Part 1

Produced by Chris Curnow, Joseph Cooper, Josephine Paolucci and the Online Distributed Proofreading Team at https://www.pgdp.net.

UNIVERSITY OF KANSAS PUBLICATIONS MUSEUM OF NATURAL HISTORY

Volume 12, No. 10, pp. 475-501, 7 figs. October 25, 1963

A New Genus of Pennsylvanian Fish (Crossopterygii, Coelacanthiformes) from Kansas

BY

JOAN ECHOLS

UNIVERSITY OF KANSAS LAWRENCE 1963

UNIVERSITY OF KANSAS PUBLICATIONS, MUSEUM OF NATURAL HISTORY

Editors: E. Raymond Hall, Chairman, Henry S. Fitch, Theodore H. Eaton, Jr.

~Volume 12, No. 10, pp. 475-501, 7 figs.~ ~Published October 25, 1963~

UNIVERSITY OF KANSAS Lawrence, Kansas

PRINTED BY JEAN M. NEIBARGER, STATE PRINTER TOPEKA, KANSAS 1963

[Transcriber's Note: Words surrounded by tildes, like ~this~ signifies words in bold. Words surrounded by underscores, like _this_, signifies words in italics.]

A New Genus of Pennsylvanian Fish (Crossopterygii, Coelacanthiformes) from Kansas

BY

JOAN ECHOLS

INTRODUCTION

In 1931 and 1932, H. H. Lane, C. W. Hibbard and W. K. McNown collected the specimens that Hibbard (1933) described and made the basis of two new species. These were from the Rock Lake shale member of the Stanton formation, six miles northwest of Garnett, Anderson County, Kansas. In 1954, from a locality (KAn-1/D, see page 480) approximately one fourth mile southwest of the first locality, specimens were quarried by F. E. Peabody, R. W. Wilson and R. Weeks. In 1955 R. R. Camp collected additional blocks of Rock Lake shale from this second locality. Study of all of the materials from the above mentioned localities reveals the existence of an hitherto unrecognized genus of coelacanth. It is named and described below.

I wish to thank Prof. Theodore H. Eaton, Jr., for suggesting the project and for much helpful advice. I am indebted to Dr. E. I. White of the British Museum (Natural History) for furnishing a cast of the endocranium of _Rhabdoderma elegans_ (Newberry) for comparison, and to Drs. Donald Baird (Princeton University), Bobb Schaeffer (American Museum of Natural History) and R. H. Denison (Chicago Natural History Museum) for loans and exchanges of specimens for comparison. I am grateful to Dr. Bobb Schaeffer for advice on the manuscript. Mr. Merton C. Bowman assisted with the illustrations. The study here reported on was made while I was a Research Assistant supported by National Science Foundation Grant G-14013.

SYSTEMATIC DESCRIPTIONS

Subclass CROSSOPTERYGII

Superorder COELACANTHI

Order Coelacanthiformes

Suborder DIPLOCERCIDOIDEI

Family DIPLOCERCIDAE

Subfamily ~Rhabdodermatinae~, new subfamily

_Type genus._--_Rhabdoderma Reis_, 1888, Paleontographica, vol. 35, p. 71.

_Referred genus._--_Synaptotylus_ new, described below.

_Horizon._--Carboniferous.

_Diagnosis._--Sphenethmoid region partly ossified, and consisting of basisphenoid, parasphenoid, and ethmoid ossifications; paired basipterygoid process and paired antotic process on basisphenoid; parasphenoid of normal size, and closely associated with, or fused to, basisphenoid; ethmoids paired in _Rhabdoderma_ (unknown in _Synaptotylus_).

_Discussion._--Because of the great differences in endocranial structure between the Devonian and Pennsylvanian coelacanths, they are here placed in new subfamilies. The two proposed subfamilies of the family Diplocercidae are the Diplocercinae and the Rhabdodermatinae. The Diplocercinae include those coelacanths having two large unpaired bones in the endocranium (at present this includes _Diplocercides_ Stensioe, _Nesides_ Stensioe and _Euporosteus_ Jaekel). The subfamily Rhabdodermatinae is composed of coelacanths having reduced endocranial ossification, as described in detail above, and now including _Rhabdoderma_ Reis and _Synaptotylus_ n. g.

Members of this subfamily differ from those of the subfamily Diplocercinae in having several paired and unpaired elements in the sphenethmoid region of the endocranium, instead of only one larger ossification. They differ from those of the suborder Coelacanthoidei in the retention of basipterygoid processes.

_Synaptotylus_ is more closely related to _Rhabdoderma_ than to the Diplocercines because the anterior portion of the endocranium contains only a basisphenoid, parasphenoid, and probably ethmoids. The sphenethmoid region was certainly not a large, unpaired unit as in the Diplocercines. Probably the posterior part, the otico-occipital region (not known in _Synaptotylus_), was much more nearly like that of _Rhabdoderma_, which consisted of unpaired basioccipital and supraoccipital, and paired prootics, exoccipitals, and anterior and posterior occipital ossifications (Moy-Thomas, 1937: figs. 3, 4). Moy-Thomas (1937:389) points out that in _Rhabdoderma_ the occipital region is "considerably more ossified" than in any coelacanths other than the Devonian forms. Berg (1940:390) thought that the Carboniferous coelacanths should be placed in a separate family because they did not have two large, unpaired bones in the endocranium. _Rhabdoderma_ and _Synaptotylus_ represent another stage in evolution of the endocranium in coelacanths, and, if classification is to be based on endocranial structure, then this stage (represented by the two genera) may later be given family rank as Berg suggested. Because _Rhabdoderma_ and _Synaptotylus_ have only part of the sphenethmoid region ossified and because they retain basipterygoid processes, they are considered to be related and are included in the subfamily Rhabdodermatinae.

~Synaptotylus~, new genus

_Type species._--_Synaptotylus newelli_ (Hibbard).

_Horizon._--Rock Lake shale member, Stanton formation, Lansing group, Missouri series, Upper Pennsylvanian.

_Diagnosis._--Late Pennsylvanian fishes of small size, having the following combination of characters: on basisphenoid, knoblike antotic processes connected by a low ridge to basipterygoid processes; entire ventral surface of parasphenoid toothed; anterior margin of parasphenoid notched and no evidence of hypophyseal opening. Dermal bones of skull smooth or with low, rounded tubercles and striae; fronto-ethmoid shield incompletely known but having one pair of large rectangular frontals with posterolaterally slanting anterior margins; intertemporals large, the lateral margins curving laterally; postorbital triangular, apex downward; subopercular somewhat triangular; squamosal carrying sensory canal that curves down posteriorly and extends onto a ventral projection; opercular generally triangular; supratemporals elongate, curving to fit lateral margin of intertemporals; circumorbital plates lightly ossified. Palatoquadrate complex consisting of endopterygoid and ectopterygoid (both toothed on medial surface), quadrate, and metapterygoid, the latter smooth and having widened border for articulation on anterodorsal margin. Pectoral girdle consisting of cleithrum and clavicle (supracleithrum not seen); small projection on medial surface of posterior portion of cleithrum; horizontal medial process on clavicle. Pelvic plate bearing three anteriorly diverging apophyses, and one denticulate ventromedian process for articulation to opposite plate. Lepidotrichia jointed distally, but not tuberculated. Scales oval, having posteriorly converging ridges on posterior exposed parts.

The name refers to the most distinctive character of the genus, the connected antotic and basipterygoid processes on the basisphenoid, and is derived from Greek, _synaptos_--joined, _tylos_ (masc.)--knob, projection.

_Synaptotylus_ is excluded from the advanced suborder Coelacanthoidei by the retention of basipterygoid processes on the basisphenoid. _Synaptotylus_ differs from _Rhabdoderma_ in several characters of the basisphenoid, the most important being: knoblike antotic processes (those of _Rhabdoderma_ are wider, more flattened and more dorsal in position); small, lateral basipterygoid processes (in _Rhabdoderma_ these are larger and farther ventral in position).

~Synaptotylus newelli~ (Hibbard)

_Coelacanthus newelli_ Hibbard, 1933, Univ. Kansas Sci. Bull., 21:280, pl. 27, figs. 2, 3.

_Coelacanthus arcuatus_ Hibbard, 1933, Univ. Kansas Sci. Bull., 21:282, pl. 26, fig. 8; pl. 27, fig. 1.

_Rhabdoderma elegans_ Moy-Thomas, 1937 (in part), Proc. Zool. Soc. London, 107(ser. B, pt. 3):399.

_Type._--K. U. no. 786F.

_Diagnosis._--Same as for the genus.

_Horizon._--Rock Lake shale member, Stanton formation, Lansing group, Missouri series, Upper Pennsylvanian.

_Localities._--The specimens studied by Hibbard (K. U. nos. 786F, 787F, 788) and no. 11457 were taken from the Bradford Chandler farm, from the original quarry in SW-1/4, SE-1/4, sec. 32, T.19S, R.19E. The remainder were collected from University of Kansas Museum of Natural History locality KAn-1/D, a quarry in sec. 5, T.19S, R.19E. Both of these are approximately six miles northwest of Garnett, Anderson County, Kansas.

_Referred specimens._--K. U. nos. 786F, 787F, 788, 9939, 11424, 11425, 11426, 11427, 11428, 11429, 11430, 11431, 11432, 11433, 11434, 11449, 11450, 11451, 11452, 11453, 11454, 11455, 11457.

_Preservation._--Preservation of many of the specimens is good, few are weathered, but most of the remains are fragmentary and dissociated. One specimen (the type, no. 786F) and half of another were nearly complete. Specimens are found scattered throughout the Rock Lake shale (see p. 498).

_Morphology._--Terminology used for bones of the skull is that of Moy-Thomas (1937) and Schaeffer (1952).

_Endocranium and parasphenoid_

The basisphenoid (see fig. 1) has been observed in only one specimen (K. U. no. 9939) in posterodorsal and ventral views. The basisphenoid, although somewhat crushed, appears to be fused to the parasphenoid. Both antotic and basipterygoid processes are present, and are connected by a low, rounded ridge. The antotic processes are large, bulbar projections. These processes in _Rhabdoderma_ are wider and more flattened (Moy-Thomas, 1937:figs. 3, 4). The antotic processes are at mid-point on the lateral surface, not dorsal as in _Rhabdoderma_, and both the processes and the ridge are directed anteroventrally. The basipterygoid processes are smaller, somewhat vertically elongated projections, situated at the end of the low connecting ridge extending anteroventrally from the antotic processes, and are not basal as are those of _Rhabdoderma_. The sphenoid condyles, seen in posterior view, issue from the dorsal margin of the notochordal socket. The margins of the socket are rounded, and slope down evenly to the center. A slight depression situated between and dorsal to the sphenoid condyles is supposedly for the attachment of the intercranial ligament (Schaeffer and Gregory, 1961:fig. 1). The alisphenoids extend upward, anterodorsally from the region above the sphenoid condyles, and may connect to ridges on the ventral surface of the frontals. The lateral laminae are not preserved, and their extent is unknown.

In viewing the changes in the endocranium of Carboniferous and Permian coelacanths, it would be well to consider the mechanical relationship of the loss of the basipterygoid processes to the effect on swallowing prey. Evidently many of the coelacanths, _Latimeria_ for example, are predators (Smith, 1939:104); to such fishes a more efficient catching and swallowing mechanism would be an adaptive improvement. Stensioe (1932:fig. 14) presents a cross section of the ethmosphenoid moiety of the endocranium of _Diplocercides kayseri_ (von Koenen) showing the metapterygoid of the palatoquadrate loosely articulated to both the antotic and basipterygoid processes. According to Tchernavin (1948:137) and Schaeffer and Rosen (1961:190) the swallowing of large prey depends on the ability of the fish to expand its oral cavity by allowing the posteroventral portion of the palatoquadrate and the posterior end of the mandible to swing outward. Where the palatoquadrate articulates with the basisphenoid at the antotic and basipterygoid processes, as in the Devonian coelacanths, it can not swing so far laterally as where it articulates with only the dorsal, antotic process. Perhaps the loss of the basipterygoid articulation reflects the development of a more efficient mechanism for swallowing prey in these fishes. Schaeffer and Rosen (1961:191, 193) show that in the evolution of the actinopterygians several changes improved the feeding mechanism: some of these changes are: (1) freeing of the maxilla from the cheek, giving a larger chamber for the action of the adductor mandibulae; (2) development of a coronoid process on the mandible; and (3) increase in torque around the jaw articulation. In coelacanths, at least some comparable changes occurred, such as: (1) loss of the maxillary, thus increasing the size of the adductor chamber; (2) development of the coronoid bone, affording a greater area for muscle attachment; (3) development of an arched dorsal margin on the angular; (4) modification of the palatoquadrate complex, with resultant loss of the basipterygoid processes. In _Synaptotylus_ the basipterygoid processes are small, not basally located, and perhaps not functional. A more efficient feeding mechanism developed rapidly during the Carboniferous and has remained almost unaltered.

The parasphenoid (see fig. 2) is a shovel-shaped bone having a wide anterior portion and a narrower posterior portion of nearly uniform width. Most of the ventral surface is covered with minute granular teeth. The anterior margin is flared and curved posteromedially from the lateral margin to a median triangular projection. The lateral margins curve smoothly from the greatest anterior width to the narrow central portion, where the margins become somewhat thickened and turned dorsally. Posterior to this the lateral margins are probably nearly straight. The external surface of the anterior section is nearly flat and has a central depressed area the sides of which slope evenly to the center. The internal surface is smooth and centrally convex. Because of the fragmentary nature of all four observed specimens, total length was not measured but is estimated to be 15 to 20 mm. The opening of the hypophyseal canal was not present, possibly because of crushing. Ethmoidal ossifications were not preserved in any of the specimens studied. The parasphenoid differs from that of _Rhabdoderma elegans_ (Newberry) in being more flared and widened anteriorly and more concave centrally.

_Dermal bones of the skull_

Various portions of the cranial roof are preserved in several specimens (see fig. 3). For comparisons with _Rhabdoderma elegans_, see Moy-Thomas (1937:fig. 1).

The premaxillaries and rostral elements are not preserved in any of the specimens. Only one pair of relatively large frontals have been observed; they are 5.5 to 9.0 mm. long and 2.0 to 3.5 mm. wide. These are nearly flat bones, with the greatest width posteriorly 0.1 to 1.0 mm. wider than the anterior portion. The midline suture is straight, the lateral margins are nearly straight, the anterior margin slopes evenly posterolaterally, and the posterior margin is slightly convex to straight. The anterior margin in _R. elegans_ is essentially straight. Ornamentation consists of sparse, unevenly spaced, coarse tubercles or short striae. In one specimen both bones have small clusters of tubercles near the lateral margins and about 2.0 mm. from the posterior margin. None of these bones has alisphenoids or ridges on the ventral surface, as Stensioe (1921:65, 97) described for _Wimania_ and _Axelia_.

Only six supraorbitals have been preserved (see fig. 3). These are nearly square, flat, thin bones lying nearly in place adjacent to a frontal on K. U. no. 788. The smallest is anterior; the margins of all are nearly straight. The bones are unornamented. Each bears a pore of the supraorbital line just below the midline. The supraorbitals of _R. elegans_ have a triangular outline and do not bear pores.

Intertemporals (fig. 3) on several specimens vary from approximately 9.0 to 15.0 mm. in length, 2.0 to 2.7 mm. in anterior width, and increase to 4.5 to 8.0 mm. in maximum posterior width. The midline suture is straight, the anterior margin is concave and the lateral margin proceeds laterally in a concave curve to the widest portion. In _R. elegans_ only the anterior half of the corresponding margin is concave. The posterior margin is slightly rounded and slopes anteriorly toward the lateral margin. Ornamentation is usually of randomly oriented tubercles and striae, although striae are more common in the posterior third and may be longitudinal, whereas tubercles occur mainly on the anterior section. No evidence of sensory pores, as seen on the intertemporal of _R. elegans_, has been found.

The supratemporals were observed on only one specimen (K. U. no. 788), (fig. 3). Sutures were difficult to distinguish but the medial margin is presumed to curve to fit and to articulate with the lateral margins of the intertemporals. Lateral margins are smoothly curved but the anterior and posterior margins were broken off. There appears to be no ornamentation on this bone. The supratemporals are much more elongated and curving than those in _R. elegans_.

The cheek region is nearly complete in one specimen (K. U. no. 788), and scattered parts occur in a few others (see fig. 3). The lacrimojugal of no. 788 is elongate, with both ends curving dorsally. It differs from the lacrimojugal in _R. elegans_, in which the anterior end extends anteriorly and is not curved dorsally. The posterior and anterior margins are not preserved; the greatest height appears to be posterior. Pores of the suborbital portion of the infraorbital sensory canal are seen on the dorsal surface of the bone. In _R. elegans_ the pores are on the lateral surface. A section of the lacrimojugal on specimen no. 11425, broken at both ends, shows a thin layer of bone perforated by the pores and covering a groove for the canal within the dorsal margin of the bone. Both specimens are unornamented.

A nearly complete postorbital (fig. 3) on specimen no. 788 is nearly triangular, with the apex ventral. The concave anterior margin bears pores of the postorbital part of the infraorbital line. Ornamentation consists of widely spaced, coarse tubercles.

Part of one squamosal is preserved. It is somewhat triangular and its apex is ventral. This bone is associated with the postorbital, subopercular and lacrimojugal on no. 788. The preopercular sensory line passes down the curving ventral margin of this bone, and extends ventrally onto a narrow projection. A low ridge, nearly vertical, passes dorsally from about mid-point of the canal to the dorsal portion. The anterior margin is nearly straight, the ventral margin is concave, and the dorsal margin is convex dorsally but may be incomplete. Perhaps the squamosal and preopercular are fused. The surface appears smooth; the view may be of the medial side. The squamosal of _R. elegans_ is nearly triangular and notably different from that of _Synaptotylus newelli_.

The subopercular (fig. 3) shows closely spaced tubercles on the lateral surface. The bone is an elongated, irregular triangle with the apex pointing anterodorsally. The margins are incomplete, except for the concave, curving anterior margin.

Numerous operculars (fig. 3) occur in the suite of specimens, both isolated and nearly in place. Each is subtriangular; the apex of the triangle is ventral. A slight convexity projects from the anterodorsal border. The posterior margin is broadly but shallowly indented. Otherwise the margins are smooth. Maximum height ranges from 8.0 to 11.0 mm., and maximum width from 8.0 to 13.0 mm. Ornamentation varies from a few widely spaced, randomly oriented tubercles to closely spaced tubercles merging posteriorly into striae. On some specimens these are parallel to the dorsal border, and oblique in the central portion. On the posterior margins of several operculars the striae break up into tubercles. A few operculars have closely spaced tubercles over much of the surface. The internal surface is smooth.

_Visceral skeleton_

The palatoquadrate complex, best seen on K. U. no. 9939 (fig. 4), consists of endopterygoid, ectopterygoid, metapterygoid and quadrate. No trace of epipterygoids, dermopalatines or autopalatines, such as Moy-Thomas (1937:392, fig. 5) described for _Rhabdoderma_, has been observed.

The endopterygoid has a long, ventral, anteriorly-directed process, and an anterodorsal process that meets the metapterygoid in forming the processus ascendens. The suture between the endopterygoid and metapterygoid, seen in lateral view, is distinct in some specimens and has an associated ridge; these bones appear to be fused in others, without regard to size. This suture curves dorsally from a point anterior to the quadrate and passes anterodorsally to the extremity of the processus ascendens. The suture is visible on the medial side only near the processus ascendens, for it is covered by a dorsal, toothed extension of the endopterygoid. The endopterygoid has a smooth lateral surface; the medial surface is covered with tiny granular teeth, in characteristic "line and dot" arrangement. The teeth extend onto the ventral surface of the ventral process.

Two long, narrow, splintlike bones covered on one surface with granular teeth are interpreted as ectopterygoids. These are 13.0 and 16.0 mm. long and each is 1.5 mm. wide. Orientation of these is unknown, but they probably fitted against the ventral surface of the ventral process of the endopterygoid (Moy-Thomas, 1937:fig. 5).

The metapterygoid has a smooth surface in both views. The dorsal edge has a thickened, flared margin that presumably articulated with the antotic process of the basisphenoid. No articular surface for the basipterygoid process has been observed.

The quadrate is distinct and closely applied to the posteroventral margin of the complex. In medial view the margin is nearly straight and continues to the ventral edge. The ventral surface is thickened and forms a rounded, knoblike articular surface. In lateral view the surface is smooth; the anterior margin is irregular (or perhaps broken on all specimens), and proceeds in an irregular convex curve from the posterior to the ventral margin.

The general shape of the palatoquadrate complex is most nearly like that of _Rhabdoderma elegans_ (Moy-Thomas, 1937:fig. 5). The orientation of the complex in the living fish was probably oblique, with the processus ascendens nearly vertical, the quadrate oblique, and the ventral process of the endopterygoid extending dorsoanteriorly and articulating with the parasphenoid.