The Réalmont Basin is located in the Quercy-Albigeois region (Tarn, France) in the southern French Massif Central and is among the southernmost Stephano-Permian basins in this massif (Fig. 1B, C; Delsahut 1981, 1989; Fauré 2022). It is a relatively small basin measuring 7 km north-south and 2.5 km east-west, covering an area of 4.5 km² (Delsahut 1981, 1989; Fauré 2022). Its deposits appear unconformably beneath the Argiles à Graviers Formation of the Cenozoic Aquitaine Basin (Fauré 2022). The Réalmont Basin is oriented by a north-south fault, the Réalmont Fault, which is the southern extension of a network of several faults starting from the Sillon Houiller fault, linking the Decazeville and Carmaux basins and extending towards Réalmont: the Salvetat-Peyralès Dislocation (Fauré 2022). The basin deposits are fragmented by a network of synsedimentary faults running predominantly east-west (Fauré 2022). The Réalmont Basin is divided into five formations (called ‘séries’ and ‘ensembles’ by Delsahut [1981] in French terminology) filled with siliciclastic sediments: the Blima (Stephanian B), L’Escoudarié, Cantemerle, Cayla (Autunian) and Réalmont (Saxonian) formations (Fig. 1C, D; Delsahut 1981, 1989; Fauré 2022). These formations are also known locally as ‘série grise’, ‘série conglomératique rosée’, ‘série gréso-pélitique grise’, ‘série gréso-pélitique grise et rouge’ and ‘série rouge’ respectively (Delsahut 1981, 1989; Fauré 2022). They rarely appear in continuity, and the best section available is that of the Saint-Genest borehole (Fig. 1D; Delsahut 1981; Fauré 2022). Based on the available structural information, the Autunian formations may potentially lie unconformably on the Blima Formation, and the Réalmont Formation lie unconformably on all other formations in the Réalmont Basin (Delsahut 1981; Fauré 2022). In terms of depositional environments, the coarse silico-clastic series correspond to alluvial and fluvial fan environments (Delsahut 1981; Fauré 2022). The finer series correspond to floodplain and lacustrine deposits (Delsahut 1981; Fauré 2022). Palaeontological discoveries in the Réalmont Basin are limited to the Blima and Cantemerle formations, and vertebrate remains were only recovered from the latter (Delsahut 1981; Fauré 2022).

The Cantemerle Formation (also designated under the name ‘Formation gréso-pélitique grise’ in Fauré 2022) is 60 meters thick (Delsahut 1981; Fauré 2022). It consists of a succession of black argillites and sandstones, with few conglomerates at the base of the formation (Fig. 1D; Delsahut 1981; Fauré 2022). A few fragments of Autunia conferta and Odontopteris sp. pteridosperm ‘ferns’ are reported from black shales levels of this formation (Vetter and Mazeaud 1976). In the locality of Cantemerle (Réalmont urban area), this formation outcrops alongside the road to a height of just over 6 metres and a length of 25 metres, alternating between sandstone and greenish-grey shales with conglomeratic and brecciated passages (Fig. 1E–H). The visible succession records, from bottom to top: sandstone with conglomerate passages and oblique stratifications, composed of sandstone and basement pebbles (G1, 1 m thick), grey shales with spherical alteration (S1, almost 1 m thick), massive sandstone with spherical alteration (G2, 70 cm thick), grey shales with a reddish layer intercalated (S2, 40 cm thick), a breccia with quartz pebbles (C1, almost 1 m thick), black shales with fine layers, including a reddish layer (S3, 80 cm thick), massive greenish sandstone with medium-sized quartz grains (G3, 1 m thick), and finally black shales with fine layers (S4, 50 cm thick) (Fig. 1E–H). The sandstone layers are organised without grain sorting, with indeterminate plant debris (‘griffes charbonneuses’ [‘carbonaceous claws’]). Each coarse sequence ends with silting of the area with shale deposits, finely laminated. We interpret here this sequence as a lacustrine palaeoenvironment, interrupted by coarse alluvial events with debris flows. The vertebrate remains discovered here come from levels S3, above and below the reddish layer, and S4 (Fig. 1E–H). The articulated remains are concentrated at the base of the level S4. Vetter and Mazeaud (1976) do not indicate the stratigraphic position of the ‘fish’ scales they mention, but it is highly probable that these come from the same levels.

The studied specimens were collected in 2025 and 2026 at the Cantemerle locality (Réalmont urban area), during prospections carried out by two of us (DG and BD; Fig. 1H). The sampled material is currently housed at the Muséum d’Histoire Naturelle de Toulouse (MHNT, France). The collections from the 2025 campaign have a collection number beginning with MHNT.PAL.2026.1, while those from the 2026 campaign begin with MHNT.PAL.2026.14. We also conducted surveys in the Najac Basin to locate the ‘schistes à poissons’ [fishy shales] mentioned in Bergeron (1889: p. 242). Like Thévenin (1903) before us, these searches were unsuccessful and these layers are probably lost.

The vertebrate specimens were measured with a calliper, and photographed using a Nikon D700 camera in artificial lighting conditions. The coprolites were photographed using a Canon EOS 90D, with artificial lighting. Several photographs were taken for each specimen at different focal planes, and combined in Photoshop CS6 to produce fully in-focus images.

The chemical composition of three of the studied coprolites; MHNT.PAL.2026.1.11, MHNT.PAL.2026.1.12 and MHNT.PAL.2026.1.13 was analysed entirely non-destructively, using a Bruker M4 Tornado Plus energy dispersive micro µXRF spectrometer at the Museum für Naturkunde Berlin. An element distribution map was obtained using a 30-W Rh metal-ceramic micro-focus tube with a Be window coupled to a polycapillary lens. The spot size of the X-ray beam on the sample surface was approximately 20 µm (calibrated for Mo-Kα radiation) and maximum energy settings of 50 kV acceleration voltage and 600 µA beam current were used for obtaining the element distribution map. The pixel size and dwell time per pixel was set to 18 µm and 50 ms, respectively, and the total map area was 1833 × 1350 pixel over an area of 33 × 24.3 mm².

To improve the depth of field of the element map, an aperture of 1000 µm, incorporated in the patented Aperture Management System of the µXRF instrument, was used. In order to avoid attenuation of fluorescence X-rays by air molecules in the sample chamber, the map was acquired at a sample chamber pressure of 2 mbar. Characteristic fluorescence X-rays were detected simultaneously by two 60-mm² silicon drift detectors with energy resolutions of ≤145 eV (full width at half maximum calibrated for Mn-Kα radiation). Deconvolution and quantification of the acquired energy-dispersive X-ray spectra and background subtraction was done using the Bruker M4 Version 1.6 operating software. A Fundamental Parameters algorithm (Rousseau 1984a, 1984b), based on the Sherman equation (Sherman 1955) and implemented in the operating system, was used to obtain elemental concentrations on the basis of net counts for each element of interest.

In this article, regional stages are preferably used instead of international stages, because of the lack of absolute dating in the Réalmont Basin. The datation of this basin is currently based on the palaeoflora, which uses the regional European stages: Namurian, Westphalian, Stephanian, and Autunian (e.g., Doubinger and Vetter 1985; Knight and Wagner 2014). These stages are difficult to correlate with international stages (e.g., Knight et al. 2023). Therefore, the already suspected diachronism of the Autunian and Stephanian subdivisions (e.g. Becq-Giraudon et al. 1995; Knight and Wagner 2014; Knight et al. 2026), documented at regional scale (Opluštil et al. 2022) could also occur at a local geographical scale (Mercuzot et al. 2023), and here we follow the most recent correlation proposed by Aretz et al. (2020) and Henderson et al. (2020).

As in several recent articles dealing with the palaeobiogeography of the Stephano-Permian basins of the Variscan Belt (e.g., Gonçalves and Luccisano 2025b; Gonçalves et al. 2026a, 2026b, 2026c), we here assign the Saar-Nahe Basin to the geological entity of the Rhenish Massif. This assignment is based on the sedimentary source material filling these basins. The Saar-Nahe Basin is situated within the Northern Phyllite Zone (Mid-German Crystalline Rise), south of the Rhenish Massif and separated from it by a major fault: the Metz-Hunsrück-Taunus Fault (e.g., Linnemann 2025). The Rhenish Massif was eroded during the late Pennsylvanian and Permian, and sediments accumulated in intra-orogenic depressions, such as the Saar-Nahe Trough (e.g., Linnemann 2025: 110).

We follow the nomenclature of Fischer et al. (2010) to describe Lissodus-type fin spines.

We follow the nomenclature of Heidtke (2011a, 2011b) to describe the acanthodid remains with modifications, using the nomenclature of Zajíc (1998) for scapulocoracoid anatomy, and Heyler’s zones for the caudal fin squamation (Heyler 1969: p. 38, fig. A).

We follow the nomenclature of Schultze et al. (2021) to describe actinopterygian remains, meaning, in particular, that the terms ‘parietal’ and ‘postparietal’ are used here instead of the terms ‘actinopterygian frontal’ and ‘actinopterygian parietal’ respectively. We follow Schultze et al. (2021: p. 12) who use the term ‘intertemporo-dermosphenotic’ to designate the bone called ‘dermosphenotic’ in previous studies (e.g., Štamberg 1991, 2018; Schindler 2006, 2018a, 2018b; Štamberg and Steyer 2021). In late-Variscan macrodont actinopterygians, a single bone contributes to the postero-dorsal margin of the orbit, bearing the infraorbital sensory canal as well as its flexion and junction with the supraorbital canal. While the name ‘intertemporo-dermosphenotic’ may suggest a bone resulting from the fusion of the intertemporal and dermosphenotic, which is never specified by Schultze et al. (2021), it should be understood here as positional name only. This usage follows the one proposed for the Aeduellidae by Gonçalves et al. (2026a). The measurements follow those of Gonçalves and Luccisano (2025b: Suppl. material 1).

Scale morphology is described using the terminology of Esin (1990), with comments. Esin (1990) divided the body into eight regions to facilitate the description of the squamation of sufficiently similar morphology: regions A (anterior flank scales), B (mid-body flank scales), C (posterior flank scales), D (caudal peduncle scales), E (dorsal scales), F (ventral scales) and G–H (dorsal and anal fin base scales). We use this division here, while noting that the transition between body regions is subtle and that many isolated scales cannot be confidently attributed to a region (see also Schultze et al. 2021: 41). Scale ornamentation is described following Bakaev and Kogan (2022), with modifications. Due to the degree of abrasion often observed on the scales, we prefer to use the term “ridge” here to refer to positive linear structures, triangular or rounded-triangular in cross-section and separated by deep negative structures, whether prominent or flattened. These ridges are the elements highlighted by the lines in the interpretative drawings.

GPIM Institute für Geowissenschaften, Johannes Gutenberg University, Geologisch-Paläontologisches Institut collections, Mainz, Germany

GPIUH Geiseltal Museum, Universität Halle, Halle, Germany

MFMN Musée de la Faille de Meyssac à Noailhac, Noailhac, France

MHK Museum of Eastern Bohemia, Hradec Králové, Czech Republic

MHNT Muséum d’Histoire Naturelle de Toulouse, Toulouse, France

NHMUK Natural History Museum, London, England

NMP National Museum of Prague, Prague, Czech Republic

RH Rhinopolis Associative Museum, Gannat, France

Cohort EUSELACHII Hay, 1902

Order HYBODONTIFORMES Maisey, 1975

Hybodontiformes indet. (= cf. ‘Lissodus ’)

Fig. 2

Material. A partially preserved fin spine (MHNT.PAL.2026.1.4 [Fig. 2A–M]).

Description. The fin spine MHNT.PAL.2026.1.4 was discovered in the S3 layer (Fig. 1E), fragmented into two parts: the proximal part is the smaller of the two, measuring 3.8 mm in length, 0.7 mm in width, 2.2 mm in proximal height, and 2.0 mm in distal height; while the distal part measures 5.0 mm in length, 0.6 mm in width, 1.8 mm in proximal height, and 1.0 mm in distal height (Fig. 2A–L). It is difficult to estimate the total length of the spine, as the proximal and distal ends are missing. Based on known complete ‘Lissodus’ spines (see Comparisons below), we can assume a total length of approximately 35–45 mm (Fig. 2M). The spine has a gently curved posterior face, and its lateral and medial faces are well ornamented with continuous smooth longitudinal costae (Fig. 2G–J). There are seven costae proximally, reducing to six distally. The intercostal spaces are wider in the anterior part of the spine at the proximal end, but these spaces are all the same size at the distal end. No foramen are observed, and the costae do not bifurcate or anastomose. One strong costa forms a keel along the anterior border of the spine. Hooked denticles are arranged in a single median row along the posterior face of the spine. The tip of the denticles are all broken, and only their bases are well preserved (Fig. 2K–L). They are laterally compressed, with each denticle measuring approximately 0.50 mm in length. The denticles are all aligned on the posterior edge of the spine, evenly spaced, with the spacing decreasing distally. Three denticles are preserved on the proximal piece and five on the distal piece. Based on their arrangement, there would be approximately 12 denticles per cm. The cross section is oval proximally and becomes more ovoid distally with a more pronounced keel (Fig. 2A–F). This section shows an outer cavernous layer and a large central cavity of oval shape, the size (width and height) of which represents half the total dimensions of the spine in cross section.

Comparisons. The Réalmont spine (MHNT.PAL.2026.1.4) displays costae ornamentation, a single row of denticles and an ovoid distal cross section, features characteristic of hybodontiforms, (Maisey 1978). Among Palaeozoic taxa, these characteristics are, so far, only found in the genus Lissodus (e.g., Soler-Gijón 1997; Milner and Kirkland 2006; Fischer et al. 2010). This genus is known from the Carboniferous to the Cretaceous and mainly diagnosed on the basis of its teeth (e.g., Duffin 1985; Rees and Underwood 2002; Fischer 2008; De Lange et al. 2023). Lissodus was long considered a member of Lonchidiidae (e.g., Duffin 1985; Rees and Underwood 2002; Ginter et al. 2010), but is now considered an indeterminate hybodontiform (Rees 2008; Koot et al. 2013; Wen et al. 2022). It is commonly accepted that the Palaeozoic forms of Lissodus belong to distinct genera, with dentition convergent with that of Lissodussensu stricto (Rees and Underwood 2002; Maisey et al. 2004; Koot et al. 2013; De Lange et al. 2023; Shell et al. 2024). However, no comprehensive review of the Palaeozoic material has been undertaken, and these specimens should therefore be considered as Hybodontiformes indet. (Koot et al. 2013). In order to determine the relationships of the Réalmont spine with Lissodussensu stricto and the Hybodontiformes indet. previously attributed to this genus (here referred to as ‘Lissodus’), comparisons are limited to taxa for which fin spines are known (see Fischer 2008 for a review).

Among the Palaeozoic taxa ‘Lissodus’, the Réalmont fin spine shares many characters with ‘L.’ lacustris, ‘L.’ lopezae and ‘L.’ sardiniensis: an anterior keel along the entire spine length, an absence of anastomosis or bifurcation of the costae, a single row of denticles and an ovoid cross section (Schneider 1986; Soler-Gijón 1997; Fischer et al. 2010). The Réalmont spine differs from these taxa in that it has 7 costae in its proximal part, compared to 6 in ‘L.’ lacustris, 5 in ‘L.’ lopezae and 4 in ‘L.’ sardiniensis (Schneider 1986; Soler-Gijón 1997; Fischer et al. 2010). It also differs in its number of denticles per cm (12), which is lower in ‘L.’ lacustris and ‘L.’ lopezae (6 denticles per cm) and higher in ‘L.’ sardiniensis (20 denticles per cm) (Schneider 1986; Soler-Gijón 1997; Fischer et al. 2010). It should be noted that differences in the distance between denticles may represent individual variation (Fischer et al. 2010). ‘Lissodus’ sp. NM (abbreviation for Niedermoschel locality used in Hampe 1996) cannot be compared to the Réalmont spine due to the absence of anatomical overlap, as it is known only from a fragment of the proximal end of the fin spine (Hampe 1996). Fin spines of ‘Lissodus’ have also been described from the Lodève and Bourbon-l’Archambault basins (Steyer et al. 2000; Fischer 2008; Štamberg and Steyer 2021), but have not been described and cannot be compared to the Réalmont spine.

In Lissodussensu stricto, only fin spines from Bonenburg (Upper Triassic, Germany) and Wealden (Early Cretaceous, England) attributed to Lissodus sp. share an absence of anastomosis or bifurcation, a single row of denticles and a small number of costae (e.g., Patterson 1966; Soler-Gijón 1997; Fischer et al. 2010; Vida et al. 2025). They differ from the Réalmont spine in having fewer costae and are insufficiently described to be compared in more detail.

All of the characters of the Réalmont spine are unique among the fin spines attributed to ‘Lissodus’ and Lissodus spp. Its fragmentary nature calls for caution and the need to wait for additional remains to confirm our observations and complete its diagnosis. We therefore assign MHNT.PAL.2026.1.4 to Hybodontiformes indet., an attribution equivalent to cf. ‘Lissodus’ if we retain the old nomenclature.

Order ACANTHODIFORMES Berg, 1940

Family ACANTHODIDAE Huxley, 1861

Acanthodidae indet.

Fig. 3

Material. Posterior portion of a skeleton (MHNT.PAL.2026.14.72 [Fig. 3A, B]); disarticulated patches of scales associated with fin spines (MHNT.PAL.2026.14.73 [Fig. 3C, D], MHNT.PAL.2026.14.69 [Fig. 3E]); a slab with a fin spine and a mandibular splint on one side, and a partial skull, a scapulocoracoid and numerous disarticulated scales on the other side (MHNT.PAL.2026.14.78 [Fig. 3F–I]); a near-complete scapulocoracoid (MHNT.PAL.2026.1.2 [Fig. 3J]); three isolated fin spines (MHNT.PAL.2026.14.70, MHNT.PAL.2026.14.75); associated patches of scales (MHNT.PAL.2026.14.74).

Description. MHNT.PAL.2026.14.72 is the posterior portion of a skeleton, preserving the almost complete caudal fin and the anal fin spine (Fig. 3A, B), and originating from the layer S4 (Fig. 1E). It measures 29.0 mm in length, with a maximum depth of 7.3 mm and a minimum depth of 3.3 mm, just before the base of the caudal fin. The anterior part of the specimen shows the beginnings of curvature, a post-mortem phenomenon common in acanthodians. The anal fin spine measures 6.5 mm in length and 0.5 mm in maximum width at its base. It is situated 7 mm from the base of the caudal fin. The anal fin spine is sabre-shaped: it is elongated proximo-distally, tapering slightly distally and slightly curved. Its surface is smooth, with a longitudinal groove along its lateral side and it bears no denticles. The tips of the two caudal fin lobes are not preserved, and there are no traces of the anal or dorsal fin. The caudal fin measures 10.1 mm in preserved length, with a dorsal lobe of 8.1 mm and a caudal lobe of 7.0 mm. Most of the body surface is irregular with poorly defined scales – too poorly preserved to distinguish Heyler’s zones in the caudal region – but a few well-preserved scales are located posteriorly to the anal fin spine, and in the ventral lobe of the caudal fin. Scales are smooth with a rhombic to subrhombic shape. The largest well-preserved scales are those located posterior to the anal fin spine, and are 350 µm in length (Fig. 3B).

MHNT.PAL.2026.14.69 is a slab containing numerous scattered scale patches, unidentified fragments, the counterpart of the distal tip of a fin spine, and a fragment of a fin spine (Fig. 3E). All these elements are associated and partially overlap. The fin spine fragment and the fin spine counterpart measure 3.0 mm and 2.3 mm in maximum width respectively. MHNT.PAL.2026.14.73 is a slab and counterslab pair containing numerous scattered scale patches associated with a fin spine. A rectangular-shaped bone associated with the proximal end of the spine is most likely a procoracoid. These elements are all surrounded by a halo of lighter-coloured sediment, a feature usually absent on slabs from this sediment, and which may represent an altered body imprint. (Fig. 3C, D). The fin spine measures 26.0 mm in length and 2.3 mm in maximum width at its base, and is identified here as a pectoral fin based on its relative size and curvature. Both MHNT.PAL.2026.14.69 and MHNT.PAL.2026.14.73 originate from layer S4 (Fig. 1E). Fin spines are sabre-shaped, with a smooth surface and a longitudinal groove along their lateral side. Scales are smooth, often rhombic to subrhombic in shape, but sometimes ovoid in MHNT.PAL.2026.14.73 (Fig. 3D). These ovoid scales are found alongside the other scales, with no distinction or zonation. The largest scales are those of MHNT.PAL.2026.14.73, and are 350 µm in length (Fig. 3D).

The slab MHNT.PAL.2026.14.78 bears acanthodians remains on both sides, and originates from the layer S4 (Fig. 1E). The first side contains a fin spine, a mandibular splint and around ten scattered gill rakers (Fig. 3H, I). The second side of the slab contains an indeterminable viscerocranial element, three associated pairs of branchiostegal rays, a scapulocoracoid counterpart and numerous scattered scale patches (Fig. 3F, G). The viscerocranial element forms a vaguely triangular mass measuring 8.6 × 6.9 mm, with clearly visible but irregular contours and no visible sutures within it. No individual elements are distinguishable, but this may represent the dorsal part of the neurocranium. The mandibular splint is thin, elongated anteroposteriorly with a total length of 8.2 mm. The mandibular splint bends antero-dorsally by 32° at the first third of its length, dividing it into two branches: a posterior branch 3.3 mm long, missing its posterior end, and an anterior branch 5.3 mm long. The width of the mandibular splint is uniform throughout (0.4 mm), until the anterior end where it tapers to form a pointed tip. Three pairs of branchiostegal rays are preserved, and each ray has the same morphology: they are smooth, sabre-shaped with a curved anterior section and a straight posterior section, and measure 1.4 mm in length and 0.11 mm in width. The gill rakers are also smooth and sabre-shaped, but are smaller: they measure a maximum of 0.82 mm in length and 0.08 mm in width. The scapulocoracoid is shaped like an inverted Y, with a broad ventral part that tapers dorsally at the shaft. The ventral part divides into two branches, the wider of which is the anterior branch, which forms the surface for the dorsal adductor muscles, and the posterior branch, which is the procoracoid process. The shaft measures 5.7 mm in height and 1.4 mm in width, whilst the anterior and posterior branches measure 3.7 mm in length and 1.7 mm in width, and 2.8 mm in length and 1.1 mm in width, respectively. The fin spine is sabre-shaped, with a smooth surface and a longitudinal groove along its lateral side. It measures 15.7 mm in length and 0.9 mm at its widest point at the base. The scales are smooth, often rhomboid to sub-rhomboid in shape.

MHNT.PAL.2026.1.2 is a bone with an eroded surface and missing dorsal and ventral ends, originating from the layer S4 (Fig. 1E). Its morphology corresponds to a scapulocoracoid, with a broad ventral part that tapers dorsally at the shaft (Fig. 3J). It measures 8.0 mm high, 3.4 mm long ventrally and 1.0 mm long dorsally. The ventral part of MHNT.PAL.2026.1.2 bifurcates into two processes, but these are very poorly preserved. It is not possible to assess the cross-section of the bone due to its state of preservation. The other elements usually associated with the acanthodian scapulocoracoid (procoracoid and suprascapula) are not preserved here.

MHNT.PAL.2026.14.70 is a distal tip of a fin spine, measuring 19.5 mm in length and 2.6 mm at its maximum width at the proximal broken end. MHNT.PAL.2026.14.75 is a slab containing a fragment of a fin spine and a distal end of a fin spine, along with a body fragment of Progyrolepis cf. heyleri (see below). Both spines are originating from the layer S4 (Fig. 1E) The fragment measures 13.2 mm in length and 2.0 mm in maximum width. The distal end of the fin spine measures 10.6 mm in length and 1.8 mm in maximum width at its proximal broken end. Fin spines are sabre-shaped, with a smooth surface and a longitudinal groove along their lateral side.

MHNT.PAL.2026.14.74 consists of associated patches of scales, which vaguely form a fusiform-shaped body, with no visible fins or spines. It was discovered in the layer S4 (Fig. 1E) The specimen measures 20.0 mm in its maximum length. A few scales are visible in some areas, but the specimen consists mainly of an irregular surface of lighter-coloured sediment, probably representing the weathered remains of a body. The scales are smooth, rhomboid to sub-rhomboid in shape.

Comparisons. The remains from Réalmont exhibit characters unique to the acanthodids of the European Stephano-Permian basins, namely Acanthodes and Westrichus (Heidtke 2011a, 2011b). These include the combination of a bend, toothless mandibular splint, an inverted Y-shaped scapulocoracoid, sabre-shaped fin spines with a smooth surface, lacking denticles and featuring a longitudinal groove, and small, smooth rhombic scales (e.g., Heyler 1969; Heidtke 1990a, 1990b, 1993, 1996, 2011a, 2011b; Zajíc 1998, 2005; Beznosov 2009). However, the specific identification of acanthodids is difficult in the absence of complete specimens, as diagnostic features are based primarily on body and fin spine ratios (e.g., Zidek’s diagram; see Zidek 1976; Zajíc 1998) and characters of the axial skeleton (e.g., Heidtke 2011a; Zajíc 1998, 2005). The characters of the mandibular splint, scapulocoracoid, and fin spines are plesiomorphic for the family (e.g., Heidtke 1996, 2011b; Zajíc 1998; Burrow 2021). Furthermore, among the European Stephano-Permian species, only a few (i.e., Acanthodes confusus, A. fritschi, A. gracilis, A. stambergi) have been sufficiently described and illustrated to be compared in detail with the Réalmont material (e.g., Miles 1973; Zajíc 1998, 2005; Dearden et al. 2025). Many well-preserved specimens from Spain and France have also not been the subject of comprehensive studies (e.g., Heyler 1969, 1980; Forey and Young 1985), and only the German and Czech specimens have precise taxonomic identifications (see Zajíc 2005 and Heidtke 2011b for a review). The ossification of the cranial visceral skeleton distinguishes the Réalmont acanthodid from numerous species for which the neurocranium is presumed to be unossified: Acanthodes bourbonensis, A. bronni, A. palatinensis, A. pollichiae, A. ruhinensis, A. splendidus, A. ultimus, and Westrichus tholeyi (Heidtke 1996, 2011a, 2011b). However, the vast majority of these species have been described briefly and based on a limited number of specimens (Heidtke 2011b), and it would be necessary to investigate in greater detail the variations in these ossifications, which may depend on ontogeny, individual variation, and taphonomic conditions. The scales of the Réalmont specimens lack posteriorly oriented spiny projections, unlike Acanthodes fritschi, A. gracilis, and A. stambergi (Zajíc 1998, 2005), nor do they have gill rakers covered by striae or tiny ridges as in Acanthodes fritschi (Zajíc 1998). Ultimately, among the European Stephanian-Permian acanthodians, only Acanthodes confusus and Westrichus kraetschmeri most closely resemble the Réalmont acanthodid, sharing all of its characters (Miles 1973; Heidtke 2011a, 2011b; Dearden et al. 2025). In the absence of solid diagnostic criteria and more complete material, it is not possible to assign the Réalmont acanthodid to a specific taxon, and for these reasons, we assign these remains to Acanthodidae indet.

Family indet.

Genus PROGYROLEPIS Fritsch, 1895

Fig. 7

Material. Four specimens (MHNT.PAL.2026.1.10 [Fig. 7C], MHNT.PAL.2026.1.11 [Fig. 7A, F], MHNT.PAL.2026.1.12 [Fig. 7D–F], MHNT.PAL.2026.1.13 [Fig. 7B, F]).

Comparative description. A total of four coprolites were found during the recent sampling of the locality. The coprolite specimens are visible in cross section on the surface of the material sampled at the locality, and appear black and shiny under natural light (Fig. 7A–E). The coprolitic matrix is characterised by very fine crystalline texture and conchoidal fractures. One of us (AR) observed that their preservational state closely resembles those of coprolites found in the early Autunian from the Merdanson and Maro rivers outcrops (Usclas-Saint-Privas Formation, Lodève Basin), suggesting a similar taphonomic pathway. MHNT.PAL.2026.1.11 and MHNT.PAL.2026.1.12 are the most complete coprolites, displaying an ovoid morphology characterised by a broader, rounded end, and a narrower, tapering end (Fig. 7A, D). Their size does not exceed 10 mm in length and 4 mm in width. None of these coprolites present a visible spiralled structure (e.g., Hunt and Lucas 2012), a feature characteristic of chondrichthyans (including acanthodians), non-tetrapod sarcopterygians and non-teleostean actinopterygians (McAllister 1987; Argyriou et al. 2016)

Elemental analysis with µXRF reveals that the coprolites are mainly composed of calcium and phosphorus, homogeneously distributed in the coprolite’s matrix (Fig. 7E, F), suggesting calcium phosphate minerals. In contrast, the surrounding sedimentary matrix is mostly composed of aluminium, silicium and iron (Fig. 7E, F). Three quadrangular scales are observable within the coprolitic matrix (Fig. 7D, E) of MHNT.PAL.2026.1.12, two of which being sub-complete, evidencing the carnivorous behaviour of its producer. The surface of the scales is weathered, highlighting the growth lines. The weathering of the scales surface, most likely due to digestive processes of the producer, has not completely erased their ornamentation, and a series of ridges can be observed. These ridges run diagonally across the scales, from the anterodorsal corner to the posteroventral corner of the scale. They divide the lateral surface of the scale into the anteroventral and posterodorsal halves. The arrangement of ridges on the two halves is different. Few ridges are recognisable on the anteroventral region of the scale, starting diagonally at the anterior margin of the scale and gently curving to become horizontal posteriorly. The posterodorsal half of the scales has few large ridges which start at the anterodorsal corner of the scale, branch posteriorly, and terminate in multiple diagonal ridges that end at the posterior edge of the scale. This scale ornamentation is identical to the second scale morphotype (mid-body flank scales) of Progyrolepis cf. heyleri described above (Figs 4, 5, 6).

Based on the coprolite size and the known fauna from the locality, potential producers likely include the herein described hybodontiform cf. “Lissodus” or the actinopterygian Progyrolepis cf. heyleri. However, the lack of spiral structure, typical of chondrichtyans (e.g., McAllister 1987), argues against the cf. “Lissodus”. Moreover, this taxa exhibits an overall heterodont arrangement of low-crowned teeth, characteristic for a crushing type dentition. Since such dentition suggest a durophagous bottom-dwelling lifestyle (Duffin 1985; Hampe 1996; Fischer et al. 2010; Vida et al. 2025), the presence of Progyrolepis cf. heyleri scale within the coprolitic matrix makes cf. “Lissodus” an unlikely candidate, even though predation (e.g., Duffin 1985; Fischer et al. 2010) and/or facultative scavenging cannot be entirely excluded. Progyrolepis cf. heyleri. seems to be the most plausible candidate as producer of the coprolites, as it is supposed to have actively hunted other actinopterygians (Štamberg 2018, 2020) and given the current knowledge on this faunal assemblage.

The ichthyofauna from the Cantemerle Formation provides the opportunity to revisit its dating (Fig. 8), which until now had been solely based on the presence of Odontopteris sp., Autunia conferta and ‘fish’ scales (Vetter and Mazeaud 1976; Delsahut 1981). This assemblage suggested an Autunian age for this formation, and its stratigraphic position and facies suggested an early Autunian age (Vetter and Mazeaud 1976; Delsahut 1981). The presence of Hybodontiformes indet. (= cf. ‘Lissodus’), Acanthodidae indet. and Progyrolepis cf. heyleri confirms this age. The taxa assigned to ‘Lissodus’ and the indeterminate acanthodids have a wide stratigraphic distribution, spanning the entire Stephano-Autunian (e.g., Brauckmann et al. 2007; Fischer 2008; Heidtke 2011b; Gonçalves et al. 2025). Progyrolepis heyleri is limited to the early Autunian of the French Massif Central, in the Bourbon-l’Archambault (Buxières Formation) and Brive (Saint-Antoine Limestones Formation) basins (Štamberg 2018; Štamberg and Steyer 2021). It is found in Autunian deposits located just above the Autunian-Stephanian C boundary (Debriette 1992; Feys 1976; Châteauneuf and Farjanel 1989; Mercuzot et al. 2023). In the Buxières Formation, the vertebrate-bearing level is radiochronologically constrained, with a calculated age of 298.592 ± 0.075 Ma, indicating an early Asselian age close to the Carboniferous-Permian (C-P) boundary (Mercuzot et al. 2023). The new discoveries presented here therefore confirm the early Autunian age proposed for the Cantemerle Formation, near the C-P transition.

Predators with varied ecologies and feeding behaviours are found within the lacustrine deposits of the Cantemerle Formation. ‘Lissodus’ hybodontiforms are considered slow, demersal durophagous chondrichthyans (Fischer et al. 2010; Vida et al. 2025), feeding on benthic hard-shelled invertebrates (e.g., gastropods, crustaceans, bivalves) but also capable of capturing other preys lacking a shell (e.g., Duffin 1985; Fischer et al. 2010). Acanthodids from the Stephano-Permian basins of Europe are edentulous zooplanktivores (e.g. Heidtke 2011b; Burrow 2021), and this was certainly the case for the Réalmont acanthodid. Finally, Progyrolepis is considered as a powerful swimmer which actively hunted other actinopterygians and tetrapods (Štamberg 2018, 2020). The coprolites described here, if they are indeed produced by P. cf. heyleri, would tend to indicate that this taxon was also cannibalistic. To this day, the faunal assemble studied here lack of taxa belonging to the lowest trophic levels, but also that of groups that were nevertheless common in most early Autunian Stephano-Permian basins in Europe (e.g., Xenacanthiformes, Aeduelliformes) (e.g., Heyler 1969, 2000; Brauckmann et al. 2007; Štamberg and Zajíc 2008; Luccisano et al. 2021; Gonçalves et al. 2025, 2026a, 2026b, 2026c). The absence of such taxa could be linked to different ecological preferences and collecting bias.

Progyrolepis is a macrodont actinopterygian reaching approximately 70 to 80 centimetres in length (Štamberg 1991, 2018; Štamberg and Steyer 2021). Two valid species are currently recognised: P. speciosus (type species; Fritsch 1895; Štamberg 1991, 2013) and P. heyleri (Poplin 1999; Štamberg 2018; Štamberg and Steyer 2021). The fossil record of Progyrolepis is often incompletely listed, with invalid occurrences (e.g., Schultze et al. 2021; Henderson et al. 2023). Given its relative abundance, Progyrolepis is a key taxon in the Cantemerle locality, particularly for understanding its palaeobiogeographic relationships. To this end, a review of the occurrences of this taxon is crucial and is presented here.

Progyrolepis speciosus is known from the Stephanian B-C of the Slaný and Líně (Kladno–Rakovník and Mšeno-Roudnice basins), and Syřenov and Semily (Krkonoše Piedmont and Mnichovo Hradiště basins) formations (Fig. 9; Štamberg 1991, 2013, 2016a; Štamberg and Zajíc 2008). P. heyleri is known from the early Autunian of the Buxières (Bourbon l’Archambault Basin) and Saint-Antoine Limestones (Brive Basin) formations (Fig. 8; Poplin 1999; Heyler 2000; Štamberg 2018; Štamberg and Steyer 2021). We report here P. cf. heyleri from the early Autunian of the Cantemerle Formation (Réalmont Basin). All these occurrences are based on the combination of diagnostic characters of Progyrolepis, absent in other actinopterygians. cf. Progyrolepis sp. is mentioned in the Stephanian C of the Bourran Formation (Decazeville Basin), based on an isolated fulcral scale (Fig. 8; Gonçalves et al. 2025). This scale is similar to those described for Progyrolepis heyleri (Štamberg 2018; Štamberg and Steyer 2021; Gonçalves et al. 2025), and differs from those of other late-Variscan actinopterygians in which these scales are well-described (Gonçalves et al. 2025). However, this attribution remains very provisional and needs to be confirmed by a better description of the fulcral scales of other taxa such as Blanzychthys prorobisoni, Briveichthys chantepieorum or Letovichthys spp.

Other specimens or taxa have been historically attributed to Progyrolepis, but these attributions are invalid or warrant revision, given the current diagnosis of the genus (Fig. 9; Štamberg 2018; Štamberg and Steyer 2021). In their work on actinopterygians from the Stephano-Permian basins of the French Massif Central, Heyler (1969: pl. XX, figs 7; 1976, 2000: 14–16, fig. 3) and Heyler and Poplin (1983: 37–38, pl. II; 1990: 229–230, fig. 3B; 1994: 210) regularly attributed ornamented scales, either isolated or as body fragments to cf. Palaeoniscum or Palaeoniscum sp. These attributions were based on the putative resemblance to the ornamentation of the scales of Palaeoniscum freieslebeni (e.g., Heyler 1976, 2000; Heyler and Poplin 1990). These specimens are from the localities of Buxières-les-Mines, La Queue de l’Etang (Bourbon-l’Archambault Basin) and Montceau-les-Mines (Montceau-les-Mines Basin). After the description of Progyrolepis heyleri from Buxières-les-Mines by Poplin (1999), Heyler (2000: 17) indicated that the material from this locality previously attributed to Palaeoniscum could belong to P. heyleri, which was confirmed by Štamberg (2018). Based on the new information provided by P. heyleri, Heyler (2000: 17) considered it likely that the other French Stephano-Permian remains assigned to Palaeoniscum belong to P. heyleri. The ornamentation of the scales of these specimens is not sufficiently visible in the figures to allow for a detailed comparison with that of Progyrolepis. On the other hand, the tuberculate ornamentation of the skull roof, the parasphenoid without basipterygoid process, and the fulcral scales of cf. ‘Palaeoniscum’ from La Queue de l’Etang locality (Heyler 1976: figs 2, 3) are identical to those described in P. heyleri (Štamberg 2018; Štamberg and Steyer 2021). The taxonomic attribution of cf. ‘Palaeoniscum’ from the French Massif Central should be treated with caution until the material has been examined, and is retained here under the identification Actinopterygii indet.

Progyrolepis speciosus appears regularly in the vertebrate fauna list of the Puertollano Basin (Spain) (Soler-Gijón 1994: Soler-Gijón and Moratalla 2001; Soler-Gijón and Díez Ruiz 2023). It is only mentioned but was never described or illustrated. Since the first publication of this list, the diagnosis of P. speciosus has been emended (Štamberg 2013, 2016a), the species P. heyleri has been described (Poplin 1999; Štamberg 2018), as have several macrodont actinopterygians with ornamented scales (i.e. Blanzychthys, Briveichthys, Letovichthys) (Heyler and Poplin 1994; Štamberg 2007; Štamberg and Steyer 2021). The record of P. speciosus in the Puertollano Basin therefore deserves to be revised in light of recent advances, and the specimens from Puertollano should in the meantime be treated cautiously as Actinopterygii indet. Palaeoniscum cf. freieslebeni also appears in the faunal list of the Puertollano Basin (Soler-Gijón 1994; Soler-Gijón and Moratalla 2001; Soler-Gijón and Díez Ruiz 2023), without being accompanied by a description or figures (Soler-Gijón 1994; Soler-Gijón and Díez Ruiz 2023). As with the material assigned to Palaeoniscum in the French Massif Central (see above), it is possible that the specimens assigned to this taxon are in fact attributable to Progyrolepis, especially if Soler-Gijón (1994) followed the works of Heyler (1969, 1976) and Heyler and Poplin (1983, 1990). Again, a description of the specimens is necessary to assign them with confidence, and in the absence of such a description, they should be considered as Actinopterygii indet.

Dunkle (1946) described the new species Progyrolepis tricessimalaris from the early Permian of the Admiral Formation (Texas, USA). This attribution to Progyrolepis was based on Fritsch (1895) work on the genus, prior to its revision by Štamberg (1991, 2016b, 2018). ‘P.’ tricessimalaris differs greatly from Progyrolepis in that it has two ventral infraorbitals; a preoperculum divided into two, a maxilla with a crescent-shaped maxillary plate, ornamented with tubercles; and mid-body flank scales with only diagonally oriented ridges. ‘P.’ tricessimalaris should therefore be excluded from the genus Progyrolepis, as recommended by Štamberg (1991, 2013). Revision of the holotype could identify unique characters that would solve its taxonomic status and relationships.

Following Dunkle’s (1946) work attesting to the presence of Progyrolepis in North America, Langston Jr (1953) assigned fragmentary remains to Progyrolepis sp. from the Wolfcampian (= early Permian) of the El Cobre Canyon Formation. This assignment was subsequently followed by various faunal lists (e.g., Lucas et al. 2005; Hodnett and Lucas 2015). No diagnostic characters of Progyrolepis are present in the remains described by Langston Jr (1953), which belong to an indeterminate macrodont actinopterygian.

More recently, Huttenlocker et al. (2024) attributed a disarticulated skeleton and an isolated dentary from the Virgilian (= Gzhelian) of the Maroon Formation (Colorado, USA) to aff. Progyrolepis. Despite the abundance of skull elements, only the teeth have been described in detail (Huttenlocker et al. 2024). The morphology of the bones is not described, and among the skull bones, only elements of the palate and jaws are illustrated (Huttenlocker et al. 2024: fig. 7). The attribution to aff. Progyrolepis was made on the basis of the presence of two rows of marginal teeth and vermiform ornamentation of the scales and bones. The presence of two rows of marginal teeth is plesiomorphic in Palaeozoic actinopterygians and is found in numerous taxa (see Schultze et al. 2021 for an overview). The ornamentation depicted in the Maroon specimens is clearly not vermiform as in Progyrolepis (Štamberg 1991, 2018; Štamberg and Steyer 2021), consisting only of ridges on the scale and maxilla (Huttenlocker et al. 2024: fig. 7A, C), and ridges associated with tubercles on the dentary (Huttenlocker et al. 2024: fig. 7D). Additionally, the maxilla is very poorly preserved: the shape of the maxillary plate cannot be determined and its ornamentation of short ridges is only visible on the anterior part of the bone (Huttenlocker et al. 2024: fig. 7A). Only one scale is depicted, described as a “ridge scale” (Huttenlocker et al. 2024: fig. 7C), but the absence of a median relief excludes this identification. It is probably a dorsal scale from the back line, whose ornamentation resembles that of Progyrolepis spp., but more elongated anteroposteriorly. Doubts about the identification by Huttenlocker et al. (2024) had been raised by Gonçalves et al. (2026b) and are confirmed here. Specimens described by Huttenlocker et al. (2024) have a plesiomorphic morphology and do not show diagnostic characters of Progyrolepis. These specimens should therefore be assigned to Actinopterygii indet. pending their more complete description.

The material from the Cantemerle locality allows us to discuss the affinities of Progyrolepis with other macrodont taxa, particularly with regard to the ornamentation of its scales. Both species of Progyrolepis are known from relatively complete skull remains, although P. heyleri is more completely known (Fig. 10; Štamberg 1991, 2018; Štamberg and Steyer 2021). However, the Progyrolepis body is more obscure: only fragments of the body are known (Štamberg 1991, 2016a, 2018; Štamberg and Steyer 2021). As a result, no scale count has been established for its species and no complete body reconstruction is possible. The material from Réalmont helps to better document the posterior region of the body of this genus, but does not allow us to establish a complete scale count. Progyrolepis differs from other actinopterygians by its lobed interparietal suture; an intertemporo-dermosphenotic-intertemporal borderline not extending beyond the parietal-postparietal borderline; a parasphenoid without a basipterygoid process; paired ventral rostro-premaxillae; a lacrimal and no additional ventral infraorbital; two suborbitals; a preoperculum conspicuously bent at an angle of around 130°; a maxilla with a trapezoidal maxillary plate and with a posteroventral process; a length-depth ratio of the maxillary plate of 1.6–2, and a maxilla/maxillary plate length ratio of 1.6–1.9; a maxillary plate with outstanding sculpture formed by vermicularly curved ridges; an oblong operculum 3.0 times deeper than it is long and 2.5 times deeper than the suboperculum; flank scales with conspicuous sculpture formed by ridges which diagonally traverse across the scale, dividing it in two halves (Štamberg 1991, 2018; Štamberg and Steyer 2021). The two species of Progyrolepis are very similar and can only be distinguished by their operculum: in P. heyleri, the operculum is rectangular without an epipreoperculum, whereas the operculum of P. speciosus is more rounded, with the sloping of the ventro-anterior part of the bone forming a space for the triangular epipreoperculum (Fig. 10; Štamberg 1991, 2016a, 2018; Štamberg and Steyer 2021).

Based on current records, Progyrolepis is limited to the Stephanian B-Autunian of the Variscan basins of the Bohemian Massif and the French Massif Central (Fig. 9). Its relationship to other actinopterygians has always been debated. Fritsch (1895) considered it as an indeterminate actinopterygian. It was first classified among the Palaeoniscidae (Dunkle 1946; Kazanceva-Selezneva 1977), a wastebasket group comprising the majority of late Palaeozoic actinopterygians, based solely on plesiomorphic characters (e.g., Schultze et al. 2021). Štamberg (1991, 2013, 2016a) proposed placing Progyrolepis within PygopteridaesensuAldinger (1937), mainly because of similarities in the cheek bones and opercular apparatus. Štamberg (2018) refuted this classification, arguing that Progyrolepis has a parasphenoid without a basipterygoid process and a lobed interparietal suture, unlike pygopterids. The diagnosis of Pygopteridae is difficult to establish, as no recent revision of the type species Pygopterus humboldti has been carried out (Agassiz 1833–1843; Schultze et al. 2021; Friedman et al. 2024), the work on Pygopterus being solely based on referred species (Aldinger 1937; Hoşgör and Štamberg 2014). The absence of ridge scales and scutes in front of the fins also appears to be a distinctive criterion, at least in Pygopterus (Aldinger 1937; Schultze et al. 2021), and differs from Progyrolepis (Štamberg 2018; Štamberg and Steyer 2021). Štamberg (2018) proposed classifying Progyrolepis within AcrolepidaesensuAldinger (1937), without comments. As with pygopterids, acrolepids are difficult to diagnose as the type species Acrolepis sedgwicki is known from fragmentary remains (Agassiz 1833–1843; Westoll 1934; Aldinger 1937; Štamberg 1991; Schultze et al. 2021), and numerous species have been incorrectly attributed to Acrolepis (Aldinger 1937; Štamberg 1991; Schultze et al. 2021). A. sedgwicki has an intertemporo-supratemporal, a supraorbital, a ventral infraorbital, a row of accessory extrascapulars, and a shallow operculum (Aldinger 1937; Schultze et al. 2021). All of these characters differ greatly from Progyrolepis (Štamberg 1991, 2018; Štamberg and Steyer 2021), suggesting that any affinity with Acrolepidae should be excluded.

Progyrolepis was only very recently included in a cladistic analysis, during work carried out on Aeduellidae (Gonçalves and Luccisano 2025b; Gonçalves et al. 2026a, 2026c). These analyses focused on late Variscan actinopterygians and are therefore not representative of relationships at the whole Actinopterygii scale, but only at the Variscan scale (Gonçalves and Luccisano 2025b; Gonçalves et al. 2026a, 2026c). Progyrolepis is placed in a clade of macrodont taxa with Briveichthys chantepieorum and Elonichthyidae (Gonçalves and Luccisano 2025b; Gonçalves et al. 2026a, 2026c). Progyrolepis shares with them characters distributed among all macrodont taxa, such as a large flat maxilla, two rows of marginal teeth, numerous branchiostegal rays and a khepesh-shaped cleithrum – see Gonçalves and Luccisano (2025b) for a definition (Gonçalves and Luccisano 2025b; Gonçalves et al. 2026a). It also shares more specific characters: ornamentation of the skull roof consisting of tubercles or short ridges; a very elongated parietal; an intertemporo-dermosphenotic, as long, or almost as long, as the supratemporal, a large L-shaped jugal; and one to three suborbitals (Štamberg 1991, 2018; Schultze et al. 2021; Štamberg and Steyer 2021). These characters are also found in Letovichthys tuberculatus but cannot be verified in ‘Acrolepis’gigas (Štamberg 1991, 2007), which were not included in the cladistic analyses (Gonçalves and Luccisano 2025b; Gonçalves et al. 2026a, 2026c). Progyrolepis is even closer to the Elonichthyidae in that it has rectangular postparietals, a parasphenoid without a basipterygoid process, and the same ornamentation of the mid-body flank scales (Schindler 2017, 2018a, 2018b; Štamberg 2018; Štamberg and Steyer 2021). However, we refrain from assigning Progyrolepis to Elonichthyidae because it has a lacrimal but no ventral infraorbital (followingSchultze et al. 2021), it also lacks a presupracleithrum, the shape of its maxilla is different, and its most dorsal branchiostegal ray is the same size as the others (Štamberg 2010, 2018; Schindler 2018a; Štamberg and Steyer 2021). The assignment of Progyrolepis to a particular family will need to be supported by a cladistic analysis including a wider range of macrodont actinopterygians from the Palaeozoic (e.g., Friedman et al. 2024). However, we note the possibility of a link between all Variscan forms, and more particularly between Progyrolepis and Elonichthyidae.

The faunal assemblage from the Cantemerle locality of the Réalmont Basin includes an undetermined acanthodid, an undetermined hybodontiform of the ‘Lissodus’ type, and the ‘macrodont’ actinopterygian Progyrolepis cf. heyleri. Acanthodids are found in almost all European Stephanian-Permian basins that have yielded vertebrate fauna (e.g., Brauckmann et al. 2007; Štamberg and Zajíc 2008; Gonçalves et al. 2025). In contrast, ‘Lissodus’ and Progyrolepis are only present in a few basins, and rarely associated (Fig. 11). The ichthyofaunal content from the Cantemerle locality is identical to that found in the Buxières-les-Mines locality (Buxières Formation, Bourbon-l’Archambault Basin), which is assumed here to be synchronous (Steyer et al. 2000; Štamberg 2018). In the Stephanian C of the Líně (Kladno–Rakovník and Mšeno-Roudnice basins) and Semily (Krkonoše Piedmont Basin) formations, a very similar association is present with ‘Lissodus’ and Progyrolepis speciosus (Štamberg and Zajíc 2008; Štamberg 2023). In the Puertollano Basin (Stephanian C, Spain), ‘Lissodus’ occurs alongside P. speciosus (Soler-Gijón 1994, 1997), but this occurrence of Progyrolepis is doubtful and warrants verification (see above). With the exception of these occurrences, the other basins where Progyrolepis is documented have not yielded any specimens of ‘Lissodus’ (Fig. 11). Specimens attributed to ‘Lissodus’ have also been reported from the Stephanian C-Autunian of the basins of Lodève (French Massif Central), Saale (Harz Massif), Saar-Nahe (Rhenish Massif), Thuringian Forest (Bohemian Massif) et Guardia Pisano (Sardinia) (Fig. 11; e.g., Fischer 2008).

A comparison of the palaeoichthyological assemblage from the Cantemerle locality with that from all the Stephano-Permian basins in Europe indicates a strong synchronous aquatic connection between the Réalmont Basin and the Bourbon-l’Archambault Basin in the early Autunian. This is further evidence of the active connection between the northern and southern basins of the French Massif Central, already suggested by the ichthyofauna of the Decazeville Basin (Gonçalves et al. 2025; Gonçalves in press). The similarity between the ichthyofauna of Réalmont and Bourbon-l’Archambault basins and that of the Bohemian basins is notable and indicates a direct connection between the French Massif Central and the Bohemian Massif between the Stephanian C and the early Autunian. These two palaeobiogeographic areas could have been more actively connected than with other basins (e.g., Rhenish Massif basins), a pattern also suggested by the chondrichthyan (Luccisano et al. 2022) and aeduellid (Gonçalves et al. 2026c) fossil record. We can explain this dispersal by continental migrations only, because none of these basins are considered to be connected by marine routes at this period (e.g., Fischer et al. 2013; Luccisano et al. 2023; Rossignol et al. 2025)

Most families of late-Variscan actinopterygians (Amblypteridae, Igornichthyidae, Elonichthyidae) show strong endemism at the species-level (e.g., Heyler 1969; Dietze 2000; Štamberg 2013, 2022; Schindler 2018a, 2018b; Gonçalves et al. 2026b, 2026c). Only a few taxa such as Aeduella blainvillei and Westollia crassa show a wide distribution on the Variscan scale (Gonçalves in press). Endemism is also observed in the small-sized xenacanthiform Triodus (e.g., Luccisano et al. 2021). Progyrolepis does not seem to follow an endemism pattern, as its two species have wide distributions across the French Massif Central and the Bohemian Massif. However, it should be noted that the genus is provincialised, with P. heyleri in the French Massif Central and P. speciosus in the Bohemian Massif. The pronounced endemism of most late-Variscan aquatic vertebrates may be linked to their small size, implying a high evolution rate (e.g., Gonçalves et al. 2026c). The large size of Progyrolepis would imply a lower evolution rate, but the landscape dynamics of the Variscan range probably contributed to the provincialisation of its species.