Scholarly article on topic 'Evolutionary relationships and systematics of Atoposauridae (Crocodylomorpha: Neosuchia): implications for the rise of Eusuchia'

Evolutionary relationships and systematics of Atoposauridae (Crocodylomorpha: Neosuchia): implications for the rise of Eusuchia Academic research paper on "Biological sciences"

0
0
Share paper
Academic journal
Zool J Linn Soc
OECD Field of science
Keywords
{""}

Academic research paper on topic "Evolutionary relationships and systematics of Atoposauridae (Crocodylomorpha: Neosuchia): implications for the rise of Eusuchia"

LINNEAN SOCIETY._

of London

ZOOLOGICAL

Journal

of the

Linnean Society

Zoological Journal of the Linnean Society, 2016, 177, 854-936. With 11 figures

Evolutionary relationships and systematics of Atoposauridae (Crocodylomorpha: Neosuchia): implications for the rise of Eusuchia

JONATHAN P. TENNANT1*, PHILIP D. MANNION1 and PAUL UPCHURCH2

1Department of Earth Science and Engineering, Imperial College London, South Kensington, London SW7 2AZ, UK

2Department of Earth Sciences, University College London, Gower Street, London WC1E 6BT, UK Received 18 August 2015; revised 5 January 2016; accepted for publication 19 January 2016

Atoposaurids are a group of small-bodied, extinct crocodyliforms, regarded as an important component of Jurassic and Cretaceous Laurasian semi-aquatic ecosystems. Despite the group being known for over 150 years, the taxonomic composition of Atoposauridae and its position within Crocodyliformes are unresolved. Uncertainty revolves around their placement within Neosuchia, in which they have been found to occupy a range of positions from the most basal neosuchian clade to more crownward eusuchians. This problem stems from a lack of adequate taxonomic treatment of specimens assigned to Atoposauridae, and key taxa such as Theriosuchus have become taxonomic 'waste baskets'. Here, we incorporate all putative atoposaurid species into a new phylogenetic data matrix comprising 24 taxa scored for 329 characters. Many of our characters are heavily revised or novel to this study, and several ingroup taxa have never previously been included in a phylogenetic analysis. Parsimony and Bayesian approaches both recover Atoposauridae as a basal clade within Neosuchia, more stemward than coelognathosuchians, bernissartiids, and paralligatorids. Atoposauridae is a much more exclusive clade than previously recognized, comprising just three genera (Alligatorellus, Alligatorium, and Atoposaurus) that were restricted to the Late Jurassic of western Europe, and went extinct at the Jurassic/Cretaceous boundary. A putative Gondwanan atoposaurid (Brillanceausuchus) is recovered as a paralligatorid. Our results exclude both Montsecosuchus and Theriosuchus from Atoposauridae. Theriosuchus is polyphyletic, forming two groupings of advanced neosuchians. Theriosuchus (restricted to Theriosuchus pusillus, Theriosuchus guimarotae, and Theriosuchus grandinaris) spanned the Middle Jurassic to early Late Cretaceous, and is known from Eurasia and North Africa. Two Cretaceous species previously assigned to Theriosuchus ('Theriosuchus' ibericus and 'Theriosuchus' sympiestodon) are shown to be nested within Paralligatoridae, and we assign them to the new genus Sabresuchus. The revised phylogenetic placement of Theriosuchus has several implications for our understanding of eusuchian evolution. Firstly, the presence of fully pterygoidean choanae, previously regarded as a defining characteristic of Eusuchia, is not found in some basal members of Eusuchia. However, eusuchians can be distinguished from Theriosuchus and other basal neosuchians in that their choanae are posteriorly positioned, with an anterior margin medial to the posterior edge of the suborbital fenestra. This feature distinguishes eusuchians from Theriosuchus and more basal neosuchians. Secondly, our refined understanding of Theriosuchus implies that this taxon possessed only amphicoelous presacral vertebrae, and therefore fully developed vertebral procoely is likely to have evolved only once in Crocodylomorpha, on the lineage leading to Eusuchia. These and other findings presented herein will provide an important framework for understanding the neosuchian-eusuchian transition.

© 2016 The Authors. Zoological Journal of the Linnean Society published by John Wiley & Sons Ltd on behalf of The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 177: 854-936 doi: 10.1111/zoj.12400

ADDITIONAL KEYWORDS: Bayesian - Cretaceous - Crocodiles - Crocodyliformes - Implied Weighting -Jurassic - Laurasia - Parsimony.

*E-mail: jon.tennant.2@gmail.com

854 © 2016 The Authors. Zoological Journal of the Linnean Society published by John Wiley & Sons Ltd on behalf of

The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 177, 854-936 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution

and reproduction in any medium, provided the original work is properly cited.

INTRODUCTION

Atoposaurids comprise a clade of extinct neosuchian crocodyliforms, often characterized by their differentiated dentition and diminutive body size (Owen, 1879; Joffe, 1967; Buscalioni & Sanz, 1990a). This group has a long history of study, with specimens first identified from Late Jurassic deposits in France and Germany in the mid-19th century (von Meyer, 1850, 1851). The current view is that atoposaurids were an important and diverse component of Eurasian Late Jurassic and Cretaceous terrestrial to semi-aquatic ecosystems, often with multiple sym-patric lineages (e.g. Wellnhofer, 1971; Thies, Windolf & Mudroch, 1997; Martin, Rabi & Csiki, 2010; Lau-prasert etal., 2011; Tennant & Mannion, 2014). A number of discoveries indicate that atoposaurids might also have been present in the Jurassic and Cretaceous of Africa (Michard et al., 1990; Flynn et al., 2006; Haddoumi et al., 2016) and North America (e.g. Cifelli etal., 1999; Eaton etal., 1999; Fiorillo, 1999).

Taxonomic composition

Despite their long history of study, the taxonomic composition of Atoposauridae remains uncertain and many putative atoposaurid species have never been incorporated into a phylogenetic analysis. In an extensive revision of the taxonomy of atoposaurids, Wellnhofer (1971) recognized three genera from the Late Jurassic of continental western Europe (Alliga-torellus, Alligatorium, and Atoposaurus), and followed previous authors by including Theriosuchus from the UK (Joffe, 1967), Shantungosuchus from China (Young, 1961), and Hoplosuchus from North America (Gilmore, 1926), in Atoposauridae (see also Steel, 1973). Alligatorium comprises the type species Alligatorium meyeri (Gervais, 1871), and Wellnhofer (1971) also considered the referred species Alligatorium depereti (Vidal, 1915), Alligatorium francon-icum (Ammon, 1906), and Alligatorium paintenense (Kuhn, 1961) to be valid, although all specimens of the latter two species were lost or destroyed during the Second World War. Subsequently, the Spanish species Alligatorium depereti was considered to be distinct enough to warrant its own genus, Montseco-suchus (Buscalioni & Sanz, 1988, 1990a,b). Efimov (1976) described the putative atoposaurid Karatausu-chus from Kazakhstan, but Buscalioni & Sanz (1988) subsequently removed this taxon, as well as Hoplosuchus and Shantungosuchus, from Atoposauridae. Benton & Clark (1988) considered only Alligatorium to be valid amongst atoposaurid taxa from the Late Jurassic of France and Germany, regarding Alliga-torellus and Atoposaurus as juvenile individuals of

this taxon. Benton & Clark (1988) also followed Joffe (1967) in considering Theriosuchus to represent an atoposaurid. Buscalioni & Sanz (1990a,b) largely followed Benton & Clark (1988), also accepting Alliga-torellus, but not Atoposaurus, as valid. Most recently, Tennant & Mannion (2014) argued that Alligatorellus, Alligatorium, and Atoposaurus all represent valid atoposaurid genera, with each comprising two species. These authors also suggested that Alligatorium paintenense is likely to be a junior synonym of Alligatorium franconicum. In addition to the type species of Theriosuchus (Theriosuchus pusillus; Owen, 1878a, 1879), four additional species have subsequently been named: Theriosuchus guimarotae (Schwarz & Salisbury, 2005), Theriosuchus grandi-naris (Lauprasert et al., 2011), Theriosuchus ibericus (Brinkmann, 1989), and Theriosuchus sympiestodon (Martin et al., 2010, 2014a). Some authors have further taken the view that Theriosuchus is distinct enough from all other atoposaurids to constitute its own clade, Theriosuchidae (Kalin, 1955; Buffetaut, 1982, 1983), but this taxonomic assignment has not been widely adopted. Two additional taxa have also been referred to Atoposauridae, with Michard et al. (1990) describing the first putative Gondwanan ato-posaurid (Brillanceausuchus babouriensis) from Cameroon, and Pachycheilosuchus trinquei described from North America (Rogers, 2003). However, subsequent studies have placed Pachycheilosuchus outside of Atoposauridae, and it is likely to be a member of Hylaeochampsidae (e.g. Buscalioni et al., 2011).

Atoposaurids in time and space

Based on our current understanding of Atoposauri-dae, the oldest diagnostic remains are: (1) a partial dentary from the Middle Jurassic (late Bajocian-Bathonian) of the Isle of Skye, UK (Young et al., 2016), ascribed to Theriosuchus sp.; (2) isolated tooth crowns from the late Bathonian of France and the UK (Evans & Milner, 1994; Kriwet, Rauhut & Gloy, 1997; Knoll et al., 2013); (3) crocodyliform teeth, possibly referable to an atoposaurid, from the Bathonian Grand Causses of France (Knoll et al., 2013; Knoll & Lopez-Antonanzas, 2014); and (4) teeth and mandibular and postcranial remains from the Batho-nian of Madagascar (Flynn et al. , 2006) and Morocco (Haddoumi et al., 2016) (Fig. 1). These remains indicate that atoposaurids had attained their characteristic small body size and heterodont dentition, along with a broad geographical distribution, by the Middle Jurassic.

In addition to the presence of Alligatorellus, Alligatorium, Atoposaurus, and Montsecosuchus in the Late Jurassic and Early Cretaceous of western Europe, Theriosuchus is known from a number of Late

Figure 1. Stratigraphic ranges for taxa previously attributed to Atoposauridae. The dashed lines represent the inferred presence of lineages.

Jurassic and Cretaceous localities in Europe (e.g. Owen, 1879; Buscalioni & Sanz, 1984, 1987a,b; Bus-calioni, 1986; Salisbury, 2002; Martin et al., 2010, 2014a; Salisbury & Naish, 2011; Tennant & Mannion, 2014; Young et al., 2016). Reports based on isolated teeth also place Theriosuchus in the middle Cretaceous of North America (Pomes, 1990; Winkler etal., 1990; Cifelli et al., 1999; Eaton et al., 1999; Fiorillo, 1999), alongside the putative atoposaurid Pachycheilosuchus (Rogers, 2003). Remains of The-riosuchus are also found in the latest Jurassic to Early Cretaceous of Thailand (Lauprasert et al., 2011) and China (Wu, Brinkmann & Russell, 1996). In addition, Brillanceausuchus from western Africa (Michard et al. , 1990) is the only putative Cretaceous Gondwanan occurrence of Atoposauridae, along with the fragmentary and poorly known Middle Jurassic remains mentioned by Flynn et al. (2006) and Had-doumi et al. (2016). Finally, fragmentary putative atoposaurid remains from the Eocene of Yemen would mean that Atoposauridae passed through the end-Cretaceous mass extinction (Stevens et al., 2013).

Evolutionary relationships

The monophyly ofAtoposauridae has not been tested at a low taxonomic level since Buscalioni & Sanz (1990a), and even species attributed to Theriosuchus have not been conclusively demonstrated to form a monophyletic genus. This is partly because of the taphonomy and preservation of these specimens, whereby incompleteness and the mode of preservation (i.e. dorsal flattening) restricts assessment of important characters. Furthermore, the generally small body size of ato-posaurids has led to an overall lack of clarity in distinguishing amongst plesiomorphic, juvenile, and paedomorphic characteristics (Joffe, 1967; Buffetaut, 1982; Clark, 1986; Buscalioni & Sanz, 1988; Tennant & Mannion, 2014), although the ontogeny of Theriosuchus is reasonably well understood (Joffe, 1967; Schwarz & Salisbury, 2005; Martin et al., 2014a).

Most phylogenetic analyses recover atoposaurids as non-eusuchian neosuchians, part of the important crocodyliform lineage that includes living crocodiles (e.g. Benton & Clark, 1988; Buscalioni & Sanz, 1990a,b; Salisbury etal., 2006; Brochu etal., 2009;

Pol & Gasparini, 2009; Martin etal., 2010; Adams, 2013, 2014; Sertich & O'Connor, 2014). Whereas some analyses have found Atoposauridae to be outside of Neosuchia (e.g. Sereno et al., 2003), this has not gained support from more recent studies. Recent analyses consider atoposaurids to be within Neo-suchia, but their position differs greatly, varying amongst: (1) basal to Goniopholididae and other neo-suchians when investigating higher neosuchian or eusuchian relationships (e.g. Pol & Norell, 2004a; Gasparini, Pol & Spalletti, 2006; Turner, 2006; Fortier & Schultz, 2009; Pol, Turner & Norell, 2009; Figueiredo et al., 2011; Adams, 2014) (Fig. 2A); (2) in an uncertain position within basal Neosuchia (Pol & Apesteguia, 2005; Larsson & Sues, 2007; Turner & Buckley, 2008; Lauprasert et al., 2009; Martin et al., 2010); (3) just outside of Eusuchia (Rogers, 2003; Salisbury etal., 2006; Fig. 2B); or (4) as the sister group to Paralligatoridae within Eusuchia (Fig. 2C) that, together with Hylaeochampsidae, comprises the sister group to crown Crocodylia (Turner, 2015; Turner & Pritchard, 2015).

The first analysis of atoposaurid inter-relationships recovered two subclades comprising (Montseco-suchus + Theriosuchus) and (Alligatorium + Alligatorellus) (Buscalioni & Sanz, 1988; Fig. 3A), although the position of Montsecosuchus was unstable. Karl et al. (2006) recovered the same topology, but also included Atoposaurus, which they placed as the most basal atoposaurid. However, these two studies predated the identification of several new species of Theriosuchus. Bronzati, Montefeltro & Langer (2012) constructed a crocodyliform supertree that included Alligatorium, Alligatorellus, Atoposaurus, Montseco-suchus, Pachycheilosuchus, and T. guimarotae, T. pusillus, and T. sympiestodon, placing them all within Atoposauridae, as the sister group to Gonio-pholididae and more advanced neosuchians (Fig. 3B). However, they were unable to fully resolve the internal relationships of the group beyond finding Alliga-torellus and Atoposaurus to be sister taxa, and that the three Theriosuchus species formed a clade. Turner (2015) included T. guimarotae, T. pusillus, and T. sympiestodon, finding them to be paraphyletic with respect to Alligatorium (Fig. 3C).

Consequently, the full plethora of putative ato-posaurid species has never previously been included in any phylogenetic analysis. In studies that have included atoposaurids, Montsecosuchus and the putative atoposaurids Brillanceausuchus and Karatausu-chus have been almost completely disregarded. As a result, neither the phylogenetic position of Atoposauridae within Neosuchia, nor its intrarela-tionships are clear at present.

Here, we undertake a full systematic reassessment of all species previously assigned to Atoposauridae,

determining the composition and internal relationships of the group, as well as its position within Neo-suchia. We present a new phylogenetic character matrix, analysed using parsimony and Bayesian approaches. Revised diagnoses are provided for all genera and species assigned to Atoposauridae, as well as the first phylogenetic definition for the clade, and we discuss the taxonomic and phylogenetic status of putative atoposaurids. Lastly, we discuss the implications of our results for the evolution of Eusuchia.

Institutional abbreviations

AMNH, American Museum of Natural History, New York City, New York, USA; BSPG, Bayerische Staatssammlung fur Paläontologie und Geologie, Munich, Germany; CHE, Cherves-de-Cognac Collection, Musée d'Angouleme, France; CM, Carnegie Museum, Pittsburgh, Pennsylvania, USA; DFMMh, Dinosaurier-Freilichtmuseum Muänchehagen, Germany; FGGUB, Faculty of Geology and Geophysics, University of Bucharest, Bucharest, Romania (LPB, Laboratory of Paleontology); GZG, Geowis-senschaftliches Museum, Georg-August-Universitaät, Gäottingen, Germany (BA, Max Ballerstedt collection; STR, stratigraphic collection); IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Beijing, China; MCDRD, Muzeul Civilizatxiei Dacice si Romane, Deva, Romania; MfN, Museum fur Naturkunde, Berlin, Germany (MB.R, Reptile Collection); MGB, Museo de Geologia del Ayuntamiento de Barcelona, Barcelona, Spain; MHNL,* Muséum d'Histoire Naturelle, Lyon, France; MO, Musee de l'île d'Oleéron, Saint-Pierre-d'Oléeron, France; MTM, Hungarian Natural History Museum, Budapest, Hungary; NHMUK, Natural History Museum, London, UK; NMS, National Museums Scotland, Edinburgh, UK; PIN, Paleontological Institute, Russian Academy of Sciences, Moscow, Russia; PMC, Laboratoire de Paleéontologie des Vertéebreés de l'Universiteé Pierre et Marie Curie; PRC, Palaeontological Research and Education Centre, Mahasarakham University, Thailand; SMU, Southern Methodist University, Dallas, Texas, USA; TM, Teyler's Museum, Haarlem, Netherlands; UP, Université de Poitiers, Poitiers, France; UT, University of Texas, Austin, Texas, USA.

MATERIAL AND METHODS

Taxon sampling

All previously identified atoposaurid species were included (Table 1), with the exception of Karatausu-chus sharovi (Efimov, 1976), which we were not able

Figure 2. Previously recovered inter-relationships between Atoposauridae and other major crocodyliform clades: (A) Adams (2014; (B) Rogers (2003); (C) Turner & Pritchard (2015).

to observe directly, and for which there is insufficient morphological data published to adequately score it from the literature. We also excluded

Shantungosuchus chuhsienensis (Young, 1961) for similar reasons, and because this taxon is likely to be either a protosuchian-grade crocodyliform

Figure 3. Previously recovered intrarelationships within Atoposauridae. (A) Buscalioni & Sanz (1988); (B) Bronzati et al. (2012); (C) Turner (2015).

(Wu, Brinkmann & Lu, 1994) or member of Shartegosuchidae (Clark, 2011). All personal observations of specimens were made by J. P. T. Measurements and ratios of key morphological characteristics are provided in Tables 2 and 3, respectively. For specimens not personally observed, measurements were acquired using reported values and via ImageJ (http://imagej.nih.gov/ij/; Schneider, Rasband & Eliceiri, 2012) from published photographs of specimens. Most measurements and ratios provided are given to two decimal places and based on the holotype specimens, as opposed to multiple specimens. As such, we do not provide ratio ranges for species. In all cases, measurements represent the maximum distance for each element measured between the proximal-most and distal-most points.

We included both species of Atoposaurus (Ato-posaurus jourdani and Atoposaurus oberndorferi) and Alligatorellus (Alligatorellus bavaricus and

Alligatorellus beaumonti), as well as Alligatorium meyeri, and Alligatorium franconicum was scored based on figures and illustrations in Wellnhofer (1971). We included all species of Theriosuchus, with T. pusillus, T. ibericus, and T. sympiestodon (supplemented by new material described by Martin et al., 2014a) and scored T. guimarotae and T. grandinaris based on Schwarz & Salisbury (2005) and Lauprasert et al. (2011), respectively. Scoring for Montseco-suchus depereti was based on the holotype specimen, and specimens described as Alligatorellus sp. (MB.R.3632, from the Late Jurassic of Germany; Schwarz-Wings et al., 2011), and Theriosuchus sp. (NMS G. 2014.52.1, from the Middle Jurassic of the Isle of Skye, UK; Young et al., 2016) were both incorporated to test their generic assignment. In addition, Brillanceausuchus babouriensis (Michard et al., 1990) and Pachycheilosuchus trinquei (Rogers, 2003) were also included as previously identified putative atoposaurids.

Table 1. Details of the atoposaurid and non-atoposaurid taxa used for the present study

Numerical age

(Mya) Distribution

References/sources

Personally

observed This study

Parent taxon

5= y ® N

§ <5.

<X> "-j

o st, p s

%% o 5.

b T2 e

a cr a p

a" o § ^

c? ® S-go

-J o -J p

Gl I-+1

Ingroups AUiga.torium

meyeri AUiga.torium franconicum Atoposa.urus jourdani Atoposa.urus oberndorferi Alliga.torellus ba. various

Alliga.torellus bea.umonti

Alliga.torellus sp.

Montsecosuchus

depereti Theriosuchus pusillus Theriosuchus guima.rotae Theriosuchus grandinaris Theriosuchus

ibericus Theriosuchus

sympiestodon Theriosuchus sp.

Brilla.nceausuchus

ba.bouriensis Outgroups Pa.chycheilosuchus

trinquei Wa.nncha.mpsus kirpa.chi

Kimmeridgian

Kimmeridgian Kimmeridgian

Early Tithonian

Early Tithonian

Kimmeridgian

Early Tithonian

157.3-152.1

154.75-152.1

157.3-152.1

152.1-148.55

152.1-148.55

157.3-152.1

152.1-148.55

Late Berriasian- 142.4-136.35 Early Valanginian Berriasian 145-139.8

Kimmeridgian

Early Aptian

Barremian

Maastrichtian

Late Bajocian-Bathonian Barremian

157.3-152.1 125-118

129.4-125 72.1-66 169.3-166.1 145-139.8

Early Albian 113-106.75

Aptian 125-114

Cerin, France

Kelheim, Germany Cerin, France

Kelheim,

Germany Eichstätt, Germany

Cerin, France

Kelheim,

Germany Montsec, Spain

Dorset, England

Guimarota, Portugal Sakon Nakhon, Thailand Serrania de Cuenca, Spain Hateg, Romania

Isle of Skye,

Scotland Babouri-Figuil, Cameroon

Texas, USA

Texas, USA

Wellnhofer (1971) Yes

Wellnhofer (1971) No

Wellnhofer (1971) Yes

Wellnhofer (1971) Yes

Wellnhofer (1971), Yes Tennant & Mannion (2014)

Wellnhofer (1971), Yes Tennant & Mannion (2014)

Schwarz-Wings et a.l. Yes (2011)

Buscalioni & Sanz Yes (1990a)

Owen (1879), Clark Yes (1986)

Schwarz & Salisbury No (2005)

Lauprasert et a.l. (2011) No

Brinkmann (1989) Yes

Martin et al. (2010) Yes

Young et al. (2016) No

Michard et a.l. (1990) Yes

Rogers (2003) Yes

Adams (2014) Yes

AUiga.torium

meyeri AUiga.torium' franconicum Atoposa.urus jourdani Atoposa.urus oberndorferi Alliga.torellus ba.varicus

Alliga.torellus bea.umonti

Alliga.torellus sp.

Montsecosuchus

depereti Theriosuchus pusillus Theriosuchus guima.rotae Theriosuchus grandinaris Sa.bresuchus

ibericus Sa.bresuchus

sympiestodon Theriosuchus sp.

Brilla.nceausuchus ba.bouriensis

Pa.chycheilosuchus

trinquei Wa.nncha.mpsus kirpa.chi

Atoposauridae

Neosuchia

Atoposauridae

Atoposauridae

Atoposauridae

Atoposauridae

Atoposauridae

Neosuchia

Neosuchia

Neosuchia

Neosuchia

Paralligatoridae

Paralligatoridae

Neosuchia

Paralligatoridae

Hylaeochampsidae Paralligatoridae

й ¿3

(В 2?

й ТЗ

й :й

о "Й

■и Я

и ш Рч И

й тЗ

CÜ й

'С й

й PL,

2 Ö Ö

-<е о

2 "sS ".es

-С "а ><е

ИЗ ö Kl ^

о Й г к а s

w tÜ2 -e

"i ei ^ ö

^ «5 "2 «.i У о ^ Ü

8 w ö ^ J S -£2 Ii о -2 2 а

-26 -g S ^ ö

£ ■e £ ^ ^

¡я 25

о о о

со u

^ Чй

(35 ш

d- Й Й

И (В

CD СО

(М (35

й со

о PL,

й "В -и <; м " со

^ р со

Й 6Ъ

3 Л ^ > 2 о И О

^ ' .is й И й

5ч (й1д щ а ° £ °

з W Й н <= £ °

со р

Й Й о

м S & "о

са js ВД м

S м si S S й я «

^ и В S со О

Ю Ю

со Ю Ю I—1 со

!> CUD Ю CUD (35

Ю ^ <N ^ Ю (35

i—1 I—1 i—1 I—1 I—1 i—1

Ki I—1 со CO со

со (35 !> !> 1-4

en Ю <N Ю Ю О

i—i I—1 i—I I—1 I—1 <N

ii Й Jy

и и й й Н PQ

й 13 й .tg i3 -й

^ ,-ti ТЗ

•¡3

й н ®

'ЕЪ й й

'(3 й а S3 й о

"iS к

-<е о

й Ii •й

а о .ts -с "а ><е

2 о (5 - ^

£2 I о

а о s

tu J? -е

Тз ^ .5

ES s: О й 2 У я я G

ü s -ц

We selected several basal 'protosuchian'-grade taxa, and a range of neosuchian taxa for outgroups. We incorporated Protosuchus richardsoni (Colbert & Mook, 1951) and Hoplosuchus kayi scored from (Gil-more (1926) as protosuchians. Within Neosuchia, we selected the goniopholidids Amphicotylus lucasii scored from Mook (1942) and Eutretauranosuchus delfsi, scored from Smith et al. (2010) and Pritchard et al. (2013), as these both preserve highly complete cranial material. In addition to these goniopholids, we also included Pholidosaurus purbeckensis (Salisbury, 2002) as a further representative of 'coelog-nathosuchian' crocodyliforms (sensu Martin et al. , 2014b). We included Koumpiodontosuchus aprosdok-iti to represent Bernissartiidae (Sweetman, Pedreira-Segade & Vidovic, 2015). We also included the advanced neosuchians Wannchampsus kirpachi (Adams, 2014) and Shamosuchus djadochtaensis, based on Pol et al. (2009), both of which are likely to belong to Paralligatoridae (e.g. Turner, 2015). Proto-suchus was constrained as the ultimate outgroup taxon in each analysis.

Data matrix

We constructed a new character matrix (Appendix S1) based on a range of primary sources, with the majority of characters derived from Clark (1994), Ortega et al. (2000), Pol et al. (2009), and Andrade et al. (2011). Ninety-two novel characters were also incorporated, following an extensive review of the literature, as well as via personal observations of specimens. Some of these were created by the splitting of previous characters. We formatted all characters to a standardized notation, and many characters were revised, quantified, and/or clarified to remove ambiguity, including removal of problematic gaps between plesiomorphic and derived character states (see Appendix 1 and Appendix S2). Our final data set comprises 329 characters (including autapomorphies - see 'Bayesian inference' below) scored for 24 operational taxonomic units (OTUs; 15 ingroup and nine outgroup taxa; Table 1). As with the majority of fossil crocodyliform data matrices, ours is dominated by cranial, mandibular, and dental characters (263), augmented with 16 axial, 24 appen-dicular, and 26 osteoderm characters (Appendix 1). We opted to use a reductive (contingent) coding approach, which treats non-applicable character states as missing data when there is no logical basis for interpreting the character for any given OTU (Strong & Lipscomb, 1999). The advantage of this approach is that it facilitates the capture of grouping information between successive transformations between particular characters and state values (Brazeau, 2011).

Phylogenetic analysis Parsimony analysis

We used TNT v. 1.1 (Goloboff, Farris & Nixon, 2000) to perform a series of parsimony analyses. Importantly, we wanted to test the effect of removal of taxa and combinations of taxa to test the stability of resulting topologies. We treated 47 multistate characters as ordered (additive; Appendix 1). Starting with a random seed, we employed 50 iterations of a ratchet search strategy, which is a repeated pseudo-sampling protocol that uses character reweighting to search tree space more effectively. No more than 20 substitutions were accepted during each phase of perturbation, and we did not auto-constrain cycles. An equal probability was used for both up-weighting and down-weighting of characters in each cycle. The ratchet search function uses the tree bisection and reconnection (TBR) branch swapping algorithm to search for the most parsimonious trees (MPTs), which we repeated 1000 times. The MPTs were then subjected to a final exhaustive search for all remaining topologies of equal length, again using the TBR algorithm. All trees reported are the strict consensus topologies of all MPTs for each analysis, and zero-length branches were collapsed by default. We also calculated the absolute Bremer branch support value (or decay index) for each node, which is a measure of the extra number of steps required to collapse a branch in the consensus topology (Bremer, 1994). We performed an additional analysis (with all OTUs included) utilizing implied weighting, using a weighting exponent (k) of 3, as a method for favouring characters that are more likely to be homologous and penalizing those more likely to be homoplastic and therefore producing a more 'reliable' topology (Goloboff, 1993; Goloboff et al., 2003). Finally, to test for the effects of unstable taxa or characters, we employed the iterpcr script of Pol & Escapa (2009).

Bayesian inference

In addition, we used Bayesian inference to test for topological congruence with our parsimony results using a different methodology, following the approach described by Lewis (2001). We used MrBayes v. 3.2.5 (Ronquist & Huelsenbeck, 2003), set to perform 10 000 000 generations running four simultaneous Markov chains, sampling every 1000 chains, and setting a burn-in fraction of 0.25. The Markov Chain process started at a random seed, and fixed the states and rate frequencies to vary with an equal probability. Our data matrix includes characters with states that are locally, ambiguously, or unambiguously resolved as autapomorphic characters. These are not informative for our parsimony

analyses, in which it is the shortest number of character state transformations (steps) leading to clades based on synapomorphies that is most important, but they can have the effect of increasing terminal branch lengths for trees obtained using Bayesian inference (Lewis, 2001). We elected to include autapomorphies in agreement with Muller & Reisz (2006), amongst others, who suggested that inclusion of all available data is important for yielding new insights, as well as having an effect on deeper node support values. Ordered characters were treated in the same way as for the parsimony analyses, using the 'ctype ordered:' command.

RESULTS

Parsimony analyses Unordered analysis

A complete analysis involving all OTUs and all characters defined as unordered resulted in seven MPTs, with a total length of 802 steps (Fig. 4A). In this topology, Atoposauridae comprises all species of Alligato-rium, Alligatorellus, and Atoposaurus, with the two species of Alligatorium (Alligatorium meyeri and Alli-gatorium franconicum) occupying an unresolved basal position. Atoposaurus and Alligatorellus are sister taxa, with each comprising two constituent species. Atoposauridae is in a more stemward position than a clade comprising Koumpiodontosuchus and coelog-nathosuchians. MB.R.3632 (Schwarz-Wings et al., 2011) does not group with other Alligatorellus species, supporting the conclusions of Tennant & Mannion (2014), but instead clusters with Montsecosuchus, forming a basal clade with Pachycheilosuchus. Theriosuchus is resolved as polyphyletic, with a clade of (T. ibericus + T. sympiestodon) nested within Paralligatoridae, along with Brillanceausuchus, Shamosuchus, and Wannchampsus. The remaining Theriosuchus species fall outside of this clade, with (T. guimarotae + T. pusillus) and (T. grandi-naris + Theriosuchus sp.) forming clades.

Ordered analyses

When all taxa are included and 47 characters are treated as ordered (Appendix 1), we recover a largely unresolved polytomy (Fig. 4B) for the strict consensus of 11 MPTs of length 830 steps. Theriosuchus remains polyphyletic, forming the same clades as in the unordered analysis, and Atoposaurus and Alligatorellus are monophyletic genera.

The iterpcr function of Pol & Escapa (2009) found Alligatorellus' sp. (MB.R.3632) and Alligatorium franconicum to be the most unstable taxa. The instability of the former taxon is most likely to be a result of a high proportion of missing data, whereas for the

Table 2. Primary cranial and postcranial measurements and counts for all OTUs analysed. All measurements in millimeters

Taxon Total length Tail length Skull length Skull width Snout length Orbit length Orbit width Supratemporal fenestra length Supratemporal fenestra width Symphysis length Intermandibular angle

"Alligatorellus" sp. NA NA NA NA NA NA NA NA NA NA NA

Alligatorellus bavaricus 288 165 40.9 18.5 13.7 13.1 9.3 5.7 4 7 45

Alligatorellus beaumonti 193.4 165 35.5 17.2 13.6 9.2 8 5.7 3.6 7 43

Alligatorium franconicum NA NA 77 27.8 42.6 14.6 11 11.4 5.4 NA 36

Alligatorium meyeri NA NA 65.6 29 25.3 18 12.5 10.2 6.1 NA 45

Amphicotylus lucasii NA NA 480 288 320 48 40 56 40 NA 34

Atoposaurus jourdani 188 111 20 15.6 8.3 8.6 5.8 NA NA NA 55

Atoposaurus oberndorferi 130 96 26 13.1 NA 9.1 5.6 NA NA 4 NA

Brillanceausuchus babourensis 800 NA 75 42 34 11 12 14 6 NA 42

Eutretauranosuchus delfsi NA NA 416.6 211.9 251.8 58 39.9 53.2 31.8 NA 33

Hoplosuchus kayi NA NA 31 23 9.5 10 6.8 NA NA 5 45

Karatausuchus sharovi 186 116.4 26.9 NA 10.4 8 5.6 3.2 1.7 NA NA

Koumpiodontosuchus aprosdokii NA NA 112 55 70 20 15 14 10 15 32

Montsecosuchus depereti NA NA 53.4 29.7 20 13.2 9.3 6 5.7 6 61

Pachycheilosuchus trinquei NA NA NA NA NA NA NA NA NA 6 NA

Pholidosaurus purbeckensis NA NA NA 117 NA 33 33 41.7 41.7 NA NA

Protosuchus richardsoni NA 133 113 86 44 25 5 17 14 NA 48

Shamosuchus djadochtaensis NA NA 128 86 64 21 18 20 15 NA 42

Theriosuchus sp. NA NA NA NA NA NA NA NA NA 12 NA

Theriosuchus grandinaris NA NA NA NA NA NA NA NA NA 12 NA

Theriosuchus guimarotae NA NA 74.2 40.8 34.9 14.9 13.6 11.4 8.8 4 44

Theriosuchus ibericus NA NA NA NA NA NA NA NA NA 12 NA

Theriosuchus pusillus NA NA 84.3 55.7 35.7 22 16.1 12.9 11.6 12 45

Theriosuchus sympiestodon NA NA NA NA NA NA NA 12 10 11 NA

Wannchampsus kirpachi NA NA 64 44.8 24.6 17.2 16.2 8.5 8.7 17.3 40

Radiale Manus Forelimb Ilium Pubis Ischium Femur

Taxon length MC I MC II MC III MC IV MC V length length length length length length

"Alligatorellus" sp. 15.7 11.1 11.2 8.6 10.7 5.5 NA NA NA NA NA 74.7

Alligatorellus bavaricus 5 NA NA NA NA NA 21.5 63.1 12 10.2 10.8 27.6

Alligatorellus beaumonti 6.5 3.8 4.5 4.6 4.1 3 19 56.1 11 9.3 NA 26.3

Alligatorium franconicum 6.8 6.8 8.1 7.7 5.5 NA 31.7 94.1 22 24 23 42.3

Alligatorium meyeri 10.3 4.9 6.6 6.7 6.5 4.2 29.5 NA NA NA NA 42

Amphicotylus lucasii NA NA NA NA NA NA NA NA NA NA NA NA

Atoposaurus jourdani 3.8 4.1 4.8 5.7 4.4 NA 10.2 31.3 NA NA NA 16.7

Atoposaurus oberndorferi 3 2.2 2.5 2.6 2.6 1.6 10.5 39.1 NA 5 NA 17.7

Brillanceausuchus babourensis 9 NA 9 11 9 5 35 123 NA NA NA 45

Eutretauranosuchus delfsi NA NA NA NA NA NA NA NA NA NA NA NA

Hoplosuchus kayi NA NA NA NA NA NA NA NA NA NA NA 25

Karatausuchus sharovi NA NA NA NA NA NA NA NA NA NA NA 12.1

Koumpiodontosuchus aprosdokii NA NA NA NA NA NA NA NA NA NA NA NA

Montsecosuchus depereti 7.5 4.3 5.3 5.3 5 0 23.3 71.5 20.1 NA NA 38.5

Pachycheilosuchus trinquei 11.1 NA NA NA NA NA NA NA 40.9 38.5 35.6 75.1

Pholidosaurus purbeckensis NA NA NA NA NA NA NA NA NA NA NA NA

Protosuchus richardsoni 14 11 11 12 11 8 52 170 51 56 30 100

Shamosuchus djadochtaensis NA NA NA NA NA NA NA NA NA NA 48 NA

Theriosuchus sp. NA NA NA NA NA NA NA NA NA NA NA NA

Theriosuchus grandinaris NA NA NA NA NA NA NA NA NA NA NA NA

Theriosuchus guimarotae NA NA NA NA NA NA NA NA 13.7 NA 20 25.2

Theriosuchus ibericus NA NA NA NA NA NA NA NA NA NA NA NA

Theriosuchus pusillus 10.3 NA NA NA NA NA NA NA NA NA NA 53.3

Theriosuchus sympiestodon NA NA NA NA NA NA NA NA NA NA NA NA

Wannchampsus kirpachi NA NA NA NA NA NA NA NA NA NA NA NA

NA, non-applicable.

Caudal

Premaxillary Maxillary Cervical Dorsal Sacral verte- Coracoid Scapula Humerus Ulna Radius Ulnare

teeth teeth vertebrae vertebrae vertebrae brae length length length length length length

NA NA NA NA NA NA NA NA NA NA NA NA

NA NA 7 15 3 40 NA 9.3 24.6 17 16.9 5

NA 9 7 17 2 40 5.8 13 19.6 17.5 16.7 5.5

NA NA 7 15 2 NA NA NA 35 27.4 24 5.4

NA 11 7 17 NA NA 8.9 NA NA 29.9 28.9 8.7

5 15 NA NA NA NA NA NA NA NA NA NA

NA NA 6 17 2 50 2.2 4.2 11.2 9.9 9.5 5.3

NA NA 7 16 2 50 NA 7 15.6 13 13.1 3

NA 15 7 19 NA NA NA 24 50 38 38 10

NA 17 NA NA NA NA NA NA NA NA NA NA

NA 14 11 13 2 NA NA 12.8 19.3 17.5 17.5 NA

NA NA 8 NA NA 46 NA NA 8.8 6.1 6.1 NA

4 19 NA NA NA NA NA NA NA NA NA NA

NA NA 7 15 3 21 NA 35 27.1 21.1 19.6 3

NA 15 7 15 2 18 36.7 41.9 81.8 49.3 47.6 NA

NA NA NA NA NA NA NA NA NA NA NA NA

4 13 9 15 2 39 25 48 66 52 52 11

NA 12 NA NA NA NA NA NA 65 57 50 NA

NA NA NA NA NA NA NA NA NA NA NA NA

5 NA NA NA NA NA NA NA NA NA NA NA

5 14 NA NA 2 NA 18.6 NA 24.8 NA NA NA

NA 14 NA NA NA NA NA NA NA NA NA NA

5 14 NA NA NA NA NA NA 40.3 32.2 29.6 8.6

NA 11 NA NA NA NA NA NA NA NA NA NA

NA 11 NA NA NA NA NA NA NA NA NA NA

Tibia Fibula Pes Hindlimb Personally

length length MT I MT II MT III MT IV MT V length length observed Specimen Sources

68.3 61.9 NA 37.9 32.6 33.8 NA NA NA Yes Ref. spec. Schwarz-Wings et al. (2011)

26.6 24.9 10 13 14.5 13 NA 29.2 83.4 Yes Holotype Wellnhofer (1971), Tennant &

Mannion (2014)

23.6 24.2 11.5 14.5 14.5 13 3.5 19 68.9 Yes Holotype Wellnhofer (1971), Tennant &

Mannion (2014)

26.9 34.8 25 29 30 28 8 60 129.2 No Holotype of Wellnhofer (1971), Tennant &

A. paintenense Mannion (2014)

39.1 40.6 21.1 17.7 NA NA NA 42.6 123.7 Yes Holotype Wellnhofer (1971), Tennant &

Mannion (2014)

NA NA NA NA NA NA NA NA NA Yes Ref. spec. Smith et al. (2010)

15.7 13.9 12.6 15.9 16.1 14.8 NA 17 49.4 Yes Holotype, ref. Wellnhofer (1971), Tennant &

spec. Mannion (2014)

17.6 15.8 6.7 7.1 7 7.5 1.3 15 50.3 Yes Holotype Wellnhofer (1971), Tennant &

Mannion (2014)

43 43 NA NA NA NA NA NA NA Yes Holotype Michard et al. (1990)

NA NA NA NA NA NA NA NA NA Yes Holotype Pritchard et al. (2013)

23 23 NA NA NA NA NA NA NA Yes Holotype Gilmore (1926)

13.1 13.1 NA NA NA NA NA NA NA No Holotype Storrs & Efimov (2000)

NA NA NA NA NA NA NA NA NA No Holotype Sweetman et al. (2015)

30.7 31.7 13.2 17.6 18.8 14.3 NA 44 113.2 Yes Holotype Buscalioni & Sanz (1990a,b)

74 72.3 41.5 45.6 0.1 NA NA NA NA Yes Holotype, Rogers (2003)

ref. spec.

NA NA NA NA NA NA NA NA NA Yes Holotype Salisbury (2002)

83 83 37 40 39 37 15 80 263 Yes Holotype Colbert & Mook (1951)

69 71 NA NA 41 38 NA NA NA No Holotype, Pol et al. (2009)

ref. spec.

NA NA NA NA NA NA NA NA NA Yes Holotype Young et al. (2016)

NA NA NA NA NA NA NA NA NA No Holotype Lauprasert et al. (2011)

31.2 NA NA NA NA NA NA NA NA No Holotype Schwarz & Salisbury (2005)

NA NA NA NA NA NA NA NA NA Yes Holotype Brinkmann (1992)

53.6 51.7 31.9 28 31.5 28.4 NA 59 165.9 Yes Holotype, Owen (1879)

paratype

NA NA NA NA NA NA NA NA NA Yes Holotype Martin et al. (2010, 2014a,b)

NA NA NA NA NA NA NA NA NA Yes Holotype Adams (2014)

Ref. spec., Referred specimen

1 ••!

J3 £ J3 £

g Ö Ö ö

& is 3 JB

+3 _ü ¿3 tg .3

cö ö „.£3 ö

© -J3 ^ -J3

Cj r->

B tg S tg

s ö S ö

ffi tS iS

CÖ r-t -+J

•S tg 3 sg

« fi Ö

H JH <2 iS

m ^ H ^

-3 tg g tg

iä ß § ß

PH £ JB

J3 33 £

C/3 CS cä

w S o S

3 M fc

J^ ö -S

W JB o

— -M ^

3 tg 3 tg M ö § ö

s - - *

w ^ ^ S

HM, CO 00 ,

oo t- <c <c co t- <c © © © ©

co <| © 15

co <1 © 15

££££££

i (35 i 00 co l> co (M 1C

co t- 00 t> <1 c- OO <1 t> <1 <1 co c- <1 <1 <1 <1 1C

© o o C2 25 o C2 o 25 C5 C2 25 C5 25 © C2 25 25 25 25 C5

_ (35 IC CO _ , t- 00 H

<;oot-oo<;<;cooO'-i

!> CO j t- t- <; © © 15

© (35

co <1 © 25

^ ^ °° ^ w ^

<3 <| es <| t- <| ¡5 ¡5 o ¡5 o ¡5

1 co O co (35 co <N 00 o (35 co 1

an an <1 (35 <1 C2 <1 00 an <1 00 <1 <1 <1 <1 o

C5 C5 o C5 C5 25 C5 C2 o 25 C2 ,-H 25 o 25 © 25 25 25 25

(35 1C (35 1C co i (35 <N 00 (35 l>

OO <1 <1 00 00 <1 <1 C- 1C <1 c- c-

© C3 C2 25 25 C5 o 25 C2 o 25 C2 o 25 o C2

<J <J <J <J ™ <J 25 25 25 25 © 25

(35 1C co (M o (35 © i 00 t> (35 1C o 00

CJ co 1C co co <1 i <1 1C t>

c4 c4 c4 c4 oo 1C 25 oo co 25 co i

<1 <1 O <1 ^ <1

25 25 co 25 co 25

, o h H o o H <J CS «3 oo ic m ¡5 c4 c4 in c4

co (35 o CQ IC CO

t> ^ co <; c4 15

<1 OD <1

15 c4 15

es a . a es & a »

k k k k o O O O O O ü

es es es es ig .ig .ig .ig

ii »<e

ö ■

o -S ^

■ [Iii - &

a. tä

u >«S Sä

a a « »k; »« u

a -«i S o

a a a a

a a a a

■2 -2 -2 -2

C C C C

^ ^ ^ ^

-e -e -e -c

a a a a

tt v: ■2 -2

C C ^ ^

co 00 1C 1C o o co o 1 i co o 1C

1 c- <1 <1 <1 <1 1(0 <1 00 <1 <1 <1 00 <1 i

ui 25 25 t!3 25 25 1C ui 25 co ui 25 25 25 25 1C

00 co 1C co t> co co i o o 1C O 1 co co

Ö <1 1 <N c- 1(0 <1 <1 co 0ID <1 <1 <1 <1 <1 1(3 <1 1(3 <1 1C

cu 25 t!3 t!3 oo 25 25 1(0 t^ 25 OO oo CC 25 25 co «3 25 25 co 25 co 25 !>

<N co l> o C3 co co (M 00 o co o 1C <N o 00 co (M

Ö <1 oo co 00 OO 1 co C3 <1 <1 1(3 <1 <1 <1 00 <1 c-

cu 25 co co ui 1 co co co 1(0 25 25 «3 25 25 25 co 25 C3

i co t- co t> 00 00 (35 l> 1C o i (35 <N i co

<N C3 «3 C3 < C3 00 < < co < 0ID 1C

c4 c4 c4 ,-H ¡5 ¡5 ¡5 ¡5 ¡5 ¡5 ¡5

•»2 S

- Hoplosuchus

Pachycheäosuchus Montsecosuchus 'Alligatorellus'sp. 'Alligatorium' franconicum Alligatorium meyeri

-Alligatorellus beaumonti

-Alligatorellus bavancus

-Atoposaurus obemdorfen

-Atoposaurus purdani

-Koumpkxtontosuchus

-PhoBdosaurus

Protosuchus -Hoplosuchus

- Eutretauranosuchus

- Amphicotylus

I-Theriosuchus pusmus

I-Theriosuchus guimarotae

I-Theriosuchus grandinaris

I-Theriosuchus sp.

- Wannchampsus

- Shamosuchusdjadochtaensis

- IManceausuchus

I-Theriosuchus'ibericus

'-Theriosuchus'sympiestodon

Pachych&bsuchus

-Montsecosuchus

-'Alligatorellus'sp.

---'Alligatorium'franconicum

—-Alligatorium meyeri

I-Alligatorellus beaumonti

I-Alligatorellus bavaricus

I-Atoposaurus obemdorfen

I-Atoposaurus purdani

Koumpkxtontosuchus

-Eutretauranosuchus

-Phohdosaurus

-Amphicotylus

-Theriosuchus pusiilus

-Theriosuchus guimarotae

--Theriosuchus grandinaris

-Theriosuchus sp.

-Wannchampsus

-Shamosuchusdjadoditaensis

-BriSanœausuchus

■-Theriosuchus'ibericus

I-Theriosuchus'sympiestodon

Figure 4. (A) Strict consensus topology for phylogenetic analysis when all taxa are included, and all characters are considered to be unordered (i.e., non-additive). Atoposauridae is marked with a red star. (B) Strict consensus for phyloge-netic analysis when all taxa are included, and selected characters are considered to be ordered (see Appendix 1).

latter this is more likely to be a result of character conflict, coupled with our inability to study the specimen first-hand. When these taxa are removed a priori, the topology is almost completely resolved (Fig. 5A), producing 11 MPTs of length 805 steps.

The strict consensus identifies the remaining source of conflict to be the relationship amongst Bril-lanceausuchus, Shamosuchus, and Wannchampsus (Fig. 5A). Atoposauridae comprises Alligatorium meyeri, Alligatorellus, and Atoposaurus. Montsecosuchus

Figure 5. (A) Strict consensus topology for phylogenetic analysis when iterpcr script is employed, and Alligatorium franconicum and 'Alligatorellus' sp. are excluded a priori. (B) Single most parsimonious tree for phylogenetic analysis when Pachycheilosuchus trinquei and 'Alligatorellus' sp. (MB.R.3632) are excluded a priori. Absolute Bremer support values are provided adjacent to nodes. Atoposauridae is marked with a red star.

is allied with Pachycheilosuchus, with this clade outside of all other neosuchian taxa. Bremer support values show that Atoposauridae is moderately well supported with a node value of 2. The two species of Alligatorellus are strongly supported with a node value of 4, and the relationship between Alligatorellus and Atoposaurus is supported by a node value of 2. The clade comprising (T. pusillus + T. guimarotae) is strongly resolved with a node support value of 4, and (T. sympiestodon + T. ibericus) has a node support value of 3.

We performed one final analysis that excluded Pachycheilosuchus along with MB.R.3632 a priori, but included Alligatorium franconicum. This was because of the unexpected placement of Pachycheilo-suchus in our topologies, given the more derived position it usually occupies (e.g. Buscalioni et al., 2011; Adams, 2014; Turner, 2015). This resulted in a single MPT of length 792 steps (Fig. 5B). Alligato-rium franconicum shifts to a position at the base of the clade comprising Koumpiodontosuchus and coel-ognathosuchians, with no other changes to the topology. Bremer support values show that Atoposauridae is overall only weakly supported (Fig. 5B), with a node value of 1. The sister-taxon relationship between T. pusillus and T. guimarotae is the most

strongly supported node on the tree, with a Bremer support value of 5, followed by (T. symp-iestodon + T. ibericus), with a support value of 4. Most other nodes have a support value of 1.

Implied weighting

Application of implied weighting on our ordered data set similarly recovers a monophyletic Atoposauridae, but one that also includes Montsecosuchus and MB.R.3632 (Fig. 6). Alligatorium meyeri remains the most basal member of this clade, followed by Montsecosuchus, and MB.R.3632 groups with the two species of Alligatorellus. Theriosuchus still remains polyphyletic, but there are now three groupings, with the clade comprising (T. grandi-naris + Theriosuchus sp.) shifting to a more basal position (Fig. 6). This possibly reflects the incompleteness of the specimens of both of these taxa. Alli-gatorium franconicum remains as the basal-most member of the clade comprising Koumpiodonto-suchus and coelognathosuchians. The Bremer node support for Atoposauridae is 0.23 (note that support values are non-integers owing to changes to character weights during the implied weighting procedure), with internal support values of 0.15-0.23. The clade comprising (T. pusillus + T. guimarotae) remains the

Figure 6. Single most parsimonious tree for phylogenetic analysis when implied weighting is employed with a weighting exponent of k = 3. Selected characters are considered to be ordered, and no taxa were excluded a priori. Absolute Bremer support values are provided adjacent to nodes.

most strongly supported clade, with a node support value of 0.81.

Bayesian analyses

When all taxa are included, the results of the Baye-sian analysis produce a largely unresolved topology (Fig. 7A). Protosuchus and Hoplosuchus are basal crocodyliforms, and Pachycheilosuchus retains a position basal to all other remaining taxa. The inter-relationships of MB.R.3632 ('AttigatoreUus' sp.), NMS G. 2014.52.1 (Theriosuchus sp.), Alligatorium, Montsec-osuchus, and Koumpiodontosuchus are all unresolved. Only AttigatoreUus and Atoposaurus are definitely members of Atoposauridae, with a posterior node support value of 0.71. Theriosuchus is poly-phyletic, as before in our parsimony analyses, with one group within Paralligatoridae, and the other group just outside of this clade. These groups are the most strongly supported nodes across the tree, with the clades (T. sympiestodon + T. ibericus) and (T. guimarotae + T. pusillus) both having a posterior node support value of 0.98.

When MB.R.3632 and NMS G. 2014.52.1 are both excluded a priori from analyses, then the topology changes, with a clade comprising Koumpiodonto-suchus and coelognathosuchians nested within Paral-ligatoridae (Fig. 7B), a result not supported by any previous crocodyliform analysis. Alligatorium meyeri

is excluded from Atoposauridae, in a slightly more crownward position at the base of a clade comprising Theriosuchus and all other higher neosuchians. Pachycheilosuchus and Montsecosuchus retain their basal positions.

Results summary

Our analyses demonstrate that Atoposauridae is a much more restricted clade than previously considered, comprising only Alligatorellus bavaricus, Alligatorellus beaumonti, Atoposaurus jourdani, Atoposaurus oberndorferi, and Alligatorium meyeri. However, this inclusion of Alligatorium meyeri is not supported by our Bayesian results. Based on this restricted taxonomic inclusion, Atoposauridae is recovered in a basal position within Neosuchia. Theriosuchus is consistently shown to be a polyphyletic taxon, comprising one set of species (T. guimarotae + (T. pusillus + T. grandinaris)) closely related to paralligatorids, and one clade (T. iberi-cus + T. sympiestodon) within Paralligatoridae. NMS G. 2014.52.1 (Theriosuchus sp.) is likely to be referable to the more basal group of Theriosuchus species. The position of MB.R.3632 (Alligatorellus sp.) cannot be conclusively determined, with an atoposaurid and a basal neosuchian placement supported in different analyses. Montsecosuchus is recovered outside of Ato-posauridae in almost all of our analyses, and might

Figure 7. (A) Results of phylogenetic analysis using Bayesian inference, when all OTUs are active and selected characters are considered to be ordered (see Appendix 1). (B) Results of phylogenetic analysis using Bayesian inference, when 'Alligatorellus' sp. (MB.R.3632) and Theriosuchus sp. (NMS G. 2014.52.1 and selected characters are considered to be ordered (see Appendix 1).

be more closely related to Pachycheilosuchus and other hylaeochampsids, although we cannot conclusively determine the position of these taxa. Alligato-rium franconicum is shown to be a non-atoposaurid taxon that is more closely related to bernissartiids and coelognathosuchians. Brillanceausuchus occupies a position within Paralligatoridae. Below, we provide revised diagnoses for each of these taxa, along with a discussion of the character states that support the revised systematic positions for all atoposaurid and non-atoposaurid taxa that we analysed.

SYSTEMATIC PALAEONTOLOGY

Crocodylomorpha Walker, 1970

Crocodyliformes Hay, 1930 Neosuchia Benton & Clark, 1988 Atoposauridae Gervais, 1871

Previous diagnoses and comments Atoposauridae was originally named as 'Ato-posaurides' by Gervais (1871). Since its genesis, there have been few attempts at providing a group diagnosis based on morphology and, to our knowledge, no phylogenetic definition has ever been proposed.

Romer (1956) provided the first morphological definition for Atoposauridae: (1) small, with a broad head and short, pointed snout; (2) external nares sometimes divided; (3) postorbital bar moderately inwardly displaced; (4) large orbits and small supratemporal fenestrae; (5) two rows of flattened dorsal osteoderms; (6) long and slender limbs; and (7) platycoelous vertebrae. However, this assessment was based exclusively on taxa known at the time from the Late Jurassic, comprising Atoposaurus, Alligatorellus, and Alligatorium, as well as Hoplo-suchus. Kuhn (1960) largely followed this diagnosis, but omitted several characteristics, whereas Steel (1973) subsequently re-incorporated them, and noted that there were at least some cases of atoposaurids with procoelous vertebrae, presumably referring to their inferred presence in T. pusillus (Joffe, 1967; see below). Although some of these characteristics, such as the relative sizes of the orbit and supratemporal fenestra, are known to occur in juvenile crocodyliform specimens (e.g. Joffe, 1967; Schwarz & Salisbury, 2005), it is noteworthy that even specimens of adult atoposaurids possess this feature. Therefore, this condition indicates that atoposaurids might consistently retain paedomorphic characteristics into adulthood, a factor that has probably contributed to difficulties in resolving their phylogenetic affinities.

Buscalioni & Sanz (1988) presented several hypotheses of atoposaurid inter-relationships by treating cranial, postcranial, and metric characters independently. They regarded several taxa as nomina dubia, including ''Alligatorium' franconicum and Alligatorium paintenense', and non-atoposaur-ids, including Karatausuchus, Shantungosuchus, and Hoplosuchus, but without explicit statements regarding their morphology to support these taxo-nomic opinions. Buscalioni & Sanz (1988) concluded that Atoposauridae could be diagnosed based on: (1) enlarged anterior maxillary teeth; (2) loss of the external mandibular fenestra; (3) reduction or loss of the antorbital fenestra; (4) squamosals not ventrally depressed; and (5) lack of dental hypertrophy. They also considered the possession of five premaxillary teeth and between 12 and 18 maxillary teeth as additional ambiguous synapomorphies for Atoposauridae, as these characteristics are not visible in either Atoposaurus or Alligatorium. However, almost all of these characteristics are more widespread in Neosuchia, or variably present within Theriosuchus and other putative ato-posaurids. Furthermore, their recovery as atoposaurid synapomorphies might largely have been a product of limited sampling of out-groups.

The most recent diagnosis for Atoposauridae was provided by Martin et al. (2010), who analysed several atoposaurids to resolve the position of T. symp-iestodon, based largely upon the Pol et al. (2009) data matrix, which focussed on early eusuchian relationships and their morphological transformation from Neosuchia. Martin et al. (2010) diagnosis of Atoposauridae consisted of the following synapo-morphies: (1) external nares facing dorsally and not separated by a premaxillary bar from anterior edge of rostrum; (2) antorbital fenestra much smaller than the orbit; (3) five premaxillary teeth; (4) basioccipital and ventral part of otoccipital facing posteriorly; (5) unsculpted region in the dentary below the tooth row absent; and (6) lateral surface of dentaries below alveolar margins at middle to posterior region of tooth row vertically orientated, continuous with rest of lateral surface of dentaries. However, this diagnosis was based only on information provided by three species of Theriosuchus and one of Alligatorium.

All of these aforementioned putative synapomor-phies are included as characters in our new data matrix and below we provide a revised diagnosis for Atoposauridae, as well as defining the group as a phylogenetic clade for the first time. Numbers in parentheses refer to characters and states (e.g. C159.1 means character 159, state 1).

Phylogenetic definition

Atoposauridae is a stem-based clade comprising all taxa more closely related to Atoposaurus jour-dani von Meyer, 1850, than Crocodylus Laurenti, 1768

Included taxa

Alligatorium meyeri (Jourdan, 1862), Alligatorellus beaumonti (Gervais, 1871), Alligatorellus bavaricus (Wellnhofer, 1971),

Atoposaurus jourdani (von Meyer, 1850), Atoposaurus oberndorferi (von Meyer, 1850)

Distribution

Late Jurassic of France and Germany.

Revised diagnosis and discussion Many of the synapomorphies below represent the quantification and demarcation of state boundaries from previously proposed characters. They also diagnose a more exclusive set of taxa, as we no longer consider features shared between Therio-suchus and definitive atoposaurids to be diagnostic for a united clade. Measurements and ratios pertaining to synapomorphies for taxa are provided in Tables 2 and 3, respectively. Atoposauridae can be diagnosed based on the following unique combination of character states (synapomorphies, S; Figs 8, 9):

(51) Complete division of the external nares dorsally by anterior projection of the nasals (C10.0): This feature is only known for Atoposauridae (Figs 8, 9A), T. guimarotae, and T. pusillus (Fig. 9B). In Alligatorium' franconicum, the external nares are completely opened and undivided, based on the reconstruction presented by Wellnhofer (1971), and similar to T. grandinaris (Lauprasert et al., 2011), Amphicotylus (Mook, 1942), Brillanceausuchus, and some notosuchians (sensu Pol et al., 2014), including the baurusuchid Campinasuchus dinizi (Carvalho et al., 2011). This division of the external nares by the anterior-most extent of the nasals was considered by Salisbury & Naish (2011) to be diagnostic for Theriosuchus, but we consider it to be independently acquired in this genus and atoposaurids. Some other taxa, such as the notosuchian Araripesuchus, also share this division, but the external nares are more anteriorly placed and face anteriorly (Buffetaut, 1981).

(52) Skull anteroposterior length to orbit length ratio <4.0 (C27.0-1): This synapomorphy is also shared by other neosuchian taxa, including

Figure 8. (A) Skull of the holotype of Alligatorellus beaumonti (MNHL 15639) in dorsal view. See text for details. (B) Skull of the holotype of Alligatorellus bavaricus (BSPG 1937 I 26) in dorsolateral view. Synapomor-phies for Atoposauridae indicated (see text for details).

T. pusillus, Wannchampsus (Adams, 2014), the 'protosuchian' Hoplosuchus (Gilmore, 1926), and possibly Karatausuchus (Efimov, 1976). Atoposaurus displays an extreme version of this morphology, possessing a relatively longer orbit, giving a ratio of <3.0 for both species. Montsecosuchus falls just outside of this range, with a ratio of 4.05. Although the latter value is similar in orbital dimensions to atoposaurids, we chose to set the state boundaries for this character at regular intervals, as opposed to selecting them towards creating inclusive groupings a priori, and therefore Montseco-suchus is distinct from atoposaurids in this respect. Longirostrine taxa, as expected, have a much higher ratio, with Amphicotylus possessing an extreme of this with a value of 10.0. Shamosuchus and Brillanceausuchus, although brevirostrine, have intermediate ratio values (6.1 and 6.8, respectively), reflecting the smaller dimensions of the orbits.

Figure 9. (A) Skull of the holotype of Alligatorium mey-eri (MNHL 15646) in dorsal view. Synapomorphies for Atoposauridae are indicated (see text for details). (B) Skull of the paratype of Theriosuchus pusillus (NHMUK PV R48330) in dorsal view. Shared characteristics with atoposaurids are indicated (see text for details).

(53) Skull mediolateral width to orbit width ratio of <2.5 (C28.0): Amongst atoposaurids, only Atoposaurus jourdani differs in having a smaller orbital width, giving a ratio of 2.69.

'Alligatorium' franconicum is also just outside this state boundary with a value of 2.53. This ratio is unknown for most specimens attributed to Theriosuchus, but the ratios are 3.0 and 3.46 for T. guimarotae and T. pusillus, respectively. Montsecosuchus falls between these two species, with a ratio of 3.19. The relatively large size of the orbits in atoposaurids, as quantified here, might relate to retention of the paedomorphic state, as it appears to also be possessed by mature representatives of species (e.g. Theriosuchus).

(54) Skull mediolateral width to external supratemporal fenestra width between 3.0 and 5.0 (C30.1): This feature, similar to the proportionally large orbits, also possibly relates to the relatively small body sizes of these taxa. This state is unknown in Atoposaurus because it is not possible to assess the morphology of

the supratemporal fenestrae owing to their preservation. Both T. guimarotae and T. pusillus share this character state with atoposaurids, with ratios of 4.64 and 4.8, respectively. The only taxon in our data matrix that has a proportionally larger supratemporal fenestra is Pholidosaurus (Salisbury, 2002), which has a ratio of 2.81. Both '''Alligatorium' franconicum and Montsecosuchus have proportionally smaller supratemporal fenestrae, with ratio values of 5.15 and 5.21, respectively.

(55) Dorsal surface of the premaxilla internarial bar projects anteriorly to main body of premaxilla (C35.1): This feature is diagnostic for atoposaurids, and possibly only shared outside the group with Alligatorium' franconicum (Wellnhofer, 1971). In all other taxa we sampled, the internarial bar of the premaxilla does not project anteriorly to the main body of the premaxilla and the external nares. This morphology is also different to that seen in goniopholidids, in which there is a distinct mediolateral constriction and an anteriorly spatulate morphology.

(56) Ventral depression on dorsal margin of postorbital, sometimes developing into a shallow sulcus (C128.2): Wu et al. (1996) considered the inwardly displaced (ventrally depressed) postorbital bar between the supratemporal fenestra to be diagnostic for Theriosuchus, but this appears to be absent in T. pusillus and its presence is questionable in T. guimarotae. It is clearly present in Alligatorellus (Fig. 8), and perhaps Alligatorium meyeri too, and therefore we consider it to be diagnostic for Atoposauridae. However, this feature might have also been acquired by the Glen Rose Form' [Turner, 2015 (probably referable to Wannchampsus; Adams, 2014)], and is similar to the condition observed in more basal crocodyliforms, including the shartgeosuchid Fruitachampsa callisoni (Clark, 2011), in which the postorbital bar is relatively poorly developed. Paralligator gradilifrons (Turner, 2015) also possesses this supratemporal-orbital groove, but the postorbital bar is well developed and robust, distinct from Alligatorellus.

(57) Quadratojugal contributes extensively to the ventral and posterior margins of the lateral temporal (infratemporal) fenestra (C152.0): This feature is present in all atoposaurids (Figs 8, 9A), but cannot be assessed in Atoposaurus owing to the preservation of the skull. It is also shared by T. pusillus. In some crocodyliforms, including goniopholidids, Protosuchus, and T. guimarotae (Schwarz &

Salisbury, 2005), the quadratojugal only contributes to the posterior margin. In Montsecosuchus, Brillanceausuchus, Araripe-suchus patagonicus (Ortega et al., 2000), Allodaposuchus precedens (Delfino et al., 2008b), and eusuchians including Wann-champsus (Adams, 2014), Acynodon iberoccitanus (Martin, 2007), and Shamosuchus (Pol et al. , 2009), the quadratojugal participates extensively in the dorsal and posterior margins of the lateral temporal fenestra.

(58) Otic aperture between squamosal and quadrate posteriorly open, not closed by the quadrate and otoccipital (C157.0): This character state appears to be shared by Alligatorium and Alligatorellus, but cannot be assessed for Atoposaurus owing to the preservation of the skull. It is also shared with several other neosuchian taxa, including Goniopholis simus (Salisbury et al., 1999), Pholidosaurus (Salisbury, 2002), Amphicotylus (Mook, 1942), Allodaposuchus precedens (Delfino et al., 2008b), and Shamosuchus [Turner, 2015 (although this aperture appears to be closed in Shamosuchus djadochtaensis; Pol et al., 2009)]. In these taxa, the cranioquadrate passage lacks a lateral wall, forming a sulcus or canalis quadratosqua-mosoexoccipitalis (Salisbury et al., 1999). This feature is distinct from most mesoeucrocodylians, including Susisuchus and Isisfordia, as well as modern crocodylians, in which there is a sharp posterior rim and the passage is enclosed by the otoccipital and quadrate (Salisbury et al., 2006; Pol et al., 2009).

(59) Homodont pseudocaniniform dentition (combination of C253.1, C254.1, C255.0, and C258.0): Although difficult to evaluate owing to the dorsal flattening of specimens, this type of dentition appears to be present for all specimens of Alligatorellus in which teeth are visible (Fig. 8), and in Alligatorium meyeri (Fig. 9A), and is synapomorphic amongst all putative atoposaurids (although see below). All species of Theriosuchus are clearly distinct from Alligatorellus, with their characteristic combinations of heterodont morphologies. The teeth of Atoposaurus are poorly known owing to preservation, although several damaged teeth preserved in the type specimen of Atoposaurus oberndorferi also appear to be pseudocaniniform. In Alligatorium meyeri, the tooth row is exposed in lateral view, but only one or two of these appear to actually be from the maxillary arcade, with the rest from the dentary, against which the maxilla is apressed. All of these exposed teeth appear to be pseudocaniniform in

morphology, and of a similar conical shape to Alligatorellus. Therefore, we tentatively regard this feature as being diagnostic for Atoposauridae, pending the further discovery and analysis of teeth in atoposaurids.

Below, we present revised diagnoses and discussions for all taxa included within our revised definition of Atoposauridae, and those previously regarded as putative atoposaurids. These are based on the character state distributions from the results of our parsimony-based phylogenetic analysis when Pachy-cheilousuchus and MB.R.3632 are excluded, and supplemented with details from original descriptions and personal observations where possible. Autapo-morphies in the diagnoses are highlighted with an asterisk (S*).

Definitive atoposaurid taxa Alligatorium Jourdan, 1862

Included species

Alligatorium meyeri (type species).

Previous diagnoses and comments Alligatorium was erected by Jourdan (1862) for an incomplete skeleton from the lithographic limestones of Cerin, France, and described and figured by Lortet (1892). This genus has also been reported from the Upper Jurassic of Bavaria, Germany (von Zittel, 1890; Kuhn, 1961), and identified as Alligatorium franconicum and Alligatorium paintenense [all specimens of both lost or destroyed during World War II (Wellnhofer, 1971)], but we consider the latter species to be a junior synonym of the former (Tennant & Mannion, 2014), and regard 'Alligatorium' franconicum to represent a non-atoposaurid taxon (see below). vidal (1915) also described a specimen from the Early Cretaceous of Spain as Alligatorium deper-eti, later recombined as Montsecosuchus depereti (Buscalioni & Sanz, 1988, 1990a).

Revised diagnosis

As for the type and only species.

Distribution

Late Jurassic of southern France.

Alligatorium meyeri Jourdan, 1862 (type species)

Type locality and horizon

Unknown bed, Kimmeridgian (Late Jurassic); Cerin, Ain, France.

Type specimen

MNHL 15646, partial skeleton and skull, with counterpart slab MNHL 15462.

Previous diagnoses and comments Joffe (1967) considered Alligatorium meyeri to be an immature specimen based on a range of ontogeneti-cally variable cranial characteristics. However, Clark (1986) and Benton & Clark (1988) considered Alligatorium to represent the most mature specimen of a lineage in which Atoposaurus and Alligatorellus represented younger growth stages. Tennant & Mannion (2014) were unable to confirm this based on the few specimens available for allometric analysis, but noted several morphological distinctions amongst the three genera. Our phylogenetic results are distinct from those of Clark (1986) and Benton & Clark (1988), as well as Buscalioni & Sanz (1990a) and Karl et al. (2006), which found a sister relationship between Alligatorium and Theriosuchus. This discrepancy is probably because of our increased sampling of paralligatorids, and our generally broader sampling of basal neosuchian taxa (sensu Benton & Clark, 1988). We recover Alligatorium meyeri at the base of Atoposauridae. However, it should be noted that our Bayesian analysis did not recover Alligatorium meyeri within a monophyletic Atoposauridae, and we were unable to resolve its position more precisely using this analytical approach.

Revised diagnosis and discussion

(51) Paired and unfused nasals (C65.0): The presence of 'paired' nasal bones has been widely used in diagnoses of atoposaurids, and noted in other mesoeucrocodylians, including the notosuchian Simosuchus (Buckley et al., 2000) and the neosuchian Paluxysuchus (Adams, 2013). However, it is not entirely clear what this means, as the nasal bones are always paired. Therefore, we consider this character to refer to whether or not the paired nasals are fused, and as such constitute a single' element, or are unfused. In Alligatorium meyeri, the nasals are fused along the midline, similar to Wannchampsus, T. guimarotae, T. grandinaris, and T. pusillus. In Alligatorellus, the nasals are only weakly fused or contact one another along the midline, comparable to goniopholidids, Montsecosuchus, and Shamosuchus (Pol et al., 2009).

(52) Lateral edges of the nasals subparallel to one another (C67.0): This morphology occurs anterior to the nasal contact with the periorbital elements, and is similar to Montsecosuchus, Brillanceausuchus, Allodaposuchus precedens (Delfino et al., 2008b), and possibly

Araripesuchuspatagonicus (Ortegaet al., 2000). Alligatorellus is distinct amongst atoposaurids in that the edges are oblique to one another and converge anteriorly (Tennant & Mannion, 2014), similar to Bernissartia (Buscalioni, Buffetaut & Sanz, 1984; Norell & Clark, 1990). They are laterally flared posteriorly in T. pusillus.

(53) Jugal and lacrimal with confluent anterior margins (C78.0): The jugal and lacrimal have confluent anterior contacts, instead of a discrete convexity in which a notch develops and is filled by the maxilla. This morphology is also seen in 'T.' sympiestodon (Martin et al., 2010, 2014a), as well as Protosuchus (Colbert & Mook, 1951), Amphicotylus (Mook, 1942), and Pholidosaurus (Salisbury, 2002).

(54) Frontal with single longitudinal ridge along midline suture (C100.1): The presence of a midline ridge on the frontal suture has often been considered diagnostic for Theriosuchus (e.g. Schwarz & Salisbury, 2005; Salisbury & Naish, 2011), being present on all specimens attributed to Theriosuchus in which the interorbital region is preserved. However, this ridge is also present in Shamosuchus djadochtaensis (Pol et al., 2009) and Wannchampsus (Adams, 2014). It is likely that the presence of this character is related to ontogeny because in T. guimarotae this frontal ridge is only developed in more mature individuals (Schwarz & Salisbury, 2005). Alligatorellus and Atoposaurus do not show any evidence of this ridge, despite appearing to have completely fused frontals.

(55) Anterior portion of frontal mediolaterally constricted, with convergent lateral margins (C109.1): We tentatively consider this feature to be diagnostic for Alligatorium meyeri, although it might also be present in Alligatorium' franconicum (Wellnhofer, 1971), T. guimarotae (Schwarz & Salisbury, 2005), and Amphicotylus (Mook, 1942). This morphology is distinct from notosuchians, in which the frontal is slightly mediolaterally constricted, but remains broad between the prefrontals (Gomani, 1997; Buckley et al., 2000), the goniopholidid Calsoyasuchus in which the frontal-nasal suture forms a valley' on the dorsal surface (Tykoski et al., 2002), the neosuchian Khoratosuchus with a nonconstricted and bifurcated frontal anterior process (Lauprasert et al., 2009), and eusuchians such as Aegisuchus in which the frontals are constant in width anterior to the orbits (Holliday & Gardner, 2012).

(56) Supratemporal rims developed along entire medial margin (C119.2): Alligatorium meyeri possesses well-developed supratemporal rims

along the entire medial border of each external fenestra, similar to Wannchampsus and T. pusillus, T. guimarotae, 'T.' ibericus, and 'T.' sympiestodon. This morphology is distinct from that in Alligatorellus beaumonti and Brillanceausuchus, in which this ridge is only developed posteriorly, and from that in Alligatorellus bavaricus and Montsecosuchus in which the medial edges are flat (Tennant & Mannion, 2014). Pachycheilosuchus, coelognatho-suchians, and Shamosuchus (Pol et al., 2009) do not appear to possess supratemporal rims.

(57) Dorsal margin groove for dorsal ear lid with a medial curvature (C137.1): This character state appears to be unique for Alligatorium meyeri within atoposaurids, but is also present in more advanced neosuchians, including 'T.' ibericus, Wannchampsus, and Brillanceausuchus. In other atoposaurids, this margin, comprising the lateral edge of the postorbital and squamosal, is straight, similar to T. guimarotae (Schwarz & Salisbury, 2005) and T. pusillus.

(58) External mandibular fenestra present (C207.1) and oval-shaped with anteroposteriorly orientated long axis (C210.0): The presence of an external mandibular fenestra is also shared with T. guimarotae (Schwarz & Salisbury, 2005) amongst those taxa included in our analysis as putative atoposaurids, but also represents the plesiomorphic crocodyliform condition, being present in goniopholidids (e.g. Halliday et al. , 2015), Protosuchus (Colbert & Mook, 1951), as well as in notosuchians such as Baurusuchus (Nascimento & Zaher, 2011) and Labidiosuchus (Kellner et al., 2011), in which the opening is enlarged. This fenestra is lost in some advanced neosuchians, including paralligatorids (Monte-feltro et al., 2013), but is present in most eusu-chians and Crocodylia (Brochu, 2004; Salisbury et al., 2006; Pol et al., 2009), although in some cases it is strongly reduced (Brochu et al., 2012).

(S9*) Individual dorsal and caudal osteoderms with unsculpted edges (C304.1), square-shaped (C308.3), and lacking a dorsal keel anteriorly (C311.0) and posteriorly (C312.0): The individual osteoderms preserved in Alligatorium meyeri have unsculpted edges on all but the nuchal-most elements, a feature that cannot be a preservation artefact as the cranial table remains sculpted, and the centre of each osteoderm remains relatively lightly sculpted. Although this feature might be present in Alligatorium' franconicum based on the figures provided in Wellnhofer (1971), we cannot assess this first-hand. Furthermore, Hoplosuchus (Gilmore, 1926) also possesses unsculpted

osteoderms, although it is likely that this species represents a protosuchian-grade taxon (sensu Wu, Sues & Dong, 1997). Therefore, we consider this feature to be autapomorphic for Alligatorium meyeri, pending future discoveries of Alligatorium' franconicum. The dorsal osteoderms of Alligatorium meyeri are also square-shaped in dorsal view, a feature shared with T. pusillus, Pachycheilosuchus (Rogers, 2003), and some of the osteoderms of Protosuchus (Colbert & Mook, 1951). The overall morphology of the osteoderm shield closely resembles that of the neosuchian Araripesuchus patagonicus (Ortega et al., 2000). In addition, the osteoderms of Alligatorium meyeri lack any presence of a dorsal keel, a feature shared with a range of neosuchian taxa, but that is distinct from Alligatorellus, in which the presence and morphology of this keel varies longitudinally axially (Tennant & Mannion, 2014).

Unnamed clade: (Atoposaurus + Alligatorellus)

Alligatorellus is united with Atoposaurus within all of our analyses (Figs 4-7). This is based on a range of character states, including: (1) a slit-like (i.e. medio-laterally narrow and anteroposteriorly elongated) external supratemporal fenestra (not visible in Ato-posaurus oberndorferi owing to preservation) (C17.2); (2) smooth lateral surface of anterior jugal process near maxillary contact, not stippled or striated (C51; note that we did not code this as a separate character state owing to potential duplication, as in Ato-posaurus, the entire external surface of the skull is smooth and unsculpted); (3) straight ventral edge of maxilla in lateral view (C52.0), similar to protosuchi-ans (not sinusoidal or convex as in other neosuchi-ans); (4) minimum mediolateral width between supratemporal fenestrae more than one-third of total width of cranial table (C126.1), acquired in parallel with Montsecosuchus, Brillanceausuchus, T. pusillus, and T.' sympiestodon; and (5) postorbital bar between orbit and supratemporal fenestra very narrow (with respect to lateral edge of postorbital lateral to supratemporal fenestra) and unsculpted, with superficial furrow on dorsal surface of postorbital connecting anterior edge of supratemporal fenestra to the posterior edge of and orbital (C128.2).

Atoposaurus von Meyer, 1850 Included species

Atoposaurus jourdani (type species, named first in von Meyer, 1850) and Atoposaurus oberndorferi.

Distribution

Late Jurassic of southern France and south-east Germany.

Previous diagnoses and comments Both species of Atoposaurus were named by von Meyer (1850), with Atoposaurus jourdani receiving a full description by von Meyer (1851). Wellnhofer (1971) was the first to present a diagnosis for Ato-posaurus based on the specimens from France and Germany, noting the lack of dermal armour, a feature that could be related to either ontogeny or taphonomy (Schwarz & Salisbury, 2005). Subsequently, Atoposaurus and its constituent species have largely been considered to be nomina dubia, and often regarded as juvenile representatives of Alligatorellus and/or Alligatorium (e.g. Clark, 1986; Buscalioni & Sanz, 1988). Clark (1986) considered Alligatorellus, Alligatorium, and Atoposaurus to be the same taxon represented by different growth stages. If this were the case, then Atoposaurus von Meyer, 1850; would retain priority, and Alligatorium Jourdan, 1862, along with Alligatorellus Gervais, 1871, would be synonymized with Atoposaurus. Almost all subsequent phylogenetic analyses have included just Alligatorium based on this conclusion, without consideration of the other taxa. The present analysis is the first to consider both potential species of Atoposaurus as independent OTUs, and finds them to be sister taxa in all cases (Figs 4-7), thus supporting their generic assignment.

Despite noting the same features in Alligatorellus, Wellnhofer (1971; see also Steel, 1973) stated that the presence of large orbits, a closed internal supratemporal fenestra, and divided external nares were all features defining Atoposaurus. The presence of an inwardly displaced postorbital bar (Steel, 1973) is not clear owing to the preservation of available specimens of Atoposaurus oberndorferi, but does appear to be a feature of Atoposaurus jourdani. Four of the five synapomorphies that we identify for Ato-posaurus (S1-3, S5) are contentious as they could be indicative of a juvenile phase of growth (e.g. Joffe, 1967), but equally probably they could represent the retention of juvenile characteristics through paedo-morphism related to the relatively small body size of Atoposaurus. Unfortunately, based on currently available specimens, it is impossible to distinguish between these two hypotheses (Tennant & Mannion, 2014).

Revised diagnosis and discussion (S1) Dorsal cranial bones comprising the skull roof unsculpted (C1.0): This lack of dermal sculpting, combined with their overall diminutive size, indicates that Atoposaurus specimens might be

represented by juveniles. However, it cannot be definitively confirmed that Atoposaurus is a juvenile representative of other contemporaneous atoposaurids based on allometric growth patterns alone (Tennant & Mannion, 2014), and it is likely that Atoposaurus, Alligatorellus, and Alligatorium represent three distinct genera, as our results indicate (Figs 4-7).

(52) External surface of snout unsculpted (C3.0): We consider this to be a distinct feature from S1, as in Alligatorellus there is a different pattern of sculpting between the cranial table and the rostrum. Atoposaurus is similar to Alligatorellus in this respect, completely lacking any evidence of cranial ornamentation, although this cannot be assessed properly in Atoposaurus oberndorferi owing to the mode of preservation of the holotype specimen. As with S1, this character is likely to be highly influenced by either ontogeny or paedomorphism (Joffe, 1967).

(53) Skull anteroposterior length to orbit length ratio <3.0 (C27.0): This feature is unique to Atoposaurus, and represents the characteristic proportionally large orbits and short snout of this taxon, noted by Joffe (1967) to be indicative of a juvenile status. Other atoposaurids have a ratio of between 3.0 and 4.0. Karatausuchus has a ratio of 3.36, slightly higher than Alligatorellus bavaricus (3.12), which approaches the state boundary for Atoposaurus, but it is likely that Karatausuchus represents a juvenile specimen of a (probably non-atoposaurid) crocodyliform (Storrs & Efimov, 2000; see below).

(54) 50 or more caudal vertebrae (C276.1): Complete axial columns are rarely preserved in specimens previously assigned to Atoposauridae, and the proportional numbers of cervical, dorsal, sacral, and caudal vertebrae remain poorly known, especially for Theriosuchus. Both species of Alligatorellus preserve complete and articulated caudal vertebral series, and have 40 vertebrae each. Montsecosuchus appears to only have 21 caudal vertebrae, and Pachycheilosuchus has just 18 (Rogers, 2003). Protosuchus richardsoni has 39 caudal vertebrae, and Karatausuchus has 46 (Storrs & Efimov, 2000), approaching the number for Atoposaurus, but no other crocodyliform taxon has 50 vertebrae. The presence of 50 or more caudal vertebrae, in all specimens of Atoposaurus in which this feature can be measured, is not known in any other mesoeucrocodylian taxon, and cannot be explained by ontogeny (Tennant & Mannion, 2014); therefore, we regard it as a diagnostic feature for Atoposaurus, irrespective of the ontogenetic stage of the specimens, and

therefore consider Atoposaurus to be a valid taxon.

(S5) Osteoderms absent: The lack of osteoderms is unlikely to be a taphonomic artefact (contra Schwarz-Wings et al., 2011), and is either a feature associated with extreme dwarfism in Atoposaurus, or relates to their lack of development in juvenile individuals. The only other putative atoposaurid that is similar in this respect is Karatausuchus, which Storrs & Efimov (2000) described as having reduced dermal ossicles.

Atoposaurus Jourdani von Meyer, 1850 (type species)

Type locality and horizon

Unknown bed, Kimmeridgian (Late Jurassic); Cerin, Ain, France.

Type specimen

MHNL 15679, articulated partial skeleton comprising dorsally flattened skeleton and skull, missing the posterior-most caudal vertebrae, with distal hin-dlimbs and distal left forearm preserved as impressions.

Referred specimen

MHNL 15680 (same locality as type specimen), posterior half of articulated skeleton, including trunk vertebrae and forearms.

Previous diagnoses and comments von Meyer (1851) named Atoposaurus jourdani, and described this taxon in a subsequent paper (von Meyer, 1851). We find that a unique combination of metric characters, almost exclusively regarding the relative proportions of the forelimb and hindlimb elements, can be used to distinguish this taxon from Atoposaurus oberndorferi, along with a single autapomorphy.

Revised diagnosis and discussion (S1) Skull mediolateral width to orbit width ratio of 1.80 (C28.0): This represents the lowest ratio for all atoposaurids, demonstrating that the orbits comprise almost the entire mediolateral width of the skull, separated by the narrow frontals. This is similar to Alligatorellus bavaricus, which also has enlarged orbits, but slightly mediolaterally wider frontals between the orbits (Tennant & Mannion, 2014). (S2*) Six cervical vertebrae (C266.0): This character state is unique amongst all putative atoposaurids, with all others possessing seven

cervical vertebrae. We were unable to determine this character state for any specimens assigned to Theriosuchus owing to their preservation and/ or incompleteness. This cervical count is distinct from Protosuchus richardsoni (nine) and Hoplosuchus (11), as well as Karatausuchus (eight; Storrs & Efimov, 2000) (Table 2).

(53) Forelimb length to hindlimb length ratio of 0.63 (C285.0): This character state is similar to Protosuchus richardsoni (0.65) and Montseco-suchus (0.63). Atoposaurus jourdani is distinct in this respect from other atoposaurids, including Alligatorellus beaumonti, which has a ratio of 0.81, and Atoposaurus oberndorferi, which has a ratio of 0.78. Alligatorellus bavaricus has an intermediate ratio of 0.76 (Table 3).

(54) Humerus length to femur length ratio of 0.67: This feature is similar to Protosuchus richardsoni (0.66) and Montsecosuchus (0.70). Alligatorellus beaumonti is similar, with a ratio of 0.75, but Alligatorellus bavaricus is quite distinct, with a ratio of 0.89, similar to Atoposaurus oberndorferi (0.89). Karata-usuchus also has similar limb proportions, with a ratio of 0.73. Theriosuchus pusillus falls within the range for atoposaurids, with a ratio of 0.76, but T. guimarotae is distinct with a ratio of 0.98, approaching that for Pachycheilosuchus (1.09) and Brillanceausu-chus (1.11) (Table 3).

(55) Radius to tibia length of 0.61 (C286.0): This character state is similar to Protosuchus richardsoni (0.63) and Montsecosuchus (0.64), along with Pachycheilosuchus (0.64). Theriosuchus pusillus has a proportionally long radius to tibia ratio (0.55), with this value only being exceeded by Karatausuchus (0.47). Based on Wellnhofer (1971), Alligatorium' franconicum has the most extreme value, with a ratio of 0.89, reflecting a proportionally long radius. This value is similar to Brillanceausuchus (0.88). Atoposaurus oberndorferi and Alligatorium meyeri each have a ratio of 0.74, similar to Shamosuchus (0.72) and Hoplosuchus (0.76) (Table 3).

(56) Metatarsals longitudinally grooved (C302.0): This feature also characterizes Alligatorellus beaumonti and Montsecosuchus (Tennant & Mannion, 2014), in contrast to the smooth and flat metatarsals that characterize most other mesoeucrocodylians. However, we are cautious in our interpretation of this feature, as there remains the possibility that it could represent post-mortem crushing of the delicate long bones in the tarsus.

Atoposaurus oberndorferi von Meyer, 1850 Type locality and horizon

Solnhofen Formation, early Tithonian (Late Jurassic, Hybonoticeras hybonotum ammonoid zone); Kelheim, Eichstatt, Bavaria, Germany.

Type specimen

TM 3956, near-complete skeleton, missing only the dorsal part of the skull and posterior portion of the tail.

Referred specimen

BSPG 1901 I 12, a counterpart specimen of a different individual comprising the impression of the complete skull and skeleton in lateral view.

Previous diagnoses and comments Wellnhofer (1971) diagnosed Atoposaurus oberndor-feri primarily on several size-based characteristics, but these are unlikely to represent diagnostic morphological characters. He also noted the presence of five premaxillary and eight maxillary teeth, but this could not be confirmed via observation of the type specimen because of the way in which it is preserved, and was not illustrated in the figure of the referred specimen in Wellnhofer (1971). Steel (1973) followed the diagnosis of Wellnhofer (1971), and also suggested that the inwardly displaced postorbital bar was diagnostic of Atoposaurus oberndor-feri; however, this feature is now recognized as characterizing Atoposauridae (see above). Furthermore, because of the lateral compression of the type specimen, it was not possible to directly confirm the presence of this feature in Atoposaurus oberndorferi, and it is not figured by Wellnhofer (1971), and therefore cannot be supported. We present a revised diagnosis based on examination of the type specimen for Atoposaurus oberndorferi, and tentatively consider it to be a valid taxon based on three ambiguous autapomorphies.

Revised diagnosis and discussion (S1) Skull anteroposterior length to width ratio ~2.00 (1.98) (C25.0): This feature is tentatively considered to be diagnostic for Atoposaurus oberndorferi, as the skull is highly incomplete and preserved only in ventrolateral aspect. This estimated skull length-to-width ratio is high, similar to Eutretauranosuchus (1.97), Montsecosuchus (1.80), which is represented by a mature specimen (Buscalioni & Sanz, 1990a), and Theriosuchus guimarotae (1.82) (Schwarz & Salisbury, 2005). Atoposaurus jourdani has a much lower ratio (1.28), more

similar to Protosuchus (1.31) and Hoplosuchus (1.35). The only taxa that have higher ratios are Alligatorellus (2.06-2.21), Alligatorium meyeri (2.26), Alligatorium' franconicum (2.77), and Koumpiodontosuchus (2.04) (Table 3).

(52) Skull anteroposterior length to orbit length ratio <3.00 (2.86) (C27.0): As with (S1), we are cautious with our interpretation of this character state based on the way in which the observed type specimen of Atoposaurus oberndorferi is preserved, exposing the enlarged orbit only in ventral aspect. The only taxon with a lower ratio is Atoposaurus jourdani (2.33), with Alligatorellus and Alligatorium possessing ratio values between 3.0 and 4.0. Protosuchus has a ratio of 4.52 (Table 3), an intermediate value between atoposaurids and higher neosuchians.

(53) Inwardly (dorsally) displaced splenial on the ventral mandibular surface (C234.1): In all other taxa we analysed in which the ventral surface of the mandible was exposed, the anterior portion of the splenial is confluent ventrally with the posterior cavity that is formed from the two posteriorly divergent mandibular rami. In Atoposaurus oberndorferi, the splenial is slightly inset at its contact with the dentary, a feature shared only with T. pusillus. The ventral side of the skull and mandibular region is not preserved in Atoposaurus jourdani, and this character state might also be present in that taxon too. Therefore, we are cautious in our retention of Atoposaurus oberndorferi as a distinct, second species of Atoposaurus.

Alligatorellus Gervais, 1871 Included species

Alligatorellus beaumonti (Gervais, 1871), Alligatorel-lus bavaricus (Wellnhofer, 1971; sensu Tennant & Mannion, 2014).

Distribution

Late Jurassic of southern France and south-east Germany.

Previous diagnoses and comments Alligatorellus was diagnosed by Wellnhofer (1971) based on its overall size, the shape of its skull, and its relatively large orbits, features that are all more widespread amongst atoposaurids and other small-bodied neosuchians. Wellnhofer (1971) originally described two subspecies of Alligatorellus beaumonti,

based on relative sizes and differences and geographical distribution. Most recently, Tennant & Mannion (2014) documented a number of distinguishing characters between Alligatorellus beaumonti beaumonti from France and Alligatorellus beaumonti bavaricus from Germany, and reranked the latter to its own species, Alligatorellus bavaricus. Several of the diagnostic synapomorphies for Alligatorellus, presented below, might be related to ontogenetic factors, such as the heterogeneity of the cranial sculpting and the closed internal supratemporal fenestra (Joffe, 1967). However, these features could also be related to the proposed dwarfism' for atoposaurids, and there are other indicators that the available specimens of Alli-gatorellus represent a reasonably mature state of growth, such as neurocentral fusion and the degree of fusion of the cranial bones (Joffe, 1967; Tennant & Mannion, 2014).

Schwarz-Wings et al. (2011) referred a partial skeleton, MB.R.3632, from the early Tithonian of Franconia, Germany (Gravesia gigas ammonoid zone) to Alligatorellus sp., but Tennant & Mannion (2014) concluded that this specimen could only be referred to as Atoposauridae indet. In most of our analyses, this specimen is recovered as an indeterminate non-atoposaurid taxon (Figs 4, 7A). However, when we used implied weighting, this specimen groups with the other species of Alligatorellus within Atoposauridae (Fig. 6). Therefore, we tentatively regard its status as Alligatorellus sp. to be valid.

Revised diagnosis and discussion

(51) Cranial table sculpting composed of homogeneous, subcircular shallow pits (C2.2): The cranial sculpting pattern for Alligatorellus is distinct from that of Atoposaurus, which has a smooth dorsal surface, and from Alligatorium and Theriosuchus in which the sculpting is much more prominent. It is similar to Wannchampsus, which is also lightly sculpted. The reduction or lack of sculpting has been noted in smaller specimens of the basal mesoeu-crocodylian Zosuchus (Pol & Norell, 2004a,b), as well as the protosuchian Gobiosuchus (Osmolska, Hua & Buffetaut, 1997).

(52) Rostrum unsculpted or relatively less than the cranial table (C3.1): Similar to (S1), the sculpting of the rostrum is relatively light compared with Alligatorium meyeri, T. pusillus, and Wannchampsus. Distinct from these taxa, however, is how the degree of sculpting appears to decrease anteriorly for Alligatorellus, with a more prominent pattern on the cranial table, and almost no sculpting on the dorsal surface of the rostrum. This morphology is similar to that seen in the

paratype of Isisfordia, which is represented by an adult specimen (Salisbury et al., 2006), and Pachycheilosuchus, which is known from mature individuals (Rogers, 2003). In other taxa represented by mature specimens, such as T. pusillus and Rugosuchus (Wu, Cheng & Russell, 2001a), sculpting patterns are homogeneous across the entire dorsal surface of the skull. Alligatorium meyeri is unusual in that the degree of sculpting remains the same between the rostrum and cranial table, but anteriorly the subcircular pits become more elongated, a feature visible in the counterpart to its holotype specimen (MNHL 15462), and which helps to distinguish it from Alligatorellus.

(53) Closed internal supratemporal fenestra (C16.1): This feature refers to the lack of opening of the internal supratemporal fenestra, as noted by Wellnhofer (1971). In all other specimens we observed, the internal supratemporal fenestra is completely open. Joffe (1967) described the opening as slit-like' for T. pusillus, and regarded it as indicative of an immature individual. However, our observations of the paratype specimen (NHMUK PV 0R48330) did not confirm this, and the internal fenestrae appear to be fully open. Because of poor preservation, we were unable to determine whether the morphology of the internal supratemporal fenestra was open or closed in any specimen of Atoposaurus.

(54) Frontal maximal mediolateral width between the orbits narrower than maximal width of nasals (C97.1): This character state relates to the proportionally large size of the orbits, which occupy the majority of the mediolateral width of the dorsal surface of the skull, with a proportionally narrow interorbital region composed of the fused frontals. Although this feature is shared by many other neosuchians, including Theriosuchus and Wannchampsus, the frontals are distinctly narrower in Alligatorellus. In protosuchians, such as Protosuchus and Hoplosuchus, the mediolateral width of the frontal is broader than the nasal, because in these taxa the orbit is more laterally facing, and therefore does not occupy as much of the mediolateral width of the skull in dorsal view.

(55) Broad frontal anterior process with parallel lateral margins, not constricted (C109.0): This feature is distinct from the morphology described in (S4), and relates exclusively to the development of the frontals anteriorly to the anterior margin of the orbits, excluding the morphology of any frontal anterior process

where present. The broad anterior edge of the frontal with parallel lateral edges in Alligatorellus is similar to paralligatorids, T. pusillus, and Montsecosuchus, but contrasts with T. guimarotae in which the mediolat-erally constricted anterior portions of the frontals distinctly underlap the nasals (Schwarz & Salisbury, 2005).

(S6) Flat and unsculpted anterior supratemporal margins (C119.0): In Alligatorellus bavaricus, the supratemporal rims are flat and unsculpted along their entire medial edge, similar to protosuchians, Pachycheilosuchus, coelognathosuchians, Montsecosuchus, and Koumpiodontosuchus. However, in Alligatorel-lus beaumonti, there is a slight posterior development of the supratemporal margins, similar to Brillanceausuchus (specimen UP BBR 203). This is distinct from Alligatorium meyeri, Wannchampsus, and all species referred to Theriosuchus, in which the supratemporal rims are consistently well developed along their entire medial margin.

(S7*) Anterior process of squamosal extends to the posterior orbital margin (C144.0): This character state appears to be diagnostic for Alligatorellus, although we are cautious in this assignment, as the postorbital region is poorly preserved, and the exact morphology of the postorbital with respect to the other posterior periorbital elements is difficult to assess. However, in the holotypes of Alligatorellus beaumonti and Alligatorellus bavaricus, there is no notable suture on the dorsal surface of the skull table, lateral to the supratemporal fenestra, which would represent the suture between the posterior process of the postorbital and the anterior process of the squamosal. We therefore infer that the anterior process of the squamosal reached the posterior orbital margin.

(S8) Posterodorsal margin of parietals and squamosals completely covers posterodorsal occipital region, excluding the supraoccipital from the dorsal surface of the skull (C197.1): This feature was proposed by Tennant & Mannion (2014) to be autapomorphic for Alligatorellus, but also appears to be present in a range of neosuchian taxa (e.g. Acynodon adriaticus; Delfino et al., 2008b) in which the supraoccipital is excluded from the posterodorsal surface of the skull roof. We therefore consider it to only be locally diagnostic for Alligatorellus. In other mesoeucrocodylians, such as

Mahajangasuchus (Turner & Buckley, 2008),

the supraoccipital is broadly visible in the midline portion of the posterodorsal region of the skull, contacting the parietals.

(S9) Smooth mandibular external surface, lacking sculpting (C201.0): This feature is difficult to observe in Alligatorellus bavaricus owing to the dorsal flattening of the holotype specimen, although what is visible indicates that the mandible, much like the anterior portion of the skull, lacks any sculpting pattern, unlike the dorsal surface of the cranial table. This is distinct from Theriosuchus and Wannchampsus, in which the sculpting pattern on the external surface of the dentaries and posterior mandibular elements is similar to that of the dorsal surface of the skull.

(510) Proximal end of the radiale 'hatchet-shaped' (C290.1): This feature also characterizes MB.R.3632, and was used to refer this specimen to Alligatorellus (Schwarz-Wings et al., 2011). However, this morphology is also shared by Wannchampsus (Adams, 2014), but is not known in Theriosuchus specimens, owing to lack of preservation of the radiales. In other atoposaurids, the proximal end of the radiale is more expanded equidimensionally, lacking the asymmetry observed in Alligatorellus.

(511) Proportionally short metatarsal I relative to metatarsals II-IV (C303.1): In other atoposaurids, metatarsals I-IV are almost equidimensional, possibly reflecting different locomotor adaptations in Alligatorellus.

(S12*)Dorsal surface of dorsal osteoderms completely sculpted (C304.0), with parallel and straight anterior and posterior margins (C308.1), and a longitudinal ridge along entire lateral margin (C311.1 and C312.1): The utility of osteoderms in atoposaurid systematics, particularly regarding Alligatorellus, was discussed by Tennant & Mannion (2014). These authors noted that the mediolateral position and anteroposterior extent of the dorsal keel, and its serial variation along the axial column, are diagnostic for Alligatorellus, as well as for other crocodyliforms (e.g. teleosauroids, eusuchians) that preserve a dorsal series of paravertebral osteoderms.

(S13) Caudal osteoderms with smooth, nonserrated edges (C327.1): This morphology is similar to the osteoderms preserved for T. guimarotae and T. pusillus, but distinct from Brillanceausuchus and Montsecosuchus in which the margins of the caudal osteoderms are serrated. Serrated edges might also be present in caudal osteoderms of MB.R.3632,

based on at least three caudal osteoderms disassociated from the main osteoderm shield preserved on the specimen slab. However, we cannot discount the possibility that these elements are accessory dorsal osteoderms, as found in Montsecosuchus and in the proximal caudal series of Alligatorellus beaumonti.

Alligatorellus beaumonti Gervais, 1871 (type species) Alligatorellus beaumonti beaumonti Wellnhofer, 1971

Type locality and horizon

Unknown bed, Kimmeridgian (Late Jurassic); Cerin, Ain, France.

Type specimen

MNHL 15639, part and counterpart slabs comprising a near-complete and articulated skeleton, missing the distal-most caudal vertebrae (preserved as impressions) and part of the left forelimb. Parts of the skull roof and a large portion of the right maxilla, along with several axial fragments, are embedded into the counterpart slab.

Referred specimen

MNHL 15638, part slab comprising a near-complete and articulated skull and skeleton, missing just the distal-most caudal vertebrae, the right forelimb, and the distal left forelimb, all of which are preserved as impressions. The skull is exposed in ventrolateral aspect.

Previous diagnoses and comments Alligatorellus beaumonti was originally named by Gervais (1871) for two specimens from the Late Jurassic of Cerin, eastern France. Subsequently, Wellnhofer (1971) diagnosed these specimens as a distinct subspecies, Alligatorellus beaumonti beau-monti. This was based largely on size differences between these and coeval specimens from Eichstätt, south-east Germany, for which Wellnhofer (1971) erected the subspecies Alligatorellus beaumonti bavaricus (see below). Together these specimens have largely been regarded as representing a single taxon, Alligatorellus beaumonti, by subsequent workers (e.g. Buscalioni & Sanz, 1988; Schwarz-Wings et al., 2011). However, Tennant & Mannion (2014) redescribed the German remains, and observed a number of morphological differences with the French material. They provided a revised diagnosis for Alli-gatorellus beaumonti, and re-ranked the German material as Alligatorellus bavaricus.

Wellnhofer (1971) noted that as in Theriosuchus, the external nares in Alligatorellus beaumonti are

divided by an anterior projection of the nasals, a feature that also appears to be shared by Alligatorium meyeri and possibly Alligatorellus bavaricus, although the anterior-most portion of the snout in the holotype of the latter is damaged. Alligatorellus beaumonti is similar to Alligatorium meyeri in the presence of an unsculpted posterolateral lobe' of the squamosal, differing from Alligatorellus bavaricus in which the posterolateral corner of the squamosal instead displays orthogonal posterior and lateral edges. Buscalioni & Sanz (1988) suggested that another distinguishing feature between Alligatorellus beaumonti and Alligatorium meyeri is the contribution of the frontal to the supratemporal fenestra in the former; however, this feature is clearly also present in Alligatorium meyeri, and therefore cannot be used to distinguish the two taxa. The presence of a biserial osteoderm shield comprising singular sculpted osteoderms is not diagnostic for Alligatorellus (contra Wellnhofer, 1971), as it also characterizes both T. pusillus and Alligatorium meyeri. Tennant & Mannion (2014) proposed that the frontal width between the orbits being mediolaterally narrower than the nasals is an autapomorphy of Alligatorellus beaumonti; however, this condition is not considered to be diagnostic here, as it is also known in a wide range of neosuchian taxa, and the width of the paired nasals in Alligatorellus bavaricus might have been underestimated. Alligatorellus beaumonti also has the reversed condition to Alligatorellus bavari-cus, in that the anterior extension of the frontal exceeds the anterior margin of the orbits, similar to almost all other neosuchian taxa.

Revised diagnosis and discussion (S1*) Frontal with unsculpted anterior and posterior portions, and sculpted medial surface: We elected not to code this as a distinct character state from that of S3 in our matrix in order to avoid duplication of character states. Nonetheless, Tennant & Mannion (2014) identified this heterogeneity in sculpting pattern as distinct from other atoposaurids and Theriosuchus, and considered it to be autapomorphic of Alligatorellus beaumonti.

(52) Surface of rostrum notably less sculpted than cranial table (C4.1): See S2 for Alligatorellus for discussion of this character state.

(53) Relatively large lateral temporal fenestra, approximately 30% of the size of the orbit (C20.1): A lateral temporal fenestra of this size with respect to the orbit represents the intermediate condition in our analyses. This relatively large size is unique amongst atoposaurids, but is also shared with T. pusillus, Koumpiodontosuchus (Sweetman

et al. , 2015), and protosuchians, as well as the advanced neosuchians Shamosuchus (Pol et al., 2009), Isisfordia (Salisbury et al., 2006), and Brillanceausuchus. In other taxa, such as Allodaposuchus precedens, the lateral temporal fenestra approaches the size of the orbit (Buscalioni et al., 2001).

(S4) Smooth contact between maxilla and jugal (C51.2): As noted above, the pattern of sculpting on the anterior portion of the dorsal surface of the skull is diagnostic for the different species of Alligatorellus. In Alligatorellus bavaricus, the entire dorsal surface is lightly sculpted, but Alligatorellus beaumonti has a smooth contact between the maxilla and jugal, similar to Atoposaurus and Hoplosuchus, although in both of these taxa the entire external surface of the skull is not ornamented. This is distinct from T. guimarotae and T. pusillus, which both have a contact in which the external surface is sculpted to the same degree as the rest of the cranial table, and from Brillanceausuchus, 'T.' ibericus, 'T.' sympiestodon, and Montsecosuchus, in which the contact is heavily striated.

(S5*) Medial longitudinal depression on posterior portion of nasal and anterior portion of frontal (C74.1): This is diagnostic of Alligatorellus beaumonti as a local autapomorphy, but is also present in the goniopholidid Amphicotylus (Mook, 1942). This condition differs from that in Theriosuchus and a range of paralligatorids, including Brillanceausuchus, in which a distinct midline longitudinal crest develops.

(S6*) Posteromedial border of supratemporal fenestra forms a low sagittal rim (C119.1): This feature is considered to be locally autapo-morphic, as it is also present in Brillan-ceausuchus. Alligatorellus bavaricus and Atoposaurus have no supratemporal rim development, and the rims are strongly developed along the whole medial edge of the external supratemporal fenestra in Alligatorium meyeri, T. pusillus, and Wannchampsus (Adams, 2014).

(S7*) Smooth and unsculpted region on anterior portion of squamosal nearing orbit and posterolateral process of squamosal (C148.1): This feature appears to be locally diagnostic, but is also shared by Khoratosuchus (Lauprasert et al., 2009). For all other OTUs for which this feature could be scored, the pattern of sculpting did not change between the main body of the squamosal and the immediate postorbital region.

(S8*) Ratio of forelimb to hindlimb length high (0.8) (C180.2): This feature is diagnostic for Alligatorellus beaumonti amongst all OTUs for which this character could be scored. Atoposaurus oberndorferi and Alligatorellus bavaricus both have similar ratios, 0.78 and 0.76, respectively, but Atoposaurus jourdani is distinct, with a ratio of 0.63 (Table 3). However, this character state could not be scored for Theriosuchus, or the majority of our outgroup taxa, because of the relative rarity with which these specimens preserve associated and complete limb material. Therefore, although these unusual ratios are diagnostic amongst atoposaurids, we cannot determine whether they are unique or only local autapomorphies.

(S9) Ratio of tibia to femur length low (0.9) (C300.0): The relative dimensions of the tibia and femur are a feature that is closely shared with MB.R.3632 (0.91), Hoplosuchus (0.92), Alligatorium meyeri (0.93), and Atoposaurus jourdani (0.94). This ratio far exceeds that for Alligatorium' franconicum (0.64), and is distinct from Alligatorellus bavaricus (0.96) (Table 3). This feature was also noted by Tennant & Mannion (2014), but those authors used a ratio of femur to tibia length.

(S10*) Nuchal osteoderms reduced, <50% of the size of the dorsal osteoderms (C307.1): This feature is distinct from the condition in Alligatorellus bavaricus, Alligatorium meyeri, Montsecosuchus, and Protosuchus, in which the preserved nuchal osteoderms retain the same size and morphology as the dorsal series, or only decrease slightly.

(S11*) Dorsal keel in dorsal osteoderms shifts laterally in more posterior dorsal osteoderms (C317.1): The position of the dorsal keel on the dorsal osteoderm series is distinct from that in Alligatorellus bavaricus, in which the morphology is more consistent along the axial column (Tennant & Mannion, 2014). This feature is not present in any other of the OTUs that we sampled, and therefore we consider it to be diagnostic for Alligatorellus beaumonti.

(S12*) Lateral ridge on sacral osteoderms forms an incipient posterior projection: The posterior development of the lateral keel (as noted in S9) into an incipient lateral projection amongst the more sacrally positioned osteoderms is diagnostic for Alligatorellus beaumonti. In Alligatorellus bavaricus, the morphology of the keel does not change anteroposteriorly (Tennant & Mannion, 2014),

and Theriosuchus, Alligatorium meyeri, and higher neosuchians do not seem to possess this keel at all. Alligatorium' franconicum and Hoplosuchus are convergently similar, in that the lateral keel appears to form an anterolateral process, distinct from the peg and socket' articulation described for goniopholidids and T. guimarotae (Schwarz & Salisbury, 2005). We did not incorporate this as a character to avoid duplication and overweighting of the observation that the morphology of the dorsal keel changes axially in Alligatorellus beaumonti (S11). (S13*) Secondary osteoderms present in caudal series (C328.0): This feature does not appear to be present in any other atoposaurid that preserves caudal osteoderms.

Additional comments

In the holotype specimen of Alligatorellus beaumonti, the posterior-most maxillary teeth have a more labi-olingually compressed, apically pointed morphology than the remaining teeth, similar to the lanceolate' morphology exhibited by several species of Therio-suchus (Schwarz & Salisbury, 2005; Lauprasert et al., 2011) and Brillanceausuchus, as well as the bernissar-tiid Koumpiodontosuchus (Sweetman et al., 2015). This is different to the homodont dentition typically reported for Alligatorellus, which is usually described as possessing simple pseudocaniniform teeth that are smooth and lack ridges or carinae (e.g. Buscalioni & Sanz, 1990a,b; Thies et al., 1997). However, we do not assign this as a local autapomorphy of Alligatorellus beaumonti as it is only visible for one or two teeth in a specimen that shows strong evidence of dorsal compression. Its validity therefore requires further investigation pending the discovery of more specimens of Alligatorellus. Although this more lanceolate morphology was also figured for the posterior teeth of Alligatorellus bavaricus by Wellnhofer (1971), we have been unable to personally validate this on the figured specimen, and it is not visible on the holotype. Therefore, we urge caution in interpreting Alligatorellus as possessing lanceolate posterior teeth that are homologous to those found in Theriosuchus. Re-running our phylogenetic analysis [excluding Alligatorellus' sp. (MB.R.3632) and Pachycheilosuchus as before] with Alligatorellus scored as possessing lanceolate posterior teeth, we achieve a single MPT with a length of 793 steps with an unchanged topology. However, the presence of a lanceolate dentition instead becomes the basal condition in the clade containing atoposaurids and higher neosuchians, secondarily lost in Alligatorium meyeri, Brillanceausuchus, 'T.' ibericus, 'T.' sympiestodon, and the clade containing Koumpi-odontosuchus and coelognathosuchians.

Alligatorellus bavaricus Wellnhofer, 1971 (re-ranked by Tennant & Mannion, 2014) Alligatorellus beaumonti bavaricus Wellnhofer, 1971

Type locality and horizon

Solnhofen Formation, early Tithonian (Late Jurassic, Hybonoticeras hybonotum zone); Eichstatt, southeast Germany.

Type specimen

BSPG 1937 I 26, a near-complete skeleton including the skull, lacking only the left forelimb, compressed onto a slab of lithographic limestone. Note that Ten-nant & Mannion (2014) incorrectly stated that the specimen number was LMU 1937 I 26.

Referred specimen

A specimen held in the private collection of E. Schopfel was described and referred to Alligatorellus bavaricus by Wellnhofer (1971), from the Wintershof Quarry (Solnhofen Formation, Eichstätt, southeast Germany).

Revised diagnosis and discussion

(51) Concave profile of dorsal surface of snout in lateral view (C8.0): This feature represents a reversion back to the plesiomorphic state known for Protosuchus (Colbert & Mook, 1951) and Hoplosuchus (Gilmore, 1926), with other atoposaurids and neosuchians usually presenting a straight profile in lateral aspect (with exceptions such as the longirostrine goniopholidid Amphicotylus; Mook, 1942).

(52) Small, slit-shaped antorbital fenestra, enclosed by nasals (C13.0 and C14.1): Alligatorellus bavaricus appears to possess a small, slit-like antorbital fenestra, similar to the notosuchians Gondwanasuchus (Marinho et al., 2013) and Malawisuchus (Gomani, 1997). In other taxa with an antorbital fenestra, including T. guimarotae (Schwarz & Salisbury, 2005), T. pusillus, and 'T.' ibericus, it is proportionally larger and rounded. Alligatorellus beaumonti does not appear to possess an antorbital fenestra, although part of the snout is embedded in the counterpart slab, with a small opening observable near the posterior margin of the nasals, which could be a diminutive fenestra. The presence of an antorbital fenestra is documented in basal crocodyliforms, including the protosuchians Hoplosuchus (Gilmore, 1926) and Protosuchus haughtoni (Gow, 2000), Zosuchus (Pol & Norell, 2004a), and thalattosuchians (Leardi

et al. , 2012), but becomes closed in shartegosuchids, including Fruitachampsa (Clark, 2011). Pachycheilosuchus might have also possessed an antorbital fenestra (Rogers, 2003).

(S3*) Extremely narrow and short skull, with a low skull width to orbit width ratio (< 2.0) (C28.0): This character state is the lowest value for all atoposaurids, and much lower than all other OTUs in which this character was measurable. For other atoposaurids, this ratio is between 2.15 (Alligatorellus beaumonti) and 2.69 (Atoposaurus jourdani), and the only other taxon that comes close to this range is Wannchampsus (2.77). Brillanceausuchus, Montsecosuchus, and Theriosuchus species all have ratios between 3.0 and 3.5, with the ratio being considerably greater in longirostrine taxa and protosuchians (Table 3). It is likely that this character state is influenced by ontogeny (Joffe, 1967), but the broad distribution of ratios amongst the sampled oTUs, which possess a range of body sizes and ontogenetic states, means that ontogeny is unlikely to entirely control this feature.

(S4) Skull anteroposterior length to supratemporal fenestra length ratio ~7.2 (7.18) (C29.2): The proportional length of the external supratemporal fenestra is similar to Wannchampsus kirpachi (7.53), but distinct from Alligatorium meyeri (6.43) and Alligatorellus beaumonti (6.23), which have proportionally larger external supratemporal fenestrae. Consequently, we consider the proportionally short anteroposterior length of the supratemporal fenestra to skull length to be diagnostic for Alligatorellus bavaricus, because in Montsecosuchus this ratio is considerably higher (8.9) (Table 3), with a much smaller supratemporal fenestra.

(S5*) Posterior surface of nasals longitudinally crenulated (C69.0): The longitudinal crenulations on the dorsal surface of the nasals are not known in any other crocodyliform, in which the nasals are dorsally flat and sculpted like the rest of the cranial dorsal surface.

(S6*) Smooth anterior region of parietal dorsal surface with a transverse frontal-parietal ridge, and shallow emargination at the posterior parietal-squamosal contact (C117.1) that develops into a thin dorsal groove connected to the supratemporal fenestra (C147.1): The morphology of the parietal is diagnostic, with a small anterior concavity at

the posterodorsal suture contact between the parietal and squamosal, leading to a shallow sulcus along this contact into the posterior margin of the supratemporal fenestra, and a smooth anterior dorsal surface. This is distinct from the condition observed in Theriosuchus in which this contact is deep and expands mediolaterally towards the supratemporal fenestra border, and from Alligatorellus beaumonti and Alligatorium meyeri in which the grooved contact is bordered by raised crests. The presence of a transverse ridge at the parietal-frontal suture distinguishes Alligatorellus bavaricus from all other species, in which this suture is flat.

(S7) Squamosal posterolateral lobe absent (C139.1):. The squamosal posterolateral lobe is completely absent in Alligatorellus bavaricus, a feature considered to be diagnostic amongst all atoposaurids.

(S8*) Distinct ridge on proximodorsal edge of scapula (C280.1): The scapula of Alligatorellus bavaricus can be distinguished from Alligatorellus beaumonti and other atoposaurids based on the presence of a distinct ridge on the proximodorsal surface. In all other specimens we analysed, the proximodorsal edge of the scapula is flat in lateral view, and confluent with the scapular shaft.

(S9) Extremely low radius proximodistal length to humerus length ratio (0.69) (C288.1): The radius to humerus ratio is extremely low, distinct from other atoposaurids in which the value is closer to 1.0. This low ratio is identical to that for 'Alligatorium' franconicum and Karatausuchus (Storrs & Efimov, 2000), but higher than in Pachycheilosuchus (0.58).

(S10*) Low radius proximodistal length to tibia length ratio (0.64) (C289.1): This value is almost identical to that for Montsecosuchus, Pachycheilosuchus, Atoposaurus jourdani (0.61), and Protosuchus richardsoni (0.63), but much higher than that for Karatausuchus (0.47) and T. pusillus (0.55). Other taxa have proportionally long radii, including Alligatorium meyeri (0.74), Alligatorellus beaumonti (0.71),

'Alligatorium' franconicum (0.89), and Brillanceausuchus (0.88).

(S11*) Dorsal osteoderms with longitudinal medial ridge, becoming more laterally placed anteriorly (C311.1 and C312.1): This feature pertains to the morphology of the dorsal osteoderm series, which are distinct from

those in Alligatorellus beaumonti (Tennant & Mannion, 2014), as well as MB.R.3632 (Schwarz-Wings et al., 2011). In Alligatorium meyeri and T. pusillus, there is no evidence of a lateral keel.

Putative atoposaurid taxa

In this section, we provide comments on the systematic position of taxa that historically have been attributed to Atoposauridae, but are here recovered as non-atoposaurids. We provide emended diagnoses for Montsecosuchus, ''Alligatorium' franconicum, Theriosuchus, 'T.' ibericus, 'T.' symp-iestodon, and Brillanceausuchus, for which a non-atoposaurid position is novel to our study, and discuss the taxonomic affinities of taxa that have previously been recognized as non-atoposaurids (e.g. Hoplosuchus).

Neosuchia Benton & Clark, 1988 Montsecosuchus Buscalioni & Sanz, 1988

included species Montsecosuchus depereti.

Revised diagnosis

As for the type and only species.

Distribution

Early Cretaceous of Spain.

Previous diagnoses and comments Originally described as a species belonging to Alli-gatorium by Vidal (1915), the differences from Alligatorium were first noted by Buffetaut (1981), and subsequently formalized in the erection of the new genus by Buscalioni & Sanz (1988). Montsecosuchus is unusual in its relatively robust and shortened forelimbs with respect to its hindlimbs, including a large, transversely expanded distal humerus and proportionally small manus (Buscalioni & Sanz, 1990a). It was originally assigned to Atoposauridae based on its overall size, and is similar to other atoposaurids in the hatchet shaped' radiale morphology (Buscalioni & Sanz, 1988, 1990a), but distinct from Atoposaurus and Alligatorellus in the absence of a reduced fifth metatarsal (Buscalioni & Sanz, 1990a).

Buscalioni & Sanz (1988) found Montsecosuchus to be either the sister taxon to Theriosuchus, or to (Alligatorium + Alligatorellus), with this uncertainty reflecting the unusual morphology of Montseco-suchus. Many of the autapomorphies defined by Buscalioni & Sanz (1988) are metric, and therefore

might not be solely reliable in generic-level diagnoses within a group in which there is much uncertainty over ontogenetic allometry and potential dwarfism (Joffe, 1967; Schwarz-Wings et al., 2011; Martin et al., 2014a; Tennant & Mannion, 2014). The only recent formal phylogenetic analysis to include Montsecosuchus found it to be the sister taxon to T. guimarotae, to the exclusion of T. pusillus (Fig-ueiredo et al., 2011), although this analysis was not designed to assess the relationships between putative atoposaurids (see also Bronzati et al., 2012). Our results mostly recover Montsecosuchus as a non-ato-posaurid taxon, but also outside of Paralligatoridae, in an uncertain position along with Pachycheilo-suchus (Figs 4B, 5A). Our analysis was not designed to constrain the phylogenetic position of non-atopo-saurids, but to recover the composition and relative position of Atoposauridae, and we consider Montseco-suchus to be Neosuchia incertae sedis. However, in the results of the analysis using implied weights, Montsecosuchus nestles within Atoposauridae (Fig. 6). These inconclusive results warrant further comparison between Montsecosuchus and other neo-suchians, to determine its affinities. Therefore, we await the inclusion of Montsecosuchus in analyses covering broader neosuchian relationships (e.g. Adams, 2014; Turner, 2015) to resolve its phyloge-netic relationships.

Sanz, Ortega & Shibata (2014) identified a specimen as Montsecosuchus sp. from the late Barremian Hueorgina Formation of Cuenca, Spain, but we have not observed this specimen directly so cannot comment on this further. Examination of this material, and further discoveries of additional material - particularly of the basicranial region - will be important in determining the relationships of this enigmatic taxon.

Montsecosuchus depereti Buscalioni & Sanz,

Alligatorium depereti Vidal, 1915 Type locality and horizon

Le Pedrera de Rubies Formation, late Berriasian-early Barremian (Early Cretaceous); Sierra del Mon-tsec, Leorida Province, Spain.

Type specimen

MGB 512, near-complete skeleton and skull, and counterpart MGB 597.

Revised diagnosis and discussion (S1) Intertemporal mediolateral width greater than interorbital width (C19.1): The relatively high proportion of the interorbital relative to

intertemporal region is a feature shared with 'Alligatorium' franconicum, Alligatorellus beaumonti (Wellnhofer, 1971; Tennant & Mannion, 2014), and Pachycheilosuchus (Rogers, 2003). In protosuchians, coelognathosuchians, paralli-gatorids, Alligatorium meyeri, and T. pusillus, the interorbital width exceeds the width of the intertemporal region. It is unlikely that variation in this feature is exclusively a result of relative growth differences through ontogeny in all of these species, as there are multiple additional lines of evidence that indicate that many specimens had reached skeletal maturity (e.g. neurocentral fusion).

(S2*) Intermandibular angle 61° (C24.3): Montsec-osuchus has an extremely anteroposteriorly short and mediolaterally wide skull for its body size, as noted by Buscalioni & Sanz (1990a). The only other taxon to come close to this characteristically wide intermandibular angle is Atoposaurus jourdani (55°), in which this state could be a result of an allometric growth factor (Schwarz-Wings et al., 2011; Tennant & Mannion, 2014). Other brevirostrine taxa, including Wannchampsus, Brillanceausuchus, Shamosuchus, and T. pusillus, have intermandibular angles in the range of 40-45°. Longirostrine taxa, including Koumpiodon-tosuchus and Amphicotylus, have a much lower intermandibular angle of 32-34° (Table 2).

(53) Skull anteroposterior length to supratemporal fenestra length ratio 8.9 (C29.3): This dimension reflects the extremely small external supratemporal fenestra of Montsecosuchus despite its small skull size, and is similar to Amphicotylus (8.57) and Koumpiodontosuchus (8.0), as well as possibly Karatausuchus (8.41). Brillanceausuchus represents the opposite end of the spectrum, with a proportionally longer supratemporal fenestra (ratio of 5.36; Table 3). Atoposaurids, T. guimarotae, T. pusillus, and Protosuchus richardsoni fall within a range of around 67.5.

(54) Longitudinal ridge on the jugal below lateral temporal fenestra (C87.1): The presence of a longitudinal ridge on the lateral surface of the jugal, just below the lateral temporal fenestra, is shared with T. guimarotae (Schwarz & Salisbury, 2005) and 'T.' ibericus. In atoposaurids, the lateral margin of the jugal is smooth, although this cannot be observed in the holotypes of Alligatorellus beaumonti or Atoposaurus jourdani owing to the dorsal flattening of these specimens.

(S5*) Lateral border of the skull roof terminates immediately dorsal to the medial-most point of contact with the quadrate (C118.1): Montseco-suchus possesses a very mediolaterally narrow dorsal skull roof compared with the infratemporal region, similar to the eusuchians Acynodon (Delfino, Martin & Buffetaut, 2008a) and possibly Hylaeochampsa (Clark & Norell, 1992). The lateral extent of the skull roof with respect to the contact with the quadrate is a feature that we consider to be locally diagnostic for Montsecosuchus.

(S6*) Flat and ungrooved parietal-squamosal suture (147.0): The lack of a parietal-squamosal sutural groove is distinct for Montsecosuchus amongst all specimens that we scored, similar to Hylaeochampsa (Clark & Norell, 1992). In atoposaurids, there is a thin groove occupying the suture, flanked by slightly raised ridges, and in Theriosuchus the groove is deeper and expands anteriorly towards the posterior border of the supratemporal fenestra.

(57) Supraoccipital exposed medially in pos-terodorsal surface of skull roof (C197.0): The dorsal exposure of the supraoccipital in the posterior margin of the skull roof is a feature shared with T. pusillus, T. guimarotae (Schwarz & Salisbury, 2005), Wannchampsus (Adams, 2014), and Brillanceausuchus. However, in the latter two taxa, the supraoccipital is restricted to a thin surface attached to the posterior-most portion of the parietal and squamosal, and is not as well exposed as it is in Montsecosuchus, Theriosuchus, and Mahajangasuchus (Turner & Buckley, 2008). In Shamosuchus (Pol et al., 2009), protosuchians, goniopholidids, atoposaurids, and Hylaeochampsa (Clark & Norell, 1992), the supraoccipital is not exposed dorsally in the posterior margin of the skull roof.

(58) Posteriorly domed occipital surface comprising the medial portion of the exoccipitals: The posteriorly domed occipital region is autapomorphic for Montsecosuchus (the dolichocephalous' condition, sensu Buscalioni & Sanz, 1990a), whereas in other taxa this surface is flat and faces posteriorly or posteroventrally. However, we are cautious in our interpretation of this character state as autapomorphic, following Buscalioni & Sanz (1990a), as it is clear that this specimen has undergone a degree of dorsoventral flattening, and therefore a component of this character could pertain to the displacement of the exoccipitals.

(S9*) Posteriorly projecting and dorsally recurved retroarticular process (C242.3): This morphology is shared only with Brillanceausuchus, and we consider it to be a local autapomorphy for Montsecosuchus. The morphology of the retroarticular process appears to be highly phylogenetically informative, with taxa such as Alligatorellus having a posteriorly projecting but ventrally recurved form, similar to Simosuchus (Buckley et al., 2000) and Stolokrosuchus (Sereno et al. , 2003), whereas in T. pusillus and T. guimarotae the process projects posteroventrally and is paddle-shaped' (Schwarz & Salisbury, 2005; Pol et al., 2009), similar to Rugosuchus (Wu et al., 2001a) and notosuchians (Gomani, 1997). More advanced neosuchians appear to have an extremely reduced, or completely absent, retroarticular process (e.g. Wannchampsus and Shamosuchus; Pol et al., 2009), a condition similar to Protosuchus (Colbert & Mook, 1951) and the notosuchian Yacarerani boliviensis (Novas et al., 2009).

(S10*) Preacetabular (anterior) process of the ilium absent (C291.2): The absence of the preacetabular process on the ilium is a feature that is shared with Alligatorium' franconicum (Wellnhofer, 1971). This process is extremely reduced in T. guimarotae (Schwarz & Salisbury, 2005), Pachycheilosuchus (Rogers, 2003), and a specimen described as Theriosuchus sp. (IVPP V10613) by Wu et al. (1996), as well as the giant crocodyliform Sarcosuchus imperator (Sereno et al., 2001), being 75% or less of the length of the postacetabular process. The reduction of the iliac anterior process is also the condition for notosuchians (Buckley & Brochu, 1999; Pol, 2005; Turner, 2006).

(S11) Three sacral vertebrae (C274.1): Montsecosuchus is unusual in that it appears to have three sacral vertebrae, a feature that seems to be shared exclusively with Alligatorellus (Buscalioni & Sanz, 1990a) within Neosuchia (Table 2), and could relate to reconfiguration of the pelvic girdle (also see S10) owing to mechanical requirements for adaptation to a more terrestrial mode of life. Other occurrences of a third sacral vertebra amongst crocodyliforms are documented in notosuchians (e.g. Pol, 2005; Riff & Kellner, 2001; 2011).

(S12*) Dorsal osteoderms not imbricated (C313.1) or sutured (C314.1), oval-shaped, and with

medially placed anteroposterior keel on dorsal surface (C311.1 and C312.1): The osteoderms of Montsecosuchus are distinct from those assigned to Theriosuchus and atoposaurids, in that they appear not to contact each other, forming two evenly spaced rows. The shape and spacing are somewhat similar to some of the dorsal osteoderms observed in Brillanceausuchus. Cervical osteoderms are not preserved, and there is no evidence of an anterolateral process. In Alligatorellus, Alligatorium, Pachycheilosuchus, and Theriosuchus, the osteoderms are subrectangular to square shaped, and form a distinct dorsal shield. Similar to Alligatorellus is the presence of the anteroposterior dorsal keel (Tennant & Mannion, 2014), but this appears to have a uniform morphology anteroposteriorly along the axial column in Montsecosuchus. (S13*) Accessory osteoderms present in dorsal series (C316.1): Montsecosuchus also possesses accessory osteoderms that do not contribute to the main dorsal dermal shield. The presence of accessory osteoderms is also known in a range of mesoeucrocodylians, including dyrosaurids (Schwarz-Wings, Frey & Martin, 2009a), the hylaeochampsid Pietraroiasuchus ormezzanoi (Buscalioni et al., 2011), the advanced neosuchians Susisuchus anatoceps (Figueiredo et al., 2011) and Isisfordia (Salisbury et al., 2006), and the eusuchian Acynodon adriaticus (Delfino et al., 2008a). (S14*) Caudal osteoderms oval-shaped (C326.0) with serrated lateral edges (C327.0): The caudal osteoderms have an oval profile in dorsal view, a feature not known in any atoposaurid or Theriosuchus. The serration of the lateral edges of each caudal osteoderm is also diagnostic when combined with the overall morphology of the caudal series, but is a feature shared with Brillanceausuchus and Pachycheilosuchus (Rogers, 2003).

Neosuchia Benton & Clark, 1988 Theriosuchus Owen, 1878a

Included species

Theriosuchus guimarotae (Schwarz & Salisbury, 2005), T. grandinaris (Lauprasert et al., 2011), T. pusillus (Owen, 1878a, 1879). We exclude 'T.' ibericus (Brinkmann, 1989) and 'T.' symp-iestodon (Martin et al., 2010) from this genus (see below).

Distribution

Late Bajocian/Bathonian (Middle Jurassic) to Ceno-manian (early Late Cretaceous) of western Europe; Bathonian (Middle Jurassic) of Morocco; Early Cretaceous (possibly latest Jurassic too) of Asia.

Note on taxonomy

Although generally attributed to Owen (1879), T. pusillus was actually first named by Owen (1878a) - see discussion of the type species below. In all of our analyses, we find Theriosuchus to be poly-phyletic within Neosuchia. Theriosuchus pusillus and T. guimarotae are sister taxa, and form a clade with T. grandinaris (Lauprasert et al., 2011) and Theriosuchus sp. (NMS G. 2014.52.1; Young et al., 2016). The most surprising result is that Therio-suchus does not group with other atoposaurids, and is more closely related to more crownwardly placed neosuchians (i.e. paralligatorids). 'Theriosuchus' iber-icus and T.' sympiestodon form a clade that is separated from the other species, and which occupies a position nested within paralligatorids. Theriosuchus pusillus was the first named species of this genus (Owen, 1878a,b, 1879), and therefore retains taxo-nomic priority for the genus name. Consequently, we erect a new genus name for 'T.' ibericus and T. ' sympiestodon (see below), and our revised diagnosis presented below is exclusively for Therio-suchus, comprising T. grandinaris, T. guimarotae, and T. pusillus (as well as remains attributed to Theriosuchus sp.; e.g. Young et al., 2016).

Previous diagnoses and comments Theriosuchus was first identified by Owen (1878a,b, 1879), based on two specimens from the Purbeck beds of England. The genus was first referred to Ato-posauridae by Joffe (1967), who noted numerous similarities between T. pusillus and atoposaurids from western Europe. Since then, its position has varied within Neosuchia, being positioned either as one of the basal-most taxa within Atoposauridae, or more recently in a much more advanced position as the sister taxon to Paralligatoridae, within Eusuchia (Turner, 2015; Turner & Pritchard, 2015). Jouve, Larochene Bouya & Amaghaz (2006) also noted the similarities between Theriosuchus and other advanced' neosuchians, including Rugosuchus and Shamosuchus, finding them to be closely related to a clade comprising bernissartiids, hylaeochampsids, and crocodylians. Although there are five named species of Theriosuchus, typically only T. pusillus (Owen, 1878a, 1879) has been included in phyloge-netic analyses involving Neosuchia, sometimes with Alligatorium meyeri as a further representative of Atoposauridae. Exceptions to this comprise the analyses of Martin et al. (2010) and Turner (2015; see

also Turner & Pritchard, 2015), which also included T. guimarotae and 'T.' sympiestodon. Whereas the analysis of Martin et al. (2010) resulted in a mono-phyletic Theriosuchus, Turner & Pritchard (2015: fig. 7) recovered Theriosuchus as paraphyletic with respect to Alligatorium meyeri, when a series of alternative palatal character state scores for Isisfordia duncani (Salisbury et al., 2006) (a possible non-eusuchian susisuchian; Turner, 2015) were applied. To our knowledge, neither 'T.' ibericus nor T. grandi-naris have ever been included in a formal phyloge-netic analysis. Therefore, the monophyly of Theriosuchus has never been fully examined, and nor have the character states that support this been tested.

Diagnoses of Theriosuchus have varied since the original description by Owen (1879). In his unpublished thesis, Clark (1986) provided a comprehensive redescription of T. pusillus, as well as an emended diagnosis. Brinkmann (1992) provided a diagnosis for Theriosuchus, but this was based only on T.' ibericus and T. pusillus. Schwarz & Salisbury (2005) included a diagnosis of Theriosuchus in their description and naming of T. guimarotae, which was repeated by Karl et al. (2006) in their description of a poorly known skull and partial skeleton from Germany, which they cautiously attributed to T. pusillus. However, Schwarz & Salisbury (2005) did not discuss 'T.' ibericus in detail (Brinkmann, 1989, 1992), and this was prior to the identification of T. grandinaris (Lauprasert et al., 2011) and 'T.' sympiestodon (Martin etal., 2010, 2014a). Salisbury & Naish (2011) presented the most comprehensive recent diagnosis of Theriosuchus, listing the following character states: (1) proportionately short and broad rostrum, with the maxillary rostrum forming between 40 and 45% of the total skull length; (2) proportionately small antorbital fenestra; (3) slit-like, horizontally orientated and rostrally pointed external nares, separated from each other by the rostral-most extent of the nasals; (4) shallow sulcus on the dorsal surface of the maxillary rostrum, immediately posterior to the junction between the maxilla, premaxilla, and nasal; (5) proportionally long jugal; (6) medial base of the postorbital process formed by the ectoptery-goid; (7) median crest on the frontal and the parietal in later ontogenetic stages; (8) frontal and parietal partially unfused in early ontogenetic stages; (9) dorsal margin of the supratemporal fenestra smaller than the orbit throughout ontogeny; (10) lateral margin of the squamosal bevelled ventrally; (11) proportionally narrow quadrate with a concave mandibular articular surface; (12) secondary choanae bounded by the palatines rostrally and separated by a median septum of the pterygoids; (13) mandibular symphysis that does not extend posteriorly beyond a point level

with the sixth dentary tooth; (14) ilium with short preacetabular process and long postacetabular process; and (15) biserial dorsal shield comprising parasagittal osteoderms. However, it was not clear how all of these character states are distributed across the five named species of Theriosuchus. Furthermore, many of these characters can be demonstrated to be more broadly present in Atoposauridae, or characterize smaller subgroups within Therio-suchus. For example, feature 1 is consistently present in all small, brevirostrine crocodyliforms. The presence of an antorbital fenestra is variable (feature 2), with T. guimarotae clearly possessing a large fen-estra, 'T.' sympiestodon and 'T.' ibericus possibly retaining one, and T. pusillus having a pinhole and dorsally placed antorbital fenestra. The presence of an antorbital fenestra is further documented for Pachycheilosuchus (Rogers, 2003), and possibly Alligatorellus bavaricus (Tennant & Mannion, 2014). The division of the external nares by an anterior projection of the nasals (feature 3) appears to be the condition for Alligatorium meyeri and ''Alligatorium' franconicum, as well as Alligatorellus (Tennant & Mannion, 2014) and Wannchampsus (Adams, 2014). Feature 4 might be diagnostic for Theriosuchus, as it was also documented by Schwarz & Salisbury (2005) for T. guimarotae, and by Martin et al. (2010) for 'T.' sympiestodon, with this sulcus in a position posterior to the maxilla-nasal-premaxilla triple junction; however, despite first-hand examination of the type specimens, we have been unable to locate this sulcus on T. pusillus, 'T.' sympiestodon, or T.' ibericus. Therefore, this feature might be diagnostic only for T. guimarotae. We discuss these characters in more detail for each taxon below. These features have all been incorporated into the present analysis to test whether they are more broadly present in crocodyliforms, or can be used to diagnose Theriosuchus or a subset of species within Theriosuchus.

The dentition of Theriosuchus

Species previously assigned to Theriosuchus have four dentition-based morphotypes (Owen, 1879; Joffe, 1967; Brinkmann, 1992; Salisbury, 2002; Schwarz & Salisbury, 2005; Schwarz-Wings, Rees & Lindgren, 2009b; Lauprasert et al., 2011; Salisbury & Naish, 2011), which are typically structured from an anteriorly to more posteriorly position in the following sequence: (1) slender and conical teeth with api-cobasally aligned striations that are largely restricted to the lingual face of the crown, located in the premaxilla, and the rostral-most maxilla and dentary (pseudocaniniform morphotype); (2) lanceolate morphotype, moderately labiolingually compressed, with a radial distribution of the marginal

lingual striations and mesial and distal carinae, situated in the middle and posterior portions of the maxilla and dentary; (3) labiolingually compressed morphotype, in which teeth are broad and strongly labiolingually compressed, with both the lingual and labial surfaces covered with vertical, straight, and subparallel striations (although fan-shaped striations are present only on the lingual face; Thies et al., 1997); and (4) a 'low-crowned' tooth morphotype that is characterized by the apical margins being orientated at <45° from the horizontal, forming a crown that is as mesiodistally broad (or broader) as it is apicobasally tall, and more posteriorly placed in the dental arcade.

In all of these morphotypes, there is variation within the profile shape, size, striation development and strength of carinae, degree of lingual curvature through asymmetrical compression, and shape of the transverse section (e.g. Thies et al., 1997). Theriosuchus guimarotae possesses pseudocaniniform and lanceolate tooth morphotypes, all of which exhibit mesial and distal carinae (Schwarz & Salisbury, 2005). Theriosuchus grandinaris possesses a combination of pseudocaniniform, lanceolate-shaped, and labiolingually flattened teeth with faintly crenulated mesial and distal carinae (Lauprasert et al., 2011). The strongly labiolingually compressed morphotype appears to be restricted to T. pusillus and 'T.' ibericus. The low-crowned morphotype is only known in 'T.' ibericus, T. pusillus, and 'T.' sympiestodon (Martin et al., 2010).

This low-crowned' morphology is distinct from the 'low-crowned' tribodont dentition of Bernissartia and Koumpiodontosuchus (Buffetaut & Ford, 1979; Sch-warz-Wings et al., 2009b; Sweetman et al., 2015), in which the teeth are multicusped. However, both morphologies probably had a similar function in crushing harder prey items (e.g. molluscs). This niche specialization fits in with the ecology and geographical distribution of bernissartiids and Theriosuchus, as these taxa represent crocodylomorphs of reduced body size constrained to island environments. In all four dental morphotypes, the apical edges range from smooth, to faintly crenulated or serrated, to possessing well-developed carinae. Ornamentation varies, but includes apicobasally orientated longitudinal ridges on the labial and lingual surfaces of the crown, sometimes more developed on the labial side, and with variation in the regularity of spacing between ridges. It is the presence of crenulations, formed from the faint ridges on the crown, that has been used to ascribe atoposaurids and Theriosuchus with their characteristic pseudoziphodont' morphology (Prasad & De Lapparent De Broin, 2002). 'The-riosuchus' ibericus is distinct in possessing more prominent serrations on the mesial and distal tooth

margins, approaching the fully ziphodont condition. The close packing of the maxillary and dentary alveoli in Theriosuchus is similar to a range of neo-suchian taxa, including goniopholidids, Bernissartia (Buffetaut & Ford, 1979), Wannchampsus (Adams, 2014), and Shamosuchus (Pol et al., 2009), but is distinct from Pachycheilosuchus, Rugosuchus, and cro-codylians, in which the alveoli are consistently well separated by interalveolar septae (Wu et al., 2001a; Rogers, 2003; Pol et al., 2009).

This dental variation has led to several differing hypotheses as to the diet of Theriosuchus, including the consumption of small mammals (Owen, 1879) or insects (Buscalioni & Sanz, 1988), to herbivory (Brinkmann, 1992), ovivory (Kirkland et al., 1994), or piscivory (Thies et al., 1997), all based around a semi-aquatic or amphibious mode of life. There is sufficient evidence to accept all of these as valid hypotheses, suggesting that Theriosuchus was adept at adapting to take advantage of whichever trophic style fitted its ecological position. However, it is likely that, based on our results, that such a morphological or dietary plasticity evolved at least twice independently within advanced neosuchian lineages, or represents a highly adaptive continuum. Finally, it is worth noting that amongst definitive atoposaurids, Alligatorium and Alliga-torellus both exclusively have smooth-surfaced teeth lacking ridges (Wellnhofer, 1971), and the teeth of Atoposaurus and Montsecosuchus are still unknown (Buscalioni & Sanz, 1988, 1990a; Thies et al., 1997).

Young et al. (2016) identified some of the oldest known diagnosable remains of Theriosuchus, and provided a list of dentary synapomorphies that might diagnose Theriosuchus. This included the presence of a heterodont dentition, with a combination of pseu-docaniniform, labiolingually compressed, and lanceolate (or leaf-shaped') tooth crown morphotypes (Schwarz & Salisbury, 2005; Lauprasert et al., 2011), which have never previously been incorporated within a phylogenetic analysis that includes ato-posaurids, despite their clear importance in diagnosing species of Theriosuchus. The more posteriorly placed teeth in Theriosuchus possess false denticles' (Prasad & De Lapparent De Broin, 2002), accompanied by a progressive reduction in alveolus size from the fourth to sixth dentary alveoli. Some of these dentary alveoli form a confluent chain, with the dental arcade occupying an anteroposterior sulcus. Young et al. (2016) also noted additional features that might be characteristic of less-inclusive subgroups within the Theriosuchus species complex, including: (1) a nonspatulate anterior dentary in lateral view (i.e. straight or slightly convex in dorsoven-tral profile); (2) a dual pair of foramina medial to the

dental arcade on the occlusal dental surface, the position of which may vary intraspecifically; (3) vertically festooned external alveolar margins; (4) raised internal alveolar margins; and (5) a symphyseal suture extending to the D5-D7 alveoli (D means den-tary). We discuss the features that unite the clades (T. pusillus + T. guimarotae), and ('T.' ibericus + T.' sympiestodon) below.

Revised diagnosis of Theriosuchus and discussion

(51) Premaxilla-maxilla suture aligned posteromedially in dorsal view (C47.1): The posteromedial alignment of the premaxilla-maxilla suture was originally regarded as an autapomorphy of T. grandinaris (Lauprasert et al., 2011), but it also characterizes T. guimarotae and T. pusillus, and we therefore consider it to be diagnostic for Theriosuchus. This feature cannot be assessed in T.' sympiestodon, but might be present in 'T.' ibericus (PIFUB 102/21.43), although the posterior end of this premaxilla is broken, and we cannot be certain of the nature of the contact with the maxilla.

(52) Absence of a maxillary occlusal pit for reception of an enlarged dentary tooth, anterior to maxillary dental arcade (C54.0): The absence of a maxillary occlusal pit is shared with Pachycheilosuchus (Rogers, 2003). There is an occlusal pit present in both 'T.' sympiestodon, for which a referred specimen (MCDRD 134) has an associated enlarged dentary tooth (Martin et al., 2014a), and 'T.' ibericus, anterior to the hypertrophied fifth maxillary tooth. The goniopholidid Amphicotylus also appears to possess this occlusal pit, visible in lateral view (Mook, 1942).

(53) Lacrimal tapers posteroventrally, not contacting jugal or only forming a point contact (C77.1): The morphology of the lacrimal is not known in T.' ibericus or T.' sympiestodon, but the morphology exhibited by Theriosuchus is distinct from atoposaurids, Wannchampsus (Adams, 2014), Brillanceausuchus, and Koumpiodontosuchus (Sweetman et al., 2015), in which the lacrimal extends posteroventrally and broadly contacts the jugal.

(54) External surface of the dentary (C201.1) and splenial (C202.1) sculpted, including grooved or rugose patterning posteriorly: The morphology of the dentary is poorly known amongst atoposaurids owing to poor preservation (Wellnhofer, 1971; Tennant & Mannion, 2014), but all species herein assigned to Theriosuchus exhibit strong ornamentation on the external surface of the dentaries, and sometimes on the splenial when preserved (Lauprasert et al.,

2011; Young et al., 2016). This sculpting pattern is shared with Shamosuchus (Pol et al., 2009) and Hsisosuchus chowi (Peng & Shu, 2005), but is distinct from Montsecosuchus, Wannchampsus (Adams, 2014),

Pachycheilosuchus (Rogers, 2003), and Brillanceausuchus (Michard et al., 1990), in which only the dentary is sculpted, and the external surface of the splenial is smooth and lacks ornamentation. (S5) Presence of a combination of pseudocaniniform and lanceolate (C253.0), pseudoziphodont maxillary teeth: Theriosuchus pusillus also possesses two additional low-crowned' and labiolingually compressed tooth morphotypes. Labiolingually compressed teeth are absent in T. guimarotae (Schwarz & Salisbury, 2005), and this combination of dental morphologies is unique amongst other heterodont crocodyli-forms (e.g. bernissartiids, notosuchians). Low-crowned teeth are also absent in T. guimarotae and T. grandinaris (Schwarz & Salisbury, 2005; Lauprasert et al., 2011). 'Theriosuchus' ibericus and 'T.' sympiestodon also possess an enlarged fifth maxillary tooth, typically with a corresponding notch on the dentary, whereas this tooth is not present in T. grandinaris, and remains only moderately enlarged in T. guimarotae and T. pusillus.

Unnamed clade: (Theriosuchus pusillus + Theriosuchus guimarotae)

There is strong evidence for a sister-taxon relationship between T. guimarotae and T. pusillus, with this topology recovered in all of our trees, and possessing a Bremer support value of 4 and posterior node probability of 0.99. Synapomorphies uniting these two species include: (1) a posteriorly divided and dorsally facing external naris, similar to ato-posaurids (Wellnhofer, 1971; Tennant & Mannion, 2014); (2) a proportionally small antorbital fenestra, less than half of the size of the orbit, similar to Alligatorellus bavaricus (Tennant & Mannion, 2014), Pachycheilosuchus (Rogers, 2003), and Hoplosuchus (Gilmore, 1926); (3) middle maxillary teeth implanted within single, confluent dental groove, similar to 'T.' ibericus, in which all maxillary teeth occupy a single groove; (4) frontal with bifurcated anterior process, penetrating the posterior border of the nasals; (5) lateral dentary surface with concavity for reception of enlarged maxillary tooth, a feature also present in 'T.' ibericus, 'T.' sympiestodon, and Brillanceausuchus; (6) transitional dentary tooth morphology posteriorly from the fifth alveolus; (7) distinct foramina on the

dentary occlusal surface lingual to the second and third dentary alveoli.

Theriosuchus pusillus Owen, 1878a (type species) Brachydectes minor Owen, 1879 Oweniasuchus minor Woodward, 1885

Type locality and horizon

Beccles' residuary marls (sensu Salisbury, 2002) Lul-worth Beds, Purbeck Group, Berriasian (Early Cretaceous); Durlston Bay, Swanage, Dorset, England.

Lectotype

NHMUK PV OR48216, a near-complete partially articulated skeleton with skull.

Paratype

NHMUK PV OR48330, a near-complete articulated and three-dimensionally preserved skull.

Referred specimens

NHMUK PV OR48328 (holotype of 'Brachydetes minor'), left mandibular ramus; NHMUK PV OR48244, an articulated lower jaw preserved in dorsal aspect and NHMUK PV OR48262, a well-preserved dentary and teeth, all from the same locality as the type series.

Previous diagnoses and comments This species name was originally erected based upon seven paravertebral osteoderms figured by Owen (1878a), which are no longer within the NHMUK collections (Salisbury, 2002). These specimens were figured again in Owen (1879), but this time they were listed as belonging to an incertae sedis crocodyliform. In the same paper, Owen (1879) also described and figured a near-complete skull (NHMUK PV 0R48330) and a near-complete skeleton (NHMUK PV 0R48216) as T. pusillus (Salisbury, 2002). Alongside this, Owen (1879) figured several additional craniomandibular elements as T. pusillus; however, until further analysis of this material, we do not consider it to be referable to T. pusillus. Clark (1986) regarded NHMUK PV 0R48330 to be the holotype specimen of T. pusillus, but Salisbury (2002) designated NHMUK PV 0R48216 and NHMUK PV 0R48330 as the lectotype and paratype, respectively, which we follow here. Until a revision of the type species is conducted, including the referral of other putative specimens, our T. pusillus 0TU is restricted to the lectotype and paratype, as well as NHMUK PV 0R48244 and NHMUK PV 0R48262, following Young et al. (2016). Salisbury (2002) followed Clark (1986) and Brinkmann (1992) in regarding 'Owenia-

suchus ('Brachydectes') minor' (Owen, 1879; Woodward 1885) as a junior synonym of T. pusillus (see also Schwarz & Salisbury, 2005), an interpretation which we follow pending further analysis of the Purbeck crocodyliform material.

Theriosuchus pusillus possesses an unusual combination of derived and plesiomorphic character states, the latter of which might be related to the retention of paedomorphic features associated with its small body size, although both of the specimens belonging to the type series are skeletally mature (Martin et al., 2014a). This heterogeneity is emphasised by the equivocal phylogenetic positions recovered for this taxon (and Atoposauridae), possibly exacerbated by the representation of the Theriosuchus complex' often as a single taxon (i.e. T. pusillus), and the lack of use of appropriately sampled character matrices to resolve its phylogenetic position. Features that might be driving the uncertainty in the phylogenetic position of Theriosuchus include:

1. The presence of a longitudinal ridge on the external surface of the angular, which is shared with the paralligatorids Rugosuchus, Shamo-suchus, and Wannchampsus (Wu et al., 2001a; Pol et al., 2009) and, according to Turner (2015), is a feature uniting (Theriosuchus + Paralliga-toridae). However, we were unable to confirm the presence of this feature on the type specimen of T. pusillus, or any other specimen assigned to Theriosuchus.

2. In T. pusillus, the splenial contributes significantly to the dorsal surface of the mandibular symphysis (Lauprasert et al., 2011; Young et al., 2016), similar to basal crocodylians and a range of basal mesoeucrocodylians (e.g. Buffetaut, 1981; Ortega, Buscalioni & Gasparini, 1996; Pol et al., 2009). However, we have been unable to observe this feature in definitive atoposaurids, because of the manner in which they are preserved.

3. The presence of a raised supraorbital ridge in T. pusillus is similar to a range of neosuchians, including Trematochampsa (Buffetaut, 1976), Bernissartia (Buffetaut, 1975; Buffetaut & Ford, 1979), Hylaeochampsa (Clark & Norell, 1992), Shamosuchus, and several crocodylians (Pol et al., 2009). This feature is absent in definitive atoposaurids, goniopholidids, dyrosaurids, and pholidosaurids, and secondarily lost amongst most crocodylians (Pol et al., 2009).

4. A preorbital lacrimal-prefrontal sutural crest might be present in T. pusillus, a feature that Turner (2015) stated is common for Therio-suchus, and shared with more advanced neo-suchians including Shamosuchus, Rugosuchus, Wannchampsus, and some goniopholidids. There

does appear to be a slight longitudinal crest on the paratype specimen of T. pusillus, although we cannot rule out that this is a result of tapho-nomic distortion, as the skull shows evidence of dorsoventral compression.

5. The morphology of the retroarticular process in T. pusillus, as well as in Alligatorellus and Alli-gatorium, is similar to goniopholidids, Shamo-suchus, and other advanced neosuchians (Pol et al. , 2009), in being reduced and paddle-shaped' (Pol et al. , 2009), and projects posteriorly or posteroventrally. This is distinct from crocody-lians in which the dorsally facing retroarticular process is more anteroposteriorly elongated and subtriangular (Pol et al., 2009).

6. The anterior ends of the palatine bar between the suborbital fenestrae are subparallel in T. pusillus, similar to some members of Eusu-chia, dyrosaurids, and the pholidosaurid Termi-nonaris robusta (Wu, Russell & Cumbaa, 2001b), and not laterally flared as in other advanced neosuchians, such as Shamosuchus (Pol et al., 2009) and Wannchampsus (Adams, 2014). Additionally, the expansion of the posterior ends of the palatines, just anterior to the choana and pterygoid contact, is similar to Shamosuchus (Pol et al. , 2009), Batrachomimus pastosbonen-sis, Rugosuchus, and Paralligator gradilifrons (Turner & Pritchard, 2015), but is distinct from Wannchampsus in which the posterior ends remain narrow and parallel (Adams, 2014).

7. Theriosuchus pusillus shares a single appendicu-lar plesiomorphic feature with Alligatorium mey-eri in that the coracoid is subequal in length to the scapula. This is distinct from paralligatorids and hylaeochampsids, in which the coracoid is proportionally smaller (Turner, 2015; note that the coracoid is about two-thirds the length of the scapula in Pachycheilosuchus, not subequal as stated by Turner, 2015).

8. Pol et al. (2009) stated that goniopholidids and T. pusillus share a well-developed anterolateral articular peg on the dorsal osteoderms (i.e. a peg and socket' articulation), a feature also noted for T. guimarotae (Schwarz & Salisbury, 2005) and Theriosuchus sp. from China (Wu et al. , 1996). However, this comparison was based on a figured osteoderm (now lost) in Owen (1878), and we agree with Joffe (1967) and Salisbury (2002) that this feature is not visible in any of the osteoderms preserved on the paratype specimen of T. pusillus (NHMUK PV OR 48216), or any other specimens definitively attributable to T. pusillus. Rare isolated and disassociated instances of osteoderms attributed to Therio-suchus (Wu et al., 1996; Schwarz & Salisbury,

2005) with this articular morphology are more likely to be referable to a small goniopholidid (Salisbury & Frey, 2001). Goniopholidids are commonly found alongside specimens of Therio-suchus, and therefore the presence of this articular peg should not be used to unite Theriosuchus with goniopholidids until it can be shown that a specimen that definitively belongs to Therio-suchus possesses this morphology.

Revised diagnosis of Theriosuchus pusillus and

discussion

(51) Skull anteroposterior length to orbit length ratio between 3.5 and 4.0 [3.83 (NHMUK PV 0R48330)] (C27.1): This feature illustrates the characteristically large orbits that T. pusillus possesses, and has often been used to support the referral of this taxon to Atoposauridae. This ratio is similar to Wannchampsus (3.72) and Alligatorellus beaumonti (3.86), but is higher than Atoposaurus (2.33-2.86), Alligatorellus bavaricus (3.12), Alligatorium meyeri (3.64), Karatausuchus (3.36) (Storrs & Efimov, 2000), and Hoplosuchus (3.10) (Gilmore, 1926). The relative sizes of the orbit and supratemporal fenestra do not appear to decrease through ontogeny in Theriosuchus, and the retention of this feature is therefore likely to be a paedomorphic state related to the generally small body size of Theriosuchus (Schwarz & Salisbury, 2005).

(52) Abrupt mediolateral expansion of the nasals adjacent to the maxilla anterior to the lacrimals and prefrontals (C70.0): This feature is also present in Koumpiodontosuchus (Sweetman et al., 2015) and Brillancea-usuchus, but is distinct from the condition in T. guimarotae, in which the lateral margins of the nasals are parallel throughout their length (Schwarz & Salisbury, 2005). In atoposaurids, Wannchampsus, Shamosuchus (Pol et al., 2009), T. grandinaris, protosuchians, and goniopholidids, the nasals gradually widen posteriorly.

(53) Posterior tips of nasals perforated by an anterior, sagittal projection of the frontals (C73.1): This contact between the frontals and the nasals is similar to Shamosuchus (Pol et al. , 2009), Brillanceausuchus, and goniopholidids (e.g. Mook, 1942), contrasting with the transverse suture that characterizes Alligatorellus and other neosuchians (e.g. Gilchristosuchus; Wu & Brinkmann, 1993). Notosuchians are similar in possessing a transversely orientated suture, but this is distinct from the simple sutures seen in some

neosuchians because it displays a complex interdigitation between the frontals and nasals (e.g. Ortega et al., 2000; Turner, 2006). Goniopholidids, including Eutretauranosuchus (Smith et al., 2010) and Amphicotylus (Mook, 1942), possess an anterior projection of the frontals, which is sometimes anteriorly bifurcated. More advanced eusuchians, including Acynodon iberoccitanus, have a posteriorly convex frontal-nasal suture, although smaller individuals of this taxon have transversely orientated sutures (Martin, 2007).

(S4) Minimum intertemporal width more than one-third of total width of cranial table: This feature describes a proportionally broad parietal-frontal region between the supratemporal fenestrae on the dorsal skull roof, and is a feature shared by a range of taxa, including atoposaurids (Wellnhofer, 1971; Tennant & Mannion, 2014), Montsecosuchus (Buscalioni & Sanz, 1990a), 'T.' sympiestodon (estimated based on an incomplete skull table; Martin et al., 2010, 2014a), and Brillanceausuchus. We did not create a new character to describe this feature, because of probable non-independence with C19, which describes the relative width of the interorbital and intertemporal regions.

(S5*) Palatines laterally diverge posteriorly, forming palatine bars around choanal groove (C176.1): The palatines of T. pusillus form the anterior and lateral borders of the choana, and laterally diverge posteriorly between the suborbital fenestrae, becoming thickened and rod-like lateral to the choana, where they overlap the anterior portion of the pterygoids that contributes to the lateral margins of the choana. Wannchampsus has a similar morphology, but possesses a deeper choanal groove, which is slightly more posteriorly placed (Adams, 2014). This is distinct from some notosuchians, in which the palatine rods are more laterally directed around the choanal groove (Godoy et al., 2014), and from bernissartiids in which the bars are formed from the anterior extension of the pterygoids.

(S6*) Choana with anterior border mid-way anteroposteriorly between suborbital fenestrae (C181.0), with a V-shaped palatine-pterygoid contact defining anterior edge, and divided anteriorly by a pterygoidean choanal septum (C183.1): The choanal morphology of T. pusillus is distinct in that the anterior edge of the choanal groove is situated relatively anteriorly between the suborbital fenestrae, compared with more advanced neosuchians (e.g. bernissartiids and

paralligatorids). The choana in T. pusillus also receives an anterior and lateral contribution from the palatine, with a V-shaped contact similar to Rugosuchus (Wu et al., 2001a), representing an intermediate morphology between basal neosuchians and eusuchians. The groove is divided by a choanal septum of the pterygoid, similar to T. guimarotae (Schwarz & Salisbury, 2005), Araripesuchus (Ortega et al., 2000; Turner, 2006), Hsisosuchus (Peng & Shu, 2005), and paralligatorids such as Batrachomimus (Montefeltro et al., 2013) and Paralligator (Turner, 2015).

(57) Absence of external mandibular fenestra (C207.0): Theriosuchus pusillus completely lacks an external mandibular fenestra, as is also the case in Shamosuchus (Pol et al., 2009), Wannchampsus (Adams, 2014), Goniopholis (Salisbury et al., 1999), and Bernissartia (Buffetaut & Ford, 1979; Norell & Clark, 1990). This is distinct from most eusuchians and crocodylians, in which the external mandibular fenestra is secondarily well developed (Salisbury et al., 2006), or reduced to a slit-like opening (Brochu, 2004).

(58) Dorsolateral edge of dentary presenting two concave 'waves' (dorsal expansions) (C232.1): The dorsolateral edge of the dentary in T. pusillus is similar to Wannchampsus (Adams, 2014) and Koumpiodontosuchus (Sweetman et al., 2015), as well as 'T.' ibericus and 'T.' sympiestodon. In Alligatorium meyeri and T. guimarotae, the dentary is straight, closer to the condition in Shamosuchus (Pol et al., 2009) and Brillanceausuchus.

(S9*) Splenial dorsally inset into symphysis in ventral view (C234.1): Where the splenial of T. pusillus enters into the symphysis, it is dorsally inset with respect to the ventral surface of the mandible, and slopes posterodorsally. This is a feature otherwise only observed in Atoposaurus oberndorferi and therefore we consider it to be locally autapomorphic. In other taxa in which the ventral surface of the mandible can be observed, the contact between the splenial and dentary portion of the symphysis is ventrally confluent.

(S10*) Heterodont dentition, possessing a combination of anteriorly positioned pseudocaniniform teeth, intermediately positioned labiolingually compressed 'lanceolate' teeth (C253.0), and posteriorly placed 'low-crowned' teeth (C254.0): This unique combination of the three dental

morphotypes in T. pusillus is diagnostic amongst all known species of Theriosuchus and all other known heterodont crocodylo-morphs. The low-crowned morphotype is not known in either T. guimarotae or T. grandinaris (Schwarz & Salisbury, 2005; Lauprasert et al., 2011). (S11) Biconvex first caudal vertebra (C278.1): This is an unusual feature of the axial bracing system in crocodylomorphs (Salisbury & Frey, 2001), and related to the development of procoely throughout the axial series in neosuchians, which has a complicated and unresolved evolutionary history (Salisbury & Frey, 2001). The presence of a biconvex first caudal vertebra is also shared with Pachy-cheilosuchus (Rogers, 2003) and Bernissartia, as well as possibly Brillanceausuchus (Michard et al., 1990), and characterizes all eusuchians (Salisbury et al., 2006; Blanco et al., 2014), including marine forms (Brochu, 2004).

(S12*) Posterior surface of tibial shaft curved, leaving a void between the tibia and fibula (C299.1): The tibia and fibula of T. pusillus are unusual in that they are not confluent, as in other crocodylomorph taxa. Similar to Montsecosuchus (Buscalioni & Sanz, 1990a), the tibial shaft is curved, which leaves a void between the tibia and fibula, the mechanical implications of which are unclear. (S13) Square-shaped dorsal osteoderms (in dorsoven-tral view) (C308.3): The square-shaped osteoderms of T. pusillus form a well-developed biserial shield, and retain a similar outline shape to Pachycheilosuchus (Rogers, 2003) and Alligatorium meyeri (Wellnhofer, 1971). This is distinct from the oval-shaped dorsal osteoderms of Montsecosuchus, and those of Alligatorellus, in which they are subrectangular.

Theriosuchus guimarotae Schwarz &

Salisbury, 2005 Type locality and horizon

Alcobaca Formation (lower Fundschichten' and upper Ruafolge' lignite coal layer), Kimmeridigian, (Late Jurassic); Guimarota Coal Mine, Guimarota, Portugal.

Type specimen

IPFUB Gui Croc 7308, partial skull and mandible, with partial isolated surangular, sacral vertebra II, and two partial osteoderms.

Referred specimens (from Schwarz & Salisbury, 2005)

A range of disarticulated cranial and postcranial material under the accession numbers IPFUB Gui Croc' (see Schwarz & Salisbury, 2005; for a complete list of individual specimens and ID numbers). All referred specimens are from the same locality as the holotype specimen. The majority are disarticulated, and were collected from at least two different horizons - the upper and lower lignite coal layers (Schwarz & Salisbury, 2005).

Previous diagnoses and comments Schwarz & Salisbury (2005) documented a range of diagnostic cranial and axial characteristics for T. guimarotae, and included a detailed comparative discussion of this taxon and other specimens assigned to Theriosuchus. However, a number of these features are more widespread amongst neo-suchians. For example, Schwarz & Salisbury (2005) noted that the squamosal of T. guimarotae is bevelled ventrally, and possesses a notch anteriorly on the lateral surface, both features that these authors regarded as autapomorphic. However, these features are also visible in T. pusillus (NHMUK PV 0R48216) and Brillanceausuchus. Additionally, Schwarz & Salisbury (2005) considered the morphology of the posterolateral corner of the squamosal to be diagnostic, in that it forms a rounded lobe', which projects posteriorly and is similarly sculpted to the rest of the cranial table. Although distinct from T. pusillus and 'T.' ibericus, which both possess an unsculpted lobe, this overall morphology is similar to Rugosuchus (Wu et al., 2001a), Alligatorium meyeri, and Alligatorellus beaumonti. Furthermore, an unsculpted posterolateral lobe is present in Shamo-suchus (Pol et al., 2009; Turner, 2015), Sunosuchus (Wu et al., 1996), and Goniopholis (Ortega et al., 2000), but might be an ontogenetic feature that occurs in younger individuals. Clark (1986) considered this feature to be synapomorphic for Ato-posauridae, but we consider the presence of this lobe to be a synapomorphy that unites Theriosuchus and Paralligatoridae (Turner, 2015), and a feature that was acquired independently in some atoposaurids.

Revised diagnosis and discussion (S1) Subrectangular-shaped external supratemporal fenestra, in dorsal view (C17.0): The shape of the supratemporal fenestra (Schwarz & Salisbury, 2005) is distinct from that of other species of Theriosuchus and advanced neosuchians, which have a circular or subcircular outline. The subrectangular morphology in T. guimarotae is similar to that of pholidosaurids, such as Pholidosaurus

(Salisbury, 2002), Sarcosuchus (Sereno et al., 2001), and Chalawan thailandicus (Martin et al., 2014b) and dyrosaurids (e.g. Jouve, Bouya & Amaghzaz, 2005a; Jouve et al., 2005b), as well as Protosuchus richardsoni (Colbert & Mook, 1951).

(52) Proportionally large lateral temporal fenestra, with an area >50% the area of the orbit (C20.2): The lateral temporal fenestra is relatively larger than that of T. pusillus, Isisfordia (Salisbury et al., 2006), and notosuchians (Buckley et al., 2000; Novas et al., 2009). Goniopholidids, Rugosuchus (Wu et al., 2001a), Wannchampsus (Adams, 2014), and the eusuchians Allodaposuchus precedens (Buscalioni et al., 2001) and Iharkutosuchus makadii (0si, Clark & Weishampel, 2007; 0si, 2008), are similar to T. guimarotae in the large proportional size of the lateral temporal fenestra to the orbit. In dyrosaurids, the fenestra becomes approximately the same size as the supratemporal fenestra, and proportionally larger than the orbit (Jouve et al., 2005a,b).

(53) Notch on the posterolateral surface of the premaxilla within the dorsal margin of the external nares (C39.1): Similar to the goniopholidid Amphicotylus (Mook, 1942), T. guimarotae possesses a shallow fossa, or notch, on the dorsolateral surface of the premaxilla, immediately adjacent to the external nares (Schwarz & Salisbury, 2005), and we consider this feature to be a local autapomorphy for this taxon.

(54) Nasal-lacrimal contact absent on dorsal surface (C71.1): The lacrimal does not contact the nasal (Schwarz & Salisbury, 2005), instead being medially restricted and only contacting the prefrontal. This contact is also absent in Protosuchus richardsoni (Colbert & Mook, 1951), Shamosuchus (Pol et al., 2009), Hylaeochampsa (Clark & Norell, 1992), and Iharkutosuchus (Turner, 2015), but is present in the notosuchian Araripesuchus gomesii (Turner, 2006) and atoposaurids (Wellnhofer, 1971), and represents a series of reconfigurations of the periorbital elements with respect to the remainder of the rostrum in advanced neosuchians.

(55) Jugal with posteriorly directed (C83.0), anteriorly placed (C84.0), and ventromedially displaced (C85.1) postorbital process (C83.0): The jugal postorbital process is slightly anteriorly placed, instead of being medially placed, as in other species of Theriosuchus, which have equally long anterior and posterior

processes of the jugal. This process also has a dorsally directed base similar to Montsecosuchus and Brillanceausuchus, instead of the posterodorsal orientation that characterizes atoposaurids, T. pusillus, and 'T.' ibericus.

(S6) Anterior process of frontal constricted between the prefrontals (C109.0): This feature excludes the sagittal projection of the frontals into the nasals anterior to the orbits, which some crocodyliforms possess, and refers to the convergence between the lateral margins of the anterior portion of the frontals. This feature is shared with Alligatorium meyeri, 'Alligatorium' franconicum, and Amphicotylus (Mook, 1942). In other crocodyliforms, including Alligatorellus and other species of Theriosuchus, the lateral edges of the nasal are not mediolaterally constricted and remain subparallel.

(S7*) Ectopterygoid with well-developed anterior process, reaching the posterior-most two maxillary teeth (C170.0): The anterior process of the ectopterygoid is extremely well developed, reaching a point level with the anterior margin of the suborbital fenestra and the posterior-most maxillary teeth. In T. pusillus, goniopholidids, and paralli-gatorids, this process is short and poorly developed in ventral aspect.

(58) Anterior margin of palatines anteriorly pointed (C173.1): The maxilla-palatine suture is posteroventrally directed towards the anterior margin of the suborbital fenestra along the midline, level with the sixth maxillary tooth (Schwarz & Salisbury, 2005). In T. pusillus, this contact is gently rounded anteriorly, similar to Koumpiodontosuchus (Sweetman et al., 2015). However, this region of the palate is not preserved in many of the specimens included in our analysis - especially atoposaurids - and therefore we consider this to be a tentative autapomorphy at present.

(59) Parallel posterolateral margins of inter-fenestral bar between suborbital fenestrae (C175.1): The interfenestral bar of T. guimarotae is formed entirely from the paired and fused palatines, and the lateral margins of the posterior portion nearing the paired choanae run parallel to one another. This is distinct from Brillanceausuchus, in which the lateral margins converge posteriorly, and T. pusillus, Wannchampsus, and Shamosuchus, in which the lateral margins flare posterolaterally.

(S10*) Pterygoids excluded from the posterior margin of suborbital fenestra by ectopterygoid-

palatine contact (C180.1): The exclusion of the pterygoid ventral lamina from the posterior margin of the suborbital fenestra is unique in T. guimarotae. In all other specimens analysed, for which the presence or absence of this feature can be assessed (including T. pusillus, 'T.' sympiesto-don, protosuchians, goniopholidids, and paralligatorids), the pterygoids contribute to the posterior margin of the suborbital fenestra.

(S11*) Completely septated choanal groove (C183.2): The choanae of T. guimarotae are completely septated, formed anteriorly by the palatines and posteriorly by the pterygoids, similar to the goniopholidids Amphicotylus (Mook, 1942) and Eutretauranosuchus (Smith et al., 2010). This morphology is distinct from in T. pusillus, in which the choanal groove is partially septated, and 'T.' sympiestodon (Martin et al., 2010, 2014a,b) and other paralligatorids, in which the groove is open and undivided.

(512) Basisphenoid ventrally exposed anteriorly to the basioccipital (C188.0), and ventral surface continuous with surrounding cranial elements (C190.0): The conformation of the basisphenoid to the remainder of the occipital plane is distinct from that in T. pusillus and other advanced neosuchians in which the main body of the basisphenoid is separated by a sulcus and posteroventral step.

(513) Mandibular symphysis of moderate posterior length, posteriorly reaching the fifth-sixth dentary tooth position (C204.1): The relative length of the symphysis to the dental arcade is highly variable within species currently and previously assigned to Theriosuchus, being short up to the fifth alveolus in 'T.' ibericus (also shared with Brillan-ceausuchus), terminating medial to the fifth and sixth alveolus in T. guimarotae, medial to the sixth in T.' sympiestodon, and medial to the seventh alveolus in T. pusillus, T. grandinaris, and Theriosuchus sp. (NMS G. 2014.52.1; Young et al., 2016).

(S14*) External mandibular fenestra present (C207.1): Theriosuchus guimarotae possesses the plesiomorphic condition in the retention of a triangular-shaped external mandibular fenestra, as also occurs in Alligatorium meyeri, Protosuchus richardsoni (Colbert & Mook, 1951), and Eutretauranosuchus (Smith et al., 2010). This morphology is not known in any atoposaurid specimen, although the posterior portions of the dentaries of

Alligatorellus, Atoposaurus, Montsecosuchus, and T. grandinaris are partially obscured or unknown.

Theriosuchus grandinaris Lauprasert et al., 2011

Type locality and horizon

Sao Khua Formation, Khorat Group, early Aptian (Early Cretaceous); Phu Phok, Kok Prasil Subdistrict, Phu Phan District, Sakon Nakhon Province, north-western Thailand.

Type specimen

PRC-2, fused anterior rostrum and mandible (note that the specimen ID is stated as PPC' in Laupra-sert et al., 2011, but the institutional abbreviation is given as PRC).

Previous diagnoses and comments Lauprasert et al. (2011) assigned T. grandinaris to Theriosuchus (and therefore to Atoposauridae) based on the possession of a brevirostrine skull with the maxilla transversely flattened, symmetrical, and pointed at the apex, which is a feature common in numerous small neosuchian crocodyliforms. The pre-maxilla-maxilla suture of T. grandinaris is aligned posteromedially in dorsal view, deemed to be diagnostic by Lauprasert et al. (2011), but this feature is also present in T. guimarotae (Schwarz & Salisbury, 2005) and T. pusillus, and is more likely to characterize the genus Theriosuchus (see above). It is not entirely clear what is meant by the weak notch' present at the premaxilla-maxilla suture (Lauprasert et al., 2011), although a notch is present ventrally in this suture in T. pusillus and possibly T. guimarotae (Schwarz & Salisbury, 2005), which is for the enlarged dentary tooth. This feature is distinct from 'T.' ibericus and 'T.' sympiestodon, in which there is a distinct occlusal pit within the dental arcade to accommodate the enlarged dental tooth. The relatively long mandibular symphysis, terminating posteriorly medial to the D7 alveolus, is a feature shared with T. pusillus, but not unique within Theriosuchus (contra Lauprasert et al., 2010; see also Young et al., 2016). The presence of an anteriorly tapering and slender prefrontal was also stated as diagnostic for T. grandinaris by Lauprasert et al. (2011), but this morphology is present in both T. pusillus and T. guimarotae, as well as Alligatorel-lus and paralligatorids.

Revised diagnosis and discussion (S1) Fully divided external nares (C10.0): The presence of paired external nares (i.e. divided

by either the premaxilla or an anterior extension of the nasals) might be synapomorphic for Atoposauridae (see above), but is clearly present in T. grandinaris too.

(52) Premaxilla-maxilla suture ventrally confluent (C42.0), with ventral diastema at the contact suture: Lauprasert et al. (2011) stated that there is a diastema at the premaxilla-maxilla suture, not a lateral concavity as in T. pusillus and T. guimarotae, which might be diagnostic for this species.

(53) Nasals gradually widen adjacent to the maxilla (C70.1): The gradual widening of the nasal bones posteriorly is distinct from the condition in T. pusillus, in which the lateral margins expand rapidly adjacent to the maxilla and anterior to the periorbital elements, and T. guimarotae in which the nasals are consistently narrow (Schwarz & Salisbury, 2005). This gradual widening of the nasals is reminiscent of longirostrine neosuchians, such as Rugosuchus (Wu et al., 2001a).

(S4*) Unique combination of pseudocaniniform, lanceolate-shaped (C253.0), and labiolingually flattened (C354.0) teeth with faintly crenulated mesial and distal carinae (C245.0): The heterodont dentition of species attributed to Theriosuchus has long been recognized as diagnostic. Heterodonty is not exclusive to Theriosuchus, also being known in bernissartiids, notosuchians, and a range of other neosuchian taxa (Оsi, 2014). However, variation in dentition differentiates the species of Theriosuchus from one another (see above).

Additional comment

Theriosuchus grandinaris shares many similarities with a fragmentary specimen described as Therio-suchus sp. (NMS G. 2014.52.1) from the Isle of Skye, known solely from the anterior portion of a right den-tary (Young et al., 2016). The two are recovered as sister taxa in our analyses, and shared features include: (1) a straight lateral margin of the dentary in dorsolateral view; and (2) a parallel dentary sym-physis to the dental arcade. Although this might indicate that Theriosuchus sp. is referable to T. grandinaris, they are widely separated from one another spatiotemporally, and other features allow us to distinguish the two taxa (Young et al., 2016).

Theriosuchus sp. European occurrences

Additional material referred to Theriosuchus has been described from a host of other continental Euro-

pean localities, but is typically fragmentary or isolated in nature, and therefore difficult to assign to a particular species. The majority of these have been assigned to Theriosuchus based on its highly diagnostic tooth morphotypes, as discussed above. These teeth are usually small, no more than 2-3 mm in either dimension. The bases of the crowns are always mesiodistally constricted, and there are varying degrees of labiolingual compression, which can be asymmetrical, leading to a convex labial face and a flat lingual face. A lanceolate tooth morphology is also diagnostic for Theriosuchus (Schwarz & Salisbury, 2005; see above). These occurrences include, in stratigraphical order from oldest to youngest:

1. Theriosuchus sp. is known from a dentary from the late Bajocian-Bathonian of the Isle of Skye, UK (Young et al., 2016). This specimen (NMS G. 2014.52.1), although based on highly fragmentary material, appears to have several unique dental characteristics that distinguish it from T. grandinaris and other species of The-riosuchus, including: (1) posterolaterally orientated crenulations on the posterior end of the dental arcade; (2) a longitudinally crenulated occlusal surface; and (3) the symphysis not contributing to the splenial (Young et al., 2016). Although additional crocodyliform material is in preparation from the Isle of Skye that might be referable to this taxon (S. Brusatte, pers. comm., 2015), the currently available material appears to possess a unique combination of character states amongst Theriosuchus species. Despite this, Young et al. (2016) did not refer it to a new species because of the fragmentary nature of the remains, a taxonomic decision that we follow here.

2. Isolated tooth crowns from the late Bathonian Forest Marble microvertebrate horizon of the UK were referred to Atoposauridae, but described as 'Theriosuchus'-like (Evans & Mil-ner, 1994). Based on the information provided in Evans & Milner (1994: fig. 18.6e, p. 315), at least one of these teeth possesses a pseu-doziphodont morphology, and characteristic labiolingually compressed and lanceolate morphology. We therefore tentatively regard these specimens as cf. Theriosuchus sp.

3. Fifty-nine variably worn or abraded teeth, from the Oker and Uppen sections of the Kimmerid-gian of north-west Germany, assigned to cf. Theriosuchus sp. by Thies et al. (1997). These teeth possess the characteristic lanceolate morphology of T. pusillus and T. guimarotae, and we therefore consider them to represent Therio-suchus sp.

4. Another specimen (DFMMh 200, the anterior part of a crushed skeleton) from the Kimmerid-gian of northern Germany was tentatively referred to T. pusillus by Karl et al. (2006), although it is probably of a different ontoge-netic age to the type material. This specimen has a posterior maxillary dental arcade situated within a confluent dental groove, similar to the feature that we identify as synapomorphic for (T. pusillus + T. guimarotae) (see above). The teeth of DFMMh 200 are morphologically similar to T. pusillus (Karl et al., 2006), owing to the presence of faint carinae, a slightly labiolin-gually compressed and lanceolate morphology, an enlarged fourth maxillary tooth, and canini-form third and fourth dentary teeth. However, the external nares are almost completely divided by an anterior projection of the nasals, a feature that we consider to be diagnostic of Atoposauridae, to the exclusion of Therio-suchus. Despite this latter feature, we tentatively refer DFMMh 200 to Theriosuchus cf. pusillus, pending a more detailed description of this potentially important specimen. Additional material referable to Theriosuchus sp. comes from the same region as DFMMh 200, including DFMMh 605 (a partial and damaged skull, probably of a hatchling); DFMMh 325 (four ventral osteoderms, two ribs, and a fragment of a dorsal vertebra; DFMMh 236 (numerous dorsal osteoderms); DFMMh 279 (single femur); and DFMMh 507 (a solitary tooth) (Karl et al. , 2006). However, the dorsal osteoderms possess an anterior process, a feature that we do not consider to be present in Theriosuchus, and therefore at least some of the osteoderms comprising DFMMH 236 are more likely to belong to a goniopholidid. The femur and axial material cannot be definitively attributed to Theriosuchus based on our revised understanding of this genus, and we consider it to belong to an indeterminate mesoeucrocodylian. We tentatively consider the partial skull and the single tooth to be referable to cf. Theriosuchus sp., owing to the dental similarities they possess.

5. Isolated teeth from two localities in the Titho-nian of north-eastern France were referred to cf. Theriosuchus sp. (Cuny et al., 1991). One of the figured teeth (PMC MO2.15) appears to possess a lanceolate and pseudoziphodont morphology. We therefore designate the specimen as Theriosuchus sp.

6. Isolated teeth (MO-CHA-30, 31, 32) from the Tithonian of western France were referred to Theriosuchus cf. pusillus (Vullo et al., 2014). These teeth possess a lanceolate morphology, as

well as carinae that are festooned' on the apical margins of each tooth, giving a pseudoziphodont appearance. This morphology is characteristic of T. pusillus, and therefore we agree with Vullo et al. (2014) in their reference to Therio-suchus cf. pusillus, pending the discovery of more complete material.

7. Theriosuchus sp. was described based on teeth from the Berriasian of south-west France (Pouech, Mazin & Billon-Bruyat, 2006; Pouech et al., 2014). A single anterior tooth is figured in lingual view by Pouech et al. (2006) (CHEm03.506), and possesses faint apicobasal striations, a mesiodistally compressed crown base, and a pointed apex, giving it a lanceolate morphology. This is characteristic of Therio-suchus, and therefore we retain the status of these specimens as Theriosuchus sp., noting that these are of the same age as T. pusillus from the UK (Owen, 1878a, 1879; Salisbury, 2002).

8. Schwarz-Wings et al. (2009b) referred 284 teeth to Theriosuchus sp. from the Berriasian of southern Scandinavia (Skyttegard Member, Rabekke Formation of Bornholm, Denmark, and Annero Formation, Vitaback Clays, of Skane, Sweden). These teeth possess the labiolingually compressed and lanceolate morphologies characteristic of T. pusillus and T. guimarotae (Schwarz & Salisbury, 2005), and also the broader low-crowned' morphotype. This third morphotype is also known in 'T.' ibericus, and therefore we consider these teeth as a whole to represent cf. Theriosuchus sp., but recognize that more than one heterodont species, including one closely related to T. pusillus, might be present.

9. Hornung (2013) described, but did not figure, a partial left mandible (GZG.BA.0139) from the Buckeberg Formation (Berriasian-Valanginian) of northern Germany (Old comital quarry, Harrl Hill, approximately 1.7 km south-east of Buckeburg). It was assigned to Theriosuchus sp. based on the presence of a ridge on the ven-trolateral surface of the angular, a posteriorly directed retroarticular process, and the absence of the external mandibular fenestra. The absence of the latter feature suggests that it is not referable to T. guimarotae (see above). This specimen is about three times the size of the holotype of T. pusillus, and might therefore represent one of the largest known individuals of Theriosuchus. Hornung (2013) also noted a distinct knob-like lateral swelling on the anterior end of the lateral shelf of the angular. Combined with its relatively large size, GZG.BA.0139 might therefore represent a novel species of Theriosuchus, but we refer it to cf. Theriosuchus sp.,

pending examination of this material. Several osteoderms were also briefly mentioned by Hornung (2013) from an unspecified locality near Sehnde (Lower Saxony) as resembling Theriosuchus, including one (GZG.STR.50293) that was identified as closely reminiscent of T. pusil-lus. Hornung (2013) also noted that complete atoposaurid' skeletons are preserved in the Cherves-de-Cognac region in south-western France (Berriasian), and remain undescribed.

10. Theriosuchus sp. has been identified from multiple localities in the Early Cretaceous (early Ber-riasian to early Barremian) Teruel region of eastern Spain (Ruiz-Omenaca et al., 2004). However, these occurrences were only documented as part of faunal lists, and no further details are given, although a brief description without figures is provided for this material by Canudo et al. (2010). Therefore, we consider these occurrences to represent aff. Theriosuchus sp., pending examination of the identified material.

11. Theriosuchus-like teeth have been described from the Cenomanian of south-western France, and assigned to Atoposauridae (Vullo & Neraudeau, 2008). Based on the figured specimen, these teeth possess a pseudoziphodont morphology, formed by the apical extension of the lingual carinae. Although the only tooth figure has a worn apex, it is clear that these teeth possess a lanceolate morphology, with evidence of labiolingual compression. We therefore assign these specimens to Theriosuchus sp., pending further examination.

12. Two teeth from the Maastrichtian of north-eastern Iberia have been referred to an indeterminate ato-posaurid (Marmi et al., 2016). These teeth have a lanceolate morphology, are labiolingually compressed, and apically blunt. Faint apicobasally orientated ridges are present and more prominent on the lingual surface, where they develop into pseu-doziphodont crenulations. Based on this combination of characteristics, we tentatively regard these teeth as representing cf. Theriosuchus sp., but based on their age it is more likely that they represent a crocodyliform more closely related to 'T.' sympiestodon (Martin et al., 2014a,b).

Non-European occurrences

Outside of Europe, reports of Theriosuchus are less frequent, but provide further evidence that this genus was a common component of Cretaceous Asian semi-aquatic ecosystems. These occurrences comprise:

1. Haddoumi et al. (2016) documented numerous small teeth from the Bathonian of eastern Morocco, and ascribed them to cf. Theriosuchus sp. based on their lanceolate crown morphology and

pseudoziphodont carinae. As a consequence of our revision of Atoposauridae and Theriosuchus, we refer these teeth to Theriosuchus sp. These specimens currently represent the only confirmed occurrences of Theriosuchus from Gondwana.

2. Lauprasert et al. (2011) assigned a partial left dentary (PRCMR CCC-1) and a single tooth (PRCMR 283) to cf. Theriosuchus sp. from the Phu Kradung Formation (latest Jurassic to Early Cretaceous) of the Nong Bua Lum Phu Province in Thailand. This belongs to a hetero-dont crocodyliform with a combination of pseudo-caniniform and lanceolate teeth that display the presence of festooned crenulations, formed by anastomosing and irregular ridges on the crown (the characteristic pseudoziphodont apical morphology). Furthermore, the presence of interalveolar septae and dental teeth occupying a single groove (Lauprasert et al., 2011) can be used to assign this specimen to Theriosuchus (Young et al., 2016).

3. A single tooth was ascribed (PRCMR 218) to cf. Theriosuchus sp. from the Early Cretaceous of Thailand (Cuny et al., 2010). This tooth has a similar morphology to other teeth assigned to Theriosuchus from Thailand (see above), and therefore the assignment to Theriosuchus sp. is supported. It might be that these teeth represent isolated occurrences of T. grandinaris based on their near-identical morphologies.

4. Mo et al. (2016) described a tooth as cf. Therio-suchus from the Aptian of southern China. This tooth is similar to the Asian occurrences of The-riosuchus, and therefore we agree with Mo et al. (2016), retaining its status as cf. Theriosuchus sp.

Neosuchia Benton & Clark, 1988 'Alligatorium' franconicum Ammon, 1906 Alligatorium paintenense Kuhn, 1961

Type locality and horizon

Unknown bed, late Kimmeridgian (Late Jurassic); Painten, 10 km north of Kelheim, south-east Germany.

Type specimen

BSPG specimen (number unknown): destroyed or lost; articulated hindlimb and pelvic girdle.

Referred specimen

BSPG specimen (number unknown): destroyed or lost; type of Alligatorium paintenense, a skull and near-complete skeleton missing most of the tail and the right hindlimb.

Previous diagnoses and comments Ammon (1906) originally named Alligatorium fran-conicum based on an articulated right hindlimb from the Late Jurassic of Painten, Bavaria. Subsequently, Broili (1931) tentatively referred a partial skeleton from the same locality to this species. Kuhn (1961) referred this latter skeleton to a new species, Alligatorium paintenense, without detailed discussion, an interpretation followed by Kuhn (1966). Wellnhofer (1971) provided emended diagnoses for both species and considered them to be valid, along with Montsec-osuchus ('Alligatorium') depereti (Vidal, 1915) and Alligatorium meyeri (Gervais, 1871). Most recently, Tennant & Mannion (2014) concluded that the differences in limb proportions were not enough to distinguish 'Alligatorium paintenense' from 'Alligatorium' franconicum, and synonymized the former with the latter, an interpretation that we follow here.

Buffetaut (1981) described a small brevirostrine skull from the Late Jurassic of Brauvilliers, Meuse, France, and assigned it to Alligatorium cf. painte-nense. This assignment was based on the significantly longer and pointed rostrum compared with other atoposaurids, the posteriorly placed orbits, a mediolaterally narrow interorbital region, and the dense external surface sculpting. Buffetaut (1981) also noted some differences, including the overall larger size and moderate development of the posterolat-eral squamosal 'lobe', as noted for Alligatorium meyeri, Alligatorellus beaumonti, and T. pusillus. Unfortunately, we have been unable to examine this specimen first-hand and, based on the figures in Buf-fetaut (1981), we cannot determine its relationship to the now lost German specimens of 'Alligatorium' franconicum (Wellnhofer, 1971).

Wellnhofer (1971) noted that the skull of 'Alligato-rium' franconicum was strongly sculpted and more longirostrine than the typically brevirostrine ato-posaurids, and that the external nares were fully open and not divided by a septum, the presence of which is a feature that we consider to be diagnostic for ato-posaurids, but also possessed by T. pusillus (Owen, 1879). Additionally, 'Alligatorium' franconicum possesses biserial paravertebral rows of sculpted dorsal osteoderms, similar to atoposaurids and T. pusillus, but with a central keel that diminishes anteriorly.

Most of our analyses found 'Alligatorium' franconicum to be united within a clade comprising Pholi-dosaurus (Salisbury, 2002), Eutretauranosuchus (Smith et al., 2010), Amphicotylus (Mook, 1942), and Koumpiodontosuchus (Sweetman et al., 2015), representing Bernissartiidae, Goniopholididae, and Pholi-dosauridae. However, resolving the position of these groups within Neosuchia is not the focus of our study, although it is noteworthy that we found them to be more closely related to paralligatorids than ato-

posaurids. The characters uniting ''Alligatorium' franconicum with these non-atoposaurid taxa include: (1) an intermandibular angle of <40° (36°); (2) a total anteroposterior skull length to snout length (measured from the anterior margin of the orbit to the anterior edge of the premaxilla in dorsal view) ratio of <2.0 (1.81); (3) a skull anteroposterior length to orbit length ratio of >5.0 (5.27), a feature also shared with Brillanceausuchus and Shamosuchus (Pol et al., 2009); (4) a nasal that only participates posteriorly in the margin of the external nares, a feature that appears to only be shared with Amphicotylus (Mook, 1942), and distinct from atoposaurids and Therio-suchus in which the nasal participates posteriorly and medially, often projecting anteriorly into and dividing the external nares; and (5) asymmetrical dorsal osteoderms in dorsal aspect (excluding any anterolateral peg articulation), a feature more widespread across Neosuchia, including Alligatorellus (Tennant & Mannion, 2014), Pachycheilosuchus (Rogers, 2003), and Wannchampsus (Adams, 2014).

Buscalioni (1986) also found Alligatorium' fran-conicum to be closely related to goniopholidids, bernissartiids, and paralligatorids, based on a combination of factors including longer rostral length, undivided external nares, proportionally broad supratemporal fenestra, and transversely broad osteoderms. However, Buscalioni & Sanz (1988) considered it to be a nomen dubium, owing to the fact that the specimen is lost, and because the single autapomorphy proposed by Wellnhofer (1971; a reduced anterior process of the ilium) is present in Montsecosuchus and atoposaurids, and more broadly within Neosuchia. We regard Alligatorium' francon-icum as a potentially diagnostic non-atoposaurid taxon, but refrain from erecting a new genus name because all specimens are lost/destroyed and our proposed autapomorphies are tentative pending the discovery of additional material. Future discoveries referable to Alligatorium' franconicum should shed light on the taxonomy of specimens attributed to Alligatorium and the evolution of the biserial osteo-derm shield in Neosuchia, as well as the early development of a longirostrine cranial morphology in non-thalattosuchian Jurassic taxa.

Revised diagnosis and discussion (S1) Minimum mediolateral width between supratemporal fenestrae broader than minimum mediolateral width between orbits (C19.0): This feature describes the relatively high intertemporal mediolateral width compared with the interorbital region, a feature that Alligatorium' franconicum shares not only with some atopo-saurids, but also Montsecosuchus (Buscalioni & Sanz, 1990a) and Pachycheilosuchus (Rogers, 2003).

(52) Skull length to width ratio >2.5 (2.77) (C25.2): The skull length to width ratio reported here is diagnostic only in the sense that we have not sampled many fully longirostrine taxa, with the exception of Pholidosaurus (Salisbury, 2002), with which it also shares this character state. Koumpiodontosuchus (Sweetman et al., 2015) and the two goniopholidid taxa are semi-longirostrine, but not to the extent of pholidosaurids and fully marine crocodyliforms. As such, Alligatorium' franconicum might represent a transitional form towards a more aquatic lifestyle, but this requires much more detailed investigation pending the discovery of new remains.

(53) Presence of a lateral keel on posterior part of dorsal osteoderms (C312.1): This character state is shared with Alligatorellus, but the morphology of the osteoderms of Alligatorium' franconicum appears to be distinct, with a central longitudinal keel, and a lateral sulcus with an anterolateral projection, similar to the peg and socket' morphology described for T. guimarotae (Schwarz & Salisbury, 2005) and goniopholidids (Salisbury & Frey, 2001). This overall geometry appears to be different from the dorsal osteoderm shield for any known atoposaurid (Tennant & Mannion, 2014), and from any goniopholidid, and identifies Alliga-torium' franconicum as a distinct neosuchian taxon.

Neosuchia Benton & Clark, 1988 Paralligatoridae Konzhukova, 1954 (sensu Turner, 2015) Sabresuchus gen. nov. (http ://zoobank. org/NomenclaturalActs/ 35B479C6-7620-428A-92A7-F613B43FEC24)

Etymology

Sabre' in reference to the enlarged and curved fifth maxillary tooth, and suchus' from the Ancient Greek, soukhos, for crocodile.

Included species

Sabresuchus ibericus (type species) and Sabresuchus sympiestodon.

Distribution Cretaceous of Europe.

Comments

'Theriosuchus' sympiestodon shares a number of features with 'T.' ibericus, and these two taxa form a sister-taxon relationship in all of our analyses, distantly related to other species definitively ascribed to Theriosuchus. These taxa have never been directly

compared or included together in a phylogenetic analysis, and based on this novel finding here we erect the name Sabresuchus gen. nov. to include them both. The name refers to the hypertrophied fifth maxillary tooth, which is at least four times the size of adjacent teeth in both species, and is the most striking feature of this genus.

Our phylogenetic analyses place Sa. sympiestodon and Sa. ibericus within Paralligatoridae. The Bremer support uniting these two species is 3-4 (Fig. 5), with a posterior node probability of nearly 1 (Fig. 7), providing strong support for their sister-taxon relationship. Sabresuchus is the sister taxon to Wannchampsus (Adams, 2014), often cited alongside T. pusillus as an important transitional form on the lineage leading to Eusuchia (Langston, 1974). A number of dental features characterize this relationship with Wannchampsus, including the maxillary teeth with denticulate carinae on the mesiodistal margins, and ridged ornamentation on the enamel surface in middle to posterior teeth. However, the instability of this relationship with Wannchampsus is highlighted by the fact that if Pachycheilosuchus (Rogers, 2003) is retained in phylogenetic analyses a priori, then Bril-lanceausuchus and Shamosuchus instead form a poly-tomy with Wannchampsus, with Sabresuchus as the sister taxon to that clade (Fig. 5A) However, our analyses were not designed to resolve paralligatorid relationships and, irrespective of this lack of consensus, we still find a sister-taxon relationship between Sa. ibericus and Sa. sympiestodon that is nestled within this paralligatorid assemblage, in a more crownward position than Theriosuchus. This relationship between Sabresuchus and other paralligatorids is further supported by a number of unequivocal synapo-morphies, including: (1) a striated external surface of the posterior portion of the maxilla (C51.0); (2) an anteroposterior ridge occupying the entire length of the frontal dorsal surface (restricted to the median portion of this surface in Sa. sympiestodon) (C101.3); and (3) the presence of an obliquely orientated ridge on the dorsal surface of the squamosal (C145.0).

Diagnosis and discussion

(SI) Ventral edge of the groove for the upper ear lid positioned directly ventral to dorsal edge (C136.1): This morphology is distinct from T. guimarotae (Schwarz & Salisbury, 2005) and T. pusillus, in which the ventral edge of the ear lid is laterally displaced with respect to the dorsal edge. However, it is similar to a range of neosuchians, including Brillanceausuchus, Wannchampsus, and Koumpiodontosuchus (Sweetman et al., 2015), Alligatorellus, Pholi-dosaurus (Salisbury, 2002), Eutretauranosuchus (Smith et al., 2010), Mahajangasuchus (Turner

& Buckley, 2008), and Shamosuchus (Pol et al., 2009; Turner, 2015), as well as Protosuchus richardsoni (Colbert & Mook, 1951).

(52) Squamosal posterolateral process ('lobe') offset from the dorsally flat skull table (C140.1): The depression of the squamosal posterolateral lobe and its confluence with the paroccipital process is similar to the condition seen in the paralligatorids Brillanceausuchus and Shamosuchus (Turner, 2015) and the eusuchian Allodaposuchus precedens (Buscalioni et al., 2001), as well as the crocodyliform Zosuchus (Pol & Norell, 2004a) (see Discussion).

(53) Squamosal with an oblique ridge on the dorsal surface (C145.0), posterior to the posterior margin of the supratemporal fenestra (C146.0): This morphology of the dorsal surface of the squamosal is distinct from Brillanceausuchus, Shamosuchus, and Wannchampsus, in which the ridge is positioned laterally to the external supratemporal fenestra. In protosuchians, coelo-gnathosuchians, atoposaurids, bernissartiids, Montsecosuchus, 'Alligatorium' franconicum, and Theriosuchus pusillus, there does not appear to be any ridge. Theriosuchus guimarotae is distinct from Sabresuchus in possessing a rounded and longitudinally orientated crest that occupies the whole of the anteroposterior length of the flat dorsal surface, separating it from the bevelled lateral portion (Schwarz & Salisbury, 2005).

(54) Lateral surface of dentary with concavity for reception of enlarged maxillary tooth (C212.1): The presence of this lateral concavity is similar to Brillanceausuchus, T. pusillus, and T. guimarotae, which all possess either an individual enlarged maxillary tooth, or a wave of enlarged teeth. To accommodate this, the lateral surface of the dentary becomes invaginated, with the maxillary teeth occluding laterally to this surface. In protosuchians, and all other neosuchians that we observed, the lateral surface of the dentary is smooth and confluent with the rest of the external surface of the dentary at the position where the anterior maxillary teeth occlude.

(55) At least some medially positioned confluent maxillary teeth, implanted in a dental groove (C217.1 and C217.2): In both species of Sabresuchus, at least some of the maxillary teeth are implanted in a confluent dental groove, instead of individual alveoli. This is shared by T. guimarotae (Schwarz & Salisbury, 2005), Eutretauranosuchus (Smith et al., 2010), and Koumpiodontosuchus (Sweetman et al., 2015), as well as by some notosuchians, in which an 'alveolar trough' develops (Gomani, 1997;

Buckley et al., 2000; O'Connor et al., 2010). This feature is distinct from the condition in thalattosuchians (Gasparini et al. , 2006), gonio-pholidids (Tykoski et al., 2002), and eusuchians (Buscalioni et al., 2001; Salisbury et al., 2006; O si et al., 2007), in which individual teeth are differentially spaced, and occupy isolated alveoli, separated by interalveolar septae.

(56) Maxillary teeth with low-crowned and strongly labiolingually compressed morphotype (С2Б4.0): The presence of a low-crowned' dental morphology has been observed and used as a defining feature for some Theriosuchus species, without ever giving a precise definition of what this characterizes. We consider this morphology to describe a dental crown that is mesiodistally broader than it is apicobasally tall, and with apical margins orientated at <450 from the horizontal. This dental morphotype is diagnostic for Sabresuchus, but also shared with T. pusillus.

(57) Maxillary tooth Б hypertrophied, at least 4.0 times the size of adjacent maxillary teeth (C2Б8.1): In all other specimens we observed (including T. pusillus and atoposaurids), the maxillary teeth were subequal in size or developed one or two enlarged waves' (e.g. goniopholidids). In protosuchians, maxillary teeth 1-3 are proportionally and variably the largest (Wu et al., 1997; Gow, 2000; Pol et al., 2004), and in notosuchians there is a progressive reduction in size posteriorly from the enlarged second maxillary tooth (Ortega et al., 2000; Pol & Apesteguia, 2005; Zaher et al., 2006; Campos et al., 2011). In Sabresuchus, the fifth maxillary tooth is larger than the fourth, which is in turn larger than the third tooth.

Sabresuchus ibericus comb. nov. Theriosuchus ibericus Brinkmann, 1989

Type locality and horizon

Lowest horizon of an abandoned lignite mine, La Huerguina Formation (Una), Upper Barremian (Early Cretaceous); south-west of the eastern edge of Un a, Serranía de Cuenca, eastern Spain.

Type specimen

MfN MB.R IPFUB 102/21.2, partial right maxilla with in situ teeth.

Referred material (specimens housed at the MfN MB.R, unless stated)

IPFUB 102/11.2 (incomplete left maxilla); IPFUB 102/11.3 (incomplete right maxilla); IPFUB 102/11.5 (frontal); IPFUB 102/11.6 (incomplete mandible);

IPFUB 102/11.7 (cranial fragments); IPFUB 102/12.1 (incomplete left dentary); IPFUB 102/12.2 (incomplete right dentary); IPFUB 102/21.1 (left maxilla fragment with a tooth); IPFUB 102/21.3 (incomplete frontals); IPFUB 102.21.4 (parietal); IPFUB 102/21.5 (incomplete left jugal); IPFUB 102/21.6 (incomplete right jugal); IPFUB 102/21.7 (incomplete right squamosal); IPFUB 102/21.8 (left quadrate fragment with incomplete left squamosal and left quadratojugal); IPFUB 102/21.9 (right quadrate fragment); IPFUB 102/21.10 (near-incomplete pterygoid and basisphe-noid); IPFUB 102/21.18 (basioccipital fragment); IPFUB 102/21.11 (incomplete left dentary); IPFUB 102/21.12 (left splenial fragment); IPFUB 102/21.13 (left angular fragment); IPFUB 102/21.14 (left fractured surangular); IPFUB 102/21.15 (incomplete right angular with right surangular fragment and right coronoid); IPFUB 102/21.16 (incomplete left articular); IPFUB 102/21.17 (right articular); IPFUB 102/21.19 (24 teeth, representing three different morphotypes); IPFUB 102/21.43 (right premaxilla); IPFUB 102/21.44 (caudal vertebra); IPFUB 102/21.45 (anterior caudal vertebrae and various other cranial bones); IPFUB 102/22.1 (two teeth and roots, one incomplete tooth crown); IPFUB 102/22.2 (two teeth and roots, one fragmentary tooth crown); IPFUB 102/ 22.3 (angular fragment); IPFUB 102/22.4 (incomplete left dentary); IPFUB 102/22.5 (one caudal and two dorsal vertebrae); IPFUB 102/22.6 (proximal left femur fragment). This material is likely to all be from the same individual, and Brinkmann (1989, 1992) regarded all of the referred material as belonging to the holotype individual, a view that we follow here.

Tentatively referred material

BUE4-NT2#25 (teeth) (La Huerguina Limestone Formation, late Barremian (Early Cretaceous); El Inglés Quarry, Serranía de Cuenca, eastern Spain); provisionally housed in the Unidad de Paleontología, Universidad Auténoma de Madrid, and will be eventually stored in the Museo de Ciencias de Castilla-La Mancha in Cuenca, Spain, at which point permanent catalogue numbers will be assigned (Buscalioni et al., 2008).

Previous diagnoses and comments [3] Both Brinkmann (1989, 1992) referred to T. ibericus as a new species, and therefore Brinkmann (1989) has taxonomic priority. The original specific designation was based on a range of features, including variation in tooth morphology and the width of the internal choanae, as well as the presence of pro-coelous caudal vertebrae. However, Schwarz & Salisbury (2005) and Martin et al. (2010) regarded this species to be of questionable validity, and it remains poorly understood. Here, we demonstrate that

T. ' ibericus is a valid taxon, with the new combination Sa. ibericus.

In spite of the differences we have recovered here, and the novel phylogenetic position for Sa. ibericus, this taxon retains some morphological similarities to T. pusillus, observations that resulted in its original generic assignment to Theriosuchus. Similar to T. pusillus, Sa. ibericus possesses a longitudinal median ridge on the parietal and frontal, and a deep groove between the squamosal and parietal on the dorsal surface of the skull table (Brinkmann, 1992). However, these features have since been found to be more widespread within crownward neosuchians, including Paralligatoridae (Turner, 2015; Turner & Pritchard, 2015), and cannot be used to unambiguously unite Sa. ibericus with Theriosuchus.

Sabresuchus ibericus retains several of the diagnostic dental morphotypes of Theriosuchus, possessing labiolingually compressed pseudocaniniform, and low-crowned teeth, but these teeth are fully zipho-dont (i.e. with denticulate carinae) in Sa. ibericus. There is a third morphotype figured in Brinkmann (1989), in which the labial surface of one of the teeth from the middle of the tooth rows is punctured by a series of small pits, which might prove to be an additional dental morphotype. The bases of the middle to posterior tooth crowns are mesiodistally constricted and, despite an overall labiolingual compression, contain a thickened central core to each tooth. The pseu-docaniniform teeth of Sa. ibericus show evidence of apicobasal striations on the labial sides of the teeth, which terminate 40% of the way towards the tip of the crown (Brinkmann, 1989).

Teeth that might be referable to Sa. ibericus (BUE4-NT2#25) have also been reported from a second locality in the Barremian of eastern Spain (Bus-calioni et al., 2008). We tentatively follow this referral, pending their further study.

Diagnosis and discussion

(S1) Palatal surface of maxilla sculpted throughout by ridged ornamentation (C57.2): The palatal surface of Sa. ibericus (IPFUB 102/21.2) is sculpted with longitudinal ridges, a feature otherwise found only in Amphicotylus (Mook, 1942). This is distinct from the condition seen in notosuchians (e.g. Pol & Powell, 2011) and other advanced neosuchians (e.g. Salisbury et al., 2006; Osi et al., 2007; Adams, 2013), as well as Theriosuchus and Sa. sympiestodon (Martin et al., 2014a), in which the maxillary palatal surface is flat and smooth. The posterior palatal surface ofthe maxilla is slightly crenulated in the mesoeucrocodylian Mahajangasuchus (Turner & Buckley, 2008), but this is not the same as the marked sculpting seen in Sa. ibericus.

(S2*) Dorsal surface of the parietal depressed relative to the squamosal (C115.2): This depression might be partially explained by mediolateral compression of this element. However, for taphonomic processes to fully explain this dorsal concavity, it would be expected for this degree of compression to be visible on other specimens from the type locality, which is not the case. In other mesoeucrocodylians, including Sa. sympiestodon (Martin et al., 2014a), the posterior portion of the dorsal surface of the skull table is horizontal and flat (e.g. Sereno et al., 2003; Osi et al., 2007; Adams, 2013).

(S3) Proportionally short mandibular symphysis, extending posteriorly medial to the fifth dentary alveolus (C205.0): The posterior extension of the symphysis in Sa. ibericus is proportionally shorter than in Theriosuchus (see also Young et al. , 2016) and notosuchians (e.g. Pol & Apesteguia, 2005), terminating medial to the D5 alveolus. This condition is only found in other brevirostrine taxa, such as Brillanceausuchus, Pachycheilosuchus (Rogers, 2003), and possibly Iharkutosuchus (Osi et al., 2007). In Sa. sympiestodon, the symphysis extends posteriorly to the sixth dentary alveolus (based on specimen MCDRD 134; Martin et al., 2014a).

(S4*) All dentary teeth occupy single, continuous, longitudinal groove (C217.2): In Theriosuchus, at least some of the dentary alveoli are independent from one another, with each tooth occupying a single alveolus and separated by interalveolar septae of varying thickness. However, Sa. ibericus is unique in that all of its dentary teeth appear to occupy a single continuous alveolar groove, along a mediola-terally narrow dentary occlusal surface. This feature is considered to be locally autapomorphic for Sa. ibericus, as it has also been documented in basal notosuchians (Pol & Apesteguia, 2005), the hylaeochampsid Iharkutosuchus (Osi et al., 2007), and the posterior dentary teeth of Isisfordia (Salisbury et al., 2006).

(S5) Occlusal dentary surface strongly medio-laterally compressed and devoid of nutrient foramina (C220.0): Sabresuchus ibericus does not appear to possess any distinctive foramina on the dorsal surface of the dentary, lingual to the mediolaterally compressed dental arcade (Young et al., 2016). By contrast, Theriosuchus, Sa. sympiestodon (Martin et al., 2014a), and notosuchians (e.g. Araripesuchus) have a mediolaterally broad dentary occlusal surface, pierced by at least one foramen (Pol & Apesteguia, 2005; Young et al., 2016). The overall morphology of the dentary of Sa. ibericus

is similar to that of Pachycheilosuchus, but the latter has a medially curved anterior end, giving a 'scimitar-shaped' profile in dorsal aspect. (S6) Grooved ornamentation (rugose patterning) present posteriorly on the external surface of the dentary (C227.1): The presence of a grooved ornamentation on the external surface of the dentary is similar to the condition observed in Theriosuchus, as well as Pachycheilosuchus (Rogers, 2003), Brillanceausuchus, and notosu-chians (Ortega et al., 2000; Pol & Apesteguia, 2005), but distinguishes Sa. ibericus from Sa. sympiestodon (Martin et al., 2014a). In basal crocodyliforms, sculpture patterns consist exclusively of evenly spaced subcircular pits (e.g. Pol et al., 2004) (S7*) Hypertrophied fifth maxillary tooth with fully caniniform morphology (C257.1), directed posteroventrally with respect to rest of maxillary tooth row (C259.0): The enlarged maxillary tooth is at least four times the total size of the adjacent teeth, possesses growth rings of varying coloration, and is poste-roventrally recurved. In Sa. sympiestodon, an enlarged tooth is also present at this position, but is directed ventrally, and not as proportionally enlarged with respect to the adjacent maxillary teeth. This level of heterodonty is not seen in Theriosuchus. (S8) Tooth crowns with denticulate carinae (fully ziphodont condition) (C245.1): Ziphodonty is known in a range of mesoeucrocodylians, including notosuchians (Pol & Powell, 2011). The difference between the heterodont morphologies of the maxillary tooth crowns of T. pusillus and Sa. ibericus is that whereas the former have slightly crenulated enamel surfaces leading to false' ziphodonty (pseudoziphodonty; Prasad & De Lapparent De Broin, 2002), the latter have 'true' serrations on the apical surfaces of the teeth, giving a fully ziphodont morphology. The false' ziphodont condition forms via the apical prolongation of the enamel ridges on the labial and lingual enamel surfaces (Prasad & De Lapparent De Broin, 2002), rather than through the development of an incisive and serrated texture.

Sabresuchus sympiestodon comb. nov. Theriosuchus sympiestodon Martin et al. , 2010

Type locality and horizon

Densus-Ciula Formation (upper part of unnamed middle member), Maastrichtian (Late cretaceous);

Oltoane Hill, Tustea, Hateg Basin, western Southern Carpathians, Romania.

Type specimen

FGGUB R.1782, a right maxilla with in situ teeth. Paratype

FGGUB R.1781, skull roof that might belong to the holotype individual.

Referred material

MCDRD 134, anterior portion of a right dentary from the Sânpetru Formation (Maastrichtian) at Cioaca Tarnovului, Sanpetru, Romania; MCDRD 793, a maxilla with teeth from the Saânpetru Formation (Maastrichtian) at La Cärare, Sânpetru, Romania; LPB (FGGUB) R.1945, a fragmentary maxilla from the Densus-Ciula Formation (Maastrichtian), Valioara-Faântaânele, in the Hateg Basin, Romania.

Previous diagnoses and comments Martin et al. (2010) assigned a series of fragmentary mesoeucrocodylian remains to Theriosuchus, erecting the new species T. sympiestodon. Additional material was subsequently referred to this taxon from nearby localities of approximately the same age (Martin et al., 2014a). A dentary (MCDRD 134) of a different provenance was referred to 'T.' sympiestodon based upon the similarity of its tooth morphology to that of T. pusillus. However, this specimen is poorly preserved and does not share any clear features to unite it with the type specimen of 'T.' sympiestodon, although we have followed its referral here. The original referral of 'T.' sympiestodon to Theriosuchus was based on: (1) the presence of a transversely (or ventrolaterally) directed groove on the anterolateral side of the maxilla; (2) a longitudinal crest on the frontal; and (3) the presence of low-crowned, labiolin-gually compressed, pseudoziphodont posterior maxillary teeth. However, the presence of this maxillary groove could not be confirmed via personal examination of the type specimen, nor was it observed in Sa. ibericus or T. pusillus. Schwarz & Salisbury (2005) noted its presence for T. guimarotae and possibly Sa. ibericus, and Turner (2015) reported its presence on the paratype specimen of T. pusillus. On NHMUK PV OR48330, the left maxilla does exhibit a longitudinal mark, but this is parallel to the nasal-maxilla suture, almost orthogonal to that figured for 'T.' sympiestodon (Martin et al., 2010, 2014a), and does not appear to occur on any of the other specimens referred to Theriosuchus in the NHMUK collections. A referred specimen of Sa. sympiestodon (MCDRD 793) has a depression of some description in this area (Martin et al., 2014a), at an oblique orientation to that of the holotype specimen, although

we have not been able to examine this specimen first-hand, and so we cannot comment on whether it is a groove, bite mark, post-mortem artefact, or a pathology. Therefore, the presence and orientation of any potential maxillary groove, and indeed its functional significance, remain uncertain, especially given that almost all specimens assigned to Therio-suchus have undergone taphonomic distortion or damage. We do not consider it to be diagnostic for 'T.' sympiestodon, or the genus Theriosuchus, although it might be autapomorphic for T. guimaro-tae (Schwarz & Salisbury, 2005). Martin et al. (2010) concluded that'T.' sympiestodon lies outside of Eusu-chia because of the anterior contribution of the palatine to the internal nares' (which we assume to mean choana, owing to the progressive posterior migration of this feature from the anterior rostrum in Crocodyliformes), a feature that is not actually visible in the holotype specimen. The entire basioc-cipital region around the choana is poorly preserved, and it is difficult to assess whether the choana was fully pterygoidean or bound anteriorly by the posterior palatines. However, it does seem to be situated close to the posterior margin of the suborbital fenes-tra, similar to paralligatorids and T. pusillus. Here, we demonstrate that 'T.' sympiestodon is a valid taxon, with the new combination Sa. sympiestodon.

Revised diagnosis and discussion (S1*) Longitudinal ridge on dorsal surface of frontal restricted to median portion (C101.1): The presence of a midline frontal crest is likely to be related to ontogeny and the fusion of the frontals, as small individuals of T. guimarotae lack this crest (Schwarz & Salisbury, 2005). Alligatorium meyeri also possesses this feature, as well as a range of non-atoposaurid neosuchians, including Wannchampsus (Adams, 2014), Shamosuchus (Pol et al., 2009), and Isisfordia (Salisbury et al., 2006). However, the position of the ridge and its continuation on the parietal appear to be phylogenetically informative; in Sa. sympies-todon, this ridge is restricted to the middle portion of the frontal, whereas in T. guima-rotae and T. pusillus it is restricted to the posterior portion (Schwarz & Salisbury, 2005). In Sa. ibericus, Wannchampsus, and Shamo-suchus (Pol et al., 2009), this ridge occupies the entire length of the frontal, and therefore it is likely that this heterogeneity in anteroposterior extent is taxonomically informative. (S2) Dentary internal alveolar margins not raised, but flat and confluent with remainder of dentary occlusal surface (C22.1): The lack of raised internal alveolar rims in

Sa. sympiestodon distinguishes it from Therio-suchus and Sa. ibericus, which have raised rims at least in the anterior alveoli (Young et al., 2016). Eusuchians (e.g. Salisbury et al., 2006), thalattosuchians, and tethysu-chians also lack raised alveolar rims (Young et al., 2014), whereas they appear to be raised in notosuchians (e.g. Campos et al., 2011). This feature is not observable in atoposaurids because of the nature of their preservation.

(S3*) Diastema present on dentary between D7 and DS alveoli (C225.0): Sabresuchus sympiestodon possesses a diastema between the D7 and D8 alveoli, in contrast with Theriosuchus, as well as Pachycheilosuchus and Wannchampsus. We therefore consider this feature to be locally diagnostic for Sa. sympiestodon. However, as a result of the lack of anatomical overlap between this dentary and the type material of Sa. sympiestodon, we are cautious in our recognition of this feature as diagnostic for the species.

(S4*) Maxillary teeth lacking striae on the labial and lingual surfaces (C250.0), with hypertrophied (C25S.0) and ventrally directed fifth maxillary tooth (C259.1): Martin et al. (2010) distinguished 'T.' sympiestodon from other species of Theriosuchus based on the presence of a single, enlarged maxillary caniniform tooth, and the anterior maxillary teeth lacking striae on the labial and lingual faces, a conclusion with which we agree. Sabresuchus ibericus also possesses this enlarged maxillary tooth, although in the one available specimen this tooth is strongly posteriorly recurved, and not directed ventrally as in Sa. sympiestodon. In Shamosuchus, the fifth maxillary tooth is also the largest, whereas in T. pusillus and Allodaposuchus the fourth tooth is the largest (Delfino et al. , 2008a), but in none of these cases does the proportional size difference come close to that for either Sa. sympiestodon or Sa. ibericus. Therefore, we consider this combination of maxillary dental character states to be diagnostic for Sa. sympiestodon.

Brillanceausuchus babouriensis

Michard et al., 1990

Type locality and horizon

Unnamed bed, ?Barremian (Early Cretaceous),

Babouri-Figuil Basin, north Cameroon.

Type specimen

UP BBR 201, skull and partial skeleton.

Previous diagnoses and comments Despite noting numerous similarities with ato-posaurids, Michard et al. (1990) assigned Brillanceausuchus to its own family, Brillanceausuchidae, within Neosuchia. However, a monogeneric family has no systematic purpose, and Brillanceausuchidae has not been used by subsequent workers. Pending the recovery of closely related taxa that do not already form a named clade, we recommend disuse of Brillanceausuchidae. Bril-lanceausuchus has remained a neglected taxon in phylogenetic and comparative analyses, despite its apparent important morphology in possessing a number of 'primitive' character states (e.g. possession of a partially septated external nares and presence of a biserial osteoderm shield) alongside more transitional' morphologies between advanced neosuchians and eusuchians (e.g. reduced ventral exposure of the basisphenoid and procoelous presacral vertebrae) (Michard et al., 1990). It was regarded as an ato-posaurid by Salisbury & Frey (2001) and Salisbury et al. (2006), and as an advanced neosuchian' by Turner (2015), without additional comment. To our knowledge, the only phylogenetic analysis to include Brillanceausuchus was conducted by Osi et al. (2007: 174), who commented that its inclusion gave much less resolution inside Eusuchia due to its incompleteness' and did not report the results.

Discussion

Brillanceausuchus possesses procoelous cervical and dorsal vertebrae (Michard et al., 1990), as well as fully pterygoidean choanae that are situated posteriorly to the posterior edge of the suborbital fenestrae, as in eusuchians (Buscalioni et al., 2001; Pol et al., 2009). Many authors have considered the presence of this combination of vertebral and palatal morphologies to imply that the eusuchian condition has evolved in parallel in several different neosuchian lineages, based on the underlying assumption that Brillanceausuchus is an atoposaurid, and therefore more basally positioned within Neosuchia (e.g. Bro-chu, 1999; Buscalioni et al., 2001; Salisbury et al., 2006). However, our preliminary results indicate that Brillanceausuchus belongs to Paralligatoridae (Figs 4B, 5, 6, 7), a clade most recently placed within Eusuchia (Turner, 2015). Therefore, the eusuchian condition might not be as homoplasious as previously regarded. Additional material assigned to Bril-lanceausuchus is currently being prepared, and comprises numerous skeletons (including skulls) preserved in three dimensions (J. Martin, pers. comm., 2015). We await the full description of this material before a comprehensive taxonomic assessment of Brillanceausuchus can be made, and preliminarily assign it to Paralligatoridae.

Preliminary emended diagnosis

(51) proportionally long supratemporal fenestra, with anteroposterior length exceeding that of the orbit, and a skull length to supratemporal fenestra length ratio of <6.0 (5.36) (C29.0), and a skull width to supratemporal fenestra width ratio of 7.0 (C30.3);

(52) sinusoidal lateral nasal borders oblique to one another (C66.3), with abrupt widening adjacent to maxilla (C70.1); (S3) base of jugal postorbital process directed dorsally (C83.1); (S4) flat frontal dorsal surface (no longitudinal crest or periorbital rims) (C100.0); (S5) parietal-postorbital suture visible on the dorsal surface of the skull roof (C112.1) and within the supratemporal fenestra (C113.1); (S6) concavity at posterodorsal edge of squamosal-parietal contact (C117.1); (S7) lateral margins of squamosal and postorbital medially concave in dorsal view (C134.2), and dorsal surface of squamosal bevelled ventrally (C138.1), becoming unsculpted anteriorly (C148.1); (S8) squamosal posterolateral process elongate, distally tapered (C143.0) and depressed from skull table (C140.1); (S9) basisphenoid ventral surface mediolaterally narrower than basioccipital (C191.0), and basioccipital with large, well-developed bilateral tuberosities (C192.1); (S10) ventrolateral surface of anterior portion of dentary strongly mediolaterally compressed and flat (C215.0), with grooved ornamentation on external surface (C227.1); (S11) retroarticu-lar process projects posteriorly and dorsally recurved (C242.3); (S12) posterior dentary teeth occlude medial to opposing maxillary teeth (C263.0); (S13) rounded and ovate dorsal osteoderm shape (C308.0).

Pachycheilosuchus trinquei, Rogers 2003

Pachycheilosuchus trinquei is known from a near-complete, disarticulated skeleton and partial skull from the Albian (Early Cretaceous) Glen Rose Formation of Erath County, Texas, USA. Initially described as a possible atoposaurid (Rogers, 2003), a position at the base of Atoposauridae was subsequently demonstrated in the analyses of Turner & Buckley (2008) and Pol et al. (2009). However, more recent analyses have placed Pachycheilosuchus outside of Atoposauridae, either within the basal eusu-chian clade Hylaeochampsidae [Buscalioni et al., 2011 (although note that this study used Theriosuchus as an outgroup, and included no definite atoposaurids); Turner & Pritchard, 2015], or just outside the eusuchian radiation (Adams, 2013; Narvaez et al., 2015). Rogers (2003) based his assignment to Atoposauridae primarily on the presence of a jugal with equally broad anterior and posterior processes, and the possession of procoelous presacral vertebrae. However, this jugal morphology is known in other neosuchians, including Paluxy-

suchus, as well as thalattosuchians (Adams, 2013). The presence of procoelous vertebrae could be more broadly distributed amongst non-neosuchian eusuchi-ans than previously recognized, and full procoely is not definitively known amongst any atoposaurid species (see also Salisbury & Frey, 2001). Hylaeochamp-sid affinities are supported by the reinterpretation of a defining character state for Atoposauridae, pertaining to whether the bar between the orbit and supratemporal fenestra is narrow, with sculpting restricted to the anterior surface (Clark, 1994). Buscalioni et al. (2011) regarded this feature to be associated more broadly with dwarfism' (as initially proposed for Pachycheilosuchus) or immature specimens, and not a synapomorphy of Atoposauridae. Pachycheilosuchus is additionally unusual in the retention of an antorbital fenestra, to which the maxilla contributes (Rogers, 2003), which is similar to T. guimarotae (Schwarz & Salisbury, 2005) and Alligatorellus bavaricus (Tennant & Mannion, 2014).

The present study was not designed to resolve the phylogenetic placement of Pachycheilosuchus, except whether or not to include it within Atoposauridae. We support its exclusion from Atoposauridae, but cannot provide further comment on its placement within Hylaeochampsidae (Buscalioni et al., 2011). It is unusual in that we recover Pachycheilosuchus in a more stemward position than Atoposauridae. We anticipate that inclusion of a broader range of ato-posaurid specimens, Theriosuchus species, hylaeochampsids (including Pietraroiasuchus ormez-zanoi; Buscalioni et al., 2011), and additional paralli-gatorids within a larger Neosuchia-focussed data matrix, will help to resolve the position of Pachy-cheilosuchus and its clearly important role in the ascent of advanced neosuchians and Eusuchia.

Wannchampsus kirpachi Adams, 2014

The Glen Rose Form' has been commonly referred to in neosuchian systematics since it was first briefly mentioned and figured by Langston (1974). Comprising a skull and lower jaw from the Early Cretaceous (late Aptian) Antlers Formation of Montague County, Texas, uSA, it was described as resembling the extant dwarfed crocodile Osteolaemus tetraspis, and Langston (1974) also noted similarities to T. pusillus. Subsequently, Adams (2014) erected Wannchampsus kirpachi for two skulls and postcranial material from the late Aptian (Early Cretaceous) Twin Mountains Formation of Comanche County (Texas, uSA), and assigned the 'Glen Rose Form' to this taxon. Adams (2014) noted that the skull of Wannchampsus was similar to that of T. pusillus, sharing features such as medial supraorbital rims, and therefore prompting its inclusion in the present analysis and discussion here.

However, Wannchampsus is distinct from T. pusillus in: (1) the possession of an enlarged third maxillary tooth (instead present at the fourth position in T. pusillus); (2) the absence of an antorbital fenestra; (3) choanae with an anterior margin close to the posterior edge of the suborbital fenestra (whereas it is more anteriorly placed in T. pusillus); and (4) the definitive presence of procoelous dorsal and caudal vertebrae. We recovered Wannchampsus in a position close to Shamosuchus [Pol et al., 2009; see also Adams (2014) and Turner (2015)], forming a paralligatorid clade with Sabresuchus and Brillanceausuchus.

Karatausuchus sharovi Efimov, 1976

Karatausuchus sharovi is known only from a single skeleton of a juvenile individual from the Late Jurassic (Oxfordian-Kimmeridgian) Karabastau Formation in southern Kazakhstan. It was considered to be an atoposaurid by Efimov (1976, 1988), but more closely related to paralligatorids by Efimov (1996). Storrs & Efimov (2000) argued that it was a relatively basal crocodyliform owing to the possession of amphiplat-yan vertebral centra, and designated it as a questionable atoposaurid. It is generally similar to atoposaurids in being small, at only 160 mm in total anteroposterior body length, but possesses reduced dermal osteoderms, suggestive of a juvenile phase of growth. Intriguingly, Storrs & Efimov (2000) observed over 90 small, labiolingually compressed teeth within the jaws, a feature unique amongst crocodyliforms. It also possesses 46 caudal vertebrae, approaching the condition known for Atoposaurus. However, it has eight cervical vertebrae, placing it intermediate to Protosuchus (nine cervical vertebrae) and the majority of other atoposaurids (seven cervical vertebrae, with the exception of Atoposaurus jour-dani, which appears to have six). Karatausuchus is similar to atoposaurids in that its skull length to orbit length ratio is relatively low, between 3.0 and 4.0 (3.37), but the other diagnostic features presented in this study for Atoposauridae cannot be assessed in the single known specimen. Therefore, we agree with Buscalioni & Sanz (1988) that Karatausuchus sharovi is currently too poorly known to be assigned to any family, including Atoposauridae, and regard it as an indeterminate crocodyliform. However, we still tentatively regard it as a valid taxon, owing to the high number of cervical and caudal vertebrae, and the possession of an anomalously high number of teeth.

Hoplosuchus kayi Gilmore, 1926

Gilmore (1926) originally recognized this taxon, based on a near-complete and articulated skeleton from the Late Jurassic Morrison Formation at Dino-

saur National Monument (Utah, USA), as a pseudo-suchian archosaur. Subsequently, several authors assigned Hoplosuchus kayi to Atoposauridae (Romer, 1956; Kuhn, 1960; Steel, 1973), based on its overall size, and the possession of relatively large, posteriorly placed, and anterolaterally facing orbits. However, Buffetaut (1982) and Osmolska et al. (1997) regarded Hoplosuchus as more similar to protosuchians, but noted that its phylogenetic affinities remained uncertain. In their phylogenetic assessment of Atoposauri-dae, Buscalioni & Sanz (1988) concluded that Hoplosuchus is a protosuchian-grade' crocodyliform.

Our examination of this taxon could not confirm any definitive atoposaurid affinities. Hoplosuchus retains features present in basal crocodyliforms, including a small and circular antorbital fenestra, and a triangular lateral temporal fenestra that is nearly as large as the orbit. Potential autapomorphies for Hoplosuchus include: (1) a steeply posteriorly inclined quadrate; (2) the pterygoid bearing a descending process that is extensively conjoined in the mid-line anterior to the basisphenoid; and (3) a lower jaw lacking the external mandibular fenestra (Steel, 1973). Hoplosuchus has slender limbs, the dorsal armour is composed of paired oblong plates, and the caudal region is completely enclosed by dermal ossifications. A full revision of pro-tosuchian crocodyliforms is currently underway (A. Buscalioni, pers. comm., 2014), and we await this before drawing any conclusions about the affinities of Hoplosuchus. Nonetheless, we exclude Hoplosuchus from Atoposauridae.

Shantungosuchus chuhsienensis Young, 1961

Young (1961) initially identified this taxon, based on a near-complete skeleton and skull from the Early Cretaceous Mengyin Formation of Shandong Province, China, as an atoposaurid. This referral was subsequently supported by Steel (1973), who provided an emended diagnosis in his discussion of Ato-posauridae. This included: (1) a triangular-shaped skull; (2) closely set teeth deeply implanted in independent alveoli; (3) seven cervical and 18 dorsal vertebrae; (4) short cervical vertebral centra; (5) relatively long dorsal vertebral centra; (6) a slightly shorter ulna than humerus; (7) the tibia significantly exceeding the femur in length; and (8) the forelimbs being proportionally long. However, Buffetaut (1981) and Buscalioni & Sanz (1988) both excluded Shantungosuchus from Atoposauridae. Wu et al. (1994) regarded much of the original interpretation of Young (1961) as incorrect, and revised Shantungo-suchus, finding it to be more closely related to proto-suchians than to atoposaurids. We concur with these authors and exclude Shantungosuchus chuhsienensis from Atoposauridae, supporting a basal position

within Crocodyliformes. However, we note that ato-posaurids do share numerous metric features with protosuchians, reflecting their small body size and paedomorphic retention of basal morphologies.

Indeterminate remains previously attributed to Atoposauridae

Alongside these named taxa, numerous additional remains (primarily teeth) have been referred to Ato-posauridae. These referrals have generally been based on the dental morphotypes that have been regarded as characteristic of Theriosuchus and, in stratigraphical order, comprise:

1. Teeth comparable to those of Theriosuchus were described from two localities from the early Bathonian of southern France (Kriwet et al., 1997). Their referral to Atoposauridae was based on the presence of different morphotypes, and the teeth were thought to represent two distinct species. The first of these 'species' includes several dozen teeth (Larnagol, IPFUB Lar-Cr 120, and Gardies, IPFUB Gar-Cr 1-20). Amongst this set of teeth, Kriwet et al. (1997) identified four gradational morphotypes, based on their inferred positions in the dental arcade. However, their referral to Atoposauridae is based mainly upon them being heterodont, a feature that is not exclusive to either Atoposauridae or Theriosuchus. The second species' (Larnagol, IPFUB Lar-Cr 21-40) differs in possessing more prominent ridges on the crown surfaces (Kriwet et al., 1997). These were referred to Atoposauridae by Kriwet et al. (1997) based on their inferred heterodonty; however, it cannot be determined whether or not all of these morphotypes belong to the same heterodont taxon, or two or more homodont or heterodont taxa. As atoposaurids are now considered to have a homodont (pseudocaniniform) dental morphology, these teeth cannot be referred to this group. They probably represent at least one (and probably more) small-bodied heterodont taxon, and therefore we consider them to be only referable to Mesoeucrocodylia indet. at present.

2. small crocodyliform teeth were noted from the Bathonian stipite' layers of the Grand Causses (France) by Knoll et al. (2013) and Knoll & López-Antonanzas (2014), and referred to an indeterminate atoposaurid. No further details were given, and therefore we regard these as representing indeterminate crocodyliforms pending further description of this material.

3. 1391 specimens comprising teeth, osteoderms, and a jaw fragment with teeth, as well as unde-scribed cranial and postcranial specimens, from

the Bathonian of Madagascar were referred to Atoposauridae (Flynn et al., 2006). Based on the description and figures provided, these teeth appear to be pseudocaniniform in morphology, with well-developed mesial and distal carinae and a ridged enamel surface. Although they vary in shape and size (up to 10 mm in apicobasal length), none can be defined as pseu-doziphodont, ziphodont, lanceolate, labiolin-gually compressed, or low-crowned. Based on the brief description, we cannot conclude that these teeth belonged to an atoposaurid, and therefore regard them as Mesoeucrocodylia indet., pending further description. Further examination of this material, along with remains identified as Theriosuchus sp. by Had-doumi et al., [2016 (see above)], will be important in examining evidence for the presence of atoposaurids and Theriosuchus in the Middle Jurassic of Gondwana.

4. Thies & Broschinksi (2001) described teeth from the Kimmeridgian of northern Germany as 'The-riosuchus-like', but identified them as belonging to a small-bodied mesosuchian. Karl et al. (2006) provisionally referred these teeth to Mesoeu-crocodylia indet., stating that their morphology is not known for any other crocodylomorph. We follow this decision of Karl et al. (2006), pending the direct comparison of this material with Theriosuchus and other small-bodied crocodyliforms.

5. A fragmentary set of specimens (IvPP v10613) from the Early Cretaceous of Inner Mongolia, including cranial and mandibular elements, were assigned to cf. Theriosuchus sp. (Wu et al., 1996). However, the material might not be referable to a single individual or even taxon, as it was collected from across an extensive outcrop. The figured osteoderm (Wu et al., 1996) is almost identical in overall morphology to Alli-gatorellus sp. (MB.R.3632; Schwarz-Wings et al., 2011), including the position and extent of the lateral keel, and the near-absence of the anterior process of the ilium is similar to that of Montsecosuchus. The dorsal vertebrae possess the semi-procoelous' condition, similar to Pachycheilosuchus. The cranioquadrate canal is closed, and therefore IVPP V10613 can be excluded from Atoposauridae. The external mandibular fenestra is absent, similar to paral-ligatorids and T. pusillus. Additionally, the parietals bear a longitudinal median ridge on the dorsal surface, which Wu et al. (1996) used to link IVPP V10613 with Theriosuchus, although this feature is herein shown to be more widespread throughout Neosuchia. Wu et al. (1996) assigned IVPP V10613 to Therio-

suchus based on the broad intertemporal region, raised supratemporal rims, and elevated medial orbital margin, but these features are found in numerous other taxa. Based on this combination of unusual characteristics, we think it likely that IVPP V10613 comprises more than one taxon, including at least one non-atoposaurid, non-Theriosuchus taxon, and one Theriosuchus-like taxon. We therefore regard IVPP V10613 as representing Neosuchia indet. pending further study of this material.

6. A skull fragment (NHMUK PV OR176) was assigned to Theriosuchus sp. from the Berri-asian-Barremian of the Isle of Wight, UK (Buf-fetaut, 1983; Salisbury & Naish, 2011). Buffetaut (1983) assigned the posterior portion of a skull to Theriosuchus sp. based on comparison with the lectotype specimen of T. pusillus (NHMUK PV OR48216). It has a median longitudinal ridge on the parietal, similar to all specimens assigned to Theriosuchus, but also to Alligatorium meyeri and paralligatorids. The otoccipitals also meet dorsal to the foramen magnum, separating it from the supraoccipitals, a feature shared with T. pusillus, T. guimaro-tae, and paralligatorids. The contact between the parietal and the squamosal on the dorsal surface, posterior to the external supratemporal fenestra, is also weakly developed, not forming the deep groove that characterizes T. pusillus. Therefore, we cannot determine whether this specimen represents Theriosuchus or another advanced neosuchian, and thus we consider this specimen to represent Neosuchia indet., pending its comparison to a broader set of neosuchi-ans. In addition, Buffetaut (1983) assigned some procoelous vertebrae (the type of Hetero-suchus valdensis') from the Early Cretaceous of the UK to Theriosuchus. The presence of pro-coely indicates that it is not referable to Therio-suchus (see below), and it is instead regarded as an indeterminate neosuchian.

7. Indeterminate remains (primarily teeth) attributed to atoposaurids, usually referred to Therio-suchus based on heterodont tooth morphotypes, have been identified from numerous sites in the Aptian-Albian (late Early Cretaceous) of North America (e.g. Pomes, 1990; Winkler et al., 1990; Cifelli et al., 1999; Eaton et al., 1999; Fiorillo, 1999; Garrison et al. , 2007; Oreska, Carrano & Dzikiewicz, 2013). However, because of our removal of Theriosuchus from Atoposauridae, it is more likely that these atoposaurid' remains represent other small-bodied taxa. We tentatively consider these remains to represent Mesoeucrocodylia indet., pending further study.

8. Theriosuchus--like teeth (MTM V 2010.243.1) were described from the Santonian of western Hungary (Iharkuot), but conservatively referred to Mesoeucrocodylia indet (Osi et al., 2012). These teeth are lanceolate in crown morphology, and possess pseudoziphodont carinae. Martin et al. (2014a) briefly mentioned the presence of two additional undescribed maxillae from the same locality, which together with the teeth might be referable to Theriosuchus. As we have recombined the Maastrichtian occurrences of The-riosuchus' into a new taxon, Sa. sympiestodon, it is best that these teeth be regarded as Neosuchia indet., pending further analysis of this material and the possibly associated maxillae.

9. A Theriosuchus-like tooth was described from the Campanian-Maastrichtian of Portugal by Galton (1996). This tooth is distinct from Theriosuchus, possessing a fully ziphodont morphology, and was suggested to instead belong to Bernissartia (Lauprasert et al., 2011). However, here we consider it to belong to an indeterminate neosuchian based on the more widespread distribution of ziphodont dentition.

10. The stratigraphically youngest material assigned to Atoposauridae comes from the middle Eocene Kaninah Formation of Yemen (Stevens et al. , 2013). This fragmentary material was tentatively designated as an atoposaurid, based on the presence of a ziphodont tooth crown, a procoelous caudal vertebral centrum, a biserial osteoderm shield (although see below), and polygonal gastral osteoderms. However, none of these characteristics is unambiguously diagnostic under our revised definition of Atoposauridae, and this material probably comprises a small, advanced eusuchian, based on the presence of procoelous caudal vertebrae. The presence of a biserial osteoderm shield is usually considered diagnostic for Atoposauri-dae; however, our analyses demonstrate that this feature is more widespread amongst small-bodied neosuchians. Furthermore, the material from Yemen is too fragmentary to confidently infer that the osteoderm shield was biserial. Therefore, we regard this material as an indeterminate eusuchian pending the discovery of more complete and better preserved specimens.

DISCUSSION

Phylogenetic relationships and

systematic implications

In all of our analyses, Atoposauridae is recovered in a basal position within Neosuchia, supporting several

recent analyses (e.g. Figueiredo et al., 2011; Adams, 2014). Atoposauridae is a much more restricted clade than previously recognized, comprising Atoposaurus (Atoposaurus jourdani and Atoposaurus oberndorferi), Alligatorellus (Alligatorellus bavaricus and Alliga-torellus beaumonti), and Alligatorium meyeri, and excluding Theriosuchus, Montsecosuchus, Bril-lanceausuchus, and Alligatorium' franconicum (as well as taxa previously demonstrated to lie outside Atoposauridae, e.g. Hoplosuchus).

The majority of previous phylogenetic analyses focused on crocodyliforms or basal neosuchians generally only incorporated Theriosuchus (usually T. pusillus) and Alligatorium meyeri, with other atoposaurid taxa rarely included (e.g. Pol et al., 2009; Adams, 2014). It is likely that this taxonomic under-sampling is at least partly responsible for the conflicting systematic positions previously recovered for Atoposauridae (Fig. 2). Alligatorium meyeri does not display any clear derived eusuchian features, whereas Therio-suchus has an unusual combination of derived, ple-siomorphic, and transitional' character states. This is the most likely explanation for the results recovered here, in which specimens traditionally assigned to Atoposauridae are split', representing clusters of basal neosuchians (i.e. true' atoposaurids), advanced non-paralligatorid neosuchians (i.e. Theriosuchus), and paralligatorids (i.e. Sabresuchus: see below). Our results further demonstrate that Theriosuchus had become a taxonomic waste-basket', to which discoveries of teeth representing small-bodied heterodont crocodyliforms were consistently attributed. It is likely that some of these teeth (and other fragmentary remains) from the Northern Hemisphere instead represent a much more taxonomically diverse group of neosuchians, including paralligatorids.

Evolutionary history of Atoposauridae, Theriosuchus, and Sabresuchus

As a result of our systematic revision, Atoposauridae is now restricted to the Late Jurassic of western Europe. Specimens previously assigned to Atoposauridae from the Middle Jurassic of Europe and Gondwana cannot be assigned to this clade, and most likely represent taxa closely related to Theriosuchus, or other small-bodied mesoeucrocodylian forms. This more restricted view demonstrates that atoposaurids were highly specialized, with a small body size, semi-aquatic lifestyle, and unusual limb proportions. They also possessed a biserial dorsal osteoderm shield, and their dentition was homodont. Atoposauridae appear to have gone extinct at the Jurassic/Cretaceous (J/K) boundary (Fig. 10), as part of an overall drop in biodiversity in both marine and terrestrial crocodyli-form groups (Mannion et al. , 2015), an event that

Figure 10. Time-scaled phylogeny showing the relationships of Atoposauridae to the other taxa analysed in the present study (based on the topology provided in Fig. 5A). Atoposauridae is marked with a red star. Created using the strap package (Bell & Lloyd, 2015), using the geoscalePhylo() function and an 'equal' time-scaling method.

was related to a major regression and the closing off of shallow marine basins in Europe across the J/K boundary (Hallam, 1986; Miller et al., 2005; Smith & McGowan, 2007; Tennant, Mannion & Upchurch, 2016; Tennant et al., in press).

The implications for our removal of Theriosuchus from Atoposauridae, and restriction of this genus to include just T. guimarotae, T. grandinaris, and T. pusillus, are complicated by the numerous referrals of teeth and poorly preserved fossils to Theriosuchus from across Laurasia (as well as less common referrals from Africa and North America). Therefore, although Theriosuchus appears to have been highly successful, spanning the Middle Jurassic-early Late Cretaceous, it was more temporally restricted than previously regarded, and did not persist into the latest Cretaceous (contra Martin et al., 2010, 2014a). Based on our current understanding, Theriosuchus was present in the Middle Jurassic to early Late Cretaceous of Europe, the Middle Jurassic of North Africa, and the Early Cretaceous of Asia. The unequivocal presence of Theriosuchus in the Cretaceous of Asia supports the hypothesis of intermittent connections between western Europe and Asia through part of the Early Cretaceous (Baraboshkin, Alekseev & Kopaevich, 2003; Lauprasert et al., 2011). Despite our taxonomic revisions, Theriosuchus remains one of the most temporally long-lived archo-

saurian lineages of all time, which could be attributed to its small body size and flexible morphology and ecology, or indicate that further splitting is required (pending the discovery of additional, well-preserved and complete specimens). Our results have recognized two separate and distinct clusters within Theriosuchus, possibly based on our more comprehensive sampling of definitive and putative ato-posaurids, resulting in the formal erection of a new genus, Sabresuchus. Given the temporal distance between the two species of Sabresuchus, we would not be surprised if new material of either species (especially the fragmentary Sa. sympiestodon) led to the recognition of two distinct genera.

The transition to Eusuchia Background

In this section, we follow the phylogenetic definition of Eusuchia provided by Brochu (1999): the last common ancestor of Hylaeochampsa vectiana and Croco-dylia and all of its descendants. Along with Wannchampsus (including the 'Glen Rose Form'), Theriosuchus (and by association, atoposaurids) has been considered to be an important taxon in understanding the transition to Eusuchia, because of the relatively advanced development of their secondary palates (e.g. Joffe, 1967; Brochu, 1999; Buscalioni

et al., 2001; Pol et al., 2009). A close relationship between Theriosuchus and Eusuchia is based on a number of features that are evident primarily in the type species T. pusillus, including: (1) an undivided external nares; (2) subdermal postorbital bars; (3) a lack of contribution from the frontals to the supratemporal fenestra; (4) relatively small internal supratemporal fenestrae; (5) development of the squamosal posterolateral process; (6) fully ptery-goidean choanae within the secondary palate; (7) an enclosed eustachian canal; (8) procoelous vertebrae; and (9) a biconvex first caudal vertebra (Joffe, 1967; Rogers, 2003; Pol et al., 2009; Turner, 2015; Turner & Pritchard, 2015). Theriosuchus shares combinations of these features with a range of more advanced neosuchians, particularly those assigned to Paralligatoridae (Turner, 2015; Turner & Pritchard, 2015). In addition, T. guimarotae shares a posteriorly opened cranioquadrate canal with Sa. sympiestodon, comparable to the paralligatorid Shamosuchus (Turner, 2015). As is also the case in the paralligatorids Sabresuchus, Shamosuchus, and Wannchampsus, the parietal of T. pusillus has a longitudinal midline ridge along its dorsal surface. Unfortunately, no specimens of Alligatorellus, Alligatorium, or Atoposaurus are preserved in a manner that would allow us to assess features such as the morphology of the choanae or the nature of presacral vertebral articulations, and we must await future examination of these specimens using 3D scanning techniques. Below, we discuss the evolution of several 'key' anatomical features that have played a significant role in discussions of the transition from basal neosuchians to Eusuchia.

Development of the squamosal posterolateral process Basal crocodyliforms and mesoeucrocodylians do not possess a posterolateral process (or lobe') on the squamosal (e.g. Pol & Norell, 2004a). Atoposaurids and other basal neosuchians, such as Paluxysuchus, have a posterolateral lobe that is in the same horizontal plane as the dorsal surface of the skull table, with this lobe sculpted in the latter taxon (Adams, 2013). In goniopholidids, the process is short, narrow, and typically unsculpted (e.g. Averianov, 2000), and in notosuchians it is proportionally longer and rhombohedral-shaped (Buckley et al., 2000), distinct from atoposaurids, Theriosuchus, or eusuchians. Theriosuchus pusillus and T. guimarotae possess this posterolateral process, but it is ventrally deflected and unsculpted in these taxa, similar to that of Rugosuchus and Shamosuchus (although in the latter taxon the lobe' is sculpted as in the rest of the dorsal surface of the cranial table; Turner, 2015). In paralligatorids (e.g. Sa. ibericus; see also Turner, 2015) and eusuchians (e.g. Allodaposuchus; Bus-

calioni et al., 2001) the posterolateral process is pos-teroventrally confluent with the paroccipital process, enclosing the otic aperture, and in most Brevirostres the posterior margin of the otic aperture is invagi-nated (Brochu, 1999). Sabresuchus ibericus is also similar to the peirosaurid Pepehsuchus (Campos et al., 2011) and to Rugosuchus (Wu et al., 2001a) in that the 'lobe' is separated by a step from the main body of the squamosal, and remains unsculpted compared with the skull roof. The ventral deflection of the posterolateral process from the plesiomorphic horizontally orientated state seems to have been an important stage in the acquisition of this eusuchian morphology. In dyrosaurids (Wu et al., 2001b; Jouve et al., 2005a,b) and gavialoids (Jouve et al., 2008), this process is ventrally directed and blade-like, forming the anterior face of the paroccipital process, and possibly relates to the transition to an aquatic lifestyle.

Development of the choanae

In crocodyliforms, the paired choanae have migrated from an anterior position within the primary palate (as in protosuchians), posteriorly through the ventral interorbital bar in neosuchians, and are positioned posteriorly within the pterygoids in eusuchians (e.g. Pol et al., 2009). This posterior positioning of the choanae, with respect to the suborbital fenestrae, coupled with the complete ventral enclosure by the pterygoids, has long been regarded as diagnostic for Eusuchia (Benton & Clark, 1988; Norell & Clark, 1990; Clark & Norell, 1992; Buscalioni et al., 2001; Salisbury et al., 2006; Pol et al., 2009).

Basal neosuchians, such as the goniopholidids Amphicotylus (Mook, 1942) Eutretauranosuchus (Smith et al., 2010; Fig. 11A), and Sunosuchus (Wu et al. , 1996), possess the plesiomorphic choanal condition, with the anterior portion of the choanae receiving a significant contribution from the palatines. In their reconstruction of T. guimarotae, Schwarz & Salisbury (2005: fig. 5B) placed the anterior border of the choanae in a more posterior position than that of T. pusillus, closer to the posterior border of the suborbital fenestra. However, based on the specimens figured, the anterior edge of the choanae and the pterygoid flanges appear to be broken (Schwarz & Salisbury, 2005: fig. 4I), and it seems more likely that the choanae extended anteriorly to a position similar to T. pusillus (see our reconstruction in Fig. 11B). Theriosuchus pusillus possesses an intermediate choanal morphology, with the anterior border of the choanae placed anteriorly with respect to the posterior border of the suborbital fenestra (Fig. 11C). The paired choanae in T. guimarotae appear to be mediolaterally narrower than those of T. pusillus, and seem to be fully bifurcated by a mid-

Figure 11. Relative positions of the choanae with respect to the main palatal bones in a range of neosuchian taxa. Citations are given were these reconstructions are based on in-text illustrations. (A) Eutretauranosuchus delfsi (Smith et al., 2010); (B) Theriosuchus guimarotae (Schwarz & Salisbury, 2005); (C) Theriosuchus pusillus; (D) Theriosuchus symp-iestodon; (E) Wannchampsus kirpachi; (F) Shamosuchus djadochtaensis (Pol et al., 2009); (G) Koumpiodontosuchus aprosdokiti (Sweetman et al., 2015); (H) Hylaeochampsa vectiana (Clark & Norell, 1992).

line pterygoidean septum, closer to the general morphology seen in goniopholidids (Mook, 1942; Wu et al., 1996; Smith et al., 2010) and notosuchians (Ortega et al., 2000; Pol & Apesteguia, 2005). However, in T. pusillus only the anterior portion of the choanal groove appears to be septated. In Sa. symp-iestodon (Martin et al., 2010, 2014a; Fig. 11D) and other paralligatorids (e.g. Rugosuchus; Wu et al., 2001a; Fig. 11E-G), the choanal groove is fully open and undivided, similar to Hylaeochampsa (Clark & Norell, 1992; Fig. 11H).

in both T. guimarotae and T. pusillus, the anterior edge of the choanae is formed by the posterior portion of the palatines, similar to Rugosuchus (Wu et al. , 2001a). However, in T. guimarotae the choa-

nae enter into the ventral lamina of the palatines (Schwarz & Salisbury, 2005), rather than just being bordered by it as in T. pusillus. in this respect, T. pusillus is similar to the advanced neosuchian Khoratosuchus, in which the palatines form only a point contact with the anterior margin of the fully open choanae (Lauprasert et al., 2009). Furthermore, the palatines contribute to the lateral margins of the choanal opening in T. pusillus, forming a bar-like, overlapping contact with the pterygoids. This morphology in T. pusillus is distinct from more crown-ward taxa [e.g. Brillanceausuchus, Gilchristosuchus (Wu & Brinkmann, 1993), Pietraroiasuchus (Bus-calioni et al., 2001), Shamosuchus (Pol et al., 2009), and Wannchampsus (Adams, 2014)], in which the

choanae are almost entirely enclosed by the pterygoids. The fact that this range of basal eusuchians (Turner, 2015; Turner & Pritchard, 2015) therefore appears to retain a palatine contribution to the anterior edge of the choana raises doubt over whether or not a fully pterygoidean choana is a synapomorphy for Eusuchia, rather than for a slightly more inclusive grouping. Theriosuchus, as the sister taxon to Paralligatoridae, exhibits a possible transitional morphology leading to the development of fully ptery-goidean choanae. The sequence appears to involve the posterolateral widening of the choanal groove, coincident with migration to a point level with the posterior border of the suborbital fenestra, and reduction and eventual loss of the pterygoidean septum. This transition might have occurred in several lineages of advanced neosuchians (e.g. Susisuchidae, Theriosuchus), and it is possible that additional features of Theriosuchus, such as the relatively anterior positions of the mandibular tooth rows (distinct from eusuchians, dyrosaurids, and Susisuchus, in which the tooth rows extend posteriorly), are related to the formation of the eusuchian palatal morphology (Salisbury et al., 2006).

Our understanding of the development of this morphology is potentially complicated by taxa such as Isisfordia, considered by Salisbury et al. (2006) to be a basal eusuchian, but placed outside of Eusuchia in many subsequent analyses (e.g. Turner & Pritchard, 2015). In this taxon, the choanae appear to have become fully enclosed by the pterygoids, with the anterior border located at the same level as the posterior edge of the suborbital fenestra (Salisbury et al. , 2006), which would indicate that the eusu-chian' condition had evolved independently in at least one other lineage. However, Turner & Pritch-ard (2015) reassessed this morphology and re-interpreted Isisfordia as possessing a palatine-pterygoid contact within the choanae, and therefore lacking completely pterygoidean choanae.

The neosuchian Bernissartia was also originally reported to have fully pterygoidean choanae based on its holotype (Buffetaut, 1975), but this specimen is poorly preserved, and Norell & Clark (1990) argued that it is unlikely that the choanae receive a pterygoidean contribution. However, the bernissar-tiid Koumpiodontosuchus does have a fully ptery-goidean choana (Sweetman et al., 2015: fig. 11G). Although the position of Bernissartia within Neo-suchia is unstable [see Turner (2015) for both a eusuchian and non-eusuchian placement], the presence of fully pterygoidean choanae in its sister taxon Koumpiodontosuchus (Sweetman et al., 2015) indicates that either: (1) it was secondarily lost in Bernissartia, or (2) that its presence in Koumpiodon-tosuchus represents an independent acquisition.

Regardless of whether the eusuchian' condition evolved more than once, it appears that the existing terminology does not describe the condition in basal eusuchians. As such, a more appropriately worded synapomorphy for Eusuchia could be the possession of posteriorly placed choanae, with an anterior margin medial to the posterior edge of the suborbital fen-estra, often with a posterior contribution from the pterygoids and the complete loss of any sagittal pterygoidean septum.

Development of vertebral procoely Norell & Clark (1990) identified the presence of pro-coelous presacral vertebrae to be diagnostic for Eusuchia. Based on a damaged anterior cervical vertebra referred to T. pusillus (NHMUK PV OR48723), Norell & Clark (1990) stated that Therio-suchus must have evolved procoely convergently (see also Clark & Norell, 1992), owing to the phylogenetic distance between Theriosuchus and Eusu-chia. Both Pachycheilosuchus and T. pusillus have been described as possessing semi-procoelous' dorsal centra, in which there is a convex posterior condyle, with a central depression sometimes filled by a plug. This intermediate morphology between amph-icoely and procoely has been hypothesized as an alternative route' to the fully procoelous condition (Rogers, 2003). The degree of procoely decreases posteriorly along the axial column in Pachycheilo-suchus, with the posterior caudal vertebrae possessing thickened centrum margins, which might represent the development of an incipient articular condyle, inferred as the possible origin of caudal procoely (Rogers, 2003). Salisbury & Frey (2001) examined the paratype of T. pusillus using X-ray scanning, and found that all presacral vertebrae are gently amphicoelous (contra Norell & Clark, 1990), as is also the case for T. guimarotae (Schwarz & Salisbury, 2005). Despite this, Salisbury & Frey (2001) tentatively concluded that at least some of the cervical vertebrae of T. pusillus were pro-coelous, based on NHMUK PV OR48273, an interpretation followed in the scoring of this taxon in many data matrices (e.g. Turner, 2015). However, NHMUK PV OR48273 is embedded in a block comprising several fragments (including a poorly preserved mandibular ramus) that was not found in association with the type specimen, and it does not preserve diagnostic features that would allow it to be unequivocally assigned to T. pusillus. Furthermore, examination of this specimen could not confirm the presence of procoely, with the vertebra instead being amphicoelous, with thickened rims along the centrum articular surfaces. Therefore we reject the hypothesis that this specimen, and thus T. pusillus, possessed any procoelous vertebrae.

By contrast, the presacral vertebrae are procoelous in the paralligatorids Wannchampsus (Adams, 2014) and Shamosuchus (Pol et al., 2009). Pachycheilo-suchus appears to possess a vertebral morphology more similar to paralligatorids than Theriosuchus, and therefore more likely to be representative of a transitional morphology between the neosuchian and eusuchian conditions. Isisfordia and Susisuchus both possess weakly or 'incipiently' procoelous vertebrae (Salisbury et al., 2006; Figueiredo et al., 2011), which has been suggested to also represent a transitional morphology from the plesiomorphic amph-icoelous condition (Turner & Pritchard, 2015).

Brinkmann (1989, 1992) described a series of fully procoelous caudal vertebrae, and others with a central depression (i.e. the 'semi-procoelous' condition), along with the material of Sa. ibericus and Bernissartia reported from the Una locality. Although the direct association of the procoelous caudal vertebrae with either of these taxa cannot be confirmed (Salisbury & Frey, 2001), Brinkmann (1989) stated that this vertebral series was found alongside the skull elements referred to Sa. ibericus. Based on the size of these vertebrae and the purported presence of a form of procoely in T. pusillus (Clark, 1986; Salisbury & Frey, 2001), coupled with the absence of this condition in Bernissartia (although some specimens of Bernissartia might have procoelous caudal vertebrae; Norell & Clark, 1990), Rogers (2003) also deemed it more probable that the fully procoelous vertebrae belong to Sa. ibericus (Brinkmann, 1989). However, their argument relies on the congeneric status between T. pusillus and Sa. ibericus, which we do not support here. Goniopholidids, also known from the Una region, can be excluded from 'ownership' of these vertebrae, because they have exclusively amphicoelous vertebrae (Brinkmann, 1989). Unasuchus reginae, also found at the Una locality, is only known from fragmentary cranial and mandibular material, and its affinities are uncertain within Neosuchia (Brinkmann, 1992). The semiprocoelous vertebrae are from a larger individual than that of the type of Sabresuchus, and might be referable to Unasuchus or an additional taxon. We have incorporated this uncertainty over the vertebral morphology of Sa. ibericus into our data matrix, electing not to code this taxon for the presence of procoely or a biconvex first caudal vertebra (see below). Nonetheless, this taxon groups with paralligatorids (which all show a degree of procoely), and therefore is the most likely candidate taxon at Una for the series of procoelous vertebrae. If this is correct, the entire vertebral column of Sa. ibericus would appear to show some form of procoely, and supports the findings of our study that Sabresuchus is closely related to eusuchians.

Pending the discovery of associated remains, we agree with Brinkmann (1989) and Salisbury & Frey (2001) that procoely, including semiprocoely, cannot currently be determined for Sabresuchus. Nevertheless, it seems likely that well-developed vertebral procoely only evolved once within the lineage leading to Eusuchia. However, the presence of weak procoely in the vertebrae of Susisuchidae (Isisfordia + Susisuchus; Turner & Pritchard, 2015) and semiprocoely in Pachycheilosuchus indicates that a number of neosuchian lineages developed some form of incipient procoely.

Biconvex first caudal vertebra

The first caudal vertebra of Pachycheilosuchus is biconvex (Rogers, 2003), a feature also proposed for Brillanceausuchus (Michard et al., 1990), but not confirmed through direct observation of the holotype specimen. A biconvex first caudal vertebra has also been documented for T. guimarotae (Schwarz & Salisbury, 2005), T. pusillus (Salisbury & Frey, 2001), and probably Sa. ibericus (Brinkmann, 1989, 1992; Salisbury & Frey, 2001; although see above regarding association). Moreover, a biconvex first caudal vertebra is a feature of all eusuchians (Salisbury et al. , 2006), and its development is probably an important morphological aspect of the neosuchian-eusuchian transition. Whether we code a biconvex first caudal vertebra as present or absent for Sa. ibericus has no effect on the topology of our tree, irrespective of whether or not we exclude Pachy-cheilosuchus a priori. This indicates that if a biconvex first caudal vertebra is indeed present in Brillanceausuchus, along with Pachycheilosuchus (Rogers, 2003) and Theriosuchus (Salisbury & Frey, 2001; Schwarz & Salisbury, 2005), then it is likely to have been independently acquired at least twice. The acquisition of this biconvex first caudal vertebra in Theriosuchus and other taxa close to the eusuchian radiation might have been important in the initiation of procoely in advanced neosuchians, and the differential acquisition of a concave posterior centrum con-dyle in different regions of the axial column.

CONCLUSION

Atoposauridae is now considered to be a much more restrictive clade of basal neosuchians, comprising only Atoposaurus jourdani, Atoposaurus oberndorferi, Alli-gatorellus beaumonti, Alligatorellus bavaricus, and Alligatorium meyeri. Based on this more exclusive taxonomic grouping, atoposaurids were restricted to the Late Jurassic of western Europe, and went extinct at the J/K boundary. We exclude Theriosuchus from Atoposauridae, recovering this genus as polyphyletic. Theriosuchus is a more crownward neosuchian than

atoposaurids, and is here restricted to T. pusillus (the type species), T. guimarotae (Schwarz-Wings et al., 2011), and T. grandinaris (Lauprasert et al., 2011). In addition, a specimen described by Young et al. (2016) from the Middle Jurassic of the UK might represent a distinct species of Theriosuchus. Therio-suchus is known from the Middle Jurassic-early Late Cretaceous, with occurrences from Europe, Asia, and North Africa. Theriosuchus' ibericus and T. ' sympiestodon are recombined under the new genus denomination, Sabresuchus gen. nov. Along with Brillanceausuchus, Sabresuchus is recovered as a paralligatorid.

'Alligatorium' franconicum is recovered outside of Atoposauridae, and is not referable to Alligatorium, but instead appears to be more closely related to lon-girostrine forms, such as bernissartiids and coelog-nathosuchians. As the only specimens referable to this taxon are lost or destroyed (Wellnhofer, 1971), we do not provide a new genus name, pending the discovery of new material. Montsecosuchus cannot be unequivocally confirmed to be an atoposaurid, and in the majority of our analyses we find it to cluster with Pachycheilosuchus.

Our revised placements of Atoposauridae and Theriosuchus have important implications for the transition from Neosuchia to Eusuchia. Theriosuchus exhibits one possible transitional route to the development of a pterygoidean-bound choana, involving the posterior migration and posterolateral widening of the choanal groove, with the reduction of the pterygoidean septum. No definitive atoposaurid possesses procoelous vertebral centra, and no specimen ascribed to Theriosuchus can be demonstrated to possess this feature. Therefore, current evidence suggests that full vertebral procoely only evolved once within the lineage leading to Eusuchia.

Future research on the systematic placement of Atoposauridae, Theriosuchus, and Paralligatoridae within Neosuchia will need to incorporate a wider set of taxa into phylogenetic analyses, such as the basal neosuchian Stolokrosuchus lapparenti (Larsson & Gado, 2000), hylaeochampsids [e.g. Paluxysuchus newmani [Adams, 2013), Hylaeochampsa vectiana (Clark & Norell, 1992), and Pietraroiasuchus ormez-zanoi (Buscalioni et al., 2011)], Bernissartia fagesii, and additional paralligatorids [e.g. Batrachomimus pastosbonensis (Montefeltro et al., 2013) and Rugosuchus nonganensis (Wu et al., 2001a)]. These additions should help to develop a clearer understanding of the Neosuchia-Eusuchia transition.

ACKNOWLEDGEMENTS

For access to specimens in their care, J. P. T. would like to thank Didier Berthet (MNHL, Centre de Con-

servation, Lyon); Zoltén Csiki-Sava (FGGUB); Géraldine Garcia (UP); Amy Henrici (CM); Carl Mehling (AMNH); Oliver Rauhut (BSPG); Chris Sagebiel (UT); Daniela Schwarz (MfN); Lorna Steel (NHMUK); Herman Voogd (TM); and Dale Winkler (SMU). We also gratefully acknowledge the Willi Hennig Society, which has sponsored the development and free distribution of TNT. For discussion throughout this research and manuscript preparation, J. P. T. would like to extend his thanks to Jeremy Martin, Thomas Adams, Mark Young, and Thomas Halliday. We thank the anonymous referee for their comments on this manuscript. Finally, for providing sustenance while preparing this paper, J. P. T. would like to thank Franziska Sattler, without whom he would be a skinnier man. J. P. T.'s research was funded by a Natural Environment Research Council PhD studentship (EATAS G013 13).

* Note that at the time of visit, the collections from the MNHL were housed at the Centre de Conservation in Lyon, but remained under the same accession numbers.

REFERENCES

Adams TL. 2013. A new neosuchian crocodyliform from the Lower Cretaceous (late Aptian) Twin Mountains Formation of north-central Texas. Journal of Vertebrate Paleontology 33: 85-101.

Adams TL. 2014. Small crocodyliform from the Lower Cretaceous (Late Aptian) of Central Texas and its systematic relationship to the evolution of Eusuchia. Journal of Palaeontology 88: 1031-1049. Ammon LV. 1906. Uber Jurassische krokodile aus Bayern.

Geognostische Jahresheften 18: 56-71. Andrade MB, Bertini RJ. 2008. A new Sphagesaurus (Mesoeucrocodylia: Notosuchia) from the Upper Cretaceous of Monte Alto City (Bauru Group, Brazil), and a revision of the Sphagesauridae. Historical Biology 20: 101-136. Andrade MB, Edmonds R, Benton MJ, Schouten R. 2011. A new Berriasian species of Goniopholis (Mesoeu-crocodylia, Neosuchia) from England, and a review of the genus. Zoological Journal of the Linnean Society 163: S66-S108.

Averianov AO. 2000. Sunosuchus sp (Crocodylomorpha, Goniopholididae) from the Middle Jurassic of Kirghisia. Journal of Vertebrate Paletonology 20: 776-779. Baraboshkin EY, Alekseev AS, Kopaevich LF. 2003. Cretaceous palaeogeography of the North- Eastern Peri-Tethys. Palaeogeography, Palaeoclimatology, Palaeoecology 196: 177-208.

Bell MA, Lloyd GT. 2015. Strap: an R package for plotting phylogenies against stratigraphy and assessing their strati-graphic congruence. Palaeontology 58: 379-389.

Benton MJ, Clark JM. 1988. Archosaur phylogeny and the relationships of the Crocodylia. In: Benton MJ, ed. The Phylogeny and Classification of Tetrapods, Volume 1: amphibians, Reptiles, Birds. Oxford: Clarendon Press, 295-338.

Blanco A, Puértolas-Pascual E, Marmi J, Vila B, Sellés AG. 2014. Allodaposuchus palustris sp. nov. from the Upper Cretaceous of Fumanya (South-Eastern Pyrenees, Iberian Peninsula): systematics, palaeoecology and palaeo-biogeography of the enigmatic allodaposuchian crocody-lians. PLoS ONE 9: e11583.

Brazeau MD. 2011. Problematic character coding methods in morphology and their effects. Biological Journal of the Linnean Society 104: 489-498.

Bremer K. 1994. Branch support and tree stability. Cla-disitics 10: 295-304.

Brinkmann W. 1989. Vorlaufige Mitteilung uber die Kroko-dilier-Faunen aus dem Ober-Jura (Kimmeridgium) der Kohlegrube Guimarota, bei Leiria (Portugal) und der Unter-Kreide (Barremium) von Una (Provinz Cuenca, Spanien). Documenta Naturae 56: 1-28.

Brinkmann W. 1992. Die krokodilier-fauna aus der UnterKreide (Ober-Barremium) von Una (Provinz Cuenca, Spanien). Berliner Geowissenschaftliche Abhandlungen 5:1-123.

Brochu CA. 1999. Phylogenetics, taxonomy, and historical biogeography of Alligatoroidea. Journal of Vertebrate Paleontology 19(Suppl. 2):1-92.

Brochu CA. 2004. A new Late Cretaceous gavialoid crocody-lian from eastern North America and the phylogenetic relationships of thoracosaurs. Journal of Vertebrate Paleontology 24: 610-633.

Brochu CA, Wagner JR, Jouve S, Sumrall CD, Dens-more LD. 2009. A correction corrected: consensus over the meaning of Crocodylia and why it matters. Systematic Biology 58: 537-543.

Brochu CA, Parris DC, Grandstaff BS, Denton RK Jr, Gallagher WB. 2012. A new species of Borealosuchus (Crocodyliformes, Eusuchia) from the Late Cretaceous-Early Paleogene of New Jersey. Journal of Vertebrate Paleontology 32: 105-116.

Broili F. 1931. Die gattung Alligatorium im oberen Jura von Franken. In: Sitzungsberichte der Bayer Akadamie der Wissenschaften. Munich: Mathematika und Naturwissenschaften, Abteilung, 63-74.

Bronzati M, Montefeltro FC, Langer MC. 2012. A species-level supertree of Crocodyliformes. Historical Biology 24: 598-606.

Buckley GA, Brochu CA. 1999. An enigmatic new crocodile from the Upper Cretaceous of Madagascar. Special Papers in Palaeontology 60: 149-175.

Buckley GA, Brochu CA, Krause DW, Pol D. 2000. A

pug-nosed crocodyliform from the Late Cretaceous of Madagascar. Nature 405: 941-944.

Buffetaut E. 1975. Sur l'anatomie et la position systématique de Bernissartia fagesii Dollo, L., 1883, crocodilien du Weal-dien de Bernissart, Belgique. Bulletin de l'Institut Royal des Sciences Naturelles de Belgique (Sciences Terre) 51: 1-20.

Buffetaut E. 1976. Osteologie et affinites de Trema-tochampsa taqueti (Crocodylia, Mesosuchia) du Séenonien

inférieur d'In Beceten (République du Niger). Geobios 9: 143-198.

Buffetaut E. 1981. Die biogeographische Geschichte der Krokodilier, mit Beschreibung einer neuen Art, Araripesu-chus wegeneri. Geologische Rundschau 70: 611-624.

Buffetaut E. 1982. Radiation evolutive, paleoecology et biogéographie des crocodiliens mesosuchians. Memoires de la Societe Geologique de France 142: 1-88.

Buffetaut E. 1983. The crocodilian Theriosuchus (Owen, 1879), in the Wealden of England. Bulletin of the British Museum of Natural History 37: 93-97.

Buscalioni AD. 1986. Cocodrilos fossiles del registro espanol. Anéilisis filogenético y Sistematico de la Familia Atoposauridae (Reptilia, Crocodylia), Unpublished PhD Thesis, Universidad Auténoma de Madrid, 363 pp.

Buscalioni AD, Sanz JL. 1984. Los Arcosaurios (Reptilia) del Jurásico Superior - Cretácico Inferior de Galve (Teruel, Espana). Separata de la Revista Teruel 71: 9-28.

Buscalioni AD, Sanz JL. 1987a. Cocodrilos del Cretacico Inferior de Galve (Teruel, Espana). Estudios Geologicos 43 (extra):23-43.

Buscalioni AD, Sanz JL. 1987b. Lista faunistica de los Vertebrados do Galve (Teruel). Estudios Geologicos 43(ex-tra):65-67.

Buscalioni AD, Sanz JL. 1988. Phylogenetic relationships of the Atoposauridae (Archosauria, Crocodylomorpha). Historical Biology 1: 233-250.

Buscalioni AD, Sanz JL. 1990a. Montsecosuchus depereti (Crocodylomorpha, Atoposauridae), new denomination for Alligatorium depereti Vidal, 195 (Early Cretaceous, Spain): redescription and phylogenetic relationships. Journal of Vertebrae Paleontology 10: 244-254.

Buscalioni AD, Sanz JL. 1990b. La familia Atoposauridae: una approximacioén a la historia de los cocodrilos enanos. Treballs del Museu de Geología de Barcelona 1: 77-89.

Buscalioni AD, Buffetaut E, Sanz JL. 1984. An immature specimen of the crocodilian from the Lower Cretaceous of Galve (Province of Teruel, Spain). Palaeontology 27: 809-813.

Buscalioni AD, Ortega F, Weishampel DBAND, Jianu CM. 2001. A revision of the crocodyliform Allodaposuchus precedens from the Upper Cretaceous of the Hateg Basin, Romania. Its relevance in the phylogeny of Eusuchia. Journal of Vertebrate Palaeontology 21: 74-86.

Buscalioni AD, Fregenal MA, Bravo AS, Poyato-Ariza FJ, Sanchiz B, Baez AM, Cambra Moo O, Martin Clo-sas C, Evans SE, Marugan Lobon J. 2008. The vertebrate assemblage of Buenache de la Sierra (Upper Barremian of Serrania de Cuenca, Spain), with insights into its taphonomy and palaeoecology. Cretaceous Research 29: 687-710.

Buscalioni AD, Piras P, Vullo R, Signore M, Barbera C. 2011. Early Eusuchia Crocodylomorpha from the vertebrate-rich Plattenkalk of Pietraroia (Lower Albian, southern Appenines, Italy). Zoological Journal of the Linnean Society 163: S199-S227.

Campos DA, Oliveira GR, Figueiredo RG, Riff D, Aze-vedo SA, Carvalho LB, Kellner AW. 2011. On a new peirosaurid crocodyliform from the Upper Cretaceous,

Bauru Group, southeastern Brazil. Anais da Academia Bra-sileira de Ciéncias 83: 317-327.

Canudo JI, Gasca JM, Aurell M, Badiola A, Blain A-H, Cruzado-Caballero P, Gomez- Fernandez D, Moreno-Azanza M, Parrilla J, Rabal-Garces R, Ruiz-Omeûaca JI. 2010. La Cantalera: an exceptional window onto the vertebrate biodiversity of the Hauterivian-Barremian transition in the Iberian Peninsula. Journal of Iberian Geology 36: 205-224.

Carvalho IS, Teixeira VPA, Ferraz MLF, Ribeiro LCB, Martinelli AG, Neto FM, Sertich JJW, Cunha GC, Cunha IC, Ferraz PF. 2011. Campinasuchus dinizi gen. et sp. nov., a new Late Cretaceous baurusuchid (Crocodyli-formes) from the Bauru Basin, Brazil. Zootaxa 2871: 19-42.

Cifelli RL, Nydam RL, Gardner JD, Weil A, Eaton JG, Kirkland JI, Madsen SK. 1999. Medial Cretaceous vertebrates from the Cedar Mountain Formation Emery County: the Mussentuchit local fauna. In: Gillette DD, ed. Vertebrate Palaeontology in Utah. Utah: Utah Geological Survey, 377-388.

Clark JM. 1986. Phylogenetic relationships of the crocodylo-morph archosaurs. Unpublished PhD Thesis, Department of Anatomy, University of Chicago, 551 pp.

Clark JM. 1994. Patterns of evolution in Mesozoic Crocodyli-formes. In: Fraser NC, Sues HD, eds. In the Shadow of Dinosaurs. Cambridge: Cambridge University, 84-97.

Clark JM. 2011. A new shartegosuchid crocodyliform from the Upper Jurassic Morrison Formation of western Colorado. Zoological Journal of the Linnean Society 163: S152-S172.

Clark JM, Norell MA. 1992. The Early Cretaceous crocody-lomorph Hylaeochampsa vectiana from the Wealden of the Isle of Wight. American Museum Novitates 3032: 1-19.

Clark JM, Sues H-D. 2002. Two new basal crocodylomorph archosaurs from the Lower Jurassic and the monophyly of the Sphenosuchia. Zoological Journal of the Linnean Society 136:77-95.

Colbert EH, Mook CC. 1951. The ancestral crocodilian Pro-tosuchus. Bulletin of the American Museum of Natural History 97: 143-182.

Cuny G, Buffetaut E, Cappetta H, Martin M, Mazin JM, Rose JM. 1991. Nouveaux restes de Vertébrés du Jurassique terminal du Boulonnais (Nord de la France). Neues Jahrbuch fur Geologie und Paläontologie, Abhandlungen 180: 323-347.

Cuny G, Laojumpon C, Cheychiw O, Lauprasert K. 2010. Fossil vertebrate remains from the Kut Island (Gulf of Thailand, Early Cretaceous). Cretaceous Research 31: 415-423.

Delfino M, Martin JE, Buffetaut E. 2008a. A new species of Acynodon (Crocodylia) from the Upper Cretaceous (San-tonian-Campanian) of Villagio del Pescatore, Italy. Palaeontology 51: 1091-1106.

Delfino M, Codrea V, Folie A, DICA P, Godefroit P, Smith T. 2008b. A complete skull of Allodaposuchus prece-dens Nopsca, 1928 (Eusuchia) and a reassessment of the morphology of the taxon based on the Romanian remains. Journal of Vertebrate Paleontology 28:111-122.

Eaton JG, Cifelli RL, Hutchison JH, Kirkland JI, Par-rish JM. 1999. Cretaceous vertebrate faunas from the Kaiparowits Plateau, South-Central Utah. In: Gillette DE, ed. Vertebrate paleontology in Utah. Salt Lake City, UT: Utah Geological Survey, Miscellaneous Publication, 345-353.

Efimov MB. 1976. The oldest crocodile on the territory of the USSR. Paleontologicheskii Zhurnal 10: 115-117.

Efimov MB. 1988. Fossil crocodiles and champsosaurs of Mongolia and the USSR. Trudy Sovmestuoi Sovetsko-Mon-gol'skoi Paleontologicheskoi Ekspeditsii 36: 1-108.

Efimov MB. 1996. The Jurassic crocodylomorphs of inner Asia. In: Morales M, ed. The Continental Jurassic. Flagstaff: Museum of Northern Arizona, 305-310.

Evans SE, Milner AR. 1994. Microvertebrate faunas from the middle Jurassic of Britain. In: Fraser NC, Sues H-D, eds. The Shadow of the Dinosaurs: Early Mesozoic Tetra-pods. Cambridge: Cambridge University Press, 303-321.

Figueiredo RG, Moreiera JKR, Saraiva AAF, Kellner AWA. 2011. Description of a new specimen of Susisuchus anatoceps (Crocodylomorpha: Mesoeucrocodylia) from the Crato Formation (Santana Group) with comments on Neo-suchia. Zoological Journal of the Linnean Society 163: S273-S288.

Fiorillo AR. 1999. Non-mammalian microvertebrate remains from the Robison Eggshell Site, Cedar Mountain Formation (Lower Cretaceous), Emery County, Utah. In: Gilette DD, ed. Vertebrate paleontology in Utah. Salt Lake City, UT: Utah Geological Survey, Miscellaneous Publications, 99: 259-268.

Flynn JJ, Fox SR, Parrish JM, Ranivoharimanana L, Wyss AR. 2006. Assessing diversity and palaeoecology of a Middle Jurassic microvertebrate assemblage from Madagascar. In: Harris DJ, Lucas SG, Spielmann JA, Lockley MG, Milner ARC, Kirkland JI, eds. The Triassic-Jurassic terrestrial transition. Albuquerque: New Mexico Museum of Natural History and Science Bulletin 37: 476-489.

Fortier DC, Schultz CL. 2009. A new neosuchian crocodylo-morph (Crocodyliformes, Mesoeucrocodylia) from the Early Cretaceous of north-east Brazil. Palaeontology 52: 991-1007.

Galton PM. 1996. Notes on Dinosauria from the Upper Cretaceous of Portugal. Neueus Jahrbuch fur Geologie und Palaontology, Mitheilung 2: 83-90.

Garrison JR Jr, Brinkman DB, Nichols DJ, Layer P, Burge DL, Thayn D. 2007. A multidisciplinary study of the Lower Cretaceous Cedar Mountain Formation, Mussen-tuchit Wash, Utah: a determination of the paleoenviron-ment and paleoecology of the Eolambia caroljonesa dinosaur quarry. Cretaceous Research 28:461-494.

Gasparini Z, Pol D, Spalletti LA. 2006. An unusual marine crocodyliform from the Jurassic- Cretaceous boundary of Patagonia. Science 311: 70-73.

Gervais P. 1871. Remarques au sujet des Reptiles provenant des calcaires lithographiques de Cerin, dans le Bugey, qui sont conservés au Musée de Lyon. Comptes Rendus des seances de l'Academie de Sciences 73: 603-607.

Gilmore CW. 1926. A new aetosaurian reptile from the Morrison Formation of Utah. Annals of Carnegie Museum 16: 325-348.

Godoy PL, Montefeltro FC, Norell MA, Langer MC.

2014. An additional baurusuchid from the Cretaceous of Brazil with evidence of interspecific predation among Crocodyliformes. PLoS ONE 9: e97138.

Goloboff PA. 1993. Estimating character weights during tree search. Cladistics 9: 83-91.

Goloboff PA, Farris S, Nixon K. 2000. TNT (Tree analysis using New Technology) (BETA) ver. 1.1 Published by the authors, Tucumaén, Argentina.

Goloboff PA, Farris JS, Kallersjo M, Oxelman B, Ramirez MJ, Szumik CA. 2003. Improvements to resampling measures of group support. Cladistics 19: 324-332.

Gomani EM. 1997. A crocodyliform from the Early Cretaceous dinosaur beds, northern Malawi. Journal of Vertebrate Paleontology 17: 280-294.

Gow CE. 2000. The skull of Protosuchus haughtoni, an Early Jurassic crocodyliform from Southern Africa. Journal of Vertebrate Paleontology 10: 49-56.

Haddoumi H, Allain R, Meslouh S, Metais G, Monbaron M, Pons D, Rage J-C, Vullo R, Zouhri S, Gheerbrant E. 2016. Guelb el Ahmar (Bathonian, Anoual Syncline, eastern Morocco): first continental flora and fauna including mammals from the Middle Jurassic of Africa. Gondwana Research 29: 290-319.

Hallam A. 1986. The Pliensbachian and Tithonian extinction events. Nature 319: 765-768.

Halliday TJD, De Andrade MB, Benton MJ, Efimov MB.

2015. A re-evaluation of goniopholidid crocodylomorph material from Central Asia: biogeographic and phylogenetic implications. Acta Palaeontological Polonica 60: 291-312.

Hay OP. 1930. Second Bibliography and Catalogue of the Fossil Vertebrata of North America, 1074. Washington, DC: Carnegie Institute Washington.

Holliday CM, Gardner NM. 2012. A new eusuchian with novel cranial integument and its significance for the origin and evolution of Crocodylia. PLoS ONE 7: e30471.

Hornung JJ. 2013. Contributions to the paleobiology of the archosaurs (Reptilia: Diapsida) from the Buckerberg Formation ( Northwest German Wealden' - Berriasian-Valangi-nian, Lower Cretaceous) of northern Germany. Unpublished PhD Thesis, Universitat Göttingen, 405 pp.

Joffe J. 1967. The dwarf crocodiles of the Purbeck Formation, Dorset: a reappraisal. Palaeontology 10: 629-639.

Jouve S. 2004. Etude des Crocodyliformes fini Créetaceé-Paléogàne du Bassin des Oulad Abdoun (Maroc) et comparaison avec les faunes africaines contemporaines: systéematique, phylogéenie et paleéobiogéeographie. Paris, Muséeum National d'Histoire Naturelle, Unpublished DPhil Thesis.

Jouve S. 2009. The skull of Teleosaurus camodensis (Crocodylomorpha; Thalattosuchia), and phylogenetic analysis of Thalattosuchia. Journal of Vertebrate Paleontology 29: 88-102.

Jouve S, Bouya B, Amaghzaz M. 2005a. A short-snouted dyrosaurid (Crocodyliformes, Mesoeucrocodylia) from the Palaeocene of Morocco/. Palaeontology 48: 359-369.

Jouve S, Larochene M, Bouya B, Amghzaz M. 2005b. A

new dyrosaurid crocodyliform from the Palaeocene of Mor-

occo and a phylogenetic analysis of Dyrosauridae. Acta Palaeontologica Polonica 50: 581-594.

Jouve S, Larochène Bouya B, Amaghaz M. 2006. A new

species of Dyrosaurus (Crocodylomorpha, Dyrosauridae) from the early Eocene of Morocco: phylogenetic implications. Zoological Journal of the Linnean Society 148: 603656.

Jouve S, Bardet N, Jalil N-E, Suberbiola XP, Bouya B, Amaghzaz M. 2008. The oldest African crocodylian: phylogeny, paleobiogeography, and differential survivorship of marine reptiles through the Cretaceous-Tertiary boundary. Journal of Vertebrate Paleontology 28: 409-421.

Kalin JA. 1955. Crocodilia. In: Piveteau J, ed. Traite de palaeontologie. Paris: Masson et Cie, Vol. 5, 695-784.

Karl H-V, Groning E, Brauckmann C, Schwarz D, Knotschke N. 2006. The Late Jurassic crocodiles of the Langengerg near Oker, Lower Saxony (Germany), and description of related materials (with remarks on the history of quarrying the "Langenberg" Limestone and "Obernkirchen Sandstone"). Clausthaler Geowissenschaften 5: 5977.

Kellner AWA, Figueiredo RG, Azevedo SAK, Campos

DA. 2011. A new Cretaceous notosuchian (Mesoeu-crocodylia) with bizarre dentition from Brazil. Zoological Journal of the Linnean Society 163: S109-S115.

Kirkland JI, Carpenter K, Hirsch K, Horner J. 1994.

Predation of dinosaur nests by terrestrial crocodilians. In: Carpenter K, Hirsch KF, Horner JR, eds. Dinosaur eggs and babies. Cambridge: Cambridge University Press, 124133.

Knoll F, Lopez-Antonanzas R. 2014. The vertebrate fauna and "stipites" layers of the Grand Causses (Middle Jurassic, France). Frontiers in Ecology and Evolution 2: 1-6.

Knoll F, Cuny G, Mojon P-O, Lopez-Antonanzas R, Huguet D. 2013. A new vertebrate-, ostracod-, and charo-phyte-bearing locality in the Middle Jurassic of the Grands Causses (southern France). Proceedings of the Geologists' Association 124: 525-529.

Konzhukova ED. 1954. New fossil crocodilians from Mongolia. Trudy Paleontologicheskogo Instituta ANSSSR 48: 171194.

Kriwet J, Rauhut OWM, Gloy U. 1997. Microvertebrate remains (Pisces, Archosauria) from the Middle Jurassic (Bathonian) of France. Neues Jahrbuch für Geologie und Paläontologie 206: 1-28.

Kuhn O. 1960. Die familien der fossilen amphibien und reptilian. Berichte der Naturforschenden Gesellschaft Bamberg 37: 20-52.

Kuhn O. 1961. Die Tier und pflanzenwelt des Solnhofener Schiefers. Geologica Bavarica 48:68 pp.

Kuhn O. 1966 Die Tierwelt des Solnhofener Schiefers. - 2. Aufl., Neue Brehmbucherei, 40 S., 144 Abb., (Ziemsen) Wittenberg.

Langston W. 1974. Nonmammalian Comanchean tetrapods. Geoscience and Man 8: 77-102.

Larsson HCE, Gado B. 2000. A new Early Cretaceous crocodyliform from Niger. Neues Jahrbuch fur Geologie und Palaontologie, Abhandlungen 217: 131-141.

Larsson HCE, Sues H-D. 2007. Cranial osteology and phy-logenetic relationships of Hamadasuchus rebouli (Crocodyli-formes: Mesoeucrocodylia) from the Cretaceous of Morocco. Zoological Journal of the Linnean Society 149: 533-567.

Lauprasert K, Cuny G, Thirakupt K, Suteethorn V. 2009. Khoratosuchus jintasakuli gen. et sp. nov., an advanced neosuchian crocodyliform from the Early Cretaceous (Aptian-Albian) of NE Thailand. In: Buffetaut E, Cuny G, Le Loeuff J, Suteethorn V, eds. Late palaeozoic and mesozoic ecosystems in SE ssia. London: The Geological Society, London, Special Publications, 315: 175-187.

Lauprasert K, Laojumpon C, Saenphala W, Cuny G, Thirakhupt K, Suteethorn V. 2011. Atoposaurid crocodyliforms from the Khorat Group of Thailand: first record of Theriosuchus from Southeast Asia. Palaeontologis-che Zeitschrift 85: 37-47.

Laurenti JN. 1768. Specimen medicum, exhibens synopsin reptilium emendatam cum experimentis circa venena et antidota reptilium austracorum, quod authoritate et con-sensu. Vienna, Joan. Thomae, 217 pp.

Leardi JM, Pol D, Fernandez MS, Idean C. 2012. The antorbital fenestra of Metriorhynchidae (Crocodyliformes, Thalattosuchia): testing its homology within a phylogenetic framework. Journal of Vertebrate Paleontology 32: 490-494.

Lewis PO. 2001. A likelihood approach to estimating phy-logeny from discrete morphological character data. Systematic Biology 50: 913-925.

Mannion PD, Benson RBJ, Carrano MT, Tennant JP, Judd J, Butler RJ. 2015. Climate constrains the history and biodiversity of crocodylians. Nature Communications 6:8438.

Marinho TS, Iori FV, Carvalho IS, Vascobcellos FM. 2013. Gondwanasuchus scabrosus gen. et sp. nov., a new terrestrial predatory crocodyliform (Mesoeucrocodylia: Bau-rusuchidae) from the Late Cretaceous Baurus Basin of Brazil. Cretaceous Research 44: 104-111.

Marmi J, Blanco A, Fondevilla V, Vecchia FMD, Selleés AG, Vicente A, Martin-Closas C, Oms O, Galobart A. 2016. The Moli del Baro-1 site, a diverse fossil assemblage from the uppermost Maastrichtian of the southern Pyrenees (north-eastern Iberia). Cretaceous Research 57: 519539.

Martin JE. 2007. New material of the Late Cretaceous glo-bidontan Acynodon iberoccitanus (Crocodylia) from Southern France. Journal of Vertebrate Paleontology 27: 367-372.

Martin JE, Rabi M, Csiki Z. 2010. Survival of Therio-suchus (Mesoeucrocodylia: Atoposauridae) in a Late Cretaceous archipelago: a new species from the Maastrichtian of Romania. Naturwissenschaften 97: 845-854.

Martin JE, Rabi M, Csiki-Sava Z, Vasile S. 2014a. Cranial morphology of Theriosuchus sympiestodon (Mesoeu-crocodylia, Atoposauridae) and the widespread occurrence of Theriosuchus in the Late Cretaceous of Europe. Journal of Palaeontology 88: 444-456.

Martin JE, Lauprasert K, Buffetaut E, Liard R, Suteethorn V. 2014b. A large pholidosaurid in the Phu Kradung Formation of north-eastern Thailand. Palaeontology 57: 757-769.

Martinelli AG. 2003. New cranial remains of the bizarre notosuchid Comahuesuchus brachybuccalis (Archosauria,

Crocodyliformes) from the late Cretaceous of Rio Negro Province (Argentina). Ameghiniana 40: 559-572.

von Meyer H. 1850. Mittheilungen an Professor Bronn gerichtet. Neues Jahrbuch für Mineralogie, Geognosie, Geologie und Petrefaktenkunde 195-204.

von Meyer H. 1851. Description de l'Atoposaurus jourdani et du Sapheosaurus thiollierei, reptiles fossiles des calcaires lithographiques du Bugey. Annales des Sciences Physiques et Naturelles, d'Agriculture et d'Industrie 2: 113-117.

Michard J-G, De Broin F, Brunet M, Hell J. 1990. Le plus ancient crocodilien néosuchien spécialisé a caracteres eusuchiens du continent Africain (Cretace inerieur, Cameroun). Comptes Rendus de l'Académie des Sciences Paris 311: 365-371.

Miller KG, Kominz MA, Browning JV, Wright JD, Mountain GS, Katz ME, Sugarman PJ, Cramer BS, Christie-Blick N, Pekar SF. 2005. The Phanerozoic record of global sea-level change. Science 310: 1293-1298.

Mo J, Buffetaut E, Tong H, Amiot R, Cabin L, Cuny G, Suteethorn V, Suteethon S, Jiang S. 2016. Early Cretaceous vertebrates from the Xinlong Formation of Guangxi (southern China): a review. Geological Magazine 153: 143159.

Montefeltro FC, Larsson HCE, Langer MC. 2011. A new

baurusuchid (Crocodyliformes, Mesoeucrocodylia) from the Late Cretaceous of Brazil and the phylogeny of Baurusuchi-dae. PLoS ONE 6: e21916.

Montefeltro FC, Larsson HCE, De Franca MAG, Langer MC. 2013. A new neosuchian with Asian affinities from the Jurassic of northeastern Brazil. Naturwissenschaften 100: 835-841.

Mook CC. 1942. Skull characters of Amphicotylus lucasii. American Museum Novitates 1165: 1-8.

Müller J, Reisz RR. 2006. The phylogeny of early eurep-tiles: comparing parsimony and Bayesian approaches in the investigation of a basal fossil clade. Systematic Biology 55: 503-511.

Narvaez I, Brochu CA, Escaso F, Pérez-García A, Ortega F. 2015. New crocodyliforms from Southwestern Europe and definition of a diverse clade of European Late Cretaceous basal eusuchians. PLoS One 10: e0140679.

Nascimento PM, Zaher H. 2011. The skull of the Upper Cretaceous baurusuchid crocodile Baurusuchus albertoi Nascimento & Zaher, 2010, and its phylogenetic affinities. Zoological Journal of the Linnean Society 163: S116-S131.

Norell MA. 1988. Cladistic approaches to paleobiology as applied to the phylogeny of alligatorids. New Haven, Connecticut, Yale University, Unpublished PhD Thesis.

Norell MA, Clark JM. 1990. A reanalysis of Bernissartia fagesii, with comments on its phylogenetic position and its bearing on the origin and diagnosis of the Eusuchia, Bulletin - Institut Royal des Sciences Naturelles de Belgique. Sciences de la Terre 60: 115-128.

Novas FE, Pais DF, Pol D, Carvalho IS, Scanferla A, Mones A, Riglos S. 2009. Bizarre notosuchian crocodyliform with associated eggs from the Upper Cretaceous of Bolivia. Journal of Vertebrate Paleontology 29: 1316-1320.

O'Connor PM, Sertich JJW, Stevens NJ, Roberts EM, Gottfried MD, Heironymous TL, Jinnah ZA, Ridgely R Ngasala SE, Temba J. 2010. The evolution of mammal-like crocodyliforms in the Cretaceous period of Gond-wana. Nature 466: 748-751.

Oreska MPJ, Carrano MT, Dzikiewicz KM. 2013. Vertebrate paleontology of the Cloverly Formation (Lower Cretaceous), I: faunal composition, biogeographic relationships, and sampling. Journal of Vertebrate Paleontology 33: 264292.

Ortega F, Buscalioni AD, Gasparini Z. 1996. Reinterpretation and new denomination of Atacisaurus crassiporatus (Middle Eocene; Issel, France) as cf. Iberosuchus (Crocody-lomorpha, Metasuchia). Geobios 29: 353-364.

Ortega F, Gasparini Z, Buscalioni AD, Calvo JO. 2000. A new species of Araripesuchus (Crocodylomorpha, Mesoeucrocodylia) from the Lower Cretaceous of Patagonia, (Argentina). Journal ofVertebrate Paleontology 20: 57-76.

Osi A. 2008. Cranial osteology of Iharkutosuchus makadii, a Late Cretaceous basal eusuchian crocodyliform from Hungary. Neues Jahrbuch fur Geologie und Paläontologie 248: 279-299.

O si A. 2014. The evolution of jaw mechanism and dental function in heterodont crocodyliforms. Historical Biology 26: 279-414.

Osi A, Clark JM, Weishampel DB. 2007. First report on a new basal eusuchian crocodyliform with multicusped teeth from the Upper Cretaceous (Santonian) of Hungary. Neues Jahrbuch fur Geologie und Palaontologie 243: 169-177.

O si A, Rabi M, Makadi L, Szentesi Z, Botfalvai G, Gulyas P. 2012. The Late Cretaceous continental vertebrate fauna from Iharkut (Western Hungary, Central Europe): a review. In: Godefroit P, ed. Tribute to Charles Darwin and the Bernissart Iguanodons: new Perspectives of Vertebrate Evolution and Early Cretaceous ecosystems. Bloomington: Indiana University Press, 533-570.

Osmolska H, Hua S, Buffetaut E. 1997. Gobiosuchus kiela-nae (Protosuchia) from the Late Cretaceous of Mongolia: anatomy and relationships. Acta Palaeontologica Polonica 42: 257-289.

Owen R. 1878a. On the fossils called 'granicones': being a contribution to the histology of the exo- skeleton in 'Reptilia'. Journal of the Royal Microscopical Society 1: 222-236.

Owen R. 1878b. Monograph on the fossil Reptilia of the Wealden and Purbeck formations. Supplement No. VIII. Crocodilia (Goniopholis, Petrosuchus and Suchosaurus). Monograph of the Palaeontolographical Society 32: 1-15.

Owen R. 1879. Monograph on the fossil Reptilia of the Weal-den and Purbeck formations. Supplement No. IX. Crocodilia (Goniopholis, Brachydectes, Nannosuchus, Theriosuchus and Nuthetes). Monograph of the Palaeontographical Society 33: 1-15.

Peng G-Z, Shu C-K. 2005. A new species of Hisosuchus from the Late Jurassic of Zigong, Sichuan, China. Vertebrata palasiatica 10: 312-324.

Pol D. 1999. El esqueleto postcraneano de Notosuchus ter-restris (Archosauria: Crocodyliformes) del Cretacico Superior de la Cuenca Neuquina y su informacién fi logenética.

Buenos Aires, Universidad de Buenos Aires, Unpublished PhD Thesis.

Pol D. 2005. Postcranial remains of Notosuchus terrestris (Archosauria; Crocodyliformes) from the Upper Cretaceous of Argentina. Ameghiniana 42: 21-38.

Pol D, Apesteguia S. 2005. New Araripesuchus remains from the early Late Cretaceous (Cenomanian-Turonian) of Patagonia. American Museum Novitates 3490: 1-38.

Pol D, Escapa IH. 2009. Unstable taxa in cladistic analysis: identification and the assessment of relevant characters. Cladistics 25: 515-527.

Pol D, Gasparini Z. 2009. Skull anatomy of Dakosaurus andiniensis (Thalattosuchia: Crocodylomorpha) and the phylogenetic position of Thalattosuchia. Journal of Systematic Palaeontology 7: 163-197.

Pol D, Norell MA. 2004a. A new crocodyliform from Zos Canyon, Mongolia. American Museum Novitates 3445:1-36.

Pol D, Norell MA. 2004b. A new gobiosuchid crocodyliform taxon from the Cretaceous of Mongolia. American Museum Novitates 3458:1-31.

Pol D, Powell JE. 2011. A new sebecid mesoeucrocodylian from the Rio Loro Formation (Palaeocene) of north-western Argentina. Zoological Journal of the Linnean Society 163: S7-S36.

Pol D, Ji S-A, Clark JM, Chiappe LM. 2004. Basal crocodyliforms from the Lower Cretaceous Tuguli Group (Xinjiang, China) and the phylogenetic position of Edento-suchus. Cretaceous Research 25: 603-622.

Pol D, Turner AH, Norell MA. 2009. Morphology of the Late Cretaceous crocodylomorph Shamosuchus djadochtaensis and a discussion of neosuchian phylogeny as related to the origin of Neosuchia. Bulletin of the American Museum of Natural History 103: 1-103.

Pol D, Leardi JM, Lecuona A, Krause DW. 2012. Postcranial anatomy of Sebecus icaeorhinus (Crocodyliformes, Sebecidae) from the Eocene of Patagonia. Journal of Vertebrate Paleontology 32: 328-354.

Pol D, Nascimento PM, Carvalho AB, Riccomini C, Pires-Domingues RA, Zaher H. 2014. A new notosuchian from the Late Cretaceous of Brazil and the phylogeny of advanced notosuchians. PLoS One 9: e93105.

Pomes ML. 1990. Morphotype of Lower Cretaceous crocodilian teeth (Archosauria) from the Cedar Mountain Formation of Utah and the Antlers Formation of Texas. Journal of Vertebrate Paleontology 10: 38.

Pouech J, Mazin J-M, Billon-Bruyat J-P. 2006. Microver-tebrate biodiversity from Cherves-de- Cognac (Lower Cretaceous, Berriasian: Charente, France). Mesozoic Terrestrial Ecosystems 96-100.

Pouech J, Amiot R, Lécuyer C, Mazin J-M, Martineau F, Fourel F. 2014. Oxygen isotope composition of vertebrate phosphates from Cherves-de-Cognac (Berriasian, France): environmental and ecological significance. Palaeo-geography, Palaeoclimatology, Palaeoecology 410: 290-299.

Prasad GVR, De Lapparent De Broin F. 2002. Late Cretaceous crocodile remains from Naskal (India): comparisons and biogeographic affinities. Annales de Pale'ontologie 22:19-71.

Pritchard AC, Turner AH, Allen ER, Norell MA. 2012. Osteology of a North American goniopholidid (Eutretaura-

nosuchus delfsi) and palate evolution in Neosuchia. American Museum Novitates 3783: 1-56.

Pritchard AC, Turner AH, Allen ER, Norell MA. 2013.

Osteology of a North American goniopholidid (Eutretauranosuchus delfsi) and palate evolution in Neosuchia. American Museum Novitates 3783: 1-56.

Rayfield EJ, Milner AC. 2008. Establishing a framework for archosaur cranial mechanics. Paleobiology 34: 494-515.

Riff D, Kellner AWA. 2001. On the dentition of Bau-rusuchus pachecoi Price (Crocodyliformes, Metasuchia) from the Upper Cretaceous of Brazil. Boletim do Museu Nacional 59: 1-15.

Riff D, Kellner AWA. 2011. Baurusuchid crocodyliforms as theropod mimics: clues from the skull and appendicular morphology if Stratiosuchus maxhechti (Upper Cretaceous of Brazil. Zoological Journal of the Linnean Society 163: S37-S56.

Rogers JV. 2003. Pachycheilosuchus trinquei, a new pro-coelous crocodyliform from the Lower Cretaceous (Albian) Glen Rose Formation of Texas. Journal of Vertebrate Paleontology 23: 128-145.

Romer AS. 1956. Osteology of the Reptiles. Chicago: Chicago University Press, 772 pp.

Ronquist F, Huelsenbeck JP. 2003. MRBAYES 3: Baye-sian phylogenetic inference under mixed models. Bioinfor-matics 19: 1572-1574.

Ruiz-Omenaca JI, Canudo JI, Aureli M, Badenas B, Barco JL, Cuenca-Bescos G, Ipas J. 2004. Estado de las investigaciones sobre los vertebrados del Jurasico Superior y Cretacicio Inferior de Galve (Teruel). Estudios Geologicos 60: 179-202.

Salisbury SW. 2002. Crocodilians from the Lower Cretaceous (Berriasian) Purbeck Limestone Group of Dorset, southern England. Special Papers in Palaeontology 68: 121-144.

Salisbury SW, Frey E. 2001. A biomechanical transformation model for the evolution of semi- spheroidal articulations between adjoining vertebral bodies in crocodilians. In: Grigg GG, Serbacher F, Franklin CE, eds. Crocodilian Biology and Evolution. Chipping Norton: Surrey Beatty and Sons, 85-134.

Salisbury SW, Naish D. 2011. Crocodilians. In: Batten DJ, ed. English Wealden Fossils. London: The Palaeontological Association, 305-369.

Salisbury SW, Willis PMA, Peitz S, Sander PM. 1999. The crocodilian Goniopholis simus from the Early Cretaceous of north-western Germany. Special Papers in Palaeontology 60: 121-148.

Salisbury SW, Molnar RE, Frey E, Willis PMA. 2006. The origin of modern crocodyliforms: new evidence from the Cretaceous of Australia. Proceedings of the Royal Society, Series B: Biological Sciences 273: 2439-2448.

Sanz JL, Ortega F, Shibata M. 2014. Dinosaurios maravillosos de espana. Cuenca: Diputación Provincial de Cuenca, Spain.

Schneider CA, Rasband WS, Eliceiri KW. 2012. NIH

Image to ImageJ: 25 years of image analysis. Nature Methods 9: 671-675.

Schwarz D, Salisbury SW. 2005. A new species of Therio-suchus (Atoposauridae, Crocodylomorpha) from the Late

Jurassic (Kimmeridgian) of Guimarota, Portugal. Geobios 38: 779-802.

Schwarz-Wings D, Frey E, Martin T. 2009a. Reconstruction of the bracing system of the trunk and tail in hypo-saurine dyrosaurids (Crocodylomorpha; Mesoeucrocodylia). Journal of Vertebrate Paleontology 29:453-472.

Schwarz-Wings D, Rees J, Lindgren J. 2009b. Lower Cretaceous mesoeucrocodylians from Scandinavia. Cretaceous Research 30: 1245-1355.

Schwarz-Wings D, Klein N, Neumann C, Resch U. 2011. A new partial skeleton of Alligatorellus (Crocodyliformes) associated with echinoids from the Late Jurassic (Titho-nian) lithographic limestones of Kelheim, S-Germany. Fossil Record 14: 195-205.

Sereno PC, Larsson HCE. 2009. Cretaceous crocodyliforms from the Sahara. ZooKeys 28: 1-143.

Sereno PC, Larsson HCE, Sidor CA, Gado B. 2001. The giant crocodyliform Sarcosuchus from the Cretaceous of Africa. Science 294: 1516-1519.

Sereno PC, Sidor CA, Larsson HCE, Gado B. 2003. A new notosuchian from the Early Cretaceous of Niger. Journal of Vertebrate Palaeontology 23: 477-482.

Sertich JJ, O'Connor PM. 2014. A new crocodyliform from the middle Cretaceous Galula formation, southwestern Tanzania. Journal of Vertebrate Paleontology 34: 576-596.

Smith AB, McGowan AJ. 2007. The shape of the Phanero-zoic palaeodiversity curve: how much can be predicted from the sedimentary rock record of western Europe. Palaeontology 50: 765-774.

Smith DK, Allen ER, Sanders RK, Stadtman KL. 2010. A new specimen of Eutretauranosuchus (Crocodyliformes; Goniopholididae) from Dry Mesa, Colorado. Journal of Vertebrate Paleontology 30: 1466-1477.

Steel R. 1973. Crocodylia. In: Kuhn O, ed. Handbuch der Palaoherpetologie, Part 16. Stuttgart: Gustav Fisher Verlag, 1-116.

Stevens NJ, Hill RV, Al-Wosabi M, Schulp A, As-Saruri M, Al-Nimey F, Jolley LA, Schulp-Stuip Y, O'Connor P. 2013. A middle Eocene mesoeucrocodylian (Crocodyli-formes) from the Kaninah Formation, Republic of Yemen. Geologos 19: 175-183.

Storrs GW, Efimov MB. 2000. Mesozoic crocodyliforms of north-central Eurasia. In: The Age of Dinosaurs in Russia and Mongolia, 402-419.

Strong EE, Lipscomb DL. 1999. Character coding and inapplicable data. Cladistics 15: 363-371.

Sweetman SC, Pedreira-Segade U, Vidovic SU. 2015. A new bernissartiid crocodyliform from the Lower Cretaceous Wessex Formation (Wealden Group, Barremian) of the Isle of Wight, southern England. Acta Palaeontological Polonica 60: 257-268.

Tennant JP, Mannion PD. 2014. Revision of the Late Jurassic crocodyliform Alligatorellus, and evidence for allo-patric speciation driving high diversity of western European atoposaurids. PeerJ 2: e599.

Tennant JP, Mannion PD, Upchurch P. 2016. Environmental drivers of crocodyliform extinction across the Jurassic/Cretaceous transition. Proceedings of the Royal Society

Series B: Biological Sciences 283: 20152840. doi: 10.1098/ rspb.2015.2840

Tennant JP, Mannion PD, Upchurch P, Sutton M, Price G. in press. Biotic and environmental dynamics through the Late Jurassic-Early Cretaceous transition: evidence for protracted faunal and ecological turnover. Biological Reviews. doi: 10.1111/brv.12255

Thies D, Broschinksi A. 2001. Teeth of a small duro-phagous crocodiles from the Late Jurassic (Kimmeridi-gian) of North Germany. Geologische Beiträge Hannover 2:65-70.

Thies D, Windolf R, Mudroch A. 1997. First record of Ato-posauridae (Crocodylia: Metamesosuchia) in the Upper Jurassic (Kimmeridgian) of northwest Germany. Neues Jahrbuch für Geologie und Paläontologie 205: 393-411.

Turner AH. 2006. Osteology and phylogeny of a new species of Araripesuchus (Crocodyliformes: Mesoeucrocodylia) from the Late Cretaceous of Madagascar. Historical Biology 18: 255-369.

Turner AH. 2015. A review of Shamosuchus and Paralliga-tor (Crocodyliformes, Neosuchia) from the Cretaceous of Asia. PLoS ONE 10: e0118116.

Turner AH, Buckley GA. 2008. Mahajangasuchus insignis (Crocodyliformes: Mesoeucrocodylia) cranial anatomy and new data on the range of the eusuchian-style palate. Journal of Vertebrate Palaeontology 28: 382-408.

Turner AH, Pritchard AC. 2015. The monophyly of Susisuchidae (Crocodyliformes) and its phylogenetic placement in Neosuchia. PeerJ 3: e759.

Turner AH, Sertich JJW. 2010. Phylogenetic history of Simosuchus clarki (Crocodyliformes: Notosuchia) from the Late Cretaceous of Madagascar. Journal of Vertebrate Paleontology Memoir 30: 177-236.

Tykoski RS, Rowe TB, Ketcham RA, Colbert MW. 2002. Calsoyasuchus valliceps, a new crocodyliform from the Early Jurassic Kayenta Formation of Arizona. Journal of Vertebrate Paleontology 22: 593-611.

Vidal LM. 1915. Nota geologica y paleontologica sobre e Jurasico Superior de la Provincia de Léerida. Boliten del Instetuto Geologico de Espana 16: 17-55.

Vullo R, Neéraudeau D. 2008. Cenomanian vertebrate assemblages from southwestern France: a new insight into the European mid-Cretaceous continental fauna. Cretaceous Research 29: 930-935.

Vullo R, Abit D, Ballèvre M, Billon-Bruyat J-P, Bour-geais R, Buffeatut E, Daviero-Gomez V, Garcia G, Gomez B, Mazin J-M, Morel S, Neéraudeau D, Pouech J, Rage J-C, Schnyder J. 2014. Palaeontology of the Purbeck-type (Tithonian, Late Jurassic) bonebeds of Chassiron (Oléeron Island, western France). Comptes Rendus Palevol 13: 421-441.

Walker AD. 1970. A revision of the Jurassic reptile Hallopus victor (Marsh) with remarks on the classification of the crocodiles. Philosophical Transactions of the Royal Society of London, Series B: Biological Sciences 257: 323-372.

Wellnhofer P. 1971. Die Atoposauridae (Crocodylia, Meso-suchia) der Oberjura-Plattenkalke Bayerns. Palaeonto-graphica Abteilung A 138: 133-165.

Wilkinson LE, Young MT, Benton MJ. 2008. A new metri-orhynchid crocodilian (Mesoeucrocodylia: Thalattosuchia) from the Kimmeridgian (Upper Jurassic) of Wiltshire, UK. Palaeontology 51: 1307-1333.

Winkler DA, Murry PA, Jacobs LL. 1990. Early Cretaceous (Comanchean) vertebrates of central Texas. Journal of Vertebrate Paleontology 10: 95-116.

Wu X-C, Brinkmann DB. 1993. A new crocodylomorph of "Mesosuchian" grade from the Upper Cretaceous Milk River Formation, Southern Alberta. Journal of Vertebrate Paleontology 13: 153-160.

Wu X-C, Sues H-D. 1996. Anatomy and phylogenetic relationships of Chimaerasuchus paradoxus, an unusual crocodyliform reptile from the Lower Cretaceous of Hubei, China. Journal of Vertebrate Paleontology 16: 688-702.

Wu X-C, Brinkmann DB, Lu J-C. 1994. A new species of Shantungosuchus from the Lower Cretaceous of Inner Mongolia (China), with comments on S.chuhsienensis Young, 1961 and the phylogenetic position of the genus. Journal of Vertebrate Paleontology 14: 210-229.

Wu X-C, Brinkmann DB, Russell AP. 1996. Sunosuchus junggarensis sp. nov. (Archosauria: Crocodyliformes) from the Upper Jurassic of Xinjiang, People's Republic of China. Canadian Journal of Earth Sciences 33: 606-630.

Wu X-C, Sues H-D, Dong Z-M. 1997. Sichuanosuchus shuhanensis, a new ?Early Cretaceous protosuchian (Archo-sauria: Crocodyliformes) from Sichuan (China) and the monophyly of Protosuchia. Journal of Vertebrate Paleontology 17: 89-103.

Wu X-C, Cheng Z-W, Russell AP. 2001a. Cranial anatomy of a new crocodyliform (Archosauria: Crocodylomorpha) from the Lower Cretaceous of Song-Liao Plain, northeastern China. Canadian Journal of Earth Sciences 38:1653-1663.

Wu X-C, Russell AP, Cumbaa SL. 2001b. Terminonaris (Archosauria: Crocodyliformes): new material from Saskatchewan, Canada, and comments on its phylogenetic relationships. Journal of Vertebrate Paleontology 21: 492-514.

Young CC. 1961. On a new crocodile from Chuhsien, E. Shantung. Vertebrata Palasiatica 1: 6-10.

Young MT, Steel L, Foffa D, Price T, Naish D, Ten-nant JP. 2014. Marine tethysuchian crocodyliform from the ?Aptian-Albian (Lower Cretaceous) of the Isle of Wight, UK. Biological Journal of the Linnean Society 113: 854-871.

Young MT, Tennant JP, Brusatte SL, Challands TJ, Fraser NC, Clark NDL, Ross DA. 2016. The first definitive Middle Jurassic atoposaurid (Crocodylomorpha, Neosuchia), and a discussion on the genus Theriosuchus. Zoological Journal of the Linnean Society 176: 443-462.

Zaher H, Pol D, Carvalho AB, Riccomini C, Campos D, Nava W. 2006. Redescription of the cranial morphology of Mariliasuchus amarali, and its phylogenetic affinities (Crocodyliformes, Notosuchia). American Museum Novitates 3512: 1-40.

von Zittel KA. 1890. Palaeozoologie, III, Vertebrata (Pisces, Reptilia, Aves). In: Handbuch der Paläoherpetologie, München and Leipzig. München and Leipzig: R. Oldenbourg, 900 pp.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the publisher's website:

Appendix S1 Full character list for phylogenetic analysis.

Appendix S2 Character state codings for taxa used for phylogenetic analysis.

APPENDIX 1

Character list

Sources for characters are provided, and new characters indicated. Where characters have been modified, details are given at the end of the character statement. All characters have been reformatted to a standardized notation to be as explicit as possible about the morphology. New characters are indicated, although some of these are the product of splitting previously used characters, or modified from statements in previously published works. Additional references have been incorporated into the main reference list above.

The following 47 characters are ordered: 3, 7, 10, 13, 20, 23-30, 33, 37, 56, 58, 84, 89, 99, 100, 103, 115, 119, 128, 133, 150, 153, 175, 179, 183, 203, 204, 217, 246, 264-266, 272, 285, 288-291, 300, 305, 321.

Cranial characters

1. Dorsal cranial bones (skull roof, cranial table), external surface: smooth (0); ornamented (1) (Clark, 1994).

2. Dorsal cranial bones (skull roof, cranial table), external surface: slightly grooved (0); heavily ornamented with deep pits and/or grooves (1); with shallow pits (2) (Clark, 1994) (character state 2 added here; added '/or' to state 1).

3. Snout, external surface, sculpting: absent (0); present but to a lesser degree than cranial table

(1); present, as prominent as on cranial table

(2) (Gasparini et al., 2006) (character state 1 added here; added as prominent as on cranial table' to character state 2) [ordered].

4. Rostrum, dorsal projection posterior to the external nares, relative to remainder of rostrum: absent, rostrum straight or low (0); rostrum upturned (1) (Andrade et al., 2011) (added posterior to the external nares, relative to remainder of rostrum').

5. Skull, lateral expansion at orbits relative to rostrum: gradual (0); abrupt (1) (Clark, 1994) (added lateral' and relative to rostrum').

6. Snout, lateral contour, in dorsal view: straight (0); sinusoidal ('festooned') (1) (Ortegaet al., 2000).

7. Snout, overall proportions: narrow oreinirostral (tall and domed) (0); nearly tubular (1); platy-

rostral (broad and flat) (2) (Clark, 1994; sensu Rayfield & Milner, 2008) [ordered].

8. Snout, profile of dorsal edge in lateral view (anterior to cranial table): concave (0); convex (1); approximately straight (2) (Sweetman et al., 2015).

9. External nares, orientation: facing anterolater-ally (0); facing dorsally (1); facing anterodor-sally (2) (Clark, 1994) (character state 2 added here).

10. External nares: completely divided by a septum (0); partially divided posteriorly (1); confluent, no indication of a septum (2) (Clark, 1994) [ordered].

11. Orbit, orientation: facing anterolaterally (0); facing fully laterally (1) (Wilkinson, Young & Benton, 2008).

12. Orbit, anterolateral border: continuous margin (0); develops as a small groove into pre-orbital elements (1) (new character).

13. External antorbital fenestra: large, >0.5 times the size of the orbit (0) small, <0.5 times the size of the orbit (1); absent (2) (Andrade et al., 2011) (changed to a multistate, adding character states 0 and 1 to replace character state 'present') [ordered].

14. Antorbital fenestra, shape: rounded or dorsoventrally high (0); dorsoventrally low and anteroposteriorly elongate, slit-like (1) (Gasparini et al., 2006).

15. External supratemporal fenestra: present (0); absent (1) (Ortega et al., 2006).

16. External supratemporal fenestra: perforated

(0); imperforated (1) (new character, adapted from Joffe, 1967).

17. External supratemporal fenestra, shape: square to subrectangular (0); circular to subcircular

(1); mediolaterally narrow and slit-like (2) (Andrade et al., 2011) (deleted character state triangular, converging medially'; added character state 2).

18. External supratemporal fenestra, maximum anteroposterior length: equal to or shorter than orbits (0); longer than orbits (1) (Clark, 1994) (added maximum' to refine character).

19. Intertemporal mediolateral width (minimum between supratemporal fenestrae), relative to interorbital mediolateral width (minimum

between orbits): intertemporal region broader (0); intertemporal region equal or narrower (1) (new character).

20. Lateral temporal fenestra in lateral view, size proportional to orbit in dorsal view: small to absent, no more than 20% of the area of the orbit (0); more than 20 to <50% of the area of the orbit (1); area is larger than 50% of the area of the orbit (2) (Andrade et al., 2011) (character state 1 added here) [ordered].

21. Lateral temporal fenestra, orientation: faces laterally (0); faces dorsolaterally (1) (Andrade et al., 2011).

22. Lateral temporal fenestra, shape: triangular (0); elliptical to subpolygonal (1) (Ortega et al., 2000).

23. Suborbital fenestra: small, <50% of orbital area

(0); between 50% and the same size as the orbit

(1); larger than the orbit (2) (Andrade et al., 2011) (quantified state 1) [ordered].

24. Intermandibular angle (degrees): lower than 40° (0); 40-45° (1); 46-50° (2); >50° (3) (new character) [ordered].

25. Skull length : skull width, ratio: <2.0 (0); 2.0 to <2.5 (1); 2.5 or greater (2) (new character) [ordered].

26. Skull length : snout length, ratio: <2.0 (0); 2.0 to <2.5 (1); 2.5 to <3.0 (2); 3.0 or greater (3) (new character) [ordered].

27. Skull length : orbit length, ratio: <3.0 (0); 3.0 to <4.0 (1) 4.0 to <5.0 (2); 5.0 or greater (3) (new character) [ordered].

28. Skull width : orbit width, ratio: <2.5 (0); 2.5 to <3.5 (1); 3.5 or greater (2) (new character) [ordered].

29. Skull length : supratemporal fenestra length, ratio: <6.0 (0); 6.0 to <7.0 (1); 7.0 to <8.0 (2) 8.0 or greater (3) (new character) [ordered].

30. Skull width : supratemporal fenestra width, ratio: <3.0 (0); 3.0 to <5.0 (1); 5.0 to <6.0 (2); 6.0 or greater (3) (new character) [ordered].

31. Premaxilla, maximum mediolateral width of paired premaxillae relative to that of the rostrum at the level of alveoli 4 or 5: premaxillae equal or narrower (0); rostrum narrower (1) (Jouve, 2009).

32. Premaxilla, anterior to nares: narrower than, or equal to, twice the anterior nasal mediolateral width (0); broader than twice the anterior nasal width (1) (Clark, 1994) (quantified).

33. Premaxilla-maxilla, distance between the anterior tip of the snout and the anterior-most position of the premaxilla-maxilla suture in dorsal view, relative to the distance between the anterior-most position of the premaxilla-maxilla suture and the posterodorsal extremity of the

premaxilla in dorsal view: distance between the tip of the snout and the anterior-most position of the premaxilla-maxilla suture larger (0); distances approximately equal (1); distance between the anterior-most position of the pre-maxilla-maxilla suture and the posterodorsal extremity of the premaxilla larger (2) (Jouve,

2004) [ordered].

34. Premaxilla-maxilla suture, small foramen in lateral surface (not for large mandibular teeth): absent (0); present (1) (Pol, 1999).

35. Premaxilla, projection of the internarial bar relative to the main body of premaxilla and narial opening: does not project anterior to the main body of the premaxilla (0); strongly projected anteriorly from narial opening, extending anterior to main body of maxilla (1) (Andrade et al., 2011).

36. Premaxilla, participation in internarial bar: forming at least the anterior half (0); with little participation (1) (Clark, 1994) (replaced 'ventral' with 'anterior' in character state 0).

37. Premaxilla, ventral edge relative to maxilla: lower than ventral edge of maxilla, with dorsal contour of anterior part of dentary strongly concave to accommodate (0); at same height as ventral edge of maxilla (1); premaxilla ventral edge dorsal to maxilla (2) (Ortega et al., 2000) (character state 2 added here) [ordered].

38. Premaxilla, perinarial crests: absent (0); present as well-defined and distinct ridges, cornering the lateral to posterior borders of the naris (1) (Andrade et al., 2011).

39. Premaxilla, notch on lateral edge of external nares: absent (0); present on the dorsal half of the lateral edge of the external nares (1) (Pol, 1999).

40. Premaxilla, perinarial fossa: absent (0); present (1) (Pol & Apesteguia, 2005).

41. Premaxilla, postnarial fossa: absent (0); present (1) (Andrade et al., 2011).

42. Premaxilla-maxilla, suture: confluent ventrally

(0); opened contact on ventral edge of rostrum

(1) (Clark, 1994).

43. Premaxilla-maxilla contact, orientation in dorsal view, whether or not posterodorsal process is present: anteromedially directed (0); postero-medially directed (1) (Schwarz & Salisbury,

2005) (added whether or not posterodorsal process is present').

44. Premaxilla, posterodorsal process: absent (0); present, extending posteriorly and wedging between maxillae and nasals (1) (Pol, 1999).

45. Premaxilla, orientation of anterior alveolar margin: vertical (0); out-turned (1) (Sereno et al., 2001) (character state 1 modified from inturned').

46. Maxillae, posterior palatal branches anterior to palatines: do not meet (0); meet (1) (Clark, 1994).

47. Maxilla-premaxilla, suture in palatal view medial to alveolar region: sinusoidal, posteromedially directed on lateral half and anteromedially directed along medial region (0); posteromedially directed (1) (Pol, 1999) (character states anteromedially directed' and premaxillae-maxil-lae suture U-shaped' removed).

48. Maxilla-premaxilla, lateral fossa excavating alveolus of last premaxillary tooth: absent (0); present (1) (Larsson & Sues, 2007).

49. Maxilla, depression on posterolateral surface, laterally positioned: absent (0); present (1) (Wu et al. , 1997) (added on posterolateral').

50. Maxilla, depression on anterolateral surface, medially positioned: absent (0); present (1) (new character).

51. Maxilla, lateral surface of jugal process (posterior portion): heavily striated (0); ornamented, like rest of rostrum (1); smooth (2) (new character).

52. Maxilla, ventral edge in lateral view: straight or convex (0); sinusoidal (1) (Ortega et al., 2000).

53. Maxilla and premaxilla, general shape of external surface: single plane facing laterally (0); with ventral region facing laterally and dorsal region facing dorsolaterally (1) (Pol, 1999).

54. Maxilla, presence of occlusal pit for reception of enlarged dentary tooth anterior to dental arcade (or M2, M stands for maxillary alveolus'): present (0); absent (1) (new character, adapted from Martin et al., 2014a,b).

55. Maxilla, evaginated alveolar edges: absent (0); present (1) (Gasparini et al., 2006).

56. Maxilla, lateral surface, unsculpted region along alveolar margin: absent (0); present (1) (Wu & Sues, 1996).

57. Maxilla, sculpturing of palatal surface: absent, palatal surface smooth (0); present anteriorly, absent posteriorly (1); present throughout, palatal surface ornamented with ridges (2) (Ortega et al., 2000) [ordered].

58. Maxilla, foramen on palatal surface, dorsome-dial to enlarged fifth tooth: absent (0); present (1); develops elongate groove (2) (new character) [ordered].

59. Maxillary teeth, dental implantation, middle teeth: confluent, located in dental groove (0); in isolated alveoli (1) (new character).

60. Maxillary teeth, dental implantation, posterior teeth: confluent, located in dental groove (0); in isolated alveoli (1) (new character).

61. Maxilla, palatine process: absent (0); present, next to the anterior border of suborbital fenes-trae (1) (Andrade & Bertini, 2008).

62. Maxilla-nasal, suture, orientation with respect to snout lateral margins: parallel (0); oblique (1) (new character).

63. Nasal, participation in dorsal margin of external nares: present (0); absent (1) (new character).

64. Nasal participation in margins of external nares: present posteriorly (0); present posteriorly and medially (1) (Clark, 1994).

65. Nasals: paired and unfused (0); partially or completely fused (1) (Gasparini et al., 2006) (added and unfused' to character state 0).

66. Nasal, lateral border posterior to external nares: concave (0); straight (1); convex (2); sinusoidal (3) (character states 2 and 3 added) (Pol, 1999).

67. Nasal, lateral edges: subparallel (0); oblique to one another, converging anteriorly (1) (Pol, 1999).

68. Nasal, participation in antorbital fenestra: present (0); absent (1) (Ortega et al., 2000).

69. Nasal, posterior portion of the dorsal surface: anteroposteriorly crenulated (0); smooth or sculpted as rest of rostrum (1) (new character).

70. Nasals, posterior mediolateral widening adjacent to the maxilla (anterior to contact with periorbital elements): abrupt (0); gradual (constant) (1); (new character, adapted from Lau-prasert et al., 2011).

71. Nasal-lacrimal contact: present (0); absent (1) (Clark, 1994).

72. Nasal-lacrimal contact: along medial surface of lacrimal (0); lacrimal forms a point contact with nasal (1) (Clark, 1994) (changed character state 1 from along medial and anterior surfaces of lacrimal' to forms a point contact with nasal').

73. Nasal, posterior tips of nasals: converge along the sagittal plane (0); separated by anterior projection of frontals (1) (Ortega et al., 2000).

74. Nasal, posterior portion and anterior portion of frontal, midline anteroposterior depression: absent (0); present (1) (Montefeltro, Larsson & Langer, 2011).

75. Lacrimal, total anteroposterior length relative to anteroposterior length of prefrontal: longer (0); shorter or equal to (1) (Brochu, 1999) (combined states 1 and 2).

76. Lacrimal, shape: anteroposteriorly longer than mediolaterally broad (0); as anteroposteriorly long as mediolaterally broad (1) (Sereno & Larsson, 2009).

77. Lacrimal, posterior extent and relationship with jugal: extending posteroventrally, widely contacting jugal (0); tapering posteroventrally, does not contact jugal or only point contact with jugal (1) (Zaher et al., 2006).

78. Lacrimal and jugal, anterior margins: confluent, with no notch at the anterior contact (0); jugal edge convex, producing an anterior notch at contact (filled with maxilla) (1) (Larsson & Sues, 2007).

79. Jugal, anterior extension below orbit, in dorso-lateral view: does not extend beyond anterior margin of orbit (0); extends beyond anterior margin (1) (Pol, 1999).

80. Jugal, dorsoventral depth of orbital portion in relation to infratemporal portion: almost the same to less than twice the depth (0); orbital portion twice the depth of the infratemporal portion (1) (Clark, 1994) (added to less than twice the depth' to character state 0).

81. Jugal, foramen on the lateral surface near the anterior margin: absent (0); present (1) (Zaher et al., 2006).

82. Jugal, anterior process length relative to infratemporal fenestrae anteroposterior length: 1.0 or less times the length (0); longer than 1.0 times the length (1) (Larsson & Sues, 2007) (changed character states 0 and 1 to define state boundary ratio).

83. Jugal, orientation of base of postorbital process: directed posterodorsally (0); directed dorsally (1) (Pol, 1999).

84. Jugal, location of postorbital process relative to main jugal body: anteriorly placed (0); in the middle (1); posteriorly positioned (2) (Pol, 1999) [ordered].

85. Jugal portion of postorbital bar, relative to lateral surface of jugal: flush with lateral surface of jugal (0); inset (1) (Ortega et al., 2000).

86. Jugal, anterior fossa: bordered by ornamented ridge (0); continuous with lateral surface (1) (new character).

87. Jugal, anteroposterior ridge on lateral surface below infratemporal fenestrae: absent (0); present (1) (Pol & Norell, 2004b).

88. Jugal-ectopterygoid, suture ridge: absent (0); present (1) (Montefeltro et al., 2011).

89. Preorbital elements, anterior palpebral bone: no notable depression or projection (0); marked depression, developing into an incipient lateral projection (1); marked depression forming a prominent lateral projection for the support of the anterior palpebral (2) (Sereno & Larsson, 2009) (character state 0 added here) [ordered].

90. Prefrontal, lateral development: reduced, no notable lateral projection (0); enlarged, extending laterally or posterolaterally over orbit (1) (Gasparini et al., 2006) (changed to 'laterally or posterolaterally' in character state 1).

91. Prefrontal, anterior morphology: tapers anteriorly to a point (0); anteriorly broad (1) (new character).

92. Prefrontal-frontal sutures, form paired dorsal crests: absent (0); present (1) (Pol & Powell, 2011).

93. Prefrontal-lacrimal suture, crest: absent (0); present, situated anterior to orbit (1) (Andrade et al., 2011) (changed 'dorsal' to 'anterior' in character state 1).

94. Prefrontal and lacrimal around orbits: forming flat rims (0); evaginated, forming elevated rims from the dorsal surface of the skull (1) (Gas-parini et al. , 2006).

95. Prefrontal pillars (ventral process): not contacting palate (0); contacting palate (1) (Clark, 1994).

96. Frontals: unfused (0); fused (1) (Clark, 1994) (changed to indicate degree of fusion).

97. Frontal, mediolateral width of paired frontals between orbits: broader than nasals (0) equal or narrower than nasals (1) (Clark, 1994) (added equal or' to character state 1).

98. Frontals, mediolateral width between orbits: narrower than posterior end (posterior end flares laterally posterior to orbits) (0); equal width or broader than posterior end (1) (new character, adapted from Sweetman et al., 2015).

99. Frontal, morphology of anterior-most border of anterior process: truncated (0); wedge-like (1); bifurcated (2) (character state 2 added here) (Andrade et al., 2011) [ordered].

100. Frontals, dorsal surface: flat (0); with antero-posterior ridge along midline suture (1) (Clark, 1994) (added along midline suture' to character state 1) [ordered] [note: see Schwarz & Salisbury (2005) for discussion of how this character relates to ontogeny].

101. Frontal, dorsal anteroposterior ridge(s): restricted to the posterior portion (0); restricted to median portion (1); restricted to anterior portion (2); occupy entire length of frontal (3) (Montefeltro et al., 2011) (character state 3 added).

102. Frontal, anterior extension of anterior margin: level with, or anterior to, the orbits (0); does not reach the anterior margin of the orbits (1) (Andrade et al., 2011).

103. Frontal, participation in orbit border: forming great part of posterior, medial, and anterior (or anteromedial) regions (0); restricted to posterior and posteromedial region (1); restricted to medial margin (2) (Montefeltro et al., 2013) (character state restricted to the posterior region' removed; character states 1 and 2 added here) [ordered].

104. Frontal, transverse ridge crossing anteromedial to the orbits: absent (0); present (1) (Pol et al., 2009).

105. Frontal, participation in supratemporal fenestra: absent (0); present (1) (new character).

106. Frontal, participation in supratemporal fenes-tra: anteromedially (0), anteriorly only (1) (new character).

107. Frontal, anterior ramus with respect to the anterior tip of the prefrontal: does not extend past the anterior tip (0); extends beyond the anterior tip of the prefrontal (1) (Sereno et al., 2001) (modified states from 'ending anteriorly' and 'ending posteriorly' to remove potential gap between states).

108. Frontal, lateral margin relative to the skull surface: flush (0); elevated, forming ridged orbital margins (1) (Brochu, 1999).

109. Frontal, anterior process constriction with respect to main body of frontal, excluding sagittal projection into nasals anterior to orbits: absent, lateral edges parallel to subparallel (0); present, anterior portion mediolaterally constricted, with convergent lateral margins (1) (Montefeltro et al., 2013) (added 'with respect to main body of frontal, excluding sagittal projection into nasals anterior to orbits').

110. Postorbital, anterolateral process: absent or poorly developed (0); well-developed, long and distally acute (1) (Clark, 1994).

111. Postorbital-jugal contact, configuration: postorbital medial to jugal (0); postorbital dorsal to jugal (1) (Clark, 1994).

112. Parietal-postorbital suture: absent from the dorsal surface of the skull roof (0); present on the dorsal surface of the skull roof (1) (Clark, 1994) (character broken down into characters 107-109).

113. Parietal-postorbital suture: absent from the supratemporal fossa (0); present within the supratemporal fossa (1) (new character).

114. Parietal-postorbital suture within the supratemporal fossa: present within the ventral region (0); broadly present (1) (new character).

115. Parietal, dorsal surface: projects dorsally, relative to the skull roof (0); same level as squamosal (1); depressed relative to the squamosal (2) (Andrade et al., 2011) (character state 2 added here) [ordered].

116. Parietal, posterior region dorsal surface: smooth (0); presenting a anteroposterior dorsal ridge (1); marked ventral deflection ( bevelled') in posterior portion (2); sculpted as with the rest of the skull table (3) (Montefeltro et al., 2013) (added anteroposterior' to character state 1; added character states 2 and 3 here).

117. Parietal-squamosal emargination (anterior concavity at suture contact), posterior margin in dorsal view: absent (0); present (1) (Wilkinson et al., 2008).

118. Supratemporal roof, dorsal surface: postorbital and squamosal with flat shelves extending laterally beyond quadrate contact (0); lateral edge terminating medial or immediately dorsal to medial-most point of contact with quadrate (1) (Clark, 1994) (character state complex' removed; character state 1 added).

119. Supratemporal fenestra, medial border: flat, sculpted region (or unsculpted if rest of cranial table unsculpted) (0); forming a low sagittal crest (rims) posteriorly (1); forming a low sagittal crest (rims) along full length (2) (Clark, 1994) (character state 1 added here) [ordered].

120. Supratemporal fenestra, dermal bone overhang: absent (0); present (1) (Norell, 1988).

121. Supratemporal fenestra, dermal bone overhang: present only medially and posteriorly (0); present about the entire edge (1) (Norell, 1988).

122. Supratemporal fenestra, shallow fossa at anteromedial corner: present (0); absent, corner smooth (1) (Brochu, 1999).

123. Supratemporal medial rims: continuous with orbital rims (0); separated from orbital rims by the postorbital bar (1) (new character).

124. Supratemporal medial rims, extend posteriorly to contact posterior skull margin: present (0); absent (1) (new character).

125. Supratemporal fenestra, relative contribution of frontal and parietal to medial margin: parietal with equal or greater contribution (0); frontal excluded from margin (1) (new character).

126. Supratemporal fenestrae, minimum width between fenestrae, with respect to maximum width of cranial table: one-third or less of total width (0); more than one-third of total width (1) (new character; adapted from Schwarz & Salisbury, 2005).

127. Orbitotemporal channel, size of the dorsal aperture: area of foramen <30% of that of the internal supratemporal fenestra (0); larger than 30% of the internal supratemporal fenes-trae area (1) (Montefeltro et al., 2013).

128. Postorbital bar between orbit and supratemporal fossa, shape: broad and solid, as broad as dorsal surface of the cranial table lateral to the supratemporal fenestra (0); much narrower (1); much narrower and connected to orbit via a thin, superficial furrow in postorbital (2) (Clark, 1994) (replaced with broadly sculpted dorsal surface if sculpture present' with size-related quantifier in character state 0; removed sculpting restricted to anterior surface' from character state 1; added character state 2) [ordered].

129. Postorbital bar between orbit and supratemporal fenestra, external texture: sculpted (0); unsculpted (1) (Clark, 1994).

130. Postorbital bar, lateral surface formed by: postorbital and jugal (0); only by postorbital (1) (Gasparini et al., 2006).

131. Postorbital bar, shape: transversely flattened (0); transversely broad, with elliptical cross-section (1); slender and cylindrical (2) (Clark, 1994).

132. Postorbital bar, shape of dorsal end nearing skull table: continuous with dorsal part of postorbital (0); dorsal part of postorbital bar constricted, distinct from the dorsal part of the postorbital (1) (Clark, 1994).

133. Cranial table (skull roof), width with respect to ventral portion of skull: as wide as ventral portion of skull (quadrates covered by squamosal) (0); narrower, but still covering more than half of the mediolateral region of quadrates (1); narrower, exposing more than half of mediolateral region of quadrate (2) (Wu et al., 1997) (added 'more than half to character states 1 and 2) [ordered].

134. Squamosal and postorbital, lateral margins, dorsal view excluding the squamosal postero-lateral process: parallel (0); diverging posteriorly (1); medially concave (2); converging posteriorly (3) (Ortega et al., 2000) (character state 2 added; added excluding the squamosal posterolateral process' to character state 0).

135. Squamosal, lateral surface, longitudinal groove for attachment of the upper ear lid: absent (0); present (1) (Clark & Sues, 2002).

136. Squamosal groove for upper ear lid: ventral edge is laterally displaced relative to dorsal edge (0); ventral edge is directly beneath dorsal edge (1) (Clark & Sues, 2002).

137. Squamosal, dorsal edge of groove for dorsal ear lid: parallel to ventral edge (0); dorsal margin with a medial curvature (1) (Montefeltro et al., 2013).

138. Squamosal, dorsolateral edge: straight and parallel to skull roof (0); bevelled ventrally, with anterolateral notch (1) (new character, adapted from Schwarz & Salisbury, 2005).

139. Squamosal, posterolateral process: present (0); absent (1) (new character, adapted from Bus-calioni & Sanz, 1990a).

140. Squamosal, posterolateral process: level with skull table (0); depressed from skull table (1) (Sereno & Larsson, 2009).

141. Squamosal, posterolateral process projection: ventrally directed, confluent with ventral rim of groove for the earflap (0); posteriorly directed and parallel to skull roof (1) (Ortega et al., 2000) (removed character state 2).

142. Squamosal, posterolateral process, ornamentation: absent (0); present (1) (Larsson & Sues, 2007).

143. Squamosal, posterolateral process, distal end: tapered and pointed (0); broad and rounded (1) (Larsson & Sues, 2007).

144. Squamosal, anterior process extending anteriorly to the orbital margin, overlapping the postorbital, in lateral view: absent (0); present (1) (Turner & Buckley, 2008).

145. Squamosal, obliquely orientated ridge on dorsal surface: present (0); absent (1) (new character).

146. Squamosal, oblique ridge on dorsal surface, position with respect to the supratemporal fen-estra: posterior to supratemporal fenestra (0); posterolateral or lateral to supratemporal fen-estra (1) (new character).

147. Squamosal-parietal suture: flat, not elevated from the skull table (0); forms a well-developed anteroposterior groove (often bounded by elevated ridges) (1) (new character, adapted from Buffetaut, 1983).

148. Squamosal, anterior portion nearing orbital edge: sculpted or unsculpted, consistent with the rest of the skull table (0); sculpting pattern changes (1) (new character).

149. Quadratojugal, ornamentation at base (dorso-lateral surface): absent (0); present (1) (Pol, 1999).

150. Quadratojugal, length of anterior process relative to the lower temporal bar: absent or less than one-third of lower temporal bar (0); one-third to one half the length of the lower temporal bar (1); long, greater than half of the lower temporal bar (2) (Larsson & Sues, 2007) (changed character states to close gap between short' and half' of length of lower temporal bar) [ordered].

151. Quadratojugal, shape of posterolateral end and relationship with quadrate: acute or rounded, tightly overlapping quadrate (0); sinusoidal ventral edge, and wide and rounded posterior edge slightly overhanging lateral surface of quadrate (1) (Pol & Norell, 2004a).

152. Quadratojugal, contribution to the lateral temporal fenestra, in dorsal view: extensive contact with the ventral and posterior margins (0); contributes to the posterior and dorsal margins (1); only contributes to the posterior margin (2) (new character).

153. Quadratojugal-postorbital contact, in lateral view: not in contact (0); small, point contact (1); broad contact between the quadratojugal and the posterior portion of the postorbital descending flange (2) (Clark, 1994) (added character state 0 here) [ordered].

154. Infratemporal fenestra, posterior margin, dorsal view: straight (0); with an anterior projection, forming an acute angle (1) (Ortega et al.,

2000) (added forming an acute angle' to character state 1).

155. Quadrate, posterior edge: broad medial to tympanum, gently concave (0); narrow dorsal to otoccipital contact, strongly concave (1) (Clark, 1994).

156. Quadrate, dorsal surface fenestration: absent (0); present (1) (Clark, 1994).

157. Otic aperture (not including additional quadrate fenestrae): open posteriorly (0); closed posteriorly by quadrate and otoccipital (1) (Clark, 1994).

158. Quadrate, distal body: anterior margin oriented at a right angle in relation to quadratojugal

(0); anterior margin gently slopes relative to quadratojugal (1) (Montefeltro et al., 2011).

159. Quadrate, pterygoid ramus: with flat ventral edge (0); with deep groove on ventral surface

(1); rod-like (2) (Clark, 1994) (character state 2 added here).

160. Quadrate, anterodorsal ramus in ventral view: developed, forming more than or equal to 50% of the lateral edge of the internal supratemporal fenestra (0); restricted, forming <50% of the lateral edge of the supratemporal fenestra (1) (Montefeltro et al., 2011).

161. Quadrate, ventral surface: smooth, with simple muscle scars (0); with multiple developed ridges (1) (Osi et al., 2007).

162. Quadrate, condyles: with poorly developed intercondylar groove (0); medial condyle expands ventrally, being separated from the lateral condyle by a deep intercondylar groove (1) (Ortega et al., 2000).

163. Quadrate, development of distal body ventral to otoccipital-quadrate contact: distinct, developing posteroventrally to contact (0); indistinct, not surpassing contact (1) (Wu et al., 1997) (added developing posteroventrally to contact' to character state 0, and not surpassing contact' to character state 1).

164. Quadrate, dorsoventral height of the proximal region: <50% of the skull roof total width (0); more than 50% of the skull roof total width (1) (Montefeltro et al., 2013) (added 'or equal to' to character state 0).

165. Cranioquadrate canal: opened laterally (0); closed laterally (1) (Clark, 1994).

166. Ectopterygoid-maxilla, contact: present (0); absent (1) (new character).

167. Ectopterygoid, contribution to postorbital bar: absent (0); present (1) (Sereno & Larsson, 2009).

168. Ectopterygoid, main axis orientation: mediolat-erally or slightly anterolaterally (0); anteropos-teriorly, subparallel to anteroposterior axis of

skull (1) (Pol et al., 2004) (changed laterally to mediolaterally in character state 0, and anteriorly to anteroposteriorly' in character state 1).

169. Ectopterygoid, extent of medial projection on the ventral surface of pterygoid flanges: barely extended (0); widely extended, covering approximately the lateral half of the ventral surface of the pterygoid flanges (1) (Zaher et al., 2006).

170. Ectopterygoid, anterior process: developed (0); reduced or absent (1) (Pol, 1999).

171. Ectopterygoid, posterior process: developed (0); reduced or absent (1) (Pol, 1999).

172. Palatines, palatal processes: do not meet on palate below narial passage (0); meet ventral to narial passage, forming part of secondary palate (1) (Clark, 1994).

173. Palatine-maxilla, suture when fused at mid-line: palatine anteriorly rounded (0); palatine anteriorly pointed (1); suture transverse to midline axis (2) (Brochu, 1999) (character state palatine invaginated' removed).

174. Interfenestral bar, anterior half between suborbital fenestrae, lateral margins: parallel to subparallel (0); flared anteriorly (1) (Pol et al., 2009) (added lateral margins').

175. Interfenestral bar, posterior half between suborbital fenestrae, lateral margins: flared posteriorly (0); parallel to subparallel (1); converge posteriorly (2) (Pol et al., 2009) (added 'lateral margins'; character state 2 added here) [ordered].

176. Palatines, anteroposterior axis: run parasagit-tally (0); diverge laterally, becoming rod-like and forming palatine bars posteriorly (1) (Mar-tinelli, 2003).

177. Palatine-pterygoid, contact on palate: run parasagittally (0); palatines firmly sutured to pterygoids (1) (Pol & Norell, 2004a).

178. Pterygoids, contact with one another on palate: not in contact anterior to basisphenoid on the palate (0); pterygoids in contact (1) (Wu et al., 1997).

179. Pterygoid, role of primary palate in forming choanal opening: forms posterior half of choa-nal opening (0); forms posterior, lateral, and part of anterior margin of choana (1); completely encloses choana (2) (Clark, 1994) [ordered].

180. Pterygoid, participation in the suborbital fenes-tra, ventral view: forms margin of suborbital fenestra (0); excluded from suborbital fenestra by ectopterygoid-palatine contact (1) (Turner & Sertich, 2010).

181. Choanae, anterior edge, location: situated between suborbital fenestrae (or anteriorly) (0); near posterior edge of suborbital fenestrae (1) (Clark, 1994).

182. Choanal opening, conformation in palate: continuous with pterygoid ventral surface except for anterior and anterolateral borders (0); opens into palate through deep choanal groove (1) (Clark, 1994).

183. Choanal groove: undivided (0); partially sep-tated (1); completely septated (2) (Clark, 1994) [ordered].

184. Pterygoid, quadrate process: well developed, extending posterolaterally beyond anterior margin of basioccipital (0); poorly developed, only present as an incipient projection (1) (Pol, 1999) (added extending posterolaterally beyond anterior margin of basioccipital' to character state 0, and only present as an incipient projection' to character state 1).

185. Pterygoid, quadrate ramus, in ventral view: narrow and bar-like (0); broad and laminar (1) (Wu et al., 1997) (added 'bar-like' to character state 0, and laminar' to character state 1).

186. Pterygoid, palatal surface: smooth (0); sculpted (1) (Clark, 1994).

187. Pterygoid flanges: mediolaterally expanded, laterally surpassing the quadrate medial condyle

(0); relatively short, and do not reach laterally to the level of the quadrate medial condyle (1) (Osi et al., 2007).

188. Basisphenoid, ventral exposure on braincase: exposed on ventral surface of braincase (0); virtually excluded from ventral surface by ptery-goid and basioccipital (1) (Clark, 1994).

189. Basisphenoid, lateral exposure on braincase: absent (0); present (1) (Pol, 1999).

190. Basisphenoid: ventral surface continuous with surrounding bones (0); body ventrally developed and separated from the remaining elements by a posteroventral step formed by a sulcus separating it from the main occipital plane, forming a postchoanal pterygoid-ba-sisphenoid tuberosity (1) (Montefeltro et al., 2011).

191. Basisphenoid, ventral surface, mediolateral size relative to basioccipital: shorter than basioccipital (0); equal or longer than basioccip-ital (1) (Clark, 1994) (added mediolateral').

192. Basioccipital: without well-developed bilateral tuberosities (0); with large, pendulous tubera

(1) (Clark, 1994).

193. Basioccipital, midline crest on basioccipital plate below occipital condyle: absent (0); present (1) (Turner & Sertich, 2010).

194. Basioccipital and ventral part of otoccipital, orientation (when skull held horizontally): posteriorly (0); posteroventrally (1) (Gomani, 1997).

195. Otoccipital, ventrolateral contact with quadrate: very narrow, otoccipital only abuts quadrate (0);

broad, ventrolateral margin of otoccipital extensively contacts ventromedial portion of quadrate (1) (Clark, 1994) (added 'otoccipital only abuts quadrate' to character state 0, and ventrolateral margin of otoccipital extensively contacts ventromedial portion of quadrate' to character state 1).

196. Supraoccipital, posterodorsal exposure in skull roof: absent (0); present (1) (Ortega et al., 2000).

197. Supraoccipital, posterodorsal exposure: exposed in midline portion of posterior region of skull table (0); restricted to a thin surface attached to posterior-most portion of parietal and squa-mosal (1) (Montefeltro et al., 2011).

198. Supraoccipital, relationship with foramen magnum: forms dorsal edge (0); otoccipitals meet dorsally, separating the foramen magnum from the supraoccipital (1) (Clark, 1994).

199. Cranial nerves IX-XI, passage through brain-case: all pass through common large foramen vagi in otoccipital (0); cranial nerve IX passes medial to nerves X and XI in separate passage

(1) (Clark, 1994).

200. Mastoid antrum, location: does not extend into supraoccipital (0); extends through transverse canal in supraoccipital to connect middle ear regions (1) (Clark, 1994).

Mandibular characters

201. Mandible, outer surface sculpture, lateral surface: absent (0); present (1) (Montefeltro et al., 2011).

202. Mandible, outer surface sculpture, ventral surface: present on dentary (0); present on dentary and splenial (1) (Montefeltro et al. , 2011).

203. Mandibular symphysis, anteroposterior length relative to mediolateral width: short, length and width subequal or shorter than wide (0); proportionally long, longer than wide (1); extremely long, length at least five times its width

(2) (Andrade et al., 2011) [ordered].

204. Mandibular symphysis, posterior extension, terminating medial to the dentary alveoli: short, up to the D5 (0); to the D5-D6 (1); to the D7 or greater in length (new character) [ordered]. D: dentary alveolus.

205. Mandibular symphysis, lateral view: shallow and tapering dorsoventrally anteriorly (0); deep and tapering dorsoventrally anteriorly (1); shallow and anterior margin convex (2) (Wu & Sues, 1996) (character state 'deep and anteriorly convex' removed).

206. Mandibular symphysis, shape, in ventral view: tapering mediolaterally anteriorly, forming an

angle (0); U-shaped, smoothly curving anteri- 217. orly (1); lateral edges anteroposteriorly orientated with convex anterolateral cornier and extensive, transversely orientated anterior 218 edge (2) (Pol, 1999).

207. External mandibular fenestra: absent (0); present (1) (Clark, 1994). 219.

208. External mandibular fenestra, size: present as a diminutive passage, <50% of the total size of

the lateral temporal fenestra (0); present as an 220 evident fenestra, 50% or greater than the total size of the lateral temporal fenestra (1) (Clark, 1994) (quantified both character states). 221.

209. External mandibular fenestra, orientation of main axis: horizontal to subhorizontal (0); inclined, directed anteroventrally-posterodor- 222 sally (Andrade et al., 2011) (added 'to subhorizontal' to character state 0).

210. External mandibular fenestra, shape: subcircu- 223 lar to elliptical (0); triangular (1) (Andrade

et al., 2011) (character states 'highly elliptic, anteroposterior axis much longer than dorso- 224 ventral axis, three time or more, but both ends rounded', slit-like, proportionally very long and both ends acute', and teardrop-like' removed.) 225

211. Jaw joint, location of dorsal edge: level with or dorsolateral to occipital condyle (0); ventrolat- 226 eral occipital condyle (1) (Wu & Sues, 1996).

212. Dentary, lateral surface adjacent to seventh 227 alveolus: smooth (0); with lateral concavity for reception of enlarged maxillary tooth (1) (Buck- 228 ley & Brochu, 1999).

213. Dentary, lateral surface below alveolar margin,

at middle to posterior region of tooth row: ver- 229 tically orientated, continuous with rest of lateral surface of the dentaries (0); flat surface exposed dorsolaterally, divided by ridge from 230 the rest of the lateral surface of the dentary (1); flat, unsculpted surface confluent with rest 231 of the lateral surface (2) (Pol & Apesteguia, 2005) (character state 2 added here).

214. Dentary, relative to external mandibular fenes- 232 tra: extends posteriorly beneath mandibular fenestra, posteriorly exceeding anterior margin

(0); does not extend beneath fenestra, either terminating anteriorly to fenestra or only forming a point contact (1) (Clark, 1994).

215. Dentary, mediolateral compression and ventro- 233. lateral surface anterior to mandibular fenestra

(or of anterior portion posterior to symphysis if fenestra is absent): compressed and flat (0); 234 uncompressed and convex (1) (Ortega et al., 1996) (added mediolateral'). 235

216. Dentary, sculpted below the tooth row: lacking sculpting (0); present (1) (Pol, 1999).

Dentary alveoli: all independent of one another

(0); some confluent (1); all confluent, within continuous alveolar groove (2) (new character) [ordered]. Dental alveoli, transitional shape morphology from circular to subcircular or oval: absent (0); present (1) (new character).

Dentary alveoli, transitional shape morphology: posteriorly from D4 (0); posteriorly from D5 (1) (new character).

Dentary, distinct foramina on occlusal surface, lingual to dental arcade: absent (0); present (1) (new character).

Dentary, distinct foramina on occlusal surface, lingual to dental arcade: at D2-D3 (0); at D4 or positioned more posteriorly (1) (new character). Dentary, external alveolar margins, dorsal edge: vertically festooned, forming raised rims about each alveolus (0); flat (1) (new character). Dentary, internal alveolar margins: forming raised rims (0); flat and confluent with dentary occlusal surface (1) (new character). Dentary, anterior portion, lateral margin shape in dorsoventral view: straight (0); distinctly spatulate, with abrupt lateral expansion (1); laterally convex (2) (new character). Dentary, diastema (gap) between D7 and D8: present (0); absent (1) (new character). Dentary, pitted ornamentation of external surface: absent (0); present (1) (new character). Dentary, grooved ornamentation of external surface: absent (0); present (1) (new character). Dentary, interalveolar septae within anterior dental arcade (D4-D8): present (0); absent (1) (new character).

Dentary, symphysis and dentary arcade lateral to symphysis, in dorsoventral view: parallel (0); oblique (1) (new character). Dentary, occlusal surface: smooth (0); antero-posteriorly crenulated (1) (new character). Dentary, obliquely inclined crenulations pos-terodorsal to D8-D9: present (0); absent (1) (new character).

Dentary, dorsolateral edge: slightly concave or straight and subparallel to anteroposterior axis of skull (0); sinusoidal, with two concave waves

(1) (Ortega et al., 1996) (character state 'with single dorsal expansion and concave posteriorly' removed).

Splenial, involvement in symphysis, in ventral view: not involved (0); involved (1) (Clark, 1994)

Splenial, contact with dentary, in ventral view: confluent (0); dorsally inset (1) (new character). Splenial, posterior to symphysis: approximately constant mediolateral thickness throughout element (0); more robust posterodorsally (1)

(Ortega et al., 1996) (changed character state thin' to approximately constant mediolateral thickness throughout element'; changed character state 1 to more robust posterodorsally').

236. Angular and posterior surangular, strong pitted pattern: absent (0); present (1); lateral surface with rugose pattern instead of pits (2) (Andrade et al., 2011) (character state 2 added here).

237. Surangular, dorsal edge in lateral view: mostly straight (0); arched dorsally, excluding articular projection (1) (Clark, 1994).

238. Surangular, anteroposterior ridge along the dorsolateral surface: absent (0); present (1) (Pol & Norell, 2004b).

239. Surangular, extension toward posterior end of retroarticular process: along entire length (0); pinched off anterior to posterior tip (Norell, 1988).

240. Articular, posterior ridge on glenoid fossa: posterior margin well developed, evidently high (0); posterior margin poorly delimited, crest absent (1) (Pol & Apesteguia, 2005).

241. Articular, medial process articulating with otoccipital and basisphenoid: absent (0); present (1) (Clark, 1994).

242. Retroarticular process: absent or extremely reduced (0); posteroventrally projecting and paddle-shaped (1); pointed, projects posteriorly and ventrally recurved (2); projects posteriorly and dorsally recurved (3) (Clark, 1994) (character states with an extensive rounded, wide, and flat (or slightly concave) surface projected posteroventrally and facing dorsomedially' and posteriorly elongated, triangular, and facing dorsally' removed; character states 2 and 3 added).

Dental characters

243. Premaxillary teeth, number: five or more (0); four or fewer (1) (Wu & Sues, 1996) (character states six', three', and two' removed, and replaced with or more' and or fewer' in remaining character states).

244. Posterior premaxillary teeth, apicobasal length: <1.5 times the size of the anterior teeth (0); 1.5 times or greater than anterior teeth (1) (Clark, 1994) (quantified and set character state boundary).

245. Maxillary teeth, mesiodistal margin carinae: absent or with smooth and crenulated carinae (0); with denticulate carinae (ziphodont condition) (1) (Ortega et al., 1996) (character state 'with tubercular heterogenic denticles' removed).

246. Maxillary tooth rows, middle to posterior elements: crowns not mesiodistally compressed, subcircular in cross-section (0); crowns slightly

compressed mesiodistally (mesiodistal to labi-olingual diameter ratio more than 0.5 at mid-height) (1); crowns highly compressed mesiodistally (mesiodistal to labiolingual diameter ratio <0.5 at mid-height) (2) (Pol, 1999) [ordered].

247. Maxillary tooth rows, mesiodistal compression of middle to posterior elements: absent, or symmetrical compression (0); asymmetrical compression, occurring only along the distal margin giving teeth a teardrop shape (1) (Andrade & Bertini, 2008) (added or symmetrical compression' to character state 0).

248. Maxillary teeth, middle to posterior elements, ridged ornamentation on enamel surface: absent (0); present (1) (Andrade et al., 2011).

249. Maxillary teeth, enamel surface: smooth or slightly crenulated (0); with ridges at base of crown (often extending apically) (1) (Turner & Sertich, 2010).

250. Maxillary teeth, striations on labial and lingual faces: present (0); absent (1) (new character, adapted from Martin et al., 2014a,b).

251. Cheek teeth, base (i.e. immediately apical to root), with respect to remainder of tooth crown: not constricted (0); constricted (1) (new character, adapted from Martin et al., 2014a,b).

252. Maxillary teeth, width of root with respect to crown: narrower (0); wider in anterior teeth and equal in posterior teeth (1) (Ortega et al., 2000) (changed character state 1 to wider in anterior teeth and equal in posterior teeth'; removed or equal' from character state 0).

253. Maxillary teeth, posterior teeth, mediolaterally compressed lanceolate-shaped morphotype (sometimes called leaf-shaped'), visible in labial or lingual view, with wide crown tapering apically to a sharp point (note that the point can often be abraded): present (0); absent (1) (new character, adapted from Schwarz & Salisbury, 2005).

254. Maxillary teeth, low-crowned and strongly labi-olingually compressed morphotypes, forming a crown that is mesiodistally broader than it is api-cobasally tall: present, apical margins orientated at <45° from horizontal (0); absent (1) (new character, adapted from Schwarz & Salisbury, 2005).

255. Tooth, present at premaxilla-maxilla contact with transitional size-based morphology: absent (0); present (1) (Turner & Sertich, 2010) (added size-based').

256. Maxillary teeth, size variation waves: absent, no tooth size variation (0); one wave of enlarged teeth (1); enlarged maxillary teeth occur in two waves (festooned) (2) (Clark, 1994).

257. Enlarged maxillary teeth (at least 1.5 times the apicobasal size of remaining teeth): present at M2 and/or M3 (0); present at M4 and/or M5 (1) (Martin et al., 2014a,b). M: maxillary alveolus.

258. Maxillary tooth 5, apicobasal size relative to adjacent maxillary teeth: subequal, or <4.0 times the size of adjacent teeth (0); hypertro-phied, at least 4.0 times the size of adjacent teeth (1) (new character).

259. Maxillary tooth 5, hypertrophied: directed pos-teroventrally (0); directed ventrally (1) (new character).

260. Maxillary teeth 6 and 7: continuous with tooth row (0); dorsally inset (1) (new character).

261. Maxillary teeth, bulbous tooth morphotype (tri-bodont): present (0); absent (1) (Sweetman et al., 2015).

262. Dentary teeth, anterior teeth (opposite premax-illa-maxilla contact) apicobasal length, relative to rest of dentary teeth: no more than twice the length (0); more than twice the length (1) (Clark, 1994).

263. Dentary teeth, posterior teeth: occlude medial to opposing maxillary teeth (0); occlude lateral to, or interlock with, opposing maxillary teeth (1) (new character, adapted from Sweetman et al., 2015).

Axial characters

264. Vertebrae, centra shape along axial column: cylindrical throughout (0); grade continuously from cylindrical to elongated spool-shaped (1); spool-shaped throughout (2) (Buscalioni & Sanz, 1988) (character state 1 added) [ordered].

265. Cervical vertebrae: amphicoelous or amphiplat-yan (0); procoelous, and posterior centrum face (condyle) with a central depression ( semi-pro-coely') (1); fully procoelous (2) (Clark, 1994) (character state 1 added) [ordered].

266. Cervical vertebrae, number: six or fewer (0); seven (1); eight or more (2) (new character) [ordered].

267. Atlas, intercentrum size: mediolaterally wider than anteroposteriorly long (0); subequal diameters or anteroposteriorly longer (Clark, 1994).

268. Cervical vertebrae, neural spine: absent, or extremely reduced (0); present, distinct from centrum body (1) (new character).

269. Cervical vertebrae, neural spines: rod-like and elongate (0); short and transversely flattened (1) (new character).

270. Cervical and dorsal vertebrae, hypapophyses or anterior keels: absent (0); present (1) (Bus-

calioni & Sanz, 1988) (character states modified to present or absent).

271. Dorsal vertebrae: amphicoelous or amphiplat-yan (0); procoelous (1) (Clark, 1994) (replaced trunk' with dorsal').

272. Dorsal vertebrae, number: 14 or fewer (0); 15-16 (1); 17 or more (2) (new character) [ordered].

273. Posterior dorsal vertebrae, transverse process shape: dorsoventrally low and laminar (0); dorsoventrally high (1) (Buscalioni & Sanz, 1988).

274. Sacral vertebrae, number: two (0); three or more (1) (Buscalioni & Sanz, 1988).

275. Sacral vertebrae, orientation of transverse processes: project laterally (horizontally) (0); deflected markedly ventrally (1) (Gasparini et al., 2006).

276. Caudal vertebrae, number: fewer than 50 (0); 50 or more (1) (new character).

277. Caudal vertebrae: amphicoelous or amphiplat-yan (0); procoelous (1) (new character, adapted from Salisbury & Frey, 2005).

278. Caudal vertebrae, first: same morphology as rest of caudal series (0); biconvex (1) (new character, adapted from Salisbury & Frey, 2001).

279. Caudal vertebrae, anteroposterior ridge/lamina separating centrum and neural arch: present (0); absent (1) (new character; note that this could be an ontogenetic feature).

Appendicular characters

280. Scapula, proximodorsal edge in lateral view: flat and confluent with scapular shaft (0); forms a distinct crest (1) (new character).

281. Coracoid, medial process: elongate posterome-dial process (0); distally expanded ventromedial process (1) (Wu & Sues, 1996).

282. Coracoid, distal expansion: equal to or larger than the proximal expansion (0); less expanded than the proximal region (1) (Pol et al., 2012).

283. Humerus, circular depression on the posterior surface of the proximal end, for the insertion of the M. scapulohumeralis caudalis: absent (0); present (1) (Pol et al., 2012).

284. Humerus, lateral and medial surfaces of distal end: flat and anteroposteriorly broad, similar in anteroposterior length to the transverse width of the distal end of the humerus (0); convex and reduced in comparison with the transverse width of the distal humerus (1) (Pol et al., 2012).

285. Forelimb : hindlimb length, ratio: <0.7 (0); 0.7 to <0.8 (1); 0.8 or greater (2) (new character) [ordered].

286. Humerus : femur length, ratio: <0.75 (0); 0.75 to <1.0 (1); 1.0 or greater (2) (new character) [ordered].

287. Ulna, morphology of olecranon process: narrow and subangular (0); wide and rounded (1) (Bro-chu, 1999).

288. Radius : humerus length, ratio: <0.6 (0); 0.6 to <0.75 (1); 0.75 or greater (2) (new character) [ordered].

289. Radius : tibia length, ratio: <0.6 (0); 0.6 to <0.7 (1); 0.7 or greater (2) (new character) [ordered].

290. Radiale, proximal end, shape: expanded symmetrically, similar to distal end (0); more expanded laterally than medially ( hatchet shaped') (1) (Buscalioni & Sanz, 1988).

291. Ilium, anterior (preacetabular) process, length relative to posterior (postacetabular) process: >75% of the length of the posterior process (0); 75% or less of the length of the posterior process (1); completely absent (2) (Clark, 1994) (changed similar in length' to >75% of the length of the posterior process' in character state 0; changed one-quarter' to 75%' in character state 1) [ordered].

292. Ilium, development of the posterior (postac-etabular) process: well-developed as a distinct process that extends anteroposteriorly for 60% or more of the acetabular length (0); extremely reduced or absent, extending anteroposteriorly <60% of the acetabular length (1) (Pol et al., 2012) (character state 1 changed to <60%' to remove gap between 50 and 60%).

293. Ilium, posterior end of the postacetabular process: tapering posteriorly to an acute tip (0); subrectangular with a vertically orientated posterior margin (1) (Pol et al., 2012) (removed with its dorsoventral height being at least 60% of the height at the origin of the postacetabular process' from character state 1).

294. Pubis, shape: rod-like without expanded distal end (0); with anterodorsally-posteroventrally expanded distal end (1) (Clark, 1994) (added anterodorsally-posteroventrally' to character state 1).

295. Pubis, anterior process: absent (0); present (1) (Clark, 1994).

296. Femur, proximal development of greater tro-chanter: prominent, ridge-like lateral border that separates the lateral surface of the proximal femur from a flat posterior surface reaching down to the level of the fourth trochanter (0); proximodistally short trochanteric surface lacking a distinct ridge, terminating well above the fourth trochanter (1) (Pol et al., 2012).

297. Femur, femoral head: mediolaterally flattened (0); hemispherical (1) (new character).

298. Tibia, distal projection of articular surfaces: medial region of distal articular surface extends further distally than the lateral region,

forming a strongly oblique distal margin of the tibia (0); medial and lateral regions subequally extended, with distal margin subhorizontally orientated (1) (Pol et al., 2012).

299. Tibia, posterior surface of shaft: flattened and confluent with fibula (0); twists posteriorly, leaving a void between the tibia and fibula (1) (new character).

300. Tibia : femur length, ratio: <0.9 (0); 0.9 to <1.0 (1); 1.0 or greater (2) (new character) [ordered].

301. Astragalus, anterior margin of the tibial facet: forming a well-defined ridge that reaches medially the ball-shaped region for the articulation of metatarsals I-II and closes the proximome-dial corner of the anterior hollow of the astragalus (0); forming a low ridge that is medially separated by a notch from the ball-shaped region for the articulation of the metatarsals III, failing to close the proximomedial corner of the anterior hollow (1) (Pol et al., 2012).

302. Distal tarsals, digits 2-4, dorsal surface: longitudinally grooved (0); smooth and flat (1) (new character).

303. Metatarsals I-IV: equidimensional (0); metatarsal I shorter than metatarsals II-IV (1) (new character).

Osteoderm characters

304. Osteoderms, dorsal surface: entirely sculpted (0); partially or completely unsculpted (1) (new character).

305. Presacral armour: cervical and dorsal trunk shields undifferentiated, morphology grading continuously (0); cervical shields clearly differentiated from dorsal trunk shields by size and general morphology (regardless of contact between nuchal and trunk series) (1); anterior-most cervical osteoderms developed into distinct shield (2) (Andrade et al., 2011) (character state 2 added) [ordered].

306. Nuchal osteoderms: consistent morphology along series (0); vary substantially in size in a random fashion (1); systematically increase in size posteriorly (2) (new character).

307. Nuchal osteoderms, with size variation: nuchals no less than half of the size of dorsal osteoderms (0); some smaller than one half of the size of the dorsal osteoderms (1) (new character).

308. Dorsal osteoderms, shape: rounded or ovate (0); subrectangular (mediolaterally wider than anteroposteriorly long) (1); subtriangular (2); square (3) (Clark, 1994) (character state 2 added).

309. Dorsal osteoderms, articular anterior process: absent (0); present (1) (Clark, 1994).

310. Dorsal osteoderms, articular anterior process: as discrete convexity on anterior margin (0);

well-developed process located anterolaterally ('peg and socket' articulation) (1) (Clark, 1994).

311. Dorsal osteoderms, anteroposterior keel on 320. anterior part of dorsal surface: absent (0); present (1) (new character).

312. Dorsal osteoderms, anteroposterior keel on pos- 321. terior part of dorsal surface: absent (0); present

(1) (new character).

313. Dorsal and cervical osteoderms: some or all 322 imbricated (0); not in contact (1) (new character).

314. Dorsal osteoderms, sutured anterior and posterior contacts: present (0); absent (1) (new char- 323. acter).

315. Dorsal primary osteoderms (sensu Frey, 1988), 324. rows: two parallel rows (0); four rows or more

(1) (Clark, 1994). 325

316. Dorsal osteoderms, accessory osteoderms (sensu Frey, 1988; i.e. osteoderms not forming part of

the dorsal shield): absent (0); present (1) 326 (Turner & Sertich, 2010).

317. Dorsal osteoderms, dorsal keel: same morphology in anterior-most dorsal osteoderms as remainder of dorsal series (0); keel shifts laterally in more posterior dorsal osteoderms (1) 327 (new character).

318. Dorsal osteoderms, anterior edge of dorsal sur- 328 face (i.e. articular surface, if present): sculpted, undifferentiated from main osteoderm body (0); 329 unsculpted (1) (new character).

319. Dorsal osteoderms, outline in dorsal aspect (excluding peg articulation): symmetrical about

anteroposterior axis (0); asymmetrical (1) (new character).

Dorsal osteoderms, mediolateral contacts: contact but not sutured (0); sutured (1) (new character).

Dorsal osteoderms, ventral to dorsal vertebrae beneath trunk: absent (0); present (1) (Clark, 1994).

Caudal osteoderms: absent (0) present on dorsal surface only (1); completely surrounding tail (2) (Clark, 1994) [ordered]. Caudal osteoderms: ovate (0); subcircular (1); subrectangular (2) (new character). Caudal osteoderms, bearing anteroposterior ridge: present (0); absent (1) (new character). Caudal osteoderms, anteroposterior ridge: present medially (0); forms a distinct lateral step in posterior-most elements (1) (new character). Caudal osteoderms, geometry: continuous from short to elongate oval (0); continuous from sub-rectangular (rounded corners) to suboval (1); isometric (equal geometry along series) (2) (new character).

Caudal osteoderms, medial and lateral edges: serrated (0); smooth (1) (new character). Caudal osteoderms, secondary osteoderms: present (0); absent (1) (new character). Caudal osteoderms, anteroposterior ridges: same morphology along series (0); becoming more pronounced posteriorly, coincident with a decrease in osteoderm size (1) (new character).