Relative age of the Devonian tetrapod Metaxygnathus, based on the associated fossil fish assemblage at Jemalong, New South Wales

Abstract

A new genus, Jemalongia, is erected for a porolepiform shoulder girdle and associated scales from the Cloghnan Shale at Jemalong Weir, near Forbes in New South Wales, Australia. There is no evidence for the Late Devonian porolepiform Holoptychius that was previously associated with the tetrapod Metaxygnathus at this site. Scales and a partial articulated specimen of Holoptychius sp. from the Famennian Worange Point Formation at Eden on the New South Wales south coast are close to both Holoptychius sp. from East Greenland and Holoptychius nobilissimus from Scotland. However, evidently a species other than Holoptychius sp. is represented by scales in the Hunter Siltstone at Grenfell, central New South Wales. The Jemalong fossils share character states with other scales from Grenfell and Bogan Gate that show features resembling coelacanth scales; a scale attributed to the Devonian coelacanth Gavinia is shown here for comparison. An isolated tooth plate demonstrates a second lungfish taxon at Jemalong, in addition to the denticulate Soederberghia sp. There is insufficient evidence for referring the skull of Soederberghia to the type species Soederberghia groenlandica. Incomplete placoderm remains suggest that a new antiarch taxon may occur at Jemalong. This new evidence suggests that the age of the Jemalong assemblage should be revised downwards to Givetian–Frasnian or older, rather than Famennian as previously interpreted.

Gavin C. Young [[email protected]; [email protected]], Department of Materials Physics, Research School of Physics, Australian National University, Canberra, Australia; Australian Museum Research Institute, 1 Williams St, Sydney 2010, New South Wales, Australia.

Keywords:

• Antiarch placoderms
• dipnoans
• porolepiform sarcopterygian scales

THE DISCOVERY of the oldest known amphibians Ichthyostega Save-Söderbergh, 1932 and Acanthostega Save-Söderbergh, 1932 in the uppermost Devonian of East Greenland was a major breakthrough regarding the origin of tetrapods (four-legged land animals). This represented fossil evidence for one of the most significant ecological transitions in vertebrate evolutionary history—from the aquatic to the terrestrial environment. The new evidence demonstrated that the ‘fish–tetrapod transition’ likely occurred sometime during the Devonian period. For subsequent decades search for more detailed fossil evidence of this transition focused on the Devonian palaeo-continents of the Northern Hemisphere.

Then the spectacular discovery of Late Devonian amphibian trackways at Genoa River in eastern Victoria (Warren & Wakefield 1972) demonstrated that tetrapods also existed on the largest land mass of the Devonian Period, the southern supercontinent of Gondwana. This was followed by the description of Metaxygnathus denticulus Campbell & Bell, 1977, a lower jaw from the Jemalong Range of central New South Wales (Fig. 1A, B), interpreted as a Devonian amphibian, and perhaps the oldest known. The tetrapod interpretation was disputed (e.g. Schultze 1986, who considered it a fish jaw), until similar lower jaw morphologies documented in European sequences supported the original interpretation. Ahlberg & Clack (1998) noted the similar slender and curved lower jaw of Metaxygnathus compared to Ventastega Ahlberg, Luksevics & Lebedev, 1994, and Acanthostega from the youngest Devonian stage (Famennian) of Laurussia. Metaxygnathus denticulus is still the only Devonian tetrapod skeletal fossil documented so far from Gondwana.

Fig. 1. Holotype lower jaw of Metaxygnathus denticulus in A, external and B, internal views (ANU 28780A; scale bar 1 cm). For comparative morphology see Ahlberg & Clack (1998). C, generalized localities for some Devonian vertebrates on a map of Eastern Australian states. Numbered localities mentioned in the text are: 1, Genoa River, Victoria; 2, Eden/Boyds Tower; 3, Pambula; 4, Forbes/Jemalong; 5, Grenfell; 6, Canowindra; 7, Parkes/Bumberry; 8, Bogan Gate; 9, Hervey Range. (BT, Bancannia Trough (subsurface Devonian); DB, Darling Basin; LFB, Lachlan Fold Belt). D, index for 1:250,000 geological map sheets in Central NSW (shaded rectangle is area covered in E). E, Devonian fish localities in the Forbes-Parkes-Grenfell-Canowindra area discussed in the text, shown on a distribution map for Hervey Group outcrop (black) on the eastern two-thirds of the Forbes sheet, and western quarter of the Bathurst sheet. Compilation from Pogson & Watkins (1998, fig. 46) and Lyons et al. (2000, fig. 11.1).

Fig. 2. A, Antiarch placoderm Remigolepis restored in dorsal view (based on Remigolepis walkeri from Canowindra). B, C, pectoral appendages of Remigolepis sp. in dorsal view: B, right appendage from Eden (ANU V1308; plaster copy of AMF 100536); C, ANU V3323, a left appendage from the Bumberry Formation in the Bumberry Syncline (loc. 7, Fig. 1E). D, antiarch left pectoral appendage from Jemalong, dorsal view (ANU V3391). E, antiarch right opercular (submarginal) bone from Jemalong, external view (ANU 28888B). F, right submarginal plate (SM; external view) attached to a skull of Remigolepis sp. from the Bumberry Formation in the Bumberry Syncline (ANU V3278). Abbreviations: Cd_1–2, dorsal central plate series of the pectoral appendage; gr, groove; Ml_2–3, lateral marginal plate series of the pectoral appendage; n.p, posterior notch; oa, overlap area; oss, ossification centre; SM, submarginal (opercular) plate.

At the time, available evidence supported Campbell & Bell’s (1977) interpretation that Metaxygnathus was the oldest tetrapod body fossil known. However, subsequent discoveries changed that perspective (reviewed by Young 2006a). Nevertheless, evidence for the age of the Jemalong fossil vertebrate assemblage remains relevant to more recent discussions concerning the time and place of origin for the first tetrapods, for example since discovery of older (Middle Devonian) Polish trackways attributed to tetrapods (Niedźwiedzki et al. 2010). These indicated a much earlier tetrapod origin, consistent with supposed tetrapod trackways (Early Devonian or older) from the Grampians of Western Victoria (Warren et al. 1986). However, both trackway interpretations have been disputed (e.g., Clack 2012, Lucas 2015), although other evidence (e.g. basal stem-tetrapod fishes in the Early Devonian of China and Australia) is also consistent with an earlier tetrapod origin (reviewed by Young & Lu 2020).

A large collection of fossil fish from the Jemalong locality is held at the ANU in Canberra, but almost all the systematic research on this material remains unpublished. The five decades since the initial systematic research by Bell (1972), has produced only two published papers (Campbell & Bell 1977, 1982). Meanwhile, there have been various comments and discussions in the scientific literature (summarized below) regarding the content of the Jemalong assemblage, in particular concerning two fossil fish groups relevant to a biostratigraphic assessment of its age.

1. ‘Placoderms’ (armoured fishes) belonging to the order Antiarchi (Figs 2, 3), a highly specialized placoderm sub-group in which the pectoral fins are enclosed in small bony plates to form crustacean-like pectoral appendages (Young 2008a, 2010).

2. Sarcopterygians (lobe-finned fishes) belonging to the Dipnoi (lungfishes), and Porolepiformes (Figs 4, 5).

Fig. 3. Comparison of bone patterns in the pectoral appendage of various antiarch placoderms, with the dorsal central series of bones shaded (left appendage, dorsal view). A, Chucinolepis (Early Devonian, China), after Zhang et al. (2001, fig. 3B); B, Remigolepis (Late Devonian, Greenland), after Stensiö (1931); C, Asterolepis ornata (Middle Devonian, Europe), after Gross (1931, pls V, VI); D, Dianolepis (Middle Devonian, China), from an unpublished restoration provided by Zhang Guorui; E, restoration of ANU V3391 from Jemalong (see Fig. 2D); F, Bothriolepis cellulosa (Upper Devonian, Europe), after Stensiö (1948). Abbreviations: Cd_1–4, dorsal central plate series of the pectoral appendage; gr, groove; Ml_2–3, lateral marginal plate series of the pectoral appendage; Mm₂, mesial marginal plate of the pectoral appendage. Bone terminology follows Stensiö (1948, table 1).

Fig. 4. A, Holoptychius, general body form in left lateral view, based on the 1956 restoration by S. Sampson (Swedish Museum of Natural History) of Holoptychius jarviki (Givetian–Frasnian of Quebec; actual length about 90 cm). B–D, scales of European Holoptychius; B, C, isolated scales attributed to H. giganteus (Agassiz 1844, pl. 24, figs 4, 8); D, part of the ventral squamation of the holotype of Holoptychius nobilissimus (from Agassiz 1844, pl. 23). The holotype (preserved length ∼71 cm) comes from Clashbennie, Perthshire, Scotland (NHMUK P6258; Woodward 1891). E–G, scales of Holoptychius sp. from Eden, NSW; E, ANU V1019; F, field specimen from Boyds Tower; G, articulated scales from Boyds Tower (AM F136010). H–L, scales of Jemalongia ritchiei gen. et sp. nov. from Jemalong (Bundaburra quarry); H, ANU V3397b; J, ANU V3398; K, ANU V3405b; L, ANU V3405a. M, lungfish tooth plate (ANU V3401). Abbreviation: str, striations on dentine ridges. (H–M, latex casts whitened with ammonium chloride; note that cavities at the apex of some teeth in Fig. 4M are bubbles in the latex. All scale images are oriented with anterior to the top.)

Fig. 5. Scales possibly referable to Jemalongia ritchiei gen. et sp. nov. from Jemalong (Bundaburra quarry); A, ANU V3395b; B, ANU V3395a. C, actinistian (coelacanth) scale referred to Gavinia sp. (ANU V1396). D, E, porolepiform scales from Bogan Gate (ANU V3252a-b). F, holotype cleithrum (internal view) of J. ritchiei gen. et sp. nov. (ANU V1899, from quarry at Jemalong Weir). Abbreviations: an, anterior notch; apr, anterior process; r.ad, anterodorsal ridge; r.av, anteroventral ridge; r.d, dorsal ridge; sc.a, anterior scapulocoracoid attachment; sc.d, dorsal scapulocoracoid attachment; sc.p, posterior scapulocoracoid attachment; vpr, ventral process. (A–E, latex casts whitened with ammonium chloride; all scale images oriented with anterior to the top.)

This paper illustrates some previously unpublished material of these fossil fish groups from the Jemalong assemblage, and discusses the implications of the new evidence regarding the relative age of the Gondwanan Devonian tetrapod Metaxygnathus.

Institutional abbreviations

AM, Australian Museum, Sydney, Australia. ANU, Research School of Earth Sciences, Australian National University, Canberra, Australia. ANU V, Research School of Physics, Australian National University, Canberra, Australia. NHMUK, The Natural History Museum, London, UK.

Materials and methods

The Cloghnan Shale producing the Jemalong fossils comprises reddish-purple siltstone and less abundant mudstone (Sherwin & Raymond 2000). It includes calcareous bands (caliche, algal structures), and abundant white bone preserved in some layers. Due to the high calcareous content, acid preparation techniques were not used for the initial undergraduate research project by M.W. Bell (1972), which only studied specimens visible on the surface of siltstone blocks. Some were further exposed by limited mechanical preparation. Much later in 2007, laboratory experimentation by ANU student G. Bell identified some non-calcareous layers on which the bone could be removed with weak hydrochloric acid. The impressions were investigated by making latex casts, which were whitened with ammonium chloride sublimate for study and photography. Unnumbered blocks in the ANU Jemalong bulk collection (presumably collected by M.W. Bell and K.S.W. Campbell in 1972) were broken up to search for new specimens. Acid-etching revealed a number of porolepiform scales and a lungfish tooth plate that were identified and latexed by ANU student D. Evans, and are illustrated in this paper.

Localities

Localities mentioned or discussed in this paper are summarized in Fig. 1C–E. There has been inconsistent usage of the term ‘Jemalong’ in previous literature. The Jemalong Range preserves the entire Upper Devonian (Hervey Group; Connolly 1965), in a westerly dipping sequence with the lowermost Cloghnan Shale on its eastern flank, and the uppermost Weddin Sandstone forming the main ridge of the Jemalong Range to the west. The main locality that produced Metaxygnathus is a quarry in the Cloghnan Shale on the eastern side of the Jemalong Range, where the Newell Highway passes through it about 30 km southwest of Forbes. This gap in the range (and the fossil site) is also named Bundaburra (Fig. 1E). About 20 km west of Forbes is another gap where the Lachlan River cuts through the Jemalong Range, evidently the fossil locality ‘Jemalong Gap’ referred to in earlier literature (e.g. Hills 1932, Sherwin 1973), but with stratigraphic level within the Hervey Group unspecified. The name ‘Jemalong Weir’ has also been applied to an adjacent quarry near a weir on the Lachlan River. This has produced fossil fish specimens from the Cloghnan Shale near the base of the stratigraphical sequence through Jemalong gap, and presumably represents a similar or the same stratigraphical level as the Bundaburra quarry vertebrate fossils of Campbell & Bell (1977, 1982). The Cloghnan Shale is a topographically recessive unit, and generally very poorly exposed except in road cuttings or quarries.

Previous age assessments for the Jemalong fossil assemblage

Campbell & Bell (1977) discussed the age of Metaxygnathus from the standpoint of correlations of the Cloghnan Shale with the stratigraphy in adjacent Upper Devonian outcrops of central NSW, named the Hervey Group by Conolly (1965). Sherwin (1973) had queried lithological alignments used by Conolly (1965) to correlate widely separated sections through the Hervey Group. Campbell & Bell agreed, and concluded that the Cloghnan Shale was ‘best regarded as Famennian, probably near the middle of the stage’, based on palaeontological evidence from the associated Devonian fish assemblage (Campbell & Bell 1977, p. 375).

The largely non-marine ‘Lambie facies’ (Hervey Group) in central New South Wales had generally been assigned a Late Devonian age, based on the occurrence of the antiarch placoderm fish Bothriolepis Eichwald, 1840, which is widely distributed in European Devonian sequences. E.S. Hills (1929, 1931) was the first to identify Bothriolepis in Australia. He relied upon some six decades of analysis of the stratigraphic range of Bothriolepis in European sequences to conclude a Late Devonian rather than Carboniferous age for the Cerberean Volcanics in the Cathedral Range of Victoria.

Hills (1932) then published on similar remains from two localities in the Hervey Group of central New South Wales (Fig. 1), notably an association in the Hervey Range of the antiarch placoderms Bothriolepis and Remigolepis Stensiö, 1931, together with scales identified as belonging to the sarcopterygian Holoptychius Agassiz, 1839. This association is also typical of the Upper Devonian of Europe. In particular, the genus Remigolepis had been erected the previous year by Stensiö (1931) for antiarch remains from the uppermost Devonian of East Greenland associated with the first described Devonian tetrapods Ichthyostega and Acanthostega (Save-Söderbergh 1932).

From ‘Jemalong Gap’ west of Forbes, Hills (1932) also illustrated a coarsely ornamented fragment that he identified as another Bothriolepis (the specimen evidently came from the Weddin Sandstone; Sherwin 1973, p. 79). Previously, Dun (1900) had identified another placoderm plate from ‘Jemalong’ to represent a new species of the European genus Asterolepis Eichwald, 1840. Asterolepis mainly occurs in the Middle Devonian of Europe, being replaced by Bothriolepis in Upper Devonian strata (e.g. Miles 1968, Young 1974, Mark-Kurik et al. 1999, Mark-Kurik & Põldvere 2012). The genus Remigolepis has a very similar morphology to Asterolepis, except that it lacks a distal ‘elbow’ joint in the pectoral appendage (Fig. 2A–C). Thus, detailed study of this aspect would be required to distinguish these taxa, to determine whether a Middle or Late Devonian age would be indicated (based on European stratigraphic ranges). However, after additional sites were documented in south-eastern Australia, Queensland (Hills 1936), and central Australia (Hills 1959), Hills (1958, p. 90) suggested that Asterolepis, Bothriolepis and Remigolepis ‘have no relative temporal significance in Australia as they have in Europe’.

Nevertheless, the Jemalong fish/tetrapod assemblage has long been grouped with the well-known fossil fish assemblage from Canowindra (about 60 km east of Jemalong) on the basis of the evidently similar Bothriolepis-Remigolepis placoderm association. Both were included as ‘Macrovertebrate Fauna 13’ by Young (1993), aligned with the lower-middle Famennian (also Young 1996). Then revised geological mapping (Young 1999, 2006b, Young et al. 2000) indicated an older (late Frasnian) age for Canowindra, with approximate alignment to the rhenana/triangularis Conodont Zone (MAV 13; Young & Turner 2000). Broad stratigraphic comparisons would suggest a similar late Frasnian age for Jemalong, where the Cloghnan Shale is also near the base of the Upper Devonian sequence (Young 2006b). Additional evidence is the discovery of a mandageriid sarcopterygian scale from Jemalong (Young et al. 2010), the type locality for this group being the Canowindra fish assemblage (Young 2008b).

Alternatively, Ahlberg & Clack (1998) supported the initial age assessment of Campbell & Bell (1977), Metaxygnathus being listed as ‘middle Famennian’ in Clack (2006). It was noted that the supposed Bothriolepis-Remigolepis-Groenlandaspis-phyllolepid placoderm assemblage at Jemalong compares closely with associated fish taxa in the famous Ichthyostega/Acanthostega tetrapod assemblage of East Greenland (late Famennian). In addition, the Cloghnan Shale has produced a single lungfish skull, described as Soederberghia sp. by Campbell & Bell (1982). The type species of this genus (Soederberghia groenlandica Lehman, 1959) also comes from the Famennian of East Greenland. Ahlberg et al. (2001) described a new smaller species from Canowindra (Soederberghia simpsoni Ahlberg, Johanson & Daeschler, 2001), noting that ‘surprisingly the Jemalong specimen is quite clearly not S. simpsoni’. They stated that ‘the Soederberghia from Jemalong is S. groenlandica’ (Ahlberg et al. 2001, pp. 4, 9), further supporting a younger (Famennian) age.

Both Ahlberg et al. (2001, p. 7) and Clack (2006, p. 181) also noted that scales of the porolepiform sarcopterygian Holoptychius occur at Jemalong, but not at Canowindra. Famennian Holoptychius scales are recorded at many localities worldwide, including the Famennian of East Greenland, and elsewhere in SE Australia. However, when the late Professor K.S.W. Campbell demonstrated to this author a supposed Holoptychius scale on a display slab from Jemalong, it was in fact the submarginal (opercular) bone of an antiarch placoderm, but with an unusually rounded shape (described below). The Jemalong material was then searched for other examples of holoptychiid scales, the results presented below.

Description of new material

Antiarch placoderms

The detached antiarch submarginal (SM) plate from Jemalong just mentioned (misidentified as a scale of Holoptychius) is illustrated here for the first time (Fig. 2E). ANU 28888B is preserved as light-coloured bone on the surface of a Jemalong slab. It is about 40 mm long, and oval in shape with height representing at least 66% of its length. This proportion can be compared with the more typical elongate shape of the SM in Famennian species, for example Remigolepis sp. from Greenland (Stensiö 1948), or Remigolepis sp. from the Bumberry Formation in the Bumberry Syncline, New South Wales (locality 7: Fig. 1E). In that specimen, the height is only 45% of length or less (SM, Fig. 2F). However, the SM of the older (Frasnian) Remigolepis walkeri Johanson, 1997a from Canowindra has height about 55% its length (Johanson 1997a), approaching the shape of the Jemalong example. This species has similar ornament of radiating ridges from a dorsal ossification centre, but the ossification centre is closer to the anterior end than in ANU 28888B, and the ridges are much finer, with many reticulations. Beneath the dorsal margin, ANU 28888B has two grooves slightly wider than the ornamental grooves (gr: Fig. 2E), which run forward and backward from the region of the ossification centre. Above this is an overlap area (oa). Johanson (1997a) suggested a dorsal spiracular opening at this position in R. walkeri, which has similar grooves. In Asterolepis ornata Eichwald, 1840 the spiracular opening is indicated by a dorsal notch about midway along the dorsal margin (Gross 1931). In Remigolepis sp. from Greenland there was only a pit in the upper edge of the SM according to Stensiö (1948, p. 60), and the angled dorsal margin of the SM fitted closely into the shallow embayment forming the lateral skull margin (Young & Zhang 1996). ANU 28888B also indicates a posterior notch (n.p), but this could be a preservational artefact.

Bell’s (1972) unpublished thesis described the antiarch remains from Jemalong as two new species in the genera Bothriolepis and Remigolepis, both based on articulated skull and trunk armours plus various isolated plates. At that time, the well-known Canowindra fossil fish fauna, comprising numerous complete articulated Bothriolepis and Remigolepis armours, had not been described. Four articulated armours on the same Jemalong block (ANU 28880) were assigned to the new Remigolepis species, ANU 28880G being selected as the holotype. These were described as of medium size (length of head plus trunk-armour about 19–20 cm), but much larger isolated trunk-armour plates were noted (ADL 16 cm long, AMD 20 cm long) that would have represented the largest species of Remigolepis, and one of the largest antiarchs, then known.

The first formal species descriptions of Bothriolepis and Remigolepis from central New South Wales were Bothriolepis grenfellensis Johanson, 1997a and Remigolepis redcliffensis Johanson, 1997a from the Hunter siltstone near Grenfell (‘Remigolepis hunterensis’ Ritchie, 2006, and ‘Remigolepis grenfellensis’ Johanson, 2002 are nomina nuda for the latter species). The thousands of articulated antiarchs in the Canowindra fossil fish assemblage (Ritchie 2006) were described by Johanson (1997b, 1998) as R. walkeri, and Bothriolepis yeungi Johanson, 1998. The only reference to the antiarchs of Jemalong was the statement by Johanson (1997a, 1998) that both B. grenfellensis and B. yeungi shared with the Jemalong Bothriolepis a pre-orbital recess of trifid shape. However, Bell (1972, p. 7) considered his Jemalong Bothriolepis species to have a pre-orbital recess like that described by Stensiö (1948) for the Canadian Bothriolepis species (the ‘simple’ recess shape of Young 1988, fig. 65). Preparation of specimens would be required to clarify this for the Jemalong Bothriolepis.

Many specimens of Remigolepis from East Greenland are preserved with the pectoral appendages attached (e.g. Stensiö 1931, pls 28, 29; Johanson 1997a). This is also the case for most of the articulated specimens of R. walkeri from Canowindra (Fig. 2A), and in articulated specimens of Famennian Remigolepis from Eden and the Bumberry Syncline (Fig. 2B, C). In lacking the distal (‘elbow’) joint (Fig. 2A), the short and robust shape of the pectoral appendage in Remigolepis perhaps made it more likely to remain articulated during fossil preservation. It is therefore surprising that for the articulated ‘Remigolepis’ specimens from Jemalong, Bell (1972, p. 7) stated that the pectoral fin was unknown. Further preparation, including acid-etching of some Jemalong Remigolepis material by G. Bell (2007–2010), similarly failed to produce any attached pectoral appendages. Bell (1972, pl. 15, fig. 3) identified in his material only one detached specimen as a very large dorsal articular plate of the pectoral appendage.

The later investigation of acid-etched specimens revealed only one detached example of an antiarch pectoral appendage (ANU V3391; Fig. 2D). This specimen suggests a bone suture pattern not previously known for antiarchs. Its proximal end articulating with the trunk-armour is missing, but the distal articulation is well preserved. The suture pattern, and prominent lateral and mesial spine rows, indicates that this is the dorsal surface of a proximal pectoral fin segment from the left side. A small second dorsal central plate (Cd2) shows its articular surface for connecting with the distal segment of the jointed pectoral appendage. Immediately adjacent on the lateral side is another small element, also showing an articular facet, and in addition carrying four prominent lateral spines. By comparison with Bothriolepis canadensis (e.g., Stensiö 1948), this is evidently part of the third lateral marginal plate (Ml3) of the distal segment, displaced from its proximal end.

The presence of a distal joint demonstrates that this pectoral appendage could not belong to Remigolepis, which lacks the distal joint (Fig. 2A). However, nor could it belong to Bothriolepis, also reported in the Jemalong assemblage, because there is a clear sutural contact between the two dorsal central plates (Cd1, Cd2; Fig. 3E). In Bothriolepis (Fig. 3F) these two bones are widely separated by a longitudinal contact between the lateral and mesial marginal plates (Ml2, Mm2). A transverse sutural contact between the two dorsal central plates of the proximal pectoral fin segment is evidently the primitive condition (character 6 in the phylogenetic analysis of Zhu 1996). It occurs in various asterolepid antiarchs (e.g., Asterolepis; Fig. 3C), and matches the suture pattern of the dorsal central series for the unjointed appendage in both Remigolepis (Figs 2B, C, 3B), and the yunnanolepid antiarchs from China (Fig. 3A; although different terminologies have been applied to the component bones of their fins; Zhu 1996, Zhang et al. 2001). The two dorsal central plates in Asterolepis and related forms are much closer in size (Cd2 slightly smaller than Cd1; Fig. 3C), whereas a provisional reconstruction of the Jemalong proximal segment (Fig. 3E) indicates that the second dorsal central plate was much smaller than the first. This is the situation in Bothriolepis. This incomplete pectoral appendage may indicate the presence of a new antiarch genus in the Jemalong assemblage, but more material would be required for a formal description to establish this.

Dipnoi (lungfishes)

The only new information provided here is a single impression of a lungfish tooth plate from Jemalong (Fig. 4M). This demonstrates at least one more lungfish taxon associated with Soederberghia sp. described by Campbell & Bell (1982), because Soederberghia had denticles in the mouth, not tooth plates. Other tooth-plated lungfish taxa associated with Soederberghia have been recorded from two Famennian Northern Hemisphere localities: the Catskill Formation of Pennsylvania (Daeschler & Mullison 2004), and the Famennian of East Greenland (Clack et al. 2019). ANU V3401 comprises poorly preserved impressions indicating at least 11 radiating rows of teeth. There is no indication of surrounding bone margins, so it is unclear if the tooth plate came from the palate, or the lower jaw. The anterior and posterior rows contained 5–6 teeth, and rows 2–7 contained 7–8 teeth. This specimen is too incomplete for further determination.

It was noted above that Ahlberg et al. (2001, p. 9) concluded that ‘the Soederberghia from Jemalong is [Soederberghia] groenlandica’. Ahlberg et al. (2001) considered their new species Soederberghia simpsoni to be primitive in some skull features relative to the younger S. groenlandica, but derived relative to others. They noted that the Jemalong specimen is nearly twice the length of the Canowindra lungfish skull. However, only a single incomplete skull is known from Jemalong, and the evidence supporting the conclusion that it is conspecific with S. groenlandica from East Greenland seems very uncertain. Contradictory arguments were put forward by Campbell & Bell (1982, pp. 143, 145), who noted several obvious differences from the Greenland type material, including ‘fewer and larger lateral line bones in the L-O series’, and a ‘rounded anterodorsal projection’ on the operculum compared to the ‘rather angular projection’ in S. groenlandica. Without more material, the Jemalong skull is best retained as Soederberghia sp., following Campbell & Bell (1982).

Porolepiform sarcopterygians

One of the most widespread fossil fish of the Late Devonian is the porolepiform Holoptychius Agassiz, 1839 (Fig. 4A). The type locality is Clashbennie, Perthshire, representing the uppermost Dura Den beds of the Old Red Sandstone in Scotland (Miles 1968). The articulated type specimen of Holoptychius nobilissimus Agassiz, 1839 was first illustrated in Murchison’s Silurian System (1839). It was the subsequent identification by Louis Agassiz of an isolated scale of Holoptychius from the ‘Silurian’ of Belgium that confirmed a correlation between the marine strata of Devon, and the Old Red Sandstone of Scotland, on which basis Sedgwick & Murchison (1839) erected the Devonian System (see Rudwick 1985, pp. 351–352).

Relatively complete preservation of articulated fish is very rare, and most sarcopterygian occurrences in the fossil record are isolated scales. A single Holoptychius (Fig. 4A) would yield over 500 large, and numerous small scales, which on decomposition and disintegration would be transported and deposited by currents in lag deposits. Distinctive morphology and ornament of isolated scales (Fig. 4B, C, E, F) may permit identification to genus or species level. In European Devonian sequences, different porolepiform genera (with distinctive scale morphologies: e.g., Lu & Zhu 2008) typify the threefold subdivision of Devonian strata: Porolepis (Lower Devonian), Glyptolepis (Middle Devonian), and Holoptychius (Upper Devonian).

For the Australian Devonian, the earliest published reference to a porolepiform is the listing by Clarke (1860, p. 288) of ‘?Glyptolepis (scale)’ in fossils from the ‘Devonian or Passage Beds’ of the southern districts of NSW. A scale called ‘cf. Glyptolepis’ was figured by Long (1991) from the Mount Howitt Devonian fish site in eastern Victoria, and similar scales belong to the fossil coelacanth Gavinia Long, 1999 from the same locality. A later letter by W.B. Clarke (January, 1870) to Professor Morrison (University of Sydney) mentions a ‘scale of Holophysius’ (sic) assigned to the Late Devonian-Carboniferous of the Williams River, Hunter Valley (Moyal 2003, p. 794). Subsequent identification of Australian porolepiform scale occurrences (e.g. ‘Holoptychius’ in the Hervey Range; Hills 1932) need to be assessed with respect to modern knowledge of the group.

Cloutier & Schultze (1996) listed a total of 22 named Holoptychius species. Eight are synonyms, and all the remainder are based on isolated scales apart from the following: H. nobilissimus (Fig. 4D), Holoptychius flemingi Agassiz, 1844, Holoptychius jarviki Cloutier & Schultze, 1996 (Fig. 4A), and a Holoptychius sp. from Greenland (Jarvik 1972). The articulated examples indicate the limits of intra-specific variation in scale morphology (Cloutier & Schultze 1996). H. nobilissimus, H. jarviki, and the Greenland Holoptychius sp. have similar ornamentation of typical flank scales (long bony ridges converging slightly posteriorly), but there are more (finer) ridges per scale in H. jarviki compared to the Greenland Holoptychius sp. In H. jarviki, typical flank scales with ridges are replaced ventrally by scales covered in tubercles (Cloutier & Schultze 1996). By contrast, the ventral scales in H. nobilissimus are also ridged (Fig. 4D), but the gular plates have tuberculate ornament (Agassiz 1844, pl. 23). This is also the case in the Greenland material (tubercles also fused into meandering bridges: Jarvik 1972, pl. 30, fig. 1). Different sized examples of H. flemingi indicate consistent scale ornamentation throughout the body, and in specimens of different sizes (Cloutier & Schultze 1996, p. 259).

Isolated scales referred to the typically Middle Devonian Glyptolepis Agassiz, 1844 were traditionally distinguished from the Late Devonian Holoptychius by having much finer ornament ridges. Ørvig (1957) did a detailed study comparing the histology and surface ornamentation of Glyptolepis and Holoptychius scales. He characterized scale ornament in Holoptychius as comprising coarse ridges of bone (compared to dentine in Glyptolepis), the ridges being generally smooth and rounded (grooved in Glyptolepis), but sometimes showing tubercle-like swellings or broken into tubercles. Holoptychius scales may also have a narrow ‘semilunar area’ immediately in front of the exposed field, comprising radiating rows of denticles (made of dentine). A corresponding area is always present in Glyptolepis scales. Here, the term ‘tubercle’ is applied to the rounded protuberances that may occur on the exposed (ornamented) part of a scale, formed by subdivision of the bony ridges. The term ‘denticle’ is applied to protuberances made of dentine, which are generally smaller, with a pointed shape. In Holoptychius scales (Ørvig 1957), the denticle rows may align with the bony ornamental ridges, and the denticles are pointed backwards and inwards, and have stellate bases and fine radiating ornament. They differ from the corresponding denticles in Glyptolepis and other forms by nowhere exhibiting any concavity on their top face. The equivalent denticles in Glyptolepis are inclined posteriorly, almost always with a distinct dorsal depression, this type variously described as ‘horseshoe-shaped’ (Ørvig 1957) ‘leaf-shaped’ (Cloutier 1996), or ‘tear-shaped’ (Johanson & Ritchie 2000). Ørvig noted that in Glyptolepis scales from the Middle Devonian of Scotland (Glyptolepis leptopterus Agassiz, 1844, Glyptolepis paucidens [Agassiz, 1844]) the dorsal concavity is seen on all denticles, but in slightly younger forms like Glyptolepis baltica Gross, 1930 (Orodesch and Podsnetgor bed, upper Middle Devonian) and Quebecius Schultze, 1973 (Frasnian, Miguasha, Canada) they are absent from the first generation denticles.

The supposed Glyptolepis scale from the Australian Devonian listed by Clarke (1860) may have been amongst fossils lost in transit when sent back to Prof. Adam Sedgwick in Cambridge (see Wright 1994). The scale fragment from the Upper Devonian Hervey Range attributed to Holoptychius sp. by Hills (1932, pl. 56, fig. 4) is too poorly illustrated for identification (the specimen cannot be located). ‘Possible holoptychiid scales’ were reported by Young (in Fergusson et al. 1979) from the Upper Devonian Worange Point Formation on the NSW south coast near Eden. Some of these scales were illustrated by Young (2007) and Young et al. (2010), and referred to Holoptychius without description. Given the rarity of articulated Holoptychius examples, it is significant that portion of an articulated fish comes from the same Boyds Tower locality, and is illustrated here for the first time (Fig. 4G). Detailed description of these Eden scales (Fig. 4E, F) provides the basis for determining new porolepiform fish scales from the Jemalong assemblage (Fig. 4H–L).

A typical isolated Holoptychius scale displays two quite different surface textures (Fig. 4B, C, E, F): an anterior portion that is mainly smooth because it was overlapped by the scales in front and to the sides, and a posterior ornamented portion, the only part that was exposed in the articulated fish (Fig. 4A). Various terminologies have been used by previous authors, but in the descriptions here the anterior area is called the ‘overlapped field’, and the posterior area the ‘exposed field’. The posterior part of the overlapped field may also sometimes show a zone of fine radiating denticle rows, the ‘semilunar area’ of Ørvig (1957) mentioned above.

The Eden example observed in outcrop (Fig. 4E) was an isolated scale at least 29 mm long (posterior edge broken), and 27 mm wide. The overlapped field shows a fine spongy bone surface without denticles. The exposed field is ornamented with about 13 bony ridges diverging anteriorly and converging posteriorly, with some anastomosing connections. The connections are more prominent in the midline of the scale with 2–3 sinuous longitudinal ridges mainly connected, with intervening rows of pores. The central pores, and indications of a median groove at the anterior edge of the exposed field, suggest this could have been a lateral line scale, by comparison with isolated scales from the Famennian of East Greenland (Stensiö 1931, pl. 35, figs 11–12). Another isolated incomplete scale (Fig. 4F) has a pointed anterior edge, with a slight midline ridge, perhaps coming from near the dorsal midline of the fish. The posterior edge is very incomplete; the scale was at least 40 mm across. The longitudinal bony ridges are more numerous than in the previous example (up to 22 across), but otherwise similarly developed, with sinuous shape and anastomosing connections. They differ in the stronger convergence towards the midline in the anterior parts of the ridges, which also show a few short tubercle rows. This scale closely resembles those on AMF136010 (Fig. 4G), a piece of red siltstone about 19 cm long, displaying at least 35 imbricated scales. Many are broken, with some displaced or rotated. In most of the scales the anterior field is overlapped by the scale in front, demonstrating an articulated fish, rather than a lag deposit. All the scales are of similar size and ornament, more complete examples being wider than long (width = ∼35 mm; length = ∼29 mm). Ornament comprises about 20 sinuous ridges with many anastomosing connections, occasionally broken up into short tubercle rows. The ridges in the middle part of each scale are more continuous running anteroposteriorly; the more lateral ridges are generally shorter, and converge posteriorly towards the middle of each scale.

These Eden scales compare closely with isolated Holoptychius scales from Upper Devonian European sequences, including some assigned by Agassiz to H. nobilissimus (Fig. 4B, C), and others from the Famennian (‘Remigolepis Series’) of East Greenland. The latter are preserved both as isolated scales (Stensiö 1931, text-fig. 92, pls 34, 35), and associated with articulated Holoptychius specimens (Jarvik 1972, pls 30, 32, fig. 1). Stensiö provisionally referred his isolated scales to two of the Agassiz species (Holoptychius nobilissimus and Holoptychius giganteus Agassiz, 1839), but he noted they could also represent variation within one species. The Greenland scales range in size downward from a maximum width of about 50 mm. Some scales have ornamental ridges more delicate and widely spaced (e.g. Jarvik 1972, pl. 32, fig. 1), but in others (Stensiö 1931, pl. 34, fig. 2, pl. 35, figs 5, 7, 11–12; Jarvik 1972, pl. 30, fig. 1) the ornament is difficult to distinguish from the Eden examples. One difference is that some of the Greenland scales, both smaller and larger than the Eden examples, show clear radiating denticles (up to 11 rows) in the posterior part of the overlapped field, not observed in any of the Eden scales.

Cloutier & Schultze (1996, p. 257) summarized the ornament on the exposed field of Holoptychius scales by recognizing four groups: (1) continuous ridges that diverge posteriorly (Holoptychius radiatus Newberry, 1889 and Holoptychius halli Newberry, 1889); (2) scales with a network of ridges (H. giganteus; Holoptychius americanus Leidy, 1843 may be a synonym); (3) tubercles replacing the ridges anteriorly (Holoptychius tuberculatus Newberry, 1889); (4) very similar ornament, with ridges converging posteriorly (H. nobilissimus, H. jarviki, and Holoptychius sp. from East Greenland).

The Holoptychius scales from Eden most closely resemble scales in the last group. The scales of H. jarviki (Givetian–Frasnian, Canada) differ from the others in this group, and the Eden scales, in having finer ridges, with a greater number on each scale. In addition, as noted above, Holoptychius sp. from East Greenland differs from the Eden scales in the common development of anterior radiating denticles on the overlapped field.

Scales referred to Holoptychius sp. were illustrated by Johanson & Ritchie (2000) from the Hunter Siltstone at Grenfell New South Wales, also of assumed Famennian age. These show clear differences from the Eden scales (radiating anterior denticle rows, thickened ornament ridges or groups of tubercles at the scale centre that are flanked by narrower ridges laterally, etc.), and evidently represent a different Holoptychius species. Other scales from the Hunter Siltstone fauna with a rather different morphology (Johanson & Ritchie 2000) were suggested to resemble an undescribed porolepiform from the Baltic Middle Devonian, and referred to as a ‘holoptychiid n.gen., n.sp.’. As noted below, these show several morphological similarities with scales described here from the Jemalong assemblage.

New isolated porolepiform scales from Jemalong (Fig. 4H–L) can be compared in detail with the Eden Holoptychius scales. Eight incomplete scales were revealed by latex casting of acid-etched Jemalong blocks done in 2008. All show an ornament of radiating anterior denticle rows and posterior ridges typical of porolepiform scales, but clearly different from the Holoptychius scales just described. Unlike Holoptychius scales, the overlapped field has two distinct zones: an anterior smooth zone, and a posterior denticulate zone (denticle field). The latter is relatively large in ANU 3397b (Fig. 4H), one of the more complete examples (at least 38/30 mm wide/long). The anterior smooth zone of the overlapped field comprises about 26% of the mid-axis length of the scale. The wider denticle field (about 44% of mid-axis length) has about 64 radiating denticle rows. In each row the anterior 7–8 denticles radiate linearly, presumably representing first generation denticles as described by Ørvig (1957). Behind this (second generation) the denticles are larger and more irregular. The exposed field of the scale comprises about 30% of the mid-axis length. It shows rounded ridges and elongate tubercles that are subparallel to slightly diverging posteriorly. Many ridges are about 3 mm long, but some are twice this length.

Another much less complete scale (ANU V3397a) is very similar to the previous scale; it is at least 27 mm across, but the anterior smooth part of the overlapped field is not preserved. The posterior part shows at least 50 radiating rows of denticles, all pointing posteriorly and many displaying the dorsal depressions. The exposed field shows elongate ridges, 1 mm thick and up to 7 mm long, interspersed with shorter ridges and some elongate tubercles. The ridges diverge slightly towards the posterior, where they break down into finer ridges. Similar ornament is seen in some of the holoptychiid n. sp. Grenfell scales (Johanson & Ritchie 2000), but in both scales from Jemalong the ornament is much less crowded, with the ridges more widely dispersed with intervening smooth areas, compared to the crowded ornament of the Grenfell scales. ANU V3404 is similar to ANU V3397a, but with ridges of the exposed field shorter and more crowded.

Another incomplete scale (ANU V3398: Fig. 4J) displays only the denticulate area, and adjacent ornament ridges of the exposed field in the central part of the scale. The ridges are sinuous and crowded, resembling somewhat one scale from the Grenfell assemblage (Johanson & Ritchie 2000). The ridges are 2 mm thick anteriorly, narrowing almost to a point posteriorly, and up to 12 mm long. The denticulate field is slightly different, with the anterior part comprising more continuous ridges with elongate dorsal grooves, subdividing posteriorly into a few large tear-shaped denticles.

Some of the anterior ridges in ANU V3397a suggest a chevron pattern of striations pointing posteriorly, as illustrated for Glyptolepis baltica by Ørvig (1957). This feature is clearly demonstrated in the better-preserved ANU V3405a-b (str; Fig. 4K, L). Similar structure has been described for the coelacanth Miguashaia Schultze, 1973 (Cloutier 1996; also Forey et al. 2000), but many other aspects of the Miguashaia scales are clearly different. The histological investigation by Ørvig (1957) showed that such striations indicate the ridges of the Jemalong scales were formed of dentine, not bone as in the scales of Holoptychius. However, there is no indication of such striations on the ornament ridges of the scales assigned by Johanson & Ritchie (2000) to ‘Holoptychiidae n. gen. and n. sp.’, and given the excellent detailed preservation in the Grenfell material, it can be assumed that they were absent.

The Jemalong material includes two incomplete cycloid scales 40–50 mm across showing scattered central tubercles on the exposed field (Fig. 5A, B). These could belong to the same taxon as the associated ridged scales, if they came from the ventral surface of a fish with tuberculate scales ventrally, and ridged scales on the flanks, as demonstrated for H. jarviki (Cloutier & Schultze 1996). One of these scales has sparse central tubercles almost completely surrounded by concentric growth rings (Fig. 5B). This resembles a scale in inner view for the primitive actinistian (coelacanth) Miguashaia (Cloutier 1996), where up to eight central ‘bumps’ have been recorded, so an alternative interpretation is that this scale displays the internal surface.

Young et al. (2010, pp. 68–69) had already noted that the distinctive Grenfell scales described as ‘Holoptychiidae n. gen. and n. sp.’ by Johanson & Ritchie (2000) showed similarities to Middle–Late Devonian actinistian scales like Miguashaia from Canada. The ridged Jemalong scales just described are also similar to the Grenfell scales, and of similar size (Johanson & Ritchie 2000). The proportions of the Grenfell scales are more variable (longer than broad, or broader than long), but this is not significant, as the fossils from the Hunter Siltstone are tectonically distorted. The denticle fields of some Grenfell scales (Johanson & Ritchie 2000) are similar to Jemalong, with 45–55 rows across, but they differ in that the rows are more crowded, and the tear-shaped depressions are less obvious. In other scales the denticles may be non-aligned (Johanson & Ritchie 2000). As noted above, the exposed field tubercles and ridges on the Grenfell scales are generally more crowded and variable than in the Jemalong scales. There is similarity to the exposed field ornament of Miguashaia scales illustrated by Cloutier (1996), but the latter are smaller scales, the denticle rows extending almost across the overlapped field, and containing 20+ denticles per row (compared to 10–15 in the Grenfell scales).

Similar scales from a third locality, near the base of the Hervey Group at Bogan Gate on the Narromine 1:250,000 geological map (locality 8: Fig. 1), were illustrated by Young et al. (2010). More detail is presented here. Both scales have similar counts of denticles and rows (Fig. 5D, E). The ornament of the exposed field comprises sub-parallel rows of elongate tubercles and short ridges about 15 rows across. The ridges may be up to 9 mm long with attenuated or pointed posterior ends (Fig. 5D), or up to 6 mm long and blunt at both ends (Fig. 5E). The upper scale has at least 20 rows across of tubercles/short ridges, which are more sinuous and irregular than the lower scale. This scale is at least 40 mm wide and long (Fig. 5E), the exposed field comprising the posterior 35–40% of mid-scale length, and the denticulate and smooth parts of the overlapped field each comprising about 30%. The denticulate area contains over 50 radiating denticle rows around its anterior edge, but this includes short distal intercalating rows of about five small denticles. More posteriorly, 25–30 proximal radiating rows of large denticles are developed. These scales differ in detail from the Jemalong scales, and lack the crowded elongate tubercles in the centre of the exposed field typical of the Grenfell scales.

In summary, this type of distinctive ‘holoptychiid/actinistian scale’ is now documented from three localities in the Hervey Group of central New South Wales. Grenfell; Bogan Gate; Jemalong. The only named Middle–Late Devonian Australian actinistian is Gavinia syntrips Long, 1999, from Mt Howitt in Victoria. One detached scale shown by Long (1999), from Pambula in southeastern New South Wales (locality 3: Fig. 1), is re-illustrated with more detail here (Fig. 5C). The same specimen shows the denticle field of a second scale exposed at a deeper level (right side of Fig. 5C). The main scale is incomplete on the right side, but if symmetrical would have been about 14 mm across. The narrow smooth part of the overlapped field is incomplete. The denticulate area has about 30 rows on one side from the midline axis, indicating a total of 50–60 rows, comparable to the Jemalong scales. Each row comprises 7–8 aligned denticles, with 2–3 less regular and larger denticles behind (clearly tear-shaped, with dorsal depressions pointed posteriorly). The exposed field has about 10 sinuous subparallel ridges across the left half (to the scale mid-axis), slightly diverging in the lateral part of the scale. In the central part they become more narrow and closer together towards the posterior edge. Between the main ridges are interpolated 1–2 finer ridges running forward from the posterior edge of the scale. Some of these finer posterior bridges are continuous to the front of the exposed field. In this detail therefore, the Gavinia scale is clearly different from the Jemalong, Grenfell and Bogan Gate scales.

For the latter, three different but related taxa are indicated, based on scale characteristics. Johanson & Ritchie (2000, p. 116) discussed similarities in the Grenfell scales to the coelacanth Miguashaia, but concluded they belonged instead to Porolepiformes, based on similar scales from Estonia (Middle Devonian) associated with undoubted holoptychiid skull and lower jaw specimens. Regarding Jemalong, it is clear that previous assumptions of Holoptychius scales in the assemblage were erroneous, the scales described and illustrated here (Fig. 4H, L) possessing features (striated dentine rather than bony ridges, tear-shaped denticles with dorsal depressions) that were used to differentiate the Middle Devonian Glyptolepis from the Late Devonian Holoptychius (Ørvig 1957). The detailed differences in morphology within this scale type also suggest differences in age between the Jemalong, Grenfell and Bogan Gate localities.

For Jemalong, an age difference with Canowindra is also indicated, because no evidence of holoptychiids or any porolepiform is known from the Canowindra assemblage. Discounting the occurrence of Holoptychius, the Jemalong scales show more resemblance to the Middle Devonian Glyptolepis, suggesting an older age. In addition, the lungfish Soederberghia shared at both localities is evidently a different species, and Jemalong possibly has a new antiarch taxon additional to the assumed co-occurrence of Bothriolepis and Remigolepis at both localities. All this new evidence supports an older Givetian–Frasnian age for Jemalong, rather than Famennian as previously suggested. Together with an associated cleithrum, and following the example of Johanson & Ritchie (2000), the Jemalong scales are provisionally assigned to a new taxon within the Porolepiformes.

The Jemalong cleithrum (Fig. 5F) was first noted by Long (1991, p. 390–391) as ‘shoulder girdle bones of Porolepiformes … from … the Famennian Jemalong Gap fauna’. ANU V1899 is a left cleithrum preserved in internal view. The posterior margin is complete only around the posteroventral angle, but the broken edges can be restored by comparison with other porolepiform cleithra (Jarvik 1980, Cloutier & Schultze 1996), indicating that not much is missing. The margins were presumably straight to slightly concave posteriorly for the dorsal lamina, and gently convex mesially for the ventral lamina. The dorsal lamina (about 48 mm) is slightly longer that the ventral (46 mm). A longer ventral lamina was said to be typical of porolepiforms (Andrews & Westoll 1970a, b, Cloutier & Schultze 1996). In Quebecius (Schultze & Arsenault 1987) the dorsal lamina is only slightly shorter than the ventral. The ventral lamina of the Jemalong cleithrum is much longer than in tristichopterids like Eusthenopteron (Andrews & Westoll 1970b) or Edenopteron (Young et al. 2013, 2019).

Highly unusual is a prominent anterior notch with largely complete bone margins, defined by anterior and ventral processes (an, apr, vpr: Fig. 5F). The upper projection (apr) is reminiscent of the anterior process of the cleithrum in tristichopterids. However, the much longer ventral projection (vpr) gives a quite different shape to that of tristichopterids. A less distinct anterior notch may also be present in another very incomplete cleithrum from the Hervey Group, named Grenfellia by Johanson & Ritchie (2000). The presence of this notch suggests a subdivision of the cleithrum into two elements, as previously recorded in some examples of Holoptychius sp. (Jarvik 1980). Jarvik (1950) illustrated subdivided cleithra in both Holoptychius sp. from East Greenland (Upper Devonian), and Glyptolepis from the Middle Devonian of Scotland. However, in these examples the sutures go straight across, so in neither could the main part of the cleithrum have had the long ventral process seen here. Jarvik (1972, p. 128) noted that the subdivided cleithrum was rare in porolepiforms, and of variable occurrence in Glyptolepis given there was no evidence for it in the Middle Devonian G. groenlandica. Actinistians also have a subdivided cleithrum, in Miguashaia (Cloutier 1996, Forey et al. 2000) deeply notched for the second dermal element, (‘extracleithrum’), but the notch is in a completely different ventral position compared to the Jemalong cleithrum.

The incomplete dorsal margin of the Jemalong cleithrum is horizontal (Fig. 5F). In Holoptychius it has been restored as sloping downwards and backwards (Jarvik 1980), and in some species it may have a shallow notch (Downs et al. 2013). The corresponding margin is more rounded in Glyptolepis (Jarvik 1972). It is incomplete in Quebecius (Schultze & Arsenault 1987), and obscured in H. jarviki (Cloutier & Schultze 1996).

Other differences from Holoptychius and Glyptolepis are indicated on the inner surface of ANU V1899. Faint radiating striations indicate the region of the ossification centre, in the standard position inside the posteroventral angle. Above this, a faint cup-shaped depression represents an attachment area for the scapulocoracoid (sc.p). Its posterodorsal elongation is the opposite to Holoptychius (posteroventral; Jarvik 1980). For the Middle Devonian porolepiform G. groenlandica the attachment was represented as a more horizontal tear-shaped scar (Jarvik 1972), again with a quite different orientation to the Jemalong specimen. A single attachment to the cleithrum was said to reflect a fundamental difference in scapulocoracoid shape in porolepiforms, compared to the tri-radiate scapulocoracoid of osteolepiforms (Jarvik 1985). However, for Glyptolepis groenlandica, Ahlberg (1989, p. 125) re-interpreted the scapulocoracoid as also tri-radiate, as he demonstrated for Glyptolepis sp. from Scotland.

Two additional scapulocoracoid attachments are indicated for V1899 (Fig. 5F), with a small elevation (sc.d), and roughened area (sc.a) preserved above and in front of the more prominent posterior attachment (sc.p). Three attachments were also described in Grenfellia Johanson & Ritchie, 2000, but the Jemalong cleithrum differs in that the posterior scapulocoracoid attachment is larger than the anterior (the opposite in Grenfellia; Johanson & Ritchie 2000). Johanson & Ritchie (2000) interpreted this as a primitive sarcopterygian feature, because it also occurs in basal tetrapodomorphs, in various Devonian lungfish, and probably in basal dipnomorphs like Youngolepis Chang & Yu, 1981. From the posterior attachment (sc.p: Fig. 5F) a ventral thickening runs forward as a ridge (r.av) to the extremity of the anterior ventral process. A faint ridge in a corresponding position is indicated for Glyptolepis (Jarvik 1972). A more prominent bifurcated ridge occurs in Grenfellia, but this differs in running from the larger anterior scapulocoracoid attachment. Another faint ridge in the Jemalong cleithrum (r.ad) extends from the ossification centre to the extremity of the anterior dorsal process (suggested also in Grenfellia). Neither ridge is recorded for Holoptychius. A third faint ridge (r.d) radiates up to the anterior margin of the dorsal lamina, evidently corresponding to the ridge in this position in both Glyptolepis and Holoptychius.

The distinctive shape of the Jemalong cleithrum suggests that it may represent a new taxon, to which the associated porolepiform-like scales can also be referred. It seems likely also that the distinctive scales from the Grenfell assemblage described as ‘Holoptychiidae n. gen. and n. sp.’ by Johanson & Ritchie (2000) could similarly belong to their genus Grenfellia.

Systematic palaeontology

OSTEICHTHYES Huxley, 1880

SARCOPTERYGII Romer, 1955

?POROLEPIFORMES Jarvik, 1942

Remarks

This new taxon is provisionally interpreted as a porolepiform. However, considering some of the scale features discussed above, referral to the Actinistia is not completely excluded. Of the small number of named actinistian genera from the Middle–Late Devonian (Diplocercides Stensiö, 1922, Gavinia, Miguashaia, Shoshonia Friedman, Coates & Anderson, 2007, ?Holopterygius), the shoulder girdle is only known for Miguashaia, where the notch for the ‘extracleithrum’ (this bone considered a coelacanth synapomorphy by Forey 1998) has a completely different ventral position.

Jemalongia gen. nov.

Jemalongia ritchiei sp. nov.

(Figs 4H–L, 5F)

1991, ‘shoulder girdle bones of Porolepiformes’, Long, p. 390.

Diagnosis

Cleithrum with a large anterior notch, and the posterior of three scapulocoracoid attachments being the largest, from which an anteroventral ridge passes to the extremity of the ventral process. Scales provisionally included typically have a posterior denticulate zone on the overlapped field more extensive then the anterior smooth zone, with 50–60 rows of tear-shaped denticles with dorsal depressions. Dentine ridges of the exposed field with striations in a chevron pattern pointing posteriorly.

Etymology

From the locality name in the Jemalong Range of central New South Wales, and in honour of the late Dr Alexander Ritchie (15 August 1935–16 November 2023), who first learnt about the Jemalong fossil site and conducted excavations there in the 1970s, when he was Curator of Fossils at the Australian Museum.

Holotype

ANU V1899, an incomplete left cleithrum preserved in internal view.

Referred material

ANU V3395, ANU V3397–ANU V3398, ANU V3404–ANU V3406, impressions of isolated scales preserved in red siltstone.

Remarks

The fine striations on the dentine ridges differentiate the referred scales from those at Grenfell and Bogan Gate discussed above. The deeper anterior notch and more acute anterior process distinguish the holotype from the Grenfellia holotype, which also differs in having the anterior scapulocoracoid attachment the largest (posterior the largest in Jemalongia).

Type locality, unit and age

The holotype comes from the quarry near Jemalong Weir, about 20 km west of Forbes, and the referred material from the Bundaburra quarry, about 30 km southwest of Forbes. Both localities are within the Cloghnan Shale, the lowest formation of the Hervey Group in the Jemalong Range, of assumed Givetian–Frasnian age (Middle-Late Devonian).

Summary and conclusion

The fossil fish assemblage in the Cloghnan Shale at Jemalong, New South Wales, was long regarded as indicating a Late Devonian (Famennian) age for the associated Devonian tetrapod Metaxygnathus denticulus (Campbell & Bell 1977). This was largely based on a supposed Bothriolepis-Remigolepis-Groenlandaspis-phyllolepid placoderm assemblage at Jemalong, closely comparable to the associated placoderms in the famous late Famennian Ichthyostega/Acanthostega tetrapod assemblage of East Greenland. However, the Jemalong fossil fish assemblage remained undescribed, except for one lungfish skull named Soederberghia sp. by Campbell & Bell (1982), which supported the Famennian age interpretation, as the type species (Soederberghia groenlandica) also comes from the East Greenland Famennian.

A similar Bothriolepis-Remigolepis-Groenlandaspis placoderm assemblage at the well-known Canowindra fossil fish site also suggested a Famennian age, until other evidence indicated it was older (Frasnian). However, a supposed key difference between the two fossil fish assemblages was the occurrence of scales of the porolepiform sarcopterygian Holoptychius at Jemalong, these being unknown at Canowindra.

Famennian Holoptychius scales are recorded at many Late Devonian localities worldwide, including East Greenland, and elsewhere in SE Australia. A detailed comparison between isolated Holoptychius scales from the Famennian at Eden (New South Wales south coast), and new porolepiform scales from Jemalong, now demonstrates clear differences in scale morphology. The Jemalong scales share characters with the porolepiform Glyptolepis (Middle Devonian in European sequences), but an associated shoulder-girdle bone (cleithrum) differs significantly from Glyptolepis, and all other porolepiforms. Thus, a new taxon, Jemalongia ritchiei has been erected for the Jemalong porolepiform material.

An isolated lungfish tooth plate demonstrates a second taxon at Jemalong additional to Soederberghia sp., which is a denticulate form lacking tooth plates. Campbell & Bell (1982) listed significant differences by which the Jemalong skull could be distinguished from the type species S. groenlandica. In addition, incomplete placoderm remains suggest that a new antiarch taxon may occur at Jemalong.

This new evidence suggests that the age of the Jemalong assemblage should be revised downwards to Givetian–Frasnian, rather than Famennian as previously interpreted. Jemalong could even be older than the Canowindra fossil fish assemblage, which has a different Soederberghia species, and lacks both porolepiform sarcopterygians, and phyllolepid placoderms. Systematic description of the Jemalong antiarchs, and detailed comparison with the two described species from Canowindra, would be required to test this hypothesis.

Note

While this article was in press, Retallack (2024, p. 1) proposed a palaeoenvironmental interpretation for the Canowindra fossil fish site as a ‘drying desert [pond]’, with the depositional setting of the Metaxygnathus type locality at Jemalong being streams in a ‘subhumid woodland’. This was based on analysis of palaeosols identified in sections measured at both sites. Retallack (2024, p. 4) further correlated the Hervey Group as Famennian age, with the occurrence of Metaxygnathus being stratigraphically higher than the Famennian Canowindra fossil fish assemblage. These conclusions are at odds with the data presented here, and will be addressed in detail in a future publication.

Acknowledgements

B. Young, D. Evans and G. Bell (ANU) are thanked for fossil preparation. Dan Evans (ANU) identified and latex cast the Jemalong lungfish toothplate and porolepiform scales, and provided a new latex cast of the Gavinia scale from Pambula. Ben Young (ANU) contributed photography and curated the fossil material. The ANU Research School of Physics (T. Senden) and Department of Materials Physics (J. Bradbury) are thanked for provision of facilities. Ken Campbell is thanked for numerous discussions about the Jemalong fossil material, and for conducting several visits to the Jemalong fossil sites with the author. Robert Jones (AM) is thanked for providing details of AM specimens from the Eden Boyds Tower locality. NSW National Parks facilitated fieldwork and fossil collecting under Scientific Licence S11982. Adolf Haupt assisted in translating publications from the German.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

Funding (2005–2007) was provided by ARC Discovery Grant DP 0558499. The Robert Day Endowment provided funding for G. Bell (ANU).