The Caseids were a group of Synapsids that appeared in the Late Carboniferous and persisted to the Middle Permian. Most are thought to have been herbivorous, some aquatic, and for a while they are thought to have been the dominant megaherbivores in many environments. They were distinguished by small cervical vertebrae, bulky, barrel-shaped bodies, and massive limbs.
In a paper published in the Annals of Carnegie Museum on 31 March 2020, David Berman of the Section of Vertebrate Paleontology at the Carnegie Museum of Natural History, Hillary Maddin of the Department of Earth Sciences at Carleton University, Amy Henrici, also of the Section of Vertebrate Paleontology at the Carnegie Museum of Natural History, Stuart of Sumida, again of the Section of Vertebrate Paleontology at the Carnegie Museum of Natural History, and of the Department of Biology at California State University, Diane Scott of the Department of Biology at the University of Toronto at Mississauga, and Robert Reiz, once again of the Section of Vertebrate Paleontology at the Carnegie Museum of Natural History, and of the Department of Biology at the University of Toronto at Mississauga, describe a new Caseid species based on four partial to nearly complete skeletons.
The new species is a member of an extensive terrestrial vertebrate assemblage from the Early Permian Artinskian Stage Tambach Formation, central Germany. Only two of the four Caseid specimens include skull material: a juvenile and smallest specimen possesses a small partially articulated portion of the skull; and the holotype and largest specimen, an adult, possesses a nearly complete but dorsoventrally crushed skull. The Bromacker Caseid specimens provide a nearly complete knowledge of the skeletal anatomy, which among caseids is surpassed only by that of Cotylorhynchus romeri. Previous age assessments assigned the formation to the partially equivalent North American Wolfcampian age based solely on its faunal composition. The Tambach Formation is exposed over an area of about 50 km² in the mid-region of the Thuringian Forest, Thuringia. To date, 12 vertebrate species, including the Caseid described by Berman et al., have been collected from the Tambach Formation. All but one, were collected from an extensive quarry site, the Bromacker quarry, located in the central area of the Tambach basin about 1.5 km north of the village of Tambach-Dietharz. The exception, a single skull of a new species of Microsaur, Tambaroter carrolli, was recovered at an excavation site for a new supermarket in the village of Tambach-Dietharz. It is from a level considered most likely contemporaneous with or slightly younger than that of the Bromacker quarry.
All the Bromacker taxa are either strictly terrestrial or semi-terrestrial, as Fish or obligatory aquatic Amphibians are noticeably absent. This agrees with the sedimentological and depositional conditions of their preservation in a small internally drained upland basin that was isolated from regional drainage systems and subject to seasonal drying, thus lacking permanent water. Detailed accounts of the geology of the Bromacker quarry and Tambach basin, as well as the taphonomy of the Vertebrates have been provided in previous studies. The Bromacker Vertebrates share a strong taxonomic commonality with those of the mixed terrestrial-aquatic assemblages of the well-documented lowland terrestrial environments of the Early Permian of the Southwest and Midwest regions of the USA. All the species from the Tambach Formation are not only unique to Europe, but also represent new genera or species except Seymouria sanjuanensis, which is otherwise well known from the Lower Permian of the USA. In terms of abundance, diversity, and superb quality of preservation, the Bromacker assemblage exceeds those of all other localities combined in the Lower Permian of Europe. The idea that late Palaeozoic assemblages of tetrapods from upland regions, including those from the upper reaches of rivers, remain completely unknown, underscores the significance of the palaeoenvironmental and palaeobiological features of the Bromacker locality.
Although the Caseasaurians Callibrachion gaudryi from France and Datheosaurus macrourus from Poland are older than the Bromacker caseid described here, they are not included in the discussion by Berman et al., because the specimens are considered to be too poorly preserved to allow their inclusion in a phylogenetic analysis. Therefore, the Bromacker Caseid is considered by Berman et al. to have been temporally preceded only by a single Caseid species, the Late Pennsylvanian Eocasea martini, from Kansas. In contrast, the remaining late Early and Middle Permian Caseids document a dramatic increase in taxonomic diversity with an expansive global distribution that includes Russia, France, Italy, and USA. Lastly, an isolated tooth from the Upper Pennsylvanian of New Mexico was identified as a Caseid. The identification, however, has been refuted as 'erroneous', as it does not conform to the well-known dental patterns seen in any of the known caseids. However, an alternative identification was not offered that would explain its Caseid-like features of a spatulate-shaped crown with a concave distal half of the lingual surface and three cuspules aligned along the apical margin of the crown.
Until recently, Caseids were considered strictly herbivorous, capable of feeding on high-fiber vegetation, as evidenced most prominently by their distinctive large, expanded, barrel-shaped trunk and spatulate or leaf-shaped teeth with apical cuspules. This characterisation, however, has been strongly contradicted by the description of the juvenile specimen of the Caseid Eocasea martini. It not only lacks an expanded rib cage, but reportedly possessed extremely small, simple, tubular teeth attributed to a non-herbivorous insectivorous diet. Unfortunately, the few preserved teeth were lost during preparation before they could be illustrated, and their description was based on the recollections of the authors. The new Bromacker Caseid specimens add greatly to our knowledge of the evolutionary history of herbivory in Caseidae.
In contrast, the Bromacker Caseid possesses a modestly expanded barrel-shaped rib cage, but most significantly the dentitions of the adult and juvenile specimens are strikingly different from one another and those of other Caseids as well. The latter feature is interpreted as indicating an ontogenetic change in dentition in which a juvenile insectivorous dentition is replaced by an adult omnivous dentition. Also, the skull of the adult Bromacker Caseid exhibits several features that characterize the crown-clade Caseids and are believed to be specializations of an herbivorous diet.
Most importantly, the non-herbivorous dentition in Eocasea martini has been interpretted as conclusive evidence that high-fiber herbivory evolved within Caseidae. This, however, discounts the possibility that its dentition might also represent an initial stage in an ontogenetic replacement to one adapted to herbivory or omnivory. Furthermore, considering the strong denial that the isolated tooth from the Late Pennsylvanian of New Mexico is a Caseid, it is ironic that an isolated tooth of Eocasea or the Bromacker caseid would have almost certainly never been identified as Caseid. This raises an additional possibility of the cryptic presence of Caseids in Permo-Carboniferous faunas currently not recognised.
The new species is named Martensius bromackerensis, where 'Martensius' honours Thomas Martens who was not only the first to discover vertebrate skeletal fossils at the Bromacker quarry in the 1970s, but also for his extraordinary efforts in initiating and maintaining an uninterrupted, highly successful, long-term program of quarrying that has resulted in the discovery and description of numerous, exceptionally well-preserved specimens representing a long list of taxa unique to Europe, and 'bromackerensis' refers to the type locality, Bromacker quarry, the only locality where the species is known to occur.
Martensius bromackerensis can be distinguished from other members of the family by the possession of the following autapomorphic character states: (1) splenials excluded from the mandibular symphysis; (2) adult marginal teeth narrowly triangular and slightly recurved; (3) lateral margins of the anterior zygapophyses of the postatlantal-precaudal column curve upward to embrace the lateral margins of the posterior zygapophyses; (4) all but possibly the first two thoracic neural arches possess a small, dimple-like excavation within the apex of the angle formed by the lateral margin of the anterior zygapophysis and anterior margin of the transverse process; (5) penultimate phalanx of digit I in the manus and pes longer than the metapodial I; and (6) penultimate phalanx of digits II–V in the manus and pes longer than the antepenultimate phalanx.
Martensius bromackerensis is distinguished from other Caseids by the following plesiomorphic characters states: (1) procumbent snout projects slightly beyond premaxillary dentition; (2) external naris greatly expanded only posteriorly but not dorsally with length equal to about twice the height; (3) cranium and mandible proportionally elongate and narrow in lateral view; (4) skull table in lateral view modestly convex over its entire length; (5) skull in dorsal view has an outline of a broadly based triangle with the lateral margins converging strongly anteriorly from the occiput to a bluntly pointed snout; (6) long, broad preorbital process of the lacrimal extends to about the midlength level of snout, ending in a very narrow contribution to the posterodorsal margin of external naris; (7) premaxilla, maxilla, and lacrimal lack shelf-like, medial extensions of the narial margin that would form a narrow continuous narial wall; (8) outline of the temporal fenestra vertically elongated with greatest horizontal width occupying about 40% of the postorbital cheek width; (9) squamosal bordering the temporal fenestra relatively wide with greatest width nearly equal to that of the fenestra; (10) adult marginal dentitions include four premaxillary, 18 maxillary, and as many as 25 dentary teeth; (11) double occipital condyle; (12) prefrontal divided longitudinally into subequal skull table and laterally exposed preorbital portions; (13) coronoid eminence of the lower jaw formed entirely by the surangular; (14) absence of deep notch on the anterior margin of the scapular blade that coincides with the dorsal extent of the clavicle contact; and (15) phalangeal formulae of the manus and pes 2-3-4-5-3 and 2-3-4-5-4, respectively.
The holotype of Martensius bromackerensis, MNG 13814 is an adult and largest of the specimens, consisting of a nearly complete, well–preserved skeleton preserved in several, closely associated, articulated sections that include: nearly complete, loosely disarticulated, dorsoventrally crushed skull; articulated presacral vertebral column except for the atlantal intercentrum being displaced and lacking the right half of the bipartite atlantal neural arch; first two sacral vertebrae with fused ribs and intercentrum of the third; three sections of the caudal series of 26, six, and 21 vertebrae of progressively more posterior positions in the tail; greater portions of the shoulder girdles, but lacking the clavicles; greater portion of the pelvic girdle that includes the right ilium and articulated paired puboischiadic plates preserved mainly as impressions; nearly complete articulated forelimbs with the left lacking mainly the humerus and proximal portion of the ulna; and complete, articulated left pes with the distal end of the fibula.
There are three paratypes of Martensius bromackerensis, MNG 14230, a juvenile and the smallest specimen, consisting of a nearly complete articulated skeleton that includes: skull consisting of partial right premaxilla and maxilla, remnants of the palate, greater anterior portion of the left mandible, and anterior portion of the right mandible; axial skeleton consisting of remnants of the posteriormost three or four cervicals that is continued posteriorly by an articulated column through the first seven caudals, which in turn is followed by strings of six and 11 vertebrae that are preceded by gaps containing poor impressions of three and four vertebrae, respectively; fragmentary shoulder girdles and complete pelvic girdles; and partial left forelimb and largely complete and articulated right forelimb and hind limbs. MNG 10552 is an adult slightly smaller than the holotype consisting of poorly preserved disarticulated and randomly scattered postcranial bones and a few cranial elements. MNG 10595 is the greater portion of an articulated postcranial skeleton of an adult slightly smaller than the holotype that includes a large portion of the dorsal vertebral column, partial forelimbs, nearly complete right hind limb, and nearly complete set of gastralia.
Martensius bromackerensis possesses numerous Caseidae synapomorphies, leaving no doubt as to its assignment to this group, which along with its sister taxon Eothyrididae comprise the monophyletic clade Caseasauria. In describing the differences of Martensius from other Caseids an emphasis is placed on those characters that uniquely characterise the entire family. In this regard, only the Late Pennsylvanian Eocasea martini, considered the oldest, most primitive, undisputed Caseid, is seriously problematic, as the holotypic skeleton exhibits at best only a couple of family synapomorphies. A more complete diagnosis was undoubtedly unavailable due to poor or incomplete preservation of the skeleton and incomplete or complete absence of ossification of bones indicating an early juvenile ontogenetic stage of development. Therefore, where it is stated that Martensius is like that of other of Caseids, Eocasea is excluded from consideration to avoid confusion and the need for repetitive explanations.However, in those instances where similarities or differences of Eocasea from other Caseids have been described, they are re-evaluated by Berman et al. Also relevant to the weak diagnosis of Eocasea is a recent exhaustive phylogenetic analysis of the basal Synapsid Varanopidae in which it is stated that 'Eocasea appears as the basalmost Caseasaurian, instead of being a Caseid. Similarly, in a cladistic analysis of Caseasauria, Eocasea was resolved as the sister taxon to Eothyris, and thus not a Caseid.
Berman et al.'s description of the skull is largely based on that of the adult holotype MNG 13814. Despite severe dorsoventral crushing of the skull, the bones are with few exceptions exposed in the same views of an articulated skull; that is, the skull roofing bones in dorsal, the palatoquadrate complex in ventral, and the occiput in posterior view. This has allowed accurate reconstructions of nearly the entire skull in dorsal and lateral views, which along with the preserved state of the holotype can be relied on for accuracy in the anatomical descriptions and discussions below. Surprisingly, in these views the overall morphology of the cranium and mandible appear to belie a Caseid identity. Rather than being relatively short and deep with subrectangular outlines, they are decidedly more comparable to Eothyridids and most carnivorous basal synapsids in having elongate slender profiles with a strongly convex skull table. Apart from Martensius the skulls of Caseids can be further chacterised by a unique three-dimensional box-like construction. This is best illustrated by comparing the outlines of the skulls in dorsal view. In Martensius the outline is a broad based triangle with the lateral margins converging strongly anteriorly to a bluntly pointed snout. This is strongly contrasted in other Caseids by a skull width that is essentially unchanged from the occiput to about the skull midlength before quickly narrowing to a bluntly pointed snout.
Severe dorsoventral crushing of the skull has resulted in most of the roofing bones being at least narrowly disarticulated and often with the margins either incomplete or obscured by adjacent bones. All the paired skull roofing bones are represented by at least one of the pair except for the postorbitals and jugals.
As is characteristic of other Caseids and Eothyridids, the snout of Martensius is procumbent, but far less pronounced, with the anterior margin of the premaxilla directed abruptly ventrally and slightly posteriorly to project only slightly beyond the premaxillary dentition. Also, uniquely characteristic of other Caseids, Martensius has a relatively short preorbital snout length that occupies about 30% of the skull length, which is well within the Caseid range of about 24 to 37%. A unique skull roof sculpturing pattern of randomly dispersed, small, shallow, well-defined, round to oblong depressions also unites Martensius, as well as Eocasea, with other Caseids.
The external naris in Martensius approaches that of other Caseids in being greatly expanded posteriorly, but in contrast is far less so dorsally, forming an oblong opening that is twice as long as high and occupies a much smaller area of the snout. In other Caseids the naris is greatly expanded dorsally into a subcircular opening that occupies most of the lateral surface of the snout to rival the orbit in size. The naris of Martensius also differs from other Caseids in the absence of shelf-like medial extensions of the narial margins of the premaxilla, maxilla, and lacrimal, which together form a narrow, continuous narial wall. As in other Caseids the orbit in Martensius is greatly enlarged and comparable in occupying about 30% of the skull length. The orbit has roughly the outline of an anteroposteriorly elongated trapezoid with the dorsal margin paralleling a much shorter ventral margin. In the few available skull reconstructions of caseids, the orbits are highly variable, including circular, oval, or triangular outlines. The temporal fenestra in Martensius has an outline of a modest-sized, vertically elongated, slightly oval opening with a greatest horizontal width equal to about 40% of the postorbital cheek length. In marked contrast, the same measurement in other Caseids may be as much as 75%, which is reflected in an extremely narrow squamosal bordering the posterior margin of the opening. The lateral exposure of the squamosal in Eocasea is similarly broad and, therefore, would have greatly reduced the width of its temporal fenestra to an outline more closely approaching that in Martensius.
The premaxillae, other than forming a slightly procumbent snout, are unremarkable. Their narrow united dorsal processes are almost entirely restricted to the anterior margin of the external naris before extending a short distance onto the skull table between the nasals. The ventral maxillary process of the premaxilla laterally overlaps the anterior end of the premaxillary process of the maxilla in a posteroventrally oblique suture that reaches the ventral margin of the skull at about the midlength level of the naris. Both premaxillae possess four teeth, with the anterior pair being about 20% larger than the posterior pair and slightly larger than the largest of the maxillary series. They are narrowly triangular and very slightly recurved. The ventrally exposed right premaxilla in the juvenile MNG 14230 also indicates four teeth, but they are represented only by their circular bases except for an empty alveolus of the third. On this basis the first three teeth exhibit a marked serial increase in size posteriorly, whereas the fourth is approximately the size of the first.
Only the left nasal can be accounted for in the holotype, but it is complete and retains its contacts except for that with its mate. It is divided longitudinally by a right-angled bend into dorsal skull table and lateral snout exposures. The skull table exposure is widest for a short distance anterior to its transverse contact with the frontal, whereas anteriorly it narrows along its lateral margin to half its posterior width before ending in a short contact with the lateral margin of the dorsal process of the premaxilla. The larger laterally exposed portion of the nasal includes a smooth, ventrally directed, flange-like expansion that appears to have bordered nearly the entire dorsal margin of the external naris. The flange is widest posteriorly, where it appears to also include a very narrow portion of the anterior margin of the lacrimal, whereas anteriorly it narrows to the dorsal rim of the naris.
The frontals are essentially complete and have a greatest overall length equal to about half that of the skull table. At the midlength of their lateral margin, a narrow, short, rectangular orbital process intervenes between the prefrontal and postfrontal, dividing the frontal into approximately subequal lengths, an anterior rectangular process and a triangular posterior process that narrows posterolaterally in an irregular margin from the midline to the medial margin of the postfrontal.
Only the left prefrontal can be accounted for, but it is essentially complete and retains its marginal contacts. It is divided longitudinally into a dorsally exposed skull table portion and a laterally exposed preorbital portion, which together border the anterodorsal corner of the orbit. The skull table portion contacts the entire lateral margin of the anterior process of the frontal and narrowly the posterolateral corner of the nasal. Due to the crushing of the skull, however, the laterally exposed portion of the prefrontal has been folded medially beneath the skull table, greatly reducing its visibility in the palatal view of the skull to a narrow exposure contacting the dorsal margin of the preorbital portion of the lacrimal. As restored, a shallow convex ventral margin of the preorbital portion of the prefrontal extends into the posterior half of the dorsal margin of the lacrimal.
Only the right postfrontal is visible, and, although disarticulated, is intact. As in other caseids it is largely confined to the skull table and has an outline of a posteriorly expanding rectangle. Medially it contacts the entire lateral margin of the posterior process of the frontal, whereas posteriorly it has a transverse margin that extends slightly into the anterior margin of the parietal. Its lateral margin forms the posterior half of the dorsal orbital rim, which is continued a very short distance posteroventrally by a small, bluntly pointed process. Although the postorbitals are lost, reconstruction of the skull roof indicates that they had a large exposure on the skull table bounded medially by the parietal, anteriorly by the postorbital and orbit, and posteriorly by the supratemporal. The original description described a large postorbital in Eocasea, which was recognised as a Caseid synapomorphy.
The parietal in Martensius is represented only by the right bone, and, although disarticulated is sufficiently preserved to roughly describe its extent and marginal contacts. It has the form of a broad, posterolaterally directed, wing-like structure. Medially the anterior margin contacts the posteromedial margin of the posterior process of the frontal. A short section of the anterolateral margin of the parietal extends laterally between the postfrontal and supratemporal, presumably contacting the missing postorbital before continuing posteriorly beneath the supratemporal. The disarticulated posterior margin of the right parietal is directed posterolaterally and was contacted by the anterior margins of the postparietal and tabular in a straight suture. The left lateral half of the pineal foramen indicates a slightly transversely elongated opening positioned at about midway along the midline union of the parietals.
The right elements of the paired postparietals and tabulars appear to be complete or nearly so, but are widely displaced and partially obscured by the over-hanging posterior margin of the parietal. They are transversely elongated, extremely thin, and other than a slight separation at their end-to-end contact, clearly formed a smoothly continuous structure between contacts with the occiput and parietal. Their anterior edges inserted into a very shallow groove on the posterior margin of the parietal, whereas an abutment contact best describes their contact with the dorsal margin of the occiput. The slightly longer postparietal ends medially at the midline in a short, ventrally directed, lobe-like expansion that inserted into a shallow depression on the occiput. Laterally the postparietal appears to have extended to the medial margin of the posttemporal foramen, which in the disarticulated skull appears as a deep V-shaped notch in the dorsal margin of the occiput. The tabular appears to have enclosed the posttemporal foramen dorsally before continuing laterally a short distance and being overlapped by the supratemporal.
Both supratemporals are present and are closely comparable to those in caseasaurians, including Eocasea martini. In Martensius they have an anteroposteriorly elongated, broadly oval outline, and occupy the posterolateral corner of the skull roof. The posterior half of the supratemporal overlapped the anterolateral corner of the opisthotic portion of the occiput and the dorsal half of a medially inflected occipital flange of the squamosal.
Both lacrimals are preserved, but only the left is essentially complete and fully exposed but visible almost entirely in the palatal view of the dorsoventrally crushed skull. It has an extensive contribution to the anteroventral corner of the orbit, and a long, broad preorbital process extends anteriorly well beyond its dorsal contact with the prefrontal. Here it ends in an irregular contact with the posteroventral margin of the nasal and a very narrow contribution to the posterodorsal margin of the external naris. A well-developed suborbital process of the lacrimal extends posteriorly to the midlength level of the orbit where it presumably contacted the anterior end of a suborbital process of the missing jugal to exclude the maxilla from the orbit. The lacrimal ventrally contacts approximately the anterior two-thirds of the postnarial dorsal margin of the maxilla. In other caseids the combination of extreme expansions of the external naris and orbit and a greatly shortened snout has resulted in striking differences of the lacrimal from that in Martensius: (1) restricted to a narrow preorbital bar without a preorbital process; (2) ventrally contacts a much shortened area of the postnarial dorsal margin of the maxilla; (3) suborbital extension reduced to an extremely short, thin process that contacts the anterior end of the suborbital process of the jugal; and 4) has an extensive contribution to the posterior margin of the external naris at the expense of loss or near loss of contact with the nasal.
Only the right squamosal can be accounted for, but is sufficiently exposed in the dorsal and palatal views of the skull to describe its extent and contacts. In lateral view of the skull roof, it has roughly the outline of a ventrally widening rectangular plate with a slightly concave anterior margin that defines the posterior border of the temporal fenestra. Dorsally the squamosal is believed to have contacted the ventral margin of the posterior process of the missing postorbital, whereas ventrally it contacts approximately the posterior half of the dorsal margin of the quadratojugal in a slightly sigmoid suture. The posterior margin of the squamosal is inflected abruptly medially as a narrow flange that vertically extends nearly the full height of the occipital margin of the cheek in forming the posterior wall of the adductor chamber.
Only the greater portion of the right quadratojugal is preserved, but its exposure is portioned between the dorsal and palatal views of the skull. It is laterally exposed as a long, narrow, anteriorly tapering element along the ventral margin of the cheek with a slightly concave ventral margin. At about the midlength along its dorsal margin, it appears to have formed a short ventral margin of the temporal fenestra. Farther anteriorly it likely contacted the ventral margin of a posterior process of the missing jugal before ending in contact with the posterior end of the maxilla. Posteriorly the quadratojugal wraps medially around the posterior surface of the ventral articular condyle of the quadrate, covering all but a very narrow exposure along its ventral margin.
Although the right maxilla is essentially fully represented, it is broken into two pieces; a greater anterior portion is laterally exposed and a smaller posterior portion is medially exposed in the palatal view of the skull. Also visible in palatal view of the skull is the laterally exposed anterior end of the left maxilla, and together they allow a nearly complete description of the maxilla. A premaxillary process of the maxilla extends nearly the entire length of the ventral margin of the external naris. From its greatest but modest height bordering the posterior margin of the external naris, the dorsal margin of the maxilla gradually slopes posteroventrally to presumably the ventral margin of the skull where it narrowly contacts the quadratojugal. Posterior to the naris the dorsal margin of the maxilla contacts the entire ventral margin of the lacrimal and likely also the ventral margin of an anterior process of the missing jugal. The maxilla possesses 18 teeth in contrast to the full counts ranging from eight to 15 in other Caseids. The maxillary teeth in Martensius gradually increase serially in size posteriorly to about the ninth tooth before gradually decreasing posteriorly with the last few being extremely small. The teeth are narrowly triangular, slightly recurved, and collectively exhibit a nearly isodont dentition. The maxillary teeth in other caseids differ strikingly from those of Martensius in being spatulate or leafshaped and usually possessing as many as five cuspules aligned along the apical margin. In the juvenile MNG 14230, the laterally exposed anterior portion of the incomplete right maxilla possesses 11 mostly intact teeth, but with the fourth tooth position represented by a gap, and the teeth are essentially identical to those in the holotype MNG 13814. The anterior four teeth increase serially in size posteriorly to the largest of the entire series. These are followed by four much smaller teeth of subequal size, whereas the remaining three teeth decrease serially in size posteriorly from a size nearly equal to the largest of the series.
The holotypic septomaxillae appear to be incomplete and limited essentially to a longitudinally elongated ventral base supported by the premaxilla and maxilla along the ventral rim of naris. At the midlength of the left bone, however, is a small, dorsal projection that likely represents a remnant of a dorsal process, which would separate the larger anterior true narial opening from an enlarged posterior septomaxillary foramen.
All that remains of the palate of the juvenile specimen MNG 14230 are three disarticulated right elements densely covered by small denticles that include a nearly complete vomer, the anterior portion of the palatine, and an unidentifiable fragment. In the holotype MNG 13814, the left side of the palate preserves only the greater posterior portion of the vomer and a small portion of the anterior end of the palatal ramus of the pterygoid. However, all the right palatal bones are essentially complete but with some disarticulation and are fully exposed except for some marginal overlap by neighboring bones. The vomers contact one another along the midline except for possibly a short distance posteriorly where they appear to narrowly diverge, presumably forming the anterior margin of the interpterygoid vacuity. A shallow, concave emargination along the midlength of the lateral margin of the right vomer presumably bordered the medial margin of the internal naris. Anterior to the narial margin, the vomer would have laterally contacted the alveolar shelf of the premaxilla. Posterior to the narial margin, the vomer expands to more than twice its anterior width and contacts the combined anterior transverse margins of a medial pterygoid and a lateral palatine in a deeply interdigitating suture. Marginal to their midline contact, the vomers support a dentition that gradually increases serially in size posteriorly from denticles to small teeth.
The palatine has a transversely broad, anteroposteriorly elongate rectangular outline with the medial margin contacting the entire lateral margin of the anterior palatal ramus of the pterygoid and the posterior margin contacting the anterior margin of the lateral transverse flange of the pterygoid. Along the lateral margin of the palatine is a partially preserved thin, low, ventrally inflected flange that contacts the medial margin of the alveolar shelf of the maxilla. The medial margin of the palatine supports a row of small denticles that slightly increase serially in size posteriorly. A small subrectangular ectopterygoid is partially interposed between the palatine and transverse process of the pterygoid at their lateral contact. Its shallow, smoothly finished concave surface thickens slightly laterally into a buttress-like contact with the posterior end of the alveolar shelf of the maxilla.
The palatal ramus of the pterygoid has an anteroposteriorly elongate rectangular outline that laterally borders a narrow interpterygoid vacuity. Its anterior contact with the vomer excludes the palatine from the interpterygoid vacuity. Densely packed small denticles extend along the medial margin of the palatal ramus and partly onto the transverse flange. The lateral margin of the palatal ramus supports a row of small, conically pointed teeth that serially increase in size posteriorly and barely continue along the anterior margin of the transverse flange. The smooth surfaced subrectangular transverse flange of the pterygoid greatly thickens toward its free posterior and lateral margins, giving it a shallow basin-like structure. The entire posterior margin supports an unbroken single row of small, conically pointed teeth that serially increase slightly in size laterally but quickly diminish as they continue a short distance along the lateral margin of the flange. A slightly concave lateral margin of the flange immediately anterior to the marginal dentition presumably borders a narrow anterior extension of the subtemporal adductor fossa. Clearly visible at the medial margin of the union of the transverse flange and palatal ramus of the pterygoid is a short, ventrally bowed, rectangular flange-like basal process, the dorsal surface of which receives the basipterygoid process of the cultriform process in a probably weakly mobile joint.
Only the right disarticulated quadrate is visible, and, although undoubtedly complete, its exposure is limited to two small partial views. A ventrally exposed articular surface is bordered posteromedially and posterolaterally by oval articular condyles, with the former being significantly larger. In dorsal view of the skull, the greater dorsal portion of the quadrate is laterally exposed as a thick platelike process that has been displaced to the dorsolateral margin of the right opisthotic of the occiput.
Ventrally exposed bones of the braincase include the parasphenoid of the fused basiparasphenoid complex, basioccipital, and sphenethmoid. Although the parasphenoid is undoubtedly complete, it is ventrally obscured anteriorly by the right pterygoid and sphenethmoid, exposing only the proximal half of the cultriform process and the right basipterygoid process. Posteriorly only a very small area of its left posterolateral corner is visible. The visible portion of the cultriform process narrows anteriorly with the lateral surfaces converging ventrally to support a slightly elevated platform that proximally bears an anteriorly converging double row of small teeth. The far lateral position of the left posterolateral corner of the parasphenoid from the midline indicates that posteriorly the main body becomes greatly expanded transversely. Protruding posteriorly from beneath the dorsal surface of the parasphenoid is a partially exposed, transversely narrow basioccipital that is marginally overlapped posteriorly by a probable atlantal cervical rib. Disarticulation of the right basicranial joint exposes the ventral articular surface of the basipterygoid process, which is divided into two laterally diverging, smoothly rounded, ridge-like condyles, the posterior of which is wider and longer.
The sphenethmoid is disarticulated, exposed in right lateral view, and largely obscured by other bones, but most aspects of its structure and original position in the skull can be safely inferred. Dorsally it consists of a broad, anterodorsally angled, blade-like structure that is distally overlain by the right palatine. Ventrally the dorsal blade is supported by a horizontally flattened base that extends anteriorly well beyond the anterior margin of the dorsal process; the presence of a similar posterior extension of the base is obscured by the right pterygoid. The sphenethmoid undoubtedly occupied the midline of the skull, extending between a contact with the dorsal surface of the cultriform process and very likely a contact with the ventral surface of the skull roof.
Braincase bones exposed in dorsal view of the skull include the posteriorly exposed occiput and the dorsally exposed disarticulated basioccipital; the exoccipitals are apparently missing. The co-ossified supraoccipital-opisthotic complex forms a massive, bilaterally expansive flat occipital plate that lacks only a lateral portion of the right opisthotic region. It is restored tilted anterodorsally and extending posterolaterally and ventrally from the midline to occupy nearly the entire posterior surface of the skull except for being bordered dorsally and laterally by dermal skull roofing bones. At about midlength along the dorsal margin of the right half of the disarticulated occiput is a deep V-shaped notch of the posttemporal fenestra, which approximates the vertical division between the two bones of the occipital plate. At the ventral midline of the occiput the dorsal margin of the foramen magnum is defined by a shallow, dorsally arched border. Ventrally the opening is abruptly expanded a very short distance ventrolaterally before being redirected ventrally by the medial margin of a slightly depressed triangular flange. The triangular flanges presumably supported the bases of the missing exoccipitals, which formed the lateral margins of the foramen magnum.
A dorsally exposed basioccipital is displaced a short distance posteriorly from its original contact with the occiput. Its nearly flat, subrectangular dorsal surface is divided along the midline by a very low ridge, which probably marked the ventromedial union with the exoccipitals. Extending anteriorly from the ridge is a narrow V-shaped notch that may mark the posterior extent of the parasphenoid contact. The lateral margins of the basioccipital are interrupted by incompletely preserved small laterally directed processes, which presumably strengthened the contact with the occiput. Medially adjacent to the right lateral process is a remnant of an extremely narrow transverse channel that probably marks the passage for cranial nerve XII and would have been enclosed dorsally by the exoccipital. The dorsal surface of the posteriorly directed double occipital condyle of the basioccipital is barely discernable as two short, convex, posterior extensions.
Preservation of the stapes is limited to the right bone in the holotype MNG 13814, which is fully exposed in occipital view of the skull and well preserved except for a very narrow break separating the footplate and shaft. It occupies its near correct anatomical position except for possibly a narrow separation of the footplate from the fenestra ovalis. Otherwise, it extends directly laterally from the fenestra ovalis to a level slightly beyond the lateral margin of the paroccipital process. The footplate is large with a greatly thickened, rounded lip-like flaring of its subcircular articular surface. The massive subcylindrical shaft thickens gradually distally with the dorsal and ventral surfaces being slightly concave and convex, respectively. The terminal margin of the shaft has a vertically elongated oval outline that is tilted slightly dorsomedial. At its terminal margin the coarsely textured core of the shaft extends very slightly beyond the smoothly finished outer bone surface, suggesting perhaps that the distal end of the shaft may have been continued by an extra-stapedial cartilage. A stapedial foramen is not detectible, but this may be due to the break at the footplate-shaft union. At about the midlength of the dorsal surface of the shaft is a small, mediolaterally oriented triangular blade-like dorsal process. The morphology of the stapes is essentially identical to those in the only two other Caseids in which it is known, Casea broilii and Euromycter rutenus (new combination of Casea rutena). Despite the small sample size, the Caseid stapes is quite distinct from those of other basal Synapsids, and, therefore, is considered a unique feature of Caseids.
The right mandible of the holotype is complete, well preserved, and laterally exposed in the ventral view of the skull. Dorsal view of the skull reveals the proximal end of the left mandible that includes small, laterally exposed portions of the angular, surangular, and splenial contacting the lateral surface of the dorsally exposed articular. Apart from proportional differences the lateral view of the right mandible differs from those of other caseids only in details. The coronoid eminence is formed entirely by a low surangular with a broadly convex dorsal margin that in other Caseids also receives a contribution from the dentary. Approximately the posterior half of the angular is broadly exposed between the ventral margin of the mandible and a slightly concave ventral contact with the surangular. Farther anteriorly the angular abruptly narrows to a splint-like projection that extends between approximately the posterior halves of the dorsal dentary and ventral splenial. Posteriorly the angular laterally overlaps all but a narrow posterior margin of the articular that is typical of caseids in lacking a definitive retroarticular process. The dentary maintains a constant depth as it extends anteriorly for about two-thirds the length of the mandible, whereas the posterior half angles dorsally, terminating in a triangular margin that laterally overlaps the anterior margin of the surangular. The splenial has an extremely thin, splint-like exposure along approximately the anterior two-thirds of the ventral margin of the mandible that narrows slightly from its midlength anteriorly and posteriorly along contacts with the dentary and angular, respectively. The splenial appears to be narrowly excluded from the symphysis by the dentary, which in other Caseids has a small to substantial contribution. The articular surface of the articular slopes steeply ventromedially and is occupied entirely, as in the left bone, by a prominent anteroposteriorly elongated condyle that greatly expands in width anteriorly.
The dentary has as many as 25 teeth, including gaps of missing teeth, which far exceeds the counts in of all other Caseids of from eight to 17. The dentition is isodont except for a serial reduction of few teeth at both ends of the series. The teeth are identical to those of the maxilla but do not quite reach their slightly greater size.
The left mandible is in great part complete in the juvenile MNG 14230, whereas only the anterior half of the right remains, and both exposed in lateral view. The sutural pattern is identical to that in the holotype, but the dentition is markedly different. Nearly the entire dental series of the dentary is dorsally exposed and, despite a few gaps of roughly two to four teeth, a total number is estimated at about 31. Unfortunately, all that remains of the teeth are the bases, which have a cylindrical or tube-like structure that likely extended to the crown. For most of the anterior extent of the dentition the tooth bases maintain an essentially constant size with a maximum diameter of slightly greater than 1.0 mm. The approximately posterior-most dozen tooth bases exhibit a marked serial reduction in the size posteriorly to barely detectible.
Description of the axial skeleton is based entirely on wellpreserved, nearly complete articulated columns in the juvenile MNG 14230 and the adult holotype MNG 13814. In the juvenile the precaudal series is dorsally exposed and complete except for the cervical series, which consist of only weathered centra of the posterior three or four vertebrae. On the other hand, the only missing elements of the precaudal series in the holotype are the right half of the bipartite atlantal neural arch, possibly the bipartite proatlas, and the third sacral centrum. The entire precaudal series is exposed dorsally, as well as lateral and ventral exposures of the anterior five cervicals and ventral exposures of the posterior 16 presacrals, anterior two sacrals, and intercentrum of the third. Three subdivisions of the presacral column are recognised based on the type of association of the ribs with the centra and whether or not the ribs reached the cartilaginous sternum: (1) the ribs of the cervical vertebrae articulate with the centra and are much too short and too far anterior in the column to reach the sternum; (2) the ribs of the thoracic vertebrae articulate with the centra, are greatly elongated, and appear to have reached the sternum; and (3) the ribs of the lumbar vertebrae are weakly fused to the centra and much too short and far posterior in the column to have reached the sternum or are absent posteriorly. Based on these definitions the presacral column in Martensius is roughly estimated to consist of six cervicals, 15 thoracics, and five lumbars for a presacral count of 26, which is within the ranges of 24 or 25 to 27 in other Caseids; 27 presacrals were reported in Eocasea. The juvenile specimen of Martensius preserves a complete sacral series of three vertebrae, whereas in the holotype the posteriormost third centrum is absent but the intercentrum is present. Except for Eocasea possessing only two sacrals vertebrae, and Angelosaurus romeri, possessing four, all other Caseids, as far as known, possess three sacrals. Intercentra are visible throughout the 19 ventrally exposed, articulated precaudals in the holotype.
In the juvenile MNG 14230, the sacral series is followed by a string of seven caudal vertebrae, which in turn is followed by strings of six and 11 vertebrae that are preceded by gaps containing poorly preserved impressions of three and four vertebrae, respectively. Although the total caudal count of 31 is low, a large distal portion of the tail is missing. In the holotype, on the other hand, an estimated ten anteriormost caudals are absent, but representing progressively more posterior portions of the tail are isolated strings of 26, six, and about 21 vertebrae; the lattermost string includes the distal or near distal end of the tail. Based on these counts the tail possessed 63 vertebrae, which slightly exceeds the average of about 60 in basal Synapsids.
In the holotype, as in all basal Synapsids, the first two cervicals are specialized as the atlas-axis complex. Only the left half of the bipartite atlantal neural arch is present, which is narrowly displaced from its contact with the dorsal surface of the atlantal centrum. The atlantal intercentrum is widely separated from its contact with anterior face of the atlantal centrum, but retains its contact with the ventral surface of the basioccipital. In dorsal view the left atlantal neural arch is anteroposteriorly elongated and dominated by the anterior and posterior zygapophyses. Both have broadly rounded margins that narrow slightly proximally and merge with one another a short distance anterior to the arch midlength. The dorsal surface of the much lager anterior zygapophysis is flat and presumably would have received the missing proatlas. The dorsal surface of the posterior zygapophysis is slightly convex in transverse section, and the ventral surface contacts the anterior zygapophysis of the axis in a horizontal plane. Interpreted as a greatly reduced epipophysis is a low, longitudinal, transversely thick ridge with rounded margins that heightens slightly posteriorly as it extends along the posterior medial margin of the neural arch half. At a short distance anterior to its midlength, the neural arch half is supported ventrally by a robust pedicel that extends far medially in a broadly convex terminal margin. The ventral surface of the pedicel is expanded transversely into an oval concave facet, which has been narrowly separated from its contact with a low, thickened, rounded ridge-like facet bordering the lateral margin of the dorsal surface of the atlantal centrum. It appears that the bases of the pedicels did not reach the midline of the atlantal centrum, exposing a narrow, slightly concave surface of the ventral floor of the neural canal. Extending posterolaterally from the lateral margin of the pedicel is a relatively long, stout transverse process that is subcircular in cross-section, as is its terminal facet for the atlantal rib.
The axial neural spine approximates those of the posterior cervicals in height, but differs most noticeably in two features: (1) in lateral view the horizontal width is much greater, and (2) in horizontal section the spine expands greatly posteriorly in a triangular outline that is sharply marginated anteriorly. On the other hand, the axial neural spine shares three features with those of cervicals 3 and 4 not seen farther posteriorly in the column: (1) a slight anteroventral inclination; (2) a prominent sharp-edged ridge that extends the entire length of the posterior margin; and (3) a low, rounded ridge on the posterior margin of the lateral surface of the spine extends between the base of the posterior zygapophysis and the spine summit.
Despite a few losses, the postaxial-precaudal vertebral series is well preserved in the holotype. In lateral view the neural spines appear as rectangular blades with a height slightly exceeding the width. In horizontal section, however, the lateral surfaces are slightly convex, narrowing anteriorly and posteriorly to keel-like edges. The spine summits of the entire series posterior to the axis are capped by low swellings that extend slightly beyond the lateral margins of the spine, which in dorsal view have outlines that vary from circular to anteroposterior narrow ovals. The zygapophyses of the postatlantal cervicals are broadly expanded in a circular outline with the articular planes only slightly tilted ventromedially from the horizontal. The zygapophyses of the postcervical-precaudal vertebrae are noticeably narrower with an articular plane tilted about 30o ventromedially from the horizontal. In the same series the lateral margins of the anterior zygapophyses curve upward to embrace the lateral margins of the posterior zygapophyses, which presumably would have partially restricted lateral flexion of the trunk. Also not described in other caseids, in all the thoracic vertebrae except possibly the first two there is a small, well-defined dimple-like excavation within the angle formed between the anterior zygapophysis and the transverse process.
The transverse processes of the postatlantal cervicals are identical to that of the left atlantal neural arch. In contrast, the transverse processes of the thoracic vertebrae are broadly expanded anteroposteriorly, flattened dorsoventrally, and have an oval cross-sectional outline that matches the terminal facets. Anteriorly in the series the transverse processes are oriented strongly anterolaterally with a rectangular outline, whereas posteriorly they are gradually redirected laterally, double in width, and expand slightly distally. The lumbar ribs are weakly fused to a low, rugose ridge-like transverse process that extends horizontally across the entire lateral surface of the centrum.
Although the atlantal intercentrum retains its contact with the ventral surface of the occipital condyle of the basioccipital, the pair is widely separated from contact with the anterior face of the atlantal centrum. In dorsal view of the skull, the atlantal intercentrum is partially exposed, protruding posteriorly from beneath the basioccipital, where it has a narrow rectangular surface with a straight posterior margin. The posterolateral corners are, however, reduced to a very short parapophysis with a posterolateral facing terminal facet for the capitulum of the atlantal rib. In posterior view, the narrowly rectangular intercentrum is slightly concave dorsally to accommodate the ventral surface of the occipital condyle of the basioccipital. The posterior surface is flat except for a low, narrow ridge that extends laterally from either side of the midline along its ventral margin, becoming slightly more developed as it merges with the rim of the rib facet. The atlantal centrum and axial intercentrum are fused into a single element with no visible sutural divisional. Rather, extending transversely across their combined convex ventral surface is a thin, low rugose ridge that presumably marks their line of fusion. Based on this division the atlantal centrum portion is block-like and ventrally fused to the anterior surface of a dorsally wedge-shaped axial intercentrum so that the complex is inclined anterodorsally. The midventral length of the atlantal centum is reduced to about 20% of those of the posterior cervicals. A small notochordal pit is positioned near the dorsal margin of the anterior surface of the centum. The lateral surfaces of cervical centra 2–4 are slightly concave in horizontal section with the length slightly exceeding the maximum vertical height of the more expanded posterior end of the centra. There is no midventral keel of the centra, and in its place is a low, wide, essentially flat elevation, which expands slightly from its midlength to merge with the base of the rounded articular lips of the circular centrum rims. The mid-ventral length of the axial intercentrum is greatly expanded to about twice that of cervicals 3 and 4.
The ventrally exposed series of posterior precaudal vertebrae in the holotype includes 11 thoracics, five lumbars, and first two of the three sacrals and the intercentrum of the third. In ventral view the thoracic centra are essentially identical to the postatlantal cervicals except in being much larger with lengths slightly exceeding the widths. This proportion, however, is reversed in the lumbars and sacrals by a marked increase in centrum widths. The centra of the entire series can be characterized as generally having an overall stout, spool-shaped appearance with the circular ends inflected laterally into a rounded articular lip. There is, however, a noticeable greater lateral expansion of the posterior articular lips of the lumbars and sacrals. Although most of the intercentra of the series are in place, a few are displaced or missing. They exhibit the standard crescent shape in end view with the outer margins curving upward to conform to the circular curvature of the adjacent articular lip of the centrum, whereas in lateral view they are slightly wedge shaped, narrowing dorsally to a blunt point.
Although the atlantal ribs are widely disarticulated and exposed in the dorsal and ventral views of the holotypic skull, the three postatlantal cervical ribs are fully exposed and in near correct articulation with the vertebra. The one complete atlantal rib, the left, lies on the ventral surface of the articulated basioccipital and atlantal intercentrum. It is dichocephalous, and the rib heads occupy half the length of the rib and appear as inline continuations of the shaft in near contact with one another, although the tuberculum is slightly longer and narrower than the capitulum. The postatlantal cervical ribs are essentially straight but short with a length equal to about one and half times that of the centrum. Nearly half the length of the rib is occupied by the heads, but with the tuberculum being significantly longer and narrower. The heads are barely separated from one another and subcircular in cross-section, as are their terminal facets. The rib shafts extend most closely in line with the capitulum, and have a very narrow, blade-like structure. The ribs have been swung posteriorly to lay against the lateral surface of the centrum, with the heads adjacent to their respective articulation sites, the capitulum near the ventral level of the intercentrum-centrum contact and the tuberculum with the transverse process.
Due to dorsoventral flattening of the rib cages in MNG 14230 and the holotype, the shafts of the thoracic ribs face mainly dorsally. Although the rib heads occupy a very short proximal length of the rib and are expanded very little in width, they are considered dichocephalous, as a very short, narrow gap is visible separating the capitulum and tuberculum. The tuberculum occupies almost the entire width of the rib head and extends proximally slightly beyond an extremely narrow capitulum. The tuberculum is oval in cross-section, as is its terminal articular facet, which matches that of the transverse process. The greater proximal lengths of the thoracic rib shafts are dorsoventrally oval in cross-section, but distally become essentially circular. A prominent ridge extends from the terminal margin of the capitulum along nearly the entire length of the ventral margin of the shaft. For a very short distance proximally, the rib shafts arch very slightly dorsally just above the level of the transverse processes before curving strongly ventrally and medially to nearly reach the ventral midline of the trunk; this contributes to the modestly expanded barrel-shaped trunk in Martensius. The distally complete ribs terminate in a cupped margin, indicating a probable short cartilaginous continuation that contacted an unossified sternum.
The rib heads of the lumbars are greatly expanded proximately with a triangular outline, but without any evidence of the dichocephalous condition in the thoracics. In addition, the rib shafts of the lumbars exhibit a serial straightening and shortening posteriorly, with the last two or three becoming straight, or nearly so, and gradually tapering distally to a point. The lumbar ribs in the juvenile MNG 14230 exhibit an early ontogenetic stage of development in being suturally attached to the centra, rather than fused. Worth noting, in the paratype MNG 10595, the posteriormost dozen or so right presacral ribs each possess one or two healing calluses due to severe breaks likely the result of a single traumatic event, perhaps a fall from a low limb while feeding on a plant.
The three dorsally exposed sacral ribs in the juvenile MNG 14230 differ from the ventrally exposed anterior two sacral ribs in the holotype in features that clearly reflect an early ontogenetic stage of development: (1) rather than being equal in size, they decrease serially in size posteriorly; (2) the proximal and distal ends are far less flared; and (3) proximally they are sutured rather than fused to the centrum. Most importantly, however, in both specimens the ribs are independently attached to the ilium.
Description of the anteriormost caudal vertebrae is limited to the laterally exposed first seven in the juvenile MNG 14230. The neural spines, although bladelike as in the anterior column, differ most noticeably in having a constant thickness in horizontal section and lacking the swollen caps of the summits in the holotype. In lateral view, (1) the centra are stout and block-like with the height and length being subequal; (2) the articular rims are vertically parallel to one another; (3) the lateral surfaces exhibit little or no obvious convexity; (4) the articular rims are joined ventrally by a horizontally straight margin of the centrum; and (5) the centrum rims are only very slightly inflected laterally into narrow, unfinished, rounded articular lips. Throughout the preserved portions of the tails in MNG 14230 and the holotype, several serial changes occur posteriorly besides an overall reduction in size that include: (1) a gradual but marked reduction in height and width of the neural spines to spike-like projections accompanied by a gradual increase in posterior inclination; (2) pre- and postzygapophyses become markedly narrower and barely separated, if at all, from one another along the midline; (3) articular facets of the zygapophyses become increasingly tilted steeply ventromedially; (4) lateral and ventral surfaces of the centra become increasingly concave; and (5) the centra gradually become proportionally longer relative to their height. Intercentral chevrons first appear in MNG 14230 at about the fifth caudal and are present nearly the entire preserved length of the tail. Their lengths slightly exceed that of the centrum and proximally they insert between ventrally beveled margins at the contact of adjacent centra. However, in the first eight caudals of the isolated series of 21 of the distal end of the tail in the holotype, the neural arches and chevrons serially diminish in size posteriorly to almost no evidence of their presence, followed by their total absence. In the absence of the neural arches the dorsal surfaces of the centra have a shallow midline channel bordered on either side by a low narrow ridge.
The first seven caudal vertebrae in MNG 14230 possess dorsally exposed holocephalous ribs, which are absent following the gap of the three missing vertebrae. The first four ribs are subequal in length and distally curved strongly posteriorly in a laterally overlapping pattern. The following three ribs are markedly smaller and diminish greatly in size with a progressive straightening of the shafts. In all the ribs, a slight expansion in width differentiates a distally short head from the shaft. They are barely disarticulated from an anteroposterior rugose ridgelike transverse process close to the ventral margin of the centrum; in adults they likely would be fused. Although the rib heads become progressively shorter posteriorly, they continue to be positioned far forward on the centrum adjacent to the rim. The remnants of neurocentral sutures in the anterior half of the caudal series is further evidence of an early ontogenetic stage of development in MNG 14230.
The shoulder girdles in the juvenile MNG 14230 are almost totally obscured by the axial skeleton and of no descriptive value. In the holotype, the only shoulder girdle elements not represented are the clavicles, and the remaining elements, although incomplete and mostly disarticulated, collectively permit description of most essential features. Typically in most primitive tetrapods, the ventrally exposed interclavicle is divisible into a broadly expanded anterior head and a narrow, flattened posterior stem or shaft that extends along the midline. The head has an isosceles triangular outline with shorter anterolateral sides converging to the midline and the longer side forming a transversely straight posterior border interrupted by the interclavicle stem. The area of ventral overlap by the missing clavicles is roughly indicated by a slightly depressed rugose surface that occupies nearly the entire anterior surface of the head except for a narrow, slightly elevated margin of finished bone along the posterior border. The length of the narrow posterior stem is about three times that of the head and has a constant width except for a slight narrowing at its distal end, which is longitudinally scored by a series of short, shallow grooves. In earlier reconstructions the stem projects posteroventrally below the level of the ventral margin of the coracoid plates.
Only the left scapula is well represented and fully exposed laterally. The dorsal half forms the typical broad, relatively thick, rectangular, slightly posteriorly inclined blade. The posterior margin is thickened and rounded, whereas the dorsal margin abruptly ends in a horizontal plane exposing a terminal margin of unfinished bone that suggests at least a short continuation by a cartilaginous suprascapula. Typically in caseids there is no evidence of a supraglenoid foramen. The posteroventral corner of the scapula is incomplete and presumably included a triangular supraglenoid buttress and possibly a small portion of the dorsal margin of the glenoid cavity. There is no deep oval notch in the anterior margin of the scapular blade, which is seen in some genera of Caseids and believed to mark the dorsal extent of the clavicular stem contact. The ventral half of the scapula is broadly expanded anteriorly into a strongly convex margin.
Typically in Caseids there is no evidence of sutural divisions between the scapula and the anterior and posterior coracoids, and the composite coracoid is referred to here as simply the coracoid plate. Both plates occupy, as in life, a near horizontal plane with their external surfaces facing ventrally and the lateral margins contacting the base of the vertically oriented scapula. Although both plates are incompletely preserved, together they indicate an originally anteroposteriorly elongated oval outline with the medial margins paralleling one another and narrowly separated from the midline on either side of the interclavicle stem. Two structures typically present in basal Synapsids are apparently absent: a dorsal process of the posterior coracoid for the triceps muscle, which in Caseids is typically either absent or little developed; and a coracoid foramen of the anterior coracoid near its contact with the posterior coracoid, which is rarely observed in Caseids. Yet, a rough approximation of the division between the two coracoids is possible based on the position of the glenoid cavity, as it is standard for the coracoids to meet in a vertical straight suture directed vertically below the cavity. The lateral margin of the more complete right plate is interrupted by a short, broad, distally incomplete expansion that terminates in a shallow, smooth concave margin that is certainly the ventral rim of the glenoid cavity. Typically in basal Synapsids and presumably also in Martensius the midlength of the glenoid approximates the division of the coracoid plate into two greatly unequal lengths, a short posterior coracoid and an anterior coracoid that is almost twice its length.
A complete left cleithrum is exposed in lateral view that is narrowly disarticulated from the anterior margin of the scapular blade. It closely conforms to those occasionally preserved in all major groups of basal Synapsids, although commonly mistaken for a rib. It is a slender, elongated triangularly splintlike bone that gradually narrows ventrally from a dorsal oval expansion to a sharply pointed ventral tip. Throughout its length it is oval in horizontal section and bowed slightly posteriorly. Along the posterior margin is a narrow cleft that deepens dorsally into which inserted the anterior margins of the scapular blade and likely also a cartilaginous suprascapula extension. A left cleithrum was recently described in Ennatosaurus tecton, which would be the first described in any caseid. However, it appears more likely to be the dorsal half of the clavicle stem, which is separated from its ventral half by an irregular thin crack. The two portions otherwise form a smoothly continuous structure articulated with the full length of the anterior margin of the scapula blade.
To avoid confusion in describing the bones of the fore- and hind limb (defined here as including the manus and pes) the anatomical views used are based on a posture in which the limbs occupy a single horizontal plane with the propodials directed laterally from the girdles and the remainder of the limb directed anteriorly. Therefore, the four primary anatomical views of the propodials are dorsal, ventral, anterior, and posterior, whereas those for the epipodials and bones of the manus and pes are dorsal, ventral, medial, and lateral. Three specimens include essentially complete or partially preserved articulated forelimbs in dorsal view, the juvenile MNG 14230 and adults MNG 13814, and MNG 10595.
The forelimbs in the juvenile MNG 14230 clearly represent an early ontogenetic stage of development. The humeri, however, exhibit the tetrahedral morphology typical of nearly all Permo-Pennsylvanian tetrapods in having greatly expanded proximal and distal heads that are twisted about the long axis of the shaft to occupy nearly right-angle planes to one another. The ectepicondyle and supinator processes are little more than short low rounded ridges separated by a shallow channel, and the posterior flaring of the entepicondyle is greatly reduced, but still possesses a well-defined, elongated entepicondylar foramen closely paralleling the posterior margin. Due to incomplete ossification of the epipodials the terminal epiphyses and articular facets are not fully ossified, surface features are either weakly developed or absent, and distinct shafts are absent. The humeri in MNG10595 are preserved mainly as impressions, but the fidelity of the dorsal aspect of the right bone is sufficient to exhibit clearly the tetrahedral morphology seen in the humerus in the juvenile MNG 14230. A nearly complete humerus is limited to the right bone in the holotype. As preserved, the ‘twist’ of the heads is almost nonexistent, however, due to dorsoventral crushing, and they occupy nearly the same horizontal plane. The only major defect of preservation is a severe horizontal compression fracture paralleling the anterior margin of the proximal head that has resulted in a deep cleft and a slight narrowing of the head. A partially preserved rugose ridge along the anterior margin of the proximal head, presumably the deltoid crest, extends distally nearly to the shaft, where it ends curving a short distance posteriorly. Paralleling the anterior margin of the distal head are two prominent ridges, a more anterior shorter narrow ridge of the supinator process and posterior to it a relatively massive ridge of the ectepicondyle. Both ridges heighten and widen distally despite partial reduction due to weathering. Proximally both structures are very narrowly separated by sharply defined opposing vertical margins that appear to define an ectepicondylar foramen with the loss of the dorsal bridge of bone. If correct, the absence of the foramen in the juvenile MNG 14230 humeri would represent an early ontogenetic feature. The entepicondyle of the holotype consists of a thick, flat plate occupying a large area of the posteromedial corner of the head that widens distally and thickens posteriorly into a rounded facet for the ulna. The entepicondylar foramen and posterior margin of the condyle, however, are obscured from direct view by overlying ribs.
Other than lacking the left humerus and proximal third of the ulna, the forelimbs of the holotype are nearly complete and mostly articulated. Fortunately, the full lengths of both bones in the right forelimb are well preserved except for missing some surface bone. They differ from those of most caseids in having a slender gracile morphology with minimum transverse widths equaling about 11% and 9% of the radius and ulna lengths, respectively. Adding to this perspective, the lengths of the radius (6.8 cm) and ulna (8.0 cm) are elongated, equaling approximately 76% and 89% of the humerus length (9.0 cm), respectively. The radius is subcircular in cross-section, and the heads gradually expand slightly mediolaterally to oval terminal facets. The proximal terminal facet of the radius is slightly concave to receive the convex surface of the radial condyle of the humerus, whereas the dorsal surface of the less expanded distal head is transversely convex, as is the terminal facet for the radiale. The olecranon process and semi-lunar or sigmoid notch of the right ulna are greatly elongated and combined occupy approximately the proximal half of the bone. In the remaining distal portion of the left ulna the shaft abruptly expands into a dorsoventrally flattened head with a strongly convex terminal facet for the intermedium, ulnare, and pisiform.
In the juvenile MNG 14230, only the right manus is nearly complete, but due to an early ontogenetic stage of development the centralia and carpal 5 are unossified, whereas the other carpals are marginally incomplete, which is also true of the articular margins of the digits. Although this has greatly reduced the descriptive value of the manus, it provides data on the sequence of ossification of the carpals. A complete description of the adult manus, however, is possible based collectively on both holotypic forelimbs and the left in the MNG 10595. In both, all 11 of the standard carpals are represented and include the pisiform, radiale, intermedium, ulnare, medial and lateral centrale, and distal carpals 1–5. Confusingly, the identifications of the medial and lateral centrale of the manus were mistakenly reversed in the diagrammatic reconstructions of basal Synapsids made by Alfred Romer and Llewellyn Price in 1940 and Robert Reiz in 1986, but not in their text descriptions. In addition, some authors refer to the lateral and medial centrale as the proximal and distal centrale, respectively.
Description of the carpus relies almost entirely on the well-preserved, complete series in the holotypic left manus. A short, squat radiale expands proximally into a concave terminal facet that tightly articulates with the convex terminal margin of the distal head of the radius, giving their union the appearance of being smoothly continuous. The distal margin of the radiale is divided into two facets: a distomedial facet for the medial centrale, and a distolateral facet for the lateral centrale and intermedium. The intermedium is proximodistally elongated with a slightly concave medial margin that apposes a convex margin at the lateral radius-radiale contact and a lateral angular projection that inserts between diverging margins at the medial ulna-ulnare contact. The distal end of the intermedium is falsely preserved as contacting the proximal dorsal surface of the lateral centrale. Their correct relationship, however, is seen in the right carpus where the contact is transverse. The ulnare, by far the largest of the carpals, is roughly rectangular with a proximodistal length that slightly exceeds the maximum transverse width. A broadly convex proximal margin is divided into three adjacent facets that include slightly convex proximomedial and proximolateral facets for the intermedium and pisiform, respectively, and a transverse facet separating them for the ulna. Distal to its contact with the intermedium a narrow area along the medial margin of the ulnare is slightly depressed with a shallow wide V-shaped emargination that is contacted by a lateral angular margin of the lateral centrale. The distal margin of the ulnare is divided into two greatly unequal adjacent facets: a much longer distomedial facet for distal carpal 4 and a very short distolateral facet for distal carpal 5. A lateral centrale occupies a central position in the carpus and has a distally pointed spade-shaped margin that is wedged between a very short proximomedial margin of distal carpal 4 and a proximolateral margin of the medial centrale. The pisiform is roughly kidney-shaped with a strongly convex free lateral margin and a concave medial margin that spans the diverging margins at the lateral contact of the ulna and ulnare. In both holotypic manus and the left in MNG 10595, the medial centrale has a mediolaterally elongated trapezoidal outline in which a short proximal margin parallels a much longer distal margin. The distal margin is contacted by distal carpals 1–3 except in the left holotypic manus where the articulated distal carpal 1 and first digit have been displaced a short distance proximally and falsely show the distal carpal contacting the otherwise free vertical medial margin of the medial centrale.
The distal carpal series is best preserved in the holotypic left manus, although some features are corroborated in both the right holotypic manus and the right in MNG 10595. Distal carpal 1 is uniquely large and exceeded in size only by distal carpal 4, typically the largest of the series in basal Synapsids. As in the left holotypic manus, the displaced distal carpal 1 has a strongly convex, free medial margin and a slightly convex lateral margin that contacted the medial margin of distal carpal 2. Distal carpals 2 and 3 are accurately preserved in the left manus of MNG 10595 as roughly triangular and narrowing proximally to bluntly pointed contacts with the distal margin of the medial centrale. The outline shapes and contacts of distal carpals 4 and 5 are well preserved in the left holotypic manus. Here distal carpal 4 is roughly quadrangular with medial and dorsolateral margins that converge proximally between distal carpal 3 and the ulnare to a narrow proximomedial margin that contacts the centralia. Typically in basal synapsids distal carpal 5 is extremely small with a proximodistal squat triangular outline in which proximally converging margins insert between diverging margins of distal carpal 4 and the ulnare at their lateral contact.
The digits of both manus (defined here as including the metapodials) in the juvenile MNG 14230 are not fully represented, ossified, or articulated. In the holotype and MNG 10595, they are also not fully preserved but collectively allow an adequate description. Typically in basal synapsids, the lengths of the metacarpals are roughly proportional to the lengths of their respective digits, increasing markedly I–IV with V being slightly shorter distally than IV. This is reflected in the proximal and distal heads of metacarpals I–IV aligning in laterally diverging, distally convex arcs. Whereas the shafts of metacarpals I–V exhibit a lateral serial proportional narrowing, this is contrasted by a progressive expansion of the lateral margin of the proximal heads of II–V that is accompanied by a lateral increase in the transverse width and convexity of the articular facet.
The non-ungual phalanges of digits II–V of the manus serially shorten distally except for the penultimate phalanx, which in apparent contrast to that in other Caseids is unique in being noticeably longer than the antepenultimate phalanx. The proximal and distal heads of the non-ungual phalanges are relatively greatly expanded transversely, the proximal heads slightly more so, and in contrast to the elongated penultimate phalanx are narrowly separated by a deep concave margin that essentially eliminates the presence of a distinct shaft. Medially marginal to the proximal heads is a small, subcylinrical, proximomedially directed process with a cup-shaped or pit-like terminal surface. As in some Caseids, the interphalangeal articulations, except those involving the unguals, are unique among basal Synapsids in not being transversely vertical, but rather strongly inclined anterodorsally in an overlapping contact that creates a larger area of contact. The contact between the penultimate phalanx and the ungual, on the other hand, is strikingly different, but typical of other Caseids. Here the distal articular facet of the penultimate phalanx is a transversely elongated narrow condyle with the terminal margins indented by a shallow cup-shaped depression or pit. The condyle contacts a vertically concave facet of the ungual of the same width but of nearly twice the vertical height to produce a hinge-like joint in which the ungula swung vertically about the penultimate condyle. This surely permitted an exceptionally large angle of vertical rotation of the ungual about the penultimate condyle. The unguals are like those in other caseids in being exceptionally long, massive, strongly curved downward, and distally pointed, as well as possessing a large triangular basal or retractor process. The holotypic right manus and left in MNG 10595 clearly indicate a phalangeal formula that differs from those of other Caseids in which it is well documented by retaining the primitive count of 2-3-4-5-3.
The pelves in the juvenile MNG 14230 are undoubtedly complete, but exposure is limited mainly to the medial surface of the right iliac blade due to the in place sacral vertebrae and ribs. The anterior and posterior processes of the ilium extend only a short distance beyond the levels of the horizontal limits of the acetabulum. Both have the form of a broadly convex flange, although the posterior process is slightly narrower and longer. In the holotype MNG 13814, only the ilium of the right pelvis can be accounted for, but unfortunately is portioned into two exposures due to a horizontal break through the matrix block containing it. Almost the entire dorsal iliac blade is exposed medially, whereas a laterally exposed ventral portion includes a dorsal neck-like constriction below which there is an expansion that includes the upper portion of the acetabulum. The anterior process of the iliac blade differs from that in MNG 14230 only in being proportionally slightly longer, but it also terminates in a broadly convex margin. Although the posterior process is missing a narrow portion along its ventral margin, it is included in the laterally exposed ventral portion of the ilium, and both portions allow a diagrammatic outline reconstruction of the ilium in lateral view. The posterior process not only differs markedly from the anterior process, but also those of most other Caseids except possibly Casea, in being greatly elongated well beyond the acetabulum and narrowing posteriorly to a blunt point.
The puboischiadic plates in the holotype MNG 13814 are united along most of their midline margins and as preserved occupy a horizontal plane due to dorsoventral crushing, exposing severely eroded dorsal or internal surfaces. The margins of the plates clearly indicate an anteroposteriorly elongated oval outline. In life they would be inclined dorsolaterally from the midline in a deep V-shaped trough and dorsally united with the vertically oriented ilia. On the lateral margins of the plates are poorly preserved, short, broad, distally incomplete expansions that presumably are the basal portions of the ilia but without any evidence of the ventral suture. Furthermore, in neither plate is there a suture indicating the union of the pubis and ischium, nor can the pubis be identified by either the presence of a pubic tubercle or an obturator foramen of the pubis typically located near its contact with the ischium. The absence of pelvic sutures is apparently typical of Caseids. However, the pubes and ischium portions can be roughly identified on the basis on the relative position of the ilium, as its ventral midlengdth typically straddles the vertical sutural divison separating the pubis and ischium, allowing their identification on the basis of contrasting features present in almost all basal Synapsids: (1) the pubes are shorter and their midline contact extends just short of their anterior margins; and (2) the ischia are longer and diverge from their midline contact far short of their posterior margins. An adult feature of the puboischiadic plates is the absence of a diamond-shaped opening at the midline intersection of the paired pubes and ischia that commonly occurs in juvenile basal Synapsids, which is attributed to a slower ossification of the endochrondral bones of the pelvis in this region.
Both femora are preserved in the juvenile MNG 14230, and although only the left is essentially complete, an early ontogenetic stage of ossification has eliminated most adult structures. The only other preserved femur is a right exposed posteriorly in the adult MNG 10595, which is distinct from those of the majority of Caseids in its slender gracile morphology with only modestly expanded heads. Although the proximal head is marginally incomplete, it is well-enough preserved to indicate that it terminated in a broadly convex articular facet. Only the larger posterior of the paired distal condyles is visible, which terminates in a slightly convex ventrolateral margin. A narrow, coarsely textured facet on the posterior margin of the condyle would have been contacted by the fibula. A short distance proximal to the posterior condyle is a slight expansion on the ventral margin of the shaft that may be either a poorly developed adductor ridge or fourth trochanter.
The epipodials are best represented by the left bones in the juvenile MNG 14230 and the right in MNG 10595 all of which are dorsally exposed except for the right tibia in the latter being ventrally exposed. They have a slender gracile appearance with narrow subcylndrical shafts that extend between gradually widening flattened heads. The minimum transverse widths of the tibia and fibula shafts in the juvenile MNG 14230 are equal to about 15% and 13% of the bone lengths, respectively, whereas the same measurements in the adult MNG 10595 are approximately12.5% and 10%. The opposing margins of the epipodials are strongly concave, whereas the outer margins are only slightly convex. This, in addition to the shafts gradually narrowing to their midlength, gives the epipodials the appearance of being bowed away from one another. The proximal head of the left tibia in MNG 14230, although incompletely ossified, is greatly expanded compared to the distal head, and the terminal margin is clearly divided into two slightly concave facets for the distal head of the femur: a medial dorsoventrally elongate narrow facet for the anterior condyle of the femur and a lateral mediolaterally elongate narrow facet for the posterior condyle. A poorly developed cnemial crest extends a short distance distally from the dorsal margin of the medial facet. In MNG 14230 and 10595 the transversely narrow oval terminal facet of the distal head contacted the astragalus. Ventral view of the right tibia in MNG 10595 exposes a prominent ridge that extends distally from the proximal mid-width of the shaft to the lateral margin of the distal head. The dorsally exposed left fibula in MNG 14230 and the right in MNG 10595 possess well-defined proximal and distal heads. The articular margin of the proximal head for the femur is slightly convex and faces strongly proximomedially, whereas the articular margin of the distal head is divided into two adjacent facets, one facing distomedially for the astragalus and the other distolaterally for the calcaneum.
Description of the dorsally exposed tarsals in the juvenile MNG 14230 are severely limited by their early incomplete ontogenetic ossification, which presumably also accounts for the absence of the single centrale and distal tarsal 5. The greater portion of the right tarsus in MNG 10595 is preserved but ventrally exposed. Fortunately, the holotypic left pes is superbly preserved in dorsal view except for a badly distorted slightly displaced astragalus and includes all the standard tarsal bones of basal Synapsids, which in addition to the astragalus include the calcaneum, a single centrale (presumably the lateral centrale has expanded into the space formerly occupied by the medial centrale), and distal tarsals 1–5, as well as complete digits. The astragalus, however, is well preserved in ventral view in the right pes of MNG 10595, which is greatly thickened throughout and has the standard reversed L-shaped outline. The narrower rectangular vertical limb is about twice the length of the stouter horizontal limb when measured along the internal margins of their union. The lateral margin of the astragalus would have contacted the medial margin of the calcaneum in a nearly straight suture except for being interrupted a short distance from its distal extent by a well-defined U-shaped notch for the perforating artery. The proximal margin of the vertical limb slopes distolaterally for the medial facet of the fibula, whereas the lateral margin of the horizontal limb is occupied by a transversely rounded condyle-like articular facet for the distal end of the tibia. The transversely flat distal margin of the astragalus contacts the full width of the combined proximal margin of the centrale and the proximomedial corner of distal tarsal 4. The calcaneum in the holotypic left pes has essentially a proximodistal elongate rectangular outline except for the lateral margin gradually flaring proximally into a broadly convex margin. The calcaneum is relatively thin throughout except for a pronounced thickening along its contact margins. The proximal margin of the calcaneum slopes slightly proximomedially for the lateral facet of the fibula, whereas distally it narrows to a transversely straight margin that is contacted by distal tarsals 4 and 5.
Typically in basal Synapsids, many aspects of the descriptions for the distal carpals apply nearly as well to the distal tarsals. The full widths of the distal margins of the five distal tarsals are restricted in contact to their respective metatarsals. Although the holotypic distal tarsals 1–3 and their respective digits have shifted a short distance distally within the pes, the original contacts of the tarsals for the most part can be safely interpreted based collectively on the holotype and MNG 10595. Distal tarsal 1 is unusually large and exceeded in size only by 4, which is typically the largest of the series in basal synapsids. Distal tarsal 1 is most accurately preserved in MNG 10595, where it extends proximally and slightly laterally from its metatarsal contact, narrowing slightly between convex margins to a bluntly convex distal margin that contacts the centrale. Whereas the medial margin is free of contact, the lateral margin contacts a slightly concave margin of distal tarsal 2. The holotype shows distal tarsals 2 and 3 as having triangular outlines, narrowing proximally to pointed contacts with the distal margin of the centrale. The greatly enlarged distal tarsal 4 is best exemplified in the holotype, where it has a pentagonal outline of contacts, which includes in clockwise succession: a proximomedial margin for the astragalus; a proximolateral margin for the calcaneum; a lateral margin for distal tarsal 5; a distolateral margin for the proximal head of metatarsal IV; and a long distomedial margin for distal tarsal 3 and the centrale. The transversely expanded subrectangular centrale laterally contacts the proximal half of the medial margin of distal tarsal 4 and distally the proximal margins of distal tarsals 1–3; its medial margin is free of contact.
Most features of the metacarpals apply equally well to the metatarsals: (1) their lengths are roughly proportional to the lengths of their respective digits; (2) II–V increase in length with V being slightly shorter distally than IV; (2) II–V exhibit progressively more slender proportions, which when calculated as a minimum width to maximum length percentage yield decreasing values of 26 to 16%; (3) metatarsal I differs from II–V in being relatively far more massive with a minimum width slightly greater than half its length and lacking a distinct shaft; and (4) the proximal and distal heads of I–IV align in laterally diverging distally convex arcs except for V being distally shorter than IV. The metapodials of the manus and pes in species of Cotylorhyncus and, in the absence of a pes, the manus in Euromycter rutenus differ from those of other Caseids, as well as basal Synapsids generally, in being short and massive and lacking a distinct shaft, but with a modest gradual serial lengthening of I–V from subequal to the non-ungual phalanges of digits I–IV to nearly doubling in length.
The pes phalanges of Martensius also share several important features with those of the manus, among which include: (1) in digits II–V the penultimate phalanx is much longer than the antepenultimate phalanx; (2) non-ungual phalanges possess a short subcylinrical process that extends proximomedially from the medial margin of the proximal head and terminates in a cup-shaped or pit-like depression; (3) inter-phalangeal articulations, except that involving the ungual, overlap in an anterodorsal inclined plane; and (4) unguals are exceptionally long, massive, strongly curved, narrow distally to a point, and possess a large triangular basal or retractor process.
In Martensius the pes phalanges proximal to the penultimate phalanx of digits II–V differ strikingly from those of the manus in being relatively much shorter, more massive, and lacking a distinct shaft. The same features characterise the non-ungual phalanges in the manus and pes of Cotylorhyncus species and the right manus of Euromycter rutenus. Generally, in other basal Synapsids, including the Caseid Alierasaurus ronchii, the non-ungual phalanges serially shorten distally. However, in Martensius and in contrast to other Caseids, as well as all basal Synapsids, the penultimate phalanges of digits II–V in the manus and pes are unique in being much longer than the antepenultimate phalanges. Also, the penultimate phalanx of digit I of the pes is unique in being not only much larger than those in digits II–V, but also longer than metatarsal I. The phalangeal formula of the Martensius pes differs from those of all other caseids in which it is well document in possessing the primitive count of 2-3-4-5-4. In the original description of Eocasea martini the claim was made also that the pes exhibits the primitive phalangeal count. However, judging from the illustration of the pes. this is uncertain, as digits II and IV are shown distally incomplete. Also questionable is the acceptance of Samuel Wendell Williston's (1910) description of the pes in Casea broilii as possessing the primitive phalangeal formula. According to Everett Olson there is no reliable substantiating evidence to support Williston’s description, and in fact the illustration of the pes by Alfred Romer and Llewelin Price and duplicated by Rober Reisz shows only digit I as complete and almost all the phalanges of digits II–V as lost.
Gastralia (also referred to as abdominal or ventral ribs) are preserved only in MNG 10595, and for the most part exhibit an individual structure and collective organization like that in other basal Synapsids. Although the individual elements are imperfectly preserved as bone and impression on part and counterpart surfaces, they clearly extended almost the entire length of the ventral surface of the trunk. The individual elements are essentially identical as slender, toothpick-shaped rods pointed at both ends that are approximately 1.0 mm thick and 1.0 cm long. They are arranged singly in tightly packed chevron rows with the apices directed anteriorly on the midline and overlapping at their ends. The ends of each element are overlapped by those of the same row. The paired elements of the apex are apparently not fused into a single V-shaped median bone as described in other basal Synapsids. This is the only well-documented example of the gastralia in a Caseid, although numerous individual elements were noted in a specimen of Cotylorhynchus romeri that match those described here in Martensius bromackerensis except for being much larger.
The skeletal and whole-body restorations of Martensius bromackerensis in lateral view are feasible based on the superb preservation of the essentially complete holotype, with only some additional information provided by the paratype MNG 10595. Skeletal restorations of Caseids in lateral view are limited to two species, the late Early Permian Casea broilii and the Late to Middle Permian Cotylorhynchus romeri. In overall appearance Martensius bromackerensis is most comparable to Casea broilii.
Immediately noticeable in the restorations of Martensius bromackerensis is the ridiculously small skull, a feature that is not only characteristic of other Caseids, but also of other herbivorous basal Synapsids. Also differing from the ordinary are the proportionally large feet. The modestly expanded barrel-shaped trunk of Martensius bromackerensis is only slightly less expanded than that in the late Early Permian Casea broilii, but greatly smaller compared to its huge build in the later occurring Permian Caseids, which is dramatically portrayed in the skeletal restoration of Cotylorhynchus romeri. Unexpected for a herbivore, the manus and pes of Martensius bromackerensis possess unusually large, strongly recurved, sharply pointed unguals. Also noteworthy, in contrast to most Caseids the plane of contact between the non-ungual phalanges is strongly inclined anterodorsally, rather than being oriented in the standard, transverse vertical plane. Uniquely, the penultimate phalanx in digits II–V is longer than the antepenultimate phalanx.
A phylogenetic analysis was performed using a character-taxon matrix developed by Neil Brocklehurst and Jörg Fröbisch. Alierasaurus ronchii was included in the matrix based on the scores provided by Marco Romano, Neil Brocklehurst, and Jörg Fröbischin 2018. Ruthenosaurus russellorum was not included in the current analysis because the holotype and only known specimen lacks a skull, which precludes it being resolved in the current analyses. On similar grounds Arisierpetom simplex was eliminated from the analysis not only because it lacks a skull, but otherwise consists only of isolated toothbearing jaw elements, including a premaxilla designated as the holotype Similarly, Phreatophasma aenigmaticum was excluded due to its lack of completeness. Brocklehurst and Fröbisch identified Ctenorhachis jacksoni, Caseopsis agilis, and Angelosaurus dolani as wild card taxa, and they were thus deleted from their subsequent analyses. This deletion is followed here. Callibrachion gaudryi and Datheosaurus macrourus were also excluded from the Berman et al.'s analysis, based on the conclusion that poor preservation of the specimens makes their placement within Caseasauria uncertain. The resulting analysis, therefore, contained 56 taxa (including four outgroup taxa) and 245 characters. Martensius bromackerensis was scored for 64% of the characters. The taxon-character matrix is available here.
A maximum parsimony analysis was conducted in Paup* 4.0b10 for Macintosh All characters were unordered and equally weighted. Multi-state taxa were treated as polymorphic and gaps were treated as missing data. Outgroups were the same those used by Roger Benson in his 2012 study of basal Synapsids, which included Limnoscelis, Protorothyris, Tseajaia, and Captorhinus. The tree-bisection-reconnection branch swapping algorithm was employed under a heuristic search option with 100 random addition replicates.
This analysis yielded three islands of optimal trees. Island one was landed upon 80 times out of 100 replicates and contained the shortest trees with 818 steps. This most optimal island contained 9,072 most parsimonious trees. The strict consensus and 50% majority rule trees resemble traditional arrangements of the larger ‘Pelycosaur’ groups, where Caseasauria forms a monophyletic clade that is the sister taxon to the rest of Synapsida. Varanopidae is the next group to diverge, followed by Ophiacodontidae plus Echinerpeton intermedium, Edaphosauridae, and Sphenacodontidae plus Therapsida.
Within the caseasaur portion of the tree a dichotomy is formed between a monophyletic Eothyrididae including Eothyris parkeyi, Oedaleops campi, and Vaughnictis smithae, and a monophyletic Caseidae. Within Caseidae, both the strict consensus tree and 50% majority- rule consensus tree yielded Eocasea martini as the basalmost branching taxon, followed by Martensius bromackerensis. In the strict consensus tree, the remaining Caseid taxa are poorly resolved. In the 50% majority rule consensus tree, Casea broilii branches off one node more crownward than Martensius bromackerensis, and is followed by a polytomy between Oromycter dolesorum, Trichasaurus texensis, and a clade of the remaining caseid taxa. Within those remaining taxa, a sequence of branching taxa begins with Casea nicholsi, followed by Euromycter rutenus, then Ennatosaurus tecton, then Angelosaurus romeri, and then an apical clade of the three Cotylorhynchus species plus Alierasaurus ronchii. In this final clade Cotylorhynchus hancocki, and Cotylorhynchus bransoni are sister taxa, and the clade forms a polytomy with Cotylorhynchus romeri and Alierasaurus ronchii.
Berman et al. (2020) report on a new caseid taxon, Martensius bromackerensis, based on four partial to nearly complete, mostly articulated, and well-preserved skeletons from the middle Early Permian Artinskian Stage of Germany. Collectively, the Martensius bromackerensis specimens present a nearly complete knowledge of the skeletal morphology. Significantly, Martensius bromackerensis occupies not only a temporal but also a phylogenetic position that link it with the Late Pennsylvanian Eocasea martini, considered the most primitive and oldest known Caseid, and the numerous crown-clade Caseids of the late Early and Middle Permian.
Until recently, all caseids were characterised as medium to large bodied, high-fiber herbivores, as evidenced by the distinctive features of a greatly expanded barrel-shaped trunk and leaf- or spatulate-shaped cuspidate teeth. This characterization, however, was challenged by Robert Reisz and Jörg Fröbisch in their description of Eocasea martini, who noted that it lacks the hallmark caseid features of herbivory and instead has an unexpanded rib cage and a dentition of extremely small simple conical teeth suggestive of a non-herbivorous Insectivorous diet. As recognised by Reisz and Fröbisch, the dentition of Eocasea demonstrates for the first time that an exclusively herbivorous diet evolved within Caseidae. In contrast, Martensius bromackerensis not only possesses a modestly expanded barrel-shaped trunk, but as argued below a dentition in which a juvenile insectivorous dentition was ontogenetically replaced in the adult by a dentition suggestive of an omnivorous diet.
Critical to this discussion is being able to establish with confidence that the adult specimen of Martensius bromackerensis was herbivorous, although likely continuing to be in small part Insectivorous. It was recognized by Romer and Price that among basal Synapsids the skulls of herbivores are proportionately much shorter than those of carnivores when compared to their snout-vent lengths. This correlation is accurately illustrated in the numerous skeletal reconstructions of basal Synapsids presented by Romer and Price. In species of the herbivorous Edaphosaurus, skull length ranges from about 12 to 14% of the snout-vent length, and in the recently described Edaphosaurid Gordodon kraineri it is roughly estimated at 18%, although a large posterior portion of the trunk is absent. In the two Caseids in which this measurement is available, Cotylorhyncus romeri and Casea broilii, it is approximately12.5% and 15.0%, respectively. In marked contrast the same measurement in the carnivorous species ranges roughly from 24 to 35% or even higher, as in Ophiacodon, at about 45%. The same proportional calculation that distinguishes between carnivorous and herbivorous basal Synapsids can be used also to speculate on the diet of Martensius bromackerensis. Here the range of lengths of the presacral vertebral column and skull are available only in the juvenile and smallest specimen MNG 14230 and the adult holotype and largest specimen MNG 13814. Although the cervicals in the juvenile are either poorly preserved or absent, the original presacral vertebral column length can be roughly estimated at 30.0 cm, whereas the skull length can be safely estimated at 5.2 cm using the nearly complete left mandible as a guide. On the other hand, in the adult holotype the presacral vertebral column is essentially complete with a length of 42.0 cm, and the skull, although severely crushed, has a length safely estimated at 7.2 cm based on the length of the well-preserved right mandible. Despite a marked difference in their snout-vent lengths of 35.2 and 49.2 cm, the percentage contributions of the skulls to the snout-vent lengths are essentially identical at about 15%, suggesting an herbivorous diet. Despite identical proportional divisions of the snout-vent lengths of the skeleton, the juvenile and adult specimens are clearly distinguishable by marked differences in their ontogenetic stages of development, which includes overall skeletal ossification, dental morphology, and proportional dimensions of skeletal elements. The holotype of Eocasea martini, although obviously a juvenile, was estimated by Reisz and Fröbisch to have a snout-vent length of 12.5 cm, which even as an adult would have been dramatically smaller than the juvenile specimen (MNG 14230) of Martensius. Based on their illustration of the skeleton, the skull occupied approximately 24% of the snout-vent length, which agrees with their assessment of the its dentition as insectivorous.
In addition to the skull of the adult specimen of Martensius bromakerensis being proportionally small, it also possesses features that are unique to the crown-clade Caseids, although not as fully developed, which are considered specialisations to an herbivorous diet: (1) relatively short snout; (2) weakly developed procumbent snout; and (3) upper jaw dentition nearly isodont, but completely so in the mandible. A possible indication of an herbivorous diet by the adult Martensius are features of its feet uniquely shared with the few Caseids in which they are known: they are exceptionally large, well developed with massive, elongated, strongly recurved bony claw supports of the unguals, and a large triangular basal or retractor process.
It has been speculated that the disproportionally large and powerful limbs of caseids, especially the bony claw supports, indicate not only a capability for digging, but also considerable digging in life apparently for plant food. This behavior is supported also in Martensius by the described cup-shaped pits of its non-ungual phalanges of the feet, which most likely served as insertion sites of tendons of powerful digital flexor muscles. The same function can be attributed to the exceptionally large basal or retractor processes of the unguals. An alternative specialized behavior of foraging by Martensius is suggested by its possession of three large massive sacral vertebrae with large ribs that are greatly expanded distally and independently contact the ilium. This may have provided the necessary sacrallumbar support to assume a semi-upright posture, allowing it to reach up with its forelimbs to forage on large plants by tearing down their higher branches.
Of major significance the marginal dentition in the juvenile specimen of Martensius bromackerensis, which consists of two series of distinctly different tooth morphologies. The larger upper marginal teeth, although serially incomplete posteriorly, are identical to those of the adult in being narrowly triangular and slightly recurved. In marked contrast, the teeth of the mandible, although represented only by their bases, are extremely small, cylindrical or tubular, and more numerous, 31 versus 25 in the adult. It is speculated that if the mandibular teeth were intact they would closely resemble the simple conical teeth reported by Reisz and Fröbisch in Eocasea. The occurrence of contrasting dentitions in the juvenile fosters the interpretation that preservation occurred during the transitional stage of the ontogenetic replacement of the juvenile dentition by that of the adult, which implies a shift in diet from insectivorous to omnivorous. The sequence of dental trait acquisition in Martensius also suggests that gut processing of vegetation preceded oral processing in the evolution of Caseid herbivory. Most importantly, however, a juvenile insectivorous diet would have been critical in providing the opportunity for successful introduction into the gut of micro-organisms capable of endosymbiotic cellulysis. It has been suggested that it is 'plausible that ingested Insects, especially herbivorous forms that house such micro-organisms in their own guts, provided the original source for fermentative endosymbionts.' It is noteworthy that the Permo-Pennsylvanian Diadectids Diadectes and Desmatodon also exhibit an ontogenetic change in dentition, which suggests a preference shift in diets from possibly omnivorous or insectivorous in the juveniles to strictly herbivorous in the adults.
Berman et al. note, the poor ossification of the Eocasea skeleton is obviously due to its very early juvenile growth stage in which many of the limb elements are either incompletely ossified or, as in the tarsals, not ossified at all. Therefore, the ontogeny-dependent pattern of tooth morphology observed in Martensius raises the possibility that Eocasea also had a similar replacement in dentition in which insectivorous teeth of the juvenile were replaced in the adult by teeth suggestive of an omnivorous diet. This would influence interpretations of when herbivory first evolved in caseids and reopen the possibility that the common ancestor of caseids had an herbivorous or omnivorous adult dentition. However, without adult material of Eocasea this interpretation is highly speculative.
Although an ontogenetic shift in dentitions correlated to a dietary change in Martensius has not been documented in other caseids, it does suggests a likely three-stage chronology of evolutionary changes in the dentitions and associated diets of caseids: the initial stage of dentition was represented by the strictly insectivorous type in the Late Pennsylvanian Eocasea; this was followed by a dentition like that attributed here to the middle Early Permian Martensius in which there was an ontogenetic replacement of a juvenile insectivorous dentition by an adult omnivorous dentition; and lastly would be the highly modified dentition of the late Early and Middle Permian caseids adapted for an exclusively high-fiber diet. The last stage would explain the dramatic speciation and global dispersal of the late Early and Middle Permian caseids, as it availed them to high-fiber plants, which during their appearance became a widespread abundant food source.
See also...
Online courses in Palaeontology.
Follow Sciency Thoughts on Facebook.
Follow Sciency Thoughts on Twitter.
