A high latitude Gondwanan species of the Late Devonian tristichopterid Hyneria (Osteichthyes: Sarcopterygii)

We describe the largest bony fish in the Late Devonian (late Famennian) fossil assemblage from Waterloo Farm near Makhanda/Grahamstown, South Africa. It is a giant member of the extinct clade Tristichopteridae (Sarcopterygii: Tetrapodomorpha) and most closely resembles Hyneria lindae from the late Famennian Catskill Formation of Pennsylvania, USA. Notwithstanding the overall similarity, it can be distinguished from H. lindae on a number of morphological points and is accordingly described as a new species, H. udlezinye sp. nov. The preserved material comprises most of the dermal skull, lower jaw, gill cover and shoulder girdle. The cranial endoskeleton appears to have been unossified and is not preserved, apart from a fragment of the hyoid arch adhering to a subopercular, but the postcranial endoskeleton is represented by an ulnare, some semi-articulated neural spines, and the basal plate of a median fin. The discovery of H. udlezinye shows that Hyneria is a cosmopolitan genus extending into the high latitudes of Gondwana, not a Euramerican endemic. It supports the contention that the derived clade of giant tristichopterids, which alongside Hyneria includes such genera as Eusthenodon, Edenopteron and Mandageria, originated in Gondwana.


Introduction
Tristichopterid fishes represent the sister group of elpistostegalians and digit bearing tetrapods.They achieved a worldwide distribution during the later part of the Devonian Period, before becoming extinct during the End Devonian Mass Extinction Event [1].Most known taxa have either been recovered from the tropical to subtropically deposited sediments of Euramerica or alternately from Australia which, towards the end of the Devonian, formed the low latitude northern rim of Gondwana.Olive et al. [2] have hypothesised a western European (Euramerican) origin for tristichopterids, suggesting that they migrated from Western Europe to Australia as part of the Late Devonian faunal exchange between Euramerica and Gondwana.Thereafter they suggest that a highly nested clade of giant Late Devonian tristichopterids, comprising Cabonnichthys burnsi [3], Mandageria fairfaxi [4], Eusthenodon wangsjoi [5], Edenopteron keithcrooki [6], and Hyneria lindae [7] most likely differentiated in Australia.Of these Cabonnichthys, Mandageria and Edenopteron are known from single localities in eastern Australia, whereas Eusthenodon is known from Greenland, Western Europe, Russia and Australia, and Hyneria is known from Pennsylvania in North America [2,8,9].Eusthenodon and Hyneria are hypothesised to have migrated to Euramerica from an Australian centre of origination during the Famennian [2].
The greatest limitation of this biogeographical scenario is that in designating Australia specifically as the centre of endemism of the proposed Gondwanan originated clade, it overlooks the possible effects of collection bias.Although Gondwana was the largest landmass during the Devonian, extending from tropical to polar regions, investigation of Late Devonian early vertebrate faunas has largely been restricted to Australia and palaeo-adjacent portions of Antarctica.
Here we describe a new species of Hyneria from the Gondwanan high-latitude lagersta ¨tte at Waterloo Farm, South Africa (Fig 1).This has important implications for our understanding of the biogeographical distribution and habitat preferences of Late Devonian tristichopterids.

Material and techniques
The fossil material was excavated from a roadside cutting outside Makhanda/Grahamstown and is now housed at the Albany Museum in that city.Specimens were prepared mechanically by hand, without the use of power tools; the soft sediment was in most cases slowly flaked off with a porcupine quill in order to protect the fragile and diagenetically altered bone compressions.The bones were photographed without any form of coating.Interpretative drawings were traced directly from the photographs, in Adobe Photoshop, with the specimens to hand for checking.Where possible, information was combined from the part and counterpart of the bone.The three-dimensional skull reconstruction was produced by fitting together paper outlines of the bones to make a physical three-dimensional model, which was then photographed and traced in Adobe Photoshop.All material was collected by the first author, with the permission of SANRAL (the South African National Roads Agency; land owner) and under permits granted by SAHRA (the South African Heritage Resource Agency) and ECPHRA (the Eastern Cape Heritage Resource Agency), before being accessioned into the Devonian collection of the Albany Museum, on which the first author is the principal researcher.One exception is specimen AM18000, collected by Ryan Nel under the first author's supervision.No permits were required for the described study, which complied with all relevant regulations.Permits were not required because the first author is employed by the Albany Museum specifically to curate and conduct research on their Devonian collection, of which this material forms a part.The second author was invited by the first author to collaborate on analysis of the material.

Nomenclatural acts
The electronic edition of this article conforms to the requirements of the amended International Code of Zoological Nomenclature, and hence the new names contained herein are available under that Code from the electronic edition of this article.This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN.The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/".The LSID for this publication is: urn:lsid:zoobank.org:pub: 3112D493-2967-448F-B095-52892107DDC1.The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital repositories: PubMed Central, LOCKSS, DiVA (Uppsala University repository).Diagnosis-Tetrapodomorph sarcopterygians with postspiracular bone present, vomers with long caudal process clasping the parasphenoid, circular scales with a median boss, and an elongate body with a trifurcate or rhombic caudal fin (modified from [3]).
Etymology-an apposition, from isiXhosa 'udlezinye', meaning 'one who eats others', referring to the inferred predatory lifestyle of the species.IsiXhosa is the widely spoken indigenous language of south-eastern South Africa where the fossil locality is located.

Description
The Hyneria material from Waterloo Farm consists predominantly of dermal bones, though some elements of the axial skeleton and paired fins are also preserved (Fig 2).Material is preserved in muddy carbonaceous metashale.During metamorphosis thicker bone became internally carbonized, though its outer surface is defined by kaolin derived from breakdown of secondary mica.The latter style of preservation also comprises thinner elements such as scales.The large size and robustness of the bones means that the dermal ornament and overlap surfaces are, for the most part, clearly visible.The suite of dermal skull bones is complete enough to allow a well-constrained reconstruction of the cranial morphology (Fig 3).Lateral line canals are generally not well preserved because of the compaction of the material, but can be traced in some areas.Curiously, the teeth are much less well preserved than the bone tissue; they tend to be black rather than silvery in colour, and even the largest of them (such as the dentary fang pair) do not show complete natural outlines.No cephalic endoskeletal elements of Hyneria have been recovered from Waterloo Farm.This may be due to preservational factors, but Hyneria appears in any case to be characterised by a lightly ossified endoskeleton.The type species Hyneria lindae from the Catskill Formation, Pennsylvania, has an ossified Meckelian element, palatoquadrate, hyomandibula, scapulocoracoid and pectoral fin skeleton, but the braincase seems to be completely unossified [17,18].Some remains of the axial and appendicular skeleton of Hyneria udlezinye have been recovered, including, generally disassociated, neural and haemal arches, a fin support, lepidotrichia and an ulnare.
Because Hyneria udlezinye is in most respects extremely similar to the type species Hyneria lindae, with only proportional differences in some bones serving to distinguish the two species, the Waterloo Farm material will be described relatively briefly and each element in turn will be compared with that of H. lindae.Comparisons with other tristichopterids will be made in the discussion.The distribution of bones on the bedding planes and their relative sizes allow us to assign certain suites of bones to single individuals with reasonable confidence.AM6528 + 6532 + 6540, which we designate as the holotype of Hyneria udlezinye, comprises an incomplete postparietal shield, quadratojugal, incomplete opercular, anocleithrum, cleithrum and clavicle.AM6535 comprises an incomplete parietal shield and a preopercular, AM6542 a postorbital and a lateral extrascapular.It has not proved possible to assign the remaining bones with confidence to these individual assemblies, though some almost certainly belong to them.
Skull roof.The parietal shield is represented by a single almost complete right parietal with an attached complete intertemporal (Fig 4A -4D).Well-defined overlap areas for the median rostral, nasal and anterior supraorbital (or prefrontal in tetrapod terminology) are present along its anterior and anterolateral margins.The contact for the posterior supraorbital (postfrontal) is a broad underlap on the ventral surface of the parietal; this feature can also be seen in Hyneria lindae, although it is not labelled [17: fig 3A].The medial margin of the parietal carries a notch for the pineal region, which appears to have been drop-shaped with a posterior point.The anterior part of the main trunk of the supraorbital lateral line is visible on the dorsal surface of the parietal, along with two smaller anterolaterally directed side branches.The intertemporal carries a large lateral overlap flange for the postorbital.
The proportions of the parietal shield are distinctive, with a notably short and broad postpineal region (Fig 3A).This feature is shared with H. lindae, and possibly with Langlieria socqueti [20], but not with other derived tristichopterids such as Eusthenodon, Mandageria, Cabonnichthys or Edenopteron [3][4][5][6]21], in which this region is proportionately longer and often tapers posteriorly towards a proportionately shorter posterior margin.
The postparietal shield is represented by a single incomplete postparietal together with the associated tabular and supratemporal (Fig 4D and 4E).The two latter bones are complete, but their sutures with the postparietal cannot be traced.A short section of the postotic lateral line canal is visible in the posterior part of the tabular.The outline of the postparietal shield differs from that of H. lindae in the smoother curve of the lateral tabular margin, which does not have a distinct lateral corner and is slightly longer with a more posteriorly positioned posterolateral corner.The anterior margin of the shield (i.e. the dermal intracranial joint) is also proportionately shorter than in H. lindae and the contact margin for the postorbital on the supratemporal correspondingly more oblique.down to contact the jugal.This condition is typical for the clade of derived tristichopterids that includes Hyneria, Eusthenodon, Edenopteron, Mandageria and Cabonnichthys [2,3].
The maxilla is almost parallel-sided.Again, this is typical for the aforementioned clade of derived tristichopterids, and differs from the condition seen in Eusthenopteron and almost all non-tristichopterid 'osteolepiforms', where the maxilla has a distinct dorsal process peaking at the jugal-squamosal junction [22,23].The maxilla of Hyneria lindae is even more extreme than that of H. udlezinye, in that the contact margin for the squamosal is concave rather than convex [17: fig 5].A parallel-sided maxilla is also present in tetrapods, although this is probably convergent with the derived tristichopterid condition.Near the anterior end of the dorsal margin, the maxilla carries a slender anterodorsally directed process.This process contributes to the posterior margin of the choana and is seen in many tetrapodomorph fishes [17,23].
The remaining bones of the cheek are of generalized tristichopterid character.Comparison with Hyneria lindae is made somewhat more difficult by the fact that the articulated H. lindae cheek-plate figured by Daeschler & Downs [17] is shown only in internal view, whereas the H. udlezinye cheek bones are exposed in external view.Tristichopterid cheekbones tend to have large overlap surfaces [22: fig 115], meaning that the internal and external suture patterns can look rather different.In the H. udlezinye material we note particularly that the squamosal has a very extensive overlap surface for the postorbital (Fig 5G and 5H).Nevertheless, comparison between the two species indicates that H. udlezinye differs from H. lindae in having a proportionately deeper preopercular (in H. lindae this bone is very narrow and bar-shaped) and a shorter, deeper, lacrimal.In both these respects, Hyneria udlezinye appears to be the more generalized of the two, matching the condition in a wide range of tristichopterids including Eusthenopteron, Cabonnichthys, Eusthenodon and Edenopteron [3,5,6,21,22].
Lower jaw.The lower jaw is poorly represented in the material.Only the dentary and an incomplete angular are known as isolated bones (Fig 6A , 6B, 6D and 6E).An articulated but partly disrupted lower jaw ramus in mesial view shows the prearticular in association with some of the smaller bones (Fig 6C ) but is in some respects challenging to interpret.The angular is of a generic tristichopterid type, but the dentary has a distinctive, almost club-shaped morphology with a deep anterior end carrying the poorly preserved remains of a large fang pair (Fig 6A and 6B).Comparison with the lower jaw of Hyneria lindae, which unfortunately is figured only in mesial view so that the complete outline of the dentary cannot be seen, shows that the dentary fang pair of H. lindae is only about half the relative size of that of H. udlezinye, and suggests that the dentary of H. lindae is shallower at the anterior end.
Palate.The palate of Hyneria udlezinye is represented by several entopterygoids of varying degrees of completeness and a complete, partially exposed parasphenoid (Fig 7).The left entopterygoid of the holotype individual is almost complete when information from the part and counterpart is combined (Fig 7B).In addition to the main corpus of the bone, which was exposed on the palate in life, the entopterygoid also has an extensive ventral flange that formed the attachment area for the ectopterygoid and dermopalatine.The two latter bones were thus not in direct contact with the palatoquadrate.The same arrangement was documented by Jarvik [22: fig 108] in the grinding series of Eusthenopteron.As the palatoquadrate complex of H. lindae (which includes entopterygoid, dermopalatine and palatoquadrate) is incomplete posteriorly, it is somewhat difficult to compare with the entopterygoid of H. udlezinye.However, in H. udlezinye the denticulated field has a more pronounced dorsal corner and extends closer to the dorsal margin of the bone in the metapterygoid region.
The parasphenoid, which also belongs to the holotype individual, is exposed in right ventrolateral view (Fig 7E and 7F).Like the corresponding bone in H. lindae it has a very long overlap area for the vomer, but the denticulated field extends further anteriorly than in H. lindae.Operculogular series.The bony gill cover is represented by specimens of the opercular, subopercular and principal gular ( Fig 8).No definite submandibulars have been recovered, although a rectangular bone with a large overlap area at one end (tentatively interpreted as a supracleithrum) may in fact come from this series.The most noteworthy aspect of the operculogular series is the great depth of the opercular and the correspondingly shallow subopercular.The opercular of H. udlezinye is nearly three times as deep as the subopercular, whereas in Eusthenopteron it is less than twice as deep [22].In Hyneria lindae the opercular is incompletely known, but the subopercular has a very different shape to that of H. udlezinye with a much deeper posteroventral region [17: fig 8C,D].As the subopercular of H. udlezinye is represented by several specimens of different sizes, all with the same distinctive shallow shape, we are confident in the taxonomic significance of this difference.The principal gular is of generalized tristichopterid type.
Pectoral girdle.Only the dermal elements of the pectoral girdle are preserved (Fig 9).These include definite examples of the lateral extrascapular, anocleithrum, cleithrum and clavicle.The lateral extrascapular has a smoothly rounded posterior margin.If its size relative to the postparietal shield has been inferred correctly, the median extrascapular was rather wide.The anocleithrum is very similar to that of Hyneria lindae [17].Its ornamented area is triangular as in H. lindae, Edenopteron and Mandageria [4,6,17], contrasting with the four-sided ornamented area of Eusthenopteron    Axial skeleton.The only preserved part of the axial skeleton, apart from isolated bones, is a partly articulated series of arches with fin radials from the posterior part of the vertebral column, associated with a triangular basal plate from either the anal or the posterior dorsal fin (Fig 11C).Interestingly, the shape of the basal plate resembles that of non-tristichopterids such as the porolepiform Glyptolepis [24,25] rather than Eusthenopteron [22].The arches comprise one evidently articulated series and a scatter of disrupted elements; we surmise that the articulated elements are most probably neural arches, held in place by the body musculature after death, whereas the scattered elements are haemal arches that have been displaced by the rupture of the body cavity.If this is correct, it implies that the neural arches of the tail region articulated distally with radials of the epichordal lobe of the caudal fin, as in Cabonnichthys [3],  contrasting with the condition in Eusthenopteron where such radials are absent [22].Vertebral centra are not preserved and were presumably cartilaginous.
Squamation and lepidotrichia.Numerous isolated scales and some areas of lepidotrichia are preserved in association with the dermal bones of Hyneria udlezinye (Fig 12).The scales are characterised by an extensive, deeply folded margin of the trailing edge, a character uniquely shared with Hyneria lindae (Fig 12B).Isolated, elongated scales with an extremely extended trailing edge (Fig 12C -12E), up to 75 mm in length, are interpreted as probable basal scutes from the fins [22].

Discussion
It is a straightforward matter to identify the largest sarcopterygian from Waterloo Farm as a giant tristichopterid.All the characteristics of the preserved bones fall squarely within the general gestalt and range of variation typical of these fishes.Assuming body proportions similar to those of Mandageria, the only large tristichopterid for which it has been possible to assemble a whole-body reconstruction [4], the total body length of the holotype individual of Hyneria udlezinye was in the range of 180-190 cm; some isolated bones, however, come from larger individuals.An isolated cleithrum (AM 6545), for example, is 50 percent larger than that of the type specimen (36 cm from dorsal to ventral extremity as opposed to 24 cm), suggesting a possible body length of at least 270 cm for this individual (Fig 10D).This size range is similar to other giant tristichopterids.A precise identification is more challenging, because the diversity of these top predators is surprisingly high: even if we restrict the analysis to known Famennian taxa we have to consider the genera Langlieria, Eusthenodon, Edenopteron and Hyneria [2].The features that assign the Waterloo Farm material to Hyneria are the broad parietal shield with its distinctively short postpineal region, the ornament (which is finer than in Eusthenodon and Edenopteron) and the morphology of the scales, which are uniquely characterized by an extensive, deeply folded  margin to the trailing edge.Hyneria udlezinye differs from H. lindae, the type species, in a number of respects.The proportions of the prearticular and lacrimal, and the morphology of the maxilla, appear less derived and closer to general conditions for the Tristichopteridae.Conversely, the shallow subopercular and exceptionally large (though incompletely preserved) dentary fangs of H. udlezinye appear to be autapomorphic.
Comparison with the published reconstructions of other giant tristichopterids suggests that Hyneria udlezinye has an unusually large branchial chamber.A measure of caution is needed, however, because the branchial region is difficult to reconstruct accurately as the bones frequently become displaced or pushed over each other.The gill cover of Eusthenodon, reconstructed by Jarvik [5], certainly seems implausibly small.On the other hand, Mandageria and Cabonnichthys have been reconstructed with gill covers approximately similar in size to each other and Eusthenopteron, all of them slightly smaller than H. udlezinye [3,4,22].Hyneria lindae appears to have gill cover bones of similar size to H. udlezinye, although the subopercular is proportionately deeper, but the skull has never been reconstructed [17].We conclude from this that Hyneria probably has a larger branchial chamber than other known tristichopterids, although the exact magnitude of the difference is uncertain.This may reflect life in oxygenpoor waters, or perhaps a very static lifestyle where the respiratory current was not significantly supplemented by forward movement through the water.
The discovery of Hyneria udlezinye increases the known diversity of large, late tristichopterids, and presents us with the top non-chondrichthyan predator of the Waterloo Farm ecosystem (Fig 13).However, arguably its main importance lies in the light that it casts on tristichopterid (and more generally, vertebrate) biogeography of the Late Devonian.Giant tristichopterids are globally distributed in the Famennian and appear to be almost ubiquitous members of non-marine to brackish ecosystems.They occur from the tropics of Euramerica (for example in East Greenland and at Andreyevka-2 in central Russia), through the southern mid-latitudes (the Catskill Formation of Pennsylvania and the Worange Point Formation of New South Wales), down to the Antarctic Circle (Waterloo Farm).Some genera were widespread.Eusthenodon is recorded from Greenland, Belgium, central Russia and Australia [2,8,9]; Langlieria from Belgium and Pennsylvania [26]; and now Hyneria from Pennsylvania and South Africa.In general terms this speaks to the interconnectedness of the Famennian world, but on a more specific level it would be interesting to know where this radiation of giants originated.
The recent study by Olive et al. [2] addressed this question by applying Bayesian Binary MCMC analysis to the geographic distribution of known tristichopterids, mapped onto a parsimony tree.The analysis suggested that the Langlieria lineage may have originated in western Europe, whereas the top-end clade containing Cabonnichthys, Mandageria, Eusthenodon, Edenopteron and Hyneria is interpreted as most probably an Australian radiation.The result undoubtedly reflects the fact that Cabonnichthys and Mandageria are known only from the Frasnian locality of Canowindra, and Edenopteron only from the Famennian Worange Point Formation, both in New South Wales.Much of Gondwana is poorly sampled for Late Devonian vertebrates, potentially skewing the results; it should be noted in this context that the Givetian to Frasnian Aztec Siltstone of Antarctica contains not only the rather primitive giant tristichopterid Notorhizodon, but also fragmentary remains of possibly more derived Eusthenodon-like forms [29].It may thus be that the apparent Australian biogeographical signature detected by Olive et al. [2] is actually a broader Gondwanan signature partly obscured by patchy sampling.The discovery of Hyneria udlezinye fits very nicely into this context by showing that Hyneria also has a Gondwanan distribution.Indeed, the great distance (both in kilometers and palaeolatitude) of Waterloo Farm from the Catskill Formation suggests that Hyneria may have been widely distributed in Gondwana.

Conclusion
The largest osteichthyan member of the Waterloo Farm vertebrate assemblage, a predatory sarcopterygian with a probable maximum length of nearly three metres, proves to be a new species of the genus Hyneria.This genus is otherwise only recorded from the late Famennian Catskill Formation of Pennsylvania.The new species, Hyneria udlezinye, differs from the type species Hyneria lindae in a number of minor but securely attested proportional characters relating to the skull roof, cheek, lower jaw and operculum.Hyneria now joins Eusthenodon and Langlieria as one of the derived, late, giant tristichopterids known from both Euramerica and Gondwana.The other confirmed members of this clade (Mandageria, Cabonnichthys and Edenopteron) are exclusively known from Gondwana.This strongly supports the contention that this clade represents a Gondwanan radiation [2].
Hyneria udlezinye is the first tristichopterid to be recorded from a high palaeolatitude, all other members of the group coming from palaeoequatorial to mid-palaeolatitude localities.All previously recorded Gondwanan members of the derived tristichopterid clade come from Australia, leading Olive et al. to argue for an Australian origin for this clade [2].The new evidence from Waterloo Farm, however, suggests that a more general Gondwanan origin for this clade is highly likely.This once again demonstrates how inferences about biogeographical patterns have historically been skewed by a paucity of data from high-palaeolatitude localities.Such  [11], the placoderms Groenlandaspis riniensis and Bothriolepis africana [19], the coelacanth Serenichthys kowiensis [27], the lungfish Isityumzi mlomomde [28], and a cyrtoctenid eurypterid.Painting by Maggie Newman, copyright R. W. Gess.https://doi.org/10.1371/journal.pone.0281333.g013data can only come from Gondwana, as no continents extended into northern high latitudes during the Devonian.The Waterloo Farm lagersta ¨tte provides a unique window into an almost unknown part of the Late Devonian world.

Fig 1 .
Fig 1.The location of Waterloo Farm.A, map of South Africa showing position of Waterloo Farm (W.F.).Brown area denotes the outcrop of the Witpoort Formation.B, palaeogeographical map of Gondwana in south polar view during the Famennian, showing the position of Waterloo Farm (W.F.) and three other Late Devonian tristichopterid localities, Eden (E), Canowindra (C) and Grenfell (G), relative to the South Pole (P).A modified after [10], B modified after [11].https://doi.org/10.1371/journal.pone.0281333.g001

Fig 2 .
Fig 2. AM6540b and AM6528a, the two main blocks of the holotype of Hyneria udlezinye.Each block also has a counterpart (not illustrated).A, AM6540b.Unlabelled bones all belong to a single large individual of the arthrodire placoderm Groenlandaspis riniensis[19].B, AM6528a.This block also carries a jugal of the tetrapod Umzantsia amazana[11] and a paranuchal of a small individual of Groenlandaspis riniensis.

Fig 4 .Fig 5 .
Fig 4. Skull roof bones.A-B, specimen photographs of parietal plus intertemporal, AM6535c,d, in part and counterpart (left-right reversed) view.C, interpretative drawing combining information from A and B. D, specimen photograph of postparietal-supratemporal-tabular (sutures not visible), AM6540a.E, interpretation of D. Thick outline denotes true margin; dotted thick outline, inferred true margin; thin outline, broken margin or edge of ornamented external surface; pale grey, ornamented external surface; dark grey, lateral line canal; black arrow, anterior.Abbreviations as in Fig 3. Specimens are shown to the same scale but the parietal represents a larger individual.https://doi.org/10.1371/journal.pone.0281333.g004

Fig 7 .
Fig 7. Palatal bones.A-B, photo and interpretative drawing of entopterygoid AM6528a.B also incorporates information from AM6528b.C-D, photos of entopterygoid AM6508b+a, part and counterpart.E-F, photo and interpretative drawing of parasphenoid AM6528a.Graphic conventions as in Fig 4.
https://doi.org/10.1371/journal.pone.0281333.g007variation in shape (Fig 10).As is typical for tristichopterids, the contact margin for the clavicle is concave.A distinctive feature shared with H. lindae is the strongly developed and finger-like anteroventral extension.Ulnare.The only bone to be preserved from the appendicular skeleton is an ulnare (Fig 11Aand 11B).It carries a large postaxial flange and is very similar to the corresponding element in Eusthenopteron [22: fig.102].

Fig 8 .
Fig 8.The gill cover.A-B, photo and interpretative drawing of opercular AM5532.C-D, photo and interpretative drawing of opercular AM6521a.E-F, photo and interpretative drawing of subopercular AM5249.G-H, photo and interpretative drawing of subopercular AM5391bi.I-J, photo and interpretative drawing of gular AM5888.Graphic conventions as in Fig 4. In H, blue indicates a partially preserved hyoid arch element adhering to the subopercular.https://doi.org/10.1371/journal.pone.0281333.g008

Fig 11 .
Fig 11.Postcranial endoskeleton.A-B, photo and interpretative drawing of ulnare AM4868.C, photo of scattered axial elements from the posterior part of the body, AM18002.Graphic conventions as in Fig 4.