Crocodylian material here studied is currently stored at IGUW (Vienna, Austria). During a first-hand study of these remains in 2016, two of us (À.H.L and M.C) realized that these exquisite materials had never been reported in detail, possibly due to the lack of information about the site where the remains were recovered. However, both original labels (UW6526A and UW6526B) and the record book of the IGUW collections confirmed that the remains came from “Böhmen” (German name of Bohemia, Czech Republic) and had been referred to Diplocynodon aff. var. ebertsi. K. A. Redlich made similar referrals for specimens recovered from Friedrich-Anna mine (see above, and Redlich, 1902), but it is unknown who ascribed the IGUW specimens to this taxon. Given that Diplocynodon ebertsi was erected by Ludwig in 1877, and taking into account the antiquity of the labels of the remains presented here (written before WWII, pers. com. K. Rauscher, curator of the IGUW), the Diplocynodon fossils were recovered and transferred to the IGUW before 1944.

Becker's long-neglected publication from 1882 merits comment. Becker was primary authority on studies on the stratigraphy close to Josef-Oswald mine. He documented coal seam and his associated deposits, and he collected fossil crocodiles from the Chomutov area that were later offered to D. Štúr (Štúr, 1873 and Štúr, 1879). However, the last alligatoroid remains reported by Becker, including “a jaw with teeth, vertebrae and osteoderms”, were never studied in detail. Although we cannot be fully sure, these very likely correspond to the specimens herein reported. It is very probable that Becker's material has been saved for years to be studied posteriorly by D. Štúr. However, Štúr worked as a vice-head of the Imperial Geological Institute of Vienna at that moment (1877), and later he became director of this institution in 1885 (Barica, 2004). Hence, Štúr most probably offered this material, still unstudied, to the IGUW. For all these reasons, specimens discovered in Josef-Oswald mine (Tušimice) by Becker (1882) most probably represent the same material in Vienna described in our study.

The crocodylian fossil record from NW Bohemia only contains one Eocene locality from Kučlín, named Trupelník hill, (Dvořák et al., 2010). However, the deposits are formed by diatomites with no coal or similar lithological admixture (Kvaček, 2002). During the Oligocene, three areas in the Eger Graben´s basins have reported crocodile material. First locality is placed in Větruše (Ústí nad Labem) and includes claystones, diatomites and lava flows (Radoň, 2001). The second locality is situated within the Doupov volcanic complex (Doupov Mountains) formed by limestone and tuffites (Fejfar and Kvaček, 1993; Kvaček et al., 2014). Given that in both previous localities coal sedimentation does not exist, we can clearly discard them. In the Sokolov Basin a coal layer Rupelian in age exists, but it is sterile and does not contain fossil fauna (Rojík, 2004). In addition, there is no middle to late Miocene localities (neither fluvial nor limnic) in Eger Graben´s basins because sedimentation ended before the middle Miocene Climatic Optimum (Grygar et al., 2014). Consequently, only three localities fit well with the specimens herein studied, all of them roughly equal in age (early Miocene, MN3) and displaying a similar lithology and depositional environment: one from the Sokolov Basin (Friedrich-Anna mine; Redlich, 1902); and two from the Most Basin, Josef-Oswald mine (Laube, 1901) and Skyřice (Laube, 1910 and Schlosser, 1910).

To confirm our evaluation of these sites, we further studied the coal matrix preserved in both skulls. Palynological analyses were not useful as all potential localities (Friedrich-Anna mine, Josef-Oswald mine, and Skyřice) have similar macroflora communities (pers. com. J. Dašková, 2018). Limited sediment samples from the skull are insufficient to run an isotopic analysis (pers. com. K. Mach, 2018). Moreover, data for some localities, in particular the exact levels at which fossils were found are currently not available or destroyed, further reinforcing our decision not to perform isotopic and palynological studies between localities. That matrix is mainly formed by clay minerals, bone splinters, and coal fragments. Around the bone fragments, organic materials (mainly coal) are dominant, whereas a red ferrous fill can be seen inside the lacunae. According to the International Classification ECE-UN (1998), coal is classified as a low-rank. It contains abundant pyrite crystals and organic fragments. Coal samples were identified as a partially weathered brown coal (oxyhumolite: Fig. 2A–B). Organic material is formed by gelled remains of plant tissue of lignin-cellulosic character (known as huminite macerals), and more specifically ulminite with clay minerals and crystalline and rhomboidal pyrites (Fig. 2A–B). Apart of huminite macerals, other liptinite components, such as sporinite, alginite, and rare inertinite macerals like funginite, were found (Table S1). Even though both rank and maceral composition of coal analysis perfectly fit with the age of the Tušimice site, this is not enough by itself to discount an alternative attribution to Friedrich-Anna mine and Skyřice sites.

We provide a brief explanation about the preservation state and conditions of both specimens. All descriptions, figures, and schematic drawings are provided in the next subsections.

UW6526A1-5: The specimen is a partial cranium (skull plus mandible) preserved in five portions: UW6526A1 (Fig. 3A–H), anterior and mid portion of the cranium; UW6526A2 (Fig. 3I–N), left fragment of the posterolateral part of skull table and three osteoderms; UW6526A3 (Fig. 3O–T), left fragment of the dentary with one dentary tooth; UW6526A4 (Fig. 3U–Z), right fragment of the angular with two dorsal osteoderms; UW6526A5 (Fig. 3A′–F′), and caudal vertebral body.

UW6526A1: The most complete specimen is dorsoventrally flattened and most of the posterior part of the braincase, posteriorly to the parietal, is missing (Fig. 3A–D). Main cranial cavities (the infratemporal fenestra and orbit) and some bones of the skull such as the basioccipital, basisphenoid, jugal, ectopterygoid, and surangular cannot be described in detail due to distortion or absence. The cranium is preserved in a slightly clockwise rotation relative to the correct anatomical position, which is clearly evident in the anterior tip of the snout. It is well prepared, but some small cracks are still filled with sediment. Nevertheless, the anterior portion of the palatal complex and some skull roof bones are preserved. In both lateral views, a convex shape is present, caused by postmortem compression (Fig. 3E–H). Both premaxillae are preserved, but the anterior margin of the left premaxilla is slightly eroded and therefore the first premaxillary teeth are not preserved (Fig. 3C–F). The left maxilla is almost complete, whereas the right one is somewhat compressed and lacks its posterior part. The central portion of the left nasal is broken. Similarly, both lacrimals are damaged and bear a couple of holes in dorsal view (Fig. 3A–D). The posterior portion of both prefrontals is slightly crushed. In any case, the preserved portion of the prefrontals is visible in dorsal view, although several microcracks are present. On the ventral side, the whole palatal area is quite crushed and further some sutures of the bones are difficult to discern (Fig. 3C–D). The palatines are partially preserved, whereas the pterygoids are heavily damaged and the choanae are not preserved. Only both dentaries and the anterior portion of the splenials are preserved, but other bones, which form a mandible, are missing. Both dentaries and splenials are not in original anatomical position, they are rather rotated 90° to the right (Fig. 3C–D). Regarding dentition, some dentary and maxillary teeth are broken away, eroded or slightly compressed.

UW6526A2: This includes three disarticulated osteoderms and the left posterolateral portion of a skull formed by the quadratojugal, squamosal and quadrate (Fig. 3I–J). The quadratojugal is strongly damaged and only the central portion of the bone is present (Fig. 3I). Similarly, just a very small portion of the left squamosal is preserved. The quadrate is better preserved than the other bones (Fig. 3M–N). Three isolated osteoderms are attached as it can be ascertained in ventral view, but they are not in their original anatomical position (Fig. 3K–L).

UW6526A3: This is a fragment of the left dentary bearing one preserved tooth (Fig. 3O–P). In lateral view, this bone fragment is rather crushed, and posteriorly, another indeterminate fragment bone (most likely from a pterygoid) appears to be attached (Fig. 3Q–T).

UW6526A4: This corresponds to a partially preserved right angular (Fig. 3U–Z). Two dorsal osteoderms are attached, which are not in anatomical position (Fig. 3W–Z).

UW6526A5: This is an incomplete caudal vertebra comprising the vertebral centrum. The neural arch is entirely missing (Fig. 3A′–F′).

UW6526B1: This is a posterior portion of the skull preserving the frontal, parietal, postorbital, squamosal, and supratemporal fenestra (Fig. 4A–B). The ventral and lateral sides are eroded and rather damaged. Bones such as the quadrate, quadratojugal, and squamosal are completely missing (Fig. 4C–E). The dorsal surface is well prepared, but the posterior surface is poorly prepared and most of the occipital bones are covered by a coal matrix.

UW6526B2: This is a partial left angular (Fig. 4F–I), which is not very well restored. The anterior part of the bone is embedded in a coal matrix and also most of the exposed bone is similarly covered with sediment (Fig. 4F–I).

Premaxilla. This bone is completely preserved in UW6526A1, but slightly turned to the right and flattened (Fig. 3A–B). It is about 1.5 times longer than wide, with a rounded anterior margin. The premaxilla is at the anteriormost part of the snout and contacts the maxilla posterolaterally and the nasal posteromedially (Fig. 3 and Fig. 5). Pits densely decorate all the dorsal surface of the bone, although it is less pronounced along the anterior edge. Sutures between the premaxilla, maxilla, and nasal are easily visible. An oval, ridge-like thickening surrounds the naris. This structure is continuous from the anterior margin to the posterior margin, where it becomes a shallow concavity (Fig. 3A–B, E–H). The external naris is almost completely surrounded by the premaxillae, with the anterior processes of the nasals forming the posterior most narial margin. The pointed posterior tip of the nasal penetrates into the maxilla at an acute angle (Fig. 3 and Fig. 5). The premaxilla bears five teeth: the first four teeth in the left premaxilla and the second, fourth and fifth in the right one. The third and fourth teeth are the largest, and the fourth is largest (Fig. 3 and Fig. 5). On both sides of this element, the posterior dorsal processes extent of the premaxillae reaches the level of the third maxillary tooth position (Fig. 3 and Fig. 5). In ventral view, the incisive foramen cannot be confidently observed.

Maxilla. It is longer than wide and contacts the premaxilla anteriorly, the nasal medially, and the lacrimal, jugal, and palatine posteriorly (Fig. 3B and D). As in all diplocynodontids, the dorsal surface of the maxilla is pitted. The snout profile is slightly constricted at the level between the maxilla and premaxilla and the third and fourth dentary teeth (Fig. 3 and Fig. 5). This notch can be discerned behind the last premaxillary–maxillary tooth in lateral view, which is divided by the premaxillary–maxillary suture (Fig. 3 and Fig. 5). The suture with the nasal and lacrimal is almost straight. Because both maxillae are not fully preserved, the complete tooth count cannot be estimated. The anteriormost eleven teeth are preserved in the left maxilla, whereas the right element bears only eight teeth (Fig. 3 and Fig. 5). With regard to the tooth size, the first maxillary tooth is the smallest, and from the second to the fifth, they gradually increase. However, the fourth and fifth teeth are instead confluent (Fig. 3 and Fig. 5).

Nasal. It is elongated and located along the midline of the snout. The nasals form most of the snout length. The anterolateral margin of the nasal contacts the premaxilla and the maxilla, and the posterior margin contacts the frontal and lacrimal (Fig. 3B). As with most bones on the dorsal surface, rounded pits densely ornament the nasals. The anterior end of the bone is narrow and clearly extends to the external naris, preventing the premaxilla from meeting posterior to the naris (Fig. 3A–B). Posteriorly, the nasals reach their maximum width in the area of contact with the lacrimal and frontal. The nasals are bifurcated posteriorly and separated by an anterior frontal projection. Each bifurcation tapers to a point between the prefrontal and frontal. The left nasal displays a wide contact with the lacrimal (UW6526A1: Fig. 3A–B).

Prefrontal. The preserved portion of each bone in UW6526A1 is triangular in shape, and each contact at least the nasal anteriorly, the lacrimal anterolaterally, and the frontal medially (Fig. 3A–B). Its anterior tip inserts between the nasal and the lacrimal. The suture between the lacrimal and the prefrontal, only seen on the right prefrontal, reaches the anteromedial margin of the orbit, and therefore the prefrontal participates in the anterior margin of the orbital rim (Fig. 3 and Fig. 5).

Lacrimal: It is roughly triangular in shape and narrows medially. The lacrimal contacts the maxilla anterolaterally, the nasal anteromedially and the prefrontal medially (Fig. 3 and Fig. 5). The lacrimal extends further anteriorly than the prefrontal, and although both lacrimals are partially preserved in UW6526A1, it can be estimated that the lacrimal is anteroposteriorly longer than the prefrontal.

Frontal. The frontal is complete in UW6526A1 and contacts the nasal and prefrontal anterolaterally, the postorbital laterally, and the parietal posteriorly. It has a subtriangular shape, is posteriorly wide, and narrows anteriorly (Fig. 3A–B). The frontal and maxillary external surfaces are much heavily pitted than other bones. A marked dorsoventral step is recognizable at the middle of the frontal, further delimitating this bone into two parts (Fig. 3 and Fig. 5): the main body, which is markedly concave at the level of the orbits and shows an elevated orbital margin, and the anterior process, which is approximately 1.5 times longer than the body. The interorbital region is rather wide, approximately equal in width, to the supratemporal fenestra. The posterior portion of the frontal in UW6526A1 and UW6526B1 is slightly concave between the orbits (Fig. 4A–B).

Postorbital. The left postorbital is preserved in UW6526A1 (Fig. 3A–B) and both are preserved in UW6526B1 (Fig. 4A–B). The postorbital is subrectangular and wider than long. It meets the frontal anteromedially, the supratemporal fenestra medially, and the squamosal posteriorly. The postorbital contributes to the posterior orbital margin, the anterolateral margin of the supratemporal fenestra and the anterodorsal margin of the infratemporal fenestra (Fig. 3, Fig. 4 and Fig. 5). However, in both postorbitals preserved in UW6526B1, sediment covers and obscures the ornamentation and sutures with the squamosal and frontal. The dorsal surface of the bone is rather flat in both individuals.

Squamosal. This is preserved only in UW6526B1. It contacts the postorbital anteriorly, the parietal medially, and the quadrate ventrolaterally (Fig. 4 and Fig. 5). The squamosal is longer than wide. There is a pronounced crest on the posterolateral margin of this bone. The posteromedial margin is concave anteriorly and contacts the postorbital and supraoccipital medially. The postorbital–squamosal suture is slightly anterior to the center of the supratemporal fenestra (Fig. 4A–B). Connection to the quadrate and paroccipital process is preserved only on the right squamosal.

Parietal. Its shape is roughly rectangular, it is longer than wide, with a distinct constriction developed between the supratemporal fenestrae (Fig. 3, Fig. 4 and Fig. 5). It contacts the frontal anteriorly, the squamosal posterolaterally, and the supraoccipital posteriorly. The parietal is completely preserved in UW6526B1 and forms part of the supratemporal fenestral margins (Fig. 4A–B). The fronto-parietal suture is clearly visible and is slightly concave posteriorly. As with the frontal, the dorsal surface of the parietal is heavily pitted, although matrix fills most of the pits. The parietal-squamosal suture is parasagittal direction, and further it is approximately equal in length to that of the fronto-parietal suture. However, the parietal does not reach the posterior margin of the skull and supraoccipitals are in contact (Fig. 4 and Fig. 5).

Supraoccipital. This small bone is preserved on the posterior mid-region to the skull table (UW6526B1: Fig. 4A–B). It meets at least the parietal anteriorly and squamosal laterally. In dorsal view, the supraoccipital has a subsemicircular shape, with a rounded posterior margin (Fig. 4 and Fig. 5).

Exoccipital. This bone is partly preserved only on UW6526B1, and it forms the posterooccipital end of the skull as preserved. It borders the dorsolateral margin of the foramen magnum (Fig. 4D–E).

Quadrate. It is partially preserved as an isolated fragment of UW6526A2 (Fig. 3I–L). The left portion of the quadrate meets the squamosal medially and the quadratojugal laterally. Contact with the exoccipital cannot be evaluated in dorsolateral view (Fig. 3I–L). The cranioquadrate groove and foramen aereum can be ascertained only in UW6526A2. In particular, the foramen aereum is placed very far from their medial margin (Fig. 3I–J). Both hemicondyles are visible in posterior view, and the lateral one appears to be slightly larger than the medial one. The dorsal surface of the lateral hemicondyle appears to be dorsal to its medial counterpart.

Quadratojugal. A small portion of the left quadratojugal is partially preserved in UW6526A2 (Fig. 3I–L). The anterior portion is missing, but the quadratojugal contacts the quadrate posteromedially. No additional descriptive details can be provided.

Palatine. Only the posterior portion of this bone is preserved in UW6526A1. It is located posteromedial to the maxilla and anterior to the pterygoid (Fig. 3 and Fig. 5). The anterior preserved portion is covered by bone fragments most likely derived from the lower jaw. A fissure between the palatines is roughly preserved.

Pterygoid. The pterygoids are poorly preserved, crushed and damaged in UW6526A1 (Fig. 3C–D). The posterior margins of the suborbital fenestra are formed by the paired pterygoids, and the medial margins are formed by the palatines (Fig. 3 and Fig. 5).

Dentary. The dentaries converge anteriorly to each other. This bone contacts the splenial medially. Contacts with the surangular and angular cannot be restored. In UW6526A1, the mandibles are displaced. For this reason, the tooth count as well as the symphyseal region cannot be entirely observed (Fig. 3C–H). Nine teeth are visible in the right dentary. The third and fourth teeth are close to each other and equally large (Fig. 3C–D). Moreover, the tooth row is concavoconvex in lateral view, not linear, and also the most pronounced concavity occurs behind the fourth dentary tooth (Fig. 3C–D). Small pits decorate the external surface of both dentaries. UW6526A3 represents a left dentary fragment with one completely preserved tooth (Fig. 3O–P) and a sediment-filled alveolus (Fig. 3S–T).

Splenial. The splenial is completely preserved on the left mandible UW6526A1 (Fig. 3C–D). It lies on the medial surface of the dentary posterior to the dentary symphysis. It medially covers the Meckelian groove. Both dorsal and ventral splenial tips can be seen on the medial surface in UW6526A1. They neither contact each other nor reach the dentary symphysis. The ventral end of the splenial is anteriorly longer than the dorsal end, and it reaches the level of the fourth dentary tooth (Fig. 3C–D).

Angular. The angular forms the entire posteroventral portion of the mandible. In lateral view, the concave shape of the angular forms an obtuse posterior process visible in both UW6526A4 (Fig. 3U–V) and UW6526B2 (Fig. 4H). The ventral margin of the external mandibular fenestra can be observed only in UW6526B2 (Fig. 4H–I). Despite of the presence of some overlapping sediment, the external surface of the angular (Fig. 3 and Fig. 4) preserves ornamentation formed by large irregular to subcircular pits.

Dentition. All teeth in both mandibles are attached to bone in a normal anatomical connection: UW6526A1 (Fig. 3A–H) and UW6526A3 (Fig. 3O–T). However, no additional isolated teeth are available. Our examination reveals the presence of five premaxillary teeth (Fig. 3C–D), more than eleven maxillary teeth (Fig. 3A–D), and at least nine dentary teeth (Fig. 3C–D). The total number of maxillary and dentary teeth is probably higher. In general, all teeth (premaxillary, maxillary, and dentary) are similarly conical, medially curved and with the slender and acute apical crowns. Furthermore, they are characterized by the conspicuous cutting edges and a lateral surface that is more convex than the medial one. The enamel is smooth, but some teeth have basal striations. Based on the state of preservation of UW6526A1, with the mandible completely attached to base of the skull, we cannot identify occlusal pits.

Vertebra. One partially procœlous vertebra is preserved. It corresponds to a neural centrum without arch (UW6526A5: Fig. 3A′–F′). The absence of the parapophysis indicates that it is not a cervical vertebra. This fragment represents most likely a caudal vertebra. In posterior view, the neural centrum is square-shaped.

Osteoderms. Five osteoderms are preserved: three in internal position (UW6526A2: Fig. 3K–L); and two dorsal in external position (UW6526A4: Fig. 3Y–Z). None are in anatomical position. They are subrectangular in shape, having an internal surface pierced by foramina, but not pitted. A dense, uniform network of subcircular pits is found on the dorsal osteoderms preserved in UW6526A4. A longitudinal keel is developed along the midline of the pitted regions. Because the anterior margins of the dorsal osteoderms are not preserved, we cannot ascertain if the anterior articular surface is smooth.

Specimens from the Tušimice site (MN3, Czech Republic), Els Casots (MN4, Spain) and Saint-Gérand-le-Puy (MN2, France) are similar in many general patterns, such as the decoration and general outline of the skull, size of the orbits, and supratemporal and suborbital fenestrae. The material from Tušimice is roughly equal in size to Spanish material, but they are notably smaller than French specimens of D. ratelii. The smaller size of both individuals at Tušimice site is most surely reflecting an earlier ontogenetic stage. In terms of intraspecific variability, UW6526A1 shows a slightly broader preorbital area (clearly discernible from dorsal outline) unlike the specimens from Els Casots (IPS951 and IPS14721). However, our inspection of D. ratelii from Saint-Gérand-le-Puy suggests that this taxon can display both nose morphologies, as in Els Casots site (MNHN SG539), or alternatively as in the Tušimice site (MNHN SG13729a). The naris tends to be elliptical, slightly longer than wide, in IPS951 from Els Casots (not evaluable in Tušimice material: UW6526A1). Similarly, both variations are present in the naris from the type locality, being rather subcircular in MNHN SG652 or even slightly heart shaped as in MNHN SG13729a or MNHN SG652. All these above-mentioned distinctions may be due to different ontogenetic stages of the individuals, and therefore are inside the intraspecific variation of the species. Given that molecular data have revealed several modern crocodile species to be cryptic species complexes (Eaton et al., 2009, Hekkala et al., 2011, McAliley et al., 2006 and Shirley et al., 2014), it is necessary to take into account all these morphological differences. More detailed comparisons of type material D. ratelii (not studied in detail yet) together with other poorly studied (Brinkman and Rauhe, 1998 and Ginsburg and Bulot, 1997) or unpublished Diplocynodon remains (Montaigu: Delfino and Rossi, 2013), would be required to evaluate the ranges of the intraspecific variation of D. ratelii from the early Miocene of Europe.

As shown in the palynological and petrological results section, the crocodylian fossil remains stored at IGUW (Vienna, Austria) were most likely recovered from the Tušimice site (Bohemia, Czech Republic). These specimens can be referred to Alligatoroidea based on a dorsomedially located foramen aereum and a small, ventrally reflected medial hemicondyle of the quadrate (Brochu, 1999, Brochu, 2003, Díaz Aráez et al., 2017, Martin, 2010, Martin and Gross, 2011 and Martin et al., 2014). The combination of enlarged, subequal and confluent alveoli in both maxillary (fourth and fifth) and the dentary (third and fourth), together with the anterior extension of the lacrimal (which is longer than the prefrontals) allows us to refer them to Diplocynodon (Brochu, 1999, Díaz Aráez et al., 2017, Martin, 2010, Martin and Gross, 2011 and Martin et al., 2014). At the species level, the material described here, in turn, displays the same combination of features as in the type species D. ratelii: the shape of the fronto-parietal suture (concavoconvex, not linear); the contour of the naris (a crest-like thickening on the margin of the external naris); the extension of the dentary symphysis (it reaches the posterior margin of the fourth alveolus); the relative position of the splenial, Meckelian groove and dentary symphysis (the splenial is excluded from the dentary symphysis and the ventral end of the splenial is anteriorly longer than the dorsal end); and the relationship between the nasals and external naris (the anterior tip of the nasals reach externally the posterior rim of the external naris). In fact, D. ratelii differs uniquely from the other species of Diplocynodon in having the nasals that do reach externally (but do not bisect) the posterior rim of the external naris (Díaz Aráez et al., 2017).

These diagnostic features were taken from Brochu et al. (2012), although codings for the above-mentioned characters were provided for five of the nine valid current species only: D. darwini (Ludwig, 1877), D. hantoniensis (Wood, 1846), D. muelleri (Kälin, 1936), D. ratelii Pomel, 1847 and D. tormis Buscalioni et al., 1992. Two recent additional added two species: D. deponiae (Frey et al., 1987) by Delfino and Smith (2012); and D. elavericus Martin, 2010, D. ungeri (Prangner, 1845), and D. remensis Martin et al., 2014 by Martin et al. (2014).

Diplocynodon represents a group of European endemic fossil alligatoroids that were rather abundant and widely distributed in the Late Paleogene to the Middle Neogene deposits of Europe (Delfino and Smith, 2012, Díaz Aráez et al., 2017, Hua, 2004, Martin, 2010 and Martin et al., 2014). According to Martin et al. (2014), up to nine species are taxonomically valid. In any case, Diplocynodon most probably arose during the Paleocene, as some of its most diagnostic features (e.g., the presence of two subequal and confluent alveoli in both the maxilla and dentary) are already present in its basal-most known members (Martin et al., 2014). During the late Miocene, Diplocynodon was restricted to South European refuges until became extinct due to a cooling event. For this reason, a short coexistence interval with Crocodylus cannot be excluded (Böhme, 2003, Delfino and Rossi, 2013 and Delfino et al., 2007).

Cranial remains from Tušimice correspond to Diplocynodon ratelii recovered from the lower Miocene of France (Vaillant, 1872) and Spain (Díaz Aráez et al., 2017). This is based on two mandibular and three skull features. Based on the left medial surface of the dentary of UW6526A1 (Fig. 3C–D; see character #54 of Brochu et al., 2012), the ventral anterior tip of the splenial is longer than the dorsal one and there is no splenial symphysis. This is true for in seven of the nine Diplocynodon species: D. darwini, D. deponiae, D. elavericus, D. hantoniensis, D. ratelii, D. tormis and D. ungeri (Brochu et al., 2012, Delfino and Smith, 2012, Díaz Aráez et al., 2017, Hofmann, 1887 and Martin et al., 2014). Although in D. muelleri the splenial is also excluded from the symphysis, its dorsal tip is longer than the ventral one. That means an inverse condition for the dorsal tip of the splenial, if compared to all the above-mentioned species (Brochu et al., 2012). According to Martin et al. (2014), the splenial participates in the mandibular symphysis in D. remensis. For this reason, the Tušimice remains cannot be allocated to D. muelleri or to D. remensis.

Concerning the symphyseal extension of the dentary (character #54 #49 of Brochu et al., 2012), UW6526A1 possesses a shorter symphysis and reaching the fourth or fifth dentary alveoli (Fig. 3C–D), as in D. ratelii, D. muelleri, and D. remensis (Brochu et al., 2012, Díaz Aráez et al., 2017 and Martin et al., 2014). In contrast, a larger dentary symphysis at the level of the sixth to eighth alveoli can be observed in D. elavericus, D. ungeri, D. hantonensis, and D. darwini (Brochu et al., 2012, Hofmann, 1887, Martin and Gross, 2011 and Martin et al., 2014). Even though the symphyseal extension of the dentary is not preserved in D. tormis or D. deponiae, this feature confirms that the Tušimice material is much consistent with D. ratelii (Brochu et al., 2012, Díaz Aráez et al., 2017 and Martin et al., 2014).

Only one specimen from Tušimice (UW6526A1: Fig. 3A–B) preserves the snout, which preserves a crest-like thickening on the margin of the external naris (character #85, Brochu et al., 2012). Just two species within Diplocynodon included in previous analyses, D. ratelii and D. muelleri, display this trait (Díaz Aráez et al., 2017 and Piras and Buscalioni, 2006). The margin of the external naris cannot be evaluated in D. elavericus, D. tormis, and D. ungeri. In any case, this feature supports an attribution to D. ratelii.

Both UW6526A1 and UW6526B1 from Tušimice (Fig. 3 and Fig. 4) preserve a modestly posteriorly convex frontoparietal suture between the supratemporal fenestrae as in D. ratelii and D. ungeri (Brochu et al., 2012, Díaz Aráez et al., 2017 and Martin et al., 2014). Conversely, an extremely concavoconvex suture is observed in other Diplocynodon species except for D. elavericus, in which the posterior skull table is damaged and therefore not preserved (Martin, 2010). Hence, this combination of features also reinforces an attribution to D. ratelii.

The relationships between the nasals and the external naris (character #82, Brochu et al., 2012) were studied in Tušimice material. UW6526A1 (Fig. 3 and Fig. 5) clearly shows that the anteromedial projection of the nasals extends to the external naris. According to Díaz Aráez et al. (2017), this condition is present exclusively in all specimens of D. ratelii. This feature cannot be evaluated for D. elavericus and D. deponiae (Delfino and Smith, 2012 and Martin, 2010). The extension of the nasal into the external naris is also consistent with referring the Tušimice material to D. ratelii rather than to any other named Diplocynodon species.

Finally, the described specimens from Tušimice site (MN3) possess the same diagnostic features as D. ratelii, and further display a set of features that discard confidently all other recognized species of Diplocynodon. The new material reported here from a latter locality represents the first report of Diplocynodon from the Most Basin (Eger Graben, Czech Republic). Therefore, this occurrence extends the geographic distribution of D. ratelii in central Europe, which was previously limited to western Europe (MN2 of France and MN4 of Spain). As stated above, the cranial differences reported at Tušimice, Els Casots, and Saint-Gérand-le-Puy may be due to younger ontogenetic stages or intraspecific variation. Based on sedimentological evidence and the abundant remains of anurans and freshwater snails, Tušimice represents a fluvio-lacustrine depositional system surrounded by bottomland swamp forest environments (Grygar et al., 2017a, Grygar et al., 2017b and Kvaček et al., 2004). This paleoenvironmental reconstruction is reinforced by a micromammal assemblage associated with forested and humid environments (Fejfar, 1989 and Fejfar and Kvaček, 1993). During the Miocene Climatic Optimum, central Europe presented high humidity and temperatures, which are optimal conditions for large ectothermic reptiles such as alligatoroids (Böhme, 2003 and Delfino and Rossi, 2013). The presence of alligatorid Diplocynodon further supports the fluvio-lacustrine depositional system surrounded by bottomland swamp forest environments inferred for the Tušimice site. In particular, Diplocynodon needed permanent water bodies to live and a relatively high mean temperature (not less than 14.2 °C) to thermoregulate (Böhme, 2003, Delfino and Rossi, 2013 and Markwick, 1998).