- Research Article
- Open Access
Taxonomic revision of the snakes of the genera Palaeopython and Paleryx (Serpentes, Constrictores) from the Paleogene of Europe
Swiss Journal of Palaeontology volume 140, Article number: 18 (2021)
Large constrictor snakes, referred to the genera Palaeopython and Paleryx, are an ecologically prominent part of the fauna of Europe during the Paleogene. Most species were named over a century ago and their taxonomy is largely based on isolated vertebrae. Furthermore, the majority of named taxa originate from imprecisely known localities within the Phosphorites du Quercy, in southern France, and thus their exact age is not known. We critically review and re-diagnose these genera based on personal examination of all existing type material, an array of new specimens, and a detailed literature review. We consider Palaeopython and Paleryx to be valid and propose vertebral characters to distinguish them. We recognize three valid species of Palaeopython, i.e. Palaeopython cadurcensis (type species) from the Phosphorites du Quercy, Palaeopython ceciliensis from Geiseltal, and Palaeopython helveticus from Dielsdorf (Switzerland), and one valid species of Paleryx, i.e. Paleryx rhombifer (type species) from Hordle Cliff (England). Four other species, which were previously treated as members of Palaeopython and Paleryx, i.e. “Palaeopython” filholii and “Palaeopython” neglectus from the Phosphorites du Quercy, “Palaeopython” fischeri from Messel, and “Paleryx” spinifer from Geiseltal, are also considered as valid but pertain to other genera. Among these four taxa, “Palaeopython” fischeri has been recently assigned to its own genus, Eoconstrictor. A new genus, Phosphoroboa gen. nov. is established to accommodate “Palaeopython” filholii. We designate a lectotype for Palaeopython cadurcensis and establish that the paralectotype maxilla and dentary are reasonably referred to this species. New material attributed to Palaeopython cadurcensis is described from the old collections of the Phosphorites du Quercy. Paleryx cayluxi, another species established from the old collections of the Phosphorites du Quercy, is synonymized here with Palaeopython cadurcensis. We further clarify important errors in the original description and figures of Paleryx cayluxi, identify the exact specimens that comprise the type series, and designate a lectotype. Much new material is described for Palaeopython ceciliensis from its type area in Geiseltal and intracolumnar variation is considered. We describe additional vertebral and cranial material of Paleryx rhombifer from its type area in Hordle Cliff. Based on this cranial material, we suggest non-booid affinities for Paleryx rhombifer. We designate a lectotype for Paleryx depressus and agree with its previous suggested synonymy with Paleryx rhombifer. We re-describe the lectotype and paralectotypes of “Palaeopython” neglectus and refer and describe new material of this species from the Phosphorites du Quercy, paying special attention to intracolumnar variation; we also defer a decision on its generic relations until more abundant and complete material can be studied. We describe new vertebral material of the booid Eoconstrictor cf. fischeri from Geiseltal; similar material was previously known only from Messel and Dielsdorf. We determine that Eoconstrictor fischeri contains two distinct and unrelated species and describe intracolumnar variation in the nominotype. We clarify certain issues regarding the type series of Paleryx spinifer, designate a lectotype, and report previously unrecognized cranial material associated with the latter specimen; we transfer this species to Eoconstrictor based on cranial features and recombine it as Eoconstrictor spinifer comb. nov. We finally describe much new vertebral and cranial material of Phosphoroboa filholii comb. nov. from the Phosphorites du Quercy (both from the old collections but also from the late Eocene localities of Escamps A and C), paying special attention to intracolumnar variation. Based on this cranial material from Escamps, we identify Phosphoroboa gen. nov. as a booid. An analytical approach is undertaken in many isolated remains in order to quantify vertebral structures and assess intracolumnar variation, as well as associating isolated cranial elements to vertebral-based taxa. 3D models of the type material of the Geiseltal and Messel taxa are presented. The importance of vertebrae in the taxonomy of fossil Constrictores is addressed, although it is acknowledged that it is cranial material that can afford the most reliable phylogenetic conclusions. The diversity, distribution, biogeographic origins, and final demise and extinction of large Constrictores in the Paleogene of Europe are discussed.
Large constrictor snakes—belonging to the lineage Constrictores Oppel, 1811a (sensu Georgalis & Smith, 2020), which encompasses boas and pythons—are a diverse and ecologically prominent part of terrestrial faunal assemblages in the European Paleogene. For instance, in the early–middle Eocene site of Messel, Germany, specimens referred to the species Palaeopython (currently Eoconstrictor) fischeri Schaal, 2004, which reached over 2 m in length, are among the most abundant squamate species and are known to have consumed higher-level carnivores such as small crocodylians (Greene, 1983), lizards (Smith & Scanferla, 2016) and carnivorous mammals (Gunnell et al. 2018). These large constrictors have been known since the middle of the nineteenth century (Owen, 1850) and partly due to their prominence (in respect of size and abundance) have been accorded a great deal of taxonomic attention. Yet, many species were established on the basis of isolated vertebrae, others on the basis of both vertebral and cranial material, and after 170 years of study their taxonomy is perhaps understandably muddled.
The aim of this paper is to disentangle the complex alpha taxonomy of the largest Constrictores from the Paleogene of Europe, which have variously been referred to the genera Palaeopython and Paleryx (Barnes, 1927; De Stefano, 1905; Filhol, 1877a, 1877b, 1877c; Georgalis & Scheyer, 2019a; Kuhn, 1939a; Lydekker, 1888a; Owen, 1850; Rage, 1974; Rochebrune, 1880, 1884; Schaal, 2004). More specifically, we critically review the type material and the taxonomic status of all nominal species in order to identify valid species and phenetically defensible genera. We view this work as a necessary first step towards clarifying the phylogenetic relationships of the species concerned, and it will facilitate continuing studies on more complete material from the German Konservat-Lagerstätten of Messel and Geiseltal. Most taxa are named on the basis of isolated vertebrae, and characters of isolated vertebrae are poorly suited to establishing phylogenetic relationships (e.g., Smith, 2013; Smith & Georgalis, in press). We also describe and figure previously undescribed cranial and vertebral material from the key Paleogene localities of the Phosphorites du Quercy (France), Geiseltal (Germany), Messel (Germany), and Hordle Cliff (United Kingdom). We further apply novel analytical practices to quantify certain vertebral structures, which can be used for disentangling intracolumnal and intraspecific variation. An analytical approach is also undertaken to infer association of isolated cranial elements to species typified by vertebral material.
Institutional abbreviations AMNH, American Museum of Natural History, New York, USA; BSPG, Bayerische Staatssammlung für Paläontologie und historische Geologie, Munich, Germany; CM, Carnegie Museum of Natural History, Pittsburgh, Pennsylvania, USA; GMH, Geiseltalmuseum of Martin-Luther Universität Halle-Wittenberg, now referred to as the Geiseltalsammlung, housed as part of the Zentralmagazin Naturwissenschaftlicher Sammlungen, Halle, Germany; HLMD, Hessisches Landesmuseum Darmstadt, Darmstadt, Germany; MBS, Naturhistorisches Museum, Basel, Switzerland; MDHC, Massimo Delfino Herpetological Collection, University of Torino, Torino, Italy; MNCN, Museo Nacional de Ciencias Naturales, Madrid, Spain; MNHN, Muséum national d’Histoire naturelle, Paris, France; MTM, Hungarian Natural History Museum, Budapest, Hungary; NHMUK, Natural History Museum, London, United Kingdom; NHMW, Naturhistorisches Museum Wien, Vienna, Austria; NMP, Národní Muzeum Praha, Prague, Czech Republic; PIMUZ, Palaeontological Institute and Museum, University of Zurich, Zurich, Switzerland; SMF-ME, Messel vertebrate collection, Senckenberg Research Institute and Natural History Museum, Frankfurt am Main, Germany; SMF-PH, Paleoherpetology collection, Senckenberg Research Institute and Natural History Museum, Frankfurt am Main, Germany; SMNK-PAL, Palaeontology collection of Staatliches Museum für Naturkunde Karlsruhe, Germany; UM, Université de Montpellier, Montpellier, France; UWr, University of Wrocław, Wrocław, Poland; YPM, Yale Peabody Museum; ZZSiD, Institute of Systematics and Evolution of Animals, Polish Academy of Sciences, Kraków, Poland.
Material and methods
The fossil specimens described herein are permanently curated at the collections of GMH, HLMD, MNHN, NHMUK, NHMW, PIMUZ, SMF-ME, SMNK-PAL, UM, and YPM. The 3D model of SMF-ME 929 was generated using smartSCAN 3D (Breuckman GmbH) with post-processing in PolyWorks v12 (InnovMetric Software Inc.). Extant material of Constrictores was used for comparative purposes and in the analytical methods, housed at the collections of CM, MDHC, MNCN, MNHN, MTM, NHMW, NMP, SMF-PH, UWr, and ZZSiD. Additionally, many comparisons were made with CT scans available at Digimorph (http://Digimorph.org 2002–2020).
Measurements and their anatomical abbreviations follow Auffenberg (1963). In particular, CL = centrum length, measured ventrally and including the condyle; CoH = cotyle height; CoW = cotyle width; NAH = neural arch height; NAW = neural arch width, measured at the maximum interzygapophyseal constriction; NSPL = neural spine length; PO-PO = distance between the outer edges of the postzygapophyses; PR-PR = distance between the edges of the prezygapophyses; ZGH = zygantrum height; ZW = zygosphene width (see Fig. 1). To facilitate intracolumnar and interspecific analysis of the degree of vaulting, we introduce a “vaulting ratio”, which is measured in posterior view. A line connecting the outer edges of the postzygapophyses (distance PO-PO of Auffenberg, 1963) serves as the baseline. The vaulting ratio is then the height of the roof of the zygantrum to the PO-PO line, measured on the midline, to the half-width of PO-PO (Fig. 1).
The taxonomic history sections, listing of type material and type localities, diagnoses, and the overall format in the “Systematic palaeontology” part follow recent taxonomic reviews of fossil turtle groups (e.g., Georgalis & Joyce, 2017; Georgalis et al., 2021b; Joyce, 2016). As such, the taxonomic histories of each species presented below include only the establishments of new species, new combinations, incorrect spellings of generic names or species epithets, and lectotype designations and thus do not include every single mention of each taxon into the literature. Taxonomy of extant snakes follows Georgalis and Smith (2020) for Constrictores, and for its inclusive clades, Pyron et al. (2014) for Booidea, and Reynolds et al. (2014) and Wallach et al. (2014) for Pythonoidea.
Localities and age
The snake material described herein originates from the Eocene of the Phosphorites du Quercy (various, unknown localities plus the precisely dated Escamps A and C [late Eocene, MP 19]), the early–middle Eocene (MP 11) of Messel, the late early or middle Eocene of Geiseltal, and the late Eocene (MP 17) of Hordle Cliff, The United Kingdom (Fig. 2). For reasons of completeness, we also refigure the holotype of Palaeopython helveticus from the late middle–late Eocene (MP 16–20) of Dielsdorf, Switzerland.
Phosphorites du Quercy, France
It is now known that the various localities of the Phosphorites du Quercy span a significant stratigraphic range, from the early Eocene (MP 8 + 9) until the early Miocene (MN 3), although the majority of them range from late middle Eocene (MP 16) to late Oligocene (MP 28) (Georgalis, 2017; Georgalis et al., 2021a; Rage, 2006; Sigé & Hugueney, 2006). Also, these localities are distributed over a considerable geographic area, extending over the current Departments of Lot, Tarn-et-Garonne, Tarn, and Aveyron, pertaining to the administrative region of Occitanie, in southern France (Sigé & Hugueney, 2006). Specimens from the so-called “old collections” from the Phosphorites du Quercy, which were mostly collected during the second half of the nineteenth century, originate from many different localities, and most lack any precise locality data. This applies to all type material of all named species of Palaeopython. As such, specimens from MNHN that were described in the nineteenth and early twentieth centuries lack precise locality data and this fact, frustratingly, concerns also the type material of Palaeopython cadurcensis, “Palaeopython” neglectus, and Phosphoroboa filholii comb. nov.
Among this type material, only for the paralectotype “mummy” of “Palaeopython” neglectus (MNHN.F QU16324) can a more precise age be estimated, but still of course accurate locality data are lacking. The age for this specimen can be estimated to pertain around the late middle–late Eocene (late Bartonian–late Priabonian [MP 16–19 or 20, about 40–34 Ma]). This assumption is based on the recent suggestion that all of the few snake and amphibian “mummified” specimens from the Phosphorites du Quercy described and figured by Filhol (1877a, 1877b, 1877c) originate from a single locality, the age of which is constrained by the “mummy” of Thaumastosaurus gezei Rage & Roček, 2007, a frog that is also known in other well-dated localities from the area (see Laloy et al., 2013; Tissier et al., 2016). In addition, we remark that in the earliest brief description of such “mummified” amphibian and snake specimens, Filhol (1873a) mentioned that these originated from the “dépôts de phosphates de chaux de l’Aveyron”, and therefore the area could be potentially more geographically constrained within the Department of Aveyron. However, an alternative interpretation was recently made by Lemmiere et al. (2021), who suggested that all “mummified” specimens could probably originate from a single, unknown, locality in the vicinity of Escamps, Department of Lot, and more particularly, potentially from one of the sites of Les Rosières (MP 17–19) or Les Tempories (MP 19). If this latter interpretation is correct, then the geographic provenance of the paralectotype “mummy” MNHN.F QU16324 is from the Department of Lot (and not Aveyron), but still its age is roughly identical (MP 17–19) with that of previous recent suggestions of Laloy et al. (2013) and Tissier et al. (2016).
This uncertainty about the exact provenance data (and therefore the age) also applies to the rich and abundant NHMW and YPM Quercy material. Many NHMW specimens (NHMW 2019/0032/0001; NHMW 2019/0032/0005; NHMW 2019/0033/0001–NHMW 2019/0033/0041; NHMW 2019/0033/0049–NHMW 2019/0033/0084; NHMW 2019/0033/0130–NHMW 2019/0033/0153) are lacking labels, besides the basic information “Quercy”. Several NHMW specimens (NHMW 2019/0032/0003–NHMW 2019/0032/0004; NHMW 2019/0033/0085–NHMW 2019/0033/0129) are accompanied by an old label written in French “Trouvés dans diverses exploitations du Tarn-et-Garonne et du Lot” which suggests the mixed nature of the material. Three specimens (NHMW 2019/0020/0001–NHMW 2019/0020/0003) had only an old French label written “Trouvés dans le Department du Lot et du Tarn et Garonne”, whereas a few more specimens (NHMW 2019/0032/0002; NHMW 2019/0033/0042–NHMW 2019/0033/0048) had a relatively new label written in German “Schlangen Oligozän Quercy” [apparently this Oligocene corresponds to an older concept of that age, which is now instead considered as late Eocene (e.g., Bonis et al., 1973)]. The most precise locality information in the NHMW sample was given for specimen NHMW 2019/0066/0001, which was accompanied by a new label termed “Mouilliac bei Caylux”. However, this information only denotes that it originated near the village of “Mouilliac” (currently called Mouillac), near Caylus, in the Department of Tarn-et-Garonne but is too general, as there are plenty of different localities in that vicinity (see map in Bonis et al., 1973: fig. 1). Overall, the NHMW Palaeopython and Phosphoroboa gen. nov. material was accompanied by a vertebra of the Eocene “erycine” Cadurceryx and an array of lizard remains, pertaining to both Eocene and Oligocene taxa, hence further suggesting the mixed nature of the localities (see Georgalis et al., 2021a for these NHMW lizards).
Regarding the YPM material, the only available information denotes that it was collected near the village of Mouillac. However, like with the Mouillac material from NHMW, with only this information at hand, it is impossible to state whether they were all collected from a single or more than one of the several known localities in the area of Mouillac. Furthermore, besides snakes, the YPM Quercy material includes squamates that are both of Eocene and Oligocene age, such as Saniwa (Eocene) and Pseudeumeces (Oligocene).
The dentary NHMUK PV R 3489 also originates from an unknown locality near the village of Mouillac. The vertebral material from the collections of NHMUK bears the basic information “Caylux, Tarn-et-Garonne”, which apparently hints towards some nearby locality(ies) near that town (modern spelling “Caylus”).
Lastly, a small number of vertebrae and skull elements of Phosphoroboa filholii comb. nov. from the collections of UM represent the only specimens from the Phosphorites du Quercy that bear precise locality data, originating from Escamps (A and C), a locality in the Department of Lot, that pertains to the late Eocene (MP 19) (Augé, 2005, 2012; Georgalis, 2017; Rage, 2006; Sigé & Hugueney, 2006). Although located in different Departments, Escamps is relatively geographically near Mouillac, the area that yielded the YPM specimens and the dentary NHMUK PV R 3489. Escamps is a rather important locality, representing the namesake of the “landmark-level” (“niveau-repère” of Crochet et al., 1981) of Escamps, the final such level that precedes the “Grande Coupure” (Crochet et al., 1981; Rage, 1984a).
The Messel Pit (colloquially “Messel”) is located about 20 km south of Frankfurt am Main, Hesse, Germany. Messel has yielded a diverse vertebrate fauna, with many specimens characterized by exceptional, exquisite preservational attributes, including gut contents and the remains of scales, plumage and fur (Smith et al., 2018). There is also a rich arthropod fauna (Wedmann, 2018) and flora (Collinson et al., 2012; Lenz & Wilde, 2018). Messel represents an ancient maar lake formed by a phreatomagmatic eruption at 48.2 Ma (Lenz et al., 2015). All fossils derive from the Middle Messel Formation, which represents the stable phase of the lake’s history, in which a permanent stratification of the water column prevailed (Goth, 1990). The excavation sites in the pit span an interval of about 22–28 m (there is some uncertainty in the exact horizon of individual specimens) corresponding to c. 160–200 ka. Astronomically calibrated pollen profiles suggest that this profile spans the Ypresian–Lutetian boundary (Lenz et al., 2015).
The former Geiseltal brown coalfield (today occupied by the Geiseltal lake), 30 km south of Halle (Saale), Sachsen-Anhalt, Germany, yielded a diverse fossil flora and fauna often characterized by exceptional preservation (e.g., Barnes, 1927; Krumbiegel et al., 1983). The fossiliferous part of the Geiseltal succession consisted of three major coal seams intercalated by silty and sandy clastic deposits with a maximal thickness of ca. 290 m. Vertebrate localities (“quarries”) of variable preservational conditions were distributed vertically and horizontally within the coal seams (e.g., Haubold & Thomae, 1990). Specimens of large Constrictores in Geiseltal, described herein, have been recovered from the quarries IX, LII, I, VI, XXII, XLI, LIX, XXXIII, XXXV–XXXVIII, Cecilie I–IV, and Leonhardt I and III; among them, quarry Cecilie I is the type locality of the two named species from Geiseltal. The succession has long been considered Eocene in age (e.g., based on the tapiromorph Lophiodon; Schroeder, 1913) and lithostratigraphic correlation with nearby coalfields grading into heterotypic marine deposits (e.g., Quitzow, 1948). Regional spore–pollen biostratigraphy suggested an age spanning the entire Lutetian (middle Eocene, Spore/Pollen Zone 14/15–15; Krutzsch, 1966, 1970, 1976; Krutzsch et al. 1992). Jaeger (1971) also suggested Lutetian age for the Geiseltal fauna (middle Lutetian for the “Mittelkohle”—“middle coal” seam) based on similarities with the Bouxwiller mammals from Alsace in France, coming from lacustrine beds that can be correlated with the marine Lutetian (middle Eocene).
The works of Krutzsch apparently inspired the subsequent biochronology using mammal faunas, which established Geiseltal as the reference locality for the middle Eocene Mammal Paleogene levels (MP) 11–13 (= Geiseltalian). Geiseltal was considered to range from MP 11 to MP 14 and span the Lutetian (Franzen & Haubold, 1987; Haubold, 1987, 1989; Haubold & Thomae, 1990; Schmidt-Kittler et al., 1987). While apparent faunal similarities with Messel reasonably assume close age proximity, the temporal duration of the Geiseltal record is in fact less straightforward. A recent revision of propalaeothere and tapiromorph perissodactyls, taxa largely underpinning mammal biochronology at Geiseltal, however, found that earlier taxonomies are not reproducible and revealed a considerably lower diversity across the section than previously thought (Ring et al., 2020). Moreover, the mammal biochronology of Geiseltal, while based on first and last occurrence of index species, does not consider apparent taphonomic and collecting bias. Rigorous assignment of the various coal seams (Unter-, Mittel-, Oberkohle—lower, middle and upper coal) to MP levels is therefore impossible at the moment.
Until the regional palynostratigraphy of Krutzsch is better correlated with marine deposits, the inferred 5–11 Myr duration of the Geiseltal record (Franzen, 2005; Haubold & Thomae, 1990; Krutzsch, 2011) is doubtful. Considering the maximum present-day tropical carbon accumulation rate in peat with modelled mass loss during coalification (Large & Marshall, 2015), the deposition of fossiliferous lignite seams with the same minimal vertical thickness as in Geiseltal (70 m) has been estimated to require just over 1 Myr (Ring et al., 2020). The maximal thickness of the fossiliferous Geiseltal lignite seams totals 210 m and therefore requires ca. 3 Myrs at this depositional rate (67 cm/10 Kyr; Large & Marshall, 2015). The expected relatively low organic carbon content of the Geiseltal lignite seams (max. 60%, but usually significantly lower; Gusterhuber, 2007) and the inferred constant paratropical climate (Ring et al., 2020 and references therein) suggest that considering the maximal accumulation rate is reasonable (Large & Marshall, 2015). The total thickness of the clastic interbeds is 80 m, which may correspond to 800 Kyr calculating with the lower end of modern-day fluvial sedimentation rate (0.02 cm/year; Ferring, 1986) and the higher end of compaction rate (0.2 cm/year; Meckel et al., 2007). Consequently, based on sedimentation and compaction rates, the maximal temporal span of the bottom and top of the Geiseltal fossil record can be conservatively estimated to ca. 3.8 Myr. These being said, fossil remains from Geiseltal are dated around the late early or middle Eocene.
Hordle Cliff, England
The Hordle Cliff (also known in old literature under the name Hordwell) area in southern England is known since the first half of the nineteenth century and yielded a diverse array of vertebrate taxa, including several species of reptiles (Benton & Spencer, 1995; Holman & Harrison, 1998a, 1998b; Holman et al., 2006; Hooley, 1905; Klembara & Green, 2010; Lydekker, 1888a; Owen, 1850; Owen & Bell, 1849; Seeley, 1876). From there, Owen (1850) described the fossil constrictors Paleryx rhombifer and Paleryx depressus. All specimens described by Owen (1850) were collected from Hordle Cliff, with no precise stratigraphical data available; however, from Hastings (1852) it can be deduced that all Hordle Cliff snake remains were probably collected either from “Hastings Bed 1” (which now corresponds to the “Rodent Bed”) and from the upper part of “Hastings Bed 15” (which now corresponds to the “Mammal Bed”) (Milner et al., 1982). In any case, both “Rodent Bed” and “Mammal Bed” of Hordle Cliff are parts of the Totland Bay Member of the Headon Hill Formation and are currently considered as coeval (MP 17) (Klembara & Green, 2010), so we can be confident at least about the age of both Paleryx rhombifer and Paleryx depressus type material.
Taxonomic history and history of discoveries of the European large fossil Constrictores
In the middle of the nineteenth century, Richard Owen (1850) established a new genus of fossil snakes, Paleryx, comprising two species, Paleryx rhombifer and Paleryx depressus, on the basis of isolated vertebrae from the late Eocene of Hordle Cliff. This act marked the first ever description of a constrictor in the history of snake palaeontology, taking into consideration that the previously described giant aquatic snakes of the genus Palaeophis Owen, 1841, from the Eocene of the United Kingdom, which were originally considered to have affinities with boas and pythons (e.g., Owen, 1841, 1850), are no longer considered to have close affinities with that lineage (e.g., Georgalis et al., 2020b; Rage et al., 2003). Only a few years later, Pictet et al. (1855–1857) described and figured vertebral material from the Eocene of mount Mormont, near Éclépens, Switzerland, which they assigned to two forms: one large-sized taxon, which they referred to Python Daudin, 1803a, and for which they estimated a total length of 3.25 to 3.5 m, and a smaller-sized one, which they suggested that it was referrable to either Python or Paleryx. Gervais (1873, 1876) figured a trunk vertebra from the Phosphorites of Quercy, which he tentatively considered as comparable to Palaeophis (misspelled as “Paléophis”); however, this material was soon later re-identified as pertaining instead to constrictors by Filhol (1877a, 1877b, 1877c), and subsequently to Palaeopython (Rage, 2006). Filhol (1877a, 1877b, 1877c) established a new extinct species of Python (i.e. Python cadurcensis) on the basis of cranial and vertebral material from the Phosphorites of Quercy. A few years, later, Rochebrune (1880) established his new genus Palaeopython in order to accommodate this Quercy form, regarding it as distinct enough from the extant species of Python and particularly Python molurus (Linnaeus, 1758) and Python sebae (Gmelin, 1789). In the same paper, Rochebrune (1880) also described a second species of this genus, Palaeopython filholii, on the basis of a single trunk vertebra, again from the Phosphorites du Quercy. Rochebrune (1884) later referred additional specimens to both Palaeopython cadurcensis and Palaeopython filholii and also established a third species of the genus, Palaeopython neglectus, again from the Phosphorites du Quercy. Zittel (1887–1890) described and figured an additional vertebra of Palaeopython cadurcensis from Quercy. The same author discussed the then-named species of both Palaeopython and Paleryx and provided emended diagnoses for both genera, but nevertheless he admitted the strong resemblance between them (Zittel 1887–1890). Lydekker (1888a) described several new specimens of both Paleryx rhombifer and Paleryx depressus and suggested that Palaeopython resembles Paleryx so greatly that the two genera should probably be considered synonyms, but he did not express this with certainty and furthermore left open the possibility that Palaeopython cadurcensis could be generically distinct as well. In another paper published in the same year, Lydekker (1888b) appeared even more confident about the synonymy of Palaeopython with Paleryx, whereas he also tentatively placed Palaeopython filholii into the synonymy of Paleryx depressus. This purported synonymy was subsequently followed by several authors who treated Paleryx as the valid genus name for the large Quercy constrictor (De Stefano, 1903, 1905; Kuhn, 1939a, 1963; Nicholson & Lydekker, 1889; Romer, 1956; Swinton, 1926).
Other authors of the early and middle twentieth century defied this synonymy opinion and chose to treat Palaeopython and Paleryx as distinct (e.g., Arldt, 1909; Auffenberg, 1963; Barnes, 1927; Gilmore, 1938; Hoffstetter, 1955; Huene, 1956; Nopcsa, 1928; Počta, 1905; Simpson, 1933; Williston, 1925; Woodward, 1932). Portis (1901) established his new species Palaeopython sardus on the basis of cranial material from the middle Miocene of Monte Albu, Sardinia, Italy, an occurrence that seemed to be the youngest record of Palaeopython, but this material has recently been demonstrated by Delfino et al. (2014) to not even to be a snake at all, but instead to represent a fish—accordingly, this taxon is omitted from consideration herein. De Stefano (1905), who accepted the synonymy of Palaeopython with Paleryx, described several new specimens from the Phosphorites du Quercy, which he referred to Paleryx rhombifer, Paleryx filholi (sic), Paleryx neglectus, but also established a fourth species of large constrictor, Paleryx cayluxi. Barnes (1927) described the first fossil constrictors from Germany by establishing two new species, Palaeopython ceciliensis and Paleryx spinifer from the middle Eocene of Geiseltal. Again from Geiseltal, Weigelt (1929) described and figured several large vertebrae, but he misidentified them for the large lizard Palaeovaranus Zittel, 1887–1890. When comparing his newly described giant Paleogene Argentinean taxon Madtsoia bai with European fossil large constrictors, Simpson (1933) pointed out important features of both Palaeopython and Paleryx that could differentiate them not only from his South American form, but also from each other. Gilmore (1938) also treated the two European genera as valid and proposed further characters that could differentiate Palaeopython from the North American Boavus Marsh, 1871. The most complete material of Palaeopython to that date was described by Kuhn (1939a), including partial skeletons with skulls from Geiseltal. The same author further properly referred Weigelt’s (1929) vertebrae to snakes and particularly to P. ceciliensis, but he assigned both this taxon and P. spinifer to a single genus, Paleryx (Kuhn, 1939a). Since then, the beautiful snake skeletons from Geiseltal have received little attention, with only sporadic mentions with figures in the next decades (Krumbiegel et al., 1983). After a relatively long period without new descriptions or new remains, fieldwork occurring at various Eocene localities within the Phosphorites du Quercy during the past five decades resulted in the recovery of new material of large snakes. Much of this material remains undescribed, but important descriptions were published (Rage & Augé, 2010; Rage, 1974, 1978, 1988b). The most comprehensive of these works was that of Rage (1974) on Quercy snakes, which allowed a better understanding of Palaeopython. In this work, Rage (1974) once again regarded Palaeopython and Paleryx as distinct, a view that has been widely accepted since then (e.g., Carroll, 1988; Georgalis & Scheyer, 2019a; Holman et al., 2006; Rage & Augé, 1993, 2010; Rage, 1984b, 1987, 2001; Szyndlar & Rage, 2003; Szyndlar & Schleich, 1993; Wallach et al., 2014). The same author described additional vertebral and cranial material from Quercy, which he referred to Palaeopython filholii (Rage, 1974). Furthermore, for the first time since the end of the nineteenth century, Rage and Ford (1980) described additional material of Paleryx rhombifer from the late Eocene of the United Kingdom and also synonymized Paleryx depressus with the former species; subsequent mentions of Paleryx rhombifer from the Eocene of the United Kingdom were also made by Milner et al. (1982), Benton and Spencer (1995), and Holman et al. (2006) but were accompanied by no figures and therefore cannot be evaluated.
Moreover, the well-known and exceptionally preserved fossil locality of Messel, Germany, began to yield beautiful Eocene specimens similar to Palaeopython, initially briefly documented by Greene (1983), based on a cast, and by Habersetzer and Schaal (1990), the latter eventually becoming one of the paratypes of the recently described Messelopython freyi Zaher and Smith, 2020. More detailed studies revealed that the Messel form was specifically distinct from all other known taxa and afforded the establishment of a species of its own, Palaeopython fischeri (Schaal, 2004; see also Smith & Scanferla, 2016; Smith et al., 2018). Rosselet (1991) briefly described Palaeopython remains from the vertebral material from the middle to late Eocene of Dielsdorf, Switzerland. That material, along with other specimens from Dielsdorf, was subsequently extensively described and figured by Georgalis and Scheyer (2019a), who established another new species, Palaeopython helveticus. Finally, Scanferla and Smith (2020b) established the genus Eoconstrictor to accommodate the Messel species Palaeopython fischeri and, on the basis of a phylogenetic analysis, they suggested booid affinities for this taxon.
Owen (1850) highlighted the close resemblance of Paleryx with the extant genera Eryx Daudin, 1803b, and Python. Subsequent early workers highlighted the strong (or even congeneric) affinities of Palaeopython and Paleryx with Python (De Stefano, 1905; Filhol, 1877a, 1877b, 1877c; Lydekker, 1888b; Nicholson & Lydekker, 1889; Palacký, 1884; Rochebrune, 1880, 1884). Nevertheless, these extinct genera were until recently almost universally considered to be boids (Kuhn, 1939b, 1963; McDowell, 1987; Rage, 1974, 1984b, 1987; Szyndlar, 1994; Szyndlar & Böhme, 1993; Szyndlar & Rage, 2003; Szyndlar & Schleich, 1993; Underwood, 1976), though in most of these taxonomic concepts of past decades, boids included both boas and pythons (see Georgalis & Smith, 2020). In its original establishment, Rochebrune (1880) placed Palaeopython into Pythonidae, for which group he followed the terminology of Duméril (1853) as Holodontiens—Pythonides. Palacký (1884), Zittel (1887–1980), Nicholson and Lydekker (1889), De Stefano (1905), and Počta (1905) also followed such placement into Pythonidae, considering it distinct from Boidae (their Boaeidae). Among recent authors, only Wallach et al. (2014) and Boundy (2021) treated Palaeopython and Paleryx as pythonids, though they did not provide any rationale for such a taxonomic arrangement. The taxonomic assignment of the two Paleogene genera to boids has been primarily made on the basis of cranial anatomy, and particularly the pterygoid, maxilla, and palatine morphology (Rage, 1974, 1984b; Szyndlar & Rage, 2003). So far, paracotylar foramina on the vertebrae of Palaeopython [a feature that is characteristic of certain booids (Georgalis, 2019; Georgalis et al., 2019b; Ivanov, 2000; Szyndlar & Rage, 2003)] had never been documented, with the exception of a record by Szyndlar and Rage (2003) who mentioned (but did not figure) a vertebra from Quercy (AMNH 242) that was clearly referable to Palaeopython and possessed a pair of paracotylar foramina. We here document the presence of paracotylar foramina in a few vertebrae of Palaeopython cadurcensis and Palaeopython ceciliensis, as well as in a few specimens of Paleryx rhombifer and Phosphoroboa filholii comb. nov., thus demonstrating that this feature, though rare, was still more widespread than previously thought. Furthermore, the extreme thickness observed in the zygosphene of the vertebrae of some members of Palaeopython is in fact more reminiscent of pythonids rather than that of similarly sized boids. Rage (1974) considered the morphology of a pterygoid from the late Eocene of Escamps, which he referred to Phosphoroboa filholii comb. nov. (then placed in Palaeopython), to be more reminiscent of booids instead of pythonoids—we here redescribe and refigure this pterygoid and concur with its booid affinities. Moreover, Habersetzer and Schaal (1990), using X-radiographs, reported the presence of premaxillary teeth (a typical feature of pythonoids) in a specimen (SMF-ME 710) similar to Palaeopython from the early–middle Eocene of Messel—this observation on that material was subsequently questioned by Baszio (2004). Zaher and Smith (2020) later referred that specimen to the pythonoid Messelopython freyi, which in fact has premaxillary teeth; therefore, this specimen does not pertain to Palaeopython.
Georgalis and Scheyer (2019a) suggested that it is more secure to assign Palaeopython and Paleryx to “Booidea” sensu lato, which currently corresponds to Constrictores sensu Georgalis and Smith (2020), the clade comprising both boas and pythons. We also here treat both these two genera as Constrictores incertae sedis, thus not implying any closer relationship to either the lineages of Booidea or Pythonoidea. We note nevertheless, that based on the cranial morphology of Paleryx rhombifer, booid affinities are excluded for this taxon (see that entry below). Similarly, “Palaeopython” neglectus is also treated as Constrictores incertae sedis. On the other hand, the two species of Eoconstrictor (i.e. Eoconstrictor fischeri and E. spinifer comb. nov.) afford a more precise taxonomic placement, as this genus has recently been demonstrated to pertain to Booidea (sensu Pyron et al., 2014), having in particular close phylogenetic affinities with extant Neotropical boas (Scanferla & Smith, 2020b). We also suggest booid affinities for Phosphoroboa filholii comb. nov., based primarily on the cranial remains we describe from Escamps.
Squamata Oppel, 1811b
Serpentes Linnaeus, 1758
Alethinophidia Nopcsa, 1923
Genus Palaeopython Rochebrune, 1880
Type species: Python cadurcensis Filhol, 1877a.
Geographic and stratigraphic distribution: see Table 3.
Diagnosis: Palaeopython is assigned to Constrictores on the basis of its vertebrae being strongly built, the vertebrae taller than long, the centrum triangular and wider than long, prezygapophyseal accessory processes small, paradiapophyses undivided or only weakly divided into diapophyseal and parapophyseal portions, and lack of parazygantral foramina. Palaeopython can be differentiated from Paleryx and Phosphoroboa gen. nov. by the following features: larger size, with CL of mid-trunk vertebrae commonly > 10 mm and up to 19 mm; very thick, strongly trapezoidal zygosphene with flat anterior face in fully grown individuals; shallow, asymmetrical interzygapophyseal constriction, extending no more medially than the mid-point of the prezygapophyseal articular facet; deeper paracotylar fossae; relatively sharp, slightly projecting ventral median keel present on centrum; small median tubercle on zygosphene usually present; and zygosphene broadly concave with anterolateral horns. Palaeopython can be differentiated from Eoconstrictor by the presence of a flared posterior process of the maxilla, a much thicker zygosphene, and relatively shorter mid-trunk vertebrae (CL/NAW ratio < 0.9). Palaeopython can further be differentiated from roughly co-occurring Phosphoroboa gen. nov. by the presence of a flared posterior process of the maxilla, a longer neural spine on mid- and posterior trunk vertebrae (NSPL/CL > 0.5), and wider zygosphene (ZW/CoW > 1.2 on mid-trunk vertebrae). Palaeopython can further be differentiated from Messelopython by its larger size, shorter palatine process of the maxilla, a longer neural spine (NSPL/CL > 0.5 on mid-trunk vertebrae), and wider zygosphene (ZW/CoW > 1.2 on mid-trunk vertebrae).
Remarks: Rage (2001:Table 1b) provided important characters to distinguish several extinct and extant constrictors, including Palaeopython from Paleryx. Regarding the latter two genera, he particularly noted differences in the shape of the prezygapophyseal articular facets, the shape of the centrum, and the shape of the posterior median notch of the neural arch. We accept that the centrum shape often differs from that of Paleryx due in part to the convex subcentral ridges. However, we find that the posteromedian notch on the neural arch is variable even in the type series of Palaeopython cadurcensis. Also, we find that the prezygapophyseal articular facets grow much broader during ontogeny in vertebrae of the same position in extant Boa constrictor Linnaeus, 1758 (SMF-PH 45 vs. SMF-PH 40), suggesting that this reported difference may just be allometric; however, the main reason that this character does not obtain in our taxonomic scheme is because we have altered the composition of Palaeopython with respect to Rage’s (1984b, 2001) conception.
Of the new features we have adduced, it is noteworthy that the shallow and asymmetrical interzygapophyseal constriction (reaching its deepest point posterior to mid-length) is seen in some extant constrictors. In Boa constrictor, it appears to be developed in a stable manner over a very large range of body size, as seen in specimens SMF-PH 46 and SMF-PH 40.
Like many other nineteenth century named genera, Palaeopython suffered from numerous misspellings throughout its long taxonomic history, with exaggerating name morphs such as “Palaeophyton” (e.g., Počta, 1905), which would literally translate as “old plant” in Greek!
Palaeopython cadurcensis (Filhol, 1877a)
(= Paleryx cayluxi De Stefano, 1905)
Taxonomic history: Python cadurcensis Filhol 1876b (nomen nudum); Python cadurcensis Filhol, 1877a (new species); Palæopython cadurcensis Rochebrune, 1880 (new combination); Paleryx rhombifer = Palaeopython cadurcensis Lydekker, 1888b (junior synonym); Paleopython candurcensis Palacký, 1898 (incorrect spelling of genus name and species epithet); Palaeopython Cadurcensis De Stefano, 1905 (emended spellings of genus name and species epithet); Palaeophyton cadurcensis Počta, 1905 (incorrect spelling of genus name).
Type material: MNHN.F QU16318 (formerly MNHN QU 318) (lectotype, herein designated), five articulated mid-trunk vertebrae (Filhol, 1877a, figs. 441–442; Filhol, 1877b, figs. 441–442; Filhol, 1877c, figs. 441–442; Georgalis & Scheyer, 2019a, fig. 15c; Figs. 3e, f, 4); MNHN.F QU16321 (formerly MNHN QU 321) (paralectotype), a complete left (not “right” as stated by Wallach et al., 2014) maxilla with teeth (Filhol, 1877a, figs. 435, 436; Filhol, 1877b, figs. 435, 436; Filhol, 1877c, figs. 435, 436; Rochebrune, 1884, pl. II.4; Rage, 1984b, fig. 12a; Figs. 3a, b, 5); MNHN.F QU16317 (formerly MNHN QU 317) (paralectotype), a right (not “left” as stated by Wallach et al., 2014) dentary with some preserved teeth (Filhol, 1877a, figs. 437, 438; Filhol, 1877b, figs. 437, 438; Filhol, 1877c, figs. 437, 438; Rochebrune, 1884, pl. II.3; Rage, 1984b, fig. 12b; Figs. 3c, d, 6); unknown collection (paralectotype), an isolated mid-trunk vertebra (Filhol, 1877a, figs. 439–440 and 443–444; Filhol, 1877b, figs. 439–440 and 443–444; Filhol, 1877c, figs. 439–440 and 443–444; Fig. 3g–j); unknown collection (paralectotypes), (unknown number of) vertebrae (Filhol, 1877a, not figured).
Type locality: Unknown precise locality, Phosphorites du Quercy, France; probably middle or late Eocene.
Previously referred material and range: Middle Eocene (MP 16), Lavergne, Quercy, France (referred material of Rage, 2013); middle Eocene (MP 16), Le Bretou, Quercy, Tarn-et-Garonne, France (referred material of Rage, 1988b); middle Eocene (MP 16), Robiac, Gard, Occitanie, France (material referred to Paleryx cayluxi by Rage, 1984b); middle or late Eocene, unknown precise locality(ies), type area, Phosphorites du Quercy, France (referred material of Rochebrune, 1880, Lydekker, 1888a, De Stefano, 1905, and Georgalis and Scheyer, 2019a; material referred to Palaeopython filholii [partim] by Rochebrune, 1884; material referred to Paleryx rhombifer by Lydekker, 1888a; material referred to Paleryx rhombifer by De Stefano, 1905; lectotype of Paleryx cayluxi De Stefano, 1905).
Referred material discussed herein: Phosphorites du Quercy, imprecise localities (Figs. 7, 8, 9, 10, 11, 12, 13, 14and 15): 26 trunk vertebrae (MNHN.F QU16319, MNHN.F QU16343–MNHN.F QU16357, NHMUK PV R 428 [two vertebrae], NHMW 2019/0032/0001–NHMW 2019/0032/0005, NHMW 2019/0033/0003, NHMW 2019/0033/0037, NHMW 2019/0033/0051) and a probable cloacal vertebra (NHMUK PV R 2799(1)). Phosphorites du Quercy, imprecise localities near Mouillac: two trunk vertebrae (NHMW 2019/0066/0001 and YPM-VPPU 29855).
Diagnosis: Palaeopython cadurcensis can be referred to the genus Palaeopython by the presence of the full list of characters given for that genus. Palaeopython cadurcensis differs from Palaeopython helveticus in being much larger (up to 19 mm centrum length) and having a less vaulted neural arch and thicker zygosphene. Palaeopython cadurcensis differs from Palaeopython ceciliensis in its larger size, in having less vertically oriented zygosphenal articular facets, and in the presence of a weak zygosphenal tubercle. Palaeopython cadurcensis differs from both Palaeopython ceciliensis and Palaeopython helveticus in having more laterally extended, squared off prezygapophyses.
Description of the lectotype (MNHN.F QU16318): MNHN.F QU16318 is a string of five large articulated mid-trunk vertebrae (Fig. 4). The three anterior vertebrae have a centrum length (CL) of c. 12.1 mm. The anterior view is discernible only in the anterior-most vertebra (Fig. 4a), showing that the zygosphene is extremely thick. There is a small tubercle at mid-height on its anterior face. The lateral edges of the zygosphene project strongly dorsally and anteriorly, achieving a horn-shape. The zygosphene is much wider than the cotyle (12.9 mm and 11.0 mm, respectively; ratio ZW/CoW = 1.17). The cotyle is deep and almost circular. No paracotylar foramina are present. The prezygapophyses project laterally and slightly dorsally, so that the angle subtended by the zygosphenal and prezygapophyseal articular facets is 43–47°. The shape of the neural canal cannot be fully evaluated, due to adhering sediment, but it appears to be roughly triangular in cross-section. The posterior view can only be studied in the posterior-most vertebra (Fig. 4b). The vaulting ratio is 0.38, i.e. moderately vaulted. The zygantrum is deep and is marked dorsally by prominent, elongate zygantral mounds (sensu Hecht in McGrew, 1959) on the dorsal surface of the neural arch. The shape of the condyle cannot be evaluated, as that portion of the vertebra is damaged. In dorsal view (Fig. 4c), the vertebrae are wider than long. The neural spines are relatively thick and they develop almost immediately after the zygosphene. The neural spine overhangs the median notch of the neural arch posteriorly. The postzygapophyses are subrectangular and project strongly laterally. The prezygapophyses also project laterally and the prezygapophyseal articular facets are oval. The prezygapophyseal accessory processes are small and acute. The zygosphene is exposed only on the anterior-most vertebra. It is concave (sensu Auffenberg, 1963). In ventral view (Fig. 4d), the centrum is subtriangular, being much widened anteriorly. The haemal keel is fairly sharp throughout its length; it extends along nearly the entire length of the centrum and grows slightly deeper posteriorly. The paradiapophyses are only weakly divided into dorsal diapophysis and ventral parapophysis; the former is weakly convex and the latter is weakly concave and projects below the centrum. In lateral view (Fig. 4e), the vertebrae are much taller than long. The neural spine is best preserved in the second vertebra. The neural spine is moderately high and posteriorly inclined.
Description of the paralectotype vertebra figured in Filhol (1877a, 1877b, 1877c): This is an isolated trunk vertebra (Fig. 3g–j). Taking into account that the specimen cannot be located, the description herein is based on the original lithographic illustration of Filhol (1877a, 1877b, 1877c). As such, certain features might be subjected to inaccuracies. Like the lectotype, it probably pertains to the mid-trunk region from the vertebral column, though possibly more anteriorly in the series. The specimen is similar to the lectotype, and therefore we focus on major differences and features not visible in the lectotype. The zygosphene is less concave than in the lectotype. A small tubercle on the zygosphene appears to be absent. The diapophysis appears to be more prominent than the lectotype, but its dorsal margin is less sharp. The angle subtended by the zygosphenal and prezygapophyseal articular facets is somewhat greater, which is consistent with the much greater vaulting of the neural arch. The neural canal is small and triangular. The neural spine is taller. Zygantral mounds may be present, but they are much less discrete than in the lectotype. The condyle is nearly circular, just slightly taller than wide, and (like the cotyle) is significantly larger than the neural canal. The lateral view is not depicted in the lithographic illustration.
Description of the paralectotype maxilla (MNHN.F QU16321): MNHN.F QU16321 is a large, incomplete left maxilla (Fig. 5). We consider the assignment of this specimen to the same species as the holotype to be justified on the basis of the size of the teeth, which is similar to the size of the paralectotype dentary teeth (see "Remarks" below). The premaxillary process is smooth and rounded. It is club-shaped, growing in width anteriorly before tapering again to a point. Like in other snakes, it lacks an articulation facet for the premaxilla. On its dorsal surface there opens a large anterior dorsal foramen [sensu Scanlon (2001), probably homologous with the “anterior superior alveolar foramen” of Smith and Gauthier (2013) in lizards], located above the anterior part of the 3rd tooth position. The foramen continues anteriorly as a deep, anterolaterally trending groove; it grows in width anteriorly as well, thus maintaining a nearly constant width ratio of 0.26–0.27 with respect to the premaxillary process. A relatively strong remnant of the lizard facial process (sensu Gauthier et al., 2012) rises at the level of the anterior dorsal foramen and diminishes gradually, disappearing at roughly the level of the palatine process. On the medial side of its posterior half is a set of smooth grooves and, posteriorly, a depression that indicate where the prefrontal articulated. The grooves suggest a more intimate connection of the maxilla and prefrontal than in Boa Linnaeus, 1758. There is a single, relatively small labial foramen at the level of the boundary between the 4th and 5th tooth positions. Near the posterior end of the facial process, at approximately the level of the palatine process, is a small, horizontal groove of unknown significance; there might be a tiny foramen at its anterior end.
The palatine process has a relatively short base (covering tooth positions 8–10), more reminiscent of booids; its distal extremity is broken. On its dorsal surface near the anterior base is a medially opening foramen, homologous with the superior alveolar foramen of lizards (Maisano & Rieppel, 2007), which receives the maxillary artery and superior alveolar nerve. There is a deep fossa on the dorsal surface of the bone level with the posterior margin of the palatine process, which might house a second foramen, although this was not ascertained during our respective visits. Note that in the bolyeriid Casarea dussumieri (Schlegel, 1837), a fossa is present (where it receives the “suborbital ossification” of Maisano and Rieppel ) but not a foramen, whereas in Boa there is a foramen in the same position that communicates with the superior alveolar canal. The posterior process (sensu Oelrich, 1956) of the maxilla is broken.
The specimen contains 13–14 tooth positions and 7 teeth. The teeth are conical. Their surfaces are generally smooth, except for a small, horizontal depression near the labial base of several middle teeth. It was not ascertained during our visits whether cutting edges (sensu Scanlon, 2001) are present. Our photographs, however, show a fine lingual cutting edge on at least the sixth tooth (Fig. 5); on the labial surfaces it is unclear. The anterior teeth are longest, and tooth length decreases posteriorly, gradually and continuously. While all teeth are recurved, the anterior teeth are less so than the posterior ones.
Description of the paralectotype dentary (MNHN.F QU16317): MNHN.F QU16317 is a large (47.0 mm long), nearly complete right dentary (Fig. 6). We consider assignment of this specimen to the same species as the lectotype to be justified by consideration of size (see "Remarks" below). The bone is tall, even at its anterior end, with a depth (measured in the middle of the jaw) to length (measured to the tip of the surangular process) ratio of about 0.22. This value is 37–275% higher than in any other measured constrictor [0.16 in Boa imperator Daudin, 1803a, 0.14 in Chilabothrus striatus (Fischer, 1856), 0.11 in Lichanura trivirgata Cope, 1861a, 0.13 in Calabaria reinhardtii (Schlegel, 1851), 0.08 in Loxocemus bicolor Cope, 1861b, 0.12 in Python bivittatus Kuhl, 1820, 0.14 in Aspidites melanocephalus (Krefft, 1864), 0.15 in Eryx colubrinus (Linnaeus, 1758) (all from Digimorph), 0.15 in Candoia carinata (Schneider, 1801) (specimen MBS 7103)] and could be an autapomorphy. However, it is notably similar to the pathological left dentary of Boa imperator FMNH 31182 (Digimorph), and ratios presented by Hoffstetter (1959: Fig. 2) suggest that other constrictors like the Madagascan Sanziniidae Romer, 1956, and the Neotropical Eunectes Wagler, 1830, may be similar.
Tooth morphology is difficult to compare with the paralectotype maxilla, because the anterior teeth in the dentary are poorly preserved. The fourth tooth from the rear is short, but its degree of curvature is not incompatible with that of the maxillary teeth. The mesial-most tooth locus is offset, directed more medially than dorsally, and the parapet of the dentary is reduced. This feature is also found in at least some Boa spp. (CM 145311), Ungaliophis continentalis Müller, 1880 (data from Digimorph), but it is apparently variable; for instance, it is found only on the left side of Boa imperator (data from Digimorph). It was not found in other Digimorph scans of constrictors. In Boa imperator, this appears to be compensated by a more recurved tooth, so that the tooth itself is not directed medially.
The specimen preserves 18 tooth positions and six partial teeth, only one of which is close to complete. A single large mental foramen opens at the level of the fourth tooth position. The lateral surface of the bone is nearly vertical, curving medially only near the ventral margin. The ventral margin runs parallel to the dorsal margin in the posterior two-thirds of the bone, but beneath the fifth tooth there is an inflection point where the margin runs abruptly more dorsally. The bone is thickened along this ventral rim anteriorly. The surangular notch is deep, separating the dentigerous posterodorsal process from the posteroventral process and exposing the intramandibular septum laterally, as in Boa and Python. Unlike in most extant boids, the posterodorsal process is much more posteriorly extensive than the posteroventral process; in this respect it is more similar to certain pythonids [e.g., Python bivittatus, Morelia riversleighensis (Smith and Plane, 1985) (see Scanlon, 2001:fig. 4)] and Xenopeltis unicolor Reinwardt in Boié, 1827.
The Meckelian groove is open throughout the length of the dentary. The suprameckelian lip is tall; anteriorly it is somewhat concave in transverse section, posteriorly somewhat convex. The posterior prong of suprameckelian lip seen in the paralectotype MNHN.F QU16317, termed the posteroventral prominence by Scanlon (2001), is also present in all suitably prepared (disarticulated) constrictors; in extant taxa it is braced medially by the splenial and coronoid and a facet is observed in this position in this fossil specimen.
Description of other vertebral material from Quercy—intracolumnar variation (Figs. 7, 8, 9, 10, 11, 12, 13, 14and 15): Two specimens (NHMW 2019/0032/0004 and NHMUK PV R 428(1)) could be securely identified as anterior trunk vertebrae (Fig. 7). They are smaller than most mid- and posterior trunk vertebrae (CL around 9.0 mm). The neural arch is vaulted (vaulting ratio = 0.50) but not in an exceeding degree for a vertebra in this position. The neural spine is slightly taller than in the lectotype, inclining posteriorly and broadening slightly distally. The zygosphene is relatively thick in anterior view, though not as thick as in the succeeding trunk vertebrae. Its width exceeds that of the cotyle on NHMW 2019/0032/0004 (as on the lectotype and all other vertebrae assigned to the species). The hypapophysis is relatively thick and long; in NHMW 2019/0032/0004 its terminus is broken but seems to not extend beyond the condyle (Fig. 7a–f), but in NHMUK PV R 428(1) it is fully preserved, is large and exceeds posteriorly the level of the condyle in lateral view (Fig. 7g–k). The centrum is widened anteriorly, though not to the same extent as in succeeding mid-trunk vertebrae. The diapophysis and parapophysis are more distinct than in the lectotype. Cotyles and condyles are nearly circular.
Mid-trunk vertebrae are all wider than long (Figs. 8, 9, 10, 11, 12, 13). Although sample size is small, they apparently show a relatively high vaulting ratio (0.39–0.49), the lectotype being at the low end of this range. The zygosphene is extremely thick in all specimens. The largest referred specimen is MNHN.F QU16319, with centrum length near 19 mm (Fig. 8); the lateral asymmetry of the zygosphene in this specimen is presumably due to deformation or pathology. A small tubercle around the mid-level of the zygosphene is usually present (e.g., lectotype MNHN.F QU16318). The zygosphene usually has more or less distinct and prominent lateral edges, resembling horns. The cotyle can be rather large and deep. The neural spine can be quite tall and is slightly inclined posteriorly. In the neural canal, a distinct raised area (“subneural process” sensu Auffenberg, 1963; “epapophysis” sensu Holman, 2000) is usually prominent, especially in posterior view. The subcentral ridges are generally laterally convex. Subcentral foramina sometimes present and can be rather large (e.g., NHMW 2019/0032/0003). Paracotylar foramina are absent with the sole exception of NHMW 2019/0033/0003, where a large such foramen lies on the right side of the cotyle (Fig. 11a).
Posterior trunk vertebrae have a much wider haemal keel (e.g., MNHN.F QU16345; NHMW 2019/0032/0002). In one large, but eroded, posterior trunk vertebra (NHMW 2019/0032/0002), the centrum is much compressed anteroposteriorly (Fig. 14). Subcentral foramina are usually present. The zygosphene is still thick in vertebrae of this region. The neural arch is less vaulted than in mid-trunk vertebrae. Neural spine foramina (sensu Georgalis & Scheyer, 2019a) can be also present here, as seen on the left side of NHMW 2019/0032/0002.
The only known probable cloacal vertebra is NHMUK PV R 2799(1) (Fig. 15). Judging from the wide and the relatively dorsoventrally high haemal keel, it seems that it probably pertains to the anterior cloacal region (or possibly the posterior-most trunk). Most of the lymphapophyses (only the left is partially preserved) is broken in this specimen. The zygosphene is here also rather thick (Fig. 15a). The interzygapophyseal constriction is shallow. Prezygapophyseal accessory processes are slightly more prominent than elsewhere in the column.
Caudal vertebrae are currently unknown for this species.
Remarks: Filhol (1877a) established Python cadurcensis, providing a description and figures of cranial and postcranial material from the Phosphorites du Quercy. Filhol (1877a, 1877b, 1877c) also tentatively referred the vertebral material that had been previously described and figured by Gervais (1873, 1876) to his new species. He already noted the strong resemblance of this material both to extant Python and to the fossil constrictors from the late Eocene of Switzerland that had been described by Pictet et al. (1855–1857) (Filhol, 1877a, 1877b, 1877c). A few years, later, Rochebrune (1880) established his new genus Palaeopython in order to accommodate this Quercy form, regarding it as distinct enough from the extant species of Python and particularly Python molurus (Linnaeus, 1758) and Python sebae (Gmelin, 1789). The same author later referred to Palaeopython cadurcensis additional specimens (Rochebrune, 1884). Since then, only a limited number of specimens have been referred to Palaeopython cadurcensis, with brief descriptions and few figures (Rage, 1988b, 2013).
A clarification regarding the appropriate authorship and authorship date of Palaeopython cadurcensis is needed. There is a broad consensus that these are attributed to “Filhol, 1877” (e.g., Georgalis & Scheyer, 2019a; Kuhn, 1939b; Rage, 1974, 1978, 1984b; Schaal, 2004; Szyndlar & Schleich, 1993; Wallach et al., 2014). In fact, the name Python cadurcensis appeared earlier. Specifically, 1 year prior to his three 1877 (a, b, c) papers, Filhol (1876b:28) mentioned the presence of vertebrae resembling those of the extant genus Python (“vertèbres … appartenant à des Pythons de grande taille”), originating from different localities within the Phosphorites du Quercy (“J’ai eu de Caylux et d’autres gisements de phosphorite”). Filhol (1876b) further stated that these Quercy vertebrae were different from the fossil “Python-like” vertebrae described by Pictet (1855–1857) from the Eocene of Switzerland, though he did not mention even a single feature that could differentiate the Swiss and French forms. Filhol (1876b:28) additionally referred a dentary “supporting” 17 teeth to the same species (“une demi-mâchoire inférieure supportant dix-sept dents qui provient de la même espèce”). In these sentences discussing this vertebral and dentary material, Filhol (1876b) applied no binomen to characterize these specimens, though in the next paragraph of the same page, he introduced the name Python cadurcensis for the first time (“Ainsi, à l’époque miocène inférieure, il existait dans le centre de la France une faune de Reptiles ayant … et le Python cadurcensis, des affinités fort remarquables avec la faune africaine actuelle”) (Filhol, 1876b:28). He therefore made only a simple mention of the name Python cadurcensis in his biogeographic conclusions (Filhol, 1876b), without a direct referral of the material, although of course it is clear that with that name, he was almost certainly referring to the snake material he was discussing in the above paragraph.
There are three major problematic issues regarding this initial introduction of the name Python cadurcensis by Filhol (1876b) in the literature: (i) there was no description of the vertebrae from Quercy, whereas the author in fact based his taxonomic identification on this material; (ii) the fact that Filhol (1876b) provided the number of teeth (17) supported by the dentary could comply as a minimum description for publications prior to 1931 (ICZN, 1999:Article 12.1), even though this is a rather trivial statement, but he treated this specimen only as a “referred” one (“qui provient de la même espèce”) and did not base his taxonomic identification on this element; and (iii) there is no direct association of the name Python cadurcensis with any specific material, the name appearing only in the next paragraph concerning biogeography. For all these reasons, we consider that, according to ICZN (1999:Article 12.1) for species named prior to 1931, Filhol (1876b) did not make the name available for nomenclatural purposes and Python cadurcensis Filhol 1876b should be considered a nomen nudum.
The name Python cadurcensis appeared in the following year in three practically indistinguishable publications by Filhol (1877a, 1877b, 1877c), where the author formally described and figured vertebral material, a maxilla, and a dentary. The last specimen is probably the same one mentioned in his short note of the previous year (Filhol, 1876b), taking into consideration the fact that it bears 17 tooth positions. It is not feasible to determine which of the three 1877 papers was published first and should thus have priority and be attributed the authority of the species. The paper of Filhol (1877b) is in fact the published thesis of that author incorporating practically the whole text of Filhol (1877a, 1877c) plus the previously published descriptions of some Quercy mammals (Filhol, 1876a). All subsequent authors dealing with the taxon Palaeopython cadurcensis have attributed authorship to Filhol, 1877a, i.e. his paper in the “Annales des Sciences géologiques” (Georgalis & Scheyer, 2019a; Georgalis et al., 2020c; Kuhn, 1939b; Lydekker, 1888a, 1888b; Rage, 1974, 1978, 1984b, 1988b; Rochebrune, 1880, 1884; Szyndlar & Rage, 2003; Szyndlar & Schleich, 1993; Wallach et al., 2014). Curiously, certain of these authors (Rage, 1974, 1984b, 1988b; Szyndlar & Rage, 2003; Wallach et al., 2014) also attributed authorship to Filhol (1877a), but in their reference list they cited the journal name as “Annales des Sciences géologiques” but the number of pages [1–561 (or 1–562 in Wallach et al. 2014)] is identical to the one of Filhol (1877b). In the absence of more precise data about the exact publication dates of the 1877 papers of Filhol (1877a, 1877c; only for the 1877b is a date of publication, 18 July 1877, provided), we follow the prevailing view of the ophidian literature and attribute authorship to Filhol, 1877a (see ICZN, 1999:Article 21.3, for dates of publication incompletely specified).
There is further ambiguity about the exact type series of this species. Filhol (1877a) stated that there were multiple vertebrae found (in different localities within Quercy), corresponding to different size ranges but with similar morphology (“Les vertèbres que j'ai réunies ont des tailles diverses et elles possèdent toutes les mêmes caractères”), providing measurements for the largest one. Filhol (1877a, 1877b, 1877c) did not designate any of the specimens as the holotype, so by definition all specimens he mentioned in this text (and not only the ones he described or figured) are the syntypes of Python cadurcensis (ICZN, 1999:Article 73.2 and Recommendation 73F). The identity and number of the cranial syntypes are clear, though how many and which vertebrae are the syntypes of this species cannot be evaluated with certainty. The figures of Filhol (1877a) clearly show a maxilla (MNHN.F QU16321; Filhol, 1877a:figs. 435, 436), a dentary (MNHN.F QU16317; Filhol, 1877a:figs. 437, 438), a string with five articulated trunk vertebrae (MNHN.F QU16318; Filhol, 1877a:figs. 441, 442), plus one other single trunk vertebra depicted in ventral (Filhol, 1877a:fig. 439), anterior (Filhol, 1877a:fig. 440), dorsal (Filhol, 1877a:fig. 443), and posterior views (Filhol, 1877a:fig. 444). To make things even more complicated, there are erroneous numbers in the figure captions of the publication of Filhol (1877b), i.e. “Figs. 436, 437” for the maxilla, “Fig. 442, 443” for the dentary, and “Fig. 440 à 445” for the vertebrae, although in fact the actual figure numbers are identical with those of Filhol (1877a). In the figure caption, Filhol (1877a) simply stated “Fig. 439 à 444. Vertèbres de Python cadurcensis”, so it cannot be ascertained, just from the published figures, whether the “single” vertebra was indeed an isolated one or its various views corresponded to certain vertebrae from the figured articulated ones. Judging from the published figures of Filhol (1877a), we tend to regard this “single” isolated trunk vertebra as indeed a distinct, different specimen from the other syntype with the five articulated trunk vertebrae (MNHN.F QU16318). Our personal observation of the string of the five articulated trunk vertebrae (MNHN.F QU16318; herein designated as the lectotype) at MNHN reveals that the anterior view of its first vertebra and the posterior view of its last vertebra are much different from Figs. 440 and 444, respectively, of Filhol (1877a), thus supporting our view that Figs. 439, 440, 443, and 444 do not pertain to the former specimen; for instance, in MNHN.F QU16318 the zygosphene is much more concave. Additionally, we also tend to consider that the four views of the “isolated” trunk vertebra of Filhol (1877a) all depict a single specimen, as it can be suggested by the distinct postzygapophyseal tubercles that are present in the ventral and dorsal views (his figs. 439 and 443, respectively) and the similar degree of the vaulting of the neural arch, the thickness of the neural spine and the haemal keel, and the overall shape of the paradiapophyses, as observable in the anterior and posterior views (his figs. 440 and 444, respectively). Rage (1984b) considered that this latter single vertebra “syntype” specimen was the isolated mid-trunk vertebra MNHN.F QU16319 (formerly MNHN QU 319), which, however, is a different specimen that was originally figured by Rochebrune (1884:pl. 2.3a–3b) and was not figured by Filhol (1877a, 1877b, 1877c). On the museum label of MNHN.F QU16319, it is stated that this specimen is a syntype and figured in Rochebrune’s (1884) work. This specimen (MNHN.F QU16319) is apparently identical to that shown in Rochebrune’s (1884) figure 3a and 3b, albeit mirrored. Indeed, it is characterized by a peculiar asymmetry, skewed to the right in anterior view, particularly the neural arch and the zygosphene, whereas the same features are visible on the left side of the figured specimen in Rochebrune’s (1884) lithograph; the fact that the figure of the publication represents a mirror image of the actual specimen was a common practice for lithographs in nineteenth century’s papers (see also Georgalis, 2017; Georgalis et al., 2020b). Wallach et al., (2014), however, erroneously stated that this vertebra is the same single vertebra that was figured by Filhol (1877a, 1877b, 1877c: figs. 439–440, 443–444), but that figure definitely corresponds to a different specimen. From the text of Rochebrune (1884), we find no indication that the vertebra he figured (MNHN.F QU16319) was part of the syntype series of Python cadurcensis; to the contrary it seems that it was a more recent find and not available at the time of the publication of Filhol (1877a, 1877b, 1877c), as Rochebrune (1884:150) stated that the fossil snake material described in that paper was found later than his 1880 work (“Récemment, M. le Dr Filhol a bien voulu nous confier l'étude des restes découverts par lui depuis notre publication. Les nouveaux documents, que nous devons à la gracieuse obligeance de notre savant confrère…”), and few pages later, in the description of his new Palaeopython cadurcensis material, he distinguished the new vertebrae from the ones previously published by Filhol (“Nos vertèbres, comme celles de M. le Dr Filhol, disions-nous…”) (Rochebrune, 1884:154). As such, contrary to the opinion of Rage (1984b), the vertebra MNHN.F QU16319 cannot be considered a syntype and should be excluded from the type series of Palaeopython cadurcensis; it is here simply considered as a referred specimen of this species (Fig. 8). Accordingly, the only vertebral material in the original syntype series which can confidently be identified are the five articulated trunk vertebrae (MNHN.F QU16318; Fig. 4). The whereabouts of the “single” mid-trunk vertebra figured by Filhol (1877a:figs. 439, 440, 443, and 444) are unknown, and this specimen could not be located in the collections of MNHN.
Finally, we consider the referral of the maxilla (MNHN.F QU16321) and dentary (MNHN.F QU16317) syntypes (now paralectotypes) to the same species as the lectotype string of five vertebrae (MNHN.F QU16318). To assess whether their assignment to the same species as the lectotype is justified, we measured the centrum length of a mid-trunk vertebra and dentary length in 1 extinct and 13 different extant taxa of Constrictores (see Appendix 3) and conducted major axis regression (Fig. 16). According to the model, an animal producing the lectotype with CL = 12.1 mm is expected to have a dentary 47.7 mm in length. This value is closely comparable to the paralectotype dentary MNHN.F QU16317 (47.0 mm). Since no other known constrictor from Quercy is known to be so large, and potentially co-occurring Phosphoroboa filholii comb. nov. is considerably smaller and also has other dentigerous bones that can reasonably be assigned to it (see that respective entry below), we consider the assignment of MNHN.F QU16317 to be justified. Furthermore, in the paralectotype maxilla (MNHN.F QU16321) the teeth have similar dimensions to those of the dentary, so that the same justification can be made here. The three specimens are therefore reasonably considered to pertain to the same species, Palaeopython cadurcensis.
At the same time there is no reason to believe that the cranial and vertebral syntypes available to Filhol (1877a, 1877b, 1877c) come from the same individual, and it also cannot be fully excluded that these syntypes are a chimaera representing more than one species (though unlikely, as our model in the previous paragraph indeed favours conspecificity). Therefore, in order to maintain nomenclatural stability, it is necessary to designate one of these syntypes as the lectotype of Palaeopython cadurcensis. We designate the five articulated trunk vertebrae MNHN.F QU16318 as the lectotype of this species, taking into consideration that it possesses important diagnostic features and can be directly compared with other fossil Constrictores (see also in Discussion for our rationale on selecting vertebral material as the lectotype). Accordingly, the maxilla MNHN.F QU16321, the dentary MNHN.F QU16317, and all other not currently located or identified vertebrae mentioned by Filhol (1877a) should be considered as paralectotypes of Palaeopython cadurcensis.
It further seems probable that these “Python-like” vertebrae and dentary that Filhol (1876b, 1877a, 1877b, 1877c) discussed were relatively newly found, as there is not a single mention of that material in his previous publications on Quercy fauna (Filhol, 1873a, 1873b). In contrast, the large lizard from Quercy (Palaeovaranus cayluxi) was mentioned in both 1873 and 1876 papers (Filhol, 1873b, 1876b; see Georgalis, 2017).
In addition to the above-mentioned material, there are other purported occurrences of this species that have never been described or figured and therefore are not treated here (see Table 3). We further consider that the four vertebrae from the late Eocene of Hordle Cliff, UK, referred to Palaeopython cadurcensis by Holman et al. (2006) should not be referred to this species. This material was only briefly described and never figured, but judging from the rather small dimensions of the English material provided by Holman et al. (2006), we think it more plausible that the material pertains to another taxon, potentially to Paleryx. Material from the late Eocene of mount Mormont, Vaud, Switzerland was originally referred to Python sp. by Pictet et al. (1855–1857) but subsequently referred to Palaeopython cadurcensis by Filhol (1877a, 1877b, 1877c); pending a redescription of the Mormont material, we treat that occurrence as Palaeopython sp.
It is further worth highlighting that Palaeopython cadurcensis is the largest snake recovered from the Phosphorites of Quercy (Rage, 1974) and furthermore the largest member of the genus Palaeopython. Among European snakes, the size of Palaeopython cadurcensis would only be surpassed by species of the aquatic genus Palaeophis. Notably, the trunk vertebra MNHN.F QU16319 (Fig. 8) represents one of the largest constrictor vertebrae known, surpassing true giants in other continents, such as Chubutophis grandis Albino, 1993 from the early Eocene of Argentina (Albino, 1993).
Palaeopython cf. cadurcensis
Material: A trunk vertebra (NHMW 2019/0032/0001) and a trunk vertebra (probably collections of BSPG, currently lost) figured by Zittel (1887–1890:fig. 559).
Locality: The NHMW specimen originates from an imprecise locality (probably middle or late Eocene), Phosphorites du Quercy, France; Zittel’s (1887–1890) material originates from Escamps (late Eocene, MP 19), Phosphorites du Quercy, Lot, France.
Remarks: NHMW 2019/0032/0001 resembles Palaeopython cadurcensis in many respects, especially in the thickness of the zygosphene and the overall shape and size (Fig. 17). Nevertheless, there are certain differences, such as the more vaulted neural arch (vaulting ratio 0.55) and the exceedingly high neural spine. The vertebra can be further differentiated from Palaeopython helveticus by its more massive zygosphene and more massive paradiapophyses.
Two additional vertebrae, figured by Zittel (1887–1890:fig. 559) and De Stefano (1905:pl. IV.12; originally referred to Paleryx rhombifer), are characterized by high neural arches and bear a resemblance to NHMW 2019/0032/0001. We tentatively place the Zittel’s specimen here as well.
Whether this taxon represents a distinct rather large form in the area of Quercy or simply some extreme variation of Palaeopython cadurcensis cannot be resolved for the moment.
Palaeopython ceciliensis Barnes, 1927
Type material: GMH Ce I-2978-1926 (holotype), an incomplete mid-trunk vertebra, missing its posteriormorst portion (posterior portion of ventral surface of centrum, condyle, both postzygapophyses, posterior part of neural arch, most parts of prezygapophyses, and most part of neural spine) (Barnes, 1927, pl. I.7; Georgalis & Scheyer, 2019a, fig. 15h; Figs. 18, 19, Additional file 1: Model 1).
Type locality: Quarry Cecilie I, Geiseltal, Saxony-Anhalt, Germany; late early or middle Eocene.
Referred material and range: Known exclusively from the late early or middle Eocene of its type area, Geiseltal, Saxony-Anhalt, Germany. Found in the following quarries [in stratigraphical order]: quarry LII; quarry XXII; quarry XLI; quarry XXXV; quarry XXXVI; quarry Cecilie IV; type locality, quarry Cecilie I; quarry Cecilie III; imprecisely known quarry(ies).
Referred material discussed herein (all from quarries of Geiseltal [in stratigraphic order]) (Figs. 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30and 31): quarry LII (Fig. 20): a mid-trunk vertebra (GMH LII-37-1971); quarry XXII (Fig. 21): two trunk vertebrae (GMH XXII-39-1965) and a small-sized skeleton on matrix (GMH XXII-556-1965); quarry XLI (Fig. 22): three mid-trunk vertebrae (GMH XLI-60a-1966, GMH XLI-60b-1966, and GMH XLI-60c-1966), and one anterior trunk vertebra (GMH XLI-314b-1968); quarry XXXV (Fig. 23): two articulated anterior mid-trunk vertebrae (GMH XXXV-485c-1963), a mid-trunk vertebra (GMH XXXV-404-1963), a mid-trunk vertebra (GMH XXXV-386-1963 [tentative referral]), and two posterior trunk vertebrae (GMH XXXV-485a-1963 and GMH XXXV-485b-1963); quarry Cecilie XXXVI (Fig. 24): articulated trunk vertebrae and remains of ribs on matrix (GMH XXXVI-41-1962); quarry Cecilie IV (Figs. 25, 26): several articulated and disarticulated trunk vertebrae and ribs on a matrix (GMH Ce IV-5854-1933 [several], GMH Ce IV-5849-1933[several], GMH Ce IV-5850-1933[several], GMH Ce IV-5858-1933[several], and GMH Ce IV-5858b-1933[several]), and a fragmentary skeleton with several disarticulated vertebrae and ribs (GMH Ce IV-5857-1933); quarry Cecilie I (type locality) (Figs. 27, 28): an anterior/mid-trunk vertebra (GMH Ce I-Y-42-1926), 19 large mid-trunk vertebrae and three tiny vertebral fragments (GMH Ce I-5837-1926; originally described as Palaeovaranus by Weigelt, 1929:fig. 2; Fig. 27), a mid-trunk vertebra (GMH Ce I-5834a-1926), and several trunk vertebrae (GMH Ce I-5835d-1929[tentative], GMH Ce I-5835e-1929, GMH Ce I-5835f-1929, GMH Ce I-5835g-1929); quarry Cecilie III (Fig. 29): an articulated 2.3 m skeleton with crushed skull and 243 vertebrae [GMH Ce III-5867-1933; figured in Kuhn (1939a:pl. I.3 and II.3), Weigelt (1934:pl. IX.1), and Krumbiegel et al. (1983:fig. 114); Fig. 29; probable referral]; unknown quarry(ies) (Figs. 30, 31): several large trunk vertebrae (GMH Y-38) and four strings of articulated trunk vertebrae (GMH Y-40-1964).
Diagnosis: Palaeopython ceciliensis can be referred to the genus Palaeopython by the presence of the full list of characters given for that genus. It can be differentiated from all other species of Palaeopython by the unique combination of the following characters: size moderate with CL around 10 mm (but rarely reaching up to 13 mm), neural arch of mid- and posterior trunk vertebrae less vaulted, neural spine of posterior mid- and posterior trunk vertebrae significantly lower than that of the mid-trunk ones, and median tubercle on the zygosphene in anterior view absent or only weakly developed. Palaeopython ceciliensis can be further differentiated from the co-occurring in Geiseltal Eoconstrictor cf. fischeri by its much larger size, the much thicker zygosphene, the relatively much lower neural spine in the posterior trunk vertebrae, the absent or poorly developed median tubercle on the zygosphene, and the more massive paradiapophyses.
Description of the holotype (GMH Ce I-2978-1926): This specimen is an incomplete mid-trunk vertebra, missing the posterior part of the centrum along with the condyle, posterior part of neural arch, most of neural spine and zygantrum, and both postzygapophyses (Figs. 18, 19, Additional file 1: Model 1). In anterior view (Figs. 18a, 19a), the zygosphene is extremely thick and slightly wider than the cotyle. The lateral edges of the zygosphene are slightly upturned, making the roof concave. A small tubercle is present on the anterior surface of the zygosphene at around mid-height. The cotyle is slightly depressed and quite deep. The neural canal is nearly triangular. No paracotylar foramina are present, but paracotylar fossae do exist and are relatively deep. The parapophyses project ventrally below the centrum, but not greatly so, consistent with the position of the vertebra in the column. Little can be said about the posterior view, including the shape of the condyle or the zygantrum (Figs. 18b, 19i). Nevertheless, a small median impression at the ventral margin of the zygantrum confirms that the succeeding vertebra also possessed a median tubercle on its zygosphene. A pair of large endozygantral foramina (sensu Head, 2005) is present. In dorsal view (Figs. 18c, 19e), the vertebra is wider than long. The anterior margin of the zygosphene is slightly concave and the lateral edges are prominent. The articular surfaces of the zygosphene are steep. The prezygapophyseal articular facets are missing but seem to have been large and broad. The anterior edge of the neural spine starts well behind the anterior margin of the zygosphene; it rises at first gradually as a thin, arching flange before the thick (broken) part of the neural spine proper develops in the posterior half of the vertebra. The neural arch width (NAW) is 15.5 mm. In ventral view (Figs. 18d, 19f), the haemal keel grew relatively sharp by mid-length. Relatively sharp subcentral ridges give the centrum a triangular shape. A rather small subcentral foramen is present on the right side of the haemal keel (the left side is unknown). The paradiapophyses are massive and elliptical. In lateral view (Figs. 18e, f, 19g, h), the vertebra is taller than long. The zygosphenal articular facets are massive. Lateral foramina are present just posterior to the diapophysis and below the interzygapophyseal ridge. The diapophyses and parapophyses are distinct but are not divided.
Description of additional vertebrae—intracolumnar variation (Figs. 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30and 31): Anterior trunk vertebrae are characterized by the presence of a hypapophysis instead of a haemal keel (Fig. 30). The shape of the hypapophysis in lateral view varies, depending (at least in part) on the position of the vertebra in the anterior trunk series, being either long and thin (e.g., GMH Y-38b) or (more usually) relatively anteroposteriorly long (e.g., GMH Y-38d; Fig. 30e). Succeeding anterior trunk or anterior mid-trunk vertebrae possess a rather vaulted neural arch and a tall neural spine (e.g., GMH Y-38j).
Mid-trunk vertebrae can achieve a large size, with CL approaching or even surpassing 10 mm (e.g., GMH LII-37-1971; GMH XLI-314b-1968); the largest measurable specimen has CL = 13 mm. The zygosphene is rather thick in anterior view (e.g., GMH Ce I-5834a-1926; GMH Ce I-Y-42-1926; GMH LII-37-–1971). The zygosphene is wider than the cotyle. There is usually no median tubercle on the zygosphene, although rarely this structure is weakly present (e.g., GMH Ce I-Y-42-1926). The zygosphene usually possesses two distinct lateral lobes, which can be rather prominent in certain large-sized specimens, resembling “bull’s horns” in anterior view (e.g., GMH LII-37-1971, GMH Ce I-5834a-1926, GMH Y-38a). They are sometimes upturned (e.g., GMH Ce I-5837b-1926), sometimes not (e.g., GMH LII-37-1971). The cotyle is usually large, deep, and slightly elliptical, though in certain specimens it is distinctly elliptical (e.g., GMH Ce I-5837f-1926). Prezygapophyseal articular facets are large and can be rather pointed in certain specimens (e.g., GMH Y-42-1926). Prezygapophyseal accessory processes are present, although usually they are small. The prezygapophyses are usually inclined. The distal edges of the prezygapophyseal articular facets are rather pointed. The paradiapophyses are massive. The centrum is much widened anteriorly. The haemal keel is thick and sharp and extends along nearly the entire length of the centrum. Subcentral foramina usually occur, though in some specimens they are absent (e.g., GMH Ce I-Y-42-1926). The neural spine is thick. The neural spine in mid-trunk vertebrae is relatively tall; however, posterior mid- and posterior trunk vertebrae have instead a much lower neural spine (e.g., GMH LII-37-1971, GMH Y-38f, and GMH Y-38i; small articulated skeleton on matrix GMH XXII-556-1965).
Posterior trunk vertebrae have a wide haemal keel (Fig. 31). In certain cases, the haemal keel expands anteriorly and posteriorly (e.g., GMH LII-37-1971; Fig. 20d). The zygosphene is still thick and prominent also in this portion of the vertebral column, and some specimens possess the prominent lateral lobes in dorsal view (e.g., GMH XXXV-485b-1963). Paracotylar foramina are absent in all vertebrae throughout the column, except for a single large posterior mid- or posterior trunk and an anterior trunk vertebra (GMH Y-38c and GMH Y-38e, respectively), though in all other associated vertebrae of the same collection number [i.e. GMH Y-38 (many vertebrae)], this feature is always lacking.
Remarks: Barnes (1927) established Palaeopython ceciliensis on the basis of an incomplete vertebra from the quarry Cecilie I, in Geiseltal. A couple of years later, additional large snake vertebrae were described from Geiseltal but were originally assigned to the large lizard taxon Palaeovaranus by Weigelt (1929); that material was subsequently referred to Palaeopython ceciliensis (Kuhn, 1939a; see also Georgalis, 2017). Kuhn (1939a) redescribed this species, reassigned it to Paleryx, and also referred to it, among others, a relatively complete, large (2.3 m long) skeleton (GMH Ce III-5867-1933) (Fig. 29). Rage (1984b) disputed the generic attribution of this Geiseltal taxon to Paleryx and followed the original concept of Barnes (1927) in considering it as a member of Palaeopython. Nevertheless, the same author considered that not a lot more can be said about this species, and he was further sceptical of Kuhn’s (1939a) referral of GMH Ce III-5867-1933 to the same taxon, a view subsequently shared by Szyndlar and Böhme (1993).
Palaeopython ceciliensis is the largest snake species in the Geiseltal assemblage. Besides the many disarticulated vertebrae, several articulated and disarticulated skeletons are also known, some of which comprising also skull remains (Figs. 21, 24, 25, 29). Certain specimens achieved truly large size, such as the 2.3 m skeleton GMH Ce III-5867-1933 (Fig. 29) and even larger sizes were probably achieved, as judged by the large size of certain vertebrae, up to 13 mm in CL.
In some vertebrae of Palaeopython ceciliensis, growth rings on the prezygapophyses (e.g., GMH Ce I-5835a-1929, GMH Ce I-5835f-1929) and/or postzygapophyses (e.g., GMH XLI-314b-1968) can be observed. Sometimes this is correlated with evidence of erosion of the bone surface, such as exposure of the zygantrum through the dorsal surface of the neural arch (e.g., GMH Ce I-Y-42-1926). A similar exposure of the zygantrum has been described also in the North American Oligocene species, Coprophis dakotaensis Parris and Holman, 1978, a taxon aptly named for having been recovered from a coprolite, thus suggesting that it arises via etching in gastric fluids (Parris & Holman, 1978). Accordingly, one can hypothesize that crocodylians, the largest reptilian predators in Geiseltal and which are known to possess highly acidic gastric fluids (Fisher, 1981), consumed larger snakes, as extant alligators are known to do in the southeastern United States (e.g., Wolfe et al., 1987). This would be ironic considering that it is large constrictors that consumed smaller crocodylians in Messel (Greene, 1983), and it remains a hypothesis to be tested. Regardless, growth rings can be used to determine the ontogenetic stage of an individual (e.g., Petermann & Gauthier, 2018; Venczel et al., 2015).
The vertebral anatomy of Palaeopython ceciliensis shows resemblance to the other two Palaeopython spp. and we agree with the original taxonomic opinion of Barnes (1927) in placing it in that genus. We further anticipate that the available skull material of this taxon that can be studied with the application of microcomputed tomography, will shed important light on its cranial anatomy and exact phylogenetic relationships.
Palaeopython helveticus Georgalis and Scheyer, 2019a
Taxonomic history: Palaeopython helveticus Georgalis and Scheyer, 2019a (new species).
Type material: PIMUZ A/III 634 (holotype), an anterior mid-trunk vertebra of an adult individual (Rosselet, 1991, fig. 23; Georgalis & Scheyer, 2019a, figs. 5, 6, 15a, 17i–l, supplementary Fig. 1c; Georgalis & Scheyer, 2019b, unnumbered subfig. on top right in fig. 1; Fig. 32).
Type locality: Fissure A, Dielsdorf, Zurich Canton, Switzerland; MP 16–20, late Bartonian–late Priabonian, late middle–late Eocene.
Referred material and range: MP 16–20, late Bartonian–late Priabonian, late middle–late Eocene, Dielsdorf, Fissures A, B, and 2, Zurich Canton, Switzerland (referred material of Georgalis & Scheyer, 2019a).
Diagnosis: See Georgalis and Scheyer (2019a).
Remarks: This taxon was recently described in detail by Georgalis and Scheyer (2019a) on the basis of isolated vertebrae from the late middle–late Eocene of Dielsdorf, Switzerland. These authors also presented 3D surface models and applied micro-CT scanning to vertebrae of Palaeopython helveticus, allowing also an in depth analysis of the ontogenetic variation (Georgalis & Scheyer, 2019a, 2019b). For detailed descriptions of this taxon, we refer the reader to Georgalis and Scheyer (2019a).
Genus Paleryx Owen, 1850
Type species. Paleryx rhombifer Owen, 1850.
Other valid species included: None.
Geographic and stratigraphic distribution: see Table 3.
Diagnosis: Paleryx is assigned to Constrictores on the basis of possessing strongly built vertebrae, the centrum wider than long, the vertebrae taller than long in lateral view, prezygapophyseal accessory processes small, paradiapophyses undivided or only weakly divided, and parazygantral foramina absent. Similar to Pythonoidea in having a palatine foramen and a maxillary process of the palatine located anterior to the pterygoid articulation. Size moderate, with CL maximally around 10 mm; ratio of neural spine length to centrum length < 0.5. Paleryx can be differentiated from all other known European Paleogene Constrictores by the following unique combination of features: ratio of zygosphene width to cotyle width < 1.1 in mid- and posterior trunk vertebrae, generally lower neural spine (especially on posterior trunk vertebrae), and posterior trunk vertebrae with depressed neural arch (vaulting ratio < 0.25). Paleryx can be further differentiated from Palaeopython by its smaller size, much thinner zygosphene without prominent anterolateral horns, deeper interzygapophyseal constriction, and shallower paracotylar fossae.
Remarks: Owen (1850) established the genus Paleryx from the late Eocene of the United Kingdom and noticed vertebral similarities with the extant genera Eryx and Python, as well as the extinct Palaeophis. The supposed close resemblance to Eryx was subsequently followed by other workers (e.g., Seeley, 1885), and Carus (1868) formally placed it into Erycidae. For a long time during the late nineteenth and the twentieth century, Paleryx was considered as the senior synonym of Palaeopython (see “Taxonomic history and history of discoveries of the European large fossil constrictors”). Swinton (1926) reported that his newly erected taxon Daunophis langi from the Pliocene of Myanmar was rather similar to Paleryx, an opinion that could not be confirmed by Szyndlar and Rage (2003). Swinton (1926) also erroneously stated that Paleryx persisted until the early Miocene. Simpson (1933) noted some resemblance of Paleryx with his newly described giant Paleogene Patagonian form Madtsoia bai, especially as far as it concerns the ratio of the width of the zygosphene to the cotyle and also the general proportions of the neural arch and the neural spine. Nevertheless, Simpson (1933) admitted the distinctiveness of Paleryx from Madtsoia. Interestingly, Romer (1956) not only considered Paleryx as the senior synonym of Palaeopython, but he tentatively synonymized with the former genus also Palaelaphis Rochebrune, 1884 (for which he erroneously provided an authorship date of 1886 instead of 1884); however, he provided no justification for such taxonomic assignment. In any case, both species of Palaelaphis (i.e. Palaelaphis antiquus Rochebrune, 1884 and Palaelaphis robustus Rochebrune, 1884) originate from unknown localities within the Phosphorites du Quercy and are currently considered as nomina dubia, with their material representing indeterminate Contrictores (Rage, 1984b; Szyndlar & Rage, 2003).
We here recognize Paleryx as a non-booid snake, comprising a single valid species, Paleryx rhombifer (with Paleryx depressus being its junior synonym). Paleryx is known with certainty only from the late Eocene of England.
Similarly to the case of many other taxa named in the nineteenth century, the genus name Paleryx suffered from important misspellings, such as Palaeryx by Massalongo (1859), Palaeeryx by Palacký (1884), and Paleoeryx again by Palacký (1898).
Paleryx rhombifer Owen, 1850
(= Paleryx depressus Owen, 1850)
Taxonomic history: Paleryx rhombifer Owen, 1850 (new species); Paleryx rhombifer = Palaeopython cadurcensis Lydekker, 1888b (senior synonym); Paleoeryx rhombifer Palacký, 1898 (incorrect spelling of genus name); Paleryx rhombifer = Paleryx depressus Rage and Ford, 1980 (senior synonym).
Type material: NHMUK PV OR 25259 (holotype), an anterior mid-trunk vertebra (probably not anterior trunk vertebra as stated by Rage, 1984b) (Owen, 1850, pl. XIII.29–32; Owen, 1849–1884, pl. 2.29–32; Pictet, 1853–1857, pl. XXX.4–6; Lydekker, 1888a, fig. 55; Benton and Spencer, 1995, fig. 9.7.e; Georgalis & Scheyer, 2019a, fig. 15k; Figs. 33a–d, 34). Cast exists at the collections of MNHN.
Type locality: Hordle (= Hordwell) Cliff, Hampshire, United Kingdom; Headon Hill Formation, Totland Bay Member, Mammal Bed or Rodent Bed, MP 17, late Eocene.
Referred material and range: Late Eocene (MP 18), Headon Hill, Isle of Wight, United Kingdom (referred material of Rage & Ford, 1980); late Eocene (MP 18/19), Fishbourne, Isle of Wight, United Kingdom (referred material of Rage & Ford, 1980); late Eocene (MP 19), type area, Hordle Cliff, Hampshire, United Kingdom (type material of Paleryx depressus Owen, 1850; referred material of Lydekker, 1888a; material referred to Paleryx depressus by Lydekker, 1888a; material referred to Placosaurus margariticeps by Lydekker, 1888a).
Referred material discussed herein (all from Hordle Cliff) (Figs. 35, 36, 37and 38): A mid-trunk vertebra [NHMUK PV OR 25261(1) (lectotype of Paleryx depressus, herein designated)], two posterior trunk vertebrae [NHMUK PV OR 25261(2) and NHMUK PV OR 25261(3) (paralectotypes of Paleryx depressus)], many trunk vertebrae (NHMUK PV OR 32838), several trunk vertebrae [NHMUK PV OR 32840 (originally described as Placosaurus margariticeps by Lydekker, 1888a)], and a left palatine (NHMUK PV R 10907).
Diagnosis: As for the genus Paleryx (see above).
Description of the holotype (NHMUK PV OR 25259): NHMUK PV OR 25259 is an anterior mid-trunk vertebra missing most of the left prezygapophysis and part of the right postzygapophysis (Fig. 34). In anterior view (Fig. 34a), the zygosphene is rather thick, with its dorsal level being significantly wider than its ventral one. The zygosphenal roof is straight. The neural spine is only moderately high. The neural canal is small and semicircular. The prezygapophyses are relatively thick, probably reflecting the occurrence of a small, now eroded prezygapophyseal accessory process; the prezygapophyseal articular facets are only slightly dorsally inclined. The cotyle is nearly circular. No paracotylar foramina are present but paracotylar fossae exist. The paradiapophyses are partly eroded, but it is obvious that they were not divided into diapophyseal and parapophyseal parts. The parapophyseal portions of the paradiapophyses project ventrally below the cotyle. In posterior view (Fig. 34b), the neural arch is moderately vaulted and possesses a distinct angulation at the dorsolateral corner of the zygantrum. The zygantrum is deep. The condyle is almost circular. In dorsal view (Fig. 34c), the zygosphene is almost straight and the two lateral lobes are poorly developed. The neural spine is relatively thick and its base commences posterior to the level of the zygosphenal facets. The prezygapophyseal articular facets are oval and have an oblique long axis. The interzygapophyseal constriction is symmetrically rounded and extends medially past the mid-point of the prezygapophyseal articular facet. In ventral view (Fig. 34d), the centrum is triangular. The haemal keel is sharp, extending along nearly the entire length of the centrum, commencing at the ventral lip of the cotyle and ending at the base of the condyle. The postzygapophyseal articular facets are subtriangular. Two small subcentral foramina are present, at around mid-length on the centrum, situated on each side of the haemal keel. In lateral view (Fig. 34e, f), the neural spine is posteriorly inclined, with parallel anterior and posterior margins and its posterodorsal edge slightly overhanging the posteromedian notch of the neural arch. A pair of tiny neural spine foramina is present. Two small lateral foramina are present, one at each side of the centrum, situated below the interzygapophyseal ridges. The subcentral ridges are almost straight. The haemal keel is tall, especially at its posterior portion, where it resembles a hypapophysis.
Description of additional vertebrae—intracolumnar variation (Figs. 35, 36and 37): In the transition from the anterior trunk vertebrae to the mid-trunk ones, the hypapophysis is substituted by a haemal keel, which gradually diminishes in dorsoventral height more posteriorly in the vertebral column. In mid-trunk vertebrae, the haemal keel takes the shape of a thin longitudinal ridge that crosses the whole midline of the centrum (e.g., Fig. 36d). The zygosphene is thick but can be thin in small vertebrae. There are deep paracotylar fossae next to the cotyle but usually without paracotylar foramina; a notable exception is NHMUK PV OR 32840(1), where a large paracotylar foramen is present on the left side (Fig. 36a). The neural arch is vaulted. The neural spine is thick and moderately high dorsoventrally. The interzygapophyseal constriction is moderately expressed, but in NHMUK PV OR 32840(1) it is rather shallow (like in Palaeopython).
In posterior trunk vertebrae, the neural arch becomes rather depressed and the haemal keel is very wide (e.g., Fig. 35g). One other notable feature of these vertebrae is that the zygosphene is much thinner in comparison with the preceding ones. The neural spine is also thick but becomes significantly dorsoventrally lower; in NHMUK PV OR 32838(1) it overhangs both anteriorly and posteriorly (Fig. 37e). The interzygapophyseal constriction is more prominent.
Cloacal and caudal vertebrae are so far unknown for this species.
Description of cranial material: A single cranial element, the left palatine NHMUK PV R 10907, has been recovered from Hordle Cliff (Fig. 38). Its size is broadly consistent with the overall vertebral dimensions of Paleryx rhombifer, the sole large constrictor recognized from Hordle as well as the whole British Eocene, and accordingly, we tentative assign it to the same species. The element has seven tooth positions, and the 5th and 7th teeth are complete. Tooth size, to judge by the tooth bases, appears to decrease from front to back. It is unlike that of any known booid in having a maxillary process located well anterior of the pterygoid articulation, namely at a level between the 5th and 6th teeth (Fig. 38c). The process is finger-like, tapering as it curves anteriorly. The broad maxillary facet is located on its anteroventral surface. A palatine foramen is present, as in Pythonoidea but not Booidea, which is noteworthy for extending a considerable distance along the lateral surface of the bone anterior to the maxillary articulation in a channel that is overhung dorsally and partly dorsolaterally (Fig. 38a, d, e).
The choanal process appears to have a broad base, but its original extent is uncertain; dorsal to it are elaborations of unknown significance. The pterygoid facet is short but deep, approximately one tooth base in length; at depth it is acutely angled in cross-section, implying a dorsally sharp palatine process of the pterygoid. Anteriorly on the lateral side of the bone is a horizontal ridge of unknown significance. The bone tapers anteriorly in height. Its medial surface is gently concave.
Remarks: Paleryx rhombifer and Paleryx depressus are the first fossil Constrictores described worldwide (Owen, 1850). The original material of Paleryx rhombifer and Paleryx depressus described by Owen (1850) was initially part of the collection of the Marchioness of Hastings, but they landed in the collections of the British Museum in 1855. Lydekker (1888a) referred to both Paleryx rhombifer and Paleryx depressus numerous vertebrae from England and Quercy, as he considered that the Palaeopython material from the latter French region in fact belonged to the British genus. However, he did not figure any of the referred specimens from the United Kingdom; one of them (NHMUK PV OR 32838(unknown sub-number)) was figured for the first time almost one century later (Rage, 1984b:fig. 13a). Notably, Lydekker (1888a) tentatively also referred to Paleryx rhombifer and Paleryx depressus trunk vertebrae from the early Miocene (MN 1/2) of Saint-Gérand-le-Puy, France, thus expanding significantly the distribution of Paleryx into the Neogene. However, it seems more probable that this material pertains to the “erycine” Bransateryx Hoffstetter and Rage, 1972, which is found in the same locality (Hoffstetter & Rage, 1972; Rage, 1974). Rage and Ford (1980) and later Rage (1984b) considered that the differences among Paleryx rhombifer and P. depressus (vertebrae more depressed, haemal keel thicker and less produced, zygosphene thinner, neural spine lower and more posteriorly inclined) could simply be attributed to intracolumnar variation, as indeed the types pertained to different regions within the vertebral column. This taxonomic opinion has been generally followed ever since, in the few subsequent mentions of these English constrictors (e.g., Benton and Spencer, 1995; Wallach et al., 2014).
Lydekker (1888a) also referred several vertebrae from the late Eocene of England to Placosaurus margariticeps (Gervais, 1876), a glyptosaurine lizard from Quercy, originally tentatively described as a species of the genus Varanus Merrem, 1820. However, our direct observation of the material confirms that these vertebrae (NHMUK PV OR 32840) can in fact be referred to Paleryx rhombifer based on their possession of the diagnostic characters given above.
The palatine from Hordle Cliff described above is the first cranial element attributed to Paleryx rhombifer. The maxillary process is reminiscent in shape and in its position anterior to the palatine articulation to that of a previously published palatine (UM PRR 2001) from the late Eocene (MP 17) of Perrière in Quercy, which was referred to “Palaeopython sp.” by Rage (1984b), but the latter specimen appears to lack a palatine foramen (see Rage, 1984b:fig. 12C). As mentioned above, the presence of a palatine foramen would allow rejection of booid affinities for Paleryx rhombifer and make pythonoid ones plausible, especially given the recognition of the stem pythonid Messelopython freyi in the Eocene of Europe (Zaher & Smith, 2020). However, as this feature also does exist in other snakes outside Constrictores (e.g., Lee & Scanlon, 2002), we refrain from formally referring Paleryx rhombifer to Pythonoidea, pending the recovery of more cranial material from the type locality.
See also Paleryx depressus below for further details about the synonymy of that species with Paleryx rhombifer, and the description of its type specimens.
Genus Incertae Sedis
“Palaeopython” neglectus Rochebrune, 1884
Taxonomic history: Palæopython neglectus Rochebrune, 1884 (new species); Paleryx neglectus De Stefano, 1905 (new combination); Palaeopython neglectus Rage, 1984b (lectotype designation); [Palaeopython neglectus] Rage and Augé, 1993 (nomen dubium).
Type material: MNHN.F QU16326 (lectotype), a trunk vertebra (Rochebrune, 1884, pl. II.5a, II.5b; Figs. 39b, c, 40); MNHN.F QU16325 (paralectotype), five articulated trunk vertebrae (Rochebrune, 1884, pl. II.5; Figs. 39d, 41); MNHN.F QU16324 (paralectotype), a “mummy” (Rochebrune, 1884, pl. I.1, I.1a; Figs. 39a, 42).
Type locality: Unknown exact locality, Phosphorites du Quercy, France; probably late middle or late Eocene.
Previously referred material and range: Probably late Eocene, type area, Quercy, France (referred material of De Stefano, 1905).
Referred material discussed herein (all from the Phosphorites du Quercy, imprecise localities near Mouillac) (Figs. 44, 45): 12 isolated trunk vertebrae (one anterior, nine mid- and two posterior trunk vertebrae; YPM-VPPU 29856–YPM-VPPU 29859 and YPM-VPPU 30360–YPM-VPPU 30367) and two articulated mid-trunk vertebrae (YPM-VPPU 30368).
Diagnosis: “Palaeopython” neglectus is assigned to Constrictores on the basis of possessing strongly built vertebrae, the centrum wider than long, the vertebrae taller than long in lateral view, prezygapophyseal accessory processes small, paradiapophyses undivided or only weakly divided, and parazygantral foramina absent. “Palaeopython” neglectus differs from Palaeopython spp. in being smaller, in having a deeper, symmetrical interzygapophyseal constriction, and in having a neural spine that extends well onto the roof of the zygosphene. “Palaeopython” neglectus differs from Paleryx and Phosphoroboa gen. nov. in having a longer neural spine and a wider zygosphene on mid- and posterior trunk vertebrae.
Description of the lectotype (MNHN.F QU16326): The lectotype is an almost complete trunk vertebra, with the edges of its prezygapophyses being eroded (Fig. 40). The zygosphene is broad, with moderately well-developed lateral lobes. A small median tubercle is found just above the neural canal (Fig. 40a). The distal ends of the prezygapophyses are eroded, but the articular surfaces appear to be nearly horizontal and elevated slightly above the top of the centrum. The neural spine is anteroposteriorly long, beginning as a short, thin, nearly vertical lamina at the level of the middle of the zygosphenal articular facet. It is slightly posteriorly inclined, with parallel anterior and posterior margins, and its dorsal edge appears rounded. Since that edge is capped by finished bone, it is unlikely that the rounding is purely artifactual. The cap also suggests that the individual was ontogenetically advanced.
The neural arch has low, elongate zygantral mounds on its posterior half. The laminae of the posterior margin of the neural arch are broadly rounded and do not extend beyond the postzygapophyses. Between them is a deep posteromedian notch that exposes the neural canal in front of the condyle. In posterior view, the neural arch is vaulted (Fig. 40b). Its dorsal edge appears curiously downturned along the posterior margin. A field of tiny foramina is present lateral to the zygantrum on either side. At depth the zygantrum is pierced by a pair of foramina.
The interzygapophyseal constriction is rounded and appears to have been moderately deep (Fig. 40c), although the erosion of the prezygapophyses makes explicit comparison to other taxa difficult. A pair of lateral foramina is present well below the interzygapophyseal ridges. The centrum is triangular and the subcentral ridges straight (Fig. 40d). The centrum does not widen as much anteriorly as in Palaeopython. There is a sharp haemal keel that grows in prominence posteriorly and reaches its maximum extent just in front of the condyle, leaving virtually no space between haemal keel and condyle; it is straight in lateral profile and extends below the condyle (Fig. 40e). The lymphatic grooves are deep and extend anteriorly between the cotyle and parapophyses. A pair of subcentral foramina is present.
The deep cotyle is nearly circular, slightly taller than wide, and has a flattened dorsal edge. The neural canal has the shape of a rounded arch. The paradiapophyses are undivided. The condyle is thick and round.
Judging from the shape and sharpness of the haemal keel, the vaulting of the neural arch, and the height of the neural spine, we consider that the lectotype vertebra originates from around the mid-trunk region of the column.
Description of the paralectotype articulated trunk vertebrae (MNHN.F QU16325): The paralectotype MNHN.F QU16325 comprises five relatively complete articulated trunk vertebrae (Fig. 41). The zygosphene is discernible only in the first vertebra. In dorsal view, it possesses two prominent lateral lobes along with a (less prominent) median one (Fig. 41a). The neural spine in dorsal view is anteroposteriorly long, with its anterior edge reaching as a short, thin, nearly vertical lamina almost at the level of zygosphene. The neural spine in lateral view is moderately high and slightly posteriorly inclined (Fig. 41c, d). This element is complete in the second and fourth vertebrae, where it is visible that its dorsal level is almost straight. The posteromedian notch of the neural arch is rather deep (Fig. 41a). The prezygapophyses are not much dorsally inclined in anterior and lateral views, while they extend anterolaterally in dorsal view. The prezygapophyseal articular facets are broad and oval. Prezygapophyseal accessory processes are present, being better preserved in the first vertebra. The interzygapophyseal constriction is rounded and deep (Fig. 41a). The paradiapophyses are massive and are only weakly divided. The ventral surface is relatively eroded and still covered with much sediment in all vertebrae (Fig. 41b). The haemal keel is sharp and seems to extend over most of the midline of the centrum. The sharpness of the haemal keel and the vaulting of the neural arch denote that these vertebrae originate from around the mid-trunk region.
Description of the paralectotype “mummy” (MNHN.F QU16324): The paralectotype “mummified” specimen MNHN.F QU16324 comprises a section of the trunk 16 cm in length (Fig. 42). It is rounded on one side in cross-section and flattened on the other. It is covered in what appear to be phosphatized scales: rhomboidal and without keels. Curiously, a row of transversely elongated ventral scales (gastrosteges), as expected in an alethinophidian snake (Lee & Scanlon, 2002), does not appear to be present. No vertebrae or ribs are discernible externally.
Description of newly referred material—intracolumnar variation (Figs. 44, 45): In general, intracolumnar variation is typical for Constrictores. The neural arch is much more vaulted in anterior trunk vertebrae (Fig. 43), and a hypapophysis is present (Fig. 44). The hypapophysis is replaced on mid-trunk vertebrae by a haemal keel, which is sharply defined by deep subcentral grooves on posterior trunk vertebrae. On anterior trunk vertebrae, the centrum is nearly parallel-sided in ventral view. On mid- and especially posterior trunk vertebrae (Fig. 45), in contrast, the centrum becomes triangular, although it does not achieve the very broad proportions seen in Palaeopython cadurcensis. Sample size for the posterior trunk is low (n = 2), and whereas vaulting in one of the specimens is just like in the mid-trunk, vaulting in the other is much lower (Fig. 43), like in Phosphoroboa filholii comb. nov. Without a larger sample size it is difficult to interpret that result.
The width of the zygosphene is poorly constrained on anterior trunk vertebrae (n = 1) but the ratio of zygosphene to cotyle width, in contrast to Phosphoroboa filholii comb. nov., is notably lower than on mid- and especially posterior trunk vertebrae in “Palaeopython” neglectus (Fig. 46). On mid- and posterior trunk vertebrae, the relative zygosphene width is greater in “Palaeopython” neglectus than in Paleryx rhombifer and Phosphoroboa filholii comb. nov. (Fig. 46). The difference in zygosphene width between “Palaeopython” neglectus and Phosphoroboa filholii comb. nov. in the YPM-VPPU sample is statistically significant. Most important is the length of the neural spine. The ratio of neural spine length to CL is everywhere > 0.51, in contrast to Phosphoroboa filholii comb. nov. (Fig. 47). While a high ratio is technically also true of Palaeopython cadurcensis, it is only in “Palaeopython” neglectus that the anterior end of the neural spine clearly lies over the zygosphene. This is true throughout the trunk column.
Remarks: Rochebrune (1884) established Palaeopython neglectus upon an isolated vertebra, five articulated vertebrae, and one “mummy” from (an) imprecise locality(ies) in the Phosphorites du Quercy. Rochebrune (1884) emphasized that this species shared several characteristics with both pythonids (his Pythonides) and booids (his Boæides), however, he chose to assign it still to the genus Palaeopython, with which he identified several features in common. He further noted that the “mummified” trunk portion (the paralectotype MNHN.F QU16324) resembled the skin of extant Python spp. De Stefano (1905) described and figured additional material from Quercy, which he referred to the same species. He moved the species to Paleryx, the genus that he considered as the valid name for both English and French large-sized constrictors (De Stefano, 1905), an attribution that was subsequently followed by Kuhn (1939b). Rage (1974, 1984b) totally dismissed the “mummified” specimen and considered that the taxon was based solely on the one isolated (MNHN.F QU16326) and the five articulated vertebrae (MNHN.F QU16325). The same author also regarded that the articulated trunk vertebrae do not pertain to the same species as the isolated trunk verteba (Rage, 1974, 1984b), while he formally designated the latter specimen as the lectotype of the species (Rage, 1984b). As for its affinities, Rage (1984b) only questionably referred “P.” neglectus to Palaeopython.
As with the case of the type and referred material of Phosphoroboa filholii comb. nov. discussed below, there is an apparent confusion arising from the plates and plate captions of Rochebrune (1884). In the explanation of the respective figures (plate II, figs. 5, 5a and 5b; Fig. 39b–d), it gives the impression that Rochebrune (1884) only had one vertebral specimen available [the “mummy” was depicted in a different plate (pl. I.1, I.1a; Fig. 39a)]. According to his caption, his “figure 5” depicts a string of five articulated vertebrae (Fig. 39d), whereas his “figures 5a and 5b” represent “les mêmes grossies” (“the same [plural, i.e. vertebrae] enlarged”) in anterior and dorsal views respectively (Fig. 39b, c). As such, it is not fully clear from the original plate of Rochebrune (1884) which and how many vertebral specimens are the syntypes of this species and it becomes even more perplexing with the statement “les mêmes” by Rochebrune (1884), implying a single specimen. However, in that case that there is depicted only a single vertebral specimen, if fig. 5a, and 5b of Rochebrune (1884) truly represented the same specimen with his fig. 5, then they must have been an abstraction of the latter, where the anterior-most vertebra in the string would be the primary object to which the condyle of the posterior-most vertebra was added. Nevertheless, Rage (1974) recognized that the isolated vertebra in the plate of Rochebrune (1884) was not the same specimen as the five articulated vertebrae. He also later designated the isolated vertebra as the lectotype as he felt that the (paralectotype) articulated trunk vertebrae do not pertain to the same species as the lectotype (Rage, 1984b). Indeed, in support of Rage’s opinion, the collections of MNHN host two different vertebral specimens that bear labelling that associate them to “Palaeopython” neglectus and Rochebrune’s (1884) plate, one isolated trunk vertebra (MNHN.F QU16326) and five articulated trunk vertebrae (MNHN.F QU16325). These two specimens bear much resemblance with the ones figured in the plate of Rochebrune (1884) and any difference may be accounted to inaccuracy of the original lithograph. That being said, the original type series of “Palaeopython” neglectus consists of an isolated vertebra (lectotype), a string of five articulated vertebrae (paralectotype), but also a “mummy” (paralectotype).
As mentioned before, Rage (1984b) suggested that the paralectotype articulated trunk vertebrae (MNHN.F QU16325) do not pertain to the same species as the lectotype. However, we here tend to disagree with this opinion. The paralectotype vertebral string bears the important diagnostic features observed on the lectotype, most prominently the neural spine extending onto the roof of the zygosphene, the deep interzygapophyseal constriction, and the deep posterior median notch of the neural arch (Fig. 41). Similarities in the height and shape of the neural spine and the shape of the zygosphene also favour conspecificity. Accordingly, we consider that Rochebrune (1884) was correct and the paralectotype MNHN.F QU16325 is aptly referred to “Palaeopython” neglectus.
As for the “mummified” paralectotype specimen MNHN.F QU16324 (Fig. 42), vertebrae cannot be examined externally, and its reference to the same species as the lectotype is unsupportable until micro-CT studies can compare the vertebrae that presumably are contained within. It is currently impossible to assess whether the paralectotype MNHN.F QU16324 does or does not pertain to “Palaeopython” neglectus. See “Localities and age” above for a discussion about the probable age of this specimen, considered to pertain to around the late middle–late Eocene (late Bartonian–late Priabonian [MP 16–19 or 20, about 40–34 Ma]).
A comment about De Stefano’s (1905) referred specimens is needed here; his specimens are currently lost and only his figures remain. De Stefano (1905) supposedly figured two different pairs of articulated vertebrae which he assigned to this species. He stated in his figure captions (p. 47) that his plate V, figs. 3 and 4, illustrate two different specimens (both being two articulated vertebrae) of this species (De Stefano, 1905). However, as we show below, the specimen in De Stefano (1905:pl. V.3) is in fact the ventral view of the holotype of Paleryx cayluxi that he figured in figs. 1, 2 of the same plate (see “Remarks” in the entry of Paleryx cayluxi below). As such, the only specimen of “Palaeopython” neglectus figured by De Stefano is that in his fig. 4 of his plate V. Because of the long neural spine, we view attribution to this species as plausible.
While there are some noteworthy features in the lectotype, it is the YPM material that proves that these features (especially the long neural spine) are characteristic. It fits Palaeopython in some respects (e.g., the ratio of NSPL to CL), but not in others (e.g., it has no thick zygosphene), and it is clearly distinct from Paleryx and Phosphoroboa gen. nov. (see also Appendix 2). That being said, the generic attribution of this species is uncertain: available names of medium-sized Constrictores from the Phosphorites du Quercy do exist, such as Palaelaphis Rochebrune, 1884, Tachyophis Rochebrune, 1884, Plesiotortrix Rochebrune, 1884, and Rageophis Wallach, 1986, all of which are currently considered nomina dubia or at least they are not adequately known (Szyndlar & Rage, 2003; Wallach et al., 2014). In any case, the vertebral morphology of the type species of these medium-sized genera, seems to deviate from that of “Palaeopython” neglectus, while the lectotype Rageophis lafonti (MNHN.F QU16342; a “mummified” trunk portion with several articulated vertebrae and ribs) does not allow direct observation of its vertebrae in dorsal view, thus it cannot be deduced whether the neural spine extends onto the roof of the zygosphene. However, we notice that there is some resemblance in the vertebral morphology with the recently described pythonoid Messelopython freyi from Messel (Zaher & Smith, 2020). In particular, the two species show a similar level of vaulting of mid-trunk vertebrae and a similar shape of the interzygapophyseal constriction. Furthermore, although the tip of the neural spine is short in M. freyi, its expanded base extends out onto the zygosphenal tectum, potentially an intermediate condition on the way to the highly elongated neural spine seen in “P.” neglectus. We provisionally leave the species “P.” neglectus in “Palaeopython” but signify with quotation marks that this attribution probably is in error. Only new anatomical data, preferably from skull elements, will be able to clarify the true phylogenetic affinities of this species.
Genus Eoconstrictor Scanferla and Smith, 2020b
Type species: Palaeopython fischeri Schaal, 2004.
Other species included: Paleryx spinifer Barnes, 1927.
Geographic and stratigraphic distribution: see Table 3.
Diagnosis: Eoconstrictor is assigned to Constrictores on the basis of possessing strongly built vertebrae, the centrum wider than long, the vertebrae taller than long in lateral view, prezygapophyseal accessory processes small, paradiapophyses undivided or only weakly divided, parazygantral foramina absent, maxillary process of premaxilla tapering to a point distally, and dentary mental foramen displaced caudally. Eoconstrictor can be differentiated from other snakes by the following combination of derived features (where known): edentulous premaxilla with bifid vomerine processes; mid-sagittal keel along the basioccipital contributing to V-shaped cross-section; 11 pterygoid teeth; the vertebral column with up to 369 vertebrae, of which up to 72 are postcloacal vertebrae. Eoconstrictor can be differentiated from Palaeopython in lacking the flared posterior process of the maxilla and having a less thick zygosphene with prominent lamellar tubercle. Eoconstrictor can be differentiated from Paleryx in lacking a palatine foramen, in having a generally taller neural spine (especially on posterior trunk vertebrae), and lacking a depressed neural arch on posterior trunk vertebrae. Eoconstrictor differs from the booid Phosphoroboa gen. nov. in having a lower pterygoid tooth count (11) and in having a U-shaped frontoparietal suture.
Remarks: Scanferla and Smith (2020b) erected the new genus Eoconstrictor to encompass Palaeopython fischeri, which they determined that was not closely related to Palaeopython. Their phylogenetic analyses found E. fischeri to be on the stem of Neotropical boas, i.e. Boidae sensu Pyron et al. (2014), i.e. exclusive of Sanziniidae Romer, 1956, and Candoia Gray, 1842. Below, we recognize the Geiseltal Paleryx spinifer as a second species of Eoconstrictor, expanding the stratigraphic and geographic distribution of the genus.
Eoconstrictor fischeri (Schaal, 2004)
Type material: SMF-ME 929 (holotype), two strings of three and four articulated trunk vertebrae, respectively (not a single trunk vertebra as stated by Wallach et al., 2014), from a single individual (Schaal, 2004, fig. 1; Smith & Scanferla, 2016, fig. 2f; Figs. 48, 49 and 50, Additional file 2: Model 2); SMF-ME 1002 (paratype), a complete skeleton with skull (Schaal, 2004, figs. 2–3).
Type locality: Messel Pit, Hesse, Germany; Middle Messel Formation, MP 11, early–middle Eocene. Both holotype and paratype derive from layers 1–3 m above marker bed Alpha and are therefore middle Eocene (Lutetian) in age following either of the two alternative astronomical calibrations of Lenz et al. (2015).
Previously referred material and range: Early–middle Eocene (MP 11; see Lenz et al., 2015), Middle Messel Formation, type locality only, Messel, Hesse, Germany (referred material of Smith & Scanferla, 2016; this paper; referred material of Schaal, 2004, in part).
The oldest known Messel specimens originate from around marker bed Alpha (either Ypresian or Lutetian boundary, depending on which of the two alternative astronomical calibrations of Lenz et al. (2015) is accepted.
Referred material discussed herein (all from Messel) (Figs. 51, 52and 53): SMF-ME 1545 (partial skeleton with skull fragments), SMF-ME 1607 (partial skeleton), SMF-ME 2504 (nearly complete skeleton with disarticulated skull), SMF-ME 11332 (complete skeleton of juvenile with skull), SMF-ME 11398 (complete skeleton). The specimens derive from layers ranging from 5 to 2 m below marker bed M (SMF-ME 1545) to 3 m below marker bed Alpha. Marker beds Alpha and M are separated by about 24 m in the profile. Note that there are several more mid-sized Constrictores known from Messel, but their comparison with the type material of Eoconstrictor fischeri is complicated by the high state of articulation. Further study may identify some of these as individuals of Eoconstrictor fischeri.
Diagnosis: Eoconstrictor fischeri can be differentiated from Eoconstrictor spinifer comb. nov., in having four labial foramina in the maxilla, 15–18 maxillary teeth, palatine with 5 teeth and a long maxillary process, dentary with 18–19 teeth, relatively thin zygosphene with strong, lamellar median tubercle on all trunk vertebrae, and a narrow and almost triangular cotyle on mid-trunk vertebrae.
Description of the holotype (SMF-ME 929): The holotype SMF-ME 929 consists of two strings consisting of three and four mid-trunk vertebrae, respectively (Figs. 48, 49 and 50). These were previously closely connected, the string with four in front of the string with three. The vertebrae are massively built. The zygosphene has large, upturned lateral lobes, which make the zygosphenal roof concave, and a distinctive median tubercle at mid-height. The tubercle, unlike in examined members of Palaeopython, Paleryx, and Phosphoroboa gen. nov., is lamellar: dorsoventrally compressed and mediolaterally wide. The large prezygapophyses have oval articular facets, with the long axis somewhat oblique. Growth rings are present on the left prezygapophyseal articular facet of the string of three (Fig. 50c). Prezygapophyseal accessory processes are very weak and scarcely visible below the prezygapophysis in dorsal view.
The neural spine begins to rise steeply behind the posterior margin of the zygosphenal articular facets. It is relatively short, with a flat top and parallel anterior and posterior margins. On one vertebra, it has a slightly overhanging posterior margin. The neural spine is mediolaterally thick, but its dorsal edge is not clearly expanded. The dorsal edge appears well-ossified, suggesting an ontogenetically advanced individual. The neural arch rises posteriorly. Several vertebrae evince weak tubercles in the position of epizygapophyseal spines, but they are low and rounded. Zygantral mounds are low and rounded to elliptical. The posterior border of the neural arch is nearly straight in posterior view, showing only a slight angulation. However, it is drawn out into a pair of broad, rounded flanges that project slightly beyond the posterior end of the postzygapophyses. Between these flanges is a moderately deep median notch that just barely reaches the anterior margin of the condyle. A pair of foramina opens at depth in the zygantrum.
The interzygapophyseal constriction is smoothly rounded and deep, extending well past the mid-point of the prezygapophyseal articular facet (Figs. 48c, 49a, 50c). Lateral foramina are located well below the interzygapophyseal ridge. The haemal keel is sharp-edged with a thick base and extends from just beyond the cotyle almost to the condyle. The subcentral ridges are straight to slightly concave. The centrum is an elongate triangle in ventral view.
The cotyle is deep and narrower than the zygosphene. The two exposed cotyles are nearly triangular (Figs. 48a, 49c, 50a). Paracotylar fossae are present (Schaal, 2004) but very shallow. Paracotylar foramina are absent (Schaal, 2004). The paradiapophyses are almost undivided and project a little below the cotyle. The one well-preserved condyle is circular.
Description of additional vertebrae—intracolumnar variation (Figs. 51, 52and 53): In SMF-ME 2504, the vertebral column is stretched out and in places disarticulated, yielding (in contrast to most Messel specimens) a better understanding of intracolumnar variation (Fig. 53). The median tubercle on the zygosphene is apparently developed throughout trunk series, although it can only be evaluated at intervals.
Distinct hypapophyses, marking the anterior trunk vertebrae, are present on at least the first 70 trunk vertebrae. Here the neural arch is highly vaulted. The anterior three trunk vertebrae have slender, rodlike, highly posteriorly inclined neural spines, but thereafter they are laterally compressed. On anterior trunk vertebrae, the anterior and posterior margins of the neural spine are parallel, but on mid- and posterior trunk vertebrae the anterior margin becomes relatively more oblique due to the development of a thin, convex anterior flange to the neural spine proper. The flange begins more anteriorly, above the posterior end of the zygosphene. The neural spine is of similar height and is particularly long on the anterior-most four or five trunk vertebrae; it is relatively shorter thereafter, but this is compensated by the tall neural arch. Neural spine height rises thereafter and apparently reaches its greatest height in the posterior portion of the anterior trunk series. It declines thereafter, but is of similar height in mid- and posterior trunk vertebrae. The projection of the neural spine over posteromedian notch is similar throughout trunk series.
On many mid-trunk vertebrae the posterodorsal margin of the neural arch is concave on either side, rather than straight, in posterior view, forming a continuous curve that extends from the lateral edge of the postzygapophysis onto the neural spine. Subcentral foramina are present on all vertebrae in which the ventral surface is exposed. Zygantral mounds are poorly developed throughout the trunk series; weak, sometimes elongated, sometimes rounded bumps are present on some vertebrae, although in some cases they can be magnified by or attributed to taphonomic artefacts. Very common, and throughout the vertebral column, are neural spine foramina.
On the posterior trunk vertebrae there is a blunt haemal keel instead of a sharp ridge; it is bounded by lymphatic grooves that grow in prominence posteriorly. The anterior end of the keel has an abrupt terminus, so that it projects very slightly below the centrum. On the posterior region of the posterior trunk vertebrae the anterior margin of the haemal keel grows into a weak but distinct and later moderate hypapophysis. The subcentral ridges also grow in prominence and grow acute posteriorly.
The prezygapophyseal accessory processes are weak throughout the trunk series, and slightly anteroposteriorly compressed, but in the caudal series they are more prominent and rounded in cross-section, taper to a sharp tip, and become more anteriorly directed. Haemapophyses are present. Distally in the tail the zygosphene disappears, and the neural spine becomes long, covering almost the entirety of the neural arch.
Remarks: This species was originally established at Messel, and more than a dozen specimens have been recognized, including several rather complete and articulated skeletons with skulls (Scanferla & Smith, 2020b; Schaal, 2004; Smith & Scanferla, 2016; Smith et al., 2018). We are faced with two difficulties in the taxonomic analysis of this species. First is the original selection of a short, isolated string of vertebrae as the holotype, because parts of a whole can present no more diagnostic features than the whole, and normally present many fewer. Second, significant cranial and minor vertebral differences shown by at least two of the specimens in the original description, HLMD-Me 15426 and SMNK-PAL 2347, indicate that these represent a large constrictor not closely related to the SMF-ME specimens with skulls.
The prominent, distinctive zygosphenal tubercle of SMF-ME 929 is also seen in SMF-ME 2504 and SMF-ME 11398, and skull form in SMF-ME 1002, SMF-ME 2504, SMF-ME 11332, and SMF-ME 11398 is quite similar. Thus, despite the limited holotype, we believe that Eoconstrictor fischeri is valid and other, more complete specimens in the SMF-ME collection were correctly referred to it. Scanferla and Smith (2020b) revised the taxonomy and elucidated the phylogenetic relations of this species. The relations of the second species, specimens of which were originally included in E. fischeri, will be the subject of a later contribution.
The knowledge of the palaeoecology of Eoconstrictor fischeri was enhanced by Smith and Scanferla (2016), who described an exceptionally preserved juvenile specimen (SMF-ME 11332) from Messel with an individual of the lizard Geiseltaliellus maarius Smith, 2009, in its stomach. Other large fossils from that locality also revealed trophic habits of these snakes (Greene, 1983; Gunnell et al., 2018; Mayr & Schaal, 2016; Smith et al., 2018). However, the attribution of isolated regurgitates to E. fischeri must now be re-assessed following the recognition of more than one large constrictor in the locality.
Eoconstrictor cf. fischeri
Referred material and range: Geiseltal, late early or middle Eocene (in stratigraphic order): quarry IX; quarry I; quarry VI (including VIa); quarry XXII; quarry XLI; quarry XXXIII; quarry XXXV; quarry XXXVI; quarry XXXVII; quarry XXXVIII; quarry Cecilie IV; quarry Leonhardt III; quarry Cecilie I (this paper). Dielsdorf, late middle–late Eocene (MP 16–20): Fissure A, Fissure 2 and imprecisely known fissure (referred material of Georgalis & Scheyer, 2019a).
Referred material discussed herein (all from Geiseltal [in stratigraphic order]) (Figs. 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66and 67): quarry IX (Fig. 54): two trunk vertebrae (GMH IX-260b1-1953 and GMH IX-260b2-1953), nine trunk vertebrae and three fragments of ribs probably pertaining to a single individual (GMH IX-696-1952); quarry I (Fig. 55): 37 trunk vertebrae (GMH I-10111-1949); quarry VI (Fig. 56): three trunk vertebrae (GMH VI-370b-1950, GMH VIa-529a-1952, and GMH VIa-529b-1952); quarry XXII (Fig. 57): two trunk vertebrae (GMH XXII-761-1966), 16 trunk vertebrae and a pterygoid fragment (GMH XXII-Y-43), 12 trunk vertebrae along with four vertebral and two rib fragments (GMH XXII-Y-44-1965), and a string of several articulated trunk vertebrae (GMH XXII-604b-1965); quarry XLI (Fig. 58): many trunk vertebrae (GMH XLI-139-1966, GMH XLI-148-1966, GMH XLI-165a-1966, GMH XLI-165b-1966, GMH XLI-165c-1966, GMH XLI-165d-1966, GMH XLI-165e-1966, GMH XLI-155-1966 [six trunk vertebrae vertebrae], GMH XLI-104-1966, GMH XLI-153-1966, GMH XLI-213-1967, GMH XLI-117-1966 [eight trunk vertebrae], GMH XLI-302-1968 [three vertebrae], GMH XLI-158-1966 [two vertebrae], GMH XLI-162-1966 [two articulated and three isolated large trunk vertebrae], GMH XLI-159-1966 [five large vertebrae], GMH XLI-314c-1968 [a trunk vertebra plus a fragment of the articulated succeeding one], GMH XLI-325-1968 [many trunk vertebrae]), a string of six articulated trunk vertebrae and remains of ribs (GMH XLI-157-1966), a string of three small articulated trunk vertebrae and ribs (GMH XLI-50-1966), two trunk vertebrae and ribs (GMH XLI-160-1966); quarry XXXIII: a trunk vertebra along with three vertebral and two rib fragments (GMH XXXIII-16-1962) and two strings of articulated mid-trunk vertebrae (GMH XXXVIII-61-1964); quarry XXXV (Fig. 59): several trunk vertebrae (XXXV-274-1963, GMH XXXV-380-1963, GMH XXXV-131a-1963, GMH XXXV-131b-1963, GMH XXXV-577-1963, GMH XXXV-552-1963 [six vertebrae], GMH XXXV-235-1963 [two trunk vertebrae], GMH XXXV-435b-1963 [two vertebrae], GMH XXXV-401-1963) three articulated mid-trunk vertebrae (GMH XXXV-310-1963), and a trunk vertebra (GMH XXXV-403-1963); quarry XXXVI (Figs. 60, 61): a string of several articulated trunk vertebrae and ribs (GMH XXXVI-530-1963), two trunk vertebrae (GMH XXXVI-376b-1963 and GMH XXXVI-476-1963 [two fragments of vertebrae]), and a partial skeleton with many articulated vertebrae and ribs [GMH XXXVI-87-1962; originally figured by Krumbiegel et al. (1983:fig. 115); Fig. 61]; quarry XXXVII (Fig. 62): several trunk vertebrae (GMH XXXVII-179a-1970, GMH XXXVII-179b-1970, GMH XXXVII-179c-1970, GMH XXXVII-179d-1970, and GMH XXXVII-94-1964 [two pairs of two articulated trunk vertebrae and four isolated trunk vertebrae]); quarry XXXVIII (Figs. 63, 64): a large, partial, relatively three dimensional skeleton including series of articulated vertebrae and ribs (GMH XXXVIII-24-1964) and several articulated and disarticulated vertebrae and fragments of a skull on a matrix (GMH XXXVIII-7-1964); quarry Cecilie IV (Fig. 65c–h): a trunk vertebra (GMH Ce IV-2912b-1933; originally referred to Paleryx spinifer and figured by Kuhn, 1939a:pl. III.2a–d; Fig. 65c–h), several fragmentary disarticulated trunk vertebrae (GMH Ce IV-2912-1933), a trunk vertebra with two disarticulated ribs (GMH Ce IV-4649-1934), a trunk vertebra and a rib (GMH Ce IV-5856-1933), and several disarticulated trunk vertebrae and ribs on a matrix (GMH Ce IV-5855-1933); quarry Leonhardt III (Fig. 65a–b): an anterior trunk vertebra (GMH Leo III-2933-1934); quarry Cecilie I (Figs. 66, 67): a small matrix with several articulated and disarticulated trunk vertebrae and remains of ribs (GMH Ce I-5852-1931), several trunk vertebrae (GMH Ce I-5836-1929), many small trunk vertebrae (GMH Ce I-1930-1931), ten anterior and mid-trunk vertebrae (GMH Ce I-5831-1926-29), eight trunk vertebrae along with vertebral fragments and ribs (GMH Ce I-5832-1929), six trunk vertebrae (GMH Ce I-5834b-1926, GMH Ce I-5834c-1926, GMH Ce I-5834d-1926, GMH Ce I-5834e-1926, GMH Ce I-5834f-1926, and GMH CI-5834g-1926), and several trunk vertebrae (GMH Ce I-5835a-1929, GMH Ce I-5835c-1929, GMH Ce I-5835h-1929 (originally figured by Kuhn, 1939a:pl. III.5; Fig. 67f–j), GMH Ce I-5835i-1929, and GMH Ce I-5835j-1929); unknown quarry: an anterior trunk vertebra, five mid-trunk vertebrae, and four vertebral fragments (GMH Y-46).
Description: Anterior trunk vertebrae have a taller neural arch and neural spine (e.g., GMH IX-696b-1952). These vertebrae bear a hypapophysis (Figs. 54a–l, 65a, b). The hypapophysis is relatively broad and short. It varies in shape and size, even within vertebrae of the same individual, depending on the relative position of the vertebra in the column (e.g., GMH IX-696b-1952 and GMH IX-696c-1952), but never surpasses the posterior border of the condyle. A prominent median tubercle on the zygosphene also occurs in the anterior trunk vertebrae. However, rarely this median tubercle is missing, and the zygosphene in anterior view has a strange, downwardly directed curved shape. Also, certain anterior trunk vertebrae have a relatively thick zygosphene (GMH Leo III-2933-1934; Fig. 65a, b). The prezygapophyses are not so inclined as in Palaeopython ceciliensis and can be almost horizontal in anterior view and level with the dorsal margin of the cotyle (e.g., GMH XXXV-310-1963). Prezygapophyseal accessory processes can be present, though small.
Mid-trunk vertebrae have a sharp haemal keel (e.g., GMH Ce I-5835c-1929). Their average CL is between 8 and 10 mm (see Appendix 1). Smaller specimens have a CL of around 6 mm (e.g., GMH XLI-302-1968), while certain specimens can achieve a relatively large vertebral size [e.g., GMH XXII-Y-43a and GMH XXII-Y-43b (Fig. 57)] and, rarely, surpass a CL of 12 mm (e.g., GMH XLI-148-1966; Fig. 58h, i). The centrum is not so widened anteriorly (but see GMH XXXV-310-1963 for an exception). In certain cases, the median tubercle on the zygosphene is extremely prominent [e.g., GMH XXII-Y-43a (Fig. 57a); GMH XXXVI-87-1962 (Fig. 61c)]. Rarely, however, the median tubercle is either extremely small or even totally absent (e.g., GMH XXXV-577-1963, GMH Ce IV-4649-1934 [but this identification is tentative]). In rare cases observed in certain large specimens, the zygosphene is dorsally concave and two prominent lateral lobes are visible in dorsal view (e.g., GMH XXXV-310-1963). The paradiapophyses are large and undivided. The zygantrum is deep. The neural spine is usually short, posteriorly inclined [e.g., GMH XXII-Y-43a (Fig. 57e), GMH XXXV-380-1963, GMH XXXV-131b-1963 (Fig. 59j); GMH XXXVII-179a-1970 (Fig. 62e); GMH XLI-162-1966; GMH Ce IV-2912-1933 (Fig. 65g, h)]. Subcentral foramina are situated usually on both sides of the haemal keel, and in certain cases they are rather prominent (e.g., GMH Ce I-5835i-1929). In certain specimens, the prezygapophyseal accessory processes are distinct, though still small [e.g., GMH XLI-155a-1966 (Fig. 58c, d); GMH XLI-314c-1968 (Fig. 58f, g)].
In posterior mid- and posterior trunk vertebrae, the neural arch becomes depressed [e.g., GMH IX-696d-1952 (Fig. 54c); GMH VIa-529b-1952 (Fig. 56g)]. The neural spine is slightly shorter than on the preceding vertebrae. As such, they are comparatively taller than the neural spines of the posterior trunk vertebrae of the co-occurring Palaeopython ceciliensis. In a single specimen (GMH XXXVI-376b-1963), the dorsal surface of the neural spine slightly overhangs both anteriorly and posteriorly, in contrast to the situation observed in all other trunk vertebrae, where this structure is overhanging only posteriorly. The haemal keel in these vertebrae is wider. It is not uniform in shape throughout its length but is markedly constricted at its middle part, resembling thus an hourglass [e.g., GMH IX-260b1-1953, GMH XXXVI-376b-1963, and GMH VIa-529b-1952 (Fig. 56i)]. In certain cases, the prezygapophyses can be relatively long.
Remarks: Outside the type locality of Messel, a form compared to Eoconstrictor fischeri was recently described from the Eocene of Dielsdorf, Switzerland, by Georgalis and Scheyer (2019a). We here agree with this taxonomic referral of that Swiss material as Eoconstrictor cf. fischeri—these specimens were described and figured in much detail by Georgalis and Scheyer (2019a), and there is no need to repeat it here. We now further recognize that a similar form was also present in Geiseltal, which we also refer as Eoconstrictor cf. fischeri. This taxon is represented in Geiseltal by numerous isolated vertebrae, originating from multiple different quarries, as well as several articulated and disarticulated skeletons (Figs. 60, 61, 63, 64, 66). The Geiseltal form shares with Eoconstrictor fischeri from Messel a relatively thin zygosphene with strong, lamellar median tubercle, and a narrow and almost triangular cotyle. In Geiseltal, these specimens can be differentiated from the co-occurring Palaeopython ceciliensis by their much smaller vertebral size, much thinner zygosphene, the relatively higher neural spine in the posterior trunk vertebrae, the presence of a prominent median tubercle on the zygosphene, and the less massive paradiapophyses. Given the great similarity of the cranial lectotype material of Eoconstrictor spinifer comb. nov. to cranial specimens of E. fischeri from Messel, we suspect that most trunk vertebrae of the former might also be similar to those of the latter. The fundamental problem here is that Eoconstrictor spinifer comb. nov. is essentially a taxon based on a cranium, which is difficult to compare to these vertebrae—the only available vertebrae of E. spinifer comb. nov. are anterior-most trunk ones, which nevertheless indeed show certain differences (i.e. no median tubercle on the zygosphene and no prezygapophyseal accessory processes) from the available respective ones of E. cf. fischeri. Associated specimens may provide new insight.
Eoconstrictor spinifer (Barnes, 1927) comb. nov.
Taxonomic history: Paleryx spinifer Barnes, 1927 (new species).
Type material: GMH Ce I-2979-1926 (lectotype [part]), a “block” including the left atlantal neural arch, three almost complete anterior trunk vertebrae, remains of ribs, fragments of other vertebrae, as well as basioccipital, and right prootic (Barnes, 1927, pl. I.1; Figs. 68a, b, 69, Additional file 3: Model 3); GMH Ce I-5823-1926 (lectotype [part]), an incomplete parietal (Barnes, 1927, pl. I.5b; Figs. 68c, d, 70, Additional file 4: Model 4); GMH Ce I-5824-1926 (lectotype [part]), a right pterygoid (Barnes, 1927, pl. I.4; Figs. 68e, f, 71, Additional file 5: Model 5); GMH Ce I-5826-1926 (lectotype [part]), a right maxilla (Barnes, 1927, pl. I.2, I.3; Figs. 68g–j, 72, Additional file 6: Model 6); GMH Ce I-5822-1926 (paralectotype), two isolated anterior trunk vertebrae (the second one has also together the anterior-most portion of a succeeding vertebra stuck in its posterior part) (Barnes, 1927, pl. I.5, I.5a, I.6, I.6a; Figs. 73, 74, Additional file 7: Model 7).
Barnes (1927:7) apparently overlooked the cranial elements in GMH Ce I-2979-1926 but noted that the skull fragments—among which he counted not only the potential parietal, but also the pterygoid (GMH Ce I-5824-1926), the maxilla (GMH Ce I-5826-1926), and a palatine fragment that we have not been able to identify—were associated with the same block (“dabei befindlich”). This suggests that most of these “syntypes”, almost all of which derive from the skull or anterior-most vertebral column, pertain to a single individual and were subsequently catalogued as different specimens. Furthermore, based on the numbering one might think that the isolated anterior trunk vertebrae (GMH Ce I-5822-1926) were also isolated from the block; Barnes (1927:9) himself noted that they were found in the same, several-centimetre-thick sediment layer, and given their state of preservation he speculated that they pertained to the same individual. We suggest, therefore, that the syntype series comprises two specimens: (1) the associated specimen (catalogued as GMH Ce I-2979-1926, GMH Ce I-5823-1926, GMH Ce I-5824-1926, and GMH Ce I-5826-1926), and (2) the two isolated anterior trunk vertebrae (GMH Ce I-5822-1926), which possibly derive from the same individual. Among these two, and for the sake of nomenclatural and taxonomic stability, we designate the first syntype as the lectotype of Eoconstrictor spinifer comb. nov.
Type locality: Quarry Cecilie I, Geiseltal "Obere Mittelkohle", Saxony-Anhalt, Germany; late early or middle Eocene.
Referred material and range: None—taxon so far exclusively known by its type material.
Diagnosis: Eoconstrictor spinifer comb. nov. differs from Eoconstrictor fischeri in having fewer labial foramina on the maxilla, and in the absence of a median tubercle on the zygosphene and prezygapophyseal accessory processes on the anterior-most trunk vertebrae.
Description of the lectotype: The lectotype comprises specimens GMH Ce I-2979-1926, GMH Ce I-5823-1926, GMH Ce I-5824-1926, and GMH Ce I-5829-1926 (Figs. 68, 69, 70, 71 and 72, Additional file 3: Model 3, Additional file 4: Model 4, Additional file 5: Model 5, Additional file 6: Model 6).
GMH Ce I-2979-1926 is a block with the left atlantal neural arch, three almost complete anterior trunk vertebrae, remains of ribs, and fragments of other vertebrae as well as cranial elements (Figs. 68a, b, 69, Additional file 3: Model 3). The dorsal portion of the atlas, the tectum (sensu Jandzík and Bartík, 2004), is a medially curving flange whose dorsal margin is rounded in lateral view. Two processes are present laterally, a posterodorsal process at the base of the tectum and a more robust transverse process. The pedicel for articulation with the oto-occipital and the odontoid process of the axis is present. In one of the vertebral fragments, the hypapophysis is perfectly preserved. The zygosphene is slightly wider than cotyle. The hypapophysis is sharp. Cotyles are more circular than elliptical. The postzygapophyses are wide; the zygantrum is triangular. The zygosphenes are thin. In two, the neural spines are broken, and even in the third, where it is preserved, it is hidden by other remains.
The ventral surface of the incomplete basioccipital is divided into an anterior part and a projecting posterior part, the condyle, which set off by a strong neck. The anterior part bears a median crest on which the inclined surfaces for muscle attachment (probably m. protractor quadrati; see Cundall, 1986) converge. The anterior margin appears to be a concave V-shape, which the basisphenoid would have overlapped. In dorsal view the concave brain cavity is pentagonal, with a transverse posterior margin and a pointed anterior one.
We identified a partial right prootic in the block. The foramen for the mandibular branch of the trigeminal nerve (V3) has a sharp, acute anterior margin, as in Boa and Ungaliophis Müller, 1880, and also Eoconstrictor fischeri (Scanferla & Smith, 2020b). It is bounded ventrally by a strong buttress, which in Ungaliophis is ventrally scored, forming a groove that continues into the posterior opening of the Vidian canal; the buttress is absent in Eryx and apparently present in Boa and Candoia, where, however, it does not have the same close relation to the Vidian canal. The expanse of bone anterior to the V3 opening suggests that an ophidiosphenoid (sensu Gauthier et al., 2012, “pleurosphenoid” of Rieppel, 1976) is present. The opening for the maxillary branch of the trigeminal nerve (V2) may be preserved anteriorly, although it cannot be determined whether it was fully enclosed by the prootic or open anteriorly. A relatively large foramen, perhaps comparable to those identified by Maisano and Rieppel (2007) in Casarea dussumieri, appears to be present in the ophidiosphenoid. Dorsally is part of the facet for the supratemporal. In medial view the internal opening of the trigemino-facialis chamber can be made out.
The element identified as a parietal by Barnes (1927), GMH Ce I-5823-1926, is very narrow (Figs. 68c, d, 70, Additional file 4: Model 4); possibly it represents the posterior, narrowed extension of a bone similar to that in Boidae. In this interpretation, the dorsal surface is keeled and three-pronged, with an elongate, tapering median process that overlapped the oto-occipitals on the midline and a pair of thinner, less extensive lateral processes (less so than in Eoconstrictor fischeri; Scanferla and Smith, 2020b), somewhat asymmetrically developed, that overlapped the dorsolateral surfaces of the oto-occipitals. The left descending flange is partially preserved where it articulated with the prootic.
GMH Ce I-5824-1926 comprises most of the right pterygoid, including its articulation facets for the palatine, ectopterygoid, and parabasisphenoid but lacking the distal end of the quadrate ramus (Figs. 68e, f, 71, Additional file 5: Model 5). The palatine process is tall with a dorsal keel and terminates bluntly, lacking a strong anterior edentulous prong, unlike in Loxocemus. Its dorsomedial surface bears a deeply incised elongate facet for the posterior prong of the palatine that extends over four tooth spaces. The ventromedial margin of that facet is drawn out into a ridge, but a distinct process (as seen in some boids; see below under Phosphoroboa filholii comb. nov.) is absent, similar to Eoconstrictor fischeri. The palatine process bears a single, inwardly curving row of approximately 11 teeth, as in E. fischeri, which grow somewhat smaller posteriorly. The last tooth is at the level of the posterior end of the ectopterygoid facet, as in ungaliophiines, Boa and Candoia but unlike in Eryx. The ectopterygoid facet is a robust, laterally directed, comma-shaped depression, taller at the posterior end and rising and tapering anterodorsally. At the basipterygoid articulation the pterygoid bears a rounded, medially expanded flange. This flange rotates onto the dorsal surface of quadrate ramus, forming a dorsal keel that bounded a deep, elongate dorsally or dorsomedially directed fossa, as in many Boidae and Pythonidae.
GMH Ce I-5826-1926 is a nearly complete right maxilla, but its anterior and posterior ends are broken (Figs. 68g, h, 72, Additional file 6: Model 6). On the dorsal surface of the premaxillary process is an elongate well, at the base of which one large and several small foramina open, similar to the paralectotype maxilla of Palaeopython cadurcensis MNHN.F QU16321. At a slightly posterior level to this well on the lateral side of the bone is a single, large labial foramen, unlike in Eoconstrictor fischeri, where there are four (Scanferla & Smith, 2020b). A weak remnant of the lizard facial process is present, as in constrictors generally, but it decays more rapidly than in MNHN.F QU16321, so that the posterior process of the maxilla is more slender. The palatine process is asymmetrical, growing gradually in width posteriorly, reaching an apex, and then decaying abruptly, similar to Boa. Foramina associated with the palatine process are unclear. The lateral margin is straight, as in E. fischeri (see Scanferla & Smith, 2020b).
The bone preserves 14–15 tooth positions and four complete and several partial teeth located in the anterior and middle portion. The teeth are recurved and conical, and everywhere short and thick, even in the anterior part of the bone. Tooth morphology in the maxilla is homogeneous in some other Constrictores, such as Loxocemus bicolor and Ungaliophiinae (Digimorph), but contrasts sharply with that seen in other constrictors, especially Corallus Daudin, 1803a.
Description of paralectotype GMH Ce I-5822-1926: This specimen comprises two isolated anterior trunk vertebrae (Figs. 73, 74, Additional file 7: Model 7). The first is almost complete, missing only the dorsal-most part of the neural spine, the left lateral edge of the zygosphene, the dorsal surface of the cotyle, and the ventral-most portion of the hypapophysis (Figs. 73a–e, 74). In anterior view (Figs. 73a, 74a), the zygosphene wider than the cotyle (4.8 vs. 3.6 mm, respectively). The neural canal is slightly irregular, probably artifactual. The cotyle is slightly depressed. The prezygapophyseal articular facets are slightly inclined. No paracotylar foramina are present, but paracotylar fossae exist. In posterior view (Figs. 73b, 74d), the neural arch is vaulted, rendering the dorsal surface of the zygantrum almost triangular. The condyle is rounded. In dorsal view (Fig. 73c), the thin anterior edge of the neural spine extends just barely onto the roof of the zygosphene. The prezygapophyses are directed anterolaterally and their distal end extends slightly beyond the anterior level of the zygosphene. The prezygapophyseal articular facets are oval. No prezygapophyseal accessory processes are present. The neural spine is relatively thick. In ventral view (Figs. 73d, 74i), the hypapophysis is thick, and its base extends over most of the centrum, but its tip is broken. The centrum is much widened anteriorly. Subcentral ridges are present. A deep subcentral foramen is present on the right side from the hypapophysis at around the middle of the centrum. The paradiapophyses are relatively robust and undivided. It is worth noting that the right postzygapophysis is smaller than the left one and is oriented differently; it is unclear whether this can be attributed to preservational factors or represents a pathology. In lateral view (Figs. 73e, 74m), the neural spine is relatively tall and posteriorly inclined. A small lateral foramen can be observed on the left side.
The second vertebra is most probably also an anterior trunk one, missing the posterior part of its ventral surface along with its condyle, whereas remains of the anterior-most portion of a succeeding vertebra are stuck in its zygantrum and around the postzygapophyses (Fig. 73f–i). The centrum is rather narrow. A hypapophysis seem to have been originally present but only its relatively steep base is now preserved. The zygosphene is wider than the cotyle. The paradiapophyses are robust and weakly divided. The dorsal-most edge of the neural spine is eroded. The neural spine appears to be strongly inclined posteriorly. The zygosphene is not especially thick, but it is still thicker than in the anterior trunk vertebra.
Remarks: The species was originally established upon both vertebral and cranial material as a second, distinct snake species from Geiseltal (Barnes, 1927). Simpson (1933) regarded this taxon as one of the best-preserved fossil constrictors described up to that time worldwide. Kuhn (1939a) referred to this taxon four additional specimens from the quarries Cecilie I and Cecilie IV of Geiseltal, including a rather complete, though crushed, skeleton with skull. Rage (1984b) only questionably followed the generic attribution of this taxon, as he demonstrated that it in fact shares features with both Paleryx and Palaeopython. Indeed, even many of the isolated vertebrae (e.g., GMH Ce IV-2912-1933) that Kuhn (1939a) described and figured as Paleryx spinifer in fact pertain to Eoconstrictor cf. fischeri (see above). We here reject Kuhn’s referral of additional specimens to this species, and Eoconstrictor spinifer is so far confined exclusively to its lectotype and paralectotype specimen.
It was suggested by Kuhn (1939a) that certain cranial features of Eoconstrictor spinifer comb. nov. resemble pythonids, especially the toothed premaxilla of a specimen (GMH Ce IV-5864-1933) referred by that author to the same species (see Szyndlar, 1994), whereas other characters were considered more reminiscent of booids (Szyndlar & Böhme, 1993). However, we were unable to identify any premaxillary teeth in GMH Ce IV-5864-1933 and furthermore, we cannot refer this specimen to Eoconstrictor spinifer comb. nov. and we rather treat it as Constrictores indet. (see that entry and the respective figure of that specimen below). In any case, we have to highlight that there is a strong cranial resemblance among the lectotype of Eoconstrictor spinifer comb. nov. and E. fischeri. Such similarities among the two species are mostly evident in the anatomy of the basioccipital, maxilla, and pterygoid. Still, there are sufficient differences that lead us to conclude that they are distinct species.
Genus Phosphoroboa nov.
Type species: Palaeopython filholii Rochebrune, 1880.
Etymology: The genus name Phosphoroboa originates from the Greek “Φώσφορος” (“Phosphoros”), referring to its presence in the Phosphorites du Quercy, and the extant genus Boa. Gender is masculine.
Diagnosis: Phosphoroboa gen. nov. is assigned to Constrictores on the basis of possessing strongly built vertebrae, the centrum wider than long, the vertebrae taller than long in lateral view, prezygapophyseal accessory processes small, paradiapophyses un-divided or only weakly divided, and the lack of parazygantral foramina. Phosphoroboa gen. nov. can be assigned to Booidea based on the projection anteromedial to the palatine articulation on the pterygoid and the deep lateral facet for the ectopterygoid articulation on the pterygoid. Size moderate, with CL maximally around 10 mm; ratio of neural spine length to centrum length < 0.5; and ratio of zygosphene width to cotyle width low, < 1.2 on mid-trunk vertebrae. Phosphoroboa gen. nov. differs from the booid Eoconstrictor in having a higher pterygoid tooth count (14) and a V-shaped frontoparietal suture. Phosphoroboa gen. nov. can be differentiated from Paleryx by possessing wider zygosphene in anterior view and neural arch less depressed on posterior trunk vertebrae. Phosphoroboa gen. nov. can be further differentiated from Palaeopython by its smaller size, more depressed neural arch, thinner and narrower zygosphene, shorter neural spine on mid- and posterior trunk vertebrae, centrum less widened anteriorly, and postzygapophyseal articular facets less expanded laterally, and the absence of a flared posterior process of the maxilla. Further comparisons with other European Constrictores are presented in “Remarks” section.
Other species included: None.
Geographic and stratigraphic distribution: see Table 3.
Remarks: The cranial and vertebral material from the locality of Escamps, as well as those from the old collections of the Phosphorites du Quercy enable a more comprehensive understanding of the anatomy and relationships of “Palaeopython” filholii. Especially the cranial remains from Escamps enable a more precise placement within Constrictores, by referring the species to Booidea (see Description of these elements below). Considering its distinction from Palaeopython and Paleryx, in search for a possible available candidate genus name for “Palaeopython” filholii, we compared this species with all named genera of Booidea from Europe, as well as all named genera of Constrictores from the Phosphorites du Quercy. As there are adequate differences with all these taxa (see details in the next paragraph), we consider it justified to establish a new genus, Phosphoroboa gen. nov., in order to accommodate “Palaeopython” filholii.
In particular, Phosphoroboa gen. nov. can be differentiated from Eoconstrictor by the morphology of the pterygoid and parietal (see Diagnoses of Eoconstrictor and Phosphoroboa gen. nov.). Phosphoroboa gen. nov. can be differentiated from Bavarioboa Szyndlar and Schleich, 1993, referred species of which are known also from the Oligocene of the Phosphorites du Quercy, in the neural spine length, the shape of the haemal keel in the posterior-most trunk, and its pterygoid morphology, with its pterygoid crest more extended medially and its ectopterygoid process being less prominent laterally and located more posteriorly. Phosphoroboa gen. nov. can be readily distinguished from the charinaine charinaid Rageryx Smith and Scanferla, 2021, from Messel by its larger size and the absence of complex accessory processes on caudal vertebrae. Phosphoroboa gen. nov. can be distinguished from the other two charinaid genera from Messel, i.e. Messelophis Baszio, 2004, and Rieppelophis Scanferla et al., 2016, by its larger size and different shape of neural spine. Other named genera of Constrictores from the Phosphorites du Quercy include Palaelaphis, Tachyophis, Plesiotortrix, Rageophis, Cadurceryx Hoffstetter and Rage, 1972, Szyndlaria Augé and Rage, 2010, Platyspondylia Rage, 1974, Cadurcoboa Rage, 1978, and, of course, Palaeopython. A number of differences enable the distinguishment of Phosphoroboa gen. nov. from the larger Palaeopython (see details in Diagnosis of Phosphoroboa gen. nov. above). Phosphoroboa gen. nov. can be readily distinguished from the “erycine” Cadurceryx by the absence of complex accessory processes on caudal vertebrae. Phosphoroboa gen. nov. can be distinguished from Platyspondylia by its much larger size, higher neural spine, and thicker zygosphene. Phosphoroboa gen. nov. can be distinguished from Szyndlaria by its much larger size and rather different shape of neural spine and haemal keel. Phosphoroboa gen. nov. can be distinguished from Cadurcoboa by its larger size, its much more depressed neural arch, and shorter and differently shaped neural spine. Phosphoroboa gen. nov. seems to be much different from Plesiotortrix in possessing a much longer centrum and neural spine that is not confined to the posterior part of the neural arch. Comparison of Phosphoroboa gen. nov. with Rageophis is difficult, as the lectotype of the type species Rageophis lafonti is a “mummified” trunk portion with several articulated vertebrae and ribs; nevertheless, the ventral view of the vertebrae of R. lafonti is directly discernible and it seems that they possess more elongated centra, more robust diapophyses, and more laterally directed prezygapophyses, in comparison with Phosphoroboa gen. nov. Phosphoroboa gen. nov. can be differentiated from the type species of Palaelaphis, i.e. Palaelaphis antiquus (currently considered to be a nomen dubium; Rage, 1984b; Szyndlar & Rage, 2003) by its more depressed neural arch, thicker zygosphene, and different shape of neural spine. Finally, Tachyophis nitidus Rochebrune, 1884, another species that is currently considered to be a nomen dubium (Rage, 1984b; Szyndlar & Rage, 2003), bears some resemblance with Phosphoroboa filholii comb. nov. in terms of the relatively depressed neural arch. However, the lectotype of T. nitidus, four articulated vertebrae (MNHN.F QU16331), which is also the only specimen that can be securely associated with this species, has a rather small size (CL around 2.5 mm) and the specimen probably pertains to a young individual, as it has been already suggested by Rage (1974, 1984b). As such, considering its early ontogenetic stage, Tachyophis is properly considered a nomen dubium. We further note that the paralectotype pterygoid (MNHN.F QU16330) of Tachyophis nitidus bears resemblance with the pterygoid (UM ECC 2501) of Phosphoroboa filholii comb. nov. from Escamps, but in any case, this former specimen cannot be referred with certainty to T. nitidus and being simply a paralectotype, it has no actual nomenclatural power. Another taxon of Constrictores from Quercy is Pylmophis gracilis Rochebrune, 1884, currently considered a nomen dubium (Rage, 1984b; Szyndlar & Rage, 2003); note, however, that Pylmophis Rochebrune, 1880, cannot be an available name, as the Quercy taxon Pylmophis gracilis is a referred species and the genus is typified by a Miocene natricid, i.e. Coluber sansaniensis Lartet, 1851 (currently placed into Natrix Laurenti, 1768; Rage, 1981). Finally, as Phosphoroboa gen. nov. is a member of Booidea, affinities with the non-booid constrictors Paleryx and Messelopython are readily discarded; as such, resemblance in the vertebral anatomy between Phosphoroboa gen. nov. and Paleryx should only be considered as superficial, not alluding to actual phylogenetic relationships (see also “Remarks” in the entry of Phosphoroboa filholii comb. nov., below).
Nomenclatural acts: This published work and the nomenclatural act it contains have been registered in ZooBank, the official registry of zoological nomenclature for the International Code of Zoological Nomenclature (ICZN). The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix ‘http://zoobank.org/’. The LSID for this publication is: urn:lsid:zoobank.org:pub:44C5A053-CDE4-477E-AEEE-63F3860FA4C0. The LSID for this new genus name Phosphoroboa gen. nov. is: urn:lsid:zoobank.org:act:880E62D3-8435-4818-A157-8068924B36F4
Phosphoroboa filholii (Rochebrune, 1880) comb. nov.
Taxonomic history: Palæopython filholii Rochebrune, 1880 (new species); Palæopython filholi Rochebrune, 1884 (variant spelling of species epithet); Paleryx filholi Lydekker, 1888a (new combination and variant spelling of species epithet); Paleryx depressus = Palaeopython filholi Lydekker, 1888b (junior synonym and variant spelling of species epithet); Paleopython filholi Palacký, 1898 (incorrect spelling of genus name and variant spelling of species epithet).
Type material: MNHN.F QU16322 (formerly MNHN QU 322) (holotype), a posterior trunk vertebra (Rochebrune, 1880, pl. XII. 5a, 5b, and 5c; Rochebrune, 1884, pl. II.4b, II.4c; Rage, 1974, fig. 3A; Krumbiegel et al., 1983, fig. 116A; Rage, 1984b, fig. 13b; Georgalis & Scheyer, 2019a, fig. 15i; Fig. 75).
Type locality: Unknown exact locality, Phosphorites du Quercy, France; probably middle or late Eocene.
Previously referred material and range: Middle or late Eocene, type area, Quercy, France (referred material of Rochebrune, 1884, Lydekker, 1888a, De Stefano, 1905, and Georgalis & Scheyer, 2019a; material referred to Paleryx depressus by Lydekker, 1888a); late Eocene (MP 19), Escamps A and C, Quercy, Lot, France (Rage, 1974).
Referred material discussed herein: Phosphorites du Quercy, Escamps A locality, Lot, France (MP 19, late Eocene): a parietal (UM ECA 2543) and a left dentary (UM ECA 2544); Phosphorites du Quercy, Escamps C locality, Lot, France (MP 19, late Eocene): six trunk vertebrae (UM ECC 2572–UM ECC 2577), two cloacal vertebrae (UM ECC 3095 and UM ECC 3096), two caudal vertebrae (UM ECC 3097 and UM ECC 3098), and a left pterygoid (UM ECC 2501); Phosphorites du Quercy, imprecisely known localities: three articulated trunk vertebrae (MNHN.F QU16323), two articulated trunk vertebrae (NHMW 2019/0033/0007), 150 trunk vertebrae (NHMW 2019/0033/0001, NHMW 2019/0033/0002, NHMW 2019/0033/0004–NHMW 2019/0033/0006, NHMW 2019/0033/0008–NHMW 2019/0033/0036, NHMW 2019/0033/0038–NHMW 2019/0033/0050, NHMW 2019/0033/0052–NHMW 2019/0033/0154), a left maxilla (NHMW 2019/0020/0001), a right dentary (NHMW 2019/0020/0002), and a left pterygoid fragment (NHMW 2019/0020/0004); Phosphorites du Quercy, imprecise localities near Mouillac: 27 trunk vertebrae (5 anterior, 17 middle and 5 posterior trunk; YPM-VPPU 30369–YPM-VPPU 30395) and two dentaries (NHMUK PV R 3489 and YPM-VPPU 30397).
Diagnosis: As for the genus Phosphoroboa gen. nov.
Description of the holotype (MNHN.F QU16322): MNHN.F QU16322 is a nearly complete posterior trunk vertebra, with only its paradiapophyses being partially eroded (Fig. 75). In anterior view (Fig. 75a), the zygosphene is only moderately thick, having a trapezoidal shape, with its dorsolateral edges extending well laterally. Its dorsal surface is almost flat, with only a small convexity appearing posteriorly on the midline. There is a small tubercle at mid-height. The zygosphene is slightly wider than the cotyle. The neural canal is triangular. The cotyle is deep and nearly circular. No paracotylar foramina are present. The prezygapohyses are only slightly inclined. In posterior view (Fig. 75b), the neural arch has a vaulting ratio of 0.34 and its posterior margin is straight, not angulated. The zygantrum is large and deep. The condyle is large and circular. In dorsal view (Fig. 75c), the vertebra is wider than long. The neural spine is moderately thick, and its anterior base commences well behind the zygosphene. The zygosphene has two slight lateral lobes and a median one. The prezygapophyses are laterally extensive, with the prezygapophyseal articular facets being oval, with the long axis oriented obliquely. Prezygapophyseal accessory processes are small and pointed. The interzygapophyseal constriction is prominent; it is smoothly and nearly symmetrically rounded and extends medially well beyond the mid-point of the prezygapophyseal articular facet. The posterior median notch of neural arch is deep. In ventral view (Fig. 75d), the centrum is triangular and its margins, the subcentral ridges, are straight or slightly concave. The haemal keel is thick and rounded and narrows at mid-length. The haemal keel commences at the ventral lip of the cotyle but terminates well before the condyle. The paradiapophyses are eroded, but it can be seen that the diapophyses and parapophyses were poorly divided. The postzygapophyseal articular facets are large and triangular. In lateral view (Fig. 75e), the vertebra is taller than long. The neural spine has parallel anterior and posterior margins and is posteriorly inclined, slightly overhanging the posteromedian notch of the neural arch. The subcentral ridges are convex. The haemal keel is prominent and its posterior edge reaches the level of the distinct condylar neck.
Note that the centrum length (CL) given here (10.36 mm) differs from the one (8.8 mm) provided by Rage (1974), which is in fact closer to the means in samples described below. We cannot account for difference. It is possible that his measurement did not include the condyle, or covered only the haemal keel, or used some less precise procedure.
Description of additional vertebrae from Quercy—intracolumnar variation (Figs. 76, 77, 78, 79, 80, 81, 82, 83, 84, 85and 86, 91, 92, 93, 94and 95): Vertebrae from practically all regions of the skeleton are available, so that individual and intracolumnar variation is adequately known. Anterior-most trunk vertebrae (e.g., UM ECC 2572; Fig. 82a–c) are smaller, they have a small, relatively thick hypapophysis, and they possess a less thick zygosphene and a larger neural canal. They have much more vaulted neural arch than succeeding vertebrae. The prezygapophyses and postzygapophyses are also smaller. The neural spine is tall, directed posteriorly [e.g., NHMW 2019/0033/0075; NHMW 2019/0033/0044 (Fig. 79j)]. However, certain specimens possess a short neural spine (e.g., NHMW 2019/0033/0018; Fig. 76). Neural spine foramina can be present (e.g., NHMW 2019/0033/0018, NHMW 2019/0033/0044). Succeeding anterior trunk vertebrae are characterized by larger size and shorter neural spines. The neural spine can be relatively thick in few large specimens (e.g., NHMW 2019/0033/0019), reminiscent of the situation in Palaeopython cadurcensis. In one anterior trunk vertebra (UM ECC 2574), the zygosphene has a peculiar, curved, shape in anterior view. The neural arch is arched and can be quite vaulted (e.g., NHMW 2019/0033/0018; Fig. 76b). The shape of the hypapophysis varies. The interzygapophyseal constriction can be shallow (e.g., NHMW 2019/0033/0018; Fig. 76c). The subcentral ridges are usually almost straight in lateral view [e.g., NHMW 2019/0033/0018 (Fig. 76e); NHMW 2019/0033/0019; NHMW 2019/0033/0042]. The subcentral grooves are rather prominent in certain specimens (e.g., NHMW 2019/0033/0019). In posterior-most anterior trunk vertebrae, the neural spine becomes broader in lateral view, and the hypapophyses are rather large.
Mid-trunk vertebrae are moderately large, with CL ranging between 4 and 10 mm, but most specimens toward the middle of this range (see Appendix 1). Mean CL in the NHMW sample was 6.9 mm (Fig. 87); in the YPM sample, 6.16 mm (Fig. 88). In mid-trunk vertebrae, the zygosphene often has three slightly or clearly distinct lobes in dorsal view (UM ECC 2576; NHMW 2019/0033/0055), although there are specimens where these lobes are indistinct (e.g., NHMW 2019/0033/0004) or even that the zygosphene is straight (e.g., NHMW 2019/0033/0037). The zygosphene can be moderately thick, though in many specimens it is thinner (e.g., NHMW 2019/0033/0049; NHMW 2019/0033/0055). On mid-, as on all other trunk vertebrae, the zygosphene is as wide as or wider than the cotyle. The zygosphene often possesses the characteristic median tubercle, although it may be very slightly developed (e.g., NHMW 2019/0033/0054) and is sometimes totally absent (e.g., NHMW 2019/0033/0056). The neural spine is relatively short in lateral view, it is inclined posteriorly, and most usually it overhangs posteriorly (e.g., NHMW 2019/0033/0004). The neural spine commences well behind the zygosphene, a feature that seems to be consistent throughout the trunk column in individuals of different size. Lateral foramina are present and usually prominent. A prominent ridge in the postzygapophyses is clearly visible in lateral view, a feature that can be observed throughout vertebrae of the trunk column. The neural arch is less vaulted than in the anterior trunk vertebrae, the vaulting ratio being varying between about 0.34 and 0.44 independently of centrum length (Figs. 89, 90), and usually the posterior margin of the neural arch is angulated above the lateral edges of the zygantrum. Nevertheless, certain small-sized specimens have much more vaulted and almost triangular-shaped neural arch (e.g., NHMW 2019/0033/0010). In the neural canal, a distinct raised area (“subneural process” sensu Auffenberg, 1963; “epapophysis” sensu Holman, 2000) is usually prominent mid-ventrally (e.g., NHMW 2019/0033/0004), though this feature is not as thick as in specimens of Palaeopython cadurcensis. In one specimen (NHMW 2019/0033/0122), the neural canal has a distinct “trefoil” shape (Fig. 86f, g), but other specimens are similar. The prezygapophyses are dorsally inclined in anterior view (though usually not the same large extent as in Palaeopython cadurcensis). The prezygapophyseal articular facets are large and, almost always, oval in shape. Prezygapophyseal accessory processes are usually small or can be invisible, but rarely are fairly large (e.g., NHMW 2019/0033/0120; NHMW 2019/0033/0126). The posterior median notch of the neural arch is prominent. The cotyle is large and deep. Paracotylar foramina are almost always absent, with few exceptions (i.e. NHMW 2019/0033/0047). The zygantrum is always deep and two deep zygantral foramina are usually visible. In certain cases, a number of small foramina can be present next to the zygantrum. The haemal keel is usually sharp on mid-trunk vertebrae and projects slightly below the condyle, even small specimens (e.g., NHMW 2019/0033/0048). Subcentral foramina are present and are occasionally large (e.g., NHMW 2019/0033/0149). A precondylar constriction is present. The condyle is massive and usually circular, and in one case it is vertically elliptical (NHMW 2019/0033/0047), a rather strange condition for snakes. Longitudinal ridges are found on some specimens on the posterior half of the neural arch (e.g., NHMW 2019/0033/0122; NHMW 2019/0033/0124), and zygantral mounds are commonly present. Additionally, in another specimen (NHMW 2019/0033/0151) there is an exceptionally large dorsal thickening of the neural spine in dorsal view, a feature not uncommon in Constrictores (e.g., the extinct Cheilophis huerfanoensis Gilmore, 1938, Albaneryx depereti Hoffstetter and Rage, 1972, Pterygoboa Holman, 1976, Cadurceryx filholi Hoffstetter and Rage, 1972, and the extant Epicrates cenchria (Linnaeus, 1758); see figures in Hoffstetter & Rage, 1972; Rage, 1984b; Szyndlar & Rage, 2003).
In posterior mid- and posterior trunk vertebrae, the haemal keel becomes wider and thicker in ventral view and the neural arch becomes gradually even more depressed (e.g., the holotype MNHN.F QU16322; UM ECC 2577; NHMW 2019/0033/0002) (Figs. 75, 82i–m, 91, 92 and 93). Certain posterior trunk vertebrae achieve a peculiarly rather vaulted neural arch (e.g., NHMW 2019/0033/0083, NHMW 2019/0033/0084). Some posterior trunk vertebrae reach considerable size, with CL surpassing 10 mm (e.g., the holotype MNHN.F QU16322; NHMW 2019/0033/0035). The subcentral grooves (paralymphatic fossae) are more prominent than on mid-trunk vertebrae. The thick subcentral ridges can be straight or slightly convex and usually show the constriction at mid-length. The neural spine becomes also gradually shorter but retains the posterior inclination. The zygantral mounds are usually found here as well, but there are specimens that lack them (e.g., NHMW 2019/0033/0036). The prezygapophyses are slightly inclined. Prezygapophyseal accessory processes are present in several specimens, especially in large ones (e.g., NHMW 2019/0033/0036). The interzygapophyseal constriction is generally smoothly rounded and moderately deep, extending medially beyond the mid-point of the prezygapophyses, although in some cases it is more shallow and/or squared off (e.g., NHMW 2019/0033/0129). The cotyle is always deep and varies in shape from almost totally circular to slightly elliptical. The paradiapophyses are massive, sometimes extending slightly more laterally than in the holotype (e.g., NHMW 2019/0033/0035). Paracotylar foramina are present in only two specimens [NHMW 2019/0033/0073; YPM-VPPU 30393 (Fig. 93a)], a feature that apparently is rather rare here as in the preceding regions.
In the posterior-most trunk vertebrae, as exemplified by YPM-VPPU 30395, a hypapophysis is present instead of a haemal keel (Fig. 93k–o), a feature known in other booids as well, including species of Bavarioboa (see figs. in Szyndlar and Rage, 2003). The hypapophysis there is relatively prominent and is ventrally exposed across a large portion of the centrum. Other notable features of the posterior-most trunk vertebra YPM-VPPU 30395 are the vaulted neural arch (in comparison with the depressed ones in the preceding posterior trunk vertebrae), as well as the thick neural spine (Fig. 93k–m). Such increase in neural arch vaulting at the posterior-most part of the vertebral column, continuing into the cloacal vertebrae, seems to be relatively widely distributed among Constrictores, both extant and extinct (e.g., Bavarioboa spp., Calamagras weigeli Holman, 1972, and Ogmophis compactus Lambe, 1908) (see Smith, 2013; Szyndlar & Rage, 2003).
Growth rings in the postzygapophyseal articular facets are observed in few well-preserved specimens [e.g., NHMW 2019/0033/0043; NHMW 2019/0033/0122 (Fig. 86h); NHMW 2019/0033/0124; NHMW 2019/0033/0127 (Fig. 91h)].
Cloacal and caudal vertebrae are available for this species (Figs. 94, 95). Anterior cloacal vertebrae (i.e. UM ECC 3095) have a short hypapophysis instead of haemal keel (Fig. 95a–e), a feature that is present also in other fossil and extant constrictors [e.g., Bavarioboa spp. (Szyndlar & Rage, 2003) and Ungaliophiinae (Smith, 2013)]. The prezygapophyseal articular surfaces are nearly horizontal. The zygosphenal roof is flat and much wider than the cotyle. The neural spine is shorter than on posterior trunk vertebrae. In posterior view, the neural arch rises more steeply from its lateral edge, but it then shows an inflection toward the medial so that, apart from the neural spine, it is nearly horizontal around the midline. The dorsal branch of the lymphapophysis extends laterally and slightly ventrally at first, turning then downward.
Proximal caudal vertebrae are similar to the cloacal vertebrae in many respects (i.e. UM ECC 3097; Fig. 96a–e). However, the neural arch is depressed and more like the holotype (without the steep sides and inflection point). The cotyle and condyle are more nearly rounded. The pleurapophysis extends straight ventrolaterally, curving sharply ventrally near its distal end. In lateral view, the pleurapophysis tapers continuously, but in anterior and posterior views it is seen to expand in width at the point where it turns ventrally; the anterior face shows an elongate groove that follows the curvature of the process, whereas the posterior face is flat. The hypapophysis is missing, replaced by anteroposteriorly short, paired hemapophyses present at the posterior-most end of the centrum. Anteriorly on the ventral surface of the centrum, just behind the edge of the cotyle is a peculiar median tubercle with a steeper anterior than posterior margin, reminiscent of the posterior end of a haemal keel on trunk vertebrae. These vertebrae are anteroposteriorly very short.
Middle caudal vertebrae are slightly longer, similar in proportion to mid-trunk vertebrae (i.e. UM ECC 3098; Fig. 96f–j). The neural spine is slightly taller, and the cotyle and condyle have the shape of compressed (dorsoventrally elongate) ovals. Just behind the cotyle, the centrum evinces slight, paired depressions, leading to the development of a structure like a haemal keel. Otherwise, they are similar to more proximal caudals.
Description of new cranial material from Quercy (Figs. 96, 97, 98, 99, 100, 101, 102and 103): In the NHMW sample there are three marginal jaw elements, of which two can be referred to Phosphoroboa filholii comb. nov.: a left maxilla with several preserved teeth (NHMW 2019/0020/0001) (Fig. 96) and a right dentary with few preserved remains of teeth (NHMW 2019/0020/0002) (Fig. 99). Furthermore, another cranial element is in the NHMUK sample: a dentary with several preserved teeth (NHMUK PV R 3489) from an imprecise locality near Mouillac, Phosphorites du Quercy (Fig. 100) is assigned to the same taxon, based on its similarity to the NHMW dentary. Based on the same data-set used above in Palaeopython cadurcensis for studying the size relationship between CL of mid-trunk vertebrae and dentary tooth row length, we calculate that for an average-sized individual in the NHMW sample (CL = 6.9 mm) the expected dentary size is 25.6 mm, only 16% larger than NHMW 2019/0020/0002 (Fig. 87). NHMW 2019/0020/0001 is 10% larger (based on the average length of three tooth loci in the middle of the jaw). Only vertebrae of Phosphoroboa filholii comb. nov. have been identified in the NHMW sample that are even close to those sizes (Fig. 87). And Phosphoroboa filholii comb. nov. is by far the most abundant species of that sample (Fig. 87). For all these reasons we consider it probable that these marginal jaw bones pertain to that species. Another dentary from Quercy, YPM-VPPU 30397 (Fig. 98), can be referred here as well on the basis of similar size and abundance considerations (Fig. 88). Furthermore, three skull elements [a dentary (UM ECA 2544; Fig. 97), a pterygoid (UM ECC 2501; Fig. 101), and a parietal (UM ECA 2543; Fig. 103)] from Escamps are referred to this taxon. A fragmentary pterygoid (NHMW 2019/0020/0004) is also present in the NHMW collection (Fig. 102).
While we refer all these specimens to Phosphoroboa filholii comb. nov. for reasons given also below, we note that there are slight, but consistent, morphological differences in the samples. In particular, as described below, the Escamps and YPM-VPPU dentaries (Figs. 97, 98) share certain features (morphotype 1), and the NHMW and NHMUK dentaries (Figs. 99, 100) share others (morphotype 2). Similarly, the pterygoid from Escamps (Fig. 101) is slightly different from the NHMW pterygoid (Fig. 102).
The only maxilla, NHMW 2019/0020/0001, retains 12 tooth positions (Fig. 96). The teeth are strongly distally reclined, even more so than in the dentary YPM-VPPU 30397. Tooth size decreases gradually but consistently from front to back. There is a strongly posteriorly inclined, parallel-sided palatine process with a hook-shaped distal tip, as in booids. On the anterodorsal surface of the base of the process is a large foramen, and probably a second foramen is present above the tooth row posterior to the level of the palatine process. In contrast to Palaeopython cadurcensis and Messelopython freyi, the lateral margin of the bone is straight.
The dentaries are similar in being relatively deep with a single, large mental foramen located beneath the 5th tooth from the front (Figs. 97, 98, 99 and 100). The foramen is oval in lateral aspect except in NHMW 2019/0020/0002, where it is circular (Fig. 99a). The shape of the subdental shelf is almost identical, as far as they are preserved. In YPM-VPPU 30397 there are 17 preserved tooth positions, and possibly one or two are missing from the posterior end (Fig. 98). In NHMW 2019/0020/0002 the bone is ventrally bowed (Fig. 99), unlike in any other specimen. There are also 17 preserved tooth positions in this specimen; probably one or more teeth are missing in this specimen as well. In UM ECA 2544 there are 18 preserved tooth positions (Fig. 97). Given that the ventral margin of that dentary is incomplete, it appears that the posteroventral and posterodorsal processes were more similar in length than in Palaeopython cadurcensis.
There are two consistent differences between dentary morphotypes 1 (UM ECA 2544 and YPM-VPPU 30397) and 2 (NHMW 2019/0020/0002 and NHMUK PV R 3489). First, the surangular notch is much deeper in morphotype 1, more similar to other constrictors. In particular, in UM ECA 2544 and YPM-VPPU 30397 (Figs. 97, 98) the notch extends as far as the 12th tooth and well anterior of the splenial facet on the posteroventral prominence (sensu Scanlon 2001). In contrast, in NHMW 2019/0020/0002 (Fig. 99) the surangular notch terminates beneath the 17th tooth and at the anterior end of the posteroventral prominence; in NHMUK PV R 3489 (Fig. 100), the surangular notch must terminate at a level at or posterior to that in NHMW 2019/0020/0002. This reduced surangular notch appears to be a derived condition. Second, in morphotype 2 (NHMW 2019/0020/0002 and NHMUK PV R 3489) the posterior end of the dentary shows a lateral bulge that is absent in morphotype 1 (UM ECA 2544 and YPM-VPPU 30397).
The most complete pterygoid is UM ECC 2501, which lacks much of the quadrate ramus (Fig. 101). The anterior end is wedge-shaped, with a small medial projection just beyond the tooth row. The palatine articulated on the medial and lateral side of this end. On the ventromedial side of the pterygoid, just below the palatine articulation, is a medial process. Such a process has been noted for booids (Rage, 1974); in particular, we find it to be present in Boa spp., Chilabothrus angulifer (Bibron, 1840 in Ramón de la Sagra, 1838–1843) (SMF-PE 61), and Sanzinia madagascariensis (Duméril & Bibron, 1844) [SMF-PH 56], but not in Corallus caninus (Linnaeus, 1758) (SMF-PH 74), Epicrates cenchria (SMF-PH 26), Eunectes notaeus Cope, 1862 (SMF-PH 60), Chilabothrus striatus (Digimorph), or Candoia spp. (Digimorph). Such a process is not known in Pythonoidea or in Eoconstrictor (see above). The ectopterygoid articulation is a lateral concavity, taller posteriorly than anteriorly. It is not set in a prominence, unlike in Eoconstrictor (see above). As in Eoconstrictor and many boids and pythonids, there is a medial flange at the basipterygoid articulation that appears to rotate into a dorsal keel, although this is incompletely preserved.
NHMW 2019/0020/0004 is the anterior tip of a left pterygoid (Fig. 102). Only the anterior-most end of the pterygoid, the palatine articulation, is preserved in this specimen. It is more robust than UM ECC 2501, and it does not have a distinct anteromedial process, only a low prominence in the same position. It contains eight tooth spaces with five partial teeth. These are relatively small and do not appear to change in size from anterior to posterior. The palatine facet is extensive on the dorsal surface, running as far as the level of the 7th tooth. It is elaborated by a small wedge-shaped facet on the medial side at the anterior end. There is strong, longitudinal dorsal keel extending from the anterior end to the level of the 3rd tooth, which would have inserted in a deep cleft at the posterior end of the palatine.
The parietal UM ECA 2543, which we refer to Phosphoroboa filholii comb. nov. on the basis of size and abundance (see below), is the sole known such snake element in the Phosphorites du Quercy (Fig. 103). The specimen is nearly complete. In dorsal view, it is nearly T-shaped, with strong postorbital processes (Fig. 103a). The frontoparietal suture is relatively narrow, only one-third the width of the bone, and V-shaped, with a much sharper midline notch than in Eoconstrictor. The swellings for the cerebral hemispheres are present, but not very prominent. The dorsomedian keel extends along the entire preserved length of the bone, up to the frontoparietal suture. Posterolateral processes appear to be much weaker than in Eoconstrictor, assuming they are reasonably complete. UM ECA 2543 preserves in ventral view the articulations with the chondrocranium (Fig. 103b). The descending flanges that embraced the parabasisphenoid rostrum approach one another closely anteriorly and diverge somewhat posteriorly. Two bilateral pairs of projections are present here, one medially at the trabeculae, and one laterally that bounded the optic foramen. The latter structure deeply notches the parietal.
Justification on the referral of the cranial material from Escamps: The late Eocene (MP 19) fossil sites of Escamps in the Phosphorites du Quercy have yielded a diverse squamate assemblage, consisting of several taxa of lizards and snakes, as presented in the faunal lists of Bonis et al. (1973) and Crochet et al. (1981). However, despite the number of snake taxa presented in those two preliminary faunal lists, descriptions and figures of snake remains from Escamps have so far only been provided by Zittel (1887–1890) and Rage (1974).
Indeed, Zittel (1887–1890) figured a vertebra that he assigned to Palaeopython cadurcensis, a specimen which we here refer to Palaeopython cf. cadurcensis (see that entry above). Under the framework of their new, stratigraphically well-constrained collections in the Phosphorites du Quercy, Bonis et al. (1973) correlated Escamps with other nearby localities and identified three fossiliferous sites, namely Escamps A, B, C. Accordingly, specimens collected previously to that time from the area of Escamps, such as Zittel’s (1887–1890) vertebra, cannot be precisely attributed to any of the three known sites of Escamps and it can even be the case that it originated from some nearby, slightly contemporary, site (see also “Remarks” below). Bonis et al. (1973) mentioned in their list four different snake taxa from Escamps, for which they used the informal denominations “Boidae A”, “B”, “C”, and “D”, without any kind of description or figure. The following year, Rage (1974) described vertebral and cranial material from Escamps A and C, which he referred to Palaeopython filholii. He additionally mentioned that ribs and a fragment of a large dentary from Escamps (none of them figured) matched in size to Palaeopython cadurcensis and could pertain to that species. Crochet et al. (1981) provided an updated faunal list of snakes from Escamps, recognizing the following ten taxa: Coniophis sp., Eoanilius europae, Palaeopython filholi (sic), Dunnophis cadurcensis, Cadurcoboa insolita, “Boidae C”, “D”, “K”, “L”, and “M”. Again, none of these was accompanied by any kind of description or figure.
In an attempt to properly refer the available cranial elements from Escamps described herein, it is essential to realize how many and which snake taxa were present in the locality. Of course, this is ultimately a difficult task, considering the total absence of descriptions or figures, and the nature and number of specimens of each taxon, particularly for forms that were not assigned to the genus level, such as all these enigmatic “Boidae C”, “D”, etc. (on these lettered taxa, see Smith and Georgalis, in press). It is first apparent that the Escamps faunal list of Crochet et al. (1981) is the updated version of the even more preliminary one of Bonis et al. (1973), following the intervening studies of Jean-Claude Rage, who participated as coauthor in both these 1973 and 1981 papers and who also in the meantime published significant results of his studies on Quercy snakes (Rage, 1974, 1978). As a result, several of the “lettered Boidae” of Bonis et al. (1973) were more precisely assigned in Crochet et al. (1981) and the total number of distinct taxa was also reconsidered.
Of these 10 taxa from Escamps listed in Crochet et al. (1981), Phosphoroboa filholii comb. nov. is by far the most well documented and the only one described and figured by Rage (1974) from that locality. The same species was relatively abundant, as Rage (1974:284) explicitly stated “Escamps a fourni de nombreux éléments (vertèbres, côtes, un ptérygoïde gauche) appartenant à P. filholi (sic)”. The referral of the pterygoid UM ECC 2501 to Phosphoroboa filholii comb. nov. by Rage (1974) was made by means of size comparison, as he stated that all other “boids” (i.e. Constrictores in the modern sense) from Escamps were significantly smaller (“sa taille ne laisse aucun doute sur son appartenance spécifique, les autres Boidae du même gisement étant beaucoup plus petits que P. filholi (sic)”; Rage, 1974:285–286). Indeed, Coniophis sp., Eoanilius europae, Dunnophis cadurcensis, and Cadurcoboa insolita are all tiny snakes, with centrum lengths not surpassing 3 mm. Boidae M was later subsequently described by Rage (1988b) as Platyspondylia sudrei, and it is indeed considerably smaller than Phosphoroboa filholii comb. nov. Our preliminary investigation of material of Boidae D, K, and L, confirms that they too are tiny snakes. For the remaining Boidae C, there is no information, as it has never been formally described, but we accept the view of Rage (1974) that this should also be smaller than Phosphoroboa filholii comb. nov. Finally, as noted above, the largest snake in Escamps should be a similar form to Palaeopython cadurcensis [though not mentioned in Crochet et al. (1981)], judging from the mention of Rage (1974) and perhaps the vertebra figured by Zittel (1887–1890).
Therefore, on the basis of size and relative abundance, the parietal UM ECA 2543 and the dentary UM ECA 2544 from Escamps A, and the pterygoid UM ECC 2501 from Escamps C, can all be referred to Phosphoroboa filholii comb. nov. The identification of these three cranial elements enables a better understanding of the relationships of this taxon within Constrictores, suggesting affinities with Booidea. In addition, they can further provide direct means of comparisons with other isolated skull elements found in imprecisely known localities in the Phosphorites du Quercy.
The populations of Escamps and YPM-PU (if the latter are indeed a single population) share the same dentary morphology (dentary morphotype 1). As both populations also preserve posterior trunk vertebrae which are nearly identical to the holotype of Phosphoroboa filholii comb. nov., they are best viewed as representing the same species. Note also that all YPM-PU specimens originate from the area around Mouillac, which is not geographically far from Escamps. The NHMW and perhaps NHMUK samples are slightly different in dentary (i.e. dentary morphotype 2) and pterygoid morphology, and it is not yet clear what these differences mean. The greater robustness of the NHMW dentary and pterygoid specimens could represent sexual dimorphism, but the difference in the depth of the surangular notch of the dentary raises the prospect that they could represent different taxa.
Remarks: Originally established as Palaeopython filholii by Rochebrune (1880) upon a single trunk vertebra (MNHN.F QU16322) from the Phosphorites du Quercy, the same author later referred to this species three articulated vertebrae (MNHN.F QU16323), as well as the syntype (now paralectotype) maxilla (MNHN.F QU16321) of Python cadurcensis that was previously described by Filhol (1877a, 1877b, 1877c) (Rochebrune, 1884). Following his suggested synonymy of Palaeopython with Paleryx, Lydekker (1888a) placed Palaeopython cadurcensis into the latter genus, thereby also moving Palaeopython filholii to Paleryx. In the very same year, the same author went further and considered Palaeopython filholii as a probable junior synonym of the British Paleryx depressus (Lydekker, 1888b). Rage (1974) concluded that the referred specimens of Rochebrune (1884), i.e. the three articulated vertebrae MNHN.F QU16323 and the maxilla MNHN.F QU16321, cannot be assigned to the same species as the holotype vertebra of Palaeopython filholii but instead were more reminiscent of Palaeopython cadurcensis. He further described additional vertebrae and the pterygoid UM ECC 2501 from Escamps, which he referred to Palaeopython filholii (Rage, 1974).
A confusion arises when one deals with the plates and plate labels of Rochebrune’s (1880,