The Oldest Articulated Ranid from Europe: A Pelophylax Specimen from the Lowest Oligocene of Chartres-de-Bretagne (N.W. France)

ABSTRACT Ranids represent an important part of the extant anuran diversity of Europe. One of the best-known genera is Pelophylax (green-water frog). This genus is considered to have arrived in Europe during the Eocene/Oligocene transition, with numerous occurrences of the genus throughout European Oligocene sites. Unfortunately, most of the specimens are isolated bones, hampering our understanding of the diversity and evolution of the genus during this time. We here present the description of an incomplete but articulated anuran skeleton from the lowest Oligocene of Chartres-de-Bretagne (western France). This specimen, missing its head, preserves almost all postcranial bones articulated and skin impressions. The osteological description allows to assign this specimen to Pelophylax kl. esculentus, making it one of the oldest known occurrences of the genus. We also suggest that specimens assigned to the late Oligocene “Rana” aquensis should be referred to the genus Pelophylax. The presence of a Pelophylax in western Europe during the early Oligocene indicates that the genus had already spread throughout Europe, no later than 5 Ma after its emergence in the eastern part of the continent. It suggests that Pelophylax benefitted from the extinction of ranoids during the “Grande Coupure.”


INTRODUCTION
Anurans have been recovered from numerous Cenozoic sites in Europe, documenting a rich and diverse history of the clade during this period. In particular, the Quercy Phosphorites (mostly Eocene/Oligocene deposits; Pélissié et al., 2021) have yielded hundreds of anuran bones, representing most extant anuran clades present in Europe (Rage, 2016). The Phosphorites also document the turnover within neobatrachians anurans during the "Grande Coupure," with the establishment of "true" ranid taxa during the Oligocene (Rage, 2016;Vasilyan, 2018). Among them, the genus Pelophylax (green-water frogs) is thought to have appeared and diversified during the early-middle Oligocene (Chan & Brown, 2017). Unfortunately, the genus is represented only by isolated bones from a single site in the lower Oligocene (Sanchíz et al., 1993). Two taxa from the upper Oligocene/Lower Miocene, "Rana" aquensis Coquand, 1845 (southern France) and "Rana" meriani Meyer, 1852 (southern Germany), known by several complete specimens, have been informally assigned to the genus, but several uncertainties remain regarding their attribution (Sanchíz, 1998). Thus, the evolution of the geographic range of Pelophylax is almost unknown during a crucial phase of its spread across western Europe, as the genus is present in most of western Europe during the Miocene (Rocěk, 2013). Discovery and identification of specimens of Pelophylax from this region would help our understanding of the dispersal and evolutionary history of the genus during the Oligocene.
Our study is focused on the description and identification of a specimen from this geographic area (Fig. 1A, B). The specimen was discovered on July 7, 1922, by the geologist Yves Milon in the Grands-Fours quarry near the city of Rennes (Brittany, western France; Fig. 1C, D). Two days later, the specimen was presented at a session of the Société Géologique et Minéralogique de Bretagne (Dangeard & Milon, 1922). Unfortunately, Milon never described the specimen, and it was considered lost for several decades. One of us (D. Gendry) recently rediscovered this specimen, along with Milon's identification (an anuran) in the collection of the geological museum of the University of Rennes 1. We took this opportunity to describe the osteology of the fossil and propose a taxonomic attribution. We also discuss its implication for the Oligocene occurrence of ranid taxa in Europe.

GEOLOGICAL CONTEXT
Chartres-de-Bretagne is an ancient complex of quarries located 5 km south of the city of Rennes, within the Cenozoic Basin of Rennes (Fig. 1A). The quarry produced limestone, with numerous open pits present (Vasseur, 1881;Trautmann et al., 2000;Lebrun et al., 2017). The sites are unfortunately backfilled today, and no further investigations are possible.
The Oligocene beds of Chartres-de-Bretagne are thick (380 m thick) and have been well-known since the end of the 19th century (Lebesconte, 1879;Vasseur, 1881;Allix, 1922;Dangeard FIGURE 1. Location of the discovery in the Grands-Fours quarry (Chartres-de-Bretagne). A, geological map of the Chartres-de-Bretagne basin with location of the quarry (black circle; modified from Bauer et al., 2016); B, stratigraphic section of the quarry (modified from Bauer et al., 2016) with C, photography of the outcrop taken by Yves Milon in March 1922, note the black clay with quarrymen in the Upper Sapropels that yielded the anuran remains just above paleosols; D, close-up on the black clay with Yves Milon and two quarrymen, position of hammers indicates first fish remains discovered March 17, 1922, the anuran comes from the left hammer level (photography Dangeard). Labels at the bottom are for the geological map of A and black star (red in online version) represents the position of IGR 144547 within the stratigraphic log. Abbreviations: Pleis., Pleistocene. & Milon, 1926;Lebrun et al., 2017). The Lower Sapropels Formation is the most basal formation. It forms the majority of the Oligocene deposits (around 280 m) and represents a major regression during the early Rupelian. This bed is overlain by the Natica crassatina marls and "Archiacina" limestone formations (52 m thick). These two formations were deposited during a transgressive event that lasted throughout the early Rupelian (Fig. 1B). The uppermost Oligocene formation within the region is the Upper Sapropels Formation (Rupelian; around 6.5 m thick). Within this formation, the foraminiferans Arenagula kerfornei and Archiacina armorica (Peneroplis armorica in Bauer et al., 2016) have been recovered (Bauer et al., 2016;Lebrun et al., 2017). In a previous study (Bauer et al., 2016), the foraminiferan Archiacina globula was also mentioned. However, it is a subjective junior synonym of Arenagula kerfornei (World Register of Marine Species: https://www.marinespecies.org). Both Arenagula kerfornei and Archiacina armorica are early Rupelian markers for the SBZ 21 biozone (Cahuzac & Poignant, 1997;Bauer et al., 2016). Notably, they are not present in the following SBZ 22 biozone (Arenagula is only present in the SBZ 22A of the Aquitaine Basin; Cahuzac & Poignant, 1997), to which a recent study (Bauer et al., 2016) attributed the Upper Sapropels Formation (at least part of it). Thus, we do not agree with this attribution, and consider the entire Upper Sapropels Formation early Rupelian in age (SBZ 21 biozone; MP21-22; Speijer et al., 2020).
Within the formation, its lower part is composed of greenish interbedded claystones, marlstones, and limestones. The flora is rich and made up of hydrophilous taxa while the fauna is composed of lacustrine taxa (fishes and mollusks; Kerforne, 1915;Milon & Dangeard, 1920;Depape, 1926;Milon, 1930Milon, , 1935Milon, , 1936Gaudant, 1989;Bauer et al., 2016;Lebrun et al., 2017). A small fragment of a crocodilian osteoderm was also collected. All these sedimentological and paleontological elements are consistent with a restricted and quiet lacustrine environment. The blackish color of the Upper Sapropels Formation is also an indication that the depositional environment was rich in humic materials, and the lake bottom was likely anoxic (Milon, 1930(Milon, , 1935Fig. 1C). Roots, rhizomes, and leaves are exceptionally well-preserved (Lebrun et al., 2017:pl. B, C). In addition, all known vertebrate specimens (except the osteoderm) are preserved as articulated skeletons with soft tissues (skin impressions) preserved (Lebrun et al., 2017:pl. B, C). Therefore, these laminated deposits likely represent a Konservat-Lagerstätte owing to the exquisite preservation of the fossils. The single anuran specimen was discovered within the Grands-Fours quarry (La Chaussairie) from the Upper Sapropels Formation (around 30 cm below the top of the bed).

MATERIALS AND METHODS
Institutional Abbreviations-IGR, Institut de Géologie de Rennes, Rennes, France; MNHN, Muséum National d'Histoire Naturelle, Paris, France; MNHN.F.AIX, paleontological collection from Aix-en-Provence, France. The specimen is stored within the collection of the geological museum of the University of Rennes under the following number: IGR 144547. Both main slab (M1) and counterpart slab (M2) are cataloged under the same number.
Reflectance Transformation Imaging (RTI)-Most bones are crushed or preserved as partial imprints and thus difficult to identify. RTI is a method that computes an "interactive specimen" in which illumination can be oriented at will (Hammer et al., 2002). This method was used with a custom-made portable light dome (same as used in Cui et al., 2022). Sets of 54 photographs under different LED sources were compiled using the RTIBuilder software. The resulting RTI files (see Supplementary File: 10.5281/zenodo.7223828) can be opened using the RTIViewer software. Both RTIBuilder and RTIViewer are freely available at www.culturalheritageimaging.org. This study is one of the few using RTI to study fossil vertebrate specimens.

Position of the Specimen on the Slabs
The preserved portion of the specimen is 56 mm long (anterior end of ?omosternum to distal end of femora) on a 101 × 107 mm slab (M1) with its 69 × 63 mm partial counterpart (M2). The slab (M1) preserves a majority of the fossil, from base of the cranium to the femora ( Fig. 2A, B). The partial counterpart (M2) preserves only the posterior region of the body and the hands (Fig. 2C, D). The majority of the bones are surrounded by a thin black veil interpreted as a skin impression (see below; Fig.  2A, C). The bones of the specimen are preserved as (1) slightly crushed bony elements, (2) crushed and fragmentary bony elements; (3) imprint of the bone; and (4) a mix of crushed bony element and imprint on both plates. As M1 is the main plate, the position and lateralization of the anuran skeleton are based on this slab. The imprint left by the humerus on the right side of M1 (see Fig. 2) preserves some detail distally (Fig. 2). Compared with the humeral imprint made from an extant Pelophylax (see Fig. S1), the fossa cubitalis and eminentia capitata are preserved as an imprint. Thus, we interpret the humeral imprint on IGR 144547 as visible in ventral view on M1. Hence, when the bone is preserved on M1, it is visible in dorsal view. Thus, IGR 144547 presents an anuran specimen in dorsal view on M1, and in ventral view on M2. When only the imprint of the bone is preserved, it is visible in ventral view (on M1, opposite for M2). Hence, on M1, the left side of the specimen is visible on the left region of the plate, while it is preserved on the right region of the plate on M2.

Soft Tissues
The soft tissues surround most of the bones, except for the skull and part of the tibiofibula on both M1 and M2 ( Fig. 2A, C). Although it is not well defined in most regions, the preservation around the hands, femora, and cloacal region allows interpretation of this feature as skin impressions. The body shape is hard to define on the torso, although it seems to widen at midlength ( Fig. 2A). The body outline then narrows towards the cloacal region.
The skin around the shoulder and forelimb is poorly preserved. Skin around the hands is well preserved on both M1 and M2 ( Fig. 2A, C). The left hand preserves the shape of the flesh around the digits, and the skin narrows towards the last phalanx of the digits. Around the femora, the skin is wider near its proximal margin, and progressively narrows distally until the end of both femora. On M2, small patches of skin imprints seem preserved around the left tibiofibula (Fig. 2C, D). FIGURE 2. IGR 144547 main slab (M1) and its counterpart (M2). A, IGR 144547 M1, B, interpretative drawing of the same specimen; C, IGR 144547 M2 and D, interpretative drawing of the same specimen. Light gray areas represent regions on the specimen where bony elements could not be delimited, dark gray areas represent skin impressions. Abbreviations: ca, carpals; cl, clavicle; co, coracoid; f, femur; hu, humerus; il, ilium; mtc, metacarpal; omst?, putative omosternum; phx I, II, III, phalanx I, II, III; rdul, radioulna; sc, scapula; tf, tibiofibula; ur, urostyle. Scale bars equal 10 mm.

Osteological Description
IGR 144547 is an anuran skeleton missing its skull, its anterior vertebrae, most of the left elements of its pectoral girdle and most of its hindlimbs (Fig. 3). The skeleton is mostly articulated, but poorly preserved, in particular in the main body region (Figs. 2, 3).

Vertebral Column
General Comments-All vertebrae are badly preserved. Most of them are hollowed out, with only faint traces of centra and transverse processes preserved. Where bone is still present, the vertebrae are exposed in dorsal view, but the neural region has been crushed or eroded on all vertebrae, leaving the centrum exposed (Fig. 4). Presacral and sacral vertebrae are only preserved on M1. A single sacral vertebra is identified, while five presacral vertebrae are preserved. The atlas, second and third vertebrae are not preserved, due to a missing piece on M1 ( Fig.  2A). IGR 144547 likely possessed eight presacral vertebrae.
Vertebra IV-The centrum and transverse processes of the fourth vertebra are preserved as crushed bone in dorsal view. The centrum is broken anteriorly. It connects posteriorly with the fifth vertebra and on its right lateral side with the sternum (Fig. 4A, B). The base of the left transverse process shows that the processes extend slightly posteriorly. They seem to expand distally ( Fig. 4B).
Vertebra V-The centrum of the fifth vertebra is preserved as crushed bone in dorsal view. It bears an anterior cotyle and a posterior condyle. It connects anteriorly to the fourth vertebra and posteriorly to the sixth vertebra. A broken transverse process might be preserved on the left lateral side of the centrum (Fig.  4B). It seems posteriorly oriented, like the transverse processes of the seventh vertebra (see below), but it is broken from the centrum and we cannot assess if this is its in situ orientation.
Vertebra VI-Only the procoelous centrum of the sixth vertebra is preserved as crushed bone in dorsal view. It connects anteriorly to the fifth vertebra and posteriorly to the seventh vertebra. No transverse processes are preserved.
Vertebra VII-This vertebra has the longest preserved centrum (2.2 mm). It bears anterior cotyle and a posterior condyle and is connected anteriorly to the sixth vertebra and posteriorly to the eighth vertebra. The transverse processes are partially preserved. They are crushed and are separated from the centrum. They are rather long and seem slightly posteriorly oriented. However, the base of the left transverse process appears to be preserved (Fig. 4A, B) and is laterally oriented.
with the remaining presacrals procoelous is diagnostic for a diplasiocoelous vertebral column. A small remnant of the left transverse process is preserved on M1 (Fig. 4A, B). It shows that the processes were straight and extended laterally.
Sacral Vertebra-The sacral vertebra is only preserved on M1. Its centrum is preserved as a hollow structure exposed dorsally (Fig. 4A, B). Nevertheless, it shows that an anterior condyle and two posterior condyles are present. The centrum is short (2 mm) and wider than longer (2.6 mm width). The centrum seems to widen posteriorly (Fig. 4B). Both sacral transverse processes are preserved on M1 as imprints (in ventral view) with bone fragments. The right sacral transverse process is incomplete. The processes are oriented posterolaterally. They barely expend distally and instead are slender rod-like structures. Thus, we interpret IGR 144547 as possessing a "Type IIA" sacrum (Emerson, 1979;Reilly & Jorgensen, 2011). The sacral vertebra possesses a small diapophyseal expansion. Both sacral processes connect distally with the ilium.
Urostyle-The urostyle is preserved on both M1 and M2. The proximal portion of the urostyle is preserved on M1 as crushed bone exposed in dorsal view on M1 and in ventral view on M2. The urostyle shaft is covered by sediment for most of its length on M1. On M2, the shaft is preserved as crushed bone (Fig.  2C). The urostyle is long (17 mm). It connects anteriorly to the sacral vertebra. Although poorly preserved, it is clear that two cotyles were present on its anterior surface. Most of the urostyle shaft is hollowed. The urostyle narrows progressively distally (Fig. 4C, D). Its distal end is crushed and covered by sediment on both M1 and M2. The urostyle seems to end slightly anterior to the acetabular region of both ilia (Fig. 2B, D). On M2, the anterior region of the urostyle is preserved as an imprint exposed dorsally, with a thin lamina of bone present medially (Fig. 4C). It suggests that a dorsal crest is present on the urostyle of IGR 144547 and extended for at least half of its length. However, we cannot assess the height of the dorsal crest, nor its total extension on the length of the urostyle.

Pectoral Girdle
Scapulae-The right scapula is preserved on M1 as an imprint with crushed bone. The left scapula is not preserved on M1. Scapulae are not preserved on M2. The scapula is elongate (7.81 mm length), with a shaft extended laterally (Fig. 5A, B). The shaft scapula expands distally, reaching its widest point at the suprascapular margin ( Fig. 5B; 4.42 mm width). The latter is straight. The anterior and posterior margins are concave. Both margins lack laminae. The pars acromialis is mostly preserved as an imprint exposed in ventral view (Fig. 5A, B), presumed near the glenoid fossa. The pars acromialis is wide (3.3 mm), and part of this processus is preserved as bone exposed in dorsal view in the glenoid region (see Fig. 5B). The glenoid region is poorly preserved, with undefined crushed bony elements visible (Fig. 5B). As the specimen is exposed dorsally, it suggests that most of the bony elements form the base of the pars glenoidalis, distinct from the crushed pars acromialis (Fig. 5B). Thus, it is not possible to assess the size of the pars glenoidalis and if the latter and the pars acromialis were separated by a sinus glenoidalis.
Clavicles-Clavicles are only preserved on M1. The clavicle articulates with the scapula and humerus (forming a part of the glenoid fossa). The right clavicle is partially preserved as an imprint (exposed in ventral view; Fig. 5A), while the left clavicle is not preserved. The extremitas medialis is not preserved. Thus, it is not possible to assess if the latter articulated with the medial margin of the coracoid. The clavicle is slender and moderately developed (5.89 mm preserved). Its shaft is slightly arched. The glenoid region is partially obscured by fragments of unidentified bone (Fig. 5A, B).
Coracoids-Both coracoids are preserved on M1 as imprints exposed in ventral view ( Fig. 2A). The right coracoid is almost complete. It forms part of the glenoid fossa laterally, thus articulating with the clavicle and humerus. It also articulates medially with the left coracoid. The coracoid is shorter (6.64 mm) than the scapula. The processus glenoidalis widens laterally. It bears two distinct articular facets on its proximal surface. The anterior facet articulates with the clavicle and the posterior facet received the humerus. The shaft is elongated transversely and narrow. The processus epicoracoidalis is anteroposteriorly elongated. It is the widest region of the coracoid (5.3 mm). The processus is flat and bears a convex medial margin (Fig. 5B). It slightly extends anteriorly, but does not seem to form an anterior hook (as seen in Thaumastosaurus; Laloy et al., 2013;Lemierre et al., 2021).
Both coracoids were likely in contact medially when accounting for the connecting cartilage of the coracoids (Ecker, 1889). This is characteristic of the firmsternal condition for the pectoral girdle (sensu Cope, 1864;Boulenger, 1886).
Sternum-The sternum is preserved as a crushed bone (exposed in dorsal view) within the space left by the vertebral column on M1 (Fig. 5B). The reconstruction shows that the bone is fully ossified (Fig. 5C). The bone is elongated anteroposteriorly (4.5 mm) and hourglass-shaped. Both anterior and posterior regions are wider than the shaft and the anterior region is wider than the posterior.
Omosternum?-A small, poorly preserved element is preserved anterior to the preserved medial region of the right clavicle on M1. It is very small (2.24 mm length) and crescentic (Fig.  5A, B). It likely represents around half of a bony element. A putative reconstruction indicates the element had an inverted Y-shape (Fig. 5C). The posterior end seems connected to the extremitas medialis of the clavicle. The remaining part of the bone is a slender stylet (Fig. 5B). The position and shape of this element is incompatible with a vertebral element. Thus, we tentatively interpreted this element as an incomplete omosternum. As for the non-overlapping coracoids, the presence of an ossified omosternum is diagnostic for a firmsternal pectoral girdle (sensu Boulanger, 1886).

Pelvic Girdle
Ilia-Both ilia are preserved on M1 and M2. M1 preserves most of the distal region and shaft of the ilia as flattened bones, exposed in medial view ( Fig. 2A). Both distal ends are in contact with the posterior region of the sacral apophyses. The acetabular region is only preserved as an imprint exposed in lateral view of the left ilium on M1 ( Fig. 2A). On M2, the left ilium is preserved in its acetabular region with most of the bone intact and exposed in lateral view (Fig. 2B, 6B). The right ilium has been broken in two. Its distal region on M1 is crushed and separated into two halves ( Fig. 2A). The shape of the displaced sediment suggests it could have been the result of damage during the splitting of M1 and M2 by the collector. The acetabular region is displaced and covers part of the right femur and the dorsal acetabular expansion of the right ilium on M2 (Fig. 2C). This part of the right ilium is preserved as a bone exposed in medial view. The distal region of both ilia is preserved on M2 as imprints exposed in medial view. The ilium is around 19.5 mm long. It is slightly longer than the urostyle. The iliac shaft is slightly compressed laterally. The shaft bears a high dorsal crest (same height as the iliac shaft) on its whole length ( Fig. 2A), as seen on M1. The dorsal protuberance ('tuber superior' in Bailon, 1999) is a large pyriform structure protruding laterally posterior to the dorsal crest (Fig. 6B, C). The supracetabular fossa is well-developed (Fig. 6B, C). The dorsal acetabular expansion is not fully preserved but seems short and oriented posteriorly (Fig. 6C). The ventral acetabular expansion has a large ventral vector (Fig. 6B, C). Its preacetabular angle is Lemierre et al.-Oligocene Pelophylax from France (e2191663-6) clearly acute (Fig. 6C). In medial view, there is no interiliac tubercle.

Forelimbs
Humeri-Both humeri are preserved only on M1. The proximal regions of both humeri are preserved as both crushed bone and imprint, while the remainder of the diaphysis and distal region is preserved only as an imprint exposed in ventral view. The right humerus is around 11 mm long. The proximal head of the right humerus is hard to identify, as the glenoid region is poorly preserved (Figs. 5, 7D). The proximal region of the left humerus is preserved mostly as crushed bone (exposed in dorsal view) but its outline is difficult to determine (Fig. 7A,  B). No crista ventralis is preserved. If present, the crista did not extend past the mid-diaphysis (region preserved as a ventral imprint; Fig. 7C). The distal region of the right humerus is preserved as an imprint exposed ventrally. The fossa cubitalis is shallow and slightly triangular shaped (Fig.  7C, D). The fossa does not extend on the diaphysis. The eminentia capitata (humeral ball in Gómez & Turazzini, 2021) is wide and large (Fig. 7C, D). This structure is partially preserved on the left humerus as eroded bone (thus visible in dorsal view; Fig. 7A) and on the right humerus as an imprint (visible in ventral view; Fig. 7C). The eminentia capitata is not shifted laterally from the central axis of the diaphysis. The epicondylus radialis (lateral epicondyle) is partially preserved on the right humeral imprint (Fig. 7D). It is distinct from the eminentia capitata but does not extend distally past the humeral ball (Fig. 7C, D). The epicondylus ulnaris (medial epicondyle) is not preserved, hidden by the imprint of the radioulna (Fig. 7C).
Radioulnae-Only the right radioulna is preserved for its entire length on M1. Most of it is preserved as a ventrally exposed imprint (Fig. 7C). Only the distal region of the left radioulna is preserved on M1 as a ventrally exposed imprint. The distalmost portion of the left radioulna is preserved mostly as crushed bone on M2. The right radioulna (around 9 mm) is slightly shorter than the humerus (around 80% of total humeral length). The proximal region is articulated with the distal region of the humerus (Fig. 7C). The remaining imprint of the articulation did not preserve the shape of the proximal region. The diaphysis widens distally (Fig. 7D). It reaches its widest point at the level of the distal margin. Although badly crushed on M2, the presence of a shallow groove at mid-width is discernible (Fig. 7D). This groove allowed to discern both radial and ulnar articulation facets with the carpals (Fig. 7D).
Hands-Both hands are not fully preserved. The left hand preserves most elements, except for one phalanx on both M1 and M2 ( Fig. 2A, C). Four digits are visible, with the following phalanx formula: 2-2-3-3. The right hand is poorly preserved on M1 ( Fig. 2A).
Carpals-All carpals are preserved on both hands ( Fig. 2A). Unfortunately, they are poorly preserved as crushed bones exposed in dorsal view. However, most of their shape has been preserved as imprint. Left and right carpals are preserved on M1 ( Fig. 2A). Only the left carpals are preserved on M2 (Fig.  2C). The description is based on carpals from the left hand, preserved in dorsal view (see Fig. 8A). The carpals of the left hand seem preserved in anatomical position. Two rows are preserved. The proximal row preserves three bones (left to right): the ulnare, radiale, and distal radiale (Fig. 8B). The ulnare (left lateral imprint) is roughly ellipsoid, slightly elongated transversely (Fig. 8B). It articulates with the ulnare part of the articular surface of the radioulna, the radiale, and the distal ulnare (Fig. 8B). Three distinct articular facets are present on the ulnare. The posteromedial region of this element is slight protruding towards the radiale (Fig. 8B). This protrusion is interpreted as the fused intermedium (following Fabrezi and Alberch, 1996).
The radiale is the largest (both in width and length) of the proximal carpals. Its shape is difficult to discern but seems circular (Fig. 8). It articulates with the radial part of the articular surface of the radioulna, the ulnare, the distal radial element, and the distal carpalia III and IV (Fig. 8B). All four articular facets are concave (Fig. 8B).
The distal radial element is free. It articulates with the radiale, the distal carpale II, and the prepollex (Fig. 8B). The distal radial element is crescent-shaped, and its lateral margin is concave (Fig.  8B). Its proximal margin seems close to the radial (Fig. 8B). This is likely an artifact of preservation, as this region is difficult to visualize (Fig. 8A).
The distal row has three bones preserved: the distal carpal element (fused distal carpalia III, IV, and ulnare element), the distal carpal II, and the prepollex (Fig. 8A). Distal carpalia III, IV, and ulnare element (distal carpale V of Fabrezi & Alberch, 1996) are all fused into a single large bone (Fig. 8B). It contacts the ulnare, radiale, metacarpals III-V and the distal carpale II (Fig. 8B). The anterior margin bears two distinct articular surfaces for the ulnare and radiale (Fig. 8B). There is no indication that distinct articular facets are present on the posterior margin (Fig. 8B). The bone is crescentic, with a convex posterior margin (Fig. 8B).
Distal carpale II is a small circular element (Fig. 8B). It contacts the distal radial element, the prepollex, the distal carpal element, and the metacarpal II (Fig. 8B).
The prepollex is barely evident. Most, if not all, neobatrachians possess a prepollex composed of two distinct bones (Fabrezi & Barg, 2001). Unfortunately, only a single imprint is visible on the left hand of IGR 144547 (Fig. 8A). On its right hand this region is not preserved (Fig. 2B). As the two bones of the prepollex are often in contact (see Rocěk et al., 2022:fig. 2L) it is possible that they left a single imprint. The prepollex is small and triangular (Fig. 8B). It contacts the distal radial element, the distal carpale II, and the metacarpal II (Fig. 8B).
Metacarpals-Metacarpals II-V are all preserved on both hands on M1 ( Fig. 2A, C). Shallow imprints of the left metacarpals are preserved on M2 (Fig. 2B). The right metacarpals are poorly preserved and poorly defined ( Fig. 2A). Metacarpal II contacts the distal carpale II, and prepollex proximally and the phalanx I of II distally (Fig. 8B). It is one of the longest metacarpals (4.7 mm). Its proximal head is wider than its distal head. FIGURE 6. Ilium of IGR144547. A, acetabular region of the left ilium exposed in lateral view on M2, B, proposed reconstruction of the acetabular region of the left ilium in lateral view and C, right ilium exposed in medial view on M2. Abbreviations: acf, acetabular fossa; acr, acetabular rim; dae, dorsal acetabular expansion; dc, dorsal crest; dpt, dorsal protuberance; ish, iliac shaft; pz, preacetabular zone; saf, supraacetabular fossa; vae, ventral acetabular expansion. Scale bars equal 2 mm. FIGURE 7. Forelimbs of IG144547. A, left humerus on M1, B, interpretative drawing of the same humerus, C, right forelimb on M1, and D, interpretive drawing of the same forelimb. Light gray area represents skin impressions. Abbreviations: ec, eminentia capitata; dia, diaphysis; fc, fossa cubitalis; recd, epicondylus radialis. Scale bars equal 5 mm.
Both proximal and distal articular facet seem flat and straight (Fig. 8B). The bone is elongated, with a rather narrow diaphysis (Fig. 8B). Metacarpal III is shorter (4.5 mm) than metacarpal II. It contacts the distal carpale III proximally and the phalanx I of digit III distally. It is preserved with bone, with both proximal and distal heads being broken, while the diaphysis seems complete (Fig. 8A). As for metacarpal II, the diaphysis is elongated and narrow. The proximal head is wider than the distal head (Fig. 8B). The proximal and distal articular facets are flat and straight (Fig. 8B). Metacarpal IV is as long as metacarpal III (4.5 mm). It contacts the distal carpale IV proximally and the phalanx I of digit IV distally. As for metacarpal III, its distal and proximal heads are broken, and most of the diaphysis is broken as well (Fig. 8A). The proximal end bears a flat articular surface (Fig. 8B). The distal head is too damaged to assess the shape of its articular facet (Fig. 8B).
Metacarpal V is as long as metacarpal II (4.8 mm). It contacts the distal ulnare element proximally and the phalanx I of digit V distally. It is complete (Fig. 8B). Its proximal head is wider than its distal head. The articular facet of the proximal head is concave (Fig. 8B). The articular facet of the distal head is flat.
Phalanges-Phalanges are only preserved on the left hand, on both M1 and M2. Phalanx I is preserved on all digits (Fig.  8A). It is preserved as an imprint on digits II and V, and as a complete bone on digits III and IV. On digits II and IV, the phalanx is elongate (3.3 and 3.4 mm, respectively). On digit II, it is only preserved as a poorly defined imprint (Fig. 8A). On digit IV, the preservation shows that the proximal head is wider than its distal head (Fig. 8B). Both articular facets are flat. On digits III and V, phalanx I is shorter (2.8 and 2.4 mm, FIGURE 8. Hand of IGR 144547. A, left hand on M1 and B, interpretative drawing of the same hand. Light gray areas represent skin impressions of the digits. Abbreviations: dist rd, distal radiale; dist ul, distal ulnare; dist II, distal carpale II; dist III, distal carpale III; dist IV, distal carpale IV; mtc, metacarpal; phx I, II, III, phalanx I, II, III; pp, prepollex; rd, radiale; rdul, radioulna; ul, ulnare. Scale bar equals 2 mm. respectively), and the proximal and distal heads are of the same width (Fig. 8B).
Phalanx II is also preserved on all digits. It is preserved as an imprint on digits II and V, and as a complete bone on digits III and IV. For digits II and III, it is the terminal phalanx. It is very short (1.5 mm). The phalanx narrows distally, before ending in a knob (Fig. 8B). On digits IV and V, the phalanx II is connected distally to the phalanx III (Fig. 9B). It is shorter than the phalanx I (2.6 and 2.2 mm, respectively). Both proximal and distal heads are not distinct from the diaphysis.
Phalanx III is preserved on digit IV only as an imprint (Fig.  8A). It is the terminal phalanx for digits IV and V (missing on IGR 144547). It is identical to the terminal phalanx of digits II and III. It is short (1.5 mm) and bears a knob on its distal end (Fig. 8B).

Hindlimbs
Femora-Both femora are preserved on M1 and M2. The left femur is incomplete on M2. Both femora are preserved as welldelimited imprints (exposed in ventral view) with remnants of bone ( Fig. 2A, C). Femora are long (22.1 mm) and slender. The proximal head of the femur is badly preserved and mostly crushed near the acetabulum (Fig. 9B). No crest nor flange is visible on its lateral and medial margins (Fig. 9B). The femur is slightly sigmoid (S-shaped). No crest is visible on the femur shaft.
Tibiofibulae-Only the left tibiofibula is incompletely preserved on both M1 and M2. The proximal region is preserved as an uncrushed bone, while the diaphysis is mostly preserved as an imprint (exposed in ventral view). The distal region is not preserved (Fig. 9A, B). The proximal head bears a shallow groove mid-width. This groove marks the separation between the fused tibia and fibula (Fig. 9A, B). The tibiofibula is long (14 mm in length) and straight. Slightly more than half of the tibiofibula appears to be preserved. Thus, the complete tibiofibula is probably slightly longer than the femur.

Taxonomic Attribution
Comparison to Extant European Anurans-The presence of a bicondylar sacro-urostylar articulation, a firmsternal pectoral girdle and a diplasiocoelous vertebral column are diagnostic for ranoid (Frost et al., 2006). Within Ranoidea, only natatanuran taxa are known in Europe in the Cenozoic (Rocěk, 2013). Furthermore, an ossified omosternum, if present, is considered a natatanuran synapomorphy (Frost et al., 2006;Lemierre et al., 2021). Within this clade, only Ranidae are known in Europe from the Oligocene (Sanchíz, 1998;Rocěk, 2013;Rage, 2016;Vasilyan, 2018). IGR144547 is similar to Ranidae in the following features: slender rod-like sacral apophyses, the eminentia capitata of the humerus is not shifted from the central axis of the humeral diaphysis, and there is no crest on the femur.
Cenozoic and extant ranids from Western Europe have been only referred to two genera, Rana and Pelophylax (Duellman & Trueb, 1994;Sanchíz, 1998;Rocěk, 2013;Rage, 2016). During the 20th and early 21st centuries, most authors considered the two genera synonymous (with Pelophylax sometimes considered a subgenus of Rana; Frost et al., 2006). In consequence, almost all known Cenozoic ranids from Europe were described as "Rana" (Sanchíz, 1998), regardless of their affinity towards the green water frogs (Pelophylax spp.) or brown water frogs (Rana spp.). In addition, osteological differences between extant European ranids and their Cenozoic counterparts are sometimes hard to find (Sanchíz, 1998). In addition, most of the European Pelophylax are grouped into the Pelophylax kl. esculentus, as hybridization is possible between species included into this complex (Lymberakis et al., 2007), and few (if any) osteological differences exists (Crochet et al., 1995). Thus, most of the taxa erected since the 19th century are considered dubious (Sanchíz, 1998;Rocěk, 2013;Blain et al., 2023).
Pelophylax in the Oligocene of France-As mentioned above, all extinct ranids from the Oligocene of Europe were described as Rana until recently. Thus, the most complete Oligocene ranid, "Rana" aquensis Coquand, 1845 has been referred to Pelophylax in the last decades (Sanchíz, 1998) on the basis of a similar osteology to Pelophylax esculentus (Sanchíz, 1998). However, there has not been any formal assignment to the genus, and the taxonomic validity of the taxon (i.e., as distinct from extant Pelophylax species placed under Pelophylax kl. esculentus) has yet to be established (Sanchíz, 1998).
"Rana" aquensis is known by several subcomplete articulated specimens (Piveteau, 1927: fig. 4) from the upper Oligocene of southern France (Coquand, 1845;Piveteau, 1927;Sanchíz, 1998;Gaudant et al., 2018). The neotype (MNHN.F.AIX355) of "Rana" aquensis should be referred to the genus Pelophylax on the following features: a high and well-developed dorsal crest on the ilium (Sanchiz et al., 1993;Bailon, 1999), a dorsal protuberance pyriform well-developed (Sanchíz et al., 1993), and a high angle (∼42°) between the dorsal protuberance and the iliac shaft (Sanchíz et al., 1993;Bailon, 1999). Thus, we consider that all specimens assigned to "Rana" aquensis should be referred to the genus Pelophylax. However, we refrain from assessing the validity of Pelophylax aquensis, as this is beyond the scope of the study. Pelophylax aquensis has been considered closely related to the Miocene "Rana" meriani (Piveteau, 1927). Furthermore, a recent redescription of Pelophylax pueyoi (Blain et al., 2023) mentioned that the latter taxon (P. peuyoi) could be a subjective junior synonym of the Lower Miocene "Rana" meriani Meyer, 1852 (Meyer, 1860). Thus, a complete revision of the Oligo-Miocene Pelophylax is needed to assess if they represent distinct taxa, and if they could be differentiated from extant Pelophylax taxa.
Comparison of IGR 144547 to Pelophylax aquensis-IGR 144547 resembles Pelophylax aquensis in features of the ilium, that are diagnostic for Pelophylax kl. esculentus. The pectoral girdle of Pelophylax aquensis is poorly preserved (Piveteau, 1927), with only the cleithrum ("omoplate" in Piveteau, 1927) mentioned. Other specimens assigned to P. aquensis are known but have not been described (A. L. pers. obs). Thus, IGR 144547 does not differ from Pelophylax aquensis, but it cannot be distinguished from extant European Pelophylax. Hence, we refrain from any generic assignment for Pelophylax aquensis.

Paleobiogeography
The palearctic water frog Pelophylax ranges throughout Europe and Asia today (Duellman & Trueb, 1994;Chan & Brown, 2017). The divergence between the western (Europe) and eastern (Asia) lineages of Pelophylax has been estimated to have occurred around 35 Ma (Chan & Brown, 2017). The genus then spread throughout Europe and underwent a diversification event during the last 15 Ma (Lymberakis et al., 2007). However, the early diversification of the genus is almost unknown, as a single occurrence has been reported from the lower Oligocene of Germany (MP22; Sanchíz et al., 1993) and the genus is currently absent from the lower Oligocene sites of the Quercy Phosphorites (Rage, 2016). The attribution of IGR 144547 to Pelophylax and its geographic location, in northwestern France is very interesting. It implies that the genus was present in northwestern Europe (from southern Germany to northwestern France) no later than 5 Ma after its arrival in Europe. Thus, Pelophylax apparently underwent a rapid dispersal throughout Europe during the early Oligocene. During the late Oligocene, the genus is known throughout western Europe (Rocěk, 2013). This wide and rapid dispersal was likely facilitated by the "Grande Coupure" extinction event. It is linked to the Eocene/Oligocene transition (33.9 Ma) when a large turnover of the amphibian fauna took place (Rage, 2016;Vasilyan, 2018), with the rise of "true" ranids (Pelophylax, Rana) in the Oligocene (Rocěk, 2013;Rage, 2016). Thus, Pelophylax benefitted from the presence of unoccupied ecological niches, and easily spread in Europe. It should be noted that this rapid and vast dispersal might also reflect a sampling bias, in particular within the Quercy Phosphorites (that represent most of the anuran occurrences from western Europe during the early Oligocene; Rage, 1984Rage, , 2016. Several possible ranids have been identified (Rage, 1984(Rage, , 2016 in the Quercy (MP22-23; Rage, 2016), but have not been studied since Rana and Pelophylax became accepted as separated genera.

CONCLUSION
In conclusion, IGR 144547 represents one of the oldest known members of the European Pelophylax kl. esculentus, and the oldest articulated specimen of Pelophylax identified. We also suggest that 'Rana' aquensis should be included within Pelophylax kl. esculentus. IGR 144547 resembles the holotype of Pelophylax aquensis, but a complete revision of this taxon must be made before any assignment can be proposed. Our specimen represents the first occurrence of Pelophylax in the lower Oligocene of France. This occurrence suggests a rapid dispersal of Pelophylax throughout western Europe during the Oligocene. This opportunistic frog genus likely benefitted from unoccupied ecological niches and the decrease in anuran diversity following the "Grande Coupure."