1. Introduction
The coelacanths, or actinistians, form a clade that split from other sarcopterygians in Early Devonian time, i.e. more than 400 Ma ago. However, the group has never been diversified and comprises today a single genus with two species. The highest, although moderate, peak of diversity was reached in Early Triassic time (Forey, Reference Forey1998; Schultze, Reference Schultze, Arratia and Tintori2004; Cavin, Furrer & Obrist, Reference Cavin, Furrer and Obrist2013; Romano et al. Reference Romano, Koot, Kogan, Brayard, Minikh, Brinkmann, Bucher and Kriwet2016), and it has been hypothesized that these fishes benefited from the decrease in diversity of the ray-finned fishes after the Permian–Triassic mass extinction to thrive a little (Wen et al. Reference Wen, Zhang, Hu, Benton, Zhou, Tao, Huang and Chen2013). During most of the history of the group, coelacanths had representatives in both marine and freshwater environments. Most of the Early Triassic record is from marine environments, but the proportion changes towards freshwater environments in Late Triassic time (Forey, Reference Forey1998). From the Late Triassic period onwards, only two families are present in the fossil record, the mawsoniids and the latimeriids (plus the stem latimeroid Coccoderma in Late Jurassic time) until Late Cretaceous time (Forey, Reference Forey1998). Here, we describe fragmentary remains of coelacanths from marine deposits of Late Triassic age in southeastern France that can be referred with some confidence to the mawsoniids. The Rhaetian deposits of the Provence region have already yielded isolated fish remains, mostly teeth and ganoid scales referable to hybodont sharks, various ‘ganoid’ ray-finned fishes (Corroy, Reference Corroy1934) and marine reptiles (Bardet & Cuny, Reference Bardet, Cuny, Mazin and Pinna1993; Bardet et al. Reference Bardet, Buffetaut, Giner and Roggero2013; Fischer et al. Reference Fischer, Cappetta, Vincent, Garcia, Goolaerts, Martin, Roggero and Valentin2014). But, as far as we are aware, the present contribution describes actinistian remains for the first time.
2. Geographical and geological setting
The Peygros quarry is located in the Var Department in Le Thoronet municipality (Fig. 1). The quarry, located in a hill called Darboussière, was exploited over several years for its bauxite deposits by the Peychiney and SABAP companies (now affiliated with the Rio Tinto quarry consortium).
The Cabasse – Le Thoronet syncline is composed of Jurassic and Triassic limestone, marl and claystone beds. The unit corresponds to a thrust sheet with sediments ranging from the Keuper to the Hettangian above an autochthonous Triassic to Bathonian unit with Cretaceous bauxite locally. Several faults cause reverse contacts between Triassic and Jurassic outcrops, and Cretaceous bauxite deposits. The outcrops present an alternation of dark grey calcareous marl and grey argillaceous beds. The cliff is unstable and several blocks collapse because of gravitational crumbling exacerbated by atmospheric perturbations and cryoclasty during cold winters. The fossils were found in the talus cone under the cliff.
The presently described specimen was discovered during the course of a field survey of the Peygros quarry in September 2015.
2.a. Material
The fossil described here is housed in the Departmental Natural History Museum of Var, SE France under the collection number PEY-2015.5.1 (PEY-2015.5.1.1 for the skull and PEY-2015.5.1.2 for the basisphenoid). The specimen was embedded in a block of coarse-grained argillaceous limestone. The only preserved bones are a series of ceratobranchials, which articulated with basibranchial supporting tooth plates, as well as the left lower jaw with both gular plates. A basisphenoid was found close to the rest of the bones and can be referred with confidence to the same specimen. The main specimen was prepared mechanically with a steel needle and final preparation was performed with diluted (c. 10%) acetic acid.
3. Systematic palaeontology
ACTINISTIA Cope, Reference Cope1871
LATIMERIOIDEI Schultze, Reference Schultze and Benton1993
Mawsoniidae Schultze, Reference Schultze and Benton1993
gen. et sp. indeterminate
3.a. Lower jaw and gular plates
The preserved part of the left lower jaw comprises the angular, prearticular and principal coronoid (Fig. 2a, d). Pieces of the dentary or the splenial are present but poorly preserved. The general shape of the lower jaw is proportionally elongate and shallow. The prearticular, located on the lingual side of the mandible is elongate and shallow (Fig. 2a). It occupies nearly the entire length of the lower jaw. Its lingual face is fully covered with tiny teeth that are regular in shape and size. Each tooth is rounded and slightly flattened and marked with fine striations radiating from the apex. The posterior part of the bone is poorly preserved, but a narrowing is present. The principal coronoid is large, elongated and poorly preserved (Fig. 2a, d). It wedges between the angular laterally and the prearticular medially at the level of the maximum depth of the angular. The angular (Fig. 2d) was likely long and shallow, but its anterior part is damaged. The dorsal margin forms a concavity at the level of the principal coronoid while the ventral margin is approximately straight. Its lateral surface is ornamented with faint grooves in its dorsal part, strong pits and deep grooves above the course of the sensory canal, and is almost smooth in the ventral part. The mandibular sensory canal is marked with at least four large oval openings located under a ridge situated beneath the ornamented part. The posterior opening of the mandibular canal, which connected to the preopercular sensory canal, is visible at the posterodorsal corner of the bone. Two gular plates are preserved, but their shape cannot be reconstructed. The outer surface of each gular is smooth.
3.b. Basisphenoid
The basisphenoid is incomplete (Fig. 2c). The processus connectens are elongated but they apparently do not to reach the parasphenoid ventrally. The sphenoid condyles are large, close to each other and protrude dorsally. The antotic processes are not preserved and the dorsum sella appears to be short and narrow. Fragments of the posterior extremity of the parasphenoid are still attached to the ventral margin of the basisphenoid.
3.c. Branchial apparatus and tooth plates
The anteroventral portion of the branchial arches is preserved (Fig. 2a, b). The anterior extremity of the left ceratohyal is visible dorsally, wedged between the basibranchials and the left lower jaw, and its posterior part is visible in lateral view. There are five arches represented by ceratobranchials, which are large, stout and curved elements. Each ceratobranchial is excavated by a deep groove ventrally for the afferent branchial artery (Fig. 2b) and is covered dorsally by tooth plates arranged along two parallel series (Fig. 2a). Each tooth plate is quadrangular and slightly curved, thus forming a zig-zag pattern along a series. Each bears tiny villiform teeth which show ridges radiating from the centre of the crown, as present on the prearticular (Fig. 2a′, close-up). The basibranchial is large and roughly trapezoidal in shape and supports the first four pairs of branchial arches. The fifth pair of ceratobranchials is located more posteriorly. The basibranchial is covered with tooth plates, which are much larger than those on the ceratobranchials. The basibranchial teeth vary in size, with the largest ones located along the medial margins of the large posterior tooth plates. Although they are not perfectly preserved, we recognized a pair of large and elongated tooth plates located posteriorly preceded by an indeterminate number of smaller tooth plates anteriorly, laterally and possibly posteriorly. The anterior extremity of the urohyal with its visible articulated head, which originally articulated with the posterior tip of the basibranchial, is preserved in between the anterior extremities of the fifth pair of ceratobranchials.
4. Discussion and conclusion
Among Osteichthyes, the coelacanths show autapomorphic features in the basisphenoid and in the lower jaw, such as the presence of processus connectens in the former and of an angular with a typical shape in the latter (Forey, Reference Forey1998). These features occur in PEY-2015.5.1, which leaves no doubt about the attribution of the Peygros remains to this clade. On the other hand, these osteological elements as well as the rest of the anatomy of the coelacanths are fairly uniform among most species of the clade, making a precise identification difficult. A few features, however, allow some systematic precisions. The presence of teeth with radiating ridges is a character present in derived mawsoniids (Mawsonia and Axelrodichthys), but also in the latimeriid Libys, and in some more basal coelacanths (Axelia, Spermatodus; Forey, Reference Forey1998.) The ornamentation of the angular (Fig. 3, top), consisting of grooves and ridges rather than tubercles, and the lateral surface with a few slit-like ventral openings of the sensory canal under a ridge, is reminiscent of the situation in most mawsoniids (Forey, Reference Forey1998; Cavin, Valentin & Garcia, Reference Cavin, Valentin and Garcia2016). In the North American Triassic mawsoniid Diplurus, the angular is smooth (Schaeffer, Reference Schaeffer1952), but it appears to be ornamented with a similar pattern in another Triassic mawsoniid from North America, Chinlea sorenseni (Schaeffer, Reference Schaeffer1967, plate 28, fig. 2). In latimeriids and in most other non-mawsoniid species without a granular ornamentation, the angular is smooth (Forey, Reference Forey1998). The basisphenoid PEY-2015.5.1.2 possesses several features pointed out by Dutel, Herbin & Clément (Reference Dutel, Herbin and Clément2015) as diagnostic of mawsoniids, such as a narrow dorsum sella and close sphenoid condyles separated by a marked notch, a pattern which differs in the latimeriids and other non-mawsoniid species (Fig. 3, bottom). In particular, the short dorsum sella and the well-defined sphenoid condyles separated by a small depression are characters present in the basal mawsoniids Chinlea and Diplurus (Schaeffer, Reference Schaeffer1967), and in the derived mawsoniids Mawsonia and Axelrodichthys (Maisey, Reference Maisey1986; Carvalho & Maisey, Reference Carvalho, Maisey, Cavin, Longbottom and Richter2008). In non-mawsoniid coelacanths, for instance Latimeria, Megalocoelacanthus or Dobrogeria, when visible the sphenoid condyles are less marked and separated by a wider gap (Forey, Reference Forey1998; Dutel et al. Reference Dutel, Maisey, Schwimmer, Janvier, Herbin and Clément2012; Cavin & Grădinaru, Reference Cavin and Grădinaru2014). Recently, the Early Jurassic Trachymetopon liassicum was included in the mawsoniids by Dutel, Herbin & Clément (Reference Dutel, Herbin and Clément2015). This genus, originally described from the Lower Toarcian marine locality of Holzmaden (Hennig, Reference Hennig1951) was subsequently recorded in the Upper Callovian of the Vaches Noires in Normandy (Dutel, Pennetier & Pennetier, Reference Dutel, Pennetier and Pennetier2014). As far as it can be assessed from the published data, the angular and the basisphenoid of Trachymetopon are reminiscent of the specimen from Peygros. However, given the fragmentary state of our specimen and pending a more precise comparison, we prefer not to include PEY-2015.5.1 in the genus Trachymetopon, but to regard it as an indeterminate mawsoniid.
Mawsoniids from Upper Triassic and Lower Jurassic strata are known exclusively from freshwater environments in North America (Chinlea and Diplurus; Schaeffer, Reference Schaeffer1967). During Cretaceous time, representatives of the family occurred mostly in continental deposits of Western Gondwana with Mawsonia and Axelrodichthys (South America and Africa; Maisey, Reference Maisey2000). The last mawsoniid genus is Parnaibaia, from the Upper Jurassic or Lower Cretaceous continental deposits of the Parnaiba Basin in Brazil (Yabumoto, Reference Yabumoto2008). In Reference Schultze, Arratia and Tintori2004, Gottfried, Rogers & Curry Rogers recorded an occurrence of a mawsoniid in the Coniacian/Santonian of Madagascar and in 2005, Cavin et al. reported an occurrence of a mawsoniid belonging to the Mawsonia/Axelrodichthys complex in the Campanian / Lower Maastrichtian of Cruzy, France, both localities corresponding to freshwater environments. Cavin et al. (Reference Cavin, Forey, Buffetaut and Tong2005) suggested that the presence of a mawsoniid in the Upper Cretaceous of Europe was the result of dispersal from Gondwana, which has yielded older and more diverse species of Mawsonia and Axelrodichthys. Following the attribution of the Jurassic Trachymetopon to the mawsoniids, Dutel, Herbin & Clément (Reference Dutel, Herbin and Clément2015) suggested that Late Cretaceous European mawsoniids might represent the upshot of the evolution of a European lineage rather than the result of a northward dispersal from Gondwana. However, the recent inclusion of the Late Cretaceous mawsoniids from France in the genus Axelrodichthys (Cavin, Valentin & Garcia, Reference Cavin, Valentin and Garcia2016) makes this hypothesis less likely because Trachymetopon is resolved as the sister genus of the pair Mawsonia–Axelrodichthys (Dutel, Herbin & Clément, Reference Dutel, Herbin and Clément2015). However, it should be pointed out that the diagnoses of most of the mawsoniid species are in need of revision and their phylogenetic relationships are still poorly supported, making any palaeobiogeographical scenario very weak. The discovery of PEY-2015.5.1 in Peygros is a confirmation that marine mawsoniids were roaming the European seas until latest Triassic time. New discoveries would help to better define mawsoniid taxa, and would help to understand their obviously complex evolutionary history.
Although Rhaetian bone beds containing remains of fishes and marine reptiles were reported from Provence as early as the 1860s (Coquand, 1868; Dieulafait, 1869), little has been published on these ichthyofaunas, with the exception of Corroy's brief note on the topic (Corroy, Reference Corroy1934). The find of a coelacanth, a taxon hitherto unreported from the Triassic of Provence, in the Rhaetian beds of the Peygros quarry, together with other vertebrate remains (Amiot et al. Reference Amiot, Séon, Bardet, Buffetaut, Cuny, Deesri, Giner, Martin and Suan2016), suggests that further fieldwork at Peygros and similar outcrops may lead to additional significant discoveries in these deposits.
Acknowledgements
This work was made possible through a grant INSU 2015 – INTERRVIE from the French Centre National de la Recherche Scientifique (CNRS). The authors would like to thank Mrs Isabelle Raignault, from Rio Tinto, for her help, interest and permitting us to access the quarry. U.D. was supported by a research fellowship grant 2015 from the French Embassy of Thailand. U.D. would like to thank Emmanuel Robert, curator of the geological collection of the University Claude Bernard Lyon 1, for giving her access to the collection under his care and his support during her stay in Lyon. We thank both anonymous reviewers for their constructive comments.