Introduction
The Pennsylvanian Jarrow coal seam from northern County Kilkenny, Ireland, preserves a diverse assemblage of plant, fish, and tetrapod fossils of historical and paleontological significance. The fauna is the stratigraphically oldest of the Pennsylvanian Coal Measure faunas, which include those of Linton, Ohio, USA, and Nýřany, Czech Republic, and it was the first to be described taxonomically. Higgs and O'Connor (Reference Higgs and O'Connor2005) recovered miospores from the coal shales of Langsettian regional substage age, which has been shown by Pointon et al. (Reference Pointon, Chew, Ovtcharova, Sevastopulo and Crowley2012, fig. 1) to equate to the later part of the Bashkirian International Stage. Huxley in Huxley and Wright (Reference Huxley and Wright1867) gave the first formal taxonomic account of tetrapods from Jarrow and described seven new species: the aïstopods Dolichosoma emersoni (Huxley in Huxley and Wright, Reference Huxley and Wright1867), now Dolichosomatites emersoni (Kuhn, Reference Kuhn1961) and Ophiderpeton brownriggii (Huxley in Wright and Huxley, Reference Wright and Huxley1866); the temnospondyl Erpetocephalus rugosus (Huxley in Huxley and Wright, Reference Huxley and Wright1867), renamed Dendrepeton rugosum (Milner, Reference Milner1980); the nectrideans Keraterpeton galvani (Huxley in Wright and Huxley, Reference Wright and Huxley1866), redescribed by Milner (Reference Milner2019), Lepterpeton dobbsii (Huxley in Wright and Huxley, Reference Wright and Huxley1866) and Urocordylus wandesfordii (Huxley in Wright and Huxley, Reference Wright and Huxley1866), reviewed in Bossy and Milner (Reference Bossy, Milner and Wellnhoer1998); and Ichthyerpeton bradleyae (Huxley in Wright and Huxley, Reference Wright and Huxley1866), the subject of this paper.
I. bradleyae was described from a single specimen, which consists of the trunk, hindlimbs, and caudal region. As is the case of some of the other six tetrapods, I. bradleyae was first briefly described and remarked upon in Etheridge (Reference Etheridge1866) and Wright and Huxley (Reference Wright and Huxley1866) before the formal description by Huxley in Huxley and Wright (Reference Huxley and Wright1867). As a result, the genus Ichthyerpeton first appeared in the literature in Etheridge (Reference Etheridge1866), despite its name being credited to Huxley by Etheridge. The original descriptions focused on the presence of “horny scales” (Huxley in Etheridge, Reference Etheridge1866, p. 4) covering the body, including most of the hindlimb (Wright and Huxley, Reference Wright and Huxley1866), the morphology of the vertebrae with “discoidal” centra (Huxley and Wright, Reference Huxley and Wright1867), and the number of digits on the pes, which varied from three (Huxley in Etheridge, Reference Etheridge1866) to five (Wright and Huxley, Reference Wright and Huxley1866).
Additional specimens from Jarrow were identified as I. bradleyae by Lydekker (Reference Lydekker1890). Lydekker (Reference Lydekker1891a) regarded Erpetocephalus rugosus (Huxley in Huxley and Wright, Reference Huxley and Wright1867), which was originally described from a single cranium, to be a junior synonym of I. bradleyae. This was based on an additional unfigured specimen of E. rugosus from Jarrow with vertebrae similar to I. bradleyae; this specimen has not been confidently identified (see the following). However, Lydekker's taxonomic treatment has not been generally accepted (Milner, Reference Milner1980). Lydekker (Reference Lydekker1891a) placed I. bradleyae in the temnospondyl family Dendrerpetontidae and identified a second species from an isolated skull and mandible from Jarrow, later named Ichthyerpeton hibernicum (Lydekker, Reference Lydekker1891b, p. 343).
A third species of Ichthyerpeton, I. squamosum (Moodie, Reference Moodie1909), was described from Linton, Ohio, but was later identified as an aïstopod (Romer, Reference Romer1930). Romer (Reference Romer1947) considered I. bradleyae to be an embolomere and suggested that there may be an affinity between it and the anthracosaur vertebrae described from Jarrow (Baily, Reference Baily1879, Reference Baily1884). If this is the case, the holotype of I. bradleyae would represent an immature form of an animal similar to the anthracosaur Pteroplax (Boyd, Reference Boyd1980). An assignment to the anthracosaurs was tentatively suggested by Panchen (Reference Panchen and Kuhn1970), who noted that the long body with short hindlimbs was anthracosaur-like. The skulls of E. rugosus and I. hibernicum were redescribed by Milner (Reference Milner1980), who placed them in the temnospondyl genus Dendrerpeton as the species D. rugosum. Milner (Reference Milner1980) noted that because of the apparently poor preservation of the holotype of I. bradleyae, the identification of other specimens later assigned to the taxon was unwarranted and restricted the binomen to the holotype, regarding it as a nomen dubium. The presence of a long presacral column and the similarity of the holotype to specimens of colosteids from Jarrow led Milner (Reference Milner1980) to suggest that I. bradleyae might be a colosteid or a temnospondyl. Clack and Milner (Reference Clack, Milner and Sues2015) also assigned I. bradleyae to the colosteids, although this was on the basis of additional specimens from Jarrow, and not from evidence derived from the holotype. The association of colosteid material from Jarrow and I. bradleyae is disputed here.
A common characteristic of tetrapod specimens from the Jarrow assemblage is that despite the fully to partially articulated bodies and scale/skin impressions, preservation of detail is poor. Specimens are commonly highly compressed, and a layer of matrix coating them results in difficulty identifying the outlines of the bones, partly because the latter are replaced to a greater or lesser degree by carbonaceous material (Rayner, Reference Rayner1971), similar to the rock matrix in which they are preserved. Manual preparation of the holotype of I. bradleyae would have been perilous because of the dense “coating” of scalation throughout the vertebral column. To elucidate further the anatomy of the holotype of I. bradleyae, the holotype was scanned using microcomputed tomography to visualize the vertebral column and associated elements surrounding it. This has allowed a comprehensive redescription of the holotype, which will contribute to further comparison between it and other material attributed to I. bradleyae.
Materials and methods
The holotype of Ichthyerpeton bradleyae (TCD.T87) was scanned at the X-ray tomography (XTM) facility at the University of Bristol.
Microcomputed tomography parameters and three-dimensional rendering
Microcomputed tomography was carried out using a Nikon XTH 225 ST computed tomography scanner. The scan was run at 220kV and 375uA with a 1 mm plate of Cu in front of the source.
Three-dimensional models of TCD.T87 were rendered using the freeware SPIERS (Sutton et al., Reference Sutton, Garwood, Siveter and Siveter2012). Individual elements were isolated using the masking function.
Systematic paleontology notations
The notation “v” in the synonymy list follows Matthews (Reference Matthews1973) to show that the senior author has studied the specimens thus indicated.
Repositories and institutional abbreviations
Specimens referred to below are deposited in the following institutions: Natural History Museum London, United Kingdom (catalogue numbers prefixed by NHMUK); National Museum of Ireland (NMI); and the Geological Museum, Trinity College Dublin (TCD).
Systematic paleontology
Infraclass Tetrapodomorpha Ahlberg, Reference Ahlberg1991
Genus Ichthyerpeton Huxley in Etheridge, Reference Etheridge1866
Type species
Ichthyerpeton bradleyae Huxley in Wright and Huxley, Reference Wright and Huxley1866
Diagnosis
As for type species by monotypy.
Ichthyerpeton bradleyae Huxley in Wright and Huxley, Reference Wright and Huxley1866
Figures 1–5
- v. Reference Etheridge1866
Ichthyerpeton sp.; Huxley in Etheridge, Reference Etheridge1866, p. 4.
- v. Reference Wright and Huxley1866
Ichthyerpeton bradleyae Huxley in Wright and Huxley, Reference Wright and Huxley1866, p. 168.
- v. Reference Huxley and Wright1867
Ichthyerpeton bradleyae; Huxley and Wright, Reference Huxley and Wright1867, p. 367, pl. 23, fig. 1.
- v. Reference Lydekker1890
Ichthyerpetum bradleyae; Lydekker, Reference Lydekker1890, p. 344, non Wright and Huxley, Reference Wright and Huxley1866.
- v. Reference Lydekker1891a
Ichthyerpetum bradleyae; Lydekker, Reference Lydekker1891a, p. 58.
- v. Reference Romer1947
Ichthyerpeton bradleyae; Romer, Reference Romer1947, p. 266.
- v. Reference Panchen and Kuhn1970
Ichthyerpeton bradleyae; Panchen, Reference Panchen and Kuhn1970, p. 63, as nomen vanum.
- v. Reference Milner1980
Ichthyerpeton bradleyae; Milner, Reference Milner1980, p.135, as nomen dubium
- v. Reference Clack, Milner and Sues2015
Ichthyerpeton bradleyae; Clack and Milner, Reference Clack, Milner and Sues2015, p. 48, non Wright and Huxley, Reference Wright and Huxley1866.
Figure 1. Rendered model of TCD.T87 Ichthyerpeton bradleyae in (1) lateral view with scalation, (2) without scalation, and (3) a drawing of the rendered model. In the rendered model, scalation is light gray while postcranial bones are dark gray. In the drawing, elements/features are given different shades of gray ranging from darkest to lightest in the following order: ribs, hemal arch, impressions of phalanges, neural arches, centra, appendicular elements, scalation. c = centrum; fe = femur; ha = hemal arch; ?il = ilium; mt = metatarsal; na = neural arch; ?na = probable neural arch; ph = phalanx; r = rib; ?ta = ?tarsal; ti = tibia. Centra are numbered in the figure starting at 1 on the anteriormost centrum, with every fifth centra marked posteriorly. Scale bar = 50 mm.
Figure 2. Axial elements from TCD.T87 Ichthyerpeton bradleyae, including (1, 2) centra 22–28 in lateral view, (3–6) centrum 41 in articular view, (7, 8) centrum 48 in articular view, (9, 10) centrum 47 in articular view, (11, 12) neural arch articulated to centrum 39 in lateral view, (13, 14) hemal arch in lateral view, (15, 16) rib head in ventral view, and (17, 18) a disarticulated rib. ap = anterior process; c = centrum; cap = capitulum; ?dia = diapophyses; dc = depression on centrum; de = distal end; ha = hemal arch; hs = hemal spine; ic = intercentral gap; na = neural arch; nf = notochordal foramen; ns = neural spine; r = rib; ve = ventral emargination. Scale bar = 10 mm.
Figure 3. Putative ilium from TCD.T87 Ichthyerpeton bradleyae (1, 2) in lateral view. Dashed line indicates where the ilium is considered to be incomplete. ib = iliac blade; pip = posterior iliac process. Scale bar = 5 mm.
Figure 4. Left femur and tibia from TCD.T87 Ichthyerpeton bradleyae. (1–10) The femur in (1, 2) ventral, (3, 4) dorsal, (5, 6) ventro-anterior, (7, 8) dorso-anterior, and (9, 10) distal views. (11–14) The tibia in (11, 12) extensor and (13, 14) dorsal views. adb = adductor blade; cn = cnemial crest; ff = fibular facet; fh = femoral head; ftt = fourth trochanter; icf = intercondylar fossa; int = internal trochanter; itf = intertrochanteric fossa; tf = tibial facet. Scale bar = 10 mm.
Figure 5. Pes elements from TCD.T87 Ichthyerpeton bradleyae. (1) The complete body with location of phalanges marked with a box. (2, 3) Phalanges: (2) from the left limb, with impressions marked in a darker shade of gray, and (3) from the right limb. fe = femur; iph = impression of phalanges; mt = metatarsal; ph = phalange; ?ta = tarsal; ti = tibia; un = ungual. (1) Scale bar = 50 mm; (2, 3) scale bars = 10mm.
Holotype
Specimen TCD.T87, consisting of articulated vertebral column from the posterior trunk, the sacral and the anterior caudal regions, and hindlimbs with articulated scalation throughout (Huxley and Wright, Reference Huxley and Wright1867, p. 367, pl. 23, fig. 1.)
Diagnosis
Elongated tetrapod with diplospondylous vertebrae consisting of cylindrical centra that are dorsoventrally tall with the notochordal canal closed. Neural spines are the same height as the centra. Tibial length is 80% of femoral length. Thick covering of scalation with ventral trunk gastralia, dorsal trunk ovoid scales, and rhombic caudal scales.
Occurrence
Jarrow (No. 4) Coal Seam, Coolbaun Coal Formation, (Langsettian, Pennsylvanian), Clogh, County Kilkenny, Ireland.
Description
The holotype (TCD.T87) is preserved in a rectangular block of coaly shale, which has been broken anteriorly through the trunk region and posteriorly through the caudal region so that only parts of the trunk, sacral, and caudal regions are present. The left lateral side is 240 mm long and, as a result of postmortem effects, is arched dorsally midway along its length, giving it a crescentic shape (Fig. 1). The articulated axial skeleton is laterally compressed, whereas the body outline, distinguished from the scalation, is dorsolaterally flattened.
Axial skeleton
TCD.T87 preserves 50 centra, six neural arches, one hemal arch, and an additional element that may be a neural arch. Centra are here numbered from anterior to posterior. Centra 1–40 are articulated and span the trunk, sacral, and anterior caudal regions (Fig. 1). Centrum 1 is broken anteriorly by the break in the slab, and the vertebral column would have continued anteriorly from this break. Centra 42–46 are preserved in articulation but are separated from the main vertebral column. There are an additional five disarticulated centra, centra 41 and 47–50. The vertebrae of I. bradleyae are diplospondylous with cylindrical, gently amphicoelous centra. Little variation between the centra, coupled with the lack of preserved neural arches and haemal arches throughout most of the column, makes differentiation of trunk, sacral, and caudal vertebrae difficult. Sacral centra are very tentatively identified around centra 30–34 on the basis of the presence of the limb elements and the limb outline in the scalation ventral of these centra. It is unclear how many vertebrae formed the sacral column. Centra posterior of centrum 34 are considered caudal centra because they are in the caudal region (Fig. 1.3).
Centra have a rectangular profile in lateral view with a height-to-length ratio of ~2:1. Length is not uniform throughout each centrum, with some of the centra being shorter toward the ventral margin. This creates an “intercentral gap” seen between centra 20 and 21, 22 and 23, 24 and 25, and 26 and 27 (Fig. 2.1, 2.2) No bone has been found in the intercentral gaps. In lateral view, the ventral margin of the centra has an emargination, seen in centrum 24 (Fig. 2.1, 2.2), which gives it a W-shape. Subcircular depressions are located in the center of the lateral faces of centra 13–15 (Fig. 1.1) and 22–28 (Fig. 2.1, 2.2). While most of these depressions are interpreted as simple depressions on the lateral walls of the centra, the depressions on centra 22, 24, and 26 are distinct, deep, and circular and may represent the diapophysis. However, this is uncertain as no ribs are observed articulating to these depressions along the articulated vertebral column. Centra are subovoid in articular view with a narrow closed notochordal canal present in the center of the articular face (Figs. 1, 2.3–2.10). In centra 47, 49, and 50, there are two prominent dorsal facets to receive the neural arch (Fig. 2.9, 2.10).
Intercentra and pleurocentra are difficult to distinguish from one another given that there is little morphological and size variation among centra in the articulated vertebral column. The presence of an intercentral gap between even- and odd-numbered centra may suggest that for centra 1–38 all even-numbered centra are intercentra, while all odd-numbered centra are pleurocentra, as a similar gap is seen between the intercentra and pleurocentra of other tetrapods (see Discussion). No intercentral gap is discernable among centra 42–45, making their homology uncertain. The presence of distinct neural arch facets in centra 47, 49, and 50 indicates that these centra are pleurocentra. Neural arch facets cannot be made out from the model of the articulated vertebral column. As centrum 41 has part of a rib articulated to it (Fig. 2.3–2.6), it is considered an intercentrum.
Six neural arches are preserved dorsal to centra 15, 18, 19, 25, 39, and 48. The morphology of the neural arches in lateral view is difficult to discern; the zygapophyses cannot be seen on any of the arches. The bases of the neural arches dorsal of centra 18 and 19 are long, with the base of the neural arch on the latter extending posteriorly onto centrum 18 (Fig. 1). In articular view, the neural arch and spine of centrum 48 is paired, with the spines only in contact dorsally (Fig. 2.7, 2.8). In the caudal vertebrae, the neural arches were paired but, it is unclear whether this was also the condition in trunk vertebrae. In lateral view, neural spine morphology varies, but this is probably a result of the altered nature of the specimen and the difficulty of identifying the border between bone and matrix in the CT slices. The neural spine associated with centrum 15 is curved 90° posteriorly with a smooth anterior and posterior surface, so it resembles a hook (Fig. 1). The morphologies of the two neural spines dorsal of centra 18 and 19 are similar. Both have ventrally expanded bases of the neural arch with the neural spine relegated to the posterior half of the arch and the neural spines of each extend dorsally with a rounded apices. The neural spine dorsal of centra 25 extends posterodorsally in a straight line with a rounded apex (Fig. 2.1, 2.2). The morphology of the neural spine dorsal of centra 39 is different again. It extends dorsally from the arch for under half its length until this projection turns posterodorsally with a triangular anterior process (Fig. 2.11, 2.12).
A single hemal arch and spine has been identified attached to centra 41 (Fig. 2.3–2.6), which is considered an intercentrum. Here the arch wraps around the keel of the centrum. The hemal arch and spine appear to be unpaired. An additional element is preserved anterior of centrum 41, which has a similar length along its base to the length of the neural arch dorsal of centrum 18. It is probable that this element is also a neural arch.
Nineteen ribs have been identified, all of which are incomplete, and the majority, which are found mostly ventral of the trunk vertebral column, preserve only part of the proximal shaft (Fig. 1.2, 1.3). The longest rib preserved is ventral of centrum 16. Its length is approximately the length of three centra. Huxley (in Huxley and Wright, Reference Huxley and Wright1867, pl. 23. fig. 1) figured ridges extending posteroventrally from centra 8–15, extending as far ventrally as the rib ventral of centrum 16. These probably represent ribs, but they cannot now be identified either on the surface of the specimen or in the CT slices. The only rib head is dorsal of centrum 22. It is expanded relative to the shaft and has a shallow depression toward the center, which marks the articulation facet. The only articulated rib is seen in centrum 41, which has a rib articulated to the midpoint of the lateral centrum wall.
Appendicular skeleton.—Posterodorsal of centrum 39 is a tall, blade-like bone that is poorly defined from the CT slices (Fig. 1). This is tentatively identified as part of the ilium. The height of this bone is approximately the length of four centra. It is widest along the ventral margin, which is considered to be incomplete and has a process projecting anteriorly (Fig. 3). A shallow depression is seen near the base on the lateral surface. The iliac blade is thin, long, and makes up half of the dorsal height anteriorly. A unipartite dorsal process extends dorsally from the posterior half of the blade, which gives the posterior margin of the ilium a gently sigmoidal shape. The dorsal process is half the length of the ilium neck and is rounded at the apex.
A left femur is preserved ventral of centra 30–35 (Fig. 1). The femur is stouter than typical for early tetrapod femora (Warren and Turner, Reference Warren and Turner2004), with an anterior–posterior width to proximal–distal length ratio of ~1:2.1. It is expanded both distally and proximally and waisted along the shaft (Fig. 4.1–4.4). The proximal end of the femoral head is concave along the ventral side. The intertrochanteric fossa is deep, with a crest forming along its distal margin (Fig. 4.5, 4.6). The adductor blade is exposed on the anterior side of the femur in both dorsal and ventral views. The blade is approximately 22% the length of the femur. Proximally, it terminates at the internal trochanter, with a groove there separating the blade from the femoral head (Fig. 4.7, 4.8.). No pit is observable on the internal trochanter; this is not surprising given the quality of preservation of the specimen. Distally, the adductor blade thickens and juts out anteriorly, forming the fourth trochanter. An adductor crest is absent. The femoral shaft is approximately half the width of the femoral head. There is no rotation between the femoral head and the distal fibular and tibial condyles along the shaft (Fig. 4.1–4.4). The fibular and tibial condyles are separated by a deep intercondylar fossa, with an angle of ~85° between the two (Fig. 4.9, 4.10). The fibula condyle is thicker and broader than the tibial condyle. Both extend to the same level distally.
A tibia lies ventral to the femur (Fig. 1). It is a broad, dorsoventrally flat bone (Fig. 4.11–4.14). The tibia has expanded proximal and distal ends with no rotation along the shaft between them. The proximal margin is slightly convex and oriented at an angle of ~45° to the tibial shaft. A low cnemial crest extends along the anterior margin of the extensor surface from the proximal femoral facet to over a third of the tibial length (Fig. 4.11–4.14). Between the cnemial crest and the posterior margin is a shallow depression. The posterior margin is concave, which would allow for an interepipodial space between the tibia and the fibula. The distal end is less expanded than the proximal end, with a concave margin orientated at ~75° to the tibial shaft. The distal facet does not curve around onto the posterior margin. On the extensor surface, the proximal end of the tibia is ventrally thicker, relative to the distal end, with a low ridge running distally (Fig. 4.13, 4.14).
Disarticulated pedal elements are located in two separate areas, one ventral of the femur and tibia in a ventral projection of the scales and the second in the caudal region spanning to outside the scale outline (Fig. 5.1). The anterior of the two is considered to represent the left foot as it is preserved on the left side of the body outline, while the posterior collection is considered to be that of the right foot as they are preserved on the right side of the body outline. Three phalanges are preserved on the anterior left foot, with the posteriormost phalanx broken in half where it exits the scalation (Fig. 5.2). Anteroventral of the phalanges are four hollow impressions of the digits. It is difficult to count the number of phalanges from these impressions. The anterior digit has at least two. The next digit posteriorly has at least three. Directly posterior to this digit is a slimmer impression that is probably a digit, but individual phalanges cannot be identified. The posteriormost digit contains an impression for the broken half of the phalanx mentioned in the preceding, together with two other phalanges, one of which is an ungual. Huxley and Wright (Reference Huxley and Wright1867, pl. 23, fig. 2) depicted the hollow impressions of the digits with the anterior digits having two, five, two, and three phalanges (from anterior to posterior), all of which had the distal phalange as an ungual. Huxley reported a phalangeal count of 3–3–3–3 (Huxley and Wright, Reference Huxley and Wright1867), but it is not clear whether these digits were present as body fossils or were interpreted from the impressions.
The second collection of pedal elements, found in the caudal region, contains two metatarsals (Fig. 5.3). Both have expanded proximal and distal ends that are double the width of the shaft in the ventralmost metatarsal. In this region, there are 15 phalanges, three of which are unguals. At least two digits are partially preserved in articulation, consisting of two and three phalanges. Four bones that do not have the morphology of phalanges are interpreted as tarsals, but their position within the foot is not clear. The dorsalmost tarsal has a crescentic shape.
Scalation
TCD.T87 is covered in tightly packed scales throughout its length. Details of scalation in the trunk region are difficult to discern (Fig. 1.1, 1.3). Dorsal of the vertebral column, scales appear to be subovoid, but alteration has distorted their boundaries, which cannot be identified from the CT slices. Ventral scalation also presents similar problems, but it is apparent that there are tightly packed thin gastralia along the underside. Anterior to the caudal region, a scaly outline protrudes from the ventral margin in which phalanges are found, suggesting part of the hindlimbs bore a thick covering of scales. The femur and tibia are preserved on the outside of this scaly outline, indicating that not all of the hindlimb was covered. The caudal scales are difficult to discern in the CT slices but are clearly seen by the naked eye. They are rhombic and are arranged in diagonal columns with scales overlapping succeeding scales along their posterior margin. Striations originating from the anteroventral edge of the scales cover the surface in some scales. This suggests that in life, the posterior trunk region, the sacral and caudal regions, and part of the limbs of I. bradleyae were covered in a thick cover of scales, leaving little to no exposed skin.
Remarks
The description of Ichthyerpeton bradleyae is based exclusively on the holotype (TCD.T87).
Discussion
This study provides a description of the holotype of Ichthyerpeton bradleyae to compare with putative specimens of I. bradleyae material within the collections from the Jarrow Assemblage. As a result, only one additional specimen is confirmed as being conspecific with the holotype. It is TCD.38351, which preserves approximately 23 articulated vertebrae, scalation outline, and a possible hindlimb. The specimen is fragile, is preserved in a thin horizon of coaly shale, and, with two cracks running through the vertebral column, is in a more damaged state than the holotype. TCD.38351 does not provide any additional characters for I. bradleyae.
NHMUK R. 8453–8455 and 8458–8459 were assigned to I. bradleyae by Lydekker (Reference Lydekker1890) (Milner, Reference Milner1980). NHMUK R. 8453 (a skull) has since been described as the temnospondyl Procochleosaurus jarrowensis (Sequeira, Reference Sequeira1996). NHMUK R. 8454, 8455, and 84588, which preserve skulls and postcranial bodies of a probable single taxon, were also attributed to I. bradleyae by Panchen (Reference Panchen and Kuhn1970) and Clack and Milner (Reference Clack, Milner and Sues2015). These identifications are considered doubtful because the vertebrae of these specimens preserve centra with large notochordal foramina and NHMUK R. 8455 shows crescentic centra. They are probably colosteids, which also have been identified from the Jarrow Assemblage (Milner, Reference Milner1987; Sequeira, Reference Sequeira1996). NHMUK R.8459 was also considered to be associated with I. bradleyae by Panchen (Reference Panchen and Kuhn1970) and Clack and Milner (Reference Clack, Milner and Sues2015), but this appears to be the vertebral column of a finned tetrapodomorph such as Megalichthys, which is common in the Jarrow Assemblage. NHMUK R. 8463 was attributed to I. bradleyae by Clack and Milner (Reference Clack, Milner and Sues2015). This specimen preserves the articulated skeleton of a tetrapod with a poorly defined skull. The centra appear to be elongated and spool-shaped, making an affinity between the specimen and I. bradleyae doubtful.
Lydekker (Reference Lydekker1891a) considered the holotype of “Erpetocephalus rugosus” (TCD.T84) to be a specimen of I. bradleyae, an association that was based on a second specimen in the NMI collection that Lydekker (Reference Lydekker1891a) mentioned preserved the cranium and part of the vertebral column, similar to I. bradleyae. The second specimen was not figured, and no museum number was provided. It might be either NMI.F14717 or NMI.F16890. NMI.F14717 preserves the skull of a tetrapod in ventrolateral view and with robust lower jaws and possibly articulated vertebrae, but these are difficult to identify. The specimen is about the size of the skulls of Dendrerpeton from Jarrow, but without further preparation, no positive identification can be made. NMI.F16890 preserves an impression of a tetrapod skull in ventral view with articulated vertebrae. Details of the vertebrae are difficult to discern; however, they do not resemble those of I. bradleyae. If neither specimen is the second specimen referred to by Lydekker (Reference Lydekker1891a), it is probable that said specimen is missing or has been destroyed, as no other specimens in the NMI collection fit the description. Lydekker (Reference Lydekker1891b) named I. hibernicum from a single specimen, |NMI.F14713. Milner (1980) described the holotype of both “E. rugosus” (TCD.T84) and I. hibernicum (NMI.F14713) as the new species Dendrerpeton rugosum, rendering I. hibernicum a junior synonym. I. squamosum from Linton, Ohio, was described from two articulated vertebral columns (Moodie, Reference Moodie1909), but was later identified by Romer (Reference Romer1930) as the aïstopod Oestocephalus (Anderson, Reference Anderson2003), which has seniority. Additional work needs to be done on the preceding undescribed tetrapod specimens to resolve their relationships and to investigate properly their affinity with I. bradleyae. Ichthyerpeton is here considered to be a monospecific genus restricted to two specimens from the Jarrow Assemblage.
Huxley's original description of Ichthyerpeton bradleyae mentions an “animal whose scaly integument and laterally compressed, fin-like, tail might easily lead one to take it for a fish, were not its true position among higher Vertebrata settled at once by the digitated hind limb” (Huxley in Huxley and Wright, Reference Huxley and Wright1867, p. 367). Here, this placement is further supported for I. bradleyae, which was a stout-limbed tetrapod with an elongated body similar to that seen in Crassigyrinus (Panchen and Smithson, Reference Panchen and Smithson1990; Herbst and Hutchinson, Reference Herbst and Hutchinson2019), colosteids (Hook, Reference Hook1983; Godfrey, Reference Godfrey1989), and tupilakosaurid temnospondyls (Warren, Reference Warren1999); however, its relationship to other tetrapods is hard to establish. Characters of I. bradleyae are compared with those of other early tetrapods in Table 1.
Table 1. List of axial, appendicular, and scalation characters of Ichthyerpeton bradleyae compared with other early tetrapods. Information obtained for the other tetrapods from the following: Acanthostega (Coates, Reference Coates1996; Pierce et al., Reference Pierce, Ahlberg, Hutchinson, Molnar, Sanchez, Tafforeau and Clack2013); Crassigyrinus (Panchen, Reference Panchen1985; Panchen and Smithson, Reference Panchen and Smithson1990; Clack and Finnay, Reference Clack and Finney2005; Herbst and Hutchinson, Reference Herbst and Hutchinson2019); Greererpeton (Godfrey, Reference Godfrey1989; Clack and Finnay, Reference Clack and Finney2005); Dendrerpeton (Holmes et al., Reference Holmes, Carroll and Reisz1998); Thabanchuia (Warren, Reference Warren1999); Keraterpeton (Milner, Reference Milner2019); and Proterogyrinus (Holmes, Reference Holmes1984; Clack and Finnay, Reference Clack and Finney2005). Where the status of a character could not be found in the literature, it is marked by the symbol “—”.
The axial skeleton of I. bradleyae is composed of diplospondylous vertebrae with dorsoventrally tall, cylindrical, amphicoelous centra (Table. 1), similar to the vertebral arrangement in embolomeres (Clack, Reference Clack1987), tupilakosaurids (Warren, Reference Warren1999; Werneburg et al., Reference Werneburg, Sébastien Steyer, Sommer, Gand, Schneider and Vianey-Liaud2007), and Acherontiscus (Carroll, Reference Carroll1969; Clack et al., Reference Clack, Ruta, Milner, Marshall, Smithson and Smithson2019). This gives the centra of I. bradleyae a rectangular shape in lateral view, as seen in the embolomeres Pholiderpeton (Clack, Reference Clack1987), Pteroplax (Boyd, Reference Boyd1980), and Archeria (Holmes, Reference Holmes1989). Not all embolomeres have embolomerous vertebrae, but when present, the articular faces of the intercentra are convex, most notable in Eogyrinus (Panchen, Reference Panchen1977). This is not seen in the intercentra of I. bradleyae, which have gently concave articular faces. The notochordal canal is closed, at least in the caudal vertebrae of I. bradleyae, a feature that is also present in the centra of some embolomeres, including Pholiderpeton (Clack, Reference Clack1987), Eobaphetes, and possible Anthracosaurus material from Airdrie, Scotland (Panchen, Reference Panchen1966, Reference Panchen1977) but is absent in Eogyrinus (Panchen, Reference Panchen1977) and Carbonoherpeton (Klembara, Reference Klembara1985), the latter of which has relatively large notochordal foramen. Acherontiscus has crescentic inter- and pleurocentra in the cervical region, but by the twenty-fourth vertebra they are cylindrical with a closed notochordal canal similar to I. bradleyae (Carroll, Reference Carroll1969; Clack et al., Reference Clack, Ruta, Milner, Marshall, Smithson and Smithson2019). However, unlike I. bradleyae but similar to embolomeres, the centra of Acherontiscus have deep pits along their lateral and ventral surfaces.
I. bradleyae lacks a notable size difference between intercentra and pleurocentra. In embolomeres and Acherontiscus, pleurocentra are larger than intercentra (Carroll, Reference Carroll1969; Panchen, Reference Panchen1977; Holmes, Reference Holmes1989). Intercentra and pleurocentra of similar size are seen in some tupilakosaurids, although intercentra in Tupilakosaurus wetlugensis (Shishkin, Reference Shishkin1961) are slightly larger than pleurocentra (Shishkin, Reference Shishkin1989; Werneburg et al., Reference Werneburg, Sébastien Steyer, Sommer, Gand, Schneider and Vianey-Liaud2007). While the centra of tupilakosaurids have closed notochordal canals (Yates and Warren, Reference Yates and Warren2000), they have a higher degree of concavity on the articular face compared with I. bradleyae. The intercentra of tupilakosaurids have a depression along the dorsal surface (Warren, Reference Warren1999; Yates and Warren, Reference Yates and Warren2000; Werneburg et al., Reference Werneburg, Sébastien Steyer, Sommer, Gand, Schneider and Vianey-Liaud2007), which is seen in centrum 47 of I. bradleyae (Fig. 2.9, 2.10) but is not seen throughout the rest of the vertebral column. No spinous process is seen toward the ventral end on any centra in I. bradleyae. In some tupilakosaurids, a spinous process is present toward the ventral end of the intercentra, which extends posteriorly in Tupilakosaurus wetlugensis and anteriorly in an unnamed specimen from La Lieude, France (Werneberg et al., Reference Werneburg, Sébastien Steyer, Sommer, Gand, Schneider and Vianey-Liaud2007). The La Lieude specimen also has an intercentral gap, similar to I. bradleyae (Werneburg et al., Reference Werneburg, Sébastien Steyer, Sommer, Gand, Schneider and Vianey-Liaud2007, fig. 2).
Rhachitomous vertebrae, composed of a single intercentrum anteriorly and paired or fused pleurocentra posteriorly, are considered to be the plesiomorphic condition for tetrapods (Romer, Reference Romer1947) and are found in Acanthostega (Jarvik, Reference Jarvik1952; Coates, Reference Coates1996), Crassigyrinus (Herbst and Hutchinson, Reference Herbst and Hutchinson2019), colosteids (Hook, Reference Hook1983; Godfrey, Reference Godfrey1989), early temnospondyls (Milner and Sequeira, Reference Milner and Sequeira1994; Holmes et al., Reference Holmes, Carroll and Reisz1998), and the cervical region of Acherontiscus (Carroll, Reference Carroll1969; Clack et al., Reference Clack, Ruta, Milner, Marshall, Smithson and Smithson2019) but are not known in I. bradleyae. Due to the incomplete nature of TCD.T87, it is unclear whether I. bradleyae had rhachitomous vertebrae in the cervical region. The absence of rhachitomous vertebrae throughout the preserved vertebral column refutes the placement of I. bradleyae within the colosteids. However, the vertebrate column of early tetrapods is considered to have a weak phylogenetic signal (Danto et al., Reference Danto, Witzmann and Fröbisch2016) and show high plasticity, supported by the presence of reverse rhachitomous arrangements in Ichthyostega (Säve-Söderbergh, Reference Säve-Söderbergh1932), Acanthostega, and Whatcheeria (Lombard and Bolt, Reference Lombard and Bolt1995) and within temnospondyls (Shishkin, Reference Shishkin1989; Lombard and Bolt, Reference Lombard and Bolt1995; Pierce et al., Reference Pierce, Ahlberg, Hutchinson, Molnar, Sanchez, Tafforeau and Clack2013), and by the polyphyly of the “lepospondyls” (Pardo et al., Reference Pardo, Szostakiwskyj, Ahlberg and Anderson2017; Clack et al., Reference Clack, Ruta, Milner, Marshall, Smithson and Smithson2019), which were traditionally grouped together on the basis of shared vertebral characters (Baird, Reference Baird1965). This makes classifying I. bradleyae using primarily vertebral characters contentious.
The neural arches in TCD.T87 are poorly developed but lack the robust rectangular neural spines seen in embolomeres (Boyd, Reference Boyd1980; Clack, Reference Clack1987; Holmes, Reference Holmes1989). While there is no evidence to suggest that neural arches in the trunk of I. bradleyae are paired, there is a paired neural arch attached to centrum 47. Paired neural arches are commonly found in finned tetrapodomorphs (Andrews and Westoll, Reference Andrews and Westoll1970) but are rarely seen in tetrapods, with the exception of Crassigyrinus (Panchen, Reference Panchen1985; Herbst and Hutchinson, Reference Herbst and Hutchinson2019), Mesanerpeton (Smithson and Clack, Reference Smithson and Clack2018), and Balanerpeton (Milner and Sequeira, Reference Milner and Sequeira1994). TCD.T87 preserves an immature individual (see the following discussion), and paired neural arches in the caudal vertebrae may be an ontogenetic feature with neural arches fusing with maturity.
If the identification of the putative ilium in TCD.T87 is correct, then I. bradleyae had a unipartite dorsal iliac process, which is also seen in colosteids (Godfrey, Reference Godfrey1989), nectrideans (Milner, Reference Milner2019), and Aytonerpeton (Clack et al., Reference Clack2016). However, in these three taxa, the dorsal process curves more posteriorly and is square-ended at its apex. The ilium in I. bradleyae has a rounded apex and is more superficially similar to the pelvises of Eusthenopteron and Tiktaalik (Shubin et al., Reference Shubin, Daeschler and Jenkins2014).
The femora of I. bradleyae lack an adductor crest, which is unusual in early tetrapods. An adductor crest is also absent in Crassigyrinus (Panchen and Smithson, Reference Panchen and Smithson1990) and is poorly developed in Pederpes (Clack, Reference Clack2002; Clack and Finney, Reference Clack and Finney2005). The ratio of tibial to femoral length is one of the lowest among early tetrapods, with the tibia being 80% the length of the femur (Table 1), and is more similar to ratios seen in Rhizodus (Andrews and Westoll, Reference Andrews and Westoll1970; Jeffery et al., Reference Jeffery, Storrs, Holland, Clifford and Ahlberg2018) and early amniotes (Spindler et al., Reference Spindler, Werneburg and Schneider2019). In Greererpeton, tibial-to-femoral length shows ontogenetic variation from 66% in smaller, subadult specimens to 45% in larger, more-mature specimens (Godfrey, Reference Godfrey1989). The low tibial-to-femoral length ratio in TCD.T87 might suggest it preserves an immature specimen of I. bradleyae, as was suggested by Romer (Reference Romer1947). This is supported by the fibular and tibial condyles extending distally to the same level in the femur of I. bradleyae, which is unusual for early tetrapods, which typically have a longer fibular condyle (Holmes, Reference Holmes1984), but is seen in subadult Greererpeton specimens (Godfrey, Reference Godfrey1989). Due to the highly altered nature of TCD.T87, no unfinished bone can be identified in the femur. TCD.T87 preserving an immature specimen of I. bradleyae would explain why neural arches are not readily preserved throughout much of the vertebral column as they had yet to ossify.
The tibia of I. bradleyae has a thickened proximal end and a broad flat shaft and distal end similar to those of Greererpeton (Godfrey, Reference Godfrey1989) and Ossinodus (Warren and Turner, Reference Warren and Turner2004) but not seen in the tibias of embolomeres or temnospondyls, which have rounded shafts (Holmes, Reference Holmes1984; Klembara, Reference Klembara1985; Godfrey et al., Reference Godfrey, Holmes and Laurin1991; Pawley and Warren, Reference Pawley and Warren2006; Gee and Reisz, Reference Gee and Reisz2018; Adams et al., Reference Adams, Mann and Maddin2020). The tibia of I. bradleyae differs from that of Greererpeton and Ossinodus as it lacks an L-shaped distal facet that extends onto the posterior margin. The concave posterior margin of the tibia in I. bradleyae indicates the presence of an interepipodial area between the tibia and the fibula, a feature not seen in finned tetrapodomorphs but that is characteristic of all tetrapods crownward of Acanthostega (Warren and Turner, Reference Warren and Turner2004). The flattened tibial shaft indicates a placement for I. bradleyae among the stem tetrapods (Warren and Turner, Reference Warren and Turner2004).
While a deep tail and closely packed covering of scales makes TCD.T87 resemble a finned tetrapodomorph, the morphology of the tibia (mentioned in the preceding), the presence of phalanges and unguals, and the scale differentiation, with gastralia along the underside of the trunk, fortify the identification of I. bradleyae as a tetrapod. A long body with diplospondylous amphicoelous vertebrae is considered to indicate terrestrial and/or aquatic anguilliform locomotion in embolomeres and tupilakosaurids (Carroll, Reference Carroll1996; Warren, Reference Warren1999). Although the presacral vertebrae count is unknown in I. bradleyae due to the incompleteness of TCD.T87, it is feasible that I. bradleyae swam using anguilliform locomotion, propelled by its deep tail.
The systematic placement of I. bradleyae is difficult because there are only two known specimens, the most informative of which preserves an immature individual. The axial skeleton of I. bradleyae is more similar to that of the embolomeres and Acherontiscus, while the appendicular skeleton is more similar to the colosteids. Clack et al. (Reference Clack, Ruta, Milner, Marshall, Smithson and Smithson2019) identified a clade of stem tetrapods consisting of taxa with long bodies and reduced limbs (colosteids and nectrideans) or lacking limbs (aïstopods, adelospondyls, and Acherontiscus). Vertebral arrangement is diverse within this clade, which includes rhachitomous vertebrae (colosteids: Hook, Reference Hook1983; Godfrey, Reference Godfrey1989), diplospondylous amphicoelous vertebrae (Acherontiscus: Carroll, Reference Carroll1969; Clack et al., Reference Clack, Ruta, Milner, Marshall, Smithson and Smithson2019), monospondylous vertebrae with sutured neural arches (adelospondyls: Andrews and Carroll, Reference Andrews and Carroll1991), and holospondylous vertebrae (aïstopods and nectrideans [Anderson, Reference Anderson2002, Reference Anderson2003; Milner, Reference Milner2019]). When present, the iliac process is unipartite in this clade, seen in colosteids (Godfrey, Reference Godfrey1989), nectrideans (Milner, Reference Milner2019), and Aytonerpeton (Clack et al., Reference Clack2016). An affinity between I. bradleyae and this clade is here postulated, but until more material is discovered, this currently remains untestable.
Conclusion
Microcomputed tomography of the holotype (TCD.T87) of Ichthyerpeton bradleyae reveals that vertebrae are diplospondylous with cylindrical, amphicoelous centra. This does not support previous placements of I. bradleyae within the colosteids. A putative ilium is identified, along with a femur, a tibia, and an incomplete set of phalanges from both the right and left foot. The appendicular skeleton does not support a placement for I. bradleyae within the embolomeres or temnospondyls. The binomen I. bradleyae is here restricted to TCD.T87 and TCD.38351 as no link could be found between the holotype and other previously attributed specimens. As both TCD.T87 and TCD.38351 preserve only partial postcranial material of an immature individual, the relationship between I. bradleyae and other early tetrapods is difficult to establish.
Acknowledgments
We thank G. Sevastopulo for helpful and constructive discussions on the manuscript and for editing preliminary drafts. We also thank T. Davis, who helped with computed tomography. This manuscript was considerably improved by reviews from J. Anderson and T. Smithson. AÓG acknowledges funding from the Irish Centre for Research in Applied Geosciences (ICRAG) and the William George Fearnsides fund provided by The Geological Society, which allowed for the work to be carried out.
Data availability statement
Data available from the Dryad Digital Repository: https://doi.org/10.5061/dryad.5qfttdz4k.
