85, fig. 1.

Patterson, Bryan

1956. Early Cretaceous mammals and the evolution of mammalian molar teeth.
Fieldiana, Geol., 13, (1), pp. 1-105, figs. 1-17.

Simpson, George Gaylord

1929. Some Cretaceous mammals from the Lance Formation. Ann. Carnegie
Mus., 29, (2), pp. 107-118, figs. 1-6.