Subdepartment of Evolution and Development, Department of Organismal Biology, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18A, 752 36 Uppsala, Sweden; e-mails: firstname.lastname@example.org, email@example.com
Subdepartment of Evolution and Development, Department of Organismal Biology, Evolutionary Biology Centre, Uppsala University, Norbyvägen 18A, 752 36 Uppsala, Sweden; e-mails: firstname.lastname@example.org, email@example.com
Abstract: A new genus and species of Devonian tetrapod has been identified from material collected in 1947 from the southern slope of Mt. Celsius, Ymer Ø, North-East Greenland. The specimen preserves both lower jaws, partial palate, premaxillae and maxillae, with a natural mould of parts of the shoulder girdle. The new taxon, Ymeria denticulata, shows differences in dentition, skull ornament and lateral line expression from both Acanthostega and Ichthyostega, but it shows a closer resemblance to the latter. A cladistic analysis not only suggests that Ymeria lies adjacent to Ichthyostega on the tetrapod stem, but also reveals substantial topological instability. As the third genus and the fifth species of tetrapod identified from North-East Greenland, it demonstrates the high diversity of Devonian tetrapods in that region.
T he area around Kejser Franz Joseph Fjord, North-East Greenland has long been famous for the earliest and richest discoveries of Devonian tetrapods (Jarvik 1996; Blom et al. 2005). Two genera are now well known. The first named was Ichthyostega, of which three species are now recognised (Blom 2005), and the second was Acanthostega with a single species (Jarvik 1952). Each is represented by a large collection of material in various states of preservation that was collected between 1929 and 1998 (see Jarvik 1996; Clack 2002a). In 1947, a team from Sweden and Denmark that included the late Professor Erik Jarvik collected a specimen that he was unable to assign to either known genus. It was a partially preserved skull that included both lower jaws and the remains of a palate together with the maxillae and premaxillae. Clack (1988) suggested the possibility that this specimen may represent a new taxon. Further preparation of this specimen and study in the light of much recent new information on Ichthyostega and Acanthostega has allowed us to recognise it as a new genus and species. One other lower jaw specimen can also tentatively be attributed to this genus. They both come from one of the less well-explored localities of the area, the southern slopes of Celsius Bjerg.
Geological Museum, Copenhagen, specimen (MGUH) VP 6088 is recorded as coming from the south side of Celsius Bjerg on Ymer Ø, where it was collected from the talus. However, unlike most specimens collected by the Swedish–Danish expeditions, no altitude is recorded. This means that we are unable to attribute it to one of the formations that comprises the Celsius Bjerg Group, the Upper Devonian strata exposed on Celsius Bjerg (Blom 2005; Blom et al. 2005). The Celsius Bjerg Group is dated as Famennian 2b (Marshall et al. 1999).
The specimen comprises a single partial skull preserved in coarse pale red sandstone with buff-coloured inclusions. Lower jaws, premaxillae and maxillae are preserved as white bone, and the palate and part of the shoulder girdle are largely present as natural mould. Sections through tooth bases and the ventral plate of the clavicle are also visible. Some of the poorly preserved palate on the right has been mechanically removed using pneumatic pen, dental mallet and mounted needle to expose the coronoid teeth and part of the prearticular of the right lower jaw. Other parts have also been surface-prepared mechanically.
A lower jaw specimen (MGUH VP 6026) that was collected earlier, in 1932, is also tentatively attributed. It too is from the southern side of Celsius Bjerg and again lacks altitude data, but with the note ‘Under E. Läktaren Löst’ (‘Under E. Balcony, Loose’). This poorly preserved specimen is in similar matrix to MGUH VP 6088. It shows an almost complete lateral view and has been partially mechanically prepared in mesial view to expose parts of the coronoid series and prearticular.
1988 unnamed and undescribed species Clack, pp. 712–713, text-fig. 8.
Derivation of name. Ymeria– refers to Ymer Ø where the specimens were found.
Holotype. MGUH VP 6088 (Figs 1–2), a partial skull with both lower jaws, maxillae, premaxillae, partial palate and shoulder girdle.
Locality and horizon. South side of Celsius Bjerg, Ymer Ø, North-East Greenland. Talus specimen from Celsius Bjerg Group, Famennian, Upper Devonian.
Diagnosis. Stem tetrapod retaining a number of plesiomorphic features with no autapomorphies preserved.
Differential diagnosis. Can be distinguished from Ichthyostega by the following: higher premaxillary tooth count (10 or 11); maxillary and dentary teeth of similar size; largest coronoid teeth similar in size to adjacent dentary teeth; tooth shape scarcely recurved, tumescent, lacking keels; dense band of denticles along dorsal rib of prearticular confined within distinct boundary; low-profile poorly defined dermal ornament; open lateral line sulcus on lower jaw.
Referred material. A lower jaw, MGUH VP 6026, similar locality and horizon. This specimen differs from MGUH VP 6088 in minor details of the dentition.
Derivation of name. Refers to the denticulated prearticular.
Diagnosis. As for genus.
Comparative description and interpretation
Skull roof and palate. Only a few skull roofing bones are present (Fig. 1). Both premaxillae are represented, though the left is only represented by its dentition and a narrow strip of broken bone. The right element shows low-profile ornament in a radiating pattern, with lateral line pores for the infraorbital line and a short groove for the supraorbital line. The anteromedial margin is excavated for reception of either one single (as in Ichthyostega) or one of a pair (as in Acanthostega) of median rostrals. Except for this region, the dorsal margin of the premaxilla is crenellated for interdigitation with the nasal bones. The posterolateral margin is somewhat embayed and inturned at its ventral edge where it forms the anterior margin of the external naris.
A small rectangular bone is impacted into the narial embayment of the premaxilla. This bears a section of lateral line, which suggests that it could be a lateral rostral, but the shape and size of the element are more compatible with an interpretation as a tectal (Fig. 1C). The ventral migration of the naris in the tetrapod stem group, from the facial position seen in ‘osteolepiforms’ to a location near the upper jaw margin in known Devonian tetrapods (Jarvik 1980, 1996; Ahlberg et al. 1994; Clack 1994), is accompanied by a narrowing and loss of the lateral rostral. The bone is certainly absent in Acanthostega (Clack 1994); in Ichthyostega, Jarvik (1980, 1996) described a lateral rostral, but this was said to be a narrow tubular bone carrying only the lateral line. The bone so described is no longer available for study. If the element in Ymeria is a lateral rostral, the morphology of the narial region must differ dramatically from those of other known Devonian tetrapods, with the naris itself in a much more dorsal position; this seems unlikely. An interpretation as a tectal makes morphological sense but implies that this bone carried a lateral line canal, a feature that would be unique among known tetrapodomorphs. However, the course of the infraorbital sensory canal is known to be variable among Devonian tetrapods, possibly because of the loss of the lateral rostral: in Acanthostega, the postnarial component of the canal enters the maxilla, where it apparently ends blindly, but this is definitely not the case in Ventastega or Ichthyostega (Ahlberg et al. 1994; Jarvik 1996). It is thus not impossible that the canal crossed from the lacrimal to the tectal in Ymeria. Alternatively, the bone may represent a displaced fragment of canal-bearing bone from elsewhere on the skull, but this does not explain its location or morphological similarity to a tectal. The anterior margin of the naris is smooth, but its posterior margin is not represented.
Both maxillae are represented; the left is present only as sections through a number of teeth; the right is visible both as sections through the teeth seen in dorsal view, and in lateral view, where most of the teeth and part of the facial lamina have been surface-prepared. In dorsal view, the right maxilla is broader at its anterior end where it bears the largest teeth. It appears to suture with the quadratojugal posteriorly, though this is only visible in a section through broken bone.
All the palatal bones are represented to some degree (Fig. 2). Part of the right vomer is visible in dorsal view, but it is incomplete and its margins are not clearly represented. A groove separates the body of the bone into two portions, and at the point where the groove meets the posterior margin, there is a semicircular embayment. Anterior to this, the margin is digitated for suture with its antimere. The anterior margin of the bone is smooth and probably formed part of the margin of an anterior palatal fenestra as in other Devonian tetrapods.
Both palatines and ectopterygoids are present as sections through their tooth rows, with a little intervening bone and some natural mould of their palatal surfaces. It is difficult to determine where the junction between them lies. The pterygoids and a short length of parasphenoid are represented only by natural mould and some remaining eroded bone. Because of the nature of the matrix, in which the grain size is approximately the same diameter as a denticle, as seen on the right coronoid, it is not possible to tell from the areas of natural mould of the palatal bones whether any of their surfaces were denticulated: possible impressions of denticles in the matrix may simply represent places where individual calcite grains have been eroded out. It is also not possible to determine where the pterygoid sutures with the marginal palatal bones.
The subtemporal fossae are rounded, as in Acanthostega and Ichthyostega, rather than elongate as in most other early tetrapods. Part of an elongate endochondral element is preserved between the palatal surface and the right lower jaw (Fig. 2). It is expanded at one end, with a concave margin. The bone has little or no perichondral lining, and it has not been completely prepared out to avoid undue damage to the natural mould of the palate. Its identity must remain uncertain; however, one possibility is that it is a branchial element. Branchial elements are known in Acanthostega and Ichthyostega (Coates and Clack 1991; Clack et al. 2003).
Lower jaw. Both lower jaws are present, though the left ramus has been crushed dorsoventrally. Of this ramus, the ventral margin, consisting of portions of the angular and prearticular, has been exposed. Sections through the tips of some dentary and coronoid teeth and a strip of dentary bone are visible in dorsal view. The right lower jaw is almost completely exposed in lateral view, except for its tip, which is inaccessible to preparation, and its posterior end, which is largely missing. The symphysial region of this ramus is quite well exposed in the palatal view of the specimen.
In striking contrast to the external surface of the lower jaw of Ichthyostega, in which the ornament is well developed but consists of evenly spaced pits and ridges (see for example that illustrated by Blom (2005)), the lower jaw of Ymeria shows only very poorly defined, indistinct and diffuse ornament. ‘Star-burst’ arrays of narrow, shallow grooves on the angular and surangular, and small, shallow pits on the splenials and along the dorsal margin of the dentary can be detected (Fig. 1). This appearance does not seem to result from erosion of the bone, as it appears in places that have been freshly prepared. Furthermore, examples of eroded dermal bone are available for Ichthyostega, and their appearance is quite unlike that of Ymeria: instead, it shows either enlarged and more open pit and ridge structure (MGUH 28377b, a specimen from south Celsius collected in 1998 and known to be Ichthyostega, formerly part of MGUH f.n. 200, Clack and Neininger 2000; part a illustrated in Clack et al. 2003) where it has been etched by rainwater, or still retains clear evidence of pitting even when the outer surface has been removed by abrasion (MGUH 28376 formerly MGUH f.n. 1381). In other taxa with high-profile ornament comparable to that of Ichthyostega, such as Sigournea, eroded ornament exposes the bone spongiosa revealing the Haversian systems on the high points (Bolt and Lombard 2006). This is not the case in Ymeria, in which the bone surface is more or less smooth.
A comparable case is seen in specimens of Anthracosaurus russelli, an anthracosaur from the Late Carboniferous of England. An isolated skull table was long referred to as that of an eogyrinid (Panchen 1972) whose surface ornament had been eroded. Further study by one of the current authors showed that this surface texture was not eroded, but rather exactly matched that of the holotype skull of Anthracosaurus, as well as differing from an eogyrinid in a number of other subtle features (Clack 1987). Thus, ornament-type in such cases can carry a taxonomic signal.
The ornament of Ymeria also contrasts with that of Acanthostega. In that genus, ornament is more like that of Ichthyostega, except that the pits are more obviously arranged in a radiating pattern (Ahlberg and Clack 1998; Clack 2002b).
The mandibular sensory line canal of Ymeria is a narrow but open groove except on the splenial, where it is enclosed and opens through a short row of pores. The degree of canal enclosure is identical in MGUH VP 6026 and 6088, supporting the assignation of these two specimens to one taxon.
Comparison with Ichthyostega is problematic. The figured specimens (Jarvik 1996, pls 30, 32–33) appear to differ greatly as to the openness of the canal, suggesting that preservational factors may be at work. Significantly, those of Jarvik’s figures that show open lateral line grooves are photographs either of natural moulds (MGUH 6055, Jarvik 1996, pl. 33, fig. 2) or of latex peels taken from such moulds (MGUH 6147, 6163, 6167, Jarvik 1996, pl. 32, fig. 1, pl. 33, figs 1, 3, 4). Surface-prepared specimens such as MGUH 6084 (Jarvik 1996, pl. 30) and MGUH f.n. 1398 (collected by the Cambridge-Copenhagen expedition and surface prepared by SMF) show pores but no open grooves. This difference is important, because if a negative preparation (whether mechanical or using hydrochloric acid) does not succeed in removing all the bone, small bone remnants can be left between the matrix infill of the sensory line canal and the main body of the matrix external to the jaw, connecting the two and creating a spurious impression of an open sensory line groove when a latex is taken from the specimen. Examination of the natural moulds from which Jarvik’s peels were taken reveals extensive patches of remaining bone associated with many of the canal segments that appear as open grooves in the peels. In these cases, there is a high probability that the apparent open groove is an artefact. However, we cannot state categorically that no open grooves occur in Ichthyostega. Interestingly, the coding of this character in Ichthyostega has a remarkable effect on the phylogenetic analysis (see below).
The dentary is narrow and tapering, with regions of unornamented bone along the anteroventral margin – described as a ‘chamfered’ margin by Ahlberg and Clack (1998). This is seen in several Devonian tetrapod taxa such as Acanthostega and Ventastega. In some Ichthyostega specimens, it seems restricted to the posterior portion of the dentary. This may reflect individual variation or a species distinction characteristic of the Britta Dal and Aina Dal formations (Blom 2005).
The prearticular is partially exposed in both rami, giving an almost complete picture of the bone, except for its anterior end (Figs 1, 2). As in other Devonian tetrapods, it does not suture with the inframandibulars posteriorly. A narrow strip of Meckelian bone is revealed between them in the left ramus, and a section through the Meckelian bone is visible in the broken posterior end of the right ramus. A number of Meckelian foramina lie bounded by the ventral margin of the prearticular and the body of the Meckelian bone, as in Ichthyostega. The nature and length of the suture between the prearticular and the splenial and anterior coronoid unfortunately is not determinable; this area would have carried important information about early tetrapod lower jaw construction.
What can be seen of the surface of the prearticular is essentially plain and appears to lack the radiating striations found in other forms such as Acanthostega and Ventastega, but only faintly developed in Ichthyostega (Jarvik 1996; Ahlberg and Clack 1998). The exception is along the thickened dorsal margin, which bears longitudinal striations as well as a conspicuous and well-defined denticulated field (Figs 1–4). The latter constitutes one of its most obvious differences from Ichthyostega. In the latter genus, there are rarely any denticles on the prearticular. Some specimens (e.g. MGUH VP 6163) show small patches, figured in the reconstruction by Ahlberg and Clack (1998), but these are the exceptions. Closely defined denticle fields are found in Acanthostega, Ventastega and Elginerpeton (Ahlberg and Clack 1998). The second specimen attributed to Ymeria, MGUH VP 6026, differs somewhat from the holotype in the greater anterior extent of its denticulated field. The densest part of the field proved difficult or impossible to prepare owing to the poor separation between bone and matrix; however, the bounds of the field could be established.
The coronoids are narrow and bear a continuous row of teeth, but the sutures between them are difficult to trace (Fig. 2). Coronoids two and three appear to be completely exposed, but the anterior part of coronoid one is inaccessible. Coronoid three contributes a small section of the margin of the adductor fossa.
At the anterior end of the right ramus, the symphysial region and part of the parasymphysial plate have been exposed, along with part of the splenial which sutures with it. However, the parasymphysial region is shown better in MGUH VP 6026. The parasymphysial plate resembles that of Ichthyostega in general terms, but it is less deep and has a proportionately shorter suture with the anterior coronoid. The mesial parasymphysial foramen is larger than that of Ichthyostega, approximately intermediate in size between that genus and Metaxygnathus (Ahlberg and Clack 1998). Its ventral margin carries a broad, shallow, posteroventrally directed gutter that probably transmitted a blood vessel. It is uncertain whether a lateral parasymphysial foramen is present. Overall, the anterior end of the jaw is more slender than in Ichthyostega and shows a distinct dorsal curvature that is absent in the latter genus.
Dentition. All upper marginal teeth are narrowly tapering, with little or no recurvature. Sections through the teeth visible in dorsal view in MGUH VP 6088 show them to be circular with no keels or carinae. Even the smallest teeth have simple ‘labyrinthodont’ infolding at their bases.
The right premaxilla bears 8 teeth with spaces for two or three others, and the left bears 5 with spaces for at least three others. However, the left element is incompletely exposed and the number could be higher, making an estimated total of 10 or 11. The largest teeth on the premaxilla lie towards the posterior margin, just before the bone turns in to accommodate the naris, which is also the case with Ichthyostega. Premaxillary and maxillary teeth are comparable in size in Ymeria.
Premaxillary tooth number in Ichthyostega has proved difficult to determine but generally appears to be between 8 and 10, usually a slightly lower number than Ymeria. The reconstruction of the skull by Jarvik (1996) exaggerates the length of the premaxillary teeth, but they are nonetheless somewhat longer than those of the maxilla (see Fig. 4E, L, M).
The maxilla of Ymeria carries 15 teeth and spaces for 9 more. This total of 24 is a little higher than is found in Ichthyostega, whose maxillary count varies between 16 and 23 (Blom 2005). In Ymeria, the largest teeth in the row lie between positions three and five; otherwise, the teeth gradually diminish in height and diameter posteriorly. In Ichthyostega, the largest teeth on the maxilla lie between positions 5 or 6–9. Ymeria is smaller than the majority of Ichthyostega specimens, but has more teeth on the maxilla. For example, MGUH VP 6123, a small Ichthyostega specimen of a similar size to Ymeria, has no more than 19 maxillary teeth. The total number of upper row teeth in Ymeria is therefore about 34 or 35, compared with Ichthyostega in which it can be between 26 and 33.
On the vomer, one fang is visible both in section in dorsal view and in ventral view, where it projects through the specimen and lies close to the narial portion of the premaxilla. Its replacement pit is represented by a hole through the matrix. More laterally, sections through two or more small teeth are visible. It is not possible to determine whether an anterior transverse vomeral tooth row is present, as in Acanthostega (Clack 1994), Panderichthys (Vorobyeva and Schultze 1991) and Eusthenopteron (Jarvik 1980), or absent as in Ichthyostega (Jarvik 1996). The palatines carry a pair of larger teeth in a row with several smaller teeth, but the exact number cannot be determined. The largest teeth, which are similar in diameter to the largest teeth of the maxillary row, form a distinct pair located at the point where the anterior end of the tooth row bends medially around the choanal margin. These can be identified as the palatine fang pair. The ectopterygoids bear a similarly sized pair in a row with at least eight smaller ones. The larger teeth are placed towards the anterior end of the row. Ectopterygoid teeth in Ichthyostega seem to be variable in size relations and numbers. Some specimens, such as MUGHVP 6055, show no ectopterygoid fangs and have a row of about eight smaller teeth in total (Jarvik 1996, pl. 26, fig. 1A.55). Others, from the same horizon, such as MGUH VP 6070 (Jarvik 1996, pl. 27, fig. 1A.70), show a pair of larger teeth towards the middle of a row of about 10 in total.
There are at least 20 teeth exposed on the right lower jaw ramus of MGUH 6088, though much of the anterior tip is inaccessible. There are spaces for a further 7 or 8 visible, and there were probably about five in the length of dentary ramus obscured by the right maxilla. This constitutes a conservative total of about 32 or 33, approximately similar to the upper row. By contrast, the small Ichthyostega specimen MGUH VP 6123 also has about 32 teeth in the dentary, constituting a discrepancy of five between the number of teeth in the upper and lower rows in that specimen, with the greater number in the lower row. The discrepancy in number reflects a corresponding difference in size and shape between upper and lower sets in Ichthyostega, compared with Ymeria in which they are similar if not identical in both size and shape (Figs 2, 4A, compare with 4E, L, M). This is discussed further below.
The dentary carries a modest symphysial fang, only a little larger than the adjacent marginal teeth, lying medial to the main tooth row. In Ichthyostega, this fang can be very large, and Ahlberg and Clack (1998) show it lying within the main tooth row. The dentition of the parasymphysial plate differs slightly in MGUH VP 6088 and 6026. Both specimens show a parasymphysial fang pair near the anterior end of the bone. However, in MGUH VP 6088, this is followed posteriorly by a short row of at least two small teeth, decreasing in size posteriorly (Fig. 1), whereas MGUH VP 6026 has no such tooth row. MGUH VP 6026, on the other hand, has a minute tooth anterior to the fang pair, which finds no equivalent in MGUH VP 6088. In Ichthyostega, the parasymphysial dentition always consists of a fang pair followed posteriorly by a single smaller tooth.
MGUH VP 6026 shows a diastema between the teeth of the parasymphysial plate and the first teeth on the anterior coronoid, similar to that of Ichthyostega and Whatcheeria (Lombard and Bolt 1995, 2006; PEA, pers. obs. 1998). The corresponding region of the holotype is not visible.
In Ymeria, a total of 22 coronoid teeth plus five spaces makes up a more or less continuous row of teeth (the most anterior tooth that is present is not visible in the figure because it is hidden by overhanging matrix). Most tooth positions are occupied in this specimen. The sutures between the coronoids are obscure, but the junctions can be inferred from the distribution of teeth and subtle deflections of the mesial coronoid margin. The anterior coronoid carries a prominent fang pair, approximately twice the height of the neighbouring teeth. A single marginal tooth can be seen anterior to the fangs, though there may be more as the anterior end of the bone is concealed by matrix; two teeth and one empty socket are present posterior to the fangs, bringing the total count to at least six. On the middle coronoid, the fang pair (consisting of a standing fang and an empty replacement pit) is slightly smaller, but still nearly twice the height of the other teeth. There are three marginal teeth and possibly a tooth-space anterior to the fangs and three posterior, making a total of eight or nine. The posterior coronoid has the greatest number of teeth and the least prominent fang pair, a pattern shared with Ventastega, Metaxygnathus, Panderichthys and some tristichopterids (Ahlberg and Clack 1998). Thirteen tooth positions are present in total. From anterior to posterior, the tooth row comprises two small teeth and two sockets, a larger tooth (about 60 per cent taller than its neighbours), another socket, and seven teeth that decrease gradually in size posteriorly. The tall tooth is clearly one member of the fang pair; the other fang must be represented by the socket anterior or posterior to this tooth, but we cannot determine which one (Figs 2, 4F, G).
Coronoid tooth number and relative size in Ichthyostega are markedly variable. MGUH VP 6123 has 12 and 11 teeth on coronoids two and three, respectively, whereas MGUH 28376 has 5 and 6 present, respectively (Fig. 4F–H). Teeth in this specimen appear to be adjacent at their bases, so no tooth-spaces are counted among that number; possibly, two more should be allowed for coronoid two. That specimen also has a row of seven tiny teeth or denticles on coronoid two anterior to the main tooth row, a feature not so far found elsewhere in the genus. Coronoid fangs are less distinct in Ichthyostega and sometimes cannot be distinguished at all. The relative size of coronoid teeth in Ichthyostega, though variable along the row and from specimen to specimen, shows the largest to be consistently smaller than the dentary teeth.
The smaller coronoid teeth of Ymeria are parallel-sided up to bluntly pointed tips, with the larger teeth being somewhat more conical. However, none is recurved. Crown height of the coronoid row is very similar to that of the marginal rows, smallest and largest on each being comparable in size. This is in strong contrast to the coronoid tooth size in Ichthyostega, in which all coronoid teeth are considerably smaller than the marginal teeth (Figs 2, 4; compare 4A with 4B and 4E with 4F).
Tooth shape as well as size provides one of the strongest contrasts between Ymeria and Ichthyostega. In the latter, teeth are strongly recurved. The marginal teeth are narrowly tapering at their bases, but recurve strongly towards their tips, and in some cases where the crowns are well preserved, have carinae along the posterior edge (e.g. MGUH 28376). The contrast is most marked in the coronoid teeth. Those of Ymeria are essentially simple blunt cones as the marginal teeth, but in Ichthyostega, they are consistently recurved with broad bases, similar to a rose thorn in shape. The shape of the coronoid teeth in Ichthyostega has been confirmed in specimens from both the Aina Dal Formation (MGUH 28376; Fig. 4F–H) and those from the Britta Dal Formation (MGUH VP 6137, 6138: Jarvik 1996 pl. 31, fig. 1 A.138 (NB. in this figure, A.138 is in fact MGUH VP 6137; MGUH VP 6138 is a different individual from a complex of two from the same site; Fig. 4L–N). Specimens from these two horizons have been shown to be different species, I. stensioi from the Aina Dal and I. eigili and I. watsoni from the Britta Dal formations, respectively (Blom 2005).
This shape has also been confirmed in MGUH 28377b. Although the marginal dention of this specimen is heavily eroded, preventing assessment of relative size, coronoid teeth are quite well preserved, freshly revealed by mechanical preparation (Fig. 4J, K). They show the same tooth shape as Ichthyostega specimens from elsewhere, so that the differences from Ymeria are not attributable to a difference in morphology between Ichthyostega from south Celsius and those from other localities or horizons.
Reconstructions of the lower jaw of Ymeria, and a new reconstruction of that of Ichthyostega incorporating new data, are given in Figure 5.
Pectoral girdle. Partial natural moulds and broken bone represent the interclavicle, clavicle and scapulocoracoid. The interclavicle shows a narrowly pointed interclavicular area similar to that of Ichthyostega (Jarvik 1996 pl. 52, fig. 1; we note here that fig. 3 of the same plate shows a lungfish parasphenoid, not a tetrapod interclavicle), though the lateral, striated areas for clavicular overlap seem relatively narrower.
Part of the ventral plate of the clavicle is exposed in section, and part of the ascending process is present in section and as natural mould. The latter shows an anteriorly placed groove and ridge at its base that is also seen in some but not all Ichthyostega clavicles. Judging by the broken cross-section the clavicles carry little or no ornament, another feature that would align them with Ichthyostega. However, this conclusion must be treated with caution as most of the external surface of the clavicle remains unknown.
We performed cladistic analyses on a matrix consisting of 22 taxa and 115 characters, based on that of Callier et al. (2009), using PAUP 4.0.10 and MacClade 3.04. A list of characters and the data matrix are given in the Appendices 1 and 2. Runs were performed with all characters unordered, or with characters 33, 54, 70, 72, 108, 109 and 115 ordered. We also tested the effect of excluding the three least complete taxa (ANSP 21350, known only from an isolated humerus (Shubin et al. 2004); Densignathus and Metaxygnathus, known only from lower jaws). This smaller data set was analysed using the ‘branch and bound’ option in PAUP, whereas the complete data set was subjected to ‘heuristic’ search.
A few character codings were corrected, a procedure that revealed a most surprising instability in the phylogenetic signal. Character 54, the degree of closure of the mandibular lateral line canal, was originally scored ‘0’ for Ichthyostega, indicating an enclosed canal opening to the surface through discrete pores. This reflected Jarvik’s (1980, p. 239) original claim that ‘the membranous sensory canals ran in closed canals in the dermal bones … not in open grooves as in post-Devonian tetrapods’. However, as we discussed above, the specimens figured in Jarvik (1996) suggest a variable degree of canal closure with some sections apparently running in open grooves. Because of the possibility that preparation artefacts might exaggerate the extent of the open grooves, we decided conservatively to re-score Ichthyostega as ‘?’.
This single state change had a dramatic effect on the topology of the Devonian tetrapods (Fig. 6); it caused almost complete loss of resolution in the strict consensus trees, with only a single internal node (placing Ichthyostega crownward to all other Devonian tetrapods) surviving in the all-taxa run and no nodes at all in the run with ANSP 21350, Densignathus and Metaxygnathus excluded. Indeed, the loss of resolution is so great that the resultant polytomy also includes the elpistostegid Elpistostege (all-taxa run) or Elpistostege and Tiktaalik (run with ANSP 21350, Densignathus and Metaxygnathus excluded). The 50 per cent majority rule consensus trees from these runs (Fig. 6C, F) are much more resolved, and in both cases separate the elpistostegids from the Devonian tetrapods, but the internal topologies of the Devonian tetrapods differ substantially from those obtained when Character 54 was scored ‘0’ for Ichthyostega. Notably, Elginerpeton moves from the crownward to the anticrownward end of the Devonian tetrapods; Metaxygnathus and Densignathus both move down-tree and swap relative positions. These runs were performed with all characters unordered, but we also tested running characters 33, 54, 70, 72, 108, 109 and 115 ordered. Ordering had no effect on tree topology, but lengthened the shortest tree by three or four steps depending on the number of taxa included. Certain branching patterns were always present, irrespective of how Character 54 was scored or how many taxa were included: notably, Ventastega, Acanthostega, Ymeria and Ichthyostega always formed a sequence of successively more crownward plesions, and post-Devonian tetrapods always formed a clade with a stable and uncontroversial topology.
The obvious explanation for the dramatic effect of re-coding Character 54 for Ichthyostega is that it uncovered a new island of tree topologies that were previously one step longer than the MPTs and thus not found by the search algorithm. The question is what it says about the underlying phylogenetic signal. It is certainly prudent to code the mandibular canal enclosure of Ichthyostega a ‘?’ rather than ‘0’, and insofar as it allows (and appears to have implemented) optimisation to ‘1’, it is probably also more accurate. Furthermore, the repositioning of Elginerpeton at the bottom of the Devonian tetrapods is indirectly supported by the fact that this is also the earliest known tetrapod (Ahlberg 1995). This might suggest that the runs with the ‘?’ scoring (Fig. 6B–C, E–F) better reflect the phylogenetic signal in the data. However, there are reasons to suspect otherwise. As Callier et al. (2009) noted, the humerus of Ichthyostega has certain primitive character states (presence of the proximal limb of the oblique ventral ridge; absence of a deltopectoral crest; absence of a defined latissimus dorsi process) that contrast with corresponding derived character states in Acanthostega and make it seem strange that the former should be positioned crownward to the latter. Callier et al. (2009) argued that the Devonian tetrapod data set may contain extensive undetected homoplasy, a contention that is certainly not contradicted by the impact of the re-coding described above.
We conclude that Ymeria belongs to the ‘Devonian tetrapod’ segment of the tetrapod stem group and that it is similar to, but not necessarily the sister taxon of, Ichthyostega. This interpretation is fully supported by a non-algorithmic descriptive consideration of the morphology. If we accept the detailed phylogenetic picture presented by our analysis, we can conclude further that Ymeria occupies a position between the less crownward Acanthostega and the more crownward Ichthyostega, but as we have explained above, there are reasons to treat this result with caution. For the present, the interrelationships of Devonian tetrapods are best regarded as unresolved.
For our comparison between Ymeria and Ichthyostega, we have used all the specimens at our immediate disposal that show characters of the dentition and inner surface of the lower jaw. The sample size is unfortunately not large, however given that only two specimens of Ymeria have been recognised, a statistically significant sample of Ichthyostega for comparison would hardly be meaningful. Ymeria, however, falls just outside the observed limit of variability in dental count for Ichthyostega. Moreover, Ymeria shows several other character states that do not occur in Ichthyostega.
Ymeria and Ichthyostega share a rather distinctive shallowly sigmoidal shape to the posterior margin of the lower jaw that is best seen in the illustration but hard to characterise in an analysis (see Fig. 5). They also share a diastema between the parasymphysial tooth row and that of the anterior coronoid. However, while the immediate impression given by Ymeria is that of a very Ichthyostega-like animal, there is no clear evidence that the two form a clade. Ichthyostega is a very distinctive animal, well supplied with autapomorphies (Clack et al. 2003; Ahlberg et al. 2005): many of these, such as the braincase morphology, single median postparietal and regionalised vertebral column, are unknown in Ymeria, but the palate and lower jaw clearly lack distinctive Ichthyostega features such as reduced denticulation and rose thorn-shaped coronoid teeth. The two are also different in several other respects, for example, in the relative sizes of upper and lower teeth. Ichthyostega is notable for the size discrepancy between its larger premaxillary and maxillary teeth and smaller dentary teeth, a discrepancy that is absent in Ymeria.
Ymeria bears a closer resemblance to Ichthyostega than to Acanthostega (Fig. 5D) in terms of tooth number and jaw shape. Jaw shape in Acanthostega is longer and more slender, with a smoother curve at the posterior margin of the angular. In Acanthostega, there are approximately twice as many teeth in each marginal row as well as very many more and smaller coronoid teeth compared to Ymeria and Ichthyostega, so that Acanthostega can be readily distinguished from the other two in the field. As in Ymeria, Acanthostega has a denticulated band along the dorsal margin of the prearticular, but this is a widely distributed and probably primitive character among Devonian tetrapods. Ichthyostega is derived in lacking such a feature. In Acanthostega, the upper teeth are only slightly larger than the lower teeth. Discrepancy between numbers and sizes of upper and lower teeth is not uncommon among early tetrapods, though in some genera, such as Greererpeton and Balanerpeton, the lower teeth are the larger. The reason is presumably related to feeding habits, but how is not clear.
The parasymphysial dentition presents an interesting coding problem among stem tetrapods. At its most complex, for example in Obruchevichthys, it comprises a complete marginal tooth row and a fang pair (Ahlberg and Clack 1998). In Ventastega, the fang pair has been lost, whereas in certain Carboniferous tetrapods such as Megalocephalus, the parasymphysial plate carries a single pair of small teeth that we interpret as the fang pair (Ahlberg and Clack 1998). Ymeria and Ichthyostega have three teeth in a row on the anterior part of the parasymphysial plate, the posterior one being noticeably smaller than the others. Because the fang pairs of the palate and coronoids have been integrated into the tooth row in these genera (rather than lying mesial to the tooth row, as for example in Eusthenopteron), and because the parasymphysial fang pair persists into more derived tetrapods, we make the judgement that the three teeth on the parasymphysial plate of Ichthyostega and Ymeria comprise the fang pair plus a single tooth from the marginal tooth row. By the same token, Ymeria is interpreted as having a parasymphysial fang pair with two posterior marginal teeth (MGUH VP 6088) or one anterior marginal tooth (MGUH VP 6026).
The small size of the Ymeria skull distinguishes it from most Ichthyostega skulls, except that there are some small examples of Ichthyostega such as MGUH VP 6005, formerly known as Ichthyostegopsis. Unfortunately, this specimen retains very little dentition and so cannot be compared with Ymeria. However, one small Ichthyostega, from North Celsius, is of a similar size to Ymeria, but has fewer teeth (MGUH VP 6123). In addition, the small, presumed sub-adult, Ichthyostega specimens (MGUH 28377) recovered from south Celsius Bjerg unequivocally show the recurved coronoid teeth of other Ichthyostega specimens. Therefore, it is unlikely that Ymeria simply represents a juvenile Ichthyostega. There is insufficient data to establish whether MGUH 28377 resembles one of the three valid species of Ichthyostega in skull proportions; however, recent fieldwork has suggested that the specimen most likely derives from the Britta Dal Formation (HB, JEA Marshall, pers. obs.).
Coronoid tooth shape, their relative size, and the distribution of denticles on the prearticular constitute the most obvious differences between Ymeria and Ichthyostega; the low-profile dermal ornament and open lateral line sulcus (probably) are also distinguishing features of the former.
Ymeria is the third genus and fifth species of Devonian tetrapod from the Famennian of North-East Greenland, maintaining the area as the richest for named Devonian tetrapod taxa so far known. The Red Hill tetrapod fauna from Pennsylvania may have included more distinguishable forms, but they are known from isolated bones and the number of separate taxa cannot be verified (Daeschler et al. 2009). Ymeria appears to be a stem group tetrapod of the same ‘grade’ as Ichthyostega, though less obviously autapomorphic. The limited material of Ymeria does not allow us to assess its postcranial morphology in relation to the extensive postcranial differences documented between Ichthyostega and Acanthostega (Coates 1996; Ahlberg et al. 2005). However, the presence of an Ichthyostega-like stalked interclavicle suggests that the postcranium of Ymeria may have borne some resemblance to Ichthyostega. The pronounced difference in tooth shape between Ymeria and Ichthyostega hints at an ecological separation between the two genera although we cannot as yet speculate what this difference might have been. Acanthostega, Ichthyostega and Ymeria, found in contemporaneous deposits in East Greenland, demonstrate that by the Famennian, diversity among tetrapods was already substantial. Acanthostega and Ichthyostega in particular represent highly divergent lifestyles and ecologies, especially evident following the revelation of the bizarre ear and axial morphology of Ichthyostega (Clack et al. 2003; Ahlberg et al. 2005). Recognition of Ymeria provides another piece in the jigsaw puzzle that is gradually transforming the story of the origin of tetrapods from a linear succession of ‘missing links’ to one of evolutionary radiation and diversity at the interface between water and land.
Acknowledgements. We dedicate this paper to Meemann Chang in recognition of the outstanding contribution she has made to vertebrate palaeontology over many years. Thanks go to the staff of the Geological Museum in Copenhagen for permission to prepare and study material of Devonian tetrapods and fishes over a long timescale.