The Phylogeny of Rays and Skates (Chondrichthyes: Elasmobranchii) Based on Morphological Characters Revisited

Elasmobranchii are relatively well-studied. However, numerous phylogenetic uncertainties about their relationships remain. Here, we revisit the phylogenetic evidence based on a detailed morphological re-evaluation of all the major extant batomorph clades (skates and rays), including several holomorphic fossil taxa from the Palaeozoic, Mesozoic and Cenozoic, and an extensive outgroup sampling, which includes sharks, chimaeras and several other fossil chondrichthyans. The parsimony and maximum-likelihood analyses found more resolved but contrasting topologies, with the Bayesian inference tree neither supporting nor disfavouring any of them. Overall, the analyses result in similar clade compositions and topologies, with the Jurassic batomorphs forming the sister clade to all the other batomorphs, whilst all the Cretaceous batomorphs are nested within the remaining main clades. The disparate arrangements recovered under the different criteria suggest that a detailed study of Jurassic taxa is of utmost importance to present a more consistent topology in the deeper nodes, as issues continue to be present when analysing those clades previously recognized only by molecular analyses (e.g., Rhinopristiformes and Torpediniformes). The consistent placement of fossil taxa within specific groups by the different phylogenetic criteria is promising and indicates that the inclusion of more fossil taxa in the present matrix will likely not cause loss of resolution, therefore suggesting that a strong phylogenetic signal can be recovered from fossil taxa.


Introduction
Batomorpha (hereafter used to refer to a level above superorder level equivalent to Selachimorpha) is the largest subgroup of the Elasmobranchii sensu [1]; presently, they comprise 26 families and approximately 633 valid species [2]. Batomorphs differ from other elasmobranchs in having their five gill slits (six in one species) located on the ventral surface of the head; presenting a mostly dorsoventrally flattened body, with their eyes situated on the dorsal surface and their pectoral fins fused with the head and trunk forming a disc; and a lack of an anal fin [3].
Throughout their evolutionary history, batomorphs have successfully colonized various niches. Currently, most batomorph species are either benthic (living on the substrate) or benthopelagic (swimming close to the bottom but not resting on the substrate), with some members of the group being active pelagic swimmers and others living on the deep continental slope [3].  Figure 3A in [51] and (C) Raja clavata modified from Steven Text- Figure 8 in [52]. Arrowheads: (red) nasal curtain, (back) nasal curtain fringes, (gray) anterior nasal lobe.   Figure 3A in [51] and (C) Raja clavata redrawn and modified from Steven Text- Figure 8 in [52]. Arrowheads: (red) nasal curtain, (back) nasal curtain fringes, (gray) anterior nasal lobe.  [7] (char. 24). The original character was split into two different ones (83)(84) to increase the grouping information regarding the variation on the canals on the coracoid bar. The coding for Pristis was changed based on Wueringer et al. Text- Figure 1 in [54].

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The lack of an abdominal canal on the coracoid bar is the plesiomorphic state for the chondrichthyan tree. Within batomorphs, the electric rays (clade 18;) and stingrays (clade 16) retain the plesiomorphic condition. The presence of a canal is widely distributed among batomorphs (Rajiformes: Raja; Torpediniformes: Platyrhina, Platyrhinoidis; Rhinopristiformes: Rhynchobatus, Glaucostegus, Rhinobatos, Pseudobatos, Trygonorrhina, Zapteryx, Aptychotrema, and Myliobatiformes: Zanobatus), and according to the present topology, this is considered a single gain between these groups.
MLtr (see discussion Maximum Likelihood tree): The loss of the canals is recovered as independent events and a synapomorphy of clades 18 [7] (char. 24). The original character was split into two different ones (83)(84) to increase the grouping information regarding the variation on the canals on the coracoid bar. The coding for Pristis was changed based on Wueringer et al. Text- Figure 1 in [54].

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The lack of an abdominal canal on the coracoid bar is the plesiomorphic state for the chondrichthyan tree. Within batomorphs, the electric rays (clade 18;) and stingrays (clade 16) retain the plesiomorphic condition. The presence of a canal is widely distributed among batomorphs (Rajiformes: Raja; Torpediniformes: Platyrhina, Platyrhinoidis; Rhinopristiformes: Rhynchobatus, Glaucostegus, Rhinobatos, Pseudobatos, Trygonorrhina, Zapteryx, Aptychotrema, and Myliobatiformes: Zanobatus), and according to the present topology, this is considered a single gain between these groups.
MLtr (see discussion Maximum Likelihood tree): The loss of the canals is recovered as independent events and a synapomorphy of clades 18  The coding for †Cobelodus and †Ozarcus follows Maisey [18] and Pradel et al. [22]. The reduced rostrum in these taxa seems to be a product of the outgrowth of the posterior portion of the ethmoidal region (possibly trabecula) with little to no contribution of the lamina orbitonasalis Text- Figures 2, 8, 10, 39 and 48A,B in [18]. The coding for hybodonts uses the topological relations of the rostral bar and caudal internasal keel Text- Figure 2 in [57], Text- Figure 3a,b in [58]. In the fossil genera †Ostarriraja and †Arechia, due to the lack of preservation, we were unable to determine the state of this character (?) (see [30,59]). However, being batomorphs, it is very likely that their rostral cartilages, although reduced, arose from the interaction between the medial area of the trabecula and the lamina orbitonasalis.
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): Rostral cartilages that topologically seem to have arisen for the medial area of the trabecula in interaction with the lamina orbitonasalis is a synapomorphy for the euselachian clade (Hybodontiformes + Elasmobranchii). The absence of these interactions in symmoriids and holocephalians is a common feature between these groups and the plesiomorphic feature for the chondrichthyan tree.
omorphic state for batomorphs. The presence of pores is recovered as shared feature for the taxa within clade 24, being present in Trygonorrhina, Zapteryx, Aptychotrema, Platyrhina, Platyrhinoidis and Zanobatus and inapplicable for the rest of myliobatiforms and electric rays.

Neurocranium 3
Rostral cartilages: (0) Arise from the medial area of the trabecula only; (1) medial area of the trabecula + lamina orbitonasalis. According to De Beer and Moy-Thomas [55] and Miyake et al. [56], no evidence suggests the homology between the rostral cartilages in elasmobranchs and holocephalians, as the rostral cartilages of holocephalians arise from the medial area of the trabecula possibly without any contribution from the lamina orbitonasalis. Conversely, in most modern elasmobranchs, these two embryological cartilages (medial trabecula and the lamina orbitonasalis) contribute to the formation of the rostral cartilages ( Figure 4).
The coding for †Cobelodus and †Ozarcus follows Maisey [18] and Pradel et al. [22]. The reduced rostrum in these taxa seems to be a product of the outgrowth of the posterior portion of the ethmoidal region (possibly trabecula) with little to no contribution of the lamina orbitonasalis Text- Figures 2, 8, 10, 39 and 48A,B in [18]. The coding for hybodonts uses the topological relations of the rostral bar and caudal internasal keel Text- Figure 2 in [57], Text- Figure 3a,b in [58]. In the fossil genera †Ostarriraja and †Arechia, due to the lack of preservation, we were unable to determine the state of this character (?) (see [30,59]). However, being batomorphs, it is very likely that their rostral cartilages, although reduced, arose from the interaction between the medial area of the trabecula and the lamina orbitonasalis.
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): Rostral cartilages that topologically seem to have arisen for the medial area of the trabecula in interaction with the lamina orbitonasalis is a synapomorphy for the euselachian clade (Hybodontiformes + Elasmobranchii). The absence of these interactions in symmoriids and holocephalians is a common feature between these groups and the plesiomorphic feature for the chondrichthyan tree.

4.
Rostral cartilage: (0) Well-developed rostral plate with various degrees of contribution from the lamina orbitonasalis; (1) reaches the tip of the snout (carried by the growth of the pectoral fin); (2) reaches the tip of the snout (growth of lamina orbitonasalis to support the cephalic fins). Modified from Aschliman et al. [7] (char. 26). This character aims to include the variation observed in Neoselachii (sensu [12]), following observations made by Miyake et al. [56], Maisey [57] and Lane [58]. Many taxa remain uncharacterized as the present coding provides grouping information for those taxa whose rostral cartilages arise from the interaction between the medial area of the trabecula and lamina orbitonasalis.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
Presence of well-developed rostral cartilages ( Figure 5B,D) is the plesiomorphic condition for euselachians. Within batomorphs, there is the appearance of two additional states: (1) rostral cartilages located between the tip of the pectoral fins that reach the tip of the  Figure 5A), which are a shared feature among stingrays being present in Gymnura, Potamotrygon, Urotrygon, Urolophus and †Asterotrygon; (2) growth of lamina orbitonasalis to support the cephalic fins ( Figure 5C), which is a synapomorphy for the Mobula + Rhinoptera clade.
lowing observations made by Miyake et al. [56], Maisey [57] and Lane [58]. Many taxa remain uncharacterized as the present coding provides grouping information for those taxa whose rostral cartilages arise from the interaction between the medial area of the trabecula and lamina orbitonasalis.
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): Presence of well-developed rostral cartilages ( Figure 5B,D) is the plesiomorphic condition for euselachians. Within batomorphs, there is the appearance of two additional states: (1) rostral cartilages located between the tip of the pectoral fins that reach the tip of the snout ( Figure  5A), which are a shared feature among stingrays being present in Gymnura, Potamotrygon, Urotrygon, Urolophus and †Asterotrygon; (2) growth of lamina orbitonasalis to support the cephalic fins ( Figure 5C), which is a synapomorphy for the Mobula + Rhinoptera clade.
MLtr (see discussion Maximum Likelihood tree): The rostral cartilages located between the tip of the pectoral fins, reaching the tip of the snout ( Figure 5A), are recovered as a synapomorphy of clade 29 with an additional gain in Urotrygon.

MLtr (see discussion Maximum Likelihood tree):
The rostral cartilages located between the tip of the pectoral fins, reaching the tip of the snout ( Figure 5A), are recovered as a synapomorphy of clade 29 with an additional gain in Urotrygon.

5.
Medial growth of rostral cartilage: (0) Inconspicuous; (1) conspicuous (noticeable). Modified from Aschliman et al. [7] (char. 26), Villalobos-Segura et al. [32] (char. 27) and Claeson et al. [23] (char. 2). De Beer and Moy-Thomas [55] named the cranial projections observed in chimaeroids ( Figure 6A-C) as "medial and lateral rostral processes". This nomenclature is kept by Claeson [60] and used to recognize the structures on the rostral cartilages of electric rays. However, the topological homology of these structures is unclear. Because of this, we coded the medial growth of rostral cartilage in chimaeroids as inapplicable (-). The coding of this character for Ginglymostoma follows Motta and Wilga Text- Figures 8 and 9 in [61]. The character state is inconspicuous (0) for both Torpedo and Hypnos ( Figure 6E,F); in Hypnos, the antorbital cartilages and nasal capsules support the anterior extension of the pectoral disc, and in Torpedo, the rostral cartilages seem to arise primary from the anterior wall of the nasal capsules and interact with the antorbital cartilages. In Narke, there are tripodal rostral cartilages with lateral and medial growths. Narcine also shows a noticeable development of its rostral cartilages. We consider the observations by Miyake [62] in Urolophus, Urotrygon and Potamotrygon to be correct, as these taxa present evident vestiges of the rostral cartilages, such as those in Gymnura Text- Figure 1 in [63], Text- Figure 1 in [64]. Aschliman et al.'s [7] coding for Urobatis was kept, as we were unable to confirm the observations of Miyake et al. [56] and McEachran et al. [6] on the presence of these vestiges. Rhinoptera and Mobula ( Figure 6F) present a lateral growth on the trabecula and lamina orbitonasalis to support the cephalic fins [56]. Zanobatus and †Plesiozanobatus show a small medial growth of the rostrum ( Figure 6D). De Carvalho et al. [65] noticed in specimens of †Asterotrygon (NHMUK P 61244; PF 15180) the rostral projections and medial growth. The rostral cartilages in †Heliobatis, Plesiobatis, Hexatrygon, Hypanus, Potamotrygon, Neotrygon, Aetobatus, Myliobatis, †Lessiniabatis, †Tethytrygon and †Promyliobatis are inconspicuous. The rostral cartilages in the most rajoids (not in Sympterygia), and in all sclerorhynchoids, extant and extinct platyrhinids and Rhinopristiformes, are noticeable.
MLtr (see discussion Maximum Likelihood tree): The lack of rostral-cartilage growth ( Figure 6A) is an independent loss and a synapomorphy for [Hypnos + Torpedo] and clade 31.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of two noticeable "types" of cartilages in the rostrum is not exclusive of †Onchopristis and †Ischyrhiza, as different types of cartilage have been observed in Lamniformes [66]. However, the presence of a highly porous peripheral cartilage at the sides of the rostrum and fibrous wood-like cartilage at the central portion pattern across the rostrum is currently restricted to †Onchopristis and †Ischyrhiza (Figure 7).  [67]. These structures are not considered homologous to the rostral appendices of Rajiformes and Rhinopristiformes [7]. The rostral appendix in skates and guitarfishes is formed de novo on the proximal sides of the growing rostral plate [68]. Meanwhile, according to Miyake et al. [56], the paired "rostral cartilages" that are equivalent to Holmgren's [69] "rostral appendices", develop during early ontogeny and arise from the ventromedial area of the lamina orbitonasalis (at least in Torpedo).
The area of development of these structures in Torpedo is topologically similar to the areas where the rostral processes of platyrhinids are formed, and might indicate a homologous relationship between the "lateral rostral cartilages" of Baranes and Randal [70] and Claeson [60], which are present in Torpedo, Electrolux, Typhlonarke, Temera and Narke, and the "rostral processes" of Platyrhina and Platyrhinoidis of de Carvalho [67]. The presence of rostral processes is unknown (?) in some fossils recognized as sister taxa (i.e., †Britobatos) or belonging to Platyrhinidae (i.e., †Tingitanius; [23]), but rostral processes are present in the Eocene platyrhinid †Eoplatyrhina [33]. We kept holocephalians as unknown (?) as we still have doubts about the homology of the "process" recognized by de Beer and Moy-Thomas [55].
Ptr (see discussion Parsimony tree): The rostral processes are a synapomorphy of Torpediniformes.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
There is uncertainty in the plesiomorphic-state reconstruction caused by the inapplicable coding

7.
Rostral processes: (0) Absent; (1) present. Modified from Aschliman et al. [7] (char. 29); Claeson et al. [23] (char. 12). The presence of hyaline and poorly calcified structures called "rostral processes", are shared features of platyrhinids according to de Carvalho [67]. These structures are not considered homologous to the rostral appendices of Rajiformes and Rhinopristiformes [7]. The rostral appendix in skates and guitarfishes is formed de novo on the proximal sides of the growing rostral plate [68]. Meanwhile, according to Miyake et al. [56], the paired "rostral cartilages" that are equivalent to Holmgren's [69] "rostral appendices", develop during early ontogeny and arise from the ventromedial area of the lamina orbitonasalis (at least in Torpedo).
The area of development of these structures in Torpedo is topologically similar to the areas where the rostral processes of platyrhinids are formed, and might indicate a homologous relationship between the "lateral rostral cartilages" of Baranes and Randal [70] and Claeson [60], which are present in Torpedo, Electrolux, Typhlonarke, Temera and Narke, and the "rostral processes" of Platyrhina and Platyrhinoidis of de Carvalho [67]. The presence of rostral processes is unknown (?) in some fossils recognized as sister taxa (i.e., †Britobatos) or belonging to Platyrhinidae (i.e., †Tingitanius; [23]), but rostral processes are present in the Eocene platyrhinid †Eoplatyrhina [33]. We kept holocephalians as unknown (?) as we still have doubts about the homology of the "process" recognized by de Beer and Moy-Thomas [55].
Ptr (see discussion Parsimony tree): The rostral processes are a synapomorphy of Torpediniformes.
Based on Claeson [60] (Supporting Information char. 50) and Claeson [71] (char. 48). This character seeks to include the variation on the articulation between the rostral processes and the neurocranium following Marramà et al. [33].

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
There is uncertainty in the plesiomorphic-state reconstruction caused by the inapplicable coding in the taxa with no rostral processes. The presence of an articulation between the ventral aspect of the rostral cartilage and the rostral process is a basal feature for Torpediniformes, with subsequent gains of the rostral process's articulation with the nasal capsule in Hypnos, Torpedo and Temera ( Figure 8A). Narcine and †Titanonarke retain the plesiomorphic state (articulation with the ventral aspect of the rostral cartilages) ( Figure 8C,D), with a tripodal structure of Narke (0 and 1) ( Figure 8B). in the taxa with no rostral processes. The presence of an articulation between the ventral aspect of the rostral cartilage and the rostral process is a basal feature for Torpediniformes, with subsequent gains of the rostral process's articulation with the nasal capsule in Hypnos, Torpedo and Temera ( Figure 8A). Narcine and †Titanonarke retain the plesiomorphic state (articulation with the ventral aspect of the rostral cartilages) ( Figure 8C,D), with a tripodal structure of Narke (0 and 1) ( Figure 8B).  Rostral appendix: (0) Absent; (1) present. Modified from Aschliman et al. [7] (char. 28). The presence of rostral appendices is considered a shared feature between rhinopristisforms and rajiforms [7]. McEachran et al. [72] and Claeson et al. [23] also recognized their presence in fossil taxa such as †"Rhinobatos"maronita, †"R." latus, †Britobatos and †Rhombopterygia. The coding from previous works [23,59] for †Spathobatis was changed to "present" after new observations were made on several fossil specimens (e.g., BMNH P. 12067, BSPG 1952-I-82, BSPG -AS-I-505) (EV, pers. observ.). Although rostral appendices might be present in Diplobatis, Benthobatis, Narcine, Discopyge and †Titanonarke [28,69,73,74], their homology with rostral appendices of skates and guitarfishes is unclear, and we therefore coded the state as (0) in these taxa. Considering that the subtriangular rostral extremity reported and observed in Urolophus, Gymnura, Urotrygon, Plesiobatis and Potamotrygon is in connection with the anterior medial outgrowth of the trabecula [67] (one of the embryological cartilages

9.
Rostral appendix: (0) Absent; (1) present. Modified from Aschliman et al. [7] (char. 28). The presence of rostral appendices is considered a shared feature between rhinopristisforms and rajiforms [7]. McEachran et al. [72] and Claeson et al. [23] also recognized their presence in fossil taxa such as †"Rhinobatos"maronita, †"R." latus, †Britobatos and †Rhombopterygia. The coding from previous works [23,59] for †Spathobatis was changed to "present" after new observations were made on several fossil specimens (e.g., BMNH P. 12067, BSPG 1952-I-82, BSPG -AS-I-505) (EV, pers. observ.). Although rostral appendices might be present in Diplobatis, Benthobatis, Narcine, Discopyge and †Titanonarke [28,69,73,74], their homology with rostral appendices of skates and guitarfishes is unclear, and we therefore coded the state as (0) in these taxa. Considering that the subtriangular rostral extremity reported and observed in Urolophus, Gymnura, Urotrygon, Plesiobatis and Potamotrygon is in connection with the anterior medial outgrowth of the trabecula [67] (one of the embryological cartilages that forming the rostral cartilages along with the lamina orbitonasalis), it is very likely that these vestigial cartilages correspond to the rostral node and rostral shaft (sensu McEachran and Compagno [75]) and that the rostral appendices are involved, considering the presence of small posterior projections parallel to the rostral shaft. However, considering the lack of agreement about the presence of this structure in the literature [56,67,72] we coded this character as unknown for these taxa (?).
Ptr (see discussion Parsimony tree): The presence of rostral appendices ( Figure 9) is a synapomorphy for Rhinopristiformes with independent gains in clade 22 and Bathyraja and Raja.
that forming the rostral cartilages along with the lamina orbitonasalis), it is very likely that these vestigial cartilages correspond to the rostral node and rostral shaft (sensu McEachran and Compagno [75]) and that the rostral appendices are involved, considering the presence of small posterior projections parallel to the rostral shaft. However, considering the lack of agreement about the presence of this structure in the literature [56,67,72] we coded this character as unknown for these taxa (?).

Ptr (see discussion Parsimony tree):
The presence of rostral appendices ( Figure 9) is a synapomorphy for Rhinopristiformes with independent gains in clade 22 and Bathyraja and Raja.
Zapteryx and Aptychotrema are located near the base of the rostral cartilage, but more anteriorly directed in Myliobatis, Aetobatus, Rhinoptera and Mobula ( Figure 11C-E). Aschliman et al. [7] coding was kept, except for Chimaera, Harriotta, Temera and Torpedo, as we could not observe the foramen (?).
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): An anterior preorbital foramen located anteriorly is a synapomorphy of Rhinopristiformes with an independent gain in the clade [Hemiscyllium + Ginglymostoma], in clade 16 and †Spathobatis.
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): An anterior preorbital foramen located anteriorly is a synapomorphy of Rhinopristiformes with an independent gain in the clade [Hemiscyllium + Ginglymostoma], in clade 16 and †Spathobatis.
MLtr (see discussion Maximum Likelihood tree): An anterior preorbital foramen located anteriorly is recovered as a synapomorphy of the clade [Hemiscyllium + Ginglymostoma] with independent gains in †Spathobatis and being the basal state in clades 10, 10 and 14.
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): Laterally expanded nasal capsules ( Figure 12A) are the plesiomorphic state for chondrichthyans, with the subsequent gain of trumpet-shaped nasal capsules ( Figure 12D) as a synapomorphy of clade 2 ). The anterolateral expansion of the nasal capsules ( Figure 12C) is a synapomorphy of [ †Pseudorhina + Squatina], and the ventrolateral expansion of the nasal capsules ( Figure 12B) is a synapomorphy for clade 18 . Ptr (see discussion Parsimony tree): There is an independent gain of the ventrolateral expansion of the nasal capsules representing the plesiomorphic feature of clade 14, except for Hexatrygon and Plesiobatis, which present the basal state.
MLtr (see discussion Maximum Likelihood tree): It presents a similar-state reconstruction for this feature. However, the additional gain of the ventrolateral expansion of the nasal capsules is a synapomorphy for clade 29. Ptr (see discussion Parsimony tree): The presence of a horn-like process ( Figure 13) is a synapomorphy of clades 11 and 24. There are independent gains of the horn process in †Tlalocbatus, †Stahlraja and †Britobatos.
MLtr (see discussion Maximum Likelihood tree): Presents a similar reconstruction for this character as in the parsimony tree, but in this case, the paraphyletic state of the thornback clade makes the recovery of this feature as a synapomorphy for the group impossible. Ptr (see discussion Parsimony tree): There is an independent gain of the ventrolateral expansion of the nasal capsules representing the plesiomorphic feature of clade 14, except for Hexatrygon and Plesiobatis, which present the basal state.
MLtr (see discussion Maximum Likelihood tree): It presents a similar-state reconstruction for this feature. However, the additional gain of the ventrolateral expansion of the nasal capsules is a synapomorphy for clade 29. Ptr (see discussion Parsimony tree): The presence of a horn-like process ( Figure 13) is a synapomorphy of clades 11 and 24. There are independent gains of the horn process in †Tlalocbatus, †Stahlraja and †Britobatos.  MLtr (see discussion Maximum Likelihood tree): Presents a similar reconstruction for this character as in the parsimony tree, but in this case, the paraphyletic state of the thornback clade makes the recovery of this feature as a synapomorphy for the group impossible. 23. Antorbital cartilages: (0) Absent; (1) present. Modified from Aschliman et al. [7] (char. 8). Shirai [35] suggests the presence of antorbital cartilages in Pristiophorus. However, no evidence of these cartilages was observed in the Pristiophorus specimens CSIRO 3731 and CAS 4942. The previously illustrated lack of antorbital cartilages in †Cyclobatis in Cappetta Text- Figure 355A in [85] seems to be caused by the position of these cartilages, which appear to be overlapped by the propterygium, like in stingrays ( Figure 15).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees)
The presence of antorbital cartilages is recovered as a synapomorphy of batomorphs. 23. Antorbital cartilages: (0) Absent; (1) present. Modified from Aschliman et al. [7] (char. 8). Shirai [35] suggests the presence of antorbital cartilages in Pristiophorus. However, no evidence of these cartilages was observed in the Pristiophorus specimens CSIRO 3731 and CAS 4942. The previously illustrated lack of antorbital cartilages in †Cyclobatis in Cappetta Text- Figure 355A in [85] seems to be caused by the position of these cartilages, which appear to be overlapped by the propterygium, like in stingrays ( Figure 15).

MLtr (see discussion Maximum Likelihood tree):
The presence of irregular-shaped antorbital cartilages is a synapomorphy of clade 26.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of small antorbital cartilages is a synapomorphy for the Myliobatiformes, with an independent gain in †Cyclobatis. 26. Anterior process of antorbital cartilage (if regular outline): (0) Absent; (1) present.
This character includes the variation observed in the anterior portion of the antorbital cartilage of batomorphs.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of an anterior process in the antorbital cartilages is a synapomorphy of clade 16 with independent gains in Platyrhina, †"Rhinobatos"whitfieldi and Zanobatus.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of small antorbital cartilages is a synapomorphy for the Myliobatiformes, with an independent gain in †Cyclobatis.

Anterior process of antorbital cartilage (if regular outline):
This character includes the variation observed in the anterior portion of the antorbital cartilage of batomorphs.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of an anterior process in the antorbital cartilages is a synapomorphy of clade 16 with independent gains in Platyrhina, †"Rhinobatos"whitfieldi and Zanobatus.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
A reduced postorbital process is recovered ( Figure 17I) as a synapomorphy of clade 16, with independent gains in †Plesiozanobatus, †Ostarriraja, Chimaera, Harriotta and †Ozarcus.  Figure 18B) is independently a gain as well as a synapomorphy for the Hybodontiformes and clade 14, and with additional independent gains as the basal feature in clade 3 and in †Britobatos as an autapomorphy. modified from Nishida Text- Figure 7G in [82]. State (1): (I) Torpedo ocellata (AMNH 4128). Black arrowheads: postorbital process; red arrowheads: triangular process.

Ptr and MLtr: (see discussion Maximum Likelihood and Parsimony trees):
A broad shelf-like postorbital process ( Figure 18B) is independently a gain as well as a synapomorphy for the Hybodontiformes and clade 14, and with additional independent gains as the basal feature in clade 3 and in †Britobatos as an autapomorphy (. MLtr (see discussion Maximum Likelihood tree): A broad postorbital process is a synapomorphy for members of clade 3. Consequently, the narrow state in Pristiophorus is interpreted as an independent gain and thus an autapomorphy. Klug [20]. The suborbital shelf is a horizontal plate on the ventral junction of the orbital wall and basal plate that is the floor of the orbit. It runs from the nasal capsule to the otic capsule and is penetrated posteriorly by the stapedial foramen and sometimes laterally by a notch, foramen, or fenestra for the palatine branch of the facial nerve.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of a suborbital process ( Figure 19A,B) is a synapomorphy for Galeomorphii, with independent gains in [ †Hybodus + †Hamiltonichthys] and †Doliodus. MLtr (see discussion Maximum Likelihood tree): A broad postorbital process is a synapomorphy for members of clade 3. Consequently, the narrow state in Pristiophorus is interpreted as an independent gain and thus an autapomorphy. 40. Suborbital shelf: (0) Absent; (1) present. Based on observations by Shirai [35,37] and Klug [20]. The suborbital shelf is a horizontal plate on the ventral junction of the orbital wall and basal plate that is the floor of the orbit. It runs from the nasal capsule to the otic capsule and is penetrated posteriorly by the stapedial foramen and sometimes laterally by a notch, foramen, or fenestra for the palatine branch of the facial nerve.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of a suborbital process ( Figure 19A Figure 7B in [87]. Arrowhead: suborbital shelf. 41. Basitrabecular process: (0) Absent, (1) present. Based on de Carvalho [38] (char. 11), de Carvalho and Maisey [15] (char. 21) and Klug [20] (char. 10). This character is interpreted as a separate feature from the suborbital shelf based on its topographic relationships and development [90]. The basitrabecular process derives from a lateral expansion of the polar cartilage just anterior to the auditory capsules and articulates anteriorly with the orbital process of the palatoquadrate [15,90,91].
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): An ethmoidal articulation ( Figure 21A) is an independent gain and a synapomorphy of the clades 15 and [ †Hybodus + †Hamiltonichthys]).
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): An ethmoidal articulation ( Figure 21A) is an independent gain and a synapomorphy of the clades 15 and [ †Hybodus + †Hamiltonichthys]). based on observations by Maisey [92]. An additional character state is proposed to include the variation observed in †Doliodus, symmoriids (2) and hexanchids (1).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees)::
The presence of a postorbital articulation located further posteroventrolaterally of the chondrified lateral commissure is recovered as the basal state for chondrichthyans, being present in †Doliodus, †Cobelodus and †Ozarcus, with the subsequent gain of a postorbital articulation where the articulation surface is in the proximal part of the process in Hexanchus (Figure 15). based on observations by Maisey [92]. An additional character state is proposed to include the variation observed in †Doliodus, symmoriids (2) and hexanchids (1).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of a postorbital articulation located further posteroventrolaterally of the chondrified lateral commissure is recovered as the basal state for chondrichthyans, being present in †Doliodus, †Cobelodus and †Ozarcus, with the subsequent gain of a postorbital articulation where the articulation surface is in the proximal part of the process in Hexanchus (Figure 15). MLtr: The absence of a postorbital articulation is a synapomorphy of holocephalians and Euselachii, with a subsequent gain of the articulation in Hexanchus (Figure 22).   The flange on the palatoquadrate is a characteristic feature in hybodontiforms ( Figure 24A). This ledge is located laterally to the mandibular cartilage and does not interact with the Meckel's cartilage. This process corresponds to the "quadrate process" of de Carvalho et al. [97] and Maisey et al. [83] in squatinids and pristiophorids ( Figure 24B,C).

MLtr:
The absence of a postorbital articulation is a synapomorphy of holocephalians and Euselachii, with a subsequent gain of the articulation in Hexanchus (Figure 22).  The flange on the palatoquadrate is a characteristic feature in hybodontiforms ( Figure 24A). This ledge is located laterally to the mandibular cartilage and does not interact with the Meckel's cartilage. This process corresponds to the "quadrate process" of de Carvalho et al. [97] and Maisey et al. [83] in squatinids and pristiophorids ( Figure 24B,C).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of this ledge-like process in the palatoquadrate is an independent gain and shared feature for Hybodontiformes and clade 5 (Figures 1 and 2).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
An unsegmented basibranchial ( Figure 26B) is recovered additionally as a synapomorphy for batomorphs with independent gains in Pristiophorus and Hemiscyllium.  (2) spiracularis subdivided proximally and inserts separately into the palatoquadrate and the hyomandibula. This character is proposed to include the character states recognized by Aschliman et al. [7] (char. 85) except for the third state, which seems to be a variation of the first state (splits into lateral and medial bundles).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
A divided spiracularis, in which one muscle bundle enters the dorsal oral membrane underlying the neurocranium, is a shared feature of Torpedo, Hypnos, Narcine, Narke and Temera. The spiracularis splits into lateral and medial bundles, with the medial bundle inserting onto the posterior surface of the Meckel's cartilage and the lateral bundle onto the dorsal edge of the hyomandibula, which is a shared feature in Urolophus, Urobatis, Urotrygon, Plesiobatis, Hypanus, Potamotrygon and Neotrygon. There is a subsequent gain of the spiracularis subdivided proximally, inserting separately onto the palatoquadrate and the hyomandibula, Rhinoptera.  The character is separated into two different characters, 70 and 71, aiming to increase the grouping information.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The plesiomorphic state for chondrichthyans is the presence of a coracohyoideus. The lack of this muscle is a synapomorphy of clade 18 (Figures 1 and 2). 71. Coracohyoideus (if present): (0) Parallel to body axis; (1) runs parallel to the body axis and is very short; (2) runs diagonally from the wall of the first two gill slits to the posteromedial aspect of the basihyal or first basibranchial; (3) each muscle fuses with its antimere at a raphe near its insertion on the first hypobranchial.

Ptr (see discussion Parsimony tree):
A coracohyoideus parallel to the body axis is recovered as the plesiomorphic condition for chondrichthyans, being present across several groups and taxa (Chimaera, Harriotta, Chlamydoselachus, Hexanchus, Heterodontus, Squatina, Pristiphoridae, Squalus, Ginglymostoma, Raja, Bathyraja, Rhynchobatus, Glaucostegus, Rhina, Rhinobatos, Pseudobatos, Trygonorrhina and Zapteryx). A very short coracohyoideus that runs parallel to the body axis is an autapomorphy of Pristis. A coracohyoideus running diagonally from the wall of the first two gill slits to the posteromedial aspect of the basihyal or first basibranchial is a synapomorphy of the clade 27, with the subsequent gain of the coracohyoideus fusing with its antimere at a raphe near its insertion on the first hypobranchial as a shared state for clade 16. MLtr (see discussion Maximum Likelihood tree): Presents a similar reconstruction for this character as the parsimony tree. The diagonal arrangement of the coracohyoideus from the wall of the first two gill slits to the posteromedial aspect of the basihyal is not a synapomorphy due to the polytomic state of the thornbacks within clade 27.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
There is uncertainty in reconstructing the basal state, as †Doliodus, symmoriids and hybodontiforms lack calcified vertebral centra. The presence of a "synarcual" formed by the fusion of the neural/basidorsal and hemal/basiventral elements is shared by the Holocephali. A synarcual characterized by the fusion of the cervicothoracic vertebral centra is a shared feature of batomorphs. All selachimorphs have unfused vertebral centra. 49. Expanded basiventral process of cervical vertebrae: (0) Absent; (1) present. Taken from Maisey et al. [63] (chars. [16][17][18].

Ptr (see discussion Parsimony tree):
The presence of expanded basiventral processes in their cervical vertebrae, in which the first process is larger than the subsequent ones, which become smaller continuously in size posteriorly (Figure 27), is a synapomorphy of clade 5.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of lateral stays unfused with the medial crest of the synarcual is the plesiomorphic feature of batomorphs ( Figure 28C,D), with the subsequent gain of the fused state ( Figure  28A,B) as a synapomorphy of clade 17. (1) free of medial crest (new). Taxa with no synarcual (i.e., outgroups) or with no lateral stays on the cervicothoracic synarcual (i.e., Chimaera and Harriotta) are coded as inapplicable (-), which makes the reconstruction of this character for basal chondrichthyans in the trees impossible.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of lateral stays unfused with the medial crest of the synarcual is the plesiomorphic feature of batomorphs ( Figure 28C,D), with the subsequent gain of the fused state ( Figure 28A,B) as a synapomorphy of clade 17. (2) unfused medially. Modified from Aschliman et al. [7] (char. 6). In some sharks, there seems to be an anterior portion of the scapular process that is detached from the scapula, referred to as suprascapular by Marramà et al. [59]. While this element is dorsal to the scapula, its interaction with other skeletal elements and its development seems to be different from that of the suprascapula of batomorphs.

Ptr (see discussion Parsimony tree):
The lack of a suprascapula is the plesiomorphic state for chondrichthyans. The presence of a suprascapula ( Figure 29A) (see also [8,9]) is a synapomorphy of clade 2), with an independent gain in Squatina. A medially developed suprascapula, as a single-element dorsal to the vertebral column connecting the scapulocoracoid antimeres ( Figure 29B-F), is a synapomorphy for the batomorph crown group (clade 6. MLtr (see discussion Maximum Likelihood tree): The basal placement of Rajiformes (rajoids and sclerorhynchoids) causes uncertainty for reconstructing the plesiomorphic state for the batomorph crown group, as the suprascapula is missing in most sclerorhynchoids. Consequently, this character is not recovered as a synapomorphy for the crown group, presenting independent gains in rajoids, †Libanopristis and clade 23. Aschliman et al. [7] (char. 6). In some sharks, there seems to be an anterior portion of the scapular process that is detached from the scapula, referred to as suprascapular by Marramà et al. [59]. While this element is dorsal to the scapula, its interaction with other skeletal elements and its development seems to be different from that of the suprascapula of batomorphs.

Ptr (see discussion Parsimony tree):
The lack of a suprascapula is the plesiomorphic state for chondrichthyans. The presence of a suprascapula ( Figure 29A) (see also [8,9]) is a synapomorphy of clade 2, with an independent gain in Squatina. A medially developed suprascapula, as a single-element dorsal to the vertebral column connecting the scapulocoracoid antimeres ( Figure 29B-F

Suprascapula interaction with axial skeleton (if fused medially):
(0) Interacts with axial skeleton (articulated or fused); (1) free from axial skeleton (new). This character is proposed to include the variation observed in the suprascapula articulation in batomorphs.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of an interaction (i.e., fusion or articulation) between the suprascapula and the axial skeleton ( Figure 30A,B) is the basal state for the batomorph crown group (Jurassic batomorphs such as †Kimmerobatis, †Asterodermus, †Spathobatis and †Belemnobatis lack the suprascapula-or at least, a calcified one). The absence of interaction between the suprascapula and the axial skeleton ( Figure 30C,D) is a synapomorphy of clade 18

MLtr (see discussion Maximum Likelihood tree):
The basal placement of Rajiformes (rajoids and sclerorhynchoids) causes uncertainty for reconstructing the plesiomorphic state for the batomorph crown group, as the suprascapula is missing in most sclerorhynchoids. Consequently, this character is not recovered as a synapomorphy for the crown group, presenting independent gains in rajoids, †Libanopristis and clade 23.

Suprascapula interaction with axial skeleton (if fused medially):
(0) Interacts with axial skeleton (articulated or fused); (1) free from axial skeleton (new). This character is proposed to include the variation observed in the suprascapula articulation in batomorphs.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of an interaction (i.e., fusion or articulation) between the suprascapula and the axial skeleton ( Figure 30A,B) is the basal state for the batomorph crown group (Jurassic batomorphs such as †Kimmerobatis, †Asterodermus, †Spathobatis and †Belemnobatis lack the suprascapula-or at least, a calcified one). The absence of interaction between the suprascapula and the axial skeleton ( Figure 30C

Suprascapula (if interacts with axial skeleton): (0) Articulates with vertebral column;
(1) fused medially to synarcual; (2) fused medially and laterally to synarcual (new). This character is proposed to account for the variation observed in the interaction between the suprascapula and axial skeleton in batomorphs.

Ptr (see discussion Parsimony tree):
The presence of a suprascapula that is fused medially to the synarcual ( Figure 31C,D) is a synapomorphy of Rajiformes (inapplicable in Asflapristis, Ptychotrygon and Sclerorhynchus). The medial and lateral fusions of the suprascapula and synarcual is a synapomorphy of Myliobatiformes (including Zanobatus).
MLtr: The presence of a suprascapula that is fused medially to the synarcual is not recovered as a synapomorphy of the Rajiformes as there is uncertainty regarding the basal state, being inapplicable for the Jurassic batoids and Asflapristis, Ptychotrygon and Sclerorhynchus, as this cartilage is missing. The medial and lateral fusion of the suprascapula and synarcual is a synapomorphy of Myliobatiformes. (1) fused medially to synarcual; (2) fused medially and laterally to synarcual (new). This character is proposed to account for the variation observed in the interaction between the suprascapula and axial skeleton in batomorphs.

Ptr (see discussion Parsimony tree):
The presence of a suprascapula that is fused medially to the synarcual ( Figure 31C,D) is a synapomorphy of Rajiformes (inapplicable in Asflapristis, Ptychotrygon and Sclerorhynchus). The medial and lateral fusions of the suprascapula and synarcual is a synapomorphy of Myliobatiformes (including Zanobatus).
MLtr (see discussion Maximum Likelihood tree): There is uncertainty regarding the basal state for the batomorph crown group, as the placement of Rajiformes at the base of this group creates conflict between curved and crenate states. For the reaming batomorphs, both topologies recovered similar character reconstructions, with the crenated articulation as the basal state for the batomorph crown group, whereas the straight and ball-andsocket articulations are synapomorphies for electric rays and Myliobatiformes, respectively. Ptr (see discussion Parsimony tree): The presence of lateral projections on the lower lobe of the suprascapula ( Figure 34A,B) is a shared feature of Platyrhina and Platyrhinoidis. We could not determine the state in fossil thornbacks ( †Tethybatis, †Tingitanius and †Eoplatyrhina), resulting in an uncertainty in the state reconstructions for the corresponding clade.

MLtr (see discussion Maximum Likelihood tree):
There is uncertainty regarding the basal state for the batomorph crown group, as the placement of Rajiformes at the base of this group creates conflict between curved and crenate states. For the reaming batomorphs, both topologies recovered similar character reconstructions, with the crenated articulation as the basal state for the batomorph crown group, whereas the straight and ball-and-socket articulations are synapomorphies for electric rays and Myliobatiformes, respectively.  98. Scapular process-scapula: (0) Fused; (1) articulated (new). The interaction between the scapula and scapular process is a rather variable within sharks.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The fusion between the scapular process and the scapula is the basal condition for chondrichthyans. Members of Squalus display variation with the species Squalus acanthias and S. megalops having a fused process ( Figure 34A,B), while S. mitsukurii and S. brevirostris present an articulation between the process and the scapula ( Figure 34D,E). There is also variation on the state in Hexanchiformes, with Hexanchus griseus presenting the fused state ( Figure 34C) and Chlamydoselachus anguineus presenting the articulated state ( Figure 34F). Within sharks, state (1) is a shared feature between Heterodontus and Chlamydoselachus.   Figure 26E in [8]. Arrowheads: articular surface between the scapula and the scapular process.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of a short and dorsally directed scapular process is the basal state for chondrichthyans ( Figure 35A). The presence of a long, U-shaped and posterodorsally directed scapular process ( Figure 35B) is a synapomorphy of clade 18. The presence of a short, posterodorsally directed scapular process ( Figure 35C) is an autapomorphy of Pseudobatos.  Figure 26E in [8]. Arrowheads: articular surface between the scapula and the scapular process. 98. Scapular process-scapula: (0) Fused; (1) articulated (new). The interaction between the scapula and scapular process is a rather variable within sharks.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The fusion between the scapular process and the scapula is the basal condition for chondrichthyans. Members of Squalus display variation with the species Squalus acanthias and S. megalops having a fused process ( Figure 34A,B), while S. mitsukurii and S. brevirostris present an articulation between the process and the scapula ( Figure 34D,E). There is also variation on the state in Hexanchiformes, with Hexanchus griseus presenting the fused state ( Figure 34C) and Chlamydoselachus anguineus presenting the articulated state ( Figure 34F). Within sharks, state (1) is a shared feature between Heterodontus and Chlamydoselachus.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of a short and dorsally directed scapular process is the basal state for chondrichthyans ( Figure 35A). The presence of a long, U-shaped and posterodorsally directed scapular process ( Figure 35B) is a synapomorphy of clade 18. The presence of a short, posterodorsally directed scapular process ( Figure 35C) is an autapomorphy of Pseudobatos. Modified from de Carvalho [38] (char. 38), based on observations by da Silva and de Carvalho [8].

Ptr (see discussion Parsimony tree):
There is uncertainty regarding the basal-state reconstruction for chondrichthyans as the character is unknown in †Doliodus, †Ozarcus, †Cobelodus, Chimaera and Harriotta. The presence of an articulation between the scapulocoracoid and pectoral elements composed by facets ( Figure 36A-D) is a synapomorphy of Scyliorhinus and Mustelus, with independent gains in †Hybodus and †Tribodus. The combination of facets and condyles in the articulation between the pterygia and the scapulocoracoid ( Figure 36F,G) is a synapomorphy of clade 4 (with an independent gain of the facet in Hexanchus. Within Elasmobranchii, the basal state is the presence of condyles as the means of articulation of the pectoral elements ( Figure 36I,J), present in selachimorphs (Heterodontus, Squatina, Pristiophorus, Ginglymostoma, Hemiscyllium) and batomorphs.

MLtr (see discussion Maximum Likelihood tree):
The presence of facets for the articulation of the proximal pectoral elements is the basal condition for euselachians. The combination of facets and condyles for the articulation of the pterygia is a synapomorphy of clade 4, with the subsequent independent gains of the full condyle articulation as a synapomorphy in clades 2 and 5, and in batomorphs. Modified from de Carvalho [38] (char. 38), based on observations by da Silva and de Carvalho [8].
Ptr (see discussion Parsimony tree): There is uncertainty regarding the basal-state reconstruction for chondrichthyans as the character is unknown in †Doliodus, †Ozarcus, †Cobelodus, Chimaera and Harriotta. The presence of an articulation between the scapulocoracoid and pectoral elements composed by facets ( Figure 36A-D) is a synapomorphy of Scyliorhinus and Mustelus, with independent gains in †Hybodus and †Tribodus. The combination of facets and condyles in the articulation between the pterygia and the scapulocoracoid ( Figure 36F Ptr (see discussion Parsimony tree): The presence of a single condyle ( Figure 37A) is a shared feature between Heterodontus and Hemiscyllium. The presence of two condyles, for the articulation of the pro + mesopterygium and the metapterygium, respectively, is a shared feature of Squatina and Ginglymostoma ( Figure 37B). The presence of a single condyle for the articulation of the meso + metapterygium and a facet for the propterygium ( Figure 37C) is a synapomorphy for clade 4 (this is not recovered in the ML tree). The presence of three separated condyles ( Figure 37D-F) is the basal feature for [Selachimorpha + Batomorpha] being present in Pristiophorus and all batomorphs and is additionally a synapomorphy as an independent gain of clade 4.
MLtr (see discussion Maximum Likelihood tree): The presence of three separated condyles is recovered as the basal state for batomorphs with an independent gain in Pristiophorus.

MLtr (see discussion Maximum Likelihood tree):
The presence of facets for the articulation of the proximal pectoral elements is the basal condition for euselachians. The combination of facets and condyles for the articulation of the pterygia is a synapomorphy of clade 4, with the subsequent independent gains of the full condyle articulation as a synapomorphy in clades 2 and 5, and in batomorphs. 102. Condyles: (0) Single condyle; (1) pro + mesocondyle; (2) meso + metacondyle; (3) three separated condyles. Modified from de Carvalho [38] (char. 38), based on observations by da Silva and de Carvalho [8].
Ptr (see discussion Parsimony tree): The presence of a single condyle ( Figure 37A) is a shared feature between Heterodontus and Hemiscyllium. The presence of two condyles, for the articulation of the pro + mesopterygium and the metapterygium, respectively, is a shared feature of Squatina and Ginglymostoma ( Figure 37B). The presence of a single condyle for the articulation of the meso + metapterygium and a facet for the propterygium ( Figure 37C) is a synapomorphy for clade 4 (this is not recovered in the ML tree). The presence of three separated condyles ( Figure 37D-F) is the basal feature for [Selachimorpha + Batomorpha] being present in Pristiophorus and all batomorphs and is additionally a synapomorphy as an independent gain of clade 4.  (1) present. The modification of this character from the multistate coding used in McEachran and Dunn [107] (char. 36) is because the three proposed states (i.e., short to moderately long; extremely long with acute tips; and extremely long with biramous tips) are difficult to interpret in fossil specimens. Consequently, binary coding (presence/absence) is used here.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The absence of lateral prepelvic processes is the plesiomorphic condition for chondrichthyans.  [107] (char. 36) is because the three proposed states (i.e., short to moderately long; extremely long with acute tips; and extremely long with biramous tips) are difficult to interpret in fossil specimens. Consequently, binary coding (presence/absence) is used here.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The absence of lateral prepelvic processes is the plesiomorphic condition for chondrichthyans. The presence of these lateral processes, conversely, can be considered as an independent gain and a synapomorphy of clades 7 and 18.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The lack of postpelvic processes is the plesiomorphic state for chondrichthyans.
Ptr (see discussion Parsimony tree): The presence of these processes ( Figure 39B-D) is a synapomorphy of Torpediniformes and Jurassic batomorphs, with additional independent gains in Hemiscyllium and Rhinopristiformes, being present in †Tlalocbatus, Pseudobatos, Rhinobatos, Glaucostegus, Zapteryx and Aptychotrema. The presence of these lateral processes, conversely, can be considered as an independent gain and a synapomorphy of clades 7 and 18.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The lack of postpelvic processes is the plesiomorphic state for chondrichthyans.
MLtr (see discussion Maximum Likelihood tree): The presence of these processes is a synapomorphy of batomorphs, with a subsequent loss in the non-Jurassic forms. Additional independent gains of the postpelvic processes are recovered as synapomorphies of clades 22 and 24, with the subsequent loss of these processes as a synapomorphy in [Temera + Narke] and in Myliobatiformes.     (Figure 2), Stumpf et al. [108] and Coates et al. [109]. Current fossil evidence suggests that the separation of two halves, or at least a not very well-mineralized mid-bar of the pelvic girdle, is the basal state across hybodontiform-like sharks ( Figure 42A,B) (SMNS 10062) (NHMUK PV P 339).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of two pelvic halves is a shared feature of holocephalians and hybodonts. A fused pelvic girdle, conversely, is a shared feature of elasmobranchs, and due to the uncertainty at the base of the tree recovered, it does not represent a synapomorphy for Elasmobranchii in the current trees. Ptr (see discussion Parsimony tree): The crustal calcification is the basal feature for chondrichthyans. The presence of catenated calcification is a synapomorphy of rajoids, with an independent gain in some stingrays, being present in †Asterotrygon, †Heliobatis,

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The presence of two pelvic halves is a shared feature of holocephalians and hybodonts. A fused pelvic girdle, conversely, is a shared feature of elasmobranchs, and due to the uncertainty at the base of the tree recovered, it does not represent a synapomorphy for Elasmobranchii in the current trees.

Ptr (see discussion Parsimony tree):
The crustal calcification is the basal feature for chondrichthyans. The presence of catenated calcification is a synapomorphy of rajoids, with an independent gain in some stingrays, being present in †Asterotrygon, †Heliobatis, Urolophus, Urobatis, Hypanus, Neotrygon, †Lessiniabatis, †Arechia and †Tethytrygon. The genera Potamotrygon and Urotrygon present variations of the type of radial calcification depending on the portions of the pectoral fins with basal radials: those closer to the propterygium, mesopterygium and metapterygium present crustal calcification, while the subsequent series display catenated calcification.
MLtr (see discussion Maximum Likelihood tree): Presents a similar reconstruction to the parsimony analysis. The catenated calcification is recovered as an independent gain and a synapomorphy of clades 7 and 30.  [15] (char. 65). The original character was proposed as a synapomorphy for platyrhinids or as a shared feature between platyrhinids and Zanobatus, which according to de Carvalho [66] also present the following condition: extension of the propterygium and its associated radials to the anterior margin of the disc on both sides of the snout and rostrum. Aschliman et al. [7] suggested that the extension of the propterygium observed in platyrhinids and Zanobatus is similar to the condition present in Myliobatiformes and Bathyraja. However, in pelagic stingrays (e.g., Myliobatis Aetobatus, Rhinoptera and Mobula), the head stands out of the pectoral disc, causing modifications to the neurocranium and pectoral disc, suggesting differences in this structure. In contrast, the condition of Rajiformes resembles that of the remaining batomorphs.
Furthermore, homology issues arise in groups in which the propterygium does not reach the anterior margin of the disc (e.g., Torpediniformes and Rhinopristiformes). These groups present different conditions affecting the extension of their propterygium. There are considerable modifications in the cephalic and anterior portions of the body in electric rays due to the branchial electric organs. In Rhinopristiformes, there is a significant development of the rostral cartilages. Whatever is the case, there seem to be different conditions in batomorphs affecting the extension propterygium. Consequently, we only coded the presence or absence of the anterior extension of the propterygium.
Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees): The lack of an anteriorly elongated propterygium is the basal state for chondrichthyans, but the presence of an anterior elongated propterygium is a synapomorphy for the batomorphs ( Figure 43B,C).
in batomorphs affecting the extension propterygium. Consequently, we only coded the presence or absence of the anterior extension of the propterygium.
MLtr (see discussion Maximum Likelihood tree): Presents a similar reconstruction to that found in the parsimony analysis. However, this topology recovers the first segment of the propterygium reaching the level of nasal capsules as a synapomorphy of clade 14 (, with a subsequent gain of the state "first segment of the propterygium extending beyond the level of nasal capsules" as synapomorphy of clade 33. MLtr (see discussion Maximum Likelihood tree): Presents a similar reconstruction to that found in the parsimony analysis. However, this topology recovers the first segment of the propterygium reaching the level of nasal capsules as a synapomorphy of clade 14, with a subsequent gain of the state "first segment of the propterygium extending beyond the level of nasal capsules" as synapomorphy of clade 33.

MLtr (see discussion Maximum Likelihood tree):
The presence of inter-radial connections (cross-braces) between radials of the pectoral fin is a synapomorphy of clade 15, with independent gains in †Britobatos, Urotrygon, Gymnura.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The lack of an overdeveloped first pelvic radial ( Figure 47A,B) is the basal feature for chondrichthyans. The presence of overdevelopment in the first pelvic radial, which in the extant taxa results in a pelvic fin with two lobes ( Figure 47C,D), is a synapomorphy of clade 7.

MLtr (see discussion Maximum Likelihood tree):
The presence of inter-radial connections (cross-braces) between radials of the pectoral fin is a synapomorphy of clade 15, with independent gains in †Britobatos, Urotrygon, Gymnura.

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
The lack of an overdeveloped first pelvic radial ( Figure 47A,B) is the basal feature for chondrichthyans. The presence of overdevelopment in the first pelvic radial, which in the extant taxa results in a pelvic fin with two lobes ( Figure 47C,D), is a synapomorphy of clade 7.  60. Pelvic basipterygium: (0) Fused to first radial; (1) separated from first radial. Based on Riley et al. [110].

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
There is uncertainty regarding the basal state for chondrichthyans, with Chimaera and Harriotta presenting a single element supporting their pelvic fin rays ( Figure 48A), and the state for †Doliodus, †Ozarcus and †Cobelodus remaining unknown. The separation between the first radial and the basipterygium is the basal feature for euselachians ( Figure 48B,C).

Ptr and MLtr (see discussion Maximum Likelihood and Parsimony trees):
There is uncertainty regarding the basal state for chondrichthyans, with Chimaera and Harriotta presenting a single element supporting their pelvic fin rays ( Figure 48A), and the state for †Doliodus, †Ozarcus and †Cobelodus remaining unknown. The separation between the first radial and the basipterygium is the basal feature for euselachians ( Figure 48B,C).  Figure 13A in [57]. State (1) (C) Zapteryx exasperta (CNPE-IBUNAM 20528). Basipterygium and first pelvic radial in gray.

Phylogenetic Analyses
Parsimony analysis recovered a total of 250 most parsimonious trees (MPTs) with a tree length of 386 steps, a CI of 0.443 and a RI of 0.841, from which a strict consensus was estimated . Maximum likelihood (ML) recovered two trees with a likelihood score of 1733.270, for which a strict consensus also was produced. Bayesian analysis ran over 190,000,000 generations, reaching an average standard deviation of split frequencies of 0.017108, a potential scale reduction factor between 0.9 and 1.2, and an average estimated sample size < 100, suggesting that convergence between the chains occurred [111] (see logs in electronic Supplemental Material). These three methodological approaches recover some variation in their topological arrangements that illustrate some of the phylogenetic uncertainties surrounding the groups included in the present work.  Figure 13A in [57]. State (1) (C) Zapteryx exasperta (CNPE-IBUNAM 20528). Basipterygium and first pelvic radial in gray.

Phylogenetic Analyses
Parsimony analysis recovered a total of 250 most parsimonious trees (MPTs) with a tree length of 386 steps, a CI of 0.443 and a RI of 0.841, from which a strict consensus was estimated. Maximum likelihood (ML) recovered two trees with a likelihood score of 1733.270, for which a strict consensus also was produced. Bayesian analysis ran over 190,000,000 generations, reaching an average standard deviation of split frequencies of 0.017108, a potential scale reduction factor between 0.9 and 1.2, and an average estimated sample size <100, suggesting that convergence between the chains occurred [111] (see logs in electronic Supplemental Material). These three methodological approaches recover some variation in their topological arrangements that illustrate some of the phylogenetic uncertainties surrounding the groups included in the present work. Parsimony analysis did not recover a monophyletic group consisting of [Holocephali + symmoriids (represented by †Cobelodus and †Ozarcus)]. The symmoriids ( †Cobelodus and †Ozarcus) and holocephalans (Chimaera and Harriotta) were placed in a polytomy along with the Euselachian clade (sensu [1,12,112]) ( Figure 49). This arrangement might suggest alternative affinities for †Cobelodus and †Ozarcus outside the Holocephali, in contrast to previous phylogenetic analyses (e.g., [25,26,113]). However, a more extensive taxon and character sampling is mandatory to recover a more reliable systematic position for these groups, which could not be resolved using the present data matrix. Torpedo, Hypnos, Narcine, Narke and Temera [6,7]) cannot be detected in torpediniforms. Both character reconstructions suggest that reductive coding might not be the best approach for these features.  A monophyletic clade representing the Euselachii (sensu [112,113]) was recovered, and within this group, a sister group relation between Hybodontiformes and Elasmobranchii (sensu [1,112]) was found. This relationship has been previously suggested by Maisey et al. [12] and Frey et al. [114]. However, before referring to any possible elasmobranch apomorphies, a more detailed sampling of taxa with the inclusion of synechodontids or xenacanthids and other extinct groups would be desirable.
The Torpediniformes are sisters to the Myliobatiformes. However, this grouping seems to be a by-product of the use of reductive coding [118], as the characters that support this assemblage present odd reconstructions [119]. †Titanonarke, Torpedo, Narcine and Temera lack a postorbital process [7,28], meaning that the possibility to recognize a "postorbital process separated from the triangular process" for this group is impossible. Similarly, the configuration of the coracohyoideus muscle plate (char. 71: which is absent in Torpedo, Hypnos, Narcine, Narke and Temera [6,7]) cannot be detected in torpediniforms. Both character reconstructions suggest that reductive coding might not be the best approach for these features.
A monophyletic clade including stingrays (Myliobatoidei) and panrays (Zanobatus and †Plesiozanobatus) was recovered and corresponds to the order Myliobatiformes. Within the Myliobatoidei, benthic marine stingrays, freshwater stingrays and butterfly rays are grouped in a polytomy, while pelagic stingrays form a monophyletic group that includes †Promyliobatis, as suggested by Marramà et al. [31].

Maximum-Likelihood Analysis (ML)
In the ML topology, †Doliodus, symmoriids and living chimaeroids form a paraphyletic assemblage, among which the monophyletic group consisting of Chimaera and Harriotta is sister to the remaining taxa ( Figure 50).  [41,42]), with Gymnura being recovered in a close relationship to Potamotrygon, separated from Urolophus, Hexatrygon and Plesiobatis. The ML analysis recovered a monophyletic Euselachii, with a sister relationship between the monophyletic †Hybodontiformes and the Elasmobranchii (sensu [13,14]), which agrees with the parsimony hypothesis.
The intrarelationships of Selachimorpha are like those recovered in the parsimony analysis, with a monophyletic Galemorphii as sister to [Squalomorphii + Squatinomorphii] group, with a close relationship between Pristiophorus and the angel sharks, Squatina and †Pseudorhina. The identification of a clade formed by [Galeomorphii + [Squalimorphii + Squatinomorphii]] and the stable placement of the extinct sharks, †Protospinax and †Pseudorhina in both parsimony and ML analyses, is promising and suggests that it is possible to include more extinct sharks in the present matrix to evaluate their systematic position.
A monophyletic Batomorpha clade with some differences compared to the parsimony tree is recovered in the ML topology (compare Figures 49 and 50). Unlike the parsimony tree, the ML did not recover a monophyletic group formed by the Jurassic batomorphs, but placed them as successive sister taxa to the remaining batoids.
Within the †Sclerorhynchoidei, the same three groups previously recovered by families are retained: †Ptychotrygonidae, †Onchopristidae and †Sclerorhynchidae. The remaining taxa are recovered in a clade in which both "Rhinopristiformes" and "Torpediniformes" being paraphyletic. Within this group there is uncertainty in the placement of †Britobatos, which is not unexpected, considering that it shares several features with various groups, including a broad postorbital process (char. 28; shared with Myliobatiformes), vertebral centra in the "synarcual" reaching only the caudal portion of the suprascapula (char. 55; like Myliobatiformes), the presence of a rostral appendix (char. 9; like Rhinopristiformes), differentiated lingual lateral uvulae in teeth (char. 56; like Rhinopristiformes) and the cross-bracing of pectoral radials (char. 111; like Gymnura, Urotrygon and pelagic stingrays).
The paraphyletic status of Rhinopristiformes is due to the placement of the Trygonorrhinidae (including †Tlalocbatos and †Stahlraja) in the myliobatiforms and torpediniforms clade, sharing with the electric rays the presence of a postpelvic process (char. 118 [1]) ( Figure 50).
The paraphyly of the "Torpediniformes" is due to the polyphyly of the thornback rays "Platyrhinidae" since Platyrhina is recovered as sister to the Myliobatiformes. From a morphological point of view, Platyrhina shows several similarities with Myliobatiformes (such as the proximal portion of the propterygium extending beyond the mesocondyle), which causes its position closer to myliobatiforms than to other "platyrhinids" in the ML tree. Within "Torpediniformes", the electric rays (or Torpedinoidei) form a monophyletic group [[Torpedo + Hypnos] + [ †Titanonarke + [Narcine + [Temera + Narke]]]].
The ML tree recovered Myliobatiformes as monophyletic, with the panrays Zanobatidae (i.e., Zanobatus and †Plesiozanobatus) as sister to Myliobatoidei. Within Myliobatoidei, the phylogenetic relations recovered in the ML tree differ from molecular analyses (see [41,42]), with Gymnura being recovered in a close relationship to Potamotrygon, separated from Urolophus, Hexatrygon and Plesiobatis.

Bayesian Inference Analysis (BI)
In the BI, the Holocephali is not recovered monophyletically if the symmoriids are included ( Figure 51). Nevertheless, the living chimaeroids Chimaera and Harriotta form a monophyletic group in a sister relationship to a monophyletic Euselachii. Within this latter clade, the hybodontiforms, selachimorphs and batomorphs are placed in a polytomous relationship. The BI analysis also recovered a monophyletic Selachimorpha, with similar relations within its components as previous topologies (see Figures 49 and 50) and those established by molecular analyses (e.g., [41]).
Batomorpha form a monophyletic group like in previous analyses (Figures 49-51). As in the ML tree, the Jurassic taxa are not monophyletic but arranged as successive sister taxa with †Spathobatis and †Belemnobatis, and †Asterodermus and †Kimmerobatis, respectively, seemingly being more closely related to each other. The remaining Cretaceous and Cenozoic batomorphs form a monophyletic group. However, all the orders within this group, i.e., Myliobatiformes, Torpediniformes, Rajiformes and a paraphyletic "Rhinopristiformes" fall in polytomy.
This analysis also supports the placement of sclerorhynchoids as sister to the remaining fossil and living rajoids. The relations within this order are more resolved than in the ML analysis with †Cyclobatis being placed as the sister taxon to a clade that includes †Ostarriraja, Raja and Bathyraja, in which †Ostarriraja is the sister group of the Raja and Bathyraja group. Within the Sclerorhynchoidei, the †Ptychotrygonidae, †Onchopristidae and †Sclerorhynchidae are also recovered, forming a monophyletic clade ( Figure 51). Similar to the ML tree in the BI analysis, the Trygonorrhinidae are again separated from the main "Rhinopristiformes" clade, which includes the fossil taxa †"Rhinobatos" whitfieldi, †"R." hakelensis, †"R." maronita, †"R." latus, †Iansan, †Pseudorhinobatos, †Eorhinobatos and †Rhombopterygia, consequently recovering a paraphyletic arrangement for this order. Interestingly, the Platyrhinoidei and the Torpedinoidei form a monophyletic group similar to the parsimony tree and molecular analyses (e.g., [41,42]). The intrarelations of the Platyrhinoidei taxa are completely unresolved. In this perspective, their polytomic arrangement suggests that the thornbacks need a more in-depth revision of their characters.
The BI tree places stingrays (Myliobatoidei) and panrays (Zanobatus and †Plesiozanobatus) in a monophyletic clade, the Myliobatiformes. Within this clade there is a large polytomy that includes most of the stingrays. Within the pelagic stingrays, a more resolved topology is recovered, with †Tethytrygon and

Phylogenetic Considerations
With the inclusion of fossil taxa and resampling of characters in the present analyses, we recovered monophyletic Hybodontiformes, Selachimorpha and Batomorpha. Unlike previous morphological analyses (e.g., [23,24,32,33,115,120]), both parsimony and maximum likelihood recovered a sister-group relationship between selchimorphs and batomorphs (Figures 50 and 52).  The present analyses also included the species Pseudobatos productus (Rhinobatos) and Pseudobatos lentiginosus (Pseudobatos), both exhibiting significant variations in their skeletal morphologies, especially of their pectoral girdles. These differences contradict their placement in a single genus, supporting the separation of the American Pacific and Atlantic "Pseudobatos" into different genera. Overall, the present results indicate that the Rhinopristiformes are a group still in urgent need of in-depth phylogenetic studies, before any taxonomic rearrangement is proposed.
The living (Raja and Bathyraja) and fossil ( †Ostarriraja and †Cyclobatis) skates are All three analyses place the Jurassic batomorphs in a sister relation to all the remaining batomorphs [2] (Figures 49-51). Interestingly, the thornbacks (Platyrhinoidei) were recovered sister to the electric rays (Torpedinoidei) forming the order Torpediniformes; in both BI and parsimony a relationship never recovered under morphology-based analyses. However, the phylogenetic relationships of Rhinopristiformes and the recognition of its monophyly, as most recent taxonomic studies suggest [2,3], continues to be problematic, with their monophyly not being consistently found even in molecular analyses (e.g., [42]) and their composition also differing (e.g., [41,42]) between these studies. In the present study, only the parsimony analysis recovered the Rhinopristiformes as a monophyletic group ( Figure 49) while maximum-likelihood (ML) and Bayesian inference (BI) analyses recovered two groups of Rhinopristiformes, with the family Trygonorrhinidae (Trygonorrhina, Zapteryx and Aptychotrema) consistently found as a separate clade (Figures 50 and 51). All the analyses suggest close relation of the fossil taxa †Stahlraja and †Tlalocbatos within the Trygonorrhinidae, therefore placing the origin of this family well into the early Cretaceous (Albian-Aptian), suggesting a long separated evolutionary history between Trygonorrhinidae and the remaining members of the Rhinopristiformes.
The present analyses also included the species Pseudobatos productus (Rhinobatos) and Pseudobatos lentiginosus (Pseudobatos), both exhibiting significant variations in their skeletal morphologies, especially of their pectoral girdles. These differences contradict their placement in a single genus, supporting the separation of the American Pacific and Atlantic "Pseudobatos" into different genera. Overall, the present results indicate that the Rhinopristiformes are a group still in urgent need of in-depth phylogenetic studies, before any taxonomic rearrangement is proposed.
The living (Raja and Bathyraja) and fossil ( †Ostarriraja and †Cyclobatis) skates are grouped with sclerorhynchoids, which is consistent with previous analyses that recovered this relationship based on features of the branchial skeleton [32,117]. Rajiformes show a rather intriguing suite of morphological characters, which are also present in several batomorph groups, including rostral appendices (also observed in Rhinopristiformes), catenated calcification on the pectoral-fin radials and lack of a postpelvic process (also observed in Myliobatiformes). The sharing of these features with guitarfishes and myliobatiforms ultimately produces the distinct systematic placement displayed by this group in the parsimony and ML topologies, with the BI analysis neither supporting nor contradicting these arrangements.
As far as the catenated calcification pattern of radials is concerned, which is of a different type in rajoids (two-chained) and some benthic stingrays (four-chained) [63], we hypothesize that "catenated" two-chained and "catenated" four-chained represent two different calcification patterns that evolved independently. Overall, the placement of the Rajiformes in the ML topology resembles that recovered by molecular analyses [41,42], favoring the placement of Rajiformes as sister group to all the other Cretaceous and Cenozoic batoids, which might be a point in favor of the ML topology. However, the ML also recovered some problematic arrangements such as that of Platyrhina as sister taxon to the Myliobatiformes.

Conclusions
The advantages and drawbacks of the various phylogenetic methods are extensively discussed in a series of papers (e.g., [121][122][123][124][125][126]. With these criteria responding differently to the patterns in the datasets, it is not surprising that different topologies were found in the present study. O'Reilly et al. [121,122] proposed that model-basis methods are more accurate than parsimony when analyzing morphological data. However, Goloboff et al. [123,124] showed that this happens under specific parameters and that parsimony analyses also have the potential to be very informative. The present analysis focusing on batomorph relationships reveals the need for an even more profound selachian sampling (extinct and extant species), when comparing the phylogenetic topologies recovered under the different criteria, or when evaluating character reconstructions and re-evaluating the codification of morphological characters.
Maximum-likelihood and parsimony analyses recovered more resolved topologies than Bayesian inference. However, ML and parsimony disagree in some arrangements (e.g., in the placement of some batomorph groups, the monophyletic status of Rhinopristiformes, the phylogenetic placement of †Britobatos and the phylogenetic relationships of Torpediniformes), these different suggest areas of interest for future works. Our results also indicate that a revision of Jurassic batomorphs and selachimorphs is of utmost importance to provide a more consistent topology, especially related to deeper nodes. Such a morphological-trait revision also has the potential to better understand the composition of some groups recognized by molecular analyses but not in morphological analyses (e.g., Rhinopristiformes and Torpediniformes).
While the relationships between selachimorphs and batomorphs are not completely resolved in the present analysis, our character and taxon sampling presents a persistent placement of most of the fossil taxa within certain clades (orders and families), and their consistent similar relationships in all tree topologies-often similar to those previously hypothesized (e.g., [23,24,41,127])-is promising and indicates that the inclusion of more fossil taxa in the present matrix likely will not cause loss of resolution, therefore suggesting that a strong phylogenetic signal can be recovered from fossil taxa.

Institutional Review Board Statement: Not applicable.
Data Availability Statement: All data used by the authors for the analysis is available in the supplementary materials, which can be downloaded at: www.mdpi.com/xxx/s1.