4.1.1. General features

The skeleton of the lectotype NMP M470 is disarticulated but the various elements have been only slightly displaced. The skull is represented by the underside of the greater part of the skull roof, with one detached bone in the orbit (Fig. 3). Unfortunately, damage to the narial region means that the shape of the external naris cannot be determined. The palate is largely missing but both mandibles are crushed together superimposed on the left cheek region. The orbits are relatively closely spaced, much more so than the slightly larger specimen of Mordex. Immediately behind the skull are several elements of the pectoral girdle together with a humerus and numerous neural arches and ribs. The counterpart posterior slabs bear more neural arches and trunk ribs, some forelimb elements, the right half of the pelvis and many elements of the hind limbs. Dorsal and gastral dermal scales are scattered throughout the slabs and are still in associated sheets around the posterior trunk region, as described by Milner & Sequeira Reference Milner and Sequeira2003. NMP M472 is a completely disarticulated and scattered specimen. It was described fully by Milner & Sequeira (Reference Milner and Sequeira2003). The description of the palate is summarised here as it includes information not found in any other specimen.

4.1.2. Skull (Fig. 3)

The skull of NMP M470 was about 33mm in midline length. The bones are described as they appear in ventral aspect. An impression of some of the sculpture on the dorsal surface of the right supratemporal is preserved and shows honeycomb-type pitting, much more robust than in the larger skull of M. calliprepes. There are squarish nasals, elongate frontals showing a slight expansion at the anterior end, and narrow prefrontals which extend back to meet the postfrontals, excluding the frontals from the orbit margin. The postfrontals have narrow anterior extensions and a relatively expanded ovoid posterior region. The parietals are large rhomboidal elements, narrower anteriorly next to the interorbital region. The pineal foramen is relatively large and situated just anterior to the middle of the interparietal suture. There are no intertemporals, and the supratemporals are large square elements, extending posteriorly behind the level of the parietals. The postparietals are wide rectangular bones, about half the area of the parietals and narrowing laterally where they extend behind the supratemporals. Each postparietal bears an occipital flange bordering the foramen magnum. The left tabular is present though slightly detached. It is a small element separated widely from the parietal. It bears a tiny posteriorly directed process on its posterolateral corner. The lateral edge overlapping the tympanic embayment region has a convex expansion. The left postorbital is a triangular bone, wider than long and apparently only slightly wedged between the supratemporal and squamosal. Part of a large left squamosal is visible. It appears to have a concave posterior edge forming the anterior margin of a large tympanic embayment but this region is damaged. In the left orbit is a bone which appears to be the right premaxilla with an attached nasal fragment. It has a curved row of tooth bases indicative of its identity. It is evidence of a robust and distinctive premaxilla such as is also present in Mordex (Fig. 1).

Little of the palate is clearly visible in NMP M470 but there are several fragments of densely denticulate bone wedged between the two mandibles. These are probably pieces of the palatine and pterygoid, or possibly ectopterygoid. They demonstrate that the lateral palatal bones bear a dense covering of denticles, also seen in Mordex. Though comprising damaged and disarticulated material, NMP M472 is a useful source of information about the palate, as vomers, a pterygoid and the parasphenoid are present. The vomers possess smooth concave edges which appear to be the anterior margins of the interpterygoid vacuities. No vomerine fangs are visible but a dense covering of denticles is present. The left pterygoid is relatively complete. The ventral face of the central region and the palatine ramus are covered in a dense sheet of denticles. The palatine ramus widens from an anterior point suggesting that the palatine narrowed posteriorly and that the ectopterygoid may have been narrow. The quadrate ramus is broad but may have been crushed from a vertical to a horizontal plane. The basipterygoid process is broad but the surface for articulation with the basisphenoid is obscured. The parasphenoid is largely visible in ventral aspect. The broadly rectangular basal plate bears a sharply defined triangle of denticles at its junction with the cultriform process. Carotid foramina with elongate entry grooves are present on either side of the denticle patch. The cultriform process is relatively slender. Anteriorly it is damaged but appears to have had a single terminal expansion.

4.1.3. Mandible (Fig. 3)

In NMP M470, parts of the middle and posterior regions of both mandibles are present. The tooth-bearing strip of bone overlying the outer edge of the left orbit appears to be the right dentary together with part of a denticle-bearing coronoid. The preserved fragment of dentary has space for about 36 teeth. The tooth bases and positions suggest that the dentary teeth were small, equal in size and closely spaced. Medial to the tooth row is a distinct shelf, apparently the medial surface of the dentary on which the coronoids would have been situated. At the anterior end is a row of denticles, distinctly smaller than the main tooth row, and this is presumably a row of coronoid denticles. The left mandible is represented by the entire posterior region which is severely crushed. The angular is visible in ventral aspect and can be recognised by its dermal ornament in the form of radiating striations. Behind it is a mould of the articular, visible as its bilobed condylar surface.

4.1.4. Axial skeleton

The axial skeleton of NMP M470 is disarticulated but not much scattered. One articulated series of four trunk vertebrae are 13.8mm in total length. The distance from the back of the skull to the ilium is 74mm, which suggests a presacral column of about 22 vertebrae. The neural arches are present as separately ossified halves with very low neural spines and distinct anterior and posterior zygapophyses. The middle trunk series bear prominent rib articulations. In the same middle trunk region, at least two isolated intercentra are present. They have been dorsoventrally flattened and appear as small ovals of bone, pointed at both ends and with a medial suture, implying that the intercentrum has recently ossified from two separate halves and that the animal was a post-metamorphic juvenile when it died. The trunk ribs are relatively small and straight or slightly curved. The anterior trunk ribs, represented by a clump of ribs just behind the skull, are larger with expanded heads. One or two pairs situated next to the pectoral girdle have slightly expanded distal shafts. No uncinate processes are visible. The posterior trunk ribs are shorter, thinner and unossified distally. Close to the pelvis are four caudal haemal arches, each comprising a pair of rod-like bones fused to give an articulating surface proximally and terminating together as a fused point distally.

4.1.5. Appendicular skeleton (Fig. 3)

The pectoral girdle of NMP M470 is represented by the partly hidden interclavicle, both clavicles, and both scapulocoracoids, also each partly hidden. No cleithra could be recognised although it is possible that the cleithral shafts might be present, unrecognised, among the anterior trunk ribs. The interclavicle is a small rhomboid, about as long as wide and fully ossified. It is visible in internal aspect and the nature of its dermal ornament is unknown. The clavicles are long structures with narrow blades and the articulated dermal pectoral girdle would have been a narrow strut-like structure rather than a large plate-like structure. The scapulocoracoids are fully ossified, an unusual feature in such a small temnospondyl. Each scapulocoracoid is a flat element with a broad crescent-shape or kidney shape (Fig. 3). There is a thickened strut extending down the posterior concave edge and a foramen on the edge of the posteroventral extension of the coracoid. The only anterior limb elements which can be identified are the humeri and a radius. The left humerus, close to the skull, is a slender element comparable in proportions to those of branchiosaurids and dissorophids. Neither end is greatly expanded and no supinator process is visible. In this, it resembles that of Mordex rather than those of more derived trematopids. The right humerus, partway down the trunk is similar and is accompanied by a slender radius.

The lateral face of the right side of the pelvic girdle is visible (Milner & Sequeira Reference Milner and Sequeira2003, fig. 3). The ilium, ischium and pubis are sutured together, an unusual feature in such a small temnospondyl. The ilium bears a large acetabulum and a relatively stout, posterodorsally directed, dorsal shaft. The dorsal shaft is simple with only a slight suggestion of dorsal expansion. The pubis is represented by a poorly defined patch of bone, as if it were just in the process of ossification. The ischium is fully ossified and is a large slightly curved plate of bone extending well behind the ilium. It appears to have extended forwards under the pubis which is wedged between it and ilium so that much of the area immediately below the acetabulum is occupied by the pubis. The visible femur is relatively large but poorly preserved. Both sets of tibiae and fibulae are present. The tibia is the larger element, relatively straight with a highly expanded proximal end and a less expanded distal end. The smaller fibula is slightly bowed and is expanded principally at the distal end. Numerous pedal phalanges are scattered around the posterior part of the specimen.

4.1.6. Scales

Large numbers of dispersed and clumped scales are associated with NMP M470. There appear to have been both dorsal scales and ventral gastralia present. The dorsal scales are very small circular structures while the gastralia are larger ovoid scales with a thickening along one edge.

4.2.1. Skull roof (Figs 4, 5a, b, 5d–f, 6)

The large skull, NMW 1898-X-47, probably belonged to an adult as the coarse reticulate pitting shows no evidence of ‘stretching' except on the anterolateral edge of the squamosal and the sutures are poorly expressed. Their position may be deduced from the ornament patterns on the bones and the furrows between ornamented areas but the actual sutures are visible only in some regions and have had to be depicted in oversimplified form in Figure 4. There is no evidence of lateral-line sulci.

Very little can be seen of the marginal tooth-bearing series of bones. Only the right premaxillary is present as a very crushed, poorly preserved structure. An alary process appears to be present as in the type skull. An isolated premaxilla NMW 1983-32-81 (Fig. 5d) may belong to this taxon. It is about 22mm along the jaw margin with a very long alary process and heavily pitted dermal ornament. The ornament identifies it as temnospondyl and the only temnospondyls from Nýřany which were large enough to possess such a premaxilla are Capetus, Mattauschia and possibly Mordex (the premaxilla of Cochleosaurus is very different in shape). The premaxilla of Capetus has finer pitting along the jaw margin and a shorter alary process (Sequeira & Milner Reference Sequeira and Milner1993, text-fig. 4) and this specimen can be deduced to be dissorophoid by elimination. A reason for positive attribution to Mattauschia is the smooth curve on the right side as preserved corresponding to the margin of the large oval external naris. A long alary process partly bordering the naris can be seen in Fedexia (Berman et al. Reference Berman, Henrici, Brezinski and Kollar2010, figs 4b, 5b). The middle section of the left maxilla is present in NMW 1898-X-47 and bears heavy ornament. The anterior region is missing and the posterior region is obscured by the left jugal. The middle region does not form part of the orbit margin, being excluded by the palatine. No premaxillary or maxillary teeth are preserved.

The external naris is a very large ovoid fenestra of a shape and size otherwise found only in Fedexia (Berman et al. Reference Berman, Henrici, Brezinski and Kollar2010, figs 4–7) and in indeterminate juvenile trematopids (e.g., MCZ 2531) from the Texas Permian (Dilkes Reference Dilkes1993, text-fig. 4). Mordex, Ecolsonia and more derived trematopids share an elongate but relatively narrow external naris with a slight waisting produced by a convexity on the distal margin of the nasal. In this specimen however, the external naris is much wider with a smoothly convex medial margin formed by a smoothly concave distal edge to the nasal. The nasal (Fig. 5a) is unusually slender for a temnospondyl, broadening slightly posteriorly. It does not resemble that of any other dissorophoid and appears to be a unique feature of Mattauschia. Even the nasals of Fedexia are broader relative to their length. The prefrontal is a typical triangular bone bordering the orbit and with a deep rim along its orbital edge. It is relatively large and primitively contacts the postfrontal posteriorly. Because the external naris is unusually large, it is bordered by the prefrontal that has a shallowly concave anterodistal face as a result. The lacrimal (Fig. 5b) is even more extremely modified in relation to the external naris and resembles that in Fedexia, Ecolsonia and some trematopids. Most of the enlargement of the naris has been at the expense of the anterior region of the lacrimal which is a compact rhomboidal bone wedged between naris, orbit and maxilla. It extends a slender sliver of bone anteriorly, dorsal to the maxillary, along the distal margin of the naris. The main sculptured dermal surface is triangular but posteriorly there is a wedge of smooth bone within the orbit margin between palatine and prefrontal. This wedge bears the two posterior openings of the lacrimal duct. Behind the lacrimal, forming the distal border of the orbit is a dorsal exposure of the palatine (Fig. 5b). It bears some dermal sculpture and forms the orbit border between the lacrimal and the jugal, thus conforming to the typical large dissorophoid condition (Bolt Reference Bolt1974b). The broad smooth medial face of the dorsal extension of the palatine forms a major part of the orbit wall. The left jugal is incompletely preserved but is of typical dissorophoid type, namely a triangular bone with a centre of ossification at the level of the back of the orbit. Because of damage to the specimen, it cannot be observed whether or not the jugal forms part of the skull margin.

The frontals are relatively short rectangular ossifications, about twice as long as wide, and widening only slightly in the region of the prefrontal–postfrontal contact which excludes them from the orbit margin. The parietals are incomplete posteriorly but appear to be of conventional temnospondyl configuration. The large pineal foramen is very slightly wider than long. The postorbital is a shallow, roughly triangular bone forming the posterior margin of the orbit. Posteriorly, it is wedged shallowly between the squamosal and the supratemporal. Its distal end forms a slightly elongate extension ahead of the squamosal. The postfrontal has a relatively small contact with the orbit margin and is expanded posteriorly. Medially it forms a shallow wedge, narrowing the contact between the frontal and the parietal. The supratemporal is an anteroposteriorly elongate rectangular bone which broadly borders the dorsal margin of the large tympanic embayment. This tympanic border bears an expansion of sculptured bone partway along the margin. This was depicted as a laterally directed expansion by Milner (Reference Milner and Roček1986, fig. 2), but appears to be the ventrally directed supratemporal contribution to the supratympanic flange crushed outwards. If so, it demonstrates that this structure bore dermal sculpture in Mordex as in Ecolsonia. The supratemporal component to the supratympanic flange is not visible and is presumably crushed under the skull table. The squamosal is of a type frequently found in dissorophoids in that the sculptured dermal region is roughly crescent shaped, bordering the anterior edge of the tympanic embayment, while there is a large smooth surface inside the tympanic embayment. The sculptured region is unusually anteroposteriorly narrow, particularly at the medial end, which is widely separated from the tabular. Distally, the squamosal extends only about halfway back along the ventral side of the tympanic embayment. The isolated left squamosal NMW 1983-32-79 (Fig. 5e, f) merits separate description at this point, as it is less crushed and abraded than that in NMW 1898-X-47 and its precise sutural margins can be seen. There is slight crushing anterior to the stapedial notch and the resulting cracks and raised wedge of bone have been depicted in Figure 5f. The squamosal is slightly larger than that in the large skull, the sculptured region having a maximum width of 32mm as against 29mm in NMW 1898-X-47. The sculptured region is crescent shaped, the semicircular concave edge forming the anterior half of the tympanic border. Within the concavity of the tympanic border is a lightly striated thick shelf of bone bordering an acute V-shaped stapedial notch. The dorsomedial region of this shelf forms the anterior region of a supratympanic flange, but does not show the expansion of this flange that one would expect in a true dissorophid (Bolt Reference Bolt1974b). Extending around the anterior and lateral edges of the bone is a lightly striated irregular shelf which formed an undercutting sutural connection with the supratemporal, postorbital, jugal and quadratojugal. It has the same relationship to the sculptured region as that in Micromelerpeton (Boy Reference Boy1972, abb. 4g).

The quadratojugal of NMW 1898-X-47 is a relatively massive bone. The anterior region is lost and so its anterior extent and possible contact with the maxilla are unknown. The dermal sculpture is exceptionally rugose and it is probable that an expanded sculptured flange of bone hung down the cheek below the level of the jaw margin. This is the condition in Ecolsonia (Berman et al. Reference Berman, Reisz and Eberth1985, figs 7c, 8c) and the quadratojugal of NMW 1898-X-47 bears a distinct resemblance to that of Ecolsonia seen in dorsal aspect (Fig. 4 and Berman et al. Reference Berman, Reisz and Eberth1985, fig. 8a). Medial to the dermal exposure of the quadratojugal is the smooth ventral face of the embayment, made up largely of the smooth squamosal flange. In some temnospondyls, the quadratojugal contributes to the posterior region of this smooth flange but no suture can be seen which might suggest a quadratojugal–squamosal boundary in this region. Posteriorly, the sculptured region of the quadratojugal seems to extend a slender process of dermal bone above the quadrate although sutures are not clear in this region. It is possible but not certain that this is part of a crushed quadrate–tabular contact enclosing the back of the tympanic embayment. The tabular is only partly preserved on the edge of the specimen. It is a quite substantial bone with at least a slight posterolateral extension. Again, this may be part of a process of bone enclosing the back of the embayment. The quadrate is not clearly visible but there are two lumps of unsculptured bone, side by side at the posterior margin of the embayment and these probably represent the bilobed condyle of the quadrate, corresponding to the bilobed cartilage-lined pockets on the articular. There is no evidence of a blade-like anterior quadrate process in the tympanic embayment as found in some dissorophids (e.g., Cacops, see Fig. 10e).

4.2.2. Palate (Fig. 4)

Because NMW 1898-X-47 is preserved in dorsal aspect, only a few parts of the palate are visible in the narial, orbital and otic regions (Fig. 4). Through the large left external naris, can be seen part of the left vomer bordering the internal naris. The vomer appears to have an upturned rim bordering the internal naris and small denticles can be seen on the edge of this rim. The choanal margin appears be bordered by denticles, a situation which also occurs in Platyrhinops lyelli from Linton (Clack & Milner Reference Clack and Milner2010). The choana is also relatively large but narrower than the external naris. The anteromedial and posteromedial faces of the internal naris are almost perpendicular to one another and are separated by a very sharp curvature through 90°.

A fragment of the dorsal face of the palatine and much of the palatine ramus of the pterygoid can be seen in the left orbit. They demonstrate that the interpterygoid vacuities are large with very curved lateral margins. The pterygoid extends well forwards of the palatine and probably contacts the vomer, excluding the palatine from the interpterygoid vacuity margin.

4.2.3. Mandible (Figs 4, 5c, 7)

The mandible of NMW 1898-X-47 is 135mm long in a straight line from the mandibular symphysis to the back of the articular. It is visible in dorsal aspect and is crushed dorsoventrally, particularly at the posterior end. Traces of most of the ten constituent bones presumed to be present are visible, although the coronoid denticles obscure the margins of the individual coronoids. The sutural relationships of the mandibular elements are, where visible, almost identical to those in Tersomius (Carroll 1964, fig. 8).

The dentary is about 110mm long and extends back to just behind the level of the anterior end of the adductor fossa. It bears a single row of marginal teeth, there being space for up to 75 teeth. About 40 teeth are present in approximate alternation, although occasional sequential pairs of teeth are visible, the whole being presumably the product of a ‘zahnreihe system'. The marginal teeth are small, no more than 3mm in height and are simple cones, slightly bucco-lingually flattened at the unicuspid tip, and slightly incurved lingually. Each tooth bears about 12 grooves around the base, the external manifestation of simple labyrinthodont infolding. This large number of small marginal teeth is a distinct character separating Mattauschia from Ecolsonia, which has about 33 larger teeth in each mandibular ramus. Both genera share the inward curvature of the marginal dentition. At the anterior end of the dentary is a thickening at the mandibular symphysis (Fig. 5c). This thickening does not bear any certain symphyseal fangs but there is a detached tooth lying on top of the thickening that also bears two pits that may represent fang pits. If so, the symphyseal fangs are no larger than the marginal teeth, a situation similar to that in Tersomius (Carroll 1964, fig. 8). Lingual to the marginal teeth is a short shelf extending up to the coronoid region. This region is largely covered by a shagreen of minute denticles commencing just lateral to the symphyseal thickening and ending about 5mm anterior of the adductor fossa. The denticle-bearing strip is about 12 denticles wide. Sutural borders of the anterior and middle coronoids are completely obscured by the denticles but the posterior margin of the third coronoid extends behind the denticles and forms the anterobuccal margin of the adductor fossa, contacting the dentary and surangular.

On the ventromedial side of the mandible, slivers of bone representing the medial edges of the splenial and postsplenial are visible. They appear to be of orthodox temnospondyl configuration. Behind the postsplenial is a more extensively exposed prearticular. Anteriorly, this bone narrows to a point wedged between the postsplenial and the coronoid series. Posteriorly, it broadens out to form the posterolingual face of the mandible and the medial border of the adductor fossa. The buccal border of the adductor fossa is formed by the surangular which is wedged between the dentary and the coronoid anteriorly and extends back to border the articular laterally. The lingual border of the adductor fossa is formed by the prearticular extending back to border the articular medially, as in Tersomius (Carroll 1964, fig. 8). There is no medial inflection of the prearticular as in the Permian trematopids described by Dilkes (Reference Dilkes1990). At the posterior end of the mandible, the articular is visible as a relatively large articular surface of ‘unfinished' bone. As in other dissorophoids, the articular surface is anteroposteriorly deep and consists of a two-lobed structure, there being a smaller buccal lobe and a larger lingual lobe separated by a ‘waist'.

5.3.1. Closure of the tympanic embayment

Comparison of the midline skull lengths and otic embayment closure gives the following result. Note that Mordex, Fedexia, Actiobates, Anconastes, Tambachia and Rotaryus are all single-specimen genera.

The Pennsylvanian genera Mattauschia and Mordex are represented by small short-snouted juveniles (33–39mm skull) with an open squamosal embayment, and a large specimen (120mm skull) with an enclosed embayment. Fedexia is represented by a single incomplete 103mm skull with an open embayment to judge by the condition of the tabulars, Ecolsonia is represented by complete or partial large skulls in the 135–150mm range with closed embayments and highly comparable with the large Mattauschia. Actiobates has a 70mm skull with an open embayment and Anconastes has a 100mm skull with an embayment of uncertain status. The Permian taxa comprise Tambachia with a 65mm skull with open embayment, Rotaryus with a 48mm skull and open embayment, Phonerpeton consisting of skulls in the 60–90mm range with open embayments and Acheloma with 150–180mm skulls with closed embayments. In summary, skulls from 30–110mm midline length have open embayments, whereas skulls from 120–180mm in length have the embayments closed by a tabular–quadratojugal strut. Is this an unqualified taxonomic feature or an ontogenetic phenomenon in all trematopids?

In this work, skulls assigned to Mattauschia, include a small skull with an open embayment and a large one with a closed embayment. A later potential instance of this is Acheloma in the Arroyo of Texas where the large skulls are accompanied by smaller 50–70mm skulls (‘Trematops willistoni' of Olson; e.g., FMNH UC 1584, MCZ 2475) with very narrow notches combined with open embayments. The former was not discussed by Dilkes (Reference Dilkes1990) and the latter was simply attributed to Phonerpeton sp. and they have been neglected in all recent literature. They have the massive horizontal cheek region of Acheloma but the otic embayment is not closed at the back. They are consistent with being half-grown post-metamorphics of Acheloma cumminsi. A third possible example of the problem is Phonerpeton pricei, described by Dilkes (Reference Dilkes1990) as a small trematopid with open otic embayments, from the Archer City to Petrolia Formations of Texas. However in the collections at the Field Museum, Chicago, is FMNH UC 481 – a 150-mm skull from the Archer City Formation which demonstrates that either a second large form lived alongside Phonerpeton, or that at least some of the Phonerpetons are juveniles of an earlier Acheloma species where the flattening of the otic embayment was less highly developed. This specimen is not mentioned by Dilkes (Reference Dilkes1990).

The possibility must be reconsidered that all trematopid genera grew to 150mm or more in skull length but that tympanic embayment closure only took place at 110–120mm midline skull size. If this is so, the types of Mordex, Actiobates, Anconastes, Tambachia and Phonerpeton may all be immature and not ontogenetically equivalent to the larger specimens of Mattauschia, Fedexia, Ecolsonia and Acheloma.

5.3.2. Nostril shape

Nostril shape must almost certainly change with ontogeny, at least in the more derived trematopids. Dilkes (Reference Dilkes1993) figured a small snout (MCZ 2531) from the Lower Permian of Texas in which the external naris is a simple oval like those of Mattauschia and Fedexia. Dilkes interpreted this as a juvenile with an estimated skull length of 53mm. This is the most logical interpretation, although the possibility exists that this is a late-surviving relic of the Mattauschia/Fedexia snout morphology. Phonerpeton skulls of 60–70mm figured by Bolt (Reference Bolt1974a fig. 1) and Dilkes (Reference Dilkes1990 figs 3, 4) show a fully elongate bilobed naris implying either a very rapid morphological change during growth or the existence of a mixture of juvenile and dwarf species in this size range. This might explain the confusing mixture of snout shapes and nostril shapes combined within Phonerpeton. It may explain the variable position of Phonerpeton in relation to Acheloma in the cladograms of Polley & Reisz (Reference Polley and Reisz2011) and Berman et al. (Reference Berman, Henrici, Martens, Sumida and Anderson2011).

5.3.3. Snout shape and length

Finally, there is the issue of whether the smaller skulls can be determined as growing into Acheloma-shaped adults or Ecolsonia-shaped adults. An example of this problem may be the status of Ecolsonia and Anconastes both from the Cutler Formation of north-central New Mexico. They were described as an aberrant dissorophid and a trematopid respectively but in some recent analyses they are apparently different grades of trematopid. However, as noted by Sumida et al. (Reference Sumida, Berman and Martens1998), Anconastes, because incomplete, possesses only two of the six characters that they used to define the Trematopidae and both of these (external naris and pseudocaniniform teeth) are present in Ecolsonia. As Anconastes and Ecolsonia also share a peculiar mosaic covering of dorsal osteoderms, not reported in any other trematopids, it is possible that Anconastes is no more than a half-grown Ecolsonia. They have never been directly compared. The German trematopid Tambachia shares three characters with Anconastes (Sumida et al. Reference Sumida, Berman and Martens1998) namely absence of internarial fenestra (also in Ecolsonia), suborbital process of lacrimal reduced also (in Ecolsonia) and no palatine dorsal exposure but maxilla entering orbit margin (Ecolsonia does have a palatine exposure excluding the maxilla from the orbit margin but this could be size linked). It is perfectly possible that a large Tambachia would be more Ecolsonia-like than Acheloma-like.

5.3.4. Summary

Clearly a simple cladistic analysis of a series of genera represented partly by juveniles and partly by adults could give flawed results if this is not taken into account. Further investigation is beyond the scope of this paper and no taxonomic changes are proposed here but this problem merits a further focussed study. If such late ontogenetic changes do take place within trematopid species, any character analysis that does not take account of this is likely to be highly compromised. In the following discussion, the relationships of Mattauschia are discussed solely in relation to equally large trematopids (midline skull length 120mm or more) which are unambiguously adult, whereas the relationships of Mordex are simply discussed with extreme caution.

5.4.1. Mattauschia

The relationships of Mattauschia will be considered first. It is compared only with genera with comparably large skulls, namely Fedexia, Ecolsonia and Acheloma, in order to avoid the complication of ontogenetic factors (Fig. 9a). The character states involved are sufficiently few that a ‘hand-worked' presentation seems sufficient to the author.

Figure 9 (a) A hypothesis of relationships of four trematopid genera. Character states defining nodes are: Node 1 (Trematopidae) small v-shaped lacrimal and enlarged external naris, flattened tympanic embayment and large quadratojugal; Node 2 reduced prefrontal and postfrontal permit frontal to enter orbit margin, reduction to 30–40 large marginal teeth with caniniforms on premaxilla and maxilla; Node 3 external naris divided into anterior and posterior lobes by nasal flange, tympanic embayment further flattened; Taxon 4 (Acheloma) elongate straight-sided snout, internarial fenestra, even fine dermal ornament. (b) Cladogram with Mordex inserted, its small size reflecting its juvenile status. Mordex shares naris shape with Node 3 and fine ornament with Acheloma, both contra Fedexia. However, it lacks the Node 2 characters that Fedexia shares with Ecolsonia and Acheloma.

Mattauschia is a trematopid because it possesses (1) the enlarged external naris reaching almost to orbit with the dermal portion of lacrimal small and triangular, and the prefrontal entering the naris margin, (2) an unornamented supratympanic flange of the tympanic embayment with a slender squamosal component and a suture indicative of a significant supratemporal component (in the isolated squamosal), the remaining characters of the flange being obscured, (3) a tympanic embayment with a ventral border sloping at less than 45° in large individuals so that at least the anterior region of embayment tends to a horizontal slit, this is combined with an enlarged deep quadratojugal (this is not seen in Fedexia which has a deep embayment) (Fig. 10).

Figure 10 Tympanic embayments of trematopids and Cacops in lateral aspect. Quadratojugal shaded. (a) Mattauschia (based on Fig. 10). (b) Ecolsonia (based on Berman et al. Reference Berman, Reisz and Eberth1985, fig. 8c). (c) Anconastes (based on Berman et al. Reference Berman, Reisz and Eberth1987, fig. 8), degree of closure of tympanic embayment is not known; (d) Acheloma (based on Dilkes & Reisz Reference Dilkes and Reisz1987, fig. 3); (e) Cacops (based on Williston Reference Williston1910, pl. 8, sutures after Fröbisch & Reisz Reference Fröbisch and Reisz2012 fig. 2).

Mattauschia is more primitive than any later trematopid because (1) it retains a prefrontal–postfrontal contact, (2) it has a marginal dentition of 75 small teeth per ramus as opposed to 30–40 large teeth per ramus, (3) it lacks the inflexion of the prearticular on the mandible (situation unknown in Fedexia). Given the visible pattern of dentition, it probably lacks the pseudocaniniform teeth on the premaxilla and the anterior maxilla, but this is not demonstrable.

Mattauschia and Fedexia are more primitive than later trematopids because (1) the external nares are large ovals not divided into anterior and posterior lobes by a lateral flange of the nasal. The nasals are long rectangular straight sided elements, (2) the tympanic embayment of is less flattened and trematopid-like than those of Ecolsonia and the Permian trematopids.

Mattauschia, Fedexia and Ecolsonia are more primitive than later trematopids such as Acheloma because (1) they lack the elongate straight-sided snout of later forms, (2) they lack the internarial fenestra (3) they lack the very fine dermal sculpture, (4) they lack the derived otic notch reduced to a slit.

Two features of uncertain significance are as follows. One is the very narrow distance between the orbit and the squamosal embayment in Mattauschia which separates it from all later trematopids but might be an autapomorphy. The other is the humerus of the small type specimen which lacks a supinator process, whereas more derived trematopids possess one. This might be a significant phylogenetic feature but could also be based on ontogenetic change with the supinator ossifying only in larger individuals.

5.4.2. Mordex

Mordex is a trematopid because it possesses (1) the enlarged external naris reaching almost to orbit with the dermal portion of lacrimal small and triangular extending a finger of bone along the lateral margin of the naris, together with the prefrontal entering the naris margin, (2) an unsculptured supratympanic flange of the tympanic embayment with a slender squamosal component, a semilunar flange of the supratemporal and a tabular process which contacts the squamosal below the supratemporal, (3) the palatine ramus of the pterygoid extending forwards to reach the vomer and exclude the palatine from the margin of the interpterygoid vacuity. A character of uncertain significance found in some but not all trematopids is the relatively massive premaxilla.

Mordex is more primitive than any later trematopid because, like Mattauschia, (1) it retains a prefrontal–postfrontal contact, (2) it has a marginal dentition of small teeth per ramus as opposed to fewer large teeth per ramus. The status of the inflexion of the prearticular on the mandible is unknown. Given the visible pattern of dentition, it probably lacks the pseudocaniniform teeth on the premaxilla and the anterior maxilla, but this is not demonstrable.

At this point, there is a character conflict between Mordex and Fedexia (Fig. 9b). Mordex is more primitive for the reasons listed above, but more derived than Mattauschia in that it possesses a nasal with a flange dividing the external naris into two lobes, and it also possesses very fine dermal sculpture like most later trematopids. Mordex also shows the very narrow distance between the orbit and the squamosal embayment like Mattauschia and a humerus which lacks a supinator process. Given that the reduction in size of the circumorbital series resulting in loss of the prefrontal–postfrontal contact occurs repeatedly in the Dissorophoidea (within Branchiosauridae, Micromelerpetontidae and Amphibamidae), it is suggested that this is not a character of high uniqueness and that Mordex might be closer to higher trematopids than is Fedexia. The ambiguity of this situation is reflected in Figure 9b.

5.4.3. General conclusions

It seems clear that these two genera are the most primitive trematopids, they have only a few of the character states that genera such as Phonerpeton and Acheloma share, but character acquisition is sequential so this is to be expected in early more primitive forms. My conclusion is that Mattauschia is certainly the most basal trematopid yet described, while Mordex might be the most primitive relative of the main trematopid clade from Ecolsonia crownwards based on the possession of the characteristic external naris and fine dermal sculpture.