1. Introduction
The vertebrate faunas of the Early Carboniferous are fundamentally different from their late Devonian predecessors. This difference is largely attributed to the Hangenberg extinction, the second of two major pulses of extinction in the late Devonian (Sallan & Coates Reference Sallan and Coates2010). The Hangenberg extinction is generally thought to have restructured vertebrate diversity in ways critical to the emergence of modern vertebrate groups by substantially reducing or eliminating diversity of jawless fishes, ‘placoderms’, ‘acanthodians’, and non-tetrapod sarcopterygians, thus creating ecological opportunity for adaptive radiations of chondrichthyans, actinopterygians, and tetrapods (Sallan & Coates Reference Sallan and Coates2010; Friedman & Sallan Reference Friedman and Sallan2012). However, new findings from the latest Devonian and Early Carboniferous show that this narrative has been oversimplified; at least early stages of key diversifications of chondrichthyans (Frey et al. Reference Frey, Coates, Ginter, Hairapetian, Rücklin, Jerjen and Klug2019), actinopterygians (Wilson et al. Reference Wilson, Pardo and Anderson2018; Giles et al. Reference Giles, Feilich, Warnock, Pierce and Friedman2023), tetrapods (Ahlberg & Clack Reference Ahlberg and Clack2020), and post-Devonian lungfishes (Pardo et al. Reference Pardo, Huttenlocker and Small2014; Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019; Clack et al. Reference Clack, Challands, Smithson and Smithson2019) appear to have occurred cryptically prior to the Devonian–Carboniferous transition. Further, increased sampling of Tournaisian-aged localities has shown that survivorship of archaic Devonian forms was more widespread than previously thought, including moythomasiid-grade actinopterygians (Wilson et al. Reference Wilson, Pardo and Anderson2018) and Acanthostega-like stem tetrapods (Anderson et al. Reference Anderson, Smithson, Mansky, Meyer and Clack2015) in the Horton Group of Blue Beach, Nova Scotia, Canada.
However, this higher survivorship generally appears to be clustered in lineages more proximate to crown groups, raising the possibility that trait-based selection across the Hangenberg may have played a critical role in shaping the emergence of modern vertebrate faunas. The strength of these inferences depends on the quality of sampling in the latest Devonian and earliest Carboniferous. In particular, the quality of the Early Carboniferous vertebrate record is comparatively poor, referred to broadly as ‘Romer's Gap’ (Smithson et al. Reference Smithson, Wood, Marshall and Clack2012; Mansky & Lucas Reference Mansky and Lucas2013), which has the potential to obscure the true biological signal of the extinction and subsequent recovery. The specific case of lungfishes is instructive. Lungfishes achieve high diversity by the end of the Middle Devonian, at which point they appear to dominate marine and estuarine durophage niches, as evidenced by their range of blunt dentitions (Ahlberg et al. Reference Ahlberg, Smith and Johanson2006) and lungfish coprolites containing hard-bodied remains (Sallan et al. Reference Sallan, Kammer, Ausich and Cook2011). In contrast, Carboniferous and Permian lungfishes appear to be derived from a few closely related Devonian ‘phaneropleuriform’ lineages primarily associated with freshwater and deltaic environments and relatively closely related to the lungfish crown group (Pardo et al. Reference Pardo, Huttenlocker and Small2014; Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019; Clack et al. Reference Clack, Challands, Smithson and Smithson2019). A few divergent and potentially phylogenetically disparate forms do persist into the Carboniferous, notably the blade-toothed Uronemus (Smith et al. Reference Smith, Smithson and Campbell1987) and the denticulate Conchopoma (Marshall Reference Marshall1988), but in general it is thought that the lungfish of the Early Carboniferous recovery fauna were far more similar to the six species of modern lungfishes than to the diverse lungfish faunas that immediately preceded them (Lloyd et al. Reference Lloyd, Wang and Brusatte2012; Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019; Clack et al. Reference Clack, Challands, Smithson and Smithson2019). However, we know relatively little about the lungfish faunas of the earliest Carboniferous aside from a few recently described forms from tetrapod-bearing freshwater deposits of the Ballagan Formation of the Scottish border region (Smithson et al. Reference Smithson, Richards and Clack2016; Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019). More extensive sampling from outside this area, and outside this particular depositional environment, are necessary to determine whether these patterns reflect the global signal of the Hangenberg extinction or regional patterns of faunal turnover in the first 15 million years of the Carboniferous.
Considering this, we report a partial lungfish occiput from the Tournaisian (Horton Bluff Formation, Middle Tournaisian) of Blue Beach, Nova Scotia, Canada. Although the specimen is incomplete and is not directly associable with dental material, it preserves several characteristics which preclude a placement within classic ‘phaneropleuriform’ groups and instead within anatomically diverse grades of marine durophagous lungfishes from the Middle to Late Devonian.
2. Geological context
The fossil reported here was collected from the vertebrate-bearing exposures of the western shore of the Avon River, Nova Scotia, in an area locally known as Blue Beach, and is accessioned at the Nova Scotia Museum (NSM). Deposits at Blue Beach comprise thinly bedded black silty shales up to 10–20 cm thick well-indurated sandstones, most concentrated in a cove located below a lighthouse, which represent a series of water-level changes between more offshore and onshore facies, including mudcracks, trackways, and remnants of lycopod forests (Anderson et al. Reference Anderson, Smithson, Mansky, Meyer and Clack2015). The vertebrate community found there is dominated by actinopterygians (Wilson et al. Reference Wilson, Pardo and Anderson2018; Wilson et al. Reference Wilson, Mansky and Anderson2021), ‘acanthodians’, chondrichthyans, and rhizodontids (Brazeau Reference Brazeau2005), as well as rarer tetrapod fossils (Anderson et al. Reference Anderson, Smithson, Mansky, Meyer and Clack2015). Due to extreme daily tidal fluctuations within the Bay of Fundy, fossils are collected ex situ from the shoreline during low tide intervals, but position in sequence can be roughly estimated by the location along the shoreline. Most vertebrate fossils at Blue Beach are black in colour and appear to have come from the Tournaisian (late Tn2 to earliest Tn3-aged) Blue Beach or Hurd Creek Member of the Horton Group (Utting Reference Utting, Keppie and Giles1989; Martel & Gibling Reference Martel, Gibling, P., L. and K.1991), where they are often preserved within storm-generated sandstone units. Preservation of NSM017GF017.002 is consistent with this preservation mode and horizon.
The broader palaeoenvironmental context of the Horton Bluff Formation of the Horton Group has been debated. The Horton Group represents a long-term subsidence and fill of a continental lacustrine basin spanning late Devonian (Famennian) through latest Tournaisian times. There has been some debate as to the salinity content of the lacustrine system represented by the Blue Beach Member (Martel & Gibling Reference Martel, Gibling, P., L. and K.1991; Tibert & Scott Reference Tibert and Scott1999; Tang et al. Reference Tang, Pe-Piper, Piper, Chen, Hou, Guo and Zhang2024), but ostracods and agglutinated foraminiferans (Tibert & Scott Reference Tibert and Scott1999), marine-influenced invertebrate macrofossils (Lerner et al. Reference Lerner, Lucas and Mansky2016), and heavy isotopes (Tang et al. Reference Tang, Pe-Piper, Piper, Chen, Hou, Guo and Zhang2024) suggest that this system was at least intermittently marine-influenced from the base of the Blue Beach Member through the early part of the overlying Hurd Creek Member. Vertebrate fossils within the Horton Bluff Formation are typically disarticulated and modified by pre-deposition transport (Mansky & Lucas Reference Mansky and Lucas2013; Anderson et al. Reference Anderson, Smithson, Mansky, Meyer and Clack2015) so it is difficult to determine whether individual taxa were euryhaline or occupied the depositional system preferentially under freshwater or saline conditions.
3. Materials and methods
The specimen was manually prepared under a binocular dissecting scope using a PaleoTools Micro Jack 4 air scribe and was consolidated with polyvinyl acetate (PVA). Complete removal from matrix was accomplished by embedding the exposed side in carbowax and preparing the other surface in the same manner. Following preparation, we used microtomography (Anderson Lab, McCaig Institute for Bone and Joint Health, University of Calgary) to image the occipital unit in 3D. Scans were produced at a voxel size of 55.383 μm by scanning on the Skyscan1173 (Bruker Inc.) cone beam scanner at 130 kV and 61 uA, using a 0.25 mm brass filter. Scans were reconstructed as a tomographic stack using the software nRecon v.1.6.9.
The resulting tomographic stack was imported into Amira 5.3.3 (Visage, Inc.) for subsequent segmentation and visualisation. Voxels were assigned to elements using the LabelField module, resampled, and 3D surfaces were produced using the SurfaceGen module.
4. Results
NSM017GF017.002 is a partial occiput spanning from the posteriormost extent of the otic capsule to the occipital surface (Fig. 1). It is diagnosable as lungfish (and distinguishable from other osteichthyan lineages) by the lack of bony separation between the foramen magnum and notochordal canal, presence of conspicuous dorsolateral cristae suspending the neurocranium below the skull roof, and presence of separate foramina for two spinal nerves posterior to the foramen serving the vagus nerve. Although this occiput superficially resembles the occipital region of some Devonian tetrapods (Clack Reference Clack1998; Clack et al. Reference Clack, Ahlberg, Finney, Dominguez Alonso, Robinson and Ketcham2003), the presence of two successive cranial nerves (probably spinal nerves 1 and 2) passing through the occiput posterior to the vagus nerve precludes a tetrapodomorph affinity.
Anatomy of NSM017GF017.002, rendered from μCT: (a) occipital unit in anterior view; (b) endocast in anterior view; (c) occiput in posterior view; (d) endocast in posterior view; (e) occiput in dorsal view; (f) endocast in dorsal view; (g) occiput in ventral view; (h) endocast in ventral view; (i) occiput in left lateral view; (j) endocast in left lateral view; (k) occiput in right lateral view; (l) endocast in left lateral view. Scale = 1 cm. Abbreviations: bv = passage of blood vessel; crdl = crista dorsolateralis; fm = foramen magnum; f.occ.l = lateral occipital fissure; nr = notochordal recess; n.X, n.sp1 = foramen serving the first spinal nerve; n.sp2 = foramen serving the second spinal nerve; foramen serving the vagus nerve; pscc = posterior semicircular canal; psph, parasphenoid.
The occiput is relatively short and high, slightly rhomboidal in lateral view and Y-shaped in occipital view (Fig. 1c, d). The element encloses a large midline cavity that encompasses both the foramen magnum dorsally and the notochordal canal ventrally. A weakly developed bony constriction partially separates these two spaces (Fig. 1a–d). The notochordal canal curves ventrally as it progresses anteriorly. Only dorsolateral cristae are present, and the dorsal surface of the occiput is relatively well-preserved, excluding the possibility that the median crista was present in life but lost post-mortem. Both cristae are broken, leaving their dorsal and posterior extent uncertain, although the left crista does appear to curve posterodorsally, similar to the condition seen in Holodipterus (Miles Reference Miles1977). Dorsolateral cristae are oriented approximately 25 ° from the dorsal plane, leaving a broad U-shaped post-temporal fossa on the dorsal surface of the neurocranium. Each dorsolateral crista is pierced by several canals. The most prominent of these is a curved large-bore canal enclosed within the anterior face of the left dorsolateral crista (Fig. 1b, d); the shape and location of this canal is consistent with the location of the posterior semicircular canal in other lungfishes (Clement et al. Reference Clement, Challands, Cloutier, Houle, Ahlberg, Collin and Long2022). The radius of curvature is relatively small for the portion of the semicircular canal preserved, suggesting either a reduced size of the semicircular canals in general or a more erratic curvature of the posterior semicircular canal in this species. A short, narrow-calibre canal appears to be the same structure as the nutrient foramen identified in a similar position in Soederberghia by Friedman (Reference Friedman2007) and probably represents a small artery. Traces of the passage for the endolymphatic ducts are not identifiable in the endocast.
Ventral to the dorsolateral crista, the lateral wall of the occiput is lined with smooth perichondral bone. This surface of the occiput is otherwise unremarkable, with the exception of a long, shallow vertical fossa just posterior to the base of the dorsolateral crista and three foramina approximately equally spaced anteroposteriorly. The anterior foramen is located at the anterior margin of the element as preserved and probably represents the course of the vagus nerve (Fig. 1f, j, l), whereas the posterior two foramina pierce the otherwise solid lateral wall of the occipital arch and therefore probably served spinal nerves 1 and 2. The foramen for the presumed first spinal nerve is situated dorsally within the vertical fossa (Fig. 1i, k).
The ventral surface of the occiput is narrow. Anteriorly, there is a short robust triangular shelf that may represent the posterior stalk of the parasphenoid (Fig. 1g). If this is the case, the posterior stalk does not reach the occipital surface. A short triangular stalk of the parasphenoid co-ossified to the base of the neurocranium is seen in ‘chirodipterids’ such as Pillararhynchus (Long Reference Long1992). No cranial rib articulations are visible adjacent to the parasphenoid stalk.
5. Affinities of NSM017GF017.002
NSM017GF017.002 is too incomplete to incorporate into a robust phylogenetic analysis. It is identifiable as a lungfish on the basis of incomplete ossification between the foramen magnum and notochordal recess and the well-developed dorsolateral cristae. This makes NSM017GF017.002 the first definitive record of a lungfish in the Horton Group and the first record of a lungfish from the Tournaisian of North America.
Resolution within Dipnoi is more ambiguous (Fig. 2). The apparent absence of a median crista is characteristic of holodontids, a diverse clade of lungfishes that includes brevirostrine forms like Holodipterus and longirostrine forms such as Griphognathus, Soederberghia, and Rhynchodipterus. This clade is also sometimes characterised by the loss of a complete bony shelf separating the foramen magnum and notochordal canal, also observed in NSM017GF017.002. This may be due to damage or overpreparation in some holodipterids, however; it is unclear whether this is also the case in NSM017GF017.002 as this specimen shows some evidence of pre-depositional damage. More importantly, the basioccipital region of holodontids is distinct from the condition seen in NSM017GF017.002; the ventral surface of the occiput is typically broad and flat and the posterior stalk of the parasphenoid typically extends to the occipital surface. In contrast, the parasphenoid of NSM017GF017.002 probably does not reach the occipital surface, and the ventral surface of the occipital region is narrow and keel-like. This resembles more closely the occipital regions of more stemward lungfishes such as Pillararhynchus longi, which typically resolves within the grade of ‘dipterid’ lungfishes along with forms such as Dipterus valenciennesi and Sorbitorhynchus deleaskitus. However, the occiput of NSM017GF017.002 is considerably taller and narrower than these taxa, although this may be a consequence of preservation.
Lungfish survivorship of the Hangenberg extinction. Phylogeny based on Challands et al. (Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019). Dashed lines indicate possible affinities of NSM017GF017.002 with either dipterids or holodontids.
Regardless, the anatomy observable here clearly distinguishes NSM017GF017.002 from known groups of lungfishes with Carboniferous occurrences (Fig. 2). Most Carboniferous lungfishes have traditionally been assigned to the Phaneropleuriformes, a group which encompasses a range of poorly ossified Devonian forms such as Scaumenacia, as well as typical Carboniferous taxa such as Ctenodus, Sagenodus, and Tranodis (Clack et al. Reference Clack, Challands, Smithson and Smithson2019). The neurocranium of these forms appears to be completely unossified, possibly as a consequence of paedomorphosis in the origin of this clade (Boirot et al. Reference Boirot, Challands and Cloutier2022), and this group typically exhibits a long paddle-like posterior stalk of the parasphenoid (Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019). In contrast, occipital ossifications are present in gnathorhizids, a group which appears to be closely allied with extant lungfishes, although those ossifications are paired and the notochordal canal is not floored by the neurocranium but rather a cylindrical posterior stalk of the parasphenoid (Pardo et al. Reference Pardo, Huttenlocker and Small2014). The condition of the occipital region in Carboniferous problematica such as Conchopoma and uronemids is not known but is unlikely to be broadly comparable with NSM017GF017.002 based on the anatomy of the parasphenoid where known (Denison Reference Denison1969; Elliott et al. Reference Elliott, Challands and Smithson2023).
We therefore conclude that it is most likely that NSM017GF017.002 belongs to a dipterid-grade lungfish, possibly of a ‘chirodipterid’ or holodontid, with the caveat that the narrow and deep morphology of NSM017GF017.002 would be unique among this grade of lungfishes. Allowing for some uncertainty of the interrelationships of Devonian lungfishes (Ahlberg et al. Reference Ahlberg, Smith and Johanson2006) and uncertainty of the relationships of Conchopoma and uronemids (Smith et al. Reference Smith, Smithson and Campbell1987; Marshall Reference Marshall1988), this probably represents the most stemward lungfish known from the Carboniferous (following the phylogenetic framework of Ahlberg et al. Reference Ahlberg, Smith and Johanson2006; Clack et al. Reference Clack, Challands, Smithson and Smithson2019, and Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019) and the first attributable to primarily marine lungfish groups.
6. Discussion
We identify NSM017GF017.002 as the first Tournaisian lungfish from North America. This diagnosis is based on the absence of a well-developed bony separation between the foramen magnum and notochordal canal and passages for two post-vagal spinal nerves passing through the occipital arch. We are tentatively able to diagnose this specimen as either a holodontid lungfish, based on the absence of a median crista of the braincase and dorsolateral cristae extending posterior to the occiput and presence of an ossified floor of the notochordal canal (characteristic of holodontids), or a dipterid-grade lungfish, on the basis of the narrow occipital base and short triangular parasphenoid stalk not extending to the occipital surface (characteristic of dipterid-grade lungfishes). Among known holodontids, general shape of the occipital region and simple linear organisation of the spinal nerve foramina are most consistent with the morphology of the tooth-plate-bearing Holodipterus rather than denticulate forms such as Griphognathus and Soederberghia, although the short parasphenoid stalk and narrow occiput is less consistent with that taxon and holodontids in general. However, we do note that the occipital region is generally poorly known even in well-preserved lungfishes, including many holodontids and dipterid-grade lungfishes, and we are therefore hesitant to attribute this specimen to lower taxa. Furthermore, the absence of reported tooth plates in the Horton Group limits comparison with both Devonian and Early Carboniferous lungfishes, many of which are known solely from isolated tooth plates (Smithson et al. Reference Smithson, Richards and Clack2016).
This record is the youngest known record of a possible dipterid-grade and presumably holodontid lungfish, and the first definitive Carboniferous occurrence of this taxonomic grade. As holodontids and dipterid-like forms are relatively distantly related to phaneropleuriforms and crown lungfishes, this indicates that lungfish survivorship of the Hangenberg event was substantially more inclusive than previously appreciated, and probably included at least some of the predominantly marine lineages (ostensibly durophagous reef specialists) which made up a large part of Devonian lungfish diversity. A single occurrence is not enough to determine whether NSM017GF017.002 represents a single zombie taxon or the first indication of a diverse marine lungfish fauna persisting into the Tournaisian alongside well-documented survivorship of freshwater lineages (Pardo et al. Reference Pardo, Huttenlocker and Small2014; Smithson et al. Reference Smithson, Richards and Clack2016; Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019; Clack et al. Reference Clack, Challands, Smithson and Smithson2019). However, we do note that the vertebrate-bearing horizons of the Horton Bluff Formation are thought to be middle to upper Tournaisian (late Tn2 to earliest Tn3; Utting et al. Reference Utting, Keppie and Giles1989) and thus do not represent an immediate post-Hangenberg recovery fauna (unlike earliest Tournaisian placoderm-bearing horizons in Greenland; Blom et al. Reference Blom, Clack, Ahlberg and Friedman2007), favouring the latter hypothesis. If so, this would suggest that the ultimate extinction of durophagous marine lungfishes within the Early Carboniferous may have been a consequence of other biotic factors, such as competitive replacement by either actinopterygians or holocephalans, both of which are highly diverse in the Tournaisian (Richards et al. Reference Richards, Sherwin, Smithson, Bennion, Davies, Marshall and Clack2018).
In this context it is possible that the strikingly different lungfish faunas of the Ballagan Formation and Blue Beach Member might reflect the respective ages or biogeographical provinces of these sediments. The Ballagan Formation appears to span the Tournaisian, based on new interpretations of palynomorph assemblages (Marshall et al. Reference Marshall, Reeves, Bennett, Davies, Kearsey, Millward, Smithson and Browne2018), whereas the Blue Beach Member appears to be approximately mid-Tournaisian, based on a palynofloral shift recorded in the underlying Curry Brook Member (Utting, Reference Utting, Keppie and Giles1989) and a shift in chemical weathering indices associated with the Tournaisian carbon isotope excursion in the middle of the Blue Beach Member (Tang et al. Reference Tang, Pe-Piper, Piper, Chen, Hou, Guo and Zhang2024). Despite this, the Blue Beach fauna generally shows greater affinities with late Devonian vertebrate communities of North America (Anderson et al. Reference Anderson, Smithson, Mansky, Meyer and Clack2015; Wilson et al. Reference Wilson, Pardo and Anderson2018) whereas the Ballagan Formation shows closer affinities with younger Visean and Serpukhovian faunas from the British Isles (Clack et al. Reference Clack, Bennett, Carpenter, Davies, Fraser, Kearsey, Marshall, Millward, Otoo, Reeves, Ross, Ruta, Smithson, Smithson and Walsh2016). Thus, it is possible that this difference in the lungfishes might simply be a consequence of diachronous faunal turnover in the Tournaisian as the post-Hangenberg recovery fauna was replaced by classic Carboniferous communities. However, it is equally possible that some of these differences are a consequence of biogeographical barriers, such as the Acadian orogeny, that would have been present in the Early Carboniferous.
A confounding factor which might contribute to the presence of a holodontid or dipterid-grade lungfish in the Blue Beach Member but absence in the Ballagan Formation is the overall difference in depositional environment between these formations. Devonian lungfish fossils are broadly found in both freshwater and marine deposits, with specialised durophages such as holodontids found predominantly in backreef deposits of the Gogo Formation (Long & Trinajstic Reference Long and Trinajstic2017), whereas Devonian and Carboniferous phaneropleuriforms are predominantly known from freshwater or estuarine deposits (Cloutier et al. Reference Cloutier, Proust and Tessier2011). This is not a strict division; more stemward forms such as Soederberghia are predominantly found in freshwater environments (Ahlberg et al. Reference Ahlberg, Johanson and Daeschler2001), but freshwater and marginal marine tetrapod-bearing localities do exhibit distinct lungfish faunas at least within the late Devonian (Friedman & Daeschler Reference Friedman and Daeschler2006), a trend that would be expected to persist into the Carboniferous. As the Blue Beach Member has some marine influence (Tang et al. Reference Tang, Pe-Piper, Piper, Chen, Hou, Guo and Zhang2024) in comparison with the freshwater Ballagan Formation (Clack et al. Reference Clack, Bennett, Carpenter, Davies, Fraser, Kearsey, Marshall, Millward, Otoo, Reeves, Ross, Ruta, Smithson, Smithson and Walsh2016, Reference Clack, Challands, Smithson and Smithson2019; Kearsey et al. Reference Kearsey, Bennett, Millward, Davies, Gowing, Kemp, Leng, Marshall and Browne2016; Bennett et al. Reference Bennett, Kearsey, Davies, Leng, Millward, Smithson, Brand, Browne, Carpenter, Marshall, Dulson and Curry2021), the occurrence of a holodontid-like form in the former and phaneropleuriforms in the latter could simply reflect conserved paleoecology of lungfish higher taxa across the Hangenberg event. Dense locality-scale sampling has shown that apparent faunal turnovers in later parts of the Carboniferous were consequences of changing sampling of depositional environments rather than global faunal turnovers (Pardo et al. Reference Pardo, Small, Milner and Huttenlocker2019; Dunne et al. Reference Dunne, Thompson, Butler, Rosindell and Close2023), so we caution against overinterpreting faunal turnover in the Tournaisian based on only a handful of newly sampled vertebrate-bearing localities. Further, the lungfish represented by NSM017GF017.002 appears to have been a very rare member of the Blue Beach fauna in contrast with the high abundance of other vertebrate groups at the locality. This rarity could reflect the environmental context of Blue Beach, reduced overall abundance of marine lungfishes following the Hangenberg mass extinction, or both. At present, the low abundance and diversity of marine lungfishes crossing the Devonian–Carboniferous boundary suggest that the apparent incumbents of the marine durophage trophic guild may have been effectively but not completely eliminated across the Hangenberg event, facilitating the diversification of holocephalans in the Mississippian (Richards et al. Reference Richards, Sherwin, Smithson, Bennion, Davies, Marshall and Clack2018).
Regardless, this new specimen adds to a growing understanding that dipnoan diversity and disparity were high in the Tournaisian in comparison with later intervals of the upper Carboniferous and Early Permian (Challands et al. Reference Challands, Smithson, Clack, Bennett, Marshall, Wallace-Johnson and Hill2019; Smithson et al. Reference Smithson, Richards and Clack2016). This also underscores the importance of sustained lungfish extinctions during the Carboniferous, whether these were a consequence of reduced evolvability of lungfishes (Lloyd et al. Reference Lloyd, Wang and Brusatte2012) or competitive replacement by other vertebrate lineages. We hope further efforts to characterise Romer's Gap, such as Tim Smithson's critical efforts in Scotland, will help clarify these dynamics.
Acknowledgements
We thank the late Sonja Wood for her tireless efforts in supporting research on the Blue Beach locality, and Tim Fedak and Katherine Ogden of the Nova Scotia Museum for ongoing support of our study of the vertebrates of Blue Beach. Work was supported by NSERC Discovery Grants 2017-04821 and 2023-04423 to JSA and a Queen Elizabeth II graduate scholarship to CDW. We would also like to thank Marcello Ruta, Tom Challands, and Stig Walsh for inviting us to share this work to commemorate the career of our friend and colleague Tim Smithson, whose contributions to Scottish Tournaisian lungfish palaeontology were an inspiration for this paper.
