Revision of the Eocene ‘Platyrhina’ species from the Bolca Lagerstätte (Italy) reveals the first panray (Batomorphii: Zanobatidae) in the fossil record

The fossil-Lagerstätte of Bolca (Italy) is well known for the diversity and exquisite preservation of its bony and cartilaginous fishes documenting tropical shallow-water marine environments associated with coral reefs in the western Tethys during the early Eocene. In this study, the taxonomic, systematic and phylogenetic position of two batoid species traditionally assigned to the living thornback ray genus Platyrhina is re-evaluated. †Platyrhina bolcensis Heckel, 1851 is recognized as a separate species of the Platyrhinidae because of its plate-like antorbital cartilage with an irregular outline and a small horn on the nasal capsules. Also, the rostral cartilage does not reach the anterior border of the disc. Support for the placement of this species within the new genus †Eoplatyrhina gen. nov. is based on a combination of morphological and meristic features (e.g. nasal capsules at right angles to the rostrum; large space between the hyomandibulae and mandibular arch; approximately 132 vertebral centra; 15–16 rib pairs; 81–87 pectoral radials; 18–21 pelvic radials; short, straight and stout claspers; 40–50 caudal-fin radials; thorns absent). A second species, †Platyrhina egertoni (De Zigno, 1876), is more closely related to the living panray Zanobatus than Platyrhina and is assigned here to †Plesiozanobatus gen. nov. because of a combination of characters that support its placement within the family Zanobatidae (tail stout and short, distinctly demarcated from disc; two dorsal fins and complete caudal fin; small dermal denticles and scattered thorns covering disc and tail; rostral cartilage absent; nasal capsules without horn-like processes; mesopterygium absent). The systematic position of a third taxon, †Platyrhina gigantea (Blainville, 1818), is currently impossible to establish due to the poor preservation of the only known specimen, and therefore we propose to consider it a nomen dubium. Palaeoecological and biogeographic features of the Eocene platyrhinids and zanobatids from Bolca are also discussed. http://zoobank.org/urn:lsid:zoobank.org:pub:B4C7A979-7972-409B-B489-A6DDD5E35FE5


Introduction
The Ypresian Konservat-Lagerst€ atte of Bolca, in northeastern Italy, is one of the few Palaeogene deposits where fossils of cartilaginous fishes (Chondrichthyes) are exquisitely preserved . Individuals include complete and fully articulated skeletal remains, which is the exception in the fossil record with chondrichthyans mostly being represented by isolated teeth (Cappetta 2012). Recent studies have contributed to the knowledge of the taxonomy and systematic position of the cartilaginous fishes from the Pesciara and Monte Postale sites of Bolca, which include about a dozen species-level taxa belonging to a variety of holocephalian, selachian and batoid lineages (Fanti et al. 2016(Fanti et al. , 2019Marram a et al. 2018aMarram a et al. , b, c, 2019a). These batoids are represented by electric rays (Torpediniformes), guitarfishes (Rhinopristiformes), stingrays (Myliobatiformes) and three batoid species that were historically assigned to the thornback ray genus Platyrhina M€ uller & Henle, 1838: †P. bolcensis Heckel, 1851, †P. egertoni (De Zigno, 1876 and †P. gigantea (Blainville, 1818). The last account of these three batoid species was provided at the end of the nineteenth century by Jaekel (1894) in his comprehensive review of the elasmobranch fishes from Bolca known at that time. In this paper, we redescribe and re-evaluate the systematic position of the fossil material from Bolca traditionally assigned to Platyrhina in the context of our current understanding of platyrhinid phylogenetics.
The higher taxonomic placement and interrelationships of the families Platyrhinidae and Zanobatidae within batoid fishes are still debated today. According to morphological studies, these families are traditionally considered successive sister taxa to the stingray order Myliobatiformes (McEachran et al. 1996;McEachran & Aschliman 2004;Aschliman et al. 2012a;Villalobos-Segura et al. 2019). Conversely, molecular analyses place the Platyrhinidae as sister taxon to the electric ray order Torpediniformes (Aschliman et al. 2012b;Naylor et al. 2012;Bertozzi et al. 2016;Last et al. 2016), whereas the Zanobatidae are regarded either as sister to Myliobatiformes (Aschliman et al. 2012b;Bertozzi et al. 2016;Last et al. 2016) or as a member of the order Rhinopristiformes (Naylor et al. 2012).
The fossil record of platyrhinids is very poor compared to the other batoid lineages, possibly because their isolated teeth are often misidentified and assigned to the genus Rhinobatos (Claeson et al. 2013). Fossil platyrhinids can be traced back to the Late Cretaceous and include extinct genera represented by articulated skeletal remains, like †Tethybatis from the Campanian/ Maastrichtian of southern Italy (Carvalho 2004) and †Tingitanius from the Turonian of Morocco (Claeson et al. 2013), isolated teeth of †Cretaplatyrhinoidis and †Pseudoplatyrhina from the Turonian-Santonian of the Anglo-Paris Basin (Guinot et al. 2012), and a few occurrences of Platyrhina and Platyrhinoidis teeth from the Eocene of Egypt (Underwood et al. 2011) and Pleistocene of California (Long 1993). †Platyrhina ypresiensis and †P. dockeryi from the Eocene of Belgium and the USA have been recently transferred to the myliobatiform genus †Hypolophodon Cappetta, 1980(Cappetta 2012Case et al. 2015). †Britobatos primarmatus (Woodward, 1889), from the Santonian of Lebanon, was suggested to belong to the Platyrhinidae by Brito & Dutheil (2004), although Claeson et al. (2013) excluded this taxon from this family, instead placing it as a sister to the family. †Protoplatyrhina, based on isolated teeth from the Late Cretaceous of North America, was considered a possible ancestor of Platyrhina by Case (1978). However, Cappetta (1987,2012) rejected this hypothesis and considered †Protoplatyrhina to be a rhinobatoid of family incertae sedis. To our knowledge, the family Zanobatidae has never been recognized in the fossil record until now.

Geological setting
Lithological features, museum catalogue registers and information from the literature suggest that all †'Platyrhina' bolcensis specimens come from the Monte Postale site, whereas specimens of †'P.' egertoni and †'P.' gigantea are from the Pesciara site; these are two of the main fossiliferous deposits of the Bolca Konservat-Lagerst€ atte located in Verona Province, north-eastern Italy (Fig. 1). The Pesciara and Monte Postale sediments represent shallow-water Eocene sequences deposited on the Lessini Shelf, a palaeogeographical feature of the Southern Alps that was uplifted during the Alpine orogeny, acting as an area of deposition of shallow-water carbonates (Doglioni & Bosellini 1987;Bosellini 1989). The Pesciara site consists of a limestone outcrop, about 20 m thick, surrounded by volcanic deposits and comprising a rhythmic alternation of finely laminated micritic limestones with fishes, plants and grainstone-bearing benthic fossils (Papazzoni & Trevisani 2006). Based on their larger benthic foraminiferan content, the Pesciara fossiliferous sediments were assigned to the †Alveolina dainelli Zone, corresponding to the late Cuisian (late Ypresian, between 48.96 and 48.5 Ma; Papazzoni & Trevisani 2006;Papazzoni et al. 2014). Quantitative palaeoecological analyses suggest that the Pesciara fish assemblage was characterized by a sharp oligarchic structure dominated by zooplanktivorous fishes, whereas the taphonomic features support the hypothesis that the fossiliferous sediments accumulated in a shallow intraplatform basin in which anoxic conditions and the development of a microbial mat at the bottom promoted the high-quality preservation of the fossils (Papazzoni & Trevisani 2006;Marram a et al. 2016).
The uppermost part of the Monte Postale succession consists of more than 130 m of massive grainstones that alternate with massive coralgal limestones and laminated wackestones with fishes and plants similar to those of the Pesciara site, although the fossiliferous laminites of the latter appear to be slightly younger (e.g. Vescogni et al. 2016;Papazzoni et al. 2017). Evidence of a coralgal rim, lagoonal deposits, and fore-reef systems were detected for the Monte Postale palaeobiotope (Vescogni et al. 2016). This interpretation is also supported by quantitative palaeoecological and taphonomic studies of the Monte Postale fish assemblage, which revealed a high degree of disarticulation of fish skeletons, unimodal dispersion of the elements, and bioturbations, which are interpreted as the result of periodic oxic bottom conditions (Marram a et al. 2016). The fossiliferous strata of the Monte Postale span the entire NP 13 (¼ CNE 5) calcareous nannoplankton zones (Papazzoni et al. 2017), corresponding to a large part of the Shallow Benthic Zone (SBZ) 11 in the time interval between 50.5 and 48.96 Ma.

Material and methods
The present study is based on three nearly complete and articulated specimens traditionally referred to †'Platyrhina' bolcensis, six specimens of †'P.' egertoni and a single individual of †'P.' gigantea. The specimens are currently housed in the Museo Civico di Storia Naturale di Verona, Museo di Geologia e Paleontologia dell'Universit a degli Studi di Padova, Museo Geologico Giovanni Capellini, Universit a degli Studi di Bologna, Mus eum National d'Histoire Naturelle, Paris, and Museum f€ ur Naturkunde, Berlin. Some of the specimens were examined under ultraviolet light in order to distinguish the preserved skeletal and soft tissues from grout or pigments. Measurements were taken to the nearest 0.1 mm. Osteological terminology primarily follows Carvalho (2004), Aschliman et al. (2012a) and Claeson et al. (2013). A dagger ( †) preceding a taxon name is used to indicate that it is extinct.
Specimens from Bolca were treated as operational terminal taxa and added to the morphological data set of Villalobos-Segura et al. (2019), which in turn was compiled from the matrices of Aschliman et al. (2012a) and Claeson et al. (2013) (Supplemental material, File 1; Appendix 1). The original characters 5, 7, 52, 72 and 91 of Villalobos-Segura et al. (2019) were excluded because they were found to be uninformative. All the codings were checked and some were corrected based on new observations or according to the most up-to-date literature. The dataset was further concatenated with the molecular matrix of Aschliman et al. (2012b) to produce a mixed-data matrix, the subject of a second phylogenetic analysis following on from a morphologyonly analysis.
The Bolca morphological matrix also differs from that of Villalobos-Segura et al. (2019) by taking into account four potential outgroups relative to the ingroup clade, crown Batoidea: Chimaera, Heterodontus, Hexanchidae and Squalus. The morphological matrix based on Villalobos-Segura et al. (2019) included Hexanchidae and Chimaeridae as outgroups, but while building the mixed-data matrix, we did not have molecular data aligned for Hexanchidae; however, we did have data for Heterodontus and Squalus. We could also generate morphological codings based on Aschliman et al. (2012a) and personal observations for Heterodontus and Squalus, so we included these in both the morphological and mixed-data matrices. †Britobatos primarmatus is excluded from our analyses because some characters were re-coded without explanation by Villalobos-Segura et al. (2019) and were discordant with respect to the codings of Brito & Dutheil (2004) and Claeson et al. (2013), suggesting that a revision of the fossil material is needed. We added 14 additional characters mostly taken from Aschliman et al. (2012a) and Claeson et al. (2013). Additional characters and updated coding are explained in the Supplemental material, File 1. The morphological matrix was compiled in Mesquite v. 3.03 (Maddison & Maddison 2008), and the phylogenetic analysis was performed in PAUP v. 4.0a (build 166) utilizing a heuristic search with stepwise addition, amb(-) and 1000 random addition replicates (Swofford 2002). All characters are unordered and given equal weight. Tree length, consistency and retention indices, and Bremer support were subsequently calculated for the strict consensus tree.
Additional variations of the morphological matrix concern character 5 (calcified suprascapulae: [0] absent, [1] present and independent). Compagno (1999) considered the scapular process to be the unsegmented dorsomedial projection from the scapulocorocoid, and articulating with the scapular process is another small cartilage, the suprascapula. In a paper by Da Silva et al. (2018, figs 1A, 3B) the scapula is defined as the projection from the scapulocorocoid in sharks (e.g. Squalus and Heterodontus) with a segmented scapular process, while in batoids, the scapular process is a non-segmented projection. To account for this variation, we have done the following: (1) retained the coding for character 5, as a suprascapula is present according to Compagno (1999), adjusting all correlated characters (designated CH coding; Supplemental material, File 2); (2) changed the coding of character 5 to follow Da Silva et al. (2018) and adjusted all correlated characters (suprascapular absent in Squalus and Heterodontus, designated DS coding; Supplemental material, File 3); and (3) ran a parsimony analysis excluding Heterodontus and Squalus as in Villalobos-Seguera et al. (2019), updating codings for all characters. In addition, given controversy over the developmental states of the hypobranchial 2 cartilage described by Miyake & McEachran (1991), we also ran an analysis excluding character 85 (hypobranchial shape: [0] straight and segmented, [1] loop/horseshoe shaped, [2] bilateral fused plates, [3] medially fused plates).
The revised morphological data sets (CH and DS codings) were concatenated with the molecular matrix published by Aschliman et al. (2012b) to produce the mixeddata matrix for total evidence analyses. When combining the morphological and molecular data sets, we opted to reduce the amount of missing data by excluding one out of three electric ray taxa, three out of eight skates and nine out of 16 stingray taxa originally included in Aschliman et al. (2012b). There is high support for the monophyly of the clades Torpediniformes, Rajiformes and Myliobatiformes (e.g. Aschliman et al. 2012a, b;Claeson 2014). From our original morphological matrix, taxa with insufficient molecular sequences were excluded: the outgroup taxon Hexanchidae, and the electric ray taxa Hypnos, Narke and Temera. The two resultant mixed matrices (see Supplemental material, File 4 [CH coding] and File 5 [DS coding]) include a total of 42 taxa and 14,108 characters. Codon positions were set per Aschliman et al. (2012b), and the matrix was run in MrBayes for 5 million generations, where variable rates were applied to molecular data as invgamma and to morphological data as gamma (Huelsenbeck & Ronquist 2001). We calculated the clade credibility, which reflects the proportion of trees in the posterior probability sample that share a given node. Parameters are pasted at the end of the Supplemental material, Files 4 and 5, to execute automatically in MrBayes.  Derivation of name. The name is derived from the Greek E os, pertaining to the sunrise, as well as to the goddess of dawn and the epoch from which the taxon is found, plus Platyrhina, a living thornback ray, therefore indicating a close relationship of this latter genus with the new taxon.

Diagnosis. As for the genus.
Description †Eoplatyrhina bolcensis (Heckel, 1851) comb. nov. is represented by three partially complete articulated specimens in part and counterpart (Figs 2, 3), including the holotype (MGP-PD 8873C/8874C) and two additional specimens (MGP-PD 26279C/26280C and MGGC 7449/ 7450). Counts and measurements are listed in the Supplemental material (File 1, Table S1). The examined specimens are similar in size. The largest one measures 84 cm TL and 38 cm DW. The pectoral disc of †Eoplatyrhina gen. nov. is notably expanded, ovoid or shovel shaped, slightly longer than wide and reaching its maximum width just posterior to its mid-length. The snout is broad and rounded. The tail is not very stout, slightly longer than disc length, with two dorsal fins inserting posteriorly on the tail. The overall body shape and proportions are similar to those of the extant thornbacks Platyrhina and Platyrhinoidis.
Neurocranium. The rostral cartilage fails to reach the anterior margin of the disc, as in all platyrhinids. This element is long and tapers gradually anteriorly (Figs 4, 5A), resembling the condition typical of Platyrhinoidis and †Tethybatis, and differs from the short rostrum observed in †Tingitanius and Platyrhina. Unlike other platyrhinids, the anterior margin of the rostral cartilage is not pointed but trough-shaped, with the rostral node slightly expanded laterally (Figs 4, 5A). Rostral appendices at the tip of the rostrum are absent. A small rodlike process lateral to the rostral cartilage and just anterior to the nasal capsule in MGGC 7449/7450 can be interpreted as one of the two rostral processes, which are uniquely present in extant thornbacks. Although McEachran et al. (1996) considered these structures homologous to the rostral appendices of skates and guitarfishes, Carvalho (2004) pointed out that the rostral processes of platyrhinids, originating ventral to the rostral cartilage, might represent outgrowths of the lamina orbitonasalis, unlike the rostral appendices that are secondary chondrifications fused laterally to the rostral node. The nasal capsules are ovoid, laterally expanded, and at right angles to the rostrum, as in †Tethybatis. A single small horn-like process (¼ tab-like process of Claeson et al. 2013) can be recognized on the anterior margin of each nasal capsule, similar to the extant platyrhinids and †Tingitanius. The antorbital cartilages are well developed and plate-like and have an irregular outline. They project laterally from the postero-lateral margin of the nasal capsules and articulate distally with the propterygia. It is difficult to distinguish the preorbital process or the jugal arch, but a small and narrow postorbital process can be recognized in the otic region, just posterior to the supraorbital crest. The orbital region is longer than wide. The anterior fontanel extends through almost the entire length of the rostral cartilage and resembles an isosceles triangle with a close and concave posterior border, similar to the condition seen in †Tingitanius, and in contrast to the oval-shaped fontanel of Platyrhina, or to the figure-eight shape typical of Platyrhinoidis.
Jaws, hyoid and gill arches. Specimens of †Eoplatyrhina bolcensis comb. nov. are mostly preserved in dorsal view, obscuring the jaws, which are displaced and difficult to describe (Fig. 4). For the same reason, teeth are not exposed in any specimen, and therefore their morphology remains unknown. It is also unclear whether the labial cartilages are present, as in mature specimens of Platyrhina. The hyomandibulae are stout, robust and slightly arched, with a concave inner margin, narrow at their medial section. They project anterolaterally. As in †Tethybatis, there is a large space between the hyomandibulae and mandibular arch, which is interpreted by Carvalho (2004) as indicative of the presence of a large spiracular opening. In radiographs, this space is not present in Platyrhinoidis or Platyrhina, while it is present in Zanobatus. The distal part of the hyomandibulae appears taphonomically separated from the Meckel's cartilage. The fifth ceratobranchials articulate with the anterior margin of the scapulocoracoid, and the remaining gill arches are poorly preserved or missing.
Synarcual and vertebral column. Although the synarcual can be identified as a tubular mineralized structure between the neurocranium and scapulocoracoid, its morphology remains ambiguous. The dorsally exposed specimens obscure the pattern of free centra. In †Tingitanius, the first exposed vertebral centrum of the synarcual is located posterior to the articulation of the suprascapular cartilage with the synarcual. In Platyrhina, the first free centrum is situated at the level of the scapulocoracoid articulation with the synarcual. In Platyrhinoidis, the first free centrum is rostral to the scapulocoracoid articulation with the synarcual. The vertebral column of †Eoplatyrhina bolcensis comb. nov. consists of about 132 vertebral centra, in the most complete specimen MGP-PD 26279C/26280C. There are 20-24 trunk centra (from the first distinguishable centrum to the anterior margin of the puboischiadic bar), and 113-118 from the puboischiadic bar to the tip of the tail (of these, about 23 are caudal). The vertebral centra are highly calcified, sub-rectangular in shape and anteroposteriorly compressed. There are about 15 or 16 pairs of ribs.
Appendicular skeleton and fins. It is difficult to describe the morphology of the coracoid bar because the specimens are mostly exposed in dorsal view, but the scapular processes of the scapulocoracoid seem to be short in MGP-PD 26279C/80C (Fig. 5B). This specimen shows a small medially fused suprascapular cartilage; this cartilage is hourglass-shaped, with concave anterior and posterior borders, exhibiting deep indentations into which the distal edges of the scapular processes of the scapulocoracoid fit. Laterally, the scapulocoracoid articulates with the proximal portion of the pterygia through equidistant condyles. The propterygium is long and arched, tapers distally and extends to the anterior disc margin (Fig. 4). The propterygium is segmented, with the first segment lying anterior to the mouth, close to the level of the antorbital cartilage. The proximal section of the propterygium does not extend far posteriorly to the procondyle, and does not articulate with the scapulocoracoid. A single unsegmented mesopterygium seems to be present. The metapterygium is as long and curved as the propterygium, but it is unclear whether it is segmented distally. The pectoral fins are clearly of the plesodic type, with radials reaching the external border of the pectoral disc. All the radials articulate with the pterygia. Each pectoral radial contains 10-12 segments and bifurcates distally only once at about the eighth segment. There are approximately 81-87 pectoral radials, of which 35-38 are propterygial, 8-10 mesopterygial, and 38-41 metapterygial. The pectoral radials of †E. bolcensis comb. nov. are robust, stiff and completely covered by mineralized tissue, forming the so-called 'crustal calcification' typical of most of batoids except the benthic stingrays and skates (Schaefer & Summers 2005). The puboischiadic bar is partly recognizable in MGGC 7449/7450, where it seems straight or slightly bent, narrow and plate-like (Fig. 6). It is difficult to recognize the postpelvic processes on the posterior margin of the puboischiadic bar that are typical for living platyrhinids. There are about 18-21 pelvic radials. The structure of the first pelvic radial is unclear but the pelvic condyles seem close together and not separated as in skates. All the specimens show straight and stout claspers, whose length represents about 10% TL (Fig. 6). As in Platyrhinoidis, their distal extremity does not reach the origin of the first dorsal fin; they differ from those characteristic of Platyrhina, whose clasper tips can extend beyond the first dorsal-fin origin (e.g. Last et al. 2016;. The clasper glands are almost entirely covered by dermal denticles, and consequently their skeletal morphology is difficult to describe. However, the axial cartilage is rod-like, possibly calcified over most of its length, and extends and inserts over the complete length of the clasper to the ventral terminal cartilage. Dorsal and caudal fins. †Eoplatyrhina bolcensis comb. nov. possesses two dorsal fins located in the posterior half of the tail. The extent of the fin radial cartilages into the fin web is not precisely ascertainable, but they are possibly aplesodic. The base of the dorsal fins has a length of about 5% TL. No impression of dorsal-fin radials is visible. The caudal fin is only preserved in   Dermal denticles. As in extant platyrhinids (see Deynat 2005), the entire body of †E. bolcensis comb. nov. is covered with numerous small dermal denticles that form a continuous and regular covering (Fig. 7). Denticle size is quite uniform across the body. Some denticles were extracted from the dorsal side of the disc of MGP-PD 26279C/80C for a detailed analysis. Their crown is about 200 lm wide and rhomboidal or lozenge-shaped (Fig. 7). The denticle root is deeper than the crown height and a nutritive foramen can be recognized near the centre. Extant thornbacks and †Tingitanius possess parallel rows of enlarged dermal denticles (thorns) over the posterior part of the disc and tail, a condition that was regarded as diagnostic for platyrhinids. However, this is not the case for †Eoplatyrhina bolcensis comb. nov. and †Tethybatis, in which thorns are completely absent (Carvalho 2004), possibly representing a feature supporting this sister-group relationship.
Diagnosis. Pectoral disc large and roughly rounded, representing 56-70% TL; tail stout and short, distinctly demarcated from the disc; two dorsal fins and caudal fin present; densely, closely set small dermal denticles forming a continuous pavement; large, rounded, scattered thorns covering the entire disc and tail; rostral cartilage absent; nasal capsules laterally expanded without horn-like processes; long propterygia extending near the anterior margin of the disc; mesopterygium absent; about 65-75 pectoral radials; puboischiadic bar narrow and moderately arched; approximately 20 pelvic-fin rays; 80-90 vertebrae; about 10 pairs of ribs.
Derivation of name. From the Ancient Greek word pkgrί o (pl es ıon) meaning 'near' or 'close', and Zanobatus, to remark upon its close relationship with the living panray genus.
Included species. Type species only.
Remarks. De Zigno (1876) considered that the overall similarity of the disc shape and the absence of a tail sting on the holotypic specimen MGP-PD 154Z justified the assignment of this species to the genus Torpedo. Later, Jaekel (1894), analysing additional, better preserved material, assigned the species †T. egertoni to Platyrhina. However, he noticed that the fossil species from Bolca might have been more closely related to Platyrhina schoenleinii than to Platyrhina sinesensis because of the general shape and proportions of the body and disc, as well as the arrangement of the pectoral radials and gill arches. Platyrhina schoenleinii is currently recognized as Zanobatus schoenleinii (see Compagno 1999).   fig. 13A, B.
Diagnosis. As for the genus.
Description †Plesiozanobatus egertoni comb. nov. is represented by six specimens showing different ontogenetic stages, with the largest individual measuring more than 1 m in length (Fig. 8). Counts and measurements are shown in the Supplemental material (File 1, Table S2). The pectoral disc is large and nearly round, representing 56-70% TL. The tail is stout and short, distinctly demarcated from the disc and measuring about 40-50% TL (Fig. 9A-B). The most complete specimens show two nearly triangular dorsal fins of similar size, located well behind the pelvics; a nearly complete caudal fin is visible exclusively in MCSNV IG.43347. Although the general body shape is still detectable, a detailed analysis of all the skeletal structures is very difficult due to the generally poor preservation of the available specimens. The rostral cartilage is clearly absent in all the specimens, and a large empty space can always be recognized between the anterior propterygial radials (Fig. 9C, D). The nasal capsules are laterally expanded and do not show evidence of the horn-like processes typical of platyrhinids. The antorbital cartilages are difficult to detect but they probably articulated with the mesial margin of the propterygia. The propterygia are long, extending close to the anterior margin of the disc, well beyond the nasal capsules. The mesopterygium appears absent, as in Zanobatus, Gymnura and some pelagic stingrays, suggesting that the mesocondyle (not visible) might have been replaced by a ridge. There are about 65-75 highly calcified pectoral radials (¼ 'crustal pattern' of Schaefer & Summers 2005). Most of them articulate with the pterygia and some others articulate directly with the scapulocoracoid. The puboischiadic bar is scarcely visible in all the specimens and appears as a narrow and moderately arched bar at least in MB.f 1608.1/2. About 20 pelvic-fin rays can be recognized in the pelvic fins of †P. egertoni comb. nov. The most complete specimens exhibit 80-90 vertebrae and around 10 pairs of ribs. Small, imbricated and densely set dermal denticles form a continuous pavement throughout the body (Fig. 10A); their crowns are roughly rhomboid or polygonal in shape, with a flat and smooth surface. Large rounded thorns are more widely spaced, sparse and cover the whole pectoral disc and tail (Fig. 10B), whereas some scattered star-shaped thorns cover the scapular region (Fig. 10C). However, parallel anteroposteriorly directed rows of thorns are clearly absent. There are no teeth preserved in the available specimens.

Phylogenetic analyses Parsimony
The tree statistics for the phylogenetic analysis of morphological data performed using PAUP are available in Table 1, and consensus tree topologies are compared in Figure 11. The consensus tree topological hypotheses recovered are identical with respect to the matrix coding of the suprascapulae according to Compagno (1999 Fig. 11A), though tree scores are different. The hypotheses are also identical (except for the placement of Hexanchidae), and much better resolved with the exclusion of character 85, which refers to the shape of the second hypobranchial (Fig. 11B). Character mapping is provided on the tree topology of Figure 11B and in the Supplemental material (File 1, Fig. S1). We also performed an analysis where Heterodontus and Squalus were excluded, following a reviewer's comments regarding the outgroups included and their coding. Of note, there is no difference with respect to the ingroup hypothesis recovered in Figure 11B.
As the morphological matrix was primarily modified from Villalobos-Segura et al.  and Myliobatiformes. However, the relationships among these major clades differ from the hypothesis of Villalobos-Segura et al. (2019). To begin with, the outgroup to all remaining batoids is the Torpediniformes, not the Jurassic batoids (see Fig. 11). In our analysis that included character 85, the Jurassic batoids, sclerorhynchoids and remaining batoids form a polytomy (Fig. 11A). With character 85 excluded, Jurassic batoids and sclerorynchoids are each other's closest sister taxa, together forming a sister relationship to Rhinopristiformes. Rajidae (Raja þ Bathyraja) are nested among Rhinopristiformes. Rajiformes as defined by Villalobos-Segura et al. (2019) was a clade of extant skates, sister taxon to the extinct clade of sclerorhyncoid batoids, with the latter being clearly separated from the phenetically similar sawfishes (e.g. Pristis). This relationship is not recovered by our parsimony analyses. As recovered in Villalobos-Segura et al. In this section, we describe the parsimony hypotheses recovered with the exclusion of character 85 in our study ( Fig. 11B and Supplemental material, File 1, Fig.  S1). Batoidea are supported to the exclusion of their outgroups by 10 unambiguous character transformations. Among Batoidea, Torpediniformes form a monophyletic clade with 14 unambiguous character transformations, which is resolved as sister taxon to all remaining batoids. The clade of remaining batoids is supported by five unambiguous character transformations.
The clade of Jurassic þ sclerorhynchoid batoids is supported by two unambiguous character transformations: calcified suprascapulae are absent (ch. 5[1 ! 0]) and a preorbital process is absent (ch. 33[0 ! 1]). Rhinopristiformes is supported by three unambiguous character transformations: a scapulocoracoid that is elongate between the mesocondyle and metacondyle (ch. 56[0 ! 1]), some pectoral-fin radials that articulate directly with the scapulocoracoid (ch. 60[0 ! 1]), and the presence of differentiated lateral uvulae on teeth 3 Figure 9. †Plesiozanobatus egertoni (De Zigno, 1876) comb. nov. from the Pesciara site. A, MB.f 1608.1; B, reconstruction of the body outline and main skeletal structures (denticles and radials omitted); C, close-up of the head and pectoral girdle of MCSNV VII.B.81; D, reconstruction. Abbreviations: af, anterior fontanel; cb, ceratobranchials; cf, caudal fin; df, dorsal fins; hyo, hyomandibula; mc, Meckel's cartilage; met, metapterygium; nc, nasal capsules; pf, pelvic fins; pq, palatoquadrate; pro, propterygium; pub, puboischiadic bar; rad, pectoral radials; sca, scapulocoracoid; sy, synarcual; ss, suprascapula; vc, vertebral centra. Scale bars ¼ 50 mm.  as well as pectoral radials that also reach beyond Figure 12. Phylogram recovered under Bayesian analyses of the total evidence data sets. A, hypothesis based on coding the suprascapula according to Compagno (1999); B, grey inset reflects how the hypothesis of relationships differs from the tree presented in A when derived from a matrix coding the suprascapula according to Da Silva et al. (2018) Figure 11. Comparative strict consensus trees from parsimony analyses run in PAUP showing the hypothetic relationships of †Eoplatyrhina gen. nov. and †Plesiozanobatus gen. nov. (in bold) among batoids. A, consensus tree of total dataset with suprascapular coding according to Compagno (1999) and to Da Silva et al. (2018); B, consensus tree of total dataset with suprascapular coding according to Compagno (1999) and Da Silva et al. (2018)  Most pertinent to our study is the position of the Bolca fossils traditionally considered members of the thornback ray family Platyrhinidae. The family forms here a monophyletic clade, sister to the grouping formed by Zanobatidae þ Myliobatiformes in both CH and DS analyses (Fig. 11A, B). This arrangement is consistent with the results of McEachran et al. (1996) and Aschliman et al. (2012a), but contrasts with the most recent molecular studies that place the platyrhinids as sister to the electric ray order Torpediniformes (Naylor et al. 2012;Bertozzi et al. 2016;Last et al. 2016). These differences between molecular and morphological analyses are justified by the absence of unambiguous morphological synapomorphies shared by Torpediniformes and Platyrhinidae (see also Villalobos-Segura et al. 2019). The relationship of platyrhinids and zanobatids forming successive sister taxa to myliobatiforms, detected in our study, also contrasts with the recent morphological analysis of Brito et al. (2019) who recovered the clade Platyrhinidae þ †Britobatos as the sister group of the node formed by the clade †Stahlraja þ ( †Tlalocbatos þ (Aptychotrema þ Zapteryx þ Trygonorrhina)), with this relationship supported by two homoplastic characters: pectoral radials extending far beyond the nasal capsules, and scapulocoracoid elongated between mesocondyle and metacondyle (ch. 34[2] and ch. 43[1] of Brito et al. 2019). However, in our study these two features appear independently derived for platyrhinids and trygonorrhinids.
In our analyses, the monophyly of Platyrhinidae is supported by the presence of two unambiguous character transformations: rostral processes (ch. 30[0 ! 1]; consistency index [CI] ¼ 1.00), and horn-like processes on the anterior margin of nasal capsules (ch. 79[0 ! 1]). The presence of well-developed antorbital cartilages, variously shaped and with irregular outline (ch. 9[1]), has been used by Villalobos-Segura et al. (2019) to provide a shared feature between platyrhinids and electric rays. However, in our analysis this feature appears independently derived for the two clades. †Eoplatyrhina gen. nov. is recovered as a genuine thornback ray that is sister to †Tethybatis. They share the absence of thorns (ch. 97[0]). †Tethybatis is distinguished from †Eoplatyrhina in possessing long claspers (ch. 67[0 ! 1]. Platyrhinoidis is recovered as sister to †Tingitanius þ Platyrhina, supported by the presence of parallel rows of enlarged denticles (ch. 80[1]; CI ¼ 1.00). †Tingitanius þ Platyrhina are distinguished from Platyrhinoidis in possessing a pair of long claspers (ch. 67[1]). Furthermore, the position of the first enclosed vertebral centrum within the synarcual of Platyrhinoidis is at the level of the suprascapular articulation with the synarcual, rather than posterior to it (ch. 78[2 ! 1]). This placement of †Tingitanius contrasts with the results of Claeson et al. (2013) who recovered †Tingitanius as sister to Platyrhinoidis because of the absence of labial cartilages and incipient lateral uvulae on teeth. However, updated coding in our matrix for the absence/ presence of lateral uvulae, following Villalobos-Segura et al. (2019), leads to a hypothesis that considers the absence of labial cartilages in Platyrhinoidis and †Tingitanius to be independently derived. Zanobatidae is recovered as sister taxon to Myliobatiformes (Fig. 11A, B) Conversely, Naylor et al. (2012) recovered Zanobatus as a genuine member of the Rhinopristiformes, although the authors pointed out that this placement was modeldependent for that dataset.
Within Zanobatidae, †Plesiozanobatus gen. nov. is recovered as sister to Zanobatus and is distinguished from Myliobatiformes by possessing pectoral radials that directly articulate with the ridge replacing the mesopterygoid (ch. 60[0 ! 1]) (see McEachran et al. 1996, fig. 9C). A similar condition where pectoral radials are directly articulated with the scapulocorocoid has been derived independently from the Rhinopristiformes. Myliobatiformes is distinguished from Zanobatidae by 14 unambiguous character transformations, mapped in Supplemental material, File 1, Fig. S1.

Bayesian analysis
All major clades of batoids are recovered in the total evidence analyses that accounted for alternate codings of the suprascapular cartilages in outgroup taxa (CH, DS) and excluded the developmentally variable character of the second hypobranchial. Jurassic batoids are recovered in a polytomy among outgroup Chimaeridae and crown Neoselachii (Fig.  12). The position of Rajidae is the primary difference between the DS and CH hypotheses of batoid relationships.
The ultimate structure of the Bayesian phylogram from the CH analysis resembles that published by Aschliman et al. (2012b), in that among crown Batoidea, Rajiformes is the sister taxon to remaining crown batoids (clade credibility CH ¼ 54; Fig. 12A). The alternative hypothesis has weak support for a monophyletic clade of Rajiformes (clade credibility DS ¼ 54; Fig. 12B) that includes the extinct sclerorhynchoid batoids. When using CH and DS coding, †Sclerorhynchus and †Libanopristis are sister taxa and in a polytomy with †Ptychotrygon þ †Aslflapristis (Fig. 12A, B). Sclerorhynchoids, whether in a sister taxon relationship with Rajidae or not, are the sister taxon to the remaining batoids.
Also resembling the Aschliman et al. (2012b) hypothesis, the position of Torpediniformes as the sister taxon to Platyrhinidae is present in both analyses. Rhinopristiformes is paraphyletic. The 'guitarfish-1' group of Aschliman et al. (2012b) now includes Aptychotrema as the sister taxon to Trygonorrhina þ Zapteryx, as in the morphological hypothesis. In the morphological hypothesis, 'guitarfish-1' is highly nested within the Rhinopristiformes and sister to Rhinobatos. The 'guitarfish-2' group is identical to that of Aschliman et al. (2012b). The position of Rajiformes (Fig. 12B) contrasts with that of Bertozzi et al. (2016), who recover Torpediniformes as the sister taxon to all remaining crown Batoidea. Within Myliobatiformes, the relative position of Urobatis changes based on CH/DS coding from sister to Myliobatidae or sister to Urolopus þ Gymnura, respectively, with CH coding more similar to the Aschliman et al. (2012b) hypothesis.
As it pertains to the new fossils from Bolca, Platyrhinidae is predicted with a clade credibility of 100 and, as in the parsimony hypothesis, †Eoplatyrhina gen. nov. is sister taxon to †Tethybatis (clade credibility CH ¼ 53; clade credibility DS ¼ 51), Platyrhina is sister taxon to †Tingitanius (clade credibility CH ¼ 61; clade credibility DS ¼ 64) and Platyrhinoidis is unresolved relative to the other thornback rays. †Plesiozanobatus is sister taxon to Zanobatus with a clade credibility of 97 in both analyses, and together they are sister taxon to a monophyletic Myliobatiformes (clade credibility ¼ 100).

Discussion
Notes on †'Platyrhina' gigantea (Blainville, 1818) A single specimen in part and counterpart (MNHN F.Bol567) housed in the Mus eum National d'Histoire Naturelle, Paris (Fig. 13) was figured and assigned by Volta (1796, pl. 61) to Raja torpedo, which is currently a junior synonym of Torpedo torpedo (Linnaeus, 1758). Blainville (1818), without further description or figure, created a new species ( †Narcobatus giganteus) based on that specimen (Narcobatus is a junior synonym of Torpedo), whereas Molin (1860) assigned it to the genus Narcine. De Zigno (1874) reported another specimen housed in MCSNV, whose measurements may correspond to those of MCSNV VII.B.80/81 (assigned herein to †Plesiozanobatus gen. nov.; Fig. 8F), and described it as Torpedo gigantea (labelled in the MCSNV as †Platyrhina gigantea, authors' pers. obs.). Jaekel (1894) was unable to locate the specimen figured by Volta (1796) and, solely based on the poorly detailed drawing provided by Volta, concluded that the species should have been assigned to Platyrhina (see Eastman 1904Eastman , 1905. The anatomical analysis of specimen MNHN F.Bol567 is extremely problematic, it being a specimen preserved in a heavy limestone slab more than 2 m long and mounted very high on a wall at the MNHN. A cursory analysis of this badly preserved and possibly deformed specimen detected a short but slender tail, two dorsal fins and a caudal fin, but the pectoral disc was unlikely to have been anteroposteriorly elongated. No cranial or postcranial structures are recognizable. In addition, the specimen seems to have been erroneously assembled, and possibly painted, making it very difficult to interpret reliable diagnostic characters and thereby preventing a possible assignment to any known batoid taxon or group. Due to the extremely problematic taxonomic interpretation of this specimen, we therefore suggest †Platyrhina gigantea (Blainville, 1818) be considered a nomen dubium.

Comparison and relationships
The monophyly of the family Platyrhinidae has been defined by the presence of rostral processes, postpelvic processes on the puboischiadic bar, plate-like irregularly-shaped antorbital cartilages, and the rostral cartilage failing to reach the tip of the snout (Carvalho 2004;McEachran & Aschliman 2004;Aschliman et al. 2012a;Claeson et al. 2013 Table S3). The number of pectoral radials in †Eoplatyrhina gen. nov. is higher than in all the extinct and living platyrhinids, whereas its claspers are short, with the distal extremity failing to reach the first dorsal fin, unlike the very long claspers characteristic of Platyrhina and †Tingitanius. Low clade credibility among the interrelationships of Platyrhinidae, in particular the fragility of the relationships of the Cretaceous taxon †Tingitanius among Platyrhinidae based on recoding single character states, may reflect the limited number of currently identified apomorphies of extinct and extant taxa. Despite this limited number of apomorphies, we recover a sister-taxon relationship between the Eocene thornback rays †Eoplatyrhina þ †Tethybatis. Based on our analyses, Zanobatidae is unambiguously recovered as sister taxon to Myliobatiformes, using parsimony and Bayesian inferences (Figs 11, 12). Zanobatidae is no longer monotypic, now defined as Zanobatus þ †Plesiozanobatus egertoni comb. nov. with a clade credibility of 97 and one unambiguous character transformation of certain pectoral fin radials articulating directly with the ridge replacing the mesopterygia (ch. 60[0 ! 1]). A similar condition, where pectoral radials are directly articulated with the scapulocorocoid, is derived independently in the Rhinopristiformes. We consider this character to warrant thorough reexamination and worthy of a developmental study to further distinguish these morphologies. The extant panray Zanobatus includes two species (Z. schoenleinii and Z. maculatus) whose meristic features and bodily proportions (S eret 2016) are considerably different from those of †Plesiozanobatus gen. nov. (see Supplemental material, File 1, Table S3). We therefore consider †Plesiozanobatus gen. nov. to be unambiguously sister to the extant Zanobatus.

Bayesian notes and outgroup impact
The CH parsimony analysis for the morphological data set that included Chimeridae, Hexanchidae, Heterodontus and Squalus resulted in 16 most parsimonious trees when character 85 is excluded. Eight most parsimonious trees resulted for the DS analysis when character 85 is excluded, the consensus trees of which (Fig. 11B) recovered the Torpediniformes as the sister taxon to all other batoids, and Platyrhinidae and Zanobatidae forming successive sister taxa to Myliobatiformes, as in Aschliman et al. (2012a). Unlike Aschliman et al. (2012a), there was more resolution among 'guitarfish' groups, in congruence with the first morphological hypothesis presented here, i.e. Rajiformes and Rhinopristiformes are recovered as monophyletic. The most novel aspect of these hypotheses is the relationship between the sclerorhynchoid batoids and a monophyletic Jurassic batoid clade, which are paraphyetic with crown batoids (CH) or sister to each other (DS). The Bayesian analysis recovers Jurassic batoids as outside Euselachia.
During the first iteration of the Bayesian analysis for this study, only sequence data for Heterodontus were added, and Hexanchidae was excluded because we had too few molecular data for this taxon. The results of that study, however, predicted that Heterodontus was nested among the extinct Rajiformes, i.e. sclerorhyncoid sawfishes. This seemed flawed considering we had yet to score the morphology for Heterodontus and could not obtain sequence data for the extinct sclerorhyncoid sawfishes. Thus, we added the morphological data for Heterodontus to the matrix and also included total evidence for Squalus. The ultimate hypothesis resulted in a sister-group relationship among Heterodontus þ Squalus and a monophyletic crown Batoidea. This remained the case after several iterations and variations of coding among outgroup and ingroup taxa, as per suggestions by an anonymous reviewer (Table 1; Figs 11,12).
Furthermore, there were several characters warranting additional scrutiny. Namely, there are differing interpretations about the pectoral morphology and branchial morphology in elasmobranchs that might impact character transformations and interpretations of phylogeny. We therefore prepared six variations of character coding and outgroup taxa included (Table 1). The variations of the morphological matrix were with regard to character 5, the presence of a scapulocoracoid. Compagno (1999) considers the scapular process to be the unsegmented dorsomedial projection from the scapulocorocoid. Articulating with the scapular process might be another small cartilage, the suprascapula. In sharks, we considered the segmented distal portion off the scapular process to be a suprascapula, as in the case of Squalus. In Da Silva et al. (2018, figs 1A, 3B), scapular morphology is discussed for Squaliformes, where the projection from the scapulocorocoid is defined as the 'scapula' in sharks (e.g. Squalus and Heterodontus) with a segmented 'scapular process', while in batoids, a nonsegmented projection is the scapular process. This segmental scapular process is what we considered to be the suprascapula of Compagno (1999). We coded Compagno (CH) and Da Silva (DS) independently; the result was reasonably well-resolved consensus trees among major clades of batoids with the CH and DS codings being nearly identical to each other, with the exception of the position of outgroups in a polytomy (Fig. 11). Further, there was a great deal of resolution when character 85 was excluded (Fig. 11B). We also ran parsimony analyses excluding Heterodontus and Squalus as outgroups, as in Villalobos-Segura et al. (2019); this had no impact on the ingroup topology once character 85 was removed (Fig. 11B). Generally, there were no major differences in the outcomes of the ingroup relationships. With this aspect of morphology in particularbranchial element developmentwe note there will be a great benefit from conducting more ontogenetic studies to understand the early life stages of elasmobranchs and their usefulness for interpreting homologies among elasmobranch species.

Palaeoecology, palaeobiogeography and evolutionary significance
The palaeoecological role of platyrhinids and zanobatids from the Bolca Lagerst€ atte has never been investigated. All the specimens of †Eoplatyrhina bolcensis comb. nov. are from the Monte Postale site. Living representatives of the Platyrhinidae are inshore batoids today represented by four species of Platyrhina, and a single species of Platyrhinoidis occurring in warm-temperate to tropical coastal marine waters of the north-western and eastern Central Pacific, mostly occurring off sandy beaches, in muddy enclosed bays, and near kelp beds and shallow mud bottoms (Compagno & Last 1999;Iwatsuki et al. 2011;Last et al. 2016). Quantitative palaeoecological and taphonomic analyses of the fish assemblage of Monte Postale suggests that the fossiliferous sediments accumulated close to an emerged coastal area characterized by mangroves and seagrass, in a coral reef context in the western Tethys (Marram a et al. 2016;Vescogni et al. 2016). From this perspective, it is reasonable to suggest that the Bolca platyrhinids inhabited the warm shallow-water habitats of the Monte Postale palaeobiotope (Marram a et al. 2016;Vescogni et al. 2016). In addition, it is interesting to note that among the coeval Tethyan and Boreal Eocene deposits, the presence of thornback rays of the family Platyrhinidae has been reported only from Bolca and Fayum in Egypt, suggesting similar palaeoecological and palaeoenvironmental features in these two Tethyan areas (Underwood et al. 2011;Marram a et al. 2018c). This hypothesis is corroborated by the shared presence of small odontaspidid and carcharhinid sharks, which are generalist feeders on small nectobenthic prey and zooplanktivorous coastal bony fishes that represented a relevant trophic resource in the Bolca palaeobiotopes .
Conversely, all the specimens of †Plesiozanobatus egertoni comb. nov. are from the Pesciara site. The presence of zanobatids in the Pesciara site is consistent with the presence of tropical marine shallow waters hypothesized for the Pesciara palaeobiotope (Marram a et al. 2016), because extant panrays inhabit the shallow coastal waters off the eastern central African coast mainly between 10 and 15 m, but also reaching depths of about 100 m S eret 2016).
The fossil record of Platyrhinidae and Zanobatidae is very poor (Fig. 14), but this is likely an artefact, since their teeth might have been misidentified as belonging to Rhinobatos, which has been traditionally used as wastebasket genus for many fossil teeth exhibiting a 'rhinobatoid' morphology (Kriwet et al. 2009;Cappetta 2012;Claeson et al. 2013). Fossils of thornback rays and panrays have been reported so far only from the Late Cretaceous to Eocene deposits of the Tethys area, if we exclude a single occurrence from the Pleistocene of California (Fig. 14). Today, platyrhinids are restricted to temperate to tropical marine coastal waters of the north-western and eastern Central Pacific Ocean, whereas zanobatids are only present along the western coast of Africa . Molecular analyses suggest that platyrhinids diverged from torpediniforms around 200-175 Ma ago, whereas the clade including Zanobatus separated from myliobatiforms around 150 Ma (Aschliman et al. 2012b;Bertozzi et al. 2016). If these hypotheses are confirmed, it is evident that a large ghost range will characterize the fossil record of these batoid lineages, being the oldest known representatives for platyrhinids and zanobatids of Turonian (c. 89 Ma) and Ypresian (c. 50 Ma) ages, respectively. The fossil records of both platyrhinids and zanobatids are concentrated in the Tethys, thereby supporting the possibility of a Tethyan origin for these clades, as suggested by Carvalho (2004) and Claeson et al. (2013).

Supplemental material
Supplemental material for this article can be accessed here: http://dx.doi.org/10.1080/14772019.2020.1783380.