Node age estimations and the origin of angel sharks, Squatiniformes (Neoselachii, Squalomorphii)

Transcription

1 Journal of Systematic Palaeontology ISSN: (Print) (Online) Journal homepage: Node age estimations and the origin of angel sharks, Squatiniformes (Neoselachii, Squalomorphii) Stefanie Klug & Jürgen Kriwet To cite this article: Stefanie Klug & Jürgen Kriwet (2013) Node age estimations and the origin of angel sharks, Squatiniformes (Neoselachii, Squalomorphii), Journal of Systematic Palaeontology, 11:1, , DOI: / To link to this article: View supplementary material Published online: 30 Nov Submit your article to this journal Article views: 2222 View related articles Citing articles: 10 View citing articles Full Terms & Conditions of access and use can be found at

2 Journal of Systematic Palaeontology, 2013 Vol. 11, Issue 1, , Node age estimations and the origin of angel sharks, Squatiniformes (Neoselachii, Squalomorphii) Stefanie Klug a and Jürgen Kriwet b a University of Bristol, Department of Earth Sciences, Wills Memorial Building, Queens Road, Bristol BS8 1RJ, UK; b Department of Paleontology, University of Vienna, Geozentrum, Althanstrasse 14, 1090 Vienna, Austria (Received 1 August 2011; accepted 10 August 2011; first published online 30 November 2012) The fossil record of angel sharks is reviewed with special focus on skeletal remains. A new family Pseudorhinidae is established for two species based on complete specimens from the Late Jurassic and a possible third species based on isolated teeth. This clade is a member of the stem lineage of Squatiniformes. Squatinidae represents the crown with a reliable fossil record extending back into the Aptian based on a partial skeleton displaying characteristic morphological traits of Squatina. We also present revised morphological descriptions of the skeletal remains of Squatina baumbergensis (Campanian of Germany) and Squatina sp. (Miocene of Japan). In this study, we used reliable skeletal remains and a modified approach to establish the origin and divergence of the Squatiniformes and the Squatinidae. Isolated teeth are considered to be unreliable because of the poor knowledge of squatiniform dental character traits and their evolution. We estimate a hard minimum age constraint of Ma and a soft maximum age constraint of Ma for the origin of the Squatiniformes. For the crown represented by the Squatinidae, we estimate a hard minimum age constraint of 114 Ma and a soft maximum age constraint of Ma. These age constraints most likely designate the timing of the origin of identifiable squatiniform and squatinid characters as currently understood rather than the origin of the Squatiniformes or the divergence between the Pseudorhinidae and the Squatinidae. The lack of pre-late Jurassic pseudorhinids and pre-cretaceous squatinids probably represents an artefact because characteristic squatiniform tooth morphologies, which generally provide only a restricted set of diagnosable features, might not have yet been fully developed. Consequently, skeletal remains of neoselachians from the Early and Middle Jurassic are crucial for establishing reliable characters of stem-lineage representatives and to avoid misinterpretations resulting from transferring morphological traits of the crown to fossil groups with unresolved interrelationships, as well as sister-group relations. Keywords: Squatinidae; Pseudorhinidae nov. fam.; fossil record; stem lineage; divergence; character evolution Introduction Living angel (or devil) sharks represent a peculiar group of bottom-dwelling sharks. They superficially resemble rays and are of moderate body size (total length about 1 2 m) with enlarged pectoral fins, laterally located gill slits, dorsally situated eyes and spiracles, and posteriorly placed spineless dorsal fins, as well as a hypocercal caudal fin but no anal fin. They are characterized by a combination of plesiomorphic and derived features, which are supposedly related to their highly adapted bottom-dwelling mode of life (e.g. de Carvalho et al. 2008). Extant angel sharks are placed in a single genus, Squatina Duméril, 1806, representing the family Squatinidae within Squalomorphii. Compagno (1973, 1984) considered the Squatinidae to be a distinct group of sharks and consequently placed them in their own order Squatiniformes (Fig. 1). To date, 22 species have been described which are considered valid (e.g. Compagno 1973, 1977; Shirai 1992a, b; Compagno et al. 2005a, b; Castro-Aguirre et al. 2006; Last & White 2008; Kriwet et al. 2010; Stelbrink et al. 2010). Angel sharks are distributed worldwide in temperate to tropical seas (Compagno et al. 2005a, b; Last & White 2008), inhabiting continental shelves and upper continental slopes down to 500 m (Compagno et al. 2005a). The majority of living species display high endemism patterns with restricted occurrences; only a few species display wider geographical ranges (e.g. Kriwet et al. 2010). Stelbrink et al. (2010) identified four well-supported geographical clades based on molecular analyses: (1) Europe North Africa Asia; (2) North and South America; (3) Australia; and (4) South Africa. Modern taxonomic diversity is the result of vicariant events, e.g. the closure of the Tethys Sea and the rise of the Isthmus of Panama. The evolutionary history of angel sharks remains incompletely known despite the progress accomplished in the last few years. This is mainly related to the nature of their fossil record, which consists predominantly of isolated teeth. Corresponding author. stefanie.klug@bristol.ac.uk C 2013 Natural History Museum Published online 30 Nov 2012

3 92 S. Klug and J. Kriwet Figure 1. Phylogenetic hypotheses of Neoselachii and the fossil record of Squatiniformes combined with stratigraphical information (Gradstein et al. 2004; Ogg et al. 2008). A, hypothesis based on morphological data; B, hypothesis based on molecular data; C, skeletal fossil record of Squatiniformes with the extinct genus Pseudorhina (including two species P. acanthoderma and P. alifera) and six different species of Squatina. Holomorphic (complete) specimens or identifiable skeletal remains are rare or have not yet been considered in detail. Fossil elasmobranch taxa based on teeth are certainly useful for reconstructing their fossil record and past diversity patterns and to a lesser degree to infer phylogenetic relationships. However, teeth of fossil squatiniforms are very similar to those of various orectolobiforms, resulting from convergent dental developments related to similar feeding and lifestyle adaptations. Additionally, morphological differences for species discrimination within Squatina and related taxa are often only subtle and mostly difficult to evaluate if only small samples are available. So far, 34 fossil Squatina species mostly based on isolated teeth have been described (Cappetta 2006). While Underwood (2002, 2006) considered the oldest representative to be of Kimmeridgian age, de Carvalho et al. (2008) showed

4 Node age estimations and the origin of angel sharks 93 that Late Jurassic squatiniforms all belong to a different taxon, Pseudorhina. All pre-kimmeridgian records that have been variously assigned to Squatina were considered to belong to orectolobiforms by Underwood (2006). However, the squatiniform fossil record does not consist entirely of isolated teeth but includes several holomorphic specimens and skeletal remains from different stratigraphical ages throughout their evolutionary history. These remains are important for identifying indisputable extinct members of Squatina and other squatiniform groups. Only with this knowledge is it possible to provide origin and diversification estimates for different clades and to reduce gaps in the fossil record (ghost-lineages). Thus the aims of this paper are: (1) to discuss the position of Squatina and related taxa within the Squatiniformes in order to provide the basis to define a new family; (2) to provide characters for ultimately distinguishing Squatina from other squatiniforms; (3) to review the fossil skeletal record of members of Squatina, providing osteological characters of extinct taxa; (4) to provide origination estimates of squatiniform groups; and (5) to discuss limitations of dental features for identifying stem-lineage representatives, which are important when sister-group relationships are ambiguous. Material and methods Institutional abbreviations AMNH: American Museum of Natural History, New York, USA; BMB: Booth Museum of Natural History, Brighton, UK; NHMUK PV: Natural History Museum, London, UK; BSPG: Bayerische Staatssammlung für Paläontologie und Geologie, Munich, Germany; GPIM: Museum of the Geological-Palaeontological Institute of Münster, Germany; GPIT: Geologisch-Paläontologisches Institut Tübingen, Germany; MFM: Mizunami Fossil Museum, Mizunami, Japan; SMNS: State Museum of Natural History Stuttgart, Germany. Material The focus of this study is on skeletal remains and holomorphic specimens from the Cenozoic, Cretaceous and Jurassic assigned to Squatina or Squatiniformes, respectively. Jurassic and Early Cretaceous taxa based on isolated teeth, however, are considered at times if they provide additional information on the stratigraphical ranges of Squatina and related taxa. A complete list of fossil squatiniform species and their stratigraphical and geographical distributions is provided in the Supplementary Online Material. The systematic arrangement of holomorphic specimens and skeletal remains considered in the systematic section is in stratigraphical order. The following specimens were studied (in stratigraphical order): GPIT 6842, GPIT 8214, SMNS 80144/24, SMNS 80431/20, skeletal remains and holomorphic specimens of Pseudorhina acanthoderma from the upper Kimmeridgian (Upper Jurassic) of Nusplingen, Germany; BSPG AS VII 3, BSPG AS I 817, BSPG AS I 1368, holomorphic specimens of Pseudorhina alifera from the lower Tithonian (Upper Jurassic) of the Solnhofen area, Germany; NHMUK PV P.65647a, b, incomplete specimen of Squatina sp. 1 from the Aptian (Lower Cretaceous) of England; BMB 7330, incomplete specimen of Squatina cranei from the Cenomanian (Upper Cretaceous) of Clayton, West Sussex, England; NHMUK PV P.12213, incomplete specimen of Squatina cranei from the Cenomanian of Halling in Kent, England; GPIM 8553, incomplete specimen of Squatina baumbergensis from the upper Campanian (Upper Cretaceous) of Baumberge in Westphalia, Germany; MFM , disarticulated and incomplete skull of Squatina sp. 2 from the lower Miocene (Cenozoic) of Japan; AMNH 55686, cleared-and-stained specimen of Squatina californica (extant), figured in de Carvalho et al. (2008, fig. 4). Node age estimation Estimating divergence times using fossil taxa is still controversial and frequently criticized (e.g. Heads 2005; Pulquério & Nichols 2007). To overcome flaws and problems associated with using selected fossils for calibrating molecular trees, Benton & Donoghue (2007) introduced a method for estimating hard minimum and soft maximum age constraints based on the fossil record of organisms and aspects of phylogenetic bracketing and stratigraphical bounding. This procedure, although straightforward, largely disregards evolutionary events, which might provide additional valuable constraints for origin estimates. We, therefore, combine their approach with aspects of divergence estimates of neoselachians derived from statistical analyses (Kriwet et al. 2009) for setting a time framework. Accordingly, the first major divergence of neoselachians (galeomorphs and squalomorphs) can be assumed to have occurred in the late Early Jurassic. The hypothesis to test here is whether the origin of Squatiniformes, which is member of Squalomorphii, correlates with this first major divergence event or occurred earlier. For this, however, we need to dismiss the phylogenetic bracketing approach using basal sister groups for inferring node ages, because this would imply a priori that squatiniforms originated before the first major diversification event of neoselachians. Conversely, we consider a posterior approach (not a strict Bayesian approach, however) for inferring possible origination dates of squatiniform groups before or after this first major diversification event and comment on the correlated hard minimum and soft maximum age constraints. To obtain a soft maximum age constraint, we calculated the quality of the fossil record of Squatinidae considering their known stratigraphical occurrences and their total stratigraphical range, which also includes ghost lineage occurrences. In a subsequent step, the 95% confidence interval

5 94 S. Klug and J. Kriwet was calculated and compared with the fossil distribution of their sister group, which is represented by Pseudorhina. Hypothetically, the derived stratigraphical range of squatinids should extend backwards at least to the oldest known fossil of their sister group. The same approach as for Squatina then is applied to the sister group and compared with the results of Kriwet et al. (2009) and molecular phylogenetic hypotheses. According to cladistic arguments, the origin of Squatina correlates with the origin of Squatiniformes. A major problem in identifying members of the groups under consideration close to their origins is the patchiness of the squatiniform fossil record and the very limited character sets for distinguishing plesiomorphic and apomorphic characters. Thus, evolutionary morphological character changes from the stem to the crown might remain undetected. We do not comment on the intrarelationships of squatinid taxa despite the fact that the recorded tooth morphologies of fossil squatinids indicate the presence of different genera in the fossil record. Systematic palaeontology In this section, only skeletal fossils of squatiniform sharks are considered, with special focus on their cranial and postcranial morphology as far as it is preserved, rather than on their dentitions. A detailed description of English Cretaceous squatinid tooth morphologies, including those of Squatina cranei, is provided by Guinot et al. (in press). Superclass Chondrichthyes Huxley, 1880 Class Elasmobranchii Bonaparte, 1838 Cohort Euselachii Hay, 1902 Subcohort Neoselachii Compagno, 1977 Superorder Squalomorphii Compagno, 1973 Order Squatiniformes Compagno, 1973 Diagnosis. (modified from Nelson 2006). Squalomorph sharks with: (1) dorsoventrally flattened, ray-like body; (2) dorsally situated eyes; (3) two spineless dorsal fins; (4) terminal to subterminal mouth; and (5) anterior margin of nostrils with barbels. Pseudorhinidae fam. nov. Type genus. Pseudorhina Jaekel, Included genera. Type genus only. Stratigraphical range. Oxfordian Tithonian. Geographical range. Only known from Europe. Diagnosis. Squatiniform sharks characterized by the following combination of characters: (1) basihyal with straight anterior margin, slightly curved laterally and very concave posteriorly; and (2) low and tightly articulated plate-like supraneurals. Differential diagnosis. Members of Pseudorhinidae fam. nov. share numerous derived characters with squatinids, such as absence of an anal fin, a strong groove in the orbital cranial roof accommodating the orbital palatoquadrate process, a massive and anterolaterally projecting postorbital process, subtriangular and transversally oriented labial cartilages, pectoral fins with anterior subtriangular lobes, and a slender and anteriorly concave puboischiadic bar. The family is defined by an autapomorphic (morphology of basihyal) and a plesiomorphic character (morphology of supraneurals), which represents the ancestral condition for neoselachians. Additionally, Pseudorhinidae differs from Squatinidae in: (1) having a proportionally longer anterior fontanelle with less rounded posterior margin; (2) more laterally orientated postorbital processes; (3) lateral capsular walls of otic region being less concave; (4) jaws more acute angled and relatively less transverse; (5) labial protuberance of teeth not supported by root; (6) comparably shorter and more tightly connected basiventral processes; (7) longer ribs; and (8) a comparably longer postpelvic tail. Comparison. The morphology of the basihyal cartilage is considered a systematically informative structure in chondrichthyans in general (de Carvalho et al. 2008). The morphology displayed by Pseudorhinidae is unique amongst neoselachians and readily separates the Late Jurassic squatinids from living Squatina species (Fig. 2). The dental structure of Jurassic squatiniforms also differs from that of Squatina species. In pseudorhinids, the tooth crown is low, the labial apron more rectangular or broadly rounded, rather massive, and well detached from the root. The root is comparably high and massive with a much more pronounced labial root depression in comparison to squatinids. In labial view, the root lobes are more differentiated and flared. In basal view, the labial margin of the root is convex in anterior and anterolateral teeth or linear in lateral and posterolateral teeth (cf. Underwood 2002, pl. 4; de Carvalho et al. 2008, fig. 14). In the extant Squatina spp., the root lobes are not differentiated and the labial margin of the root is broadly concave. In some species, such as S. squatina, the root lobe extremities might be directed posteriorly in basal view (cf. Herman et al. 1992, pl. 45; pers. obs.). Additionally, the lateral crown shoulders strongly overhang the root in pseudorhinids. Teeth of juvenile pseudorhinids exhibit well-developed lateral cusplets. In general appearance, teeth of pseudorhinids resemble teeth of various orectolobiforms or even heterodontids if juvenile teeth are considered. This dental morphology is regarded here to be plesiomorphic for squatiniforms. Other morphological features of Pseudorhina, such as the basihyal and the form of the supraneurals, display the plesiomorphic condition compared to Squatina (de Carvalho et al. 2008) within Squatiniformes. Based on these observations, we interpret Pseudorhina as the sister taxon of Squatina.

6 Node age estimations and the origin of angel sharks 95 Figure 2. Morphological comparison of the basihyal of Squatina and Pseudorhina. A, cleared-and-stained specimen Squatina californica (AMNH 55686) in ventral view; B, Pseudorhina acanthoderma (SMNS 80431/20) from the lower Kimmeridgian of Nusplingen, S Germany; close-up of skull in ventral view. Abbreviation: bh, basihyal. Genus Pseudorhina Jaekel, 1898 Included species. Pseudorhina acanthoderma (Fraas, 1854) from the late Kimmeridgian; P. alifera (Münster, 1842) from the early Tithonian; P. frequens (Underwood, 2002) from the early Kimmeridgian. Remarks. de Carvalho et al. (2008) presented detailed morphological accounts of Pseudorhina and we refer to this paper. The most conspicuous differences between both species known by holomorphic specimens ( P. acanthoderma, P. alifera; see Fig. 3 herein) include more massive teeth with proportionally lower cusps and broader labial protuberances in P. alifera compared to P. acanthoderma and a shorter vertebral column in P. alifera, comprising about 120 compared to centra in P. acanthoderma. Pseudorhina frequens (Underwood, 2002) differs from P. alifera and P. acanthoderma in its proportionally higher and more delicate teeth. We consider these differences significant for species discrimination, although there seems to be some morphological overlap between P. frequens and P. alifera. Pseudorhina represents an extinct lineage of angel sharks, which is seemingly restricted to the Late Jurassic (Fig. 1C). All isolated squatiniform teeth from Jurassic strata (e.g. Pseudorhina frequens) display the same tooth root structure as in the holomorphic specimens and accordingly are assigned to this genus. So far, no isolated teeth of modern squatiniform morphology have been recovered from Jurassic strata. Age. The stratigraphically oldest holomorphic specimen of Pseudorhina alifera is from the lowermost Tithonian, Hybonoticeras hybonotum Zone, Lithacoceras riedense Subzone of Painten in southern Germany. The stratigraphically oldest holomorphic specimen of P. acanthoderma is from the late Kimmeridgian, Hybonoticeras beckeri Zone, Lithacoceras ulmense Subzone ( Liegende Bankkalk-Formation ) of Nusplingen in southern Germany. These records consequently are and Ma old, respectively. The oldest record of P. frequens is from the upper Rasenia cymodoce Zone, Rasenia evoluta Subzone of the early Kimmeridgian, which ranges from to Ma. The oldest fossil record of Pseudorhina, however, is Pseudorhina sp. from the lower Epipeltoceras bimmamatum Zone of the late Oxfordian of Southwest Germany (Thies 1983), which ranges from to Ma. Thus, we consider the oldest verified fossil record of Pseudorhina to be Ma. Family Squatinidae Bonaparte, 1838 Type genus. Squatina Duméril, Included genera. Squatina Duméril, 1806 (extant). Stratigraphical range. Valanginian Recent. Geographical range. Cosmopolitan. Diagnosis. For characters defining Squatinidae, see Compagno (1973, 1984).

7 96 S. Klug and J. Kriwet Figure 3. Pseudorhina from the Upper Jurassic of South Germany. A, holotype of Pseudorhina alifera (BSPG AS VII 3; adult specimen) from the Tithonian of Solnhofen area; B, juvenile specimen of Pseudorhina alifera (BSPG AS I 1368), previously designated as P. speciosa, from the Tithonian of Solnhofen area; C, Pseudorhina acanthoderma (SMNS 86214/41; adult specimen) from the lower Kimmeridgian of Nusplingen.

8 Node age estimations and the origin of angel sharks 97 Remarks. All extant angel sharks are placed into a single genus, Squatina Duméril, 1806, representing a single family, Squatinidae, within the Squatiniformes (Compagno 1973, 1984). Fossil angel sharks traditionally also are assigned to Squatina. Herman (1982) described Parasquatina cappettai based on a single isolated tooth from the Maastrichtian of north-western Germany, which he tentatively included in Squatinidae because of some overall resemblances (e.g. labial apron, crown morphology). This species also occurs in the Maastrichtian of southern Germany (F. Pfeil, pers. com.). Two additional species are present in the Late Cretaceous of England (C. Underwood, pers. com.). Although an evaluation of the dental morphology of Parasquatina is beyond the scope of this study, which primarily focuses on skeletal fossil remains, it nevertheless is necessary to address the systematic affinities of this genus. One of us (JK) was able to study the type specimen, which is housed in the Bundesanstalt für Geowissenschaften und Rohstoffe, Hanover (NLfB Hannover, collection number kma 249 ). The single specimen most probably represents a lateral tooth based on the distally inclined cusp. The tooth crown resembles that of squatinids but also of Squatirhina, as already indicated by Herman (1982). A labial protuberance as in squatinids is not developed. There is only a minor bulge at the basal edge of the labial face. The root is hemiaulacorhize. Although Herman (1982) stated that although squatinid-like, the root strongly resembles that of several orectolobiforms. We consider Parasquatina to be an orectolobiform rather than a squatiniform based on the morphology of the crown and root. Cappetta (2006) also assigned Parasquatina to galeomorphs but without any further systematic attribution. We nevertheless agree with Guinot et al. (in press) that only skeletal material might reveal its real systematic affinities. Genus Squatina Duméril, 1806 Diagnosis. For characters defining Squatina, see Compagno (1984) and de Carvalho (1996). Remarks. The genus Squatina is characterized by numerous derived characters, including posterodorsally projecting orbital processes of the palatoquadrates (which terminate dorsally in close contact with the orbit, medial to the eyeball), a complete postorbital wall penetrated by jugular canals, a robust and subtriangular inferior labial cartilage, a triangular basihyal and a cross-shaped basibranchial copula among other features (Iselstöger 1937; Holmgren 1941; Compagno 1973, 1977; de Carvalho et al. 2008). Most significant is the basihyal, which is considered to be a systematically informative structure for chondrichthyans in general, as noted above (e.g. de Carvalho et al. 2008). In Squatina it is triangular and almost an inverted T-shape, which distinguishes it from all other neoselachians. A stable dental character of Squatina is the labial knob-like apron that is well supported by the root in basal view, a flat basal root face and the characteristic outline of the root in basal view (see below). The skeletal fossil record of Squatina is very scant and restricted to the late Early and Late Cretaceous, and Miocene. Casier (1966) described two isolated neurocrania which he assigned to Squatina prima (NHMUK PV OR and NHMUK PV P.2278) from the lower Eocene London Clay Formation of Sheppey, UK. An examination of both specimens could not conclusively confirm their taxonomic identity and therefore these are excluded from the current study. The arrangement of taxa in this section is in stratigraphical order from oldest to youngest. Squatina sp. 1 Material. This is the oldest Cretaceous holomorphic angel shark specimen (Fig. 4). It was recovered from the abandoned and filled Cambridge Sandpit, which is situated south of the A31 at the Coxbridge Roundabout, near Farnham in Surrey, UK. The specimen comes from the Folkestone Beds of the Lower Greensand (Aptian). So far, no anatomical description of this specimen has been published. Description. The specimen, NHMUK PV P.65647a, b (part and counterpart), is preserved in a calcareous nodule. It lacks the skull, the anterior portion of the body, most of the pectoral girdle, the anterior part of the right pectoral fin and most of the posterior body including the caudal tips of the basipterygia and the tail. The specimen shows soft tissue preservation in calcium phosphate such as muscle fibres and organs. The spiral anal sphincter valve is preserved on the part as well as the counterpart and is situated anterior to the pelvic girdle next to the vertebral column (Fig. 4A, C). The liver might be preserved according to the position and the size of a clearly visible soft tissue structure on the left side of the specimen (Fig. 4A). The coracoid bar of the pectoral girdle is very fragmentary and only the left scapular process is preserved. It is long, comparatively narrow, medially concave and curved medially towards the vertebral column. As far as can be ascertained, it extends posteriorly beyond the mid-length of the pectoral fin. It is not possible to determine whether the suprascapular cartilage, which is present in living squatinids, is lacking, fused to the scapular or was not fossilized. There are two separate condyles present for articulation with the pectoral fin basals. The meta-, meso- and propterygium are well preserved in the left pectoral fin; in the right pectoral fin, the propterygium is completely and the mesopterygium largely missing but identifiable by impressions left in the sediment. Accordingly, the propterygium extends anteriorly and is concave and slender in comparison to the mesopterygium. The mesopterygium broadens distally and is almost cuneiform with a slightly concave mesial margin. The metapterygium is almost triangular with a medially concave margin and slender anterior and posterior sections. The posterior portion is elongated, slender and

9 98 S. Klug and J. Kriwet Figure 4. Squatina sp. 1 (NHMUK PV P.65647a, b) from the Aptian of England, UK; plate (A) and counterplate (B). A, pectoral girdle and anterior part of the pelvic girdle; B, complete pelvic girdle; C, close-up from the pelvic girdle. Abbreviations: co, coracoid bar; f, foramen; l, liver; msp, mesopterygium; mtp, metapterygium; pb, puboischiadic bar; pop, propterygium; ppp, prepelvic process; rpg, radial pterygophore; sasv, spiral anal sphincter valve; scp, scapular process. tapers caudally into a point. It was probably subdivided as in living squatinids as indicated by fractured zones on the left and right sides (Fig. 4A). Radial elements articulate with all three basals, but the number articulating with the pro- and metapterygium cannot be established; radials articulate with the mesopterygium. The pectoral fin was plesodic as in living squatinids but the number of radial series and the width of the pectoral fin cannot be determined because of the poor preservation of the specimen. The puboischiadic bar of the pelvic girdle is well preserved (Fig. 4B, C). It is distally elongated, very narrow and anteriorly distinctly concave. A short and acute prepelvic process is present on each anterolateral edge of the puboischiadic bar. Posteriorly projecting processes are not developed. There seems to be at least one nerve foramen on each side of the puboisciadic bar (Fig. 4A). The basipterygium is curved, concave medially, slender and tapering posteriorly, and almost as long as the puboischiadic

10 Node age estimations and the origin of angel sharks 99 bar (as inferred from the reconstructed left element). The radials articulate with the undulating distal margin. Two series of radials are recognizable; the exact number cannot be established due to preservational limitations. The basal series comprises the broadest elements. The anteriormost radial is enlarged as in living squatinids and articulates with the posterolateral corner of the puboischiadic bar. It is not possible to determine whether claspers were present. The vertebral column is partly preserved between the girdles and a short distance (c.4 5 centra) caudally behind the pelvic girdle. The centra are well calcified, subcircular in anterior and posterior view with slightly levelled dorsal and ventral margins, comparatively short and more or less spool-shaped. They consist of numerous concentric calcification rings (tectospondylic type). All preserved centra tightly articulate with each other. The transition from mono- to diplospondyly is concealed by preserved soft tissue (remnants of the liver?). Anteriorly, just behind the pectoral girdle, faint structures laterally to the vertebral column might represent remnants of processes projecting from the centra. Supraneurals are not preserved. Remarks. This specimen displays a suite of skeletal features, which are characteristic for living squatinids. Apomorphic features, such as the specific basihyal shape and the form and arrangement of supraneurals, are unfortunately not preserved. Nevertheless, its assignment to Squatina is beyond doubt, based on the morphology of the girdles and associated fin elements. Species discrimination, however, is impossible because of the very incomplete nature of this specimen. Age. The Folkestone Beds of the Lower Greensand belong mainly to the late Aptian Hypacanthoplites jacobi Zone, which ranges from to Ma. Thus, the age of the fossil is considered to be 114 Ma. Squatina cranei Woodward, Teeth of squaloid fish ; Dixon: 12, pl. 30, fig Squatina cranei Woodward: 144, pl. 7, figs Squatina cranei Woodward; Woodward: Squatina cranei Woodward: Woodward: 224, pl. 4, figs Squatina cranei Woodward; Herman: 123, pl. 5, fig Squatina cranei Woodward; Müller & Diedrich: 21. Description. Both known skeletal remains of this species, which are housed in the Booth Museum, Brighton (BMB 7330, holotype) and the Natural History Museum, London (NHMUK PV P.12213), were examined. These specimens have been exhaustively described (e.g. Guinot et al. inpress for cranial anatomy; Woodward 1888, 1912; Herman 1972; Müller & Diedrich 1991). Here, we only refer to those characters identifying this species as belonging to Squatina rather than to any other squatiniform. Both specimens are very fragmentary and only parts of the skull and anterior body are preserved. The Brighton specimen comprises cranial elements in part and counterpart so that most elements only are partially visible or are heavily fragmented. Unfortunately, it is not possible to identify the basihyal. The most completely preserved skeletal elements belong to the mandibular and hyoid arches (Fig. 5A). The broad orbital process of the palatoquadrate, which displays the characteristic morphology for squatiniforms in general, is preserved. Numerous teeth are still preserved along the oral border of the Meckel s cartilage and the palatoquadrate. Some detached teeth are accessible in basal view (Fig. 5B). They display the characteristic root-supported labial protuberance, which is rounded in cross section, and the concave labial root margin without any distinct labial root depression lingual to the protuberance. The central root foramen is large as it is in teeth of extant Squatina spp. In NHMUK PV P.12213, a single, slightly damaged element is located anterior to the vertebral column, more or less in the region of the branchial chamber (Fig. 5C). This element displays the characteristic triangular outline of the basihyal as found in extant Squatina spp., but is seemingly rotated to the left. Hexagonal phosphatic prisms densely cover this element so that its morphological identification is not completely supported. Remarks. This species is well known from few skeletal remains and abundant isolated teeth from the Cenomanian (Holaster subglobosus Zone) of Sussex and Kent in southern England. The taxonomic identification by Woodward (1888) was based on general similarities of the cranial cartilages in comparison to those of extant species and the general morphology of teeth, especially the anterior upper and lower ones. These characters are not helpful for identification because of the very similar cranial and dental morphologies found in squatiniforms in general. Nevertheless, several characters (see above) identify this species as an undoubted member of Squatina. Age. The Holaster subglobosus Zone of the Lower Chalk is equivalent to the Grey Chalk or Plenus Marls Member in England (Hopson 2005) and covers most of the Metoicoceras geslinianum Zone, corresponding to the Nigericeras scotti + Neocardioceras juddii Zones of the North American biostratigraphical scheme. This zone ranges from to Ma. We consequently assume the age of this fossil to be Ma. Squatina baumbergensis von der Marck, Squatina baumbergensis von der Marck: 264, pl. 25, figs Squatina baumbergensis von der Marck; Woodward: 68.

11 100 S. Klug and J. Kriwet Figure 5. Squatina cranei from the Cenomanian of South England. A, skull of the holotype (BMB ) in dorsal view (Clayton, West Sussex,UK); B, articulated specimen (NHMUK PV P.12213) displaying the disarticulated skull and anteriormost vertebrae (Halling, Kent, UK). Abbreviations: hm, hyomandibular; lpq, left palatoquadrate; nc, neurocranium; o, orbita; orp, orbital process; rmc, right Meckel s cartilage; rpq, right palatoquadratum Rhina baumbergensis (von der Marck); Siegfried: 8, pl. 1, fig Squatina baumbergensis von der Marck; Müller: 32, pl. 6, fig. 5; pl. 7, fig Squatina baumbergensis von der Marck; Thies & Müller: 101, fig. 18. Description. The single holomorphic specimen known of this species is preserved in ventral view and measures c.50 cm in length. It is not complete and lacks the caudal and dorsal fins. Cranial elements are largely obscured by the characteristic prismatic surface rendering the identification of most elements very difficult. A single tooth is visible that enables the assignment of isolated teeth to this species. The coracoid bar is slender with a feeble sigmoidal curvature. The scapular processes are comparatively short and massive with broad bases and are medially curved towards the vertebral column. The pectoral fin is plesodic but the number of radial series could not be established because of the general poor preservation of the specimen. The propterygium of the pectoral fin is short and rectangular. The mesopterygium is conspicuously slender with an elongated posterior segment. The metapterygium is slender, subtriangular and mesially curved (Fig. 6A). The puboischiadic bar of the pelvic girdle is well preserved. It is distally elongated, very narrow and anteriorly concave. A short and acute prepelvic process is present at each anterolateral edge (Fig. 6B). The basipterygium is curved, concave medially, slender, posteriorly tapering and almost as long as the puboischiadic bar (as inferred from the reconstructed left element). The anteriormost radial is enlarged as in living squatinids and articulates with the posterolateral corner of the puboischiadic bar. There are elongated structures mesiodistal to the pelvic fins that represent remains of the clasper organs (Fig. 6A). The vertebral column is well preserved and consists of tightly articulated centra of tectospondylly type. It is most parsimonious to assume that this species also possessed supraneurals as in living and extinct forms, although their arrangement and morphology remain vague. The skull is preserved in general outline, but distinct structures are poorly conserved. The contour of the upper and lower jaws with a single preserved tooth is recognizable. This tooth displays the general dental morphology as found in extant Squatina species but is also very similar to the tooth architecture of Late Jurassic forms. The hyoid and branchial apparatus are fragmentary (Fig. 6C). The hypobranchials are the only branchial elements that can be distinguished. Indistinguishable elements and placoid scales obscure the form of the basibranchial, to which the hypobranchials are attached. The hyomandibulae are short and massive, but it is not possible to identify any median

12 Node age estimations and the origin of angel sharks 101 Figure 6. Holotype of Squatina baumbergensis from the upper Campanian of Baumberge in Westphalia (NW Germany). A, ventral view; B, close-up of the left pelvic girdle; C, close-up of the skull displaying the branchial apparatus and the basihyal. Abbreviations: ba, branchial apparatus; cl, clasper organs. ridge, as is present in extant forms. The ceratohyals are curved anteriorly, long and slender, wide where they articulate to the basihyal, but their posterior ends, where the hyomandibulae articulate, are more slender. The basihyal (Fig. 6C) is positioned in its original position articulating with the two ceratohyals and has the characteristic morphology as found in extant squatinids, even though its margins are partly covered by shagreen and cartilage fragments. Remarks. This species originally was based on the single known holomorphic specimen of a squatinid from the upper Campanian of Baumberge in Westphalia, northwestern Germany by von der Marck (1885; Fig. 6). In its general morphology, the specimen resembles the extant Squatina californica and another specimen from the Aptian of England (see above; Squatina sp. 1). However, the German specimen displays a suite of dental differences to all other fossil squatiniforms and extant members,

13 102 S. Klug and J. Kriwet supporting its assignment to a different species (Müller 1989). Age. Unfortunately, the exact provenance of the holotype and single holomorphic specimen of S. baumbergensis is unknown (Siegfried 1954). It is assumed here to have come from the so-called Werksteinbank, which occurs in the upper part of the Bostrychoceras polyplocum Zone (= Baumberge Kalk-Sandstein, untere Baumbergeschichten). The B. polyplocum Zone of the Tethys corresponds to the Baculites macleari Didymoceras stevensoni zones of the North American ammonite biostratigraphical scheme and thus ranges from to Ma. Since the specimen comes from the upper part of the zone, which represents the upper Campanian, it is most parsimonious to assume that its age as Ma. We assume the mean age of S. baumbergensis to be Ma. Squatina sp Squatina sp. Itoigawa et al.: 23, pls 1 2, pl. 3, figs 1 18, text-fig. 4. Remarks. So far, 11 species assigned to Squatina have been described from Neogene deposits, three of which are extant, S. africana, S. californica and S. squatina (e.g. Long 1993; Supplementary Online Material). Squatina subserrata (Münster, 1846), which was originally based on isolated teeth from the Miocene of the Vienna Basin (Austria) under the name Sphyrna subserrata, isthemost commonly cited species, displaying a global distribution in the Miocene. This species evidently represents a wastebasket for incorporation of different taxa with very similar tooth structures. The only cranial and postcranial skeletal elements of a Cenozoic Squatina species, which are thus the youngest records, come from the lower Miocene of Japan and have been referred to as Squatina sp. by Itoigawa et al. (1985). The figured teeth resemble those of S. subserrata and S. prima to some extent. Nevertheless, dental characters for species discrimination are only subtle or absent. Consequently, we suggest refraining from assigning any Miocene Squatina fossil to any specific species pending detailed revisions in the future. Description. Specimen MFM is embedded in a calcareous block. It is incomplete and lacks the whole postcranial portion of the body. The skull and the branchial apparatus are highly disarticulated, missing major parts of the neurocranium, the left Meckel s cartilage, the left ceratohyal, both hyomandibulars and parts of the branchial cartilages (Fig. 7). Two partly preserved vertebrae are visible measuring c cm in diameter. The calcification pattern is not observable. An irregular cartilaginous element is preserved in the upper part of the specimen and potentially presents parts of the neurocranium. The left Meckel s cartilage is well preserved in lingual view, but lacks the most posterior part. It displays a long dental groove where the dental lamina would have been situated. The dental ridge is also easily recognized tapering from the rounded anterior tip of the Meckel s cartilage to the posterior part extending nearly three-quarters of way along the preserved cartilage. The dual articulation surfaces of the upper and lower jaws are partly visible at the posterior dorsal end of the Meckel s cartilage. However, only the articular knob the condyle of the medial quadratomandibular joint is completely preserved. The lateral quadratomandibular joint, built by the articular cotylus in the lower jaw, is only observable by its anteriormost tip. The right palatoquadrate is preserved in lingual view and rotated on its axis. The slender posterior part tapers to a rounded tip showing the conspicuous quadrate process at the dorsal surface. This triangular shaped process is located on the proximal third of the palatoquadrate and lost its suspensory function in angel sharks (de Carvalho et al. 2008). In living squatinids, the eye is positioned medial to this process, lying in between it and the orbital process. In this specimen, the orbital process is also preserved in the anterior part of the upper jaw. Sediment covers the symphysis of the left palatoquadrate and therefore only the posteriormost parts of the dental ridge and dental lamina are visible. The left palatoquadrate is preserved in dorsal view showing the orbital process and on the most anterior tip the dental groove, although this is almost completely covered by the sediment. The left quadrate process is not very well preserved but can be seen close to the posterior end of the right palatoquadrate. While the right ceratohyal is not preserved, nor both hyomandibulae, the left slightly bent ceratohyal is visible in lateral view, displaying the posterior articulation surfaces. The branchial apparatus is disarticulated but some elements could be identified. Three slender and nearly straight pharyngobranchials are preserved as well as a single nearly semicircular hypobranchial. Five comparatively long and slender elements with slightly curved tips could be identified as ceratobranchials. The five preserved epibranchials are slightly shorter. Three of the latter and two ceratobranchials are still articulated. Age. This specimen was recovered from siltstones of the Early Miocene Yamanouchi Member of the Akeyo Formation, Mizunami Group, near Mizunami City in the Gifu Prefecture of Japan. This member is Ma based on fission track data according to Kobayashi (1989). Discussion Phylogenetic implications Identification of stem and crown groups of Squatiniformes as well as their sister group is crucial for establishing node ages for the origin of angel sharks. Surprisingly, however, little effort has been made to distinguish between stem and

14 Node age estimations and the origin of angel sharks 103 Figure 7. Squatina sp. 2 (MFM ) from the lower Miocene of Japan. Abbreviations: artk, articular knob; cb, ceratobranchial; deng, dental groove; denl, dental lamina; denr, dental ridge; eb, epibranchial; hb, hypobranchial; lch, left ceratohyal; lmc, left Meckel s cartilage; lpq, left palatoquadrate; pb, pharyngobranchial; qp, quadrate process; v, vertebra. crown clades when the evolution of neoselachians is analysed. According to Hennig (1950), the crown is a monophyletic assembly consisting of all living members of a group, their common ancestor and all taxa that are nested phylogenetically within this group regardless of whether these are extant or extinct. The stem-group, conversely, is paraphyletic and comprises extinct clades that are more closely related to the crown under consideration than to any other (Donoghue 2005). The members of the stem-group thus are the fossils which form the framework for establishing divergence estimates (Benton et al. 2009). The sister group of Squatiniformes consisting of crown- and stemgroups also provides a possibility for inferring node ages within Squatiniformes. Squatiniformes is represented by numerous living species, which are all placed in the single genus Squatina and family Squatinidae. The systematic position of Squatinidae within Neoselachii has been a major controversy for more than 170 years (e.g. Müller & Henle ; Duméril ; Hasse ; Regan 1906; Goodrich 1909; Garman 1913; Jordan 1923; Iselstöger 1937; White 1937; Moy-Thomas 1939; Holmgren 1941; Melouk 1954; Norman 1966; Compagno 1973, 1977; Maisey 1980, 1984a, b; Shirai 1992a, b; de Carvalho 1996; de Carvalho & Maisey 1996; Maisey et al. 2004; Naylor et al. 2005). Maisey (1980), by employing the concept of shared (homologous) characters, concluded that Squatina is sister to a clade including hexanchiforms, pristiophoriforms and squaliforms based on the presence of a distinct orbital process (= orbitostylic sharks). Compagno (1984) drew a similar conclusion and hypothesized that Squatiniformes is closely related to Pristiophoriformes and Squaliformes based on morphological features. The use of strict cladistic principles and of morphological data changed the systematic arrangement of major living neoselachian clades in several aspects (e.g. Shirai 1992a, b, 1996; de Carvalho 1996; de Carvalho & Maisey 1996). Most importantly, Squatiniformes was found to be sister of Pristiophoriformes and Batoidea, a clade termed Hypnosqualea, with Squaliformes being sister to this clade (Fig. 1A). Molecular data, however, do not support the hypnosqualean hypothesis and place batoids outside the sharks (Douady et al. 2003; Winchell et al. 2004). In the analysis of Douady et al. (2003), Squatiniformes and Pristiophoriformes form a monophyletic grouping, which is sister to the more derived Squaliformes. This arrangement, however, requires rather long ghost lineages if calibrated with the occurrence of fossil representatives. In the analyses of Winchell et al. (2004), conversely, Squatiniformes is sister to a clade consisting of Squaliformes and Pristiophoriformes (Fig. 1B). Thus, three squalomorph clades, Hexanchiformes, Squaliformes and Pristiophoriformes, are thought to be the sister of squatinids by different authors. Each taxon has different impacts on origin estimates for Squatiniformes. Turner & Young (1987) tentatively assigned the genus Mcmurdodus, which is known from the Devonian and Permian, to Hexanchiformes based on noticeable

15 104 S. Klug and J. Kriwet similarities in tooth morphology and tooth root vascularization pattern. However, a multilayered enameloid layer covering the tooth crown is absent, although a parallelfibred layer is developed (e.g. Burrow et al. 2008). Consequently, Mcmurdodus, which is the only genus in the family Mcmurdodontidae, is not regarded as hexanchiform but currently is considered to be a neoselachian of uncertain affinities (Ginter et al. 2010). Therefore the oldest fossil record of Hexanchiformes is from the Early Jurassic (e.g. Thies 1983), whereas the oldest fossil record of Squaliformes is from the Early Cretaceous (e.g. Kriwet & Klug 2008; Klug & Kriwet 2010) and the oldest representative of Pristiophoriformes occurs in the Cenomanian (Cappetta 2006). Thus, identification of the basal sister group of Squatiniformes is not trivial but central if diversification events are used for setting an evolutionary framework. Compagno (1973, 1977) was not able to identify the systematic position of Squatiniformes within Neoselachii, whereas Maisey (1980) identified Squatiniformes to be the most basal member of Squalomorphii. Shirai (1992a, b, 1996) and de Carvalho (1996), conversely, placed Squatiniformes in a clade Hypnosqualea with Squaliformes being the basal sister group to this clade. Douady et al. (2003) found a similar arrangement of squalomorph taxa in their molecular study of neoselachian interrelationships. In a more extensive molecular study of neoselachian phylogenetics, Winchell et al. (2004) identified hexanchiform sharks represented by Chlamydoselachidae to be the basal sister group of Squatiniformes. This arrangement and placement of Squaliformes and Pristiophoriformes as being more derived in comparison to Squatiniformes is in better agreement with the fossil record, reducing the number of ghost lineages of Squatiniformes and more derived squalomorphs considerably (Fig. 1). The most parsimonious hypothesis is that Hexanchiformes represents the basal sister group of Squatiniformes. The relatively large stratigraphical gap between the known first fossil occurrences of hexanchiforms in the Early Jurassic, conversely, renders the strict application of the phylogenetic bracketing approach difficult, because the stem of Hexanchiformes at the moment is not resolved. Thus, the stem group of Squatiniformes turns out to be significant for reducing the ghost lineage implied by the stratigraphical occurrence of the known total group Hexanchiformes. It also enables us to derive more reliable origination dates for squatiniforms although the stem of the sister group remains obscure. Within squatiniforms, two distinct lineages can be recognized, which can be distinguished readily by several morphological characters as they are present in holomorphic specimens of Pseudorhina from the Late Jurassic and skeletal remains of Squatina from the Cretaceous and Cenozoic (de Carvalho et al. 2008; see above). Dental characters, although recognizable, are difficult to evaluate. The morphological traits as displayed by known holomorphic specimens identify Pseudorhina as belonging to the stem of Squatiniformes. Consequently, a new family is introduced for this clade. Within Squatinidae, two different groups might be distinguished as indicated by dental features based on cranial remains of Squatina cranei (Guinot et al. in press). One group displays the characteristic dental architecture of living members of Squatina, whereas some species of Cretaceous age display a different heterodonty pattern. The systematic status of both groups remains unresolved for the moment and is irrelevant for our considerations, because we consider all Cretaceous squatinids represented by skeletal material as members of Squatina. Stem and crown groups of Squatiniformes display different known stratigraphical occurrences. The known fossil record of Pseudorhina ranges from the late Oxfordian (Thies 1983) to the early Tithonian (Kriwet & Klug 2004, 2009; de Carvalho et al. 2008). The oldest record of Cretaceous squatiniforms is from the late Valanginian (Early Cretaceous) and is represented by isolated teeth assigned to Squatina cranei by Rees (2005). Although the teeth bear numerous resemblances to teeth of Cretaceous squatinids, their assignment to Squatina remains dubious pending further analyses of Cretaceous squatiniform dental characters. These teeth (as in other Early Cretaceous records) might represent members of extinct squatinids or even of stem squatinids. The fossil record of squatiniforms seemingly is not continuous across the Jurassic Cretaceous boundary, resulting in a ghost lineage of about 8 10 Ma. Recognition of potential Early Cretaceous occurrences of pseudorhinids, however, is rendered difficult by the fact that all pre-aptian squatiniform records consist of few isolated teeth. The identification of isolated teeth remains problematic because of very similar and conservative morphological features, which have only changed slightly since the Early Cretaceous (e.g. Cappetta 1975; Siverson 1995). The origin of squatinids corresponds to the origin of the stem of Squatiniformes, which is represented by Pseudorhinidae if cladistic principles are employed. This interpretation fits well with previous assumptions that squatinids originated in the Late Jurassic, although the argument was based solely on dental features and conservative taxonomic approaches. It is, however, not possible to provide hard minimum and soft maximum constraints for the divergence of Pseudorhinidae and Squatinidae, which needs a modified approach (see below), because pseudorhinids are probably not the most basal member of the stem-lineage. The origin of Squatiniformes subsequently corresponds to the oldest member of the squatiniform stem in combination with the oldest occurrence of the stem of the sister group, which, however, remains unresolved. Stem taxa are considered here to provide additional evidence towards soft maximum age constraints. The Late Jurassic hexanchiform Notidanoides may represent a member of the stem but