vp15553w.jpg (124726 bytes)




Copyright © 2000-2009 by Mike Everhart

Last updated 03/26/2009




LEFT: The anterior portion of the skull of Martinichthys brevis (FHSM VP-15553) in right and left lateral views in the collection of the Sternberg Museum, found by Pam Everhart in southeastern Gove County in 1990.

Martinichthys is an obscure genus of plethodid fish whose remains have been found only in the Smoky Hill Member, Niobrara Chalk, of Western Kansas. The genus was originally described by C. E. McClung (1926) and was named by him in honor of H. T. Martin who had collected one of the most complete specimens in 1909. Earlier workers, including Cope (1877) and Hay (1903), had assigned the fragmentary and enigmatic remains to the genus Protosphyraena. While several of the specimens of Martinichthys collected by McClung and others do include locality data, there has been no accurate information published on the stratigraphic interval in which these rare fossils were found. Field work by the author between 1988 and 1992, and examination of more recent and better documented specimens in the Kansas University and Fort Hays University collections have shown that Martinichthys occurs only in the lower one third of the Smoky Hill Chalk, and more precisely, in the stratigraphic interval from about 3 meters below Hattin's (1982) Marker Unit 4 to one meter above Marker Unit 5. The 7 meter interval corresponds to the upper portion of the biostratigraphic Zone of Protosphyraena perniciosa proposed by Stewart (1990) and is late Coniacian in age (roughly 86 mya).


he Smoky Hill Member of the Niobrara Chalk of Western Kansas was deposited as marine sediments in the Western Interior Sea that covered the central United States during the late Cretaceous period, 87 to 82 million years ago. The formation lies conformably between the Fort Hays Member of the Niobrara Chalk and the Sharon Springs Member of the Pierre Shale, and is about 182 meters (600 feet) in thickness. It has been a rich haven for hunters of Cretaceous marine fossils for more than a hundred years and has been the source of many significant discoveries. Unfortunately, many if not most of the fossils taken from the chalk were not associated with good stratigraphic or locality data, and their relative scientific value is therefore somewhat diminished.

Early workers attempted to divide the formation on the basis of the fossils which had been found in it. Logan (1897) referred to the entire Smoky Hill Member as the "Pteranodon" Beds. Williston (1897) further sub-divided the Pteranodon Beds into the upper "Hesperornis Beds" and the lower "Rudistes Beds". Williston (1897) also commented that "I need not call the attention of future collectors to the importance of locating the horizon of specimens more accurately than has been done before." Despite that admonition, few collectors in the chalk since that time have been able to document the stratigraphic location of their finds.

More recently, several authors have attempted to describe the stratigraphic sequences within the Smoky Hill Chalk. Hattin (1982) divided the Smoky Hill Chalk member of the Niobrara Chalk Formation (Upper Cretaceous) of Western Kansas on the basis of 23 specific groupings of bentonites and distinctive layers of chalk which he referred to as Marker Units. He also expanded earlier work on the biostratigraphy of the chalk by correlating invertebrate remains that are associated with the intervals between various marker units. Others, such as Bardack (1965) and Stewart (1978, 1988) have added observations on the biostratigraphy, noting where certain species of vertebrates and invertebrates were or were not found. A definitive representation of the biostratigraphy of the Smoky Hill Chalk has yet to be completed. The data provided here by the authors is only a small addition to information that is available.

Martinichthys was first described by C. E. McClung in 1926 as a new genus of fish from the Cretaceous of Kansas. Prior to the publication of his paper, the fossil remains had been included in the genus Erisichthe by E. D. Cope (1877) and Protosphyraena by O. P. Hay (1903). Locality data for this specimen and subsequent finds is somewhat vague and accurate stratigraphic information is not available. Early specimens noted to have been found in the Niobrara chalk of Gove or Trego counties do not have the information necessary to assign them to specific biostratigraphic zones. Generalized stratigraphic data can be proposed if a locality was recorded but many specimens lack sufficient site information.

Early authors tended to be more concerned with the identification of the unusual fossil remains than they were with the determination of the stratigraphic location or association with other species. These workers, including Woodward, Hay, and Loomis (McClung, 1926) had some difficulty in assigning remains of this genus to a specific family. This was due, in large part, to the lack of definitive diagnostic material. Even now, the genus is only known by relatively rare, but robust and well preserved rostra. The type specimen (AMNH 2131) of Erisichthe ziphioides described by Cope in 1877 consists of a single, 10.5 cm long rostrum. Two other specimens, (KU497 and KU498), described by McClung (1926), consist of crushed but well preserved skulls and some vertebra. Another skull (FHSM VP 3248) is more or less complete, but is disarticulated and may actually represent another genus of Plethodid. Only two specimens in the Everhart collection have portions of the skull attached to the rostrum. The remaining 60 or more known specimens consist of pieces of individual rostra.

In the genus Martinichthys, the rostrum, and a number of the bones in the mouth appear to be covered with denticles. This distinctive feature places the genus to the Order Osteoglossiforme (bony tongue). McClung (1926) noted that the rostra are generally worn on the cephalic end and Stewart (personal communication) noted that the wear pattern differs somewhat between individuals, with rostra being worn or polished at different angles. The denticle covered parasphenoid, palatines, and glossohyal also show some affinity to similar structures in plethodids such as Anogmius and Bananogmius.

The rostra of Martinichthys do have some outward similarities to the swordfish-like rostra of the genus Protosphyraena, with the notable exception that all of the rostra of Martinichthys are rounded and worn. Cope (1877) concluded erroneously that the specimen of Erisichthe ziphioides (AMNH 2131) was "the muzzle of an old individual, which has lost a good deal of it's apex by attrition", inferring that the pointed tip of the rostrum and the distinctive teeth of the genus had been worn away long before the fish had died. Hay (1903) did not agree with Cope's explanation and stated that he believed instead that the specimen belonged to a distinct species of Protosphyraena having a short and blunt snout.

After study of the more complete specimens (McClung, 1926), it was apparent that there were other notable differences between the two genera. The skulls of Martinichthys were much lighter and more delicate than those of Protosphyraena and do not have the robust, blade like teeth. In addition, they do have well ossified vertebrae which are typical of the plethodids and which have not yet been found in association with the remains of Protosphyraena. Further, the caudal and pectoral fins of Martinichthys are unknown at this time, while the pectoral and caudal fins of Protosphyraena are well documented.

martin3a.jpg (3997 bytes)

A variety of Martinichthys rostra collected from the Smoky Hill Chalk

LEFT AND RIGHT: Dorsal and ventral views of 6 specimens of Martinichthys rostra: FHSM VP-15555 (EPC 1990-23), FHSM VP-15551 (EPC 1989-29), FHSM VP-15554 (EPC1990-25), FHSM VP 15563 (EPC 1995-40), FHSM VP-15559 (EPC 1992-10, mis-labeled as 1993-10) and FHSM VP-15560 (EPC 1992-35).

martin4a.jpg (4158 bytes)

Hay (1903) recognized the differences and included Martinichthys in the Plethodidae. McClung (1926) stated that Martinichthys and several other related genera belonged to no well determined family. Romer (1966) included Martinichthys in the family Plethodontidae but indicated that its exact status was still a matter of conjecture. Russell (1988) includes the seven species described by McClung (1926) under the Plethodidae along with Bananogmius and Anogmius while the species of Protosphyraena are separated into their own distinct family, the Protosphyraenidae. Stewart (1993, personal communication) suggested that the rostra attributed to Martinichthys may actually belong to several species of Bananogmius or Anogmius since the skulls of these species are found without the rostra (ethmoid process) attached. Data to be presented in this paper appears to indicate, however, that while Bananogmius and Anogmius occur over a wide span of time in the Niobrara formation, Martinichthys remains are restricted to a relatively small interval. More recently, Taverne (1999) has reduced the number of species of Martinichthys to two, M. ziphioides and M. brevis.


Hattin (1982) published the most complete description to date of the stratigraphy and invertebrate biostratigraphy of the Smoky Hill Chalk, dividing the chalk on the basis of 23 distinct marker units. Stewart (1990) incorporated these marker units into his interpretation of the vertebrate biostratigraphy of the Smoky Hill Chalk. Considered together, these two systems represent the beginnings of a reliable methodology for determining the actual horizon for fossils collected in the chalk. Problems are encountered when Hattin's (1982) Marker Units are not readily visible or identifiable in some exposures. Stewart's (1990) biostratigraphy relies on the collection and accurate identification of vertebrate and invertebrate fossils, some of which are not easily found in all localities.

More recent work on the biostratigraphy of the chalk by Stewart (1988, 1990) divided the unit into faunal zones based on the remains of both vertebrates and invertebrates. These studies were based on the stratigraphic work done by Hattin (1982) and others, which rely on interbedded seams of altered volcanic ashes, or bentonites to define the limits of specific intervals. The advantage of using these bedding features is that they are persistent and are generally identifiable across wide areas of exposed chalk in Western Kansas.

Stewart (1978, 1988, 1990) has shown a correlation between the presence or absence of specific faunal remains and the Marker Units proposed by Hattin (1982). This correlation provides an additional tool for the relative dating of remains, and the further accumulation of data on the composition and relative species diversity of the faunal communities which existed in the Western Interior Sea during the Late Cretaceous period. Additional field work is required to validate the association of Stewart's grouping of species with Hattin's stratigraphic markers.

Stewart's Zone of Protosphyraena perniciosa includes the lowest units of the Smoky Hill Chalk. The zone is of Late Coniacian age and lies comformably on the upper surface of the Fort Hays Limestone. This zone is approximately 32 meters thick and includes Hattin's Marker Units 1-3. It ends at or just below Hattin's Marker Unit 4 which is also the beginning of Stewart's Zone of Spinaptychus. Stewart (1990) indicates that collected remains of an obscure genus of Plethodids (Martinichthys) come from only the uppermost levels of the Zone of Protosphyraena perniciosa and do not transition into the Zone of Spinaptychus. The author found sixteen Martinichthys rostra at this general elevation in the Smoky Hill Chalk during 1989-93 and confirmed the relative biostratigraphy of the genera as it relates both to Hattin's Marker Units and Stewart's faunal zones. The locality records of two other rostra (KUVP 66104 and KUVP 66105), found by J.D. Stewart in 1980, and KUVP 652651, found by D. Hattin, indicate that all three came from the same site and probably the same stratigraphic level as the authors' material.

The strata in which Martinichthys occurs is approximately six meters in thickness and is located at an elevation of between 32 and 39 meters (105-129 feet) above the base of the Smoky Hill member in the composite section of Hattin (1982). The interval is roughly bounded by the strata from approximately 3 meters below Marker Unit 4 to about a meter above Marker Unit 5 as proposed by Hattin (1982) and is probably of late Coniacian age. If the average deposition rate of 0.036 mm per year suggested by Hattin (1982) is valid, then the seven meter depositional interval in which Martinichthys remains occur would have lasted about 190 thousand years.

The localities where Martinichthys remains have been found are located geographically along a line trending northeast from northern Lane County to eastern Phillips County. This corresponds generally with outcrops from the lower portion of the Smoky Hill chalk. The major sites where the remains have been found include Wildcat Canyon in Trego County, Martin's Canyon and the Babcock Ranch in Gove County. Hattin (1982) described the stratigraphy of Wildcat Canyon and the Babcock Ranch. The stratigraphy of the Babcock Ranch was confirmed by J. D. Stewart and the authors during fieldwork in 1990 and 1992. Work by the authors on the stratigraphy of Wildcat Canyon in 1992 indicated that at least some portions of this large exposure are somewhat lower in elevation than proposed by Hattin (1982). Field work done by J. D. Stewart and the authors in 1992 also confirmed the biostratigraphy of the upper portions of Martin's Canyon.

Kansas Academy of Sciences Abstract - Notes on the Biostratigraphy of Martinichthys

Biostratigraphically, Martinichthys is associated with the large bivalves Volviceramus grandis and Platyceramus platinus and occurs several meters below the first occurrence of the distinctive bivalve Cladoceramus undulatoplicatus. The proposed interval also corresponds with the lower portion of Stewart's (1990) Zone of Spinaptychus n.sp and lies just above his Zone of Protosphyraena perniciosa. Based on the authors observations, the actual dividing line between these two biostratigraphic zones is not well defined and some overlap is to be expected.

The authors have also found flattened, elliptical masses of shell fragments consisting almost entirely of oyster shells in the same biostratigraphic zone as the Martinichthys rostra. These masses are fairly common in this zone but are only rarely found slightly above or below the narrow biostratigraphic zone occupied by the genus Martinichthys.

Two representative oyster shell structures (another here, including a cross-section) interpreted by the author as Martinichthys coprolites. Scale is in millimeters

Stewart (personal communication, 1990), and the observations of the authors, suggest that these shell structures are coprolites from an unknown species of molluscivouris fish which apparently fed on the abundant oyster (Pseudoperna congesta) communities found on the Inoceramid molluscs. Examination of the material has shown little or no evidence of the prismatic shell fragments which would indicate predation on the much larger inoceramids such as Volviceramus grandis. It is reasonable to assume that this predator could have been Martinichthys and that the shape and form of the rostrum, as well as the unique bony plate structure of the mouth of plethodids in general could have been an adaptation to feeding on this abundant food supply. It is also plausible that the evidence of wear which has been noted on the Martinichthys rostra and the presence of shell containing coprolites in the same biostratigraphic zone may be linked to the feeding habits of the genus.

The concept of a medium sized fish using it's robust, blunt rostra to batter at colonies of oysters on larger Inoceramid shells and then ingesting the loose fragments of shell and the attached flesh has some merit in the view of the authors. This theory, though largely untestable due to the current lack of sufficient evidence linking the rostra with the shell masses, would explain the unusual association of remains found to date.  It would also appear to be a reasonable exploitation of an otherwise rich and unoccupied ecological niche.


Evidence gathered by the authors between 1989 and 1995 appears to indicate that the Plethodid genus Martinichthys is found only in a limited biostratigraphic zone in the Smoky Hill Chalk of Western Kansas. Using the stratigraphic system proposed by Hattin (1982), the limits of this interval are defined as from three meters below Marker Unit 4 to one meter above Marker Unit 5. In Stewart's (1990) proposed biostratigraphic system, the genus occurs only in the upper portion of the Zone of Protosphyraena perniciosa and is of late Coniacian Age.

Kansas Academy of Sciences Abstract - Evidence of Predation on Martinichthys


Bardack, D. 1965. Localities of fossil vertebrates obtained from the Niobrara Formation (Cretaceous) of Kansas. University of Kansas Publications, Museum of Natural History 17(1):1-14.

Cope, E.D. 1877. On the Genus Erisichthe. Bulletin of the United States Geological and Geographical Survey 3(4):821-823.

Hattin, D.E. 1982. Stratigraphy and depositional environment of Smoky Hill Chalk Member, Niobrara Chalk (Upper Cretaceous) of the type area, western Kansas. Kansas Geological Survey, Bulletin 225

Hay, O.P. 1903. On certain genera and species of North American Cretaceous Actinopterous fishes, Bulletin of the American Museum of Natural History 19:22-24.

Logan, W.N. 1897. The upper Cretaceous of Kansas; with an introduction by Erasmus Haworth, University Geological Survey of Kansas, Vol.2, p. 195-234

McClung, C.E. 1926. Martinichthys - A new genus of Cretaceous fish from Kansas, with descriptions of six new species. Transactions of the American Philosophical Society 65(5):20-26.

Romer, A.S. 1966. Vertebrate Paleontology. 3rd Ed. University of Chicago Press, p.63-64, 354-355

Russell, D.A. 1988. A check list of North American marine Cretaceous vertebrates including fresh water fishes. Occasional Paper of the Tyrrell Museum of Palaeontology #4.

Schultze, H.P., Stewart, J.D., Neuner, A.M. and Coldiron, R.W. 1982. Type and figured specimens of fossil vertebrates in the collection of the University of Kansas Museum of Natural History, Part I., Fossil Fishes. University of Kansas, p. 34-35.

Stewart, J.D. 1988. Stratigraphic distribution of Late Cretaceous Protosphyraena in Kansas and Alabama. Fort Hays State University Studies, Third series, Science Series, (10):80-94.

Stewart, J.D. 1990. Niobrara Formation vertebrate stratigraphy. 1990 Society of Vertebrate Paleontology Niobrara Chalk Excursion Guidebook, p.19-30.

Taverne, L. 1999. Révision du genre Martinichythys, poisson marin (Teleostei, Tselfatiirormes) du Crétecé supérior du Kansas (États-Unis). Geobios 33(2):211-222. (Revision of the genus Martinichthys, marine fish (Teleostei, Tselfatiiformes) from the Late Cretaceous of Kansas (United States))

Williston, S.W. 1897. The Kansas Niobrara Cretaceous: University Geological Survey of Kansas 2:235-246.

Credits: The drawing of the Martinichthys rostrum was adapted from Figure 13 (AMNH 2131) of "On certain genera and species of North American Cretaceous Actinopterous fishes", Bulletin of the American Museum of Natural History, Vol. XIX, p.22-24, 1903, by O.P. Hay.