Historical Biology, 2014 http://dx.doi.org/10.1080/08912963.2014.918968

A new blunt-snouted dyrosaurid, Anthracosuchus balrogus gen. et sp. nov. (, ), from the Palaeocene of Alexander K. Hastingsa,b*, Jonathan I. Blocha and Carlos A. Jaramilloc aFlorida Museum of Natural History, University of Florida, Gainesville, FL, USA; bGeiseltalmuseum, Zentralmagazin Naturwissenschaftlicher Sammlungen, Martin Luther Universita¨t Halle-Wittenberg, Halle (Saale), ; cSmithsonian Tropical Research Institute, Balboa-Ancon, Panama (Received 23 March 2014; accepted 24 April 2014)

A new exceptionally brevirostrine dyrosaurid is described from the middle Palaeocene (58–60 million ago) Cerrejo´n Formation, northeastern Colombia, based on four partial and associated postcrania. This taxon is unique among dyrosaurids not only in shape, but also in having orbital tuberosities, and that are dorsoventrally thick and unpitted, a trait otherwise unknown in Crocodylomorpha. Results from a cladistic analysis of suggest that the new taxon, together with –Palaeocene Chenanisuchus lateroculi from and Cerrejonisuchus improcerus also from the Cerrejo´n Formation, are the most basal members of the family. Results from a biogeographic analysis indicate at least three independent dispersals of dyrosaurids from Africa to the New World occurred in the or early Palaeocene. Widely set orbits in the new taxon indicate a deviation from surface-based predation, characteristic of other dyrosaurids, to sub-surface predation, as in modern Gavialis. impressions found on turtle shells recovered from the same locality match well with teeth of the new taxon indicating possible predation.

http://www.zoobank.org/urn:lsid:zoobank.org:pub:AB2B24A5-27CC-4D3F-B580-F11F17851CE6 Keywords: Dyrosauridae; Crocodylomorpha; Palaeocene; Colombia; Cerrejo´n; biogeography

Introduction dyrosaurids that expand our understanding of the ecological Crocodylomorpha has long been considered a ‘living ’ diversity of the group, with a previously published dwarf group (McGregor 2005). This largely stems from the fact that (Cerrejonisuchus improcerus;Hastingsetal.2010) they have displayed conservative morphology in maintain- residing in freshwater habitats in adulthood, and a large- ing at least some members within the same body plan as bodied longirostrine taxon (Acherontisuchus guajiraensis; extant crocodylians since shortly after the group arose (Ross Hastings et al. 2011) also in freshwater sediments deposited 1989). However, numerous instances of incredibly varied in a large river system that drained into what is now the morphology have evolved time and again within the Caribbean Sea. crocodylomorph lineage. Forms have varied from slender Dyrosauridae originated and diversified during the late terrestrial ancestors (e.g. ; Walker 1990), to Cretaceous, somehow survived the K–P event heavily armoured forms (Armadillosuchus; Marinho and that wiped out other large-bodied marine (Benson Carvalho 2009) to fully marine with paddled feet and short et al. 2010) and ultimately became more diverse in the early Downloaded by [University of Florida] at 08:16 26 May 2014 powerful jaws skulls (; Gasparini et al. 2006). Cenozoic (Jouve, Bardet, et al. 2008). The ability of Most dyrosaurids are thought to have been relatively limited dyrosaurids to inhabit both marine and may in ecological diversity, characterised as large near-shore have contributed to their preferential survival (Jouve, marine longirostrine piscivores with some variation Bardet, et al. 2008). Isotopic values for carbon and oxygen associated with juveniles possibly inhabiting freshwater may provide empirical evidence for a freshwater habitat, environments (Jouve, Bardet, et al. 2008). Freshwater supporting this hypothesis (Wheatley 2010). Described deposits (Jaramillo et al. 2007; Figure 1) of the middle here is a new and species of Dyrosauridae with a very Palaeocene-aged (58–60 million years ago) Cerrejo´n different skull shape relative to other dyrosaurids, likely Formation in northeastern Colombia have yielded not only reflecting a unique associated with the invasion of plant documenting the earliest evidence of a the freshwater ecosystem in tropical . neotropical rainforest (Wing et al. 2009), but also vertebrate The first skull of this new dyrosaurid (referred fossils including side-necked turtles (Cadena et al. 2010; specimen UF/IGM 69) was discovered in 2005, but the Cadena, Bloch, et al. 2012; Cadena, Ksepka, et al. 2012), a skull was missing its anterior end. Preservation of the giant boid snake (Head et al. 2009a, 2009b) and a diversity of posterior external nares indicated that the individual likely

*Corresponding author. Email: [email protected]

q 2014 Taylor & Francis 2 A. K. Hastings et al.

has been assigned a unique number using the abbreviation UF/IGM. Casts of these specimens are also deposited at the collections of the Florida Museum of Natural History (UF).

Institutional abbreviations IGM, Colombian Geological Survey, Bogota´, Colombia; NJSM, State Museum, Trenton, NJ, USA; UF, University of Florida, Florida Museum of Natural History, Vertebrate Palaeontology Collection, Gainesville, FL, USA; USGS SAP, US Geological Survey – Saudi Arabian collection.

Terminology and anatomical abbreviations Teeth and alveoli will be referred to by number with 1 being the most anterior. Premaxillary teeth and alveoli will have the initial ‘pm’, maxillary teeth and alveoli will use ‘m’, and dentary teeth and alveoli will use ‘d’. For example, the first tooth or alveolus of the premaxilla will be referred to as ‘pm1’ and the second tooth or alveolus of the will be referred to as ‘m2’. These follow the abbreviations of Hastings et al. (2010).

Systematic palaeontology Crocodylomorpha Walker, 1970 Hay, 1930 Mesoeucrocodylia Whetstone and Whybrow, 1983 Dyrosauridae de Stefano, 1903 Anthracosuchus balrogus gen. et sp. nov.

Etymology. Anthraco-, meaning coal, referring to the coal mine in which the type locality is located; -suchus,

Downloaded by [University of Florida] at 08:16 26 May 2014 meaning . The specific term, balrogus, is named from a literary beast discovered within a deep mine, Figure 1. (Colour online) Locality map and stratigraphic originally written by Tolkien (1954). column for locality where all known Anthracosuchus balrogus Diagnosis. Possesses the following apomorphies: (1) a gen. et sp. nov. fossils are found. (A) Northern South America snout that is 44–53% of overall skull length; (2) rugose with star indicating Cerrejo´n fossil locality in northeastern Colombia. (B) Stratigraphic column of Cerrejo´n Formation (from orbital tuberosities anterior to medial margin of orbits; (3) Jaramillo et al. 2007). Fossil locality of Anthracosuchus is osteoderms that are dorsoventrally thick, unpitted and not underclay below Coal Seam 90, indicated with arrow above. imbricated; (4) premaxillae that are wider than long and had a short snout. A second skull with the anterior snout (5) external nares that are wider than long. preserved was discovered in early 2007, and indeed this Further differs from all other dyrosaurids in the skull exhibited a very short snout while maintaining other following unique combination of characters: (1) posterior diagnostic characters of Dyrosauridae. Two more skulls parietal margin straight; (2) interfenestral bar wide, were discovered later in 2007 bringing the minimum contributes to .10% of the skull table width, as measured number of individuals discovered thus far to 4. at anteroposterior midpoint of supratemporal fenestrae; (3) All specimens of the new taxon are deposited at the 8–10 maxillary teeth, of which 5–7 are antorbital; (4) Museo Geolo´gico Jose´ RoyoyGo´mez, Colombian strong constriction at the premaxilla–maxilla suture; (5) Geological Survey, Bogota´, Colombia. Each specimen nasal–prefrontal and nasal–lacrimal sutures subequal in Historical Biology 3

length; (6) orbits laterally placed and (7) teeth that are Comparative description robust, round and weakly striated. General. The rostrum is very stout and short in all Holotype. UF/IGM 67, nearly complete skull from specimens of Anthracosuchus, representing between 44% premaxilla to occipital condyle (missing medial portion of and 53% of the total skull length (Table 1). The rostrum of pterygoids and ventral braincase), nine ribs, seven Anthracosuchus is much shorter than that of any other vertebrae and unidentifiable bone. known dyrosaurid (Table 1). Cerrejonisuchus and Paratype. UF/IGM 68, skull from partial premaxilla to Chenanisuchus have much more longirostrine skulls occipital condyle (missing pterygoids and ventral brain- (Jouve, Bouya, et al. 2005; Hastings et al. 2010). The case), right articular, five osteoderms, five vertebrae, eight anterior rostrum of Anthracosuchus is much wider than ribs, distal pubis, distal ischium, haemal arch, three that of the other short-snouted dyrosaurids, even in the isolated teeth, proximal phalanx, sacral rib and unidentifi- case of UF/IGM 68 that has a snout only 2.5% shorter than able bone. Cerrejonisuchus (Hastings et al. 2010). The premaxilla Referred specimens. UF/IGM 69, skull from peri- and external nares of Anthracosuchus are both wider than orbital snout to posterior parietal (missing all but a long (Figures 2 and 3), a very atypical character for fragment of the pterygoids and the ventral braincase). UF/ Dyrosauridae. Even the relatively short-snouted forms of IGM 70, skull from premaxilla to quadratic condyles Cerrejonisuchus and Chenanisuchus have anteroposter- (missing pterygoids and ventral braincase). iorly elongate premaxillae and external nares (Jouve, Type locality and horizon. The holotype and all Bouya, et al. 2005; Hastings et al. 2010). The overall shape of the skull is more square than known specimens of Anthracosuchus balrogus were triangular, which is most apparent in the holotype recovered from a single stratigraphic layer, broadly (Figure 2) and the paratype (Figure 3). The holotype exposed within the Cerrejo´n coal mine, Guajira Depart- skull has experienced dorsoventral compression, but the ment, northeastern Colombia. The fossils were discovered overall shape is similar in the uncompressed paratype, in the underlying clay layer below Coal Seam 90 indicating that compression did not significantly alter (Figure 1). The geologic setting is of fine-grained, grey these features. The skull table is flat and broad, comprising clay and mud with fragmented lignite interspersed most of the posterior half of the skull in dorsal view. throughout (Jaramillo et al. 2007). The geologic setting Again, this general feature does not appear to have been is indicative of a large fluvial floodplain, likely supporting significantly altered by compression due to its presence in abundant large plant life. The age of the layer is middle- the uncompressed paratype (Figure 3). The wide, large and late Palaeocene, 58–60 million years old (Jaramillo et al. flat skull table of Anthracosuchus (Figure 2) is similar to 2007). that of Phosphatosaurus (Buffetaut 1978a), Sokotosuchus

Table 1. Snout proportions of all members of Dyrosauridae with material complete enough for skull length estimation.

DL (cm) PreoL (cm) R (PreoL/DL) £ 100 TBL (m) Anthracosuchus balrogus (UF/IGM 67) 62 27.5 44.35 4.58 Downloaded by [University of Florida] at 08:16 26 May 2014 Anthracosuchus balrogus (UF/IGM 68) 68 35.5 52.21 5.05 Cerrejonisuchus improcerus (UF/IGM 31)a 31.4 17.2 54.78 1.72–2.22 Cerrejonisuchus improcerus (UF/IGM 29)a 25.8 15.3 59.30 1.22–1.79 Chenanisuchus lateroculib 57.6 36.5 63.37 3.57–4.24 Acherontisuchus guajiraensis (UF/IGM 35)c 77.5–86.3 43.2/52.1 55.74–60.37 5.04–6.46 Acherontisuchus guajiraensis (UF/IGM 34)c 72.4–78.7 41.9/50.8 57.87–64.52 4.66–5.87 Congosaurus bequaerti d 63 41 65.08 3.97–4.66 Guarinisuchus munizi e 47.1 30.8 65.41 2.79–3.43 H. rogersiib 42.9 28.2 65.73 2.48–3.11 Sokotosuchus ianwilsonib 60.1 39.7 66.06 3.75–4.44 P. gavialoidesb 107 72 67.29 7.22–8.05 Arambourgisuchus khouribgaensisf 100 71.5 71.50 6.71–7.51 phosphaticusb 104 75 72.12 7.00–7.82 Dyrosaurus maghribensisg 89–97 – 73.00 5.89–7.28 R. keiniensisb 73.1 53.4 73.05 4.72–5.44 Atlantosuchus coupatezih 104.5 83 79.43 7.04–7.86

Notes: DL, dorsal skull length; PreoL, preorbital skull length; R, ratio of preorbital skull length to dorsal skull length; TBL, estimated total body length using method by Sereno et al. (2001). Citations are marked by superscript alphabets: a, Hastings et al. (2010); b, Jouve, Bouya, et al. (2005); c, Hastings et al. (2011); d, an estimation from Jouve and Schwarz (2004); e, an estimation from Figure 2 in Barbosa et al. (2008); f, Jouve, Iaroche`ne, et al. (2005); g, Jouve et al. (2006); h, Jouve, Bouya, et al. (2008). 4 A. K. Hastings et al.

Figure 2. (Colour online) Holotype (UF/IGM 67) of Anthracosuchus balrogus gen. et sp. nov. from Cerrejo´n locality in northeastern Colombia, middle-late Palaeocene: (A, B) in dorsal view, (C, D) in ventral view and (C) insert displays an in situ tooth in lingual and anterior views; scale bar: 1 cm. bf, bone fragment; bo, basioccipital; bsp, basisphenoid; en, external nares; eo, exoccipital; ep, ectopterygoid; f, frontal; j, jugal; l, lacrimal; max, maxilla; m1, 4, 5 and 8, maxillary teeth/alveoli; n, nasal; or t, orbital tuberosity; ost, (displaced); o t, occipital tuberosities; p, parietal; pal, palatine; pm2–4, premaxillary teeth/alveoli; pmx, premaxilla; po, postorbital; prf, prefrontal; pt, pterygoid; q, quadrate; qj, quadratojugal; sq, squamosal; stf, supratemporal fenestra; vert, (displaced). Scale bar: 10 cm.

(Buffetaut 1979), Chenanisuchus (Jouve, Bouya, et al. not accompanied by surrounding unpitted surfaces as in 2005) and Cerrejonisuchus as opposed to the elongate and Anthracosuchus. These may have been keratinised, but narrow skull table present in Hyposaurinae. The posterior likely did not support large structures. margin of the skull table in dorsal view is transversally Cranial openings. The external nares are wider than straight in Anthracosuchus (Figure 2), but highly indented long in dorsal view and completely surrounded by the anteriorly in Rhabdognathus keiniensis (Jouve 2007), premaxillae with lateral margins that are ventral to both rogersii (Troxell 1925) and Atlantosuchus the anterior and posterior margins (Figures 2 and 3). The Downloaded by [University of Florida] at 08:16 26 May 2014 coupatezi (Jouve, Bouya, et al. 2008). external nares of all other dyrosaurids are anteroposter- Each of the skulls of Anthracosuchus possesses wide iorly elongate in dorsal view. No portion of the incisive and rugose tuberosities immediately anterior to the orbits. foramen is preserved on any specimen. The orbits of These orbital tuberosities are most pronounced in Anthracosuchus are widely placed from the midline, in a dorsolateral view in UF/IGM 69 (Figure 4) and appear to lateral position (Figures 2–4) like only one other be variable as to whether they are fully on the prefrontal, dyrosaurid, Chenanisuchus lateroculi from the Palaeocene lacrimal or split between the two (Figures 2 and 3). The of (Jouve, Bouya, et al. 2005). All other orbital tuberosities are dorsoventrally short and roughly dyrosaurids have much more medially positioned orbits. conform to the contour of the skull (Figure 4). The rugose The supratemporal fenestrae of Anthracosuchus are very tuberosities near the anteromedial margins of the orbits, large, much larger than the orbits and anteroposteriorly visible in dorsal view (Figure 4), are unusual for elongate (Figures 2–4). The supratemporal fenestrae are crocodyliforms. The only other known instance of generally wider in dorsal view (Figures 2 and 4) than in preorbital tuberosities within Crocodylomorpha is that of hyposaurine dyrosaurids, which have much more elongate Caryonosuchus pricei from the Late Cretaceous of , fenestrae. The infratemporal fenestrae of Anthracosuchus but these are more akin to small horns (Kellner et al. 2011). are laterally placed, with minimal dorsal exposure and are Alligatorids have highly rugose anteromedial orbital bound anteroventrally by the jugals and posterodorsally by margins, roughly where the palpebral attaches, but this is the quadratojugals (Figures 2–4). The infratemporal Historical Biology 5

Figure 3. (Colour online) Paratype (UF/IGM 68) of Anthracosuchus balrogus gen. et sp. nov. from Cerrejo´n locality in northeastern Colombia, middle-late Palaeocene: (A, B) skull in dorsal view, (C) associated tooth in lingual view, (D) associated tooth in posterior view and (E, F) skull in ventral view. bo, basioccipital; en, external nares; eo, exoccipital; eo a, articulation for exoccipital; ep, ectopterygoid; f, frontal; j, jugal; l, lacrimal; max, maxilla; m1–8, maxillary teeth/alveoli; n, nasal; orb, orbit; or t, orbital tuberosity; ot, occipital tuberosities; p, parietal; pal, palatine; pm2–4, premaxillary teeth/alveoli; pmx, premaxilla; po, postorbital; prf, prefrontal; pt, pterygoid; q, quadrate; qj, quadratojugal; qc, quadratic condyle; sq, squamosal; stf, supratemporal fenestra. Scale bar for A–B and E–F equals 10 cm. Scale bar for C–D equals 1 cm.

fenestrae are large in hyposaurine dyrosaurids, but more at the premaxilla–maxilla suture, noticeable in dorsal and transversally compressed and laterally placed in Anthra- ventral views (Figures 2 and 3). cosuchus, which is apparent in the uncompressed UF/IGM Maxilla. The maxilla makes broad contact with the 68 (visible in dorsal view, Figure 3). The suborbital premaxilla, nasal and lacrimal in Anthracosuchus along its fenestrae of Anthracosuchus are bordered by the palatines dorsal snout (Figures 2–4). The maxilla broadly contacts and maxillae and are anteriorly concave in ventral view the palatine on the palatal surface in ventral view Downloaded by [University of Florida] at 08:16 26 May 2014 (Figures 2–4). (Figures 2–4). The maxilla of Anthracosuchus comprises Premaxilla. The premaxillae are wider than long and a similar proportion of the snout to other dyrosaurids, even completely surround the external nares, excluding the if its overall shape is very different. Eight maxillary alveoli nasal from contact with the external nares (Figures 2 and are preserved in ventral view of UF/IGM 67 and 68 3). All dyrosaurids except Anthracosuchus possess (Figures 2 and 3). Aventral portion of the posterior maxilla premaxillae that are longer than wide. Anthracosuchus may be missing in all of the specimens, so a complete instead has wide and stout premaxillae and a constriction count is not certain (Figures 2–4). The first and third at the premaxilla–maxilla suture that is very pronounced maxillary alveoli appear to be the largest (Table 2). in both dorsal and ventral views (Figures 2 and 3), as in The fifth to seventh alveoli are smaller, but the eighth Phosphatosaurus (Buffetaut 1978a), and very different alveolus appears to be more robust. Even the posteromost from the gentle curvature seen in hyposaurines such as alveoli appear uncompressed and retain the rounded shape Dyrosaurus (Jouve 2005). The premaxilla of Anthraco- of the anterior alveoli. The tooth count is smaller than suchus possesses four alveoli, although they are not well most dyrosaurids. The tooth count of typical dyrosaurids preserved in any specimen (see ventral views of Figures 2 such as Dyrosaurus maghribensis is 19–21 (Jouve et al. and 3). The third premaxillary tooth is by far the largest, 2006), as opposed to 8 in Anthracosuchus and 11 in and the fourth premaxillary tooth appears to be severely Cerrejonisuchus (Hastings et al. 2010). Most dyrosaurids reduced. There is a strong indentation of the lateral margin have approximately homodont teeth, but the maxilla of 6 A. K. Hastings et al.

Figure 4. (Colour online) Referred skull (UF/IGM 69) of Anthracosuchus balrogus gen. et sp. nov. from Cerrejo´n locality in northeastern Colombia, middle-late Palaeocene: (A, B) skull in dorsal view, (C, D) skull in ventral view and (E, F) oblique view of right orbital tuberosity. f, frontal; j, jugal; l, lacrimal; max, maxilla; m2, 5 and 6, maxillary teeth/alveoli; n, nasal; orb, orbit; or t, orbital tuberosity; p, parietal; pal, palatine; pm2 and 3, premaxillary teeth/alveoli; pmx, premaxilla; po, postorbital; prf, prefrontal; sq, squamosal. Scale bar: 10 cm.

Anthracosuchus, Cerrejonisuchus and Phosphatosaurus Bouya, et al. 2008), Rhabdognathus aslerensis (Jouve Downloaded by [University of Florida] at 08:16 26 May 2014 (Buffetaut 1978a; Hastings et al. 2010) has notably 2007) and Sokotosuchus ianwilsoni (Buffetaut 1979) all enlarged third maxillary teeth. In Anthracosuchus, the have unfused nasals. The largest two specimens of maxillary margin is modestly sinuous (nonlinear) in dorsal Anthracosuchus have unfused nasals (UF/IGM 67 and view (Figures 2 and 3). The maxilla of Anthracosuchus is 68; Figures 2 and 3), while the smaller UF/IGM 69 has somewhat intermediate in terms of degree to which the fully fused nasals (Figure 4), visible in dorsal view. The lateral margin undulates, or is ‘festooned’, in dorsal view bone is broad and medially placed in Anthracosuchus, (Figures 2 and 3). In hyposaurines, the margin is linear, but expanding from a narrow tip within the premaxilla and in phosphatosaurines, it is much more sinuous (Buffetaut extending posteriorly into a rather straight margin with 1979). Anthracosuchus instead has only a slight expansion the frontal. The nasal does not reach the external nares posterior to the premaxilla–maxilla suture. in dorsal view. Nasal. The nasal is variably fused in Anthracosuchus Lacrimal. The lacrimal forms much of the anterior in dorsal view. The nasals are unfused in UF/IGM 67 margin of the orbit in dorsal view and is anteroposteriorly and 68 (Figures 2 and 3), but fully fused in UF/IGM 69 thin (Figures 2–4). The lacrimal–prefrontal contact (Figure 4). The nasal is fused in most dyrosaurids, appears to be greater than the lacrimal–nasal contact. including Dyrosaurus (Jouve 2005), Congosaurus (Jouve The length of contact is similar between the prefrontal– and Schwarz 2004) and adult specimens of Cerrejoni- nasal and the lacrimal–nasal sutures. The prefrontal– suchus. However, Atlantosuchus coupatezi (Jouve, nasal suture is longer than the lacrimal–nasal suture in Historical Biology 7

Table 2. Measurements of alveoli for Anthracosuchus balrogus gen. et sp. nov.

Left Right Alveolar Anteroposterior Medeolateral Alveolar Anteroposterior Medeolateral Specimen position length width position length width UF/IGM 67 pm1 – – pm1 – – pm2 17.98 18.27 pm2 17.52 18.65 pm3 25.92 27.01 pm3 27.58 25.04 pm4 – – pm4 – – m1 29.47 32.55 m1 – – m2 – – m2 – – m3 – – m3 – – m4 16.70 26.92 m4 – – m5 – – m5 11.71 11.69 m6 – – m6 – – m7 – – m7 – – m8 18.43 – m8 15.41 14.63 UF/IGM 68 pm1 – – pm1 – – pm2 – – pm2 – – pm3 14.49 22.83 pm3 – – pm4 – – pm4 – – m1 – – m1 20.21 19.26 m2 – – m2 – – m3 26.19 21.76 m3 – – m4 14.04 12.61 m4 12.88 12.52 m5 16.43 15.38 m5 – – m6 – – m6 21.49 20.56 m7 – – m7 – – m8 – – m8 – – UF/IGM 69 pm1 – – pm1 – – pm2 – 13.34 pm2 – – pm3 16.95 25.41 pm3 15.71 17.75 pm4 – – pm4 – – m1 – – m1 – – m2 – – m2 – – m3 – – m3 – – m4 – – m4 – – m5 – – m5 – – m6 – – m6 – – m7 – – m7 – – m8 – – m8 – –

Notes: Alveolar walls that were either not preserved at all or not preserved well enough for measurement are denoted with a ‘–’. All measurements are in millimetres. Abbreviations for the alveoli follow those of Hastings et al. (2010). Downloaded by [University of Florida] at 08:16 26 May 2014

Dyrosaurus (Jouve et al. 2006), but the reverse is true of suchus lateroculi (Jouve, Bouya, et al. 2005). Anthraco- Congosaurus (Jouve and Schwarz 2004). Anthracosuchus suchus is also most similar to Chenanisuchus lateroculi in appears to have an intermediate condition with roughly prefrontal contribution to interorbital width, which is equal proportions of contact to the nasal, in dorsal view broad in both taxa. (Figures 2 and 3). Jugal. The jugal makes up the lateral margin of Prefrontal. The prefrontal contributes largely to the the orbit, visible in dorsal view (Figures 2 and 3). The medial margin of the orbit in dorsal view as well as to the postorbital bar is missing or obscured in each of the overall interorbital width in Anthracosuchus (Figures 2– specimens. The jugal extends posteriorly to its contact 4). The prefrontal–frontal contact on the dorsal surface is with the quadratojugal in the posterolateral corner of the extensive and penetrates broadly into the frontal, rather infratemporal fenestra, visible in dorsal view (Figures 2 than being slender and reduced. The prefrontal–frontal and 3). The jugal of Anthracosuchus is thicker and more contact of Anthracosuchus in dorsal view is much longer robust than that of all hyposaurine dyrosaurids (Troxell and extensive than the reduced state of most dyrosaurids 1925; Jouve and Schwarz 2004; Jouve 2005, 2007; Jouve, such as Congosaurus (Jouve and Schwarz 2004) and Iaroche`ne, et al. 2005; Jouve et al. 2006; Jouve, Bouya, Dyrosaurus (Jouve 2005), and most similar to Chenani- et al. 2008; Barbosa et al. 2008). 8 A. K. Hastings et al.

Frontal. The frontal is roughly cruciform in dorsal Cerrejonisuchus improcerus and Chenanisuchus latero- view, with an anterior projection penetrated by the paired culi, the interfenestral bar is much more square shaped and nasals in UF/IGM 67, slightly penetrated in UF/IGM 68 robust, but still with a bit of an overhanging lip onto the and with a fairly linear anterior connection to the nasals in supratemporal fenestra (Jouve, Bouya, et al. 2005). The UF/IGM 69 (Figures 2–4). The frontal in most dyrosaurids straight posterior parietal margin of Anthracosuchus in comes to a point anteriorly, bifurcating the nasals, which dorsal view (Figures 2–4) is most similar to that of other can be found in Dyrosaurus maghribensis (Jouve et al. dyrosaurids such as Dyrosaurus (Jouve et al. 2006) and 2006) and Congosaurus bequaerti (Jouve and Schwarz Phosphatosaurus (Buffetaut 1978a), and differs from the 2004). However, Anthracosuchus has nasals that instead anteriorly indented margins of R. keiniensis (Jouve 2007) slightly bifurcate the frontal, in dorsal view, at least in the and H. rogersii (Troxell 1925). holotype (Figures 2). Alternatively, in Cerrejonisuchus Postorbital. The postorbital forms the anterolateral improcerus, the nasal terminates along a relatively straight portion of the supratemporal fenestrae and does not appear suture with the frontal, as in UF/IGM 69 (Figure 4). The to contact the parietal on the dorsal surface in any of the frontal contributes to the posteromedial margin of the orbit specimens (Figures 2–4). The anterolateral process is well and has a broad contact with the prefrontal. The left orbit developed in UF/IGM 68 (Figure 3) and UF/IGM 69 of UF/IGM 67 has been slightly modified by compression, (Figure 4), and modest in UF/IGM 67 (Figure 2). making the prefrontal appear to exclude the frontal from Dyrosaurids typically have well-developed anterolateral the orbit (Figure 2), but the better preserved right orbit postorbital processes, as in Dyrosaurus phosphaticus shows the same relationship as the other two skulls, and (Jouve 2005) but are reduced in Phosphatosaurus seems to more accurately reflect the character in gavialoides (Buffetaut 1978a) and practically non-existent Anthracosuchus. The frontal is slightly ornamented and in Chenanisuchus lateroculi (Jouve, Bouya, et al. 2005). forms a broad bar on the lateral side where it joins with the The ventral portion of the postorbital is not preserved well postorbital on the posterior margin of the orbit. The frontal enough in any Anthracosuchus specimen to determine contributes to about one-quarter of the overall dorsal whether it contacted the infratemporal fenestra anteriorly. length of the interfenestral bar in UF/IGM 67 (Figure 2) Squamosal. The squamosal forms the posterolateral and closer to one-half in UF/IGM 69 (Figure 4). portion of the supratemporal fenestrae in dorsal view Palatine. The palatine extends anteriorly on the palate (Figures 2–4). The squamosal and postorbital appear to in ventral view to the level of the m3 in the holotype (UF/ contribute roughly equally to the lateral bridges of the IGM 67; Figure 2) but only to the m6 in UF/IGM 68 skull table. The posterodorsal prong of the squamosal is (Figure 3). The suture of UF/IGM 69 appears to reach as most pronounced in UF/IGM 68 and extends posteriorly to far forward as m2, but the suture is difficult to identify in the level of the posterior end of the occipital tuberosities this specimen (Figure 4). The palatine forms the wide bar (Figure 3). However, the other two specimens have much extending posteriorly from the palate towards the brain- more reduced posterodorsal squamosal projections case and also forms the medial margin of the suborbital (Figures 2 and 4). The degree of posterior projection of fenestra in ventral view (Figures 2 and 3). The maxilla– the squamosals in Anthracosuchus (Figures 2–4) is most palatine suture occurs at the anteromedial corner of the similar to those of Dyrosaurus (Jouve et al. 2006), and not suborbital fenestra, with the palatine contributing very nearly as pronounced and elongate as R. aslerensis minimally to the anterior margin. The palatine of (Brochu et al. 2002; Jouve 2007). Downloaded by [University of Florida] at 08:16 26 May 2014 Dyrosaurus (Jouve et al. 2006) and Rhabdognathus Pterygoid. The pterygoid is flattened onto the brain- (Jouve 2007) meets the maxilla slightly more laterally case of UF/IGM 67, but clearly shows in ventral view a than in Anthracosuchus, coming to the anterior point of the well-developed wing extending laterally from the midline suborbital fenestra. No part of the choanal opening is (Figure 2). The pterygoidian wing of UF/IGM 67 is broad preserved in any of the described skulls. and flat (Figure 2), and its relative thickness may be similar Parietal. The parietal forms the largest contribution to to that of Dyrosaurus maghribensis (Jouve et al. 2006), but the skull table in dorsal view, which is ornamented with a preservation makes it difficult to discern. A fragment of shallow divot on the triangular section between the this wing is also preserved on the ventral surface of UF/ supratemporal fenestrae, posterior to the interfenestral bar IGM 68 (Figure 3). The suture with the ectopterygoid (Figures 2 and 3). The parietal contributes between half forms a distinct, posteriorly directed angle (Figure 2). and three-quarters to the dorsal length of the interfenestral Ectopterygoid. The ectopterygoid is firmly sutured to bar. The interfenestral bar is square in cross section, with a the pterygoid on the ventral surface and connects this bone slight overhang. The interfenestral bar of dyrosaurids is to the maxilla and jugal (Figure 2). The bone forms the typically T shaped and thin, as in Dyrosaurus phosphaticus posterolateral portion of the suborbital fenestra in ventral (Jouve 2005)andArambourgisuchus khouribgaensis view. The ectopterygoid of UF/IGM 69 (Figure 4) does not (Jouve, Iaroche`ne, et al. 2005) or forms a thin sagittal appear twisted as in Dyrosaurus maghribensis (Jouve et al. crest as in R. aslerensis (Jouve 2007). In Anthracosuchus, 2006) but is instead thick and linear. Historical Biology 9

Exoccipital. The exoccipital has wide lateral expansion crests on the ventral surface are understated in and constitutes much of the posterior surface of the Anthracosuchus, if present at all (Figure 2), much like braincase. Due to oblique deformation of this part of the that of Cerrejonisuchus improcerus, but differing from the skull, it is visible on the dorsal surface of UF/IGM 68 prominent crest in Arambourgisuchus khouribgaensis (Figure 3) and to a lesser degree on the ventral surface of (Jouve, Iaroche`ne, et al. 2005), R. aslerensis (Brochu UF/IGM 67 (Figure 2). The exoccipital has a large et al. 2002) and R. keiniensis (Jouve 2007). contribution of roughly one-third from each side to the Dentition. The in situ teeth of UF/IGM 67 are wide, occipital condyle, which can be observed in dorsal view in blunt and at least one tooth has a well-preserved carina that UF/IGM 68 (Figure 3). The occipital condyle of UF/IGM defines the labial and lingual surfaces (Figure 2). The 67 is largely obscured in dorsal view and difficult to associated, yet unattached teeth preserved with UF/IGM discern exoccipital contribution (Figure 2). The exoccipi- 68 also have striations and well-developed carinae tals of all dyrosaurids contribute largely to the occipital (Figure 3). All associated teeth are blunt and low crowned. condyle (Jouve et al. 2006) and that of Anthracosuchus is The teeth vary in size, but all are roughly circular with no exception. Well-developed, wide and flat occipital minimal labiolingual compression towards the posterior tuberosities extend from the occipital surface, just ventral alveoli (Figures 2 and 3). The dentition of Anthracosuchus to the skull table (Figures 2–4). The occipital tuberosities (Figures 2 and 3) is most similar to that described for are developed to varying degrees within Dyrosauridae. Phosphatosaurus (Buffetaut 1978a) in being round and Anthracosuchus is similar to Chenanisuchus and Sokoto- blunt, and differs from the elongate recurved teeth typical suchus in having wide and flat tuberosities (Figure 2), but of Hyposaurus (Denton et al. 1997), Dyrosaurus (Jouve differs from Rhabdognathus and Hyposaurus that have et al. 2006) and the anterior dentition of Acherontisuchus. elongate and narrow tuberosities (Jouve, Bouya, et al. The teeth of Anthracosuchus are more low crowned and 2005). not spade shaped as in the posterior dentition of Basioccipital. The basioccipital forms most of the Cerrejonisuchus. wide occipital condyle, as seen in both dorsal and ventral Articular. An isolated articular (but no other portion of views (Figures 2–4). It has an oblong, smooth and rounded the mandible) was recovered with UF/IGM 68 (Figure 5). surface with a strong ventral lip where the surface recurves The articular surface where it would fuse with the dorsally, best seen in ventral view. The basioccipital of surangular is preserved well, but much of the retroarticular Anthracosuchus differs from that of Cerrejonisuchus in process is missing. Although the surangular is missing, due being less arched dorsally and lacking the wide flat to the location of articular–surangular suture marks along anteroventrally directed tuberosity on the ventral surface the lateral margin of the articular, the surangular likely did (Figure 2). No portion of the basioccipital tubera is contribute to the articular surface. The lateral shelf of the preserved in any specimen. retroarticular process is well developed in dorsal view and Basisphenoid. A small portion of the basisphenoid is the glenoid fossa is smooth and deeply curved. The lateral preserved in UF/IGM 67, but lacks any part of the shelf of the retroarticular process forms an L shape in eustachian foramen in ventral view (Figure 2). It is situated dyrosaurids (Jouve, Bouya, et al. 2005) and is well between the pterygoid and the exoccipitals and makes at represented in UF/IGM 68 (Figure 5). The anteromedial least some contact with the basioccipital. The basi- wing described for Dyrosaurus maghribensis (Jouve et al. sphenoid is not well preserved in Anthracosuchus, but does 2006) and Congosaurus bequaerti (Jouve and Schwarz Downloaded by [University of Florida] at 08:16 26 May 2014 appear to be mediolaterally narrow in ventral view 2004) is also present in Anthracosuchus, although much (Figure 2) as that of Dyrosaurus maghribensis (Jouve et al. more rounded and robust (Figure 5). 2006). Dorsal vertebrae. A total of seven dorsal vertebrae Quadratojugal. The quadratojugal reaches the postero- were preserved with UF/IGM 67 (Figure 6). These appear most portion of the skull in dorsal and ventral views and to be more or less from a continuous series from dorsal 4 to contributes at least one-quarter to the craniomandibular dorsal 9, as well as a more posterior vertebra from around joint (Figures 2 and 3), a similar capacity as Dyrosaurus the position of dorsal 16. Even the anteromost of these maghribensis (Jouve et al. 2006). This bone forms the vertebrae have fully fused neurocentral sutures, indicating posteromedial margin of the infratemporal fenestra, in maturity (sensu Brochu 1996). The neurocentral sutures of dorsal view, and extends anteriorly to the postorbital, Anthracosuchus were likely straight prior to compression, ventral to the skull table (Figures 2 and 3). as in Hyposaurinae (Schwarz et al. 2006). Quadrate. The quadrate forms approximately three- The hypapophysis of the dorsal 4 is robust, short and quarters of the craniomandibular condyle as well as much the vertebra lacks a parapophysis, indicating a transition to of the braincase, seen in both dorsal and ventral views of the rib cage (Figure 6). Dorsals 6–9 show clear attachment UF/IGM 67 and 68 (Figures 2 and 3). The quadrate points for ribs on the transverse processes, indicating extends dorsally to at least the squamosal, but possibly the bifurcated attachment surfaces. The dorsal 4 vertebra of postorbital in dorsal view (Figures 2 and 3). Quadratic Anthracosuchus bears a hypapophysis that is roughly half 10 A. K. Hastings et al.

as high as the vertebral body, as described for Hyposaurinae (Schwarz et al. 2006). Dorsals 6–9 may have possessed the ventral keel on the centrum described for Congosaurus (Schwarz et al. 2006), but preservation makes this difficult to discern. The transverse processes of Anthracosuchus dorsals 4 and 6–9 have faintly apparent parapophyses and diapophyses, making this taxon more similar to Dyrosaurus, Hyposaurus and Congosaurus than to cf. Rhabdognathus, which has divergent parapophyses and diapophyses until dorsal 5 (Schwarz et al. 2006). The dorsal 16 vertebra shows no articular surface for ribs and likely represents the lumbar region of the , posterior to the rib cage (Figure 6). These vertebrae are all wider, larger and more robust (Figure 6) than any described thus far for Dyrosauridae (Schwarz et al. 2006). Vertebral series are best known for Dyrosaurus, but have also been described for Rhabdog- nathus (Langston 1995) and Congosaurus (Jouve and Schwarz 2004). The neural spines are more mediolaterally thick and anteroposteriorly wide than those of other dyrosaurids, but not nearly as expanded as those of (Sereno et al. 2001). Compression has made it difficult to tell whether the transverse processes widen posteriorly as in Hyposaurinae, but they do appear to lengthen laterally (Schwarz et al. 2006). From what can be determined from incomplete specimens, the neural spines do appear to be of similar length, as in Hyposaurinae (Schwarz et al. 2006). Caudal vertebrae. Three caudal vertebrae were found in association with UF/IGM 68 (Figure 6) and represent positions between Caudal 28 and 32. One of these specimens has the entire neural spine preserved, which is very long and anteroposteriorly narrow in lateral view (Figure 6). The distal tip of the neural spine does not appear swollen or enlarged relative to the rest of the shaft. The caudal vertebrae from this section of the tail typically have a swelling at the distal tip in Dyrosaurus and Congosaurus, while UF/IGM 68 does not (Figure 6). The Downloaded by [University of Florida] at 08:16 26 May 2014 anterior and posterior centra of the caudal vertebrae are quadrangular in cross section as well as their articular facets (Figure 6) and appear very similar to those described for Dyrosaurus and Congosaurus (Schwarz et al. 2006). Cervical ribs. Two cervical ribs were discovered with UF/IGM 68 (Figure 7) and represent positions around articulation with cervicals 4–6. Although compressed, the widely divergent tuberculum and capitulum, best seen in medial view (Figure 7), indicate a relatively vertical position of the cervical ribs in this portion of the neck. The lateral surface is smooth, with an anterior surface that likely imbricated with the adjacent anterior cervical rib. Figure 5. (Colour online) Associated articular bone of UF/IGM The internal surface of the cervical ribs forms a concave 68: (A, B) in medial view, (C, D) in lateral view and (E, F) in trough in Anthracosuchus (Figure 6) and in all hyposaurine dorsal view. as sa, articular–surangular surface of articulation; dyrosaurids (Schwarz et al. 2006). gf, glenoid fossa; lsrap, lateral shelf of retroarticular process; rap, Dorsal ribs. Most ribs associated with UF/IGM 67 and retroarticular process. Scale bar equals 10 cm. 68 are poorly preserved and only represent partial shafts Historical Biology 11

Figure 6. (Colour online) Vertebrae associated with skulls of Anthracosuchus balrogus gen. et sp. nov. (A, B) Three articulated dorsal vertebrae associated with the skull of UF/IGM 67, likely positions 6 to 8. (C, D) Dorsal vertebra associated with the skull of UF/IGM 67, likely position 16. (E, F) Dorsal vertebra associated with the skull of UF/IGM 67, likely position 4. (G, H) Anterior view of caudal vertebra associated with the skull of UF/IGM 68, likely position 28. (I) Same as (G), but in lateral view. (J) Lateral view of caudal vertebra associated with the skull of UF/IGM 68, likely position 30. cen, centrum; hyp, hypopophysis; it, isolated tooth; nc, neural canal; ncs, traces of neurocentral suture; ns, neural spine; przy, prezygapophyses; pozy, postzygapophyses; tp, transverse process. Scale bar for A–F: 10 cm; scale bar for G–J: 5 cm.

with little to no discernible morphology. Two large dorsal articulate with any of the three preserved caudal vertebrae. ribs were well preserved with UF/IGM 67 though and are The left and right articular surfaces do not appear as fully shown in Figure 7. These are large and robust, much like fused in Anthracosuchus (Figure 7)asinCongosaurus the dorsal vertebrae, with thick dorsoventral shafts. These bequaerti (Jouve and Schwarz 2004) and seem most have clearly distinct tubercula and capitula, indicating similar to the unfused condition described for Rhabdog- Downloaded by [University of Florida] at 08:16 26 May 2014 positions around dorsals 7 and 8. The dorsal ribs of nathus sp. (Langston 1995). Anthracosuchus (Figure 7) appear more thick and robust Sacral rib. The second right sacral rib is preserved with than those figured for either Dyrosaurus or Congosaurus UF/IGM 68 (Figure 8). The iliac surface flares poster- (Schwarz et al. 2006). Anthracosuchus has dorsoventrally olaterally and appears to receive an anteroposteriorly tall ribs, as in all other described Dyrosauridae, and oriented ridge towards the posterior end of the pelvis. UF/ both Congosaurus and Anthracosuchus have thickened IGM 68 exhibits the roughly hourglass shape in ventral distal tips. view (Figure 8) typical of Dyrosauridae (Schwarz et al. Haemal arch. The single-preserved haemal arch 2006). As in Rhabdognathus sp. (USGS SAP 37-CR-1), associated with a skull (UF/IGM 68) is from the mid- the medial surface articulates with the sacral vertebra, but caudal section, with a likely position between the 11th and does not participate in the centrum (Langston 1995). 15th caudal vertebrae (Figure 7). The vertebral attachment Ischium. The left distal ischium is preserved with UF/ points are split between anterior (13.6 mm long) and IGM 68 (Figure 8). The anterior and posterior margins are posterior facets (9.1 mm long), indicating its position in well preserved and do not indicate flaring in either the tail section. The bone is 73.7 mm long, is constricted in direction, but instead a relatively narrow termination. The lateral view ventral to the haemal canal (10.5 mm wide) bone is also very dorsoventrally straight, with little to no and flares ventrally forming a spatulate shape (max ventral curvature laterally. A modest, laterally deflected crest width of 16.7 mm). The haemal arch does not appear to extends from the anterior distal edge, which is slightly 12 A. K. Hastings et al.

complete, possessing a more narrow and pointed distal extremity than the anteroposteriorly long ventral margin of Dyrosaurus maghribensis (Jouve et al. 2006). Pubis. The right distal pubis was preserved with UF/ IGM 68 (Figure 8). The entire symphyseal surface that would have joined the other pubis is preserved and reflects an angle between the pubes at around 408, narrower than that seen in H. rogersii of (,508; Troxell 1925). However, the overall distal pubic shape is still much more similar between Anthracosuchus and Cretaceous Hyposaurus of New Jersey than either is to Palaeocene– Dyrosaurus or cf. Rhabdognathus of Africa (Langston 1995). The pubis of Anthracosuchus is remarkably flat, with very little arching. Phalanx. A proximal phalanx was preserved with UF/ IGM 68 (Figure 9), likely pertaining to the pes, due to its elongate form. The dorsal surface is tubular with a slightly flattened lateral surface. The ventral (plantar) side has a wide longitudinal groove along the bone’s centreline. The articular facet is roughly ovular, with a width of 24.8 mm and height of 15.7 mm. The proximal articular surface of the phalanx of Anthracosuchus (Figure 9) appears more ovular than does the triangular shape described for hyposaurine dyrosaurids (Schwarz et al. 2006). The preserved portion is 57.1 mm long and lacks any part of the distal condyle. Osteoderms. Five osteoderms were found associated with the skull of UF/IGM 68. The osteoderms of UF/IGM 68 are very thick and completely unpitted (Figure 10). Both dorsal and ventral surfaces are smooth, with faint grooves radiating from the centre, particularly evident around the edges. The osteoderms swell to the thickest at the middle and are not curved upward, but are either flat ventrally or swell ventrally as well as dorsally. None of the preserved osteoderms have any indication of an imbricat- ing surface. These osteoderms are very different from any Figure 7. (Colour online) Ribs and haemal arch associated with other published dyrosaurid osteoderms (see Schwarz et al. skulls of Anthracosuchus balrogus gen. et sp. nov. (A) Lateral 2006). Downloaded by [University of Florida] at 08:16 26 May 2014 view of dorsal ribs associated with UF/IGM 67, likely positions 7 and 8. (B) Medial view of cervical rib associated with UF/IGM Dyrosaurids typically have thin, slightly dorsally 68, likely between positions 4 and 6. (C) Same as in (B), in lateral curved osteoderms with wide shallow pitting. Pitting is view. (D) Posterior view of haemal arch associated with UF/IGM consistent on dorsal, accessory and gastral osteoderms. 68, likely positions between 11 and 15. (E) Same as in (D), in Pitting is unlikely to have been lost in Anthracosuchus due lateral view. capm, capitulum; hc, haemal canal; inc ct, incisura capitulotubercularis (capitulotubercular incision); p art f, to taphonomic weathering as more subtle features such as posterior articular facet; tubm, tuberculum. cranial ornamentation were preserved with the same specimen. Moreover, smaller Cerrejonisuchus osteoderms from the same locality have been preserved with pitting laterally offset from the anterior margin, but then curves intact as well as a more typically dyrosaurid thin cross slightly medially to form the anterior edge. The well- section (Hastings et al. 2010). Hyposaurus rogersii is very preserved narrow distal end of the ischium of UF/IGM 68 typical for a dyrosaurid in this respect and a representative (Figure 8) is consistent with the preserved partial ischium osteoderm is shown in Figure 10. Osteoderms may be a of Acherontisuchus (Hastings et al. 2011) and interpret- little thicker in Congosaurus than other dyrosaurids (Jouve ations of H. rogersii (Troxell 1925), and less consistent and Schwarz 2004; Schwarz et al. 2006) but cross- with Dyrosaurus (Jouve et al. 2006). While known ischia sectional diagrams of these in Schwarz-Wings et al. (2009) of Acherontisuchus and Hyposaurus are incomplete, the still appear much thinner than those of Anthracosuchus. distal portion of the ischium of Anthracosuchus is Many dyrosaurid osteoderms are imbricated and have a Historical Biology 13

Figure 8. (Colour online) Portions of the pelvis of Anthracosuchus balrogus gen. et sp. nov., UF/IGM 68. (A) Distal left ischium in lateral view. (B) Distal left ischium in posterior view. (C) Distal right pubis in ventral view. (D) Distal right pubis in medial (articular) view. (E) Diagram showing appropriate position of ischium and pubis in ventral view with muscle attachments overlaid, outline indicates position in the body. (F) Second right sacral rib in ventral view. (G) Same as (F), in lateral view. ant, anterior; art pub, articular surface for attachment to other pubis bone; f sym il, facies symphysialis ilii (articular surface of the sacral rib with the ilium); gas, gastralia; is, ischium; isc, ischiocaudalis muscle; isp, ischiopubis muscle; ist, ischiotruncus muscle; pub, pubis; pub car, pubic cartilage; r ab, rectus abdominis muscle. Scale bar for A–D and E equals 10 cm; scale bar for F–G equals 5 cm.

thick articular band where one osteoderm overlaps the crocodyliform groups, but none resemble the unpitted and next-most posterior one. thickened form seen in Anthracosuchus. Even outside of The osteoderms of Anthracosuchus are not only Crocodylomorpha, osteoderms are well known in testu- atypical for Dyrosauridae, but they are highly unusual for dines, but are not as square, inflated and untextured as Crocodylomorpha, or any armoured vertebrate. Osteo- these (AKH). Osteoderms in sauropods are much more derms from all parts of the body are known in most conical or shaped like a thickened disc, and not rectangular (Dodson et al. 1998). Somewhat similar forms are seen in dorsal osteoderms of ankylosaurs (Burns 2008), but not nearly as square as in UF/IGM 68. Osteoderms in xenarthrans such as and have

Downloaded by [University of Florida] at 08:16 26 May 2014 surface patterns and also have very different overall shapes (Hill 2006).

Phylogenetic analysis Relationship to other dyrosaurids The morphologic data generated by Hastings et al. (2010, 2011) were combined with new data for Anthracosuchus to better understand the relationships of the Cerrejo´n taxa within Dyrosauridae (Table 3). A cladistic analysis was conducted using branch and bound searches with the program PAUP version 4.0b10 (Swofford 2003). The data- set was small enough that branch and bound searches were practical, while exhaustive searches (which are more Figure 9. (Colour online) Proximal phalanx of Anthracosuchus balrogus gen. et sp. nov. UF/IGM 68. (A) Dorsal view, (B) comprehensive) were far too computationally intensive. plantar view and (C) proximal view. Scale bar for A–B equals The cladistics analysis was rooted with Sarcosuchus 5 cm; scale bar for C equals 1 cm. imperator, cherifiensis and 14 A. K. Hastings et al.

Figure 10. (Colour online) Dyrosaurid osteoderms. (A, C) Dorsal osteoderm, UF/IGM 68, in dorsal view; (B, D) same specimen as (A, C) in ventral view; (E) another dorsal osteoderm, UF/IGM 68, in dorsal view; (F) same specimen as (E), in ventral view; (G) lateral osteoderm, UF/IGM 68, in dorsal view; (H) same specimen as G, in ventral view; (I) dorsal osteoderm, UF/IGM 68 in dorsal view; (J) same specimen as (I), in ventral view; (K) same specimen as (I), in cross-sectional view at fracture point visible in (I) and (J); (L) osteoderm of H. rogersii, NJSM 12293, in dorsal view; (M) same specimen as (L), in ventral view; (N) Same specimen as (L), in lateral view. Scale bar equals 5 cm.

robusta treated as outgroup taxa. The characters were unresolved status of these two taxa is that each has ordered as in Hastings et al. (2010, 2011) and multistate characters that are both derived and primitive and that codings were treated as uncertain. For the initial analysis, there are not enough coded characters to discern which we included all 14 ingroup taxa, as in Hastings et al. is more prevalent in the taxon. Acherontisuchus coded (2011), with the addition of the new taxon described in this primitively for characters: 1 (snout ,68% of dorsal skull study (Table 3). length), 72 (minimal occlusal pits) and 74 (symphysis With all taxa included, the initial results are presented wider than high). This taxon also possessed the derived Downloaded by [University of Florida] at 08:16 26 May 2014 in a strict consensus cladogram of 44 trees (Figure 11). state for characters: 23 (linear maxillary margin), 25 This topology is not well resolved, with a large polytomy (alveolar walls level with maxillary surface) and 71 including all 15 dyrosaurid species, i.e. the ingroup. Some (mandibular symphysis ends posterior to anterior 3/4 resolution was recovered with a monophyletic Rhabdog- alveoli). This character conflict, combined with only nathus and Dyrosaurus, and a pairing of Sokotosuchus 18.3% of the characters coded for the taxon, resulted in ianwilsoni and P. gavialoides, but no other relationships poor resolution within Dyrosauridae. Acherontisuchus is were retained. known from fragmentary cranial remains and mandibles The initial results revealed two wildcard taxa, (Hastings et al. 2011), and Congosaurus is known only Congosaurus and Acherontisuchus. These two taxa from the snout section of the skull and not from the were recovered as wildcard taxa in the analysis of braincase (Jouve and Schwarz 2004). Despite the lack of Hastings et al. (2011) as well. An Adams consensus material, an approximation of the snout-to-skull length cladogram placed Acherontisuchus at an unresolved ratio of Congosaurus was determined from what cranial polytomy at the base of Dyrosauridae with Chenani- material was preserved (Jouve and Schwarz 2004) and suchus and a monophyletic group including the 13 other from two nearly complete mandibles of Acherontisuchus species of Dyrosauridae. Congosaurus placed as sister to (Hastings et al. 2011)(seevaluesinTable 1). a uniting a monophyletic Rhabdognathus and Congosaurus too possessed primitive and derived Atlantosuchus þ Guarinisuchus. The reason for the characters. The primitive states were characters: 1 Historical Biology 15

(snout ,68% of dorsal skull length), 3 (thick anterior margin of external nares) and 14 (absence of lateral expansion of premaxilla). Congosaurus possessed more derived characters: 4 (medially positioned orbits), 13 (posterodorsal premaxillary process proximal/anterior to second alveolus), 22 (premaxilla nearly three times longer than wide), 23 (linear maxillary margin), 33 (nasal ceases posterior to first maxillary tooth) and 81 (alveoli more widely spaced in posterior versus anterior snout). When the two wildcard taxa are removed from the analysis, the result is a single-most parsimonious ). cladogram (Figure 11). This topology shows Chenanisu-

2010 chus lateroculi from Africa as the most primitive dyrosaurid. The next most basal dyrosaurids are Anthracosuchus balrogus and Cerrejonisuchus impro- cerus, respectively. The next most basal clade includes H. rogersii as the sister taxon to Sokotosuchus ianwilsoni and P. gavialoides. Arambourgisuchus khouribgaensis is the sister taxon to a clade that includes a monophyletic Dyrosaurus, and another nested clade that includes Atlantosuchus coupatezi þ Guarinisuchus munizi and a monophyletic Rhabdognathus.ForFigure 11, the wildcard taxa have been given tentative positions based on the

0/1. analysis from Hastings et al. (2011), from a 50% majority

¼ rule consensus cladogram where they were placed within a polytomy with the clade including Atlantosuchus þ 1/2; B Guarinisuchus and a monophyletic Rhabdognathus,a ¼

following the character matrix of Hastings et al. ( clade also recovered in the current study. As Hyposaurus was recovered as more primitive than the analysis of

????????? ????????1?0? ? Hastings et al. (2011), it is not presented as part of this sp. nov. polytomy. Instead, the monophyletic Dyrosaurus is retained in its position as sister to the clade including

gen. et Atlantosuchus þ Guarinisuchus and a monophyletic Rhabdognathus. The positions of Acherontisuchus and Congosaurus are denoted with a dashed line to indicate that they were not part of this particular analysis, but are included to provide an approximation of where they likely Downloaded by [University of Florida] at 08:16 26 May 2014 fit within Dyrosauridae. Three characters unambiguously support the clade including all dyrosaurids except the shortest-snouted

Anthracosuchus balrogus genera Chenanisuchus, Anthracosuchus and Cerrejonisu- chus (characters 7, 8 and 71; see snout proportions in Table 1). These characters are a narrow interfenestral bar, an unornamented interfenestral bar and a mandibular symphysis that ends posterior to the anterior three-quarters of the alveoli. The pairing of Hyposaurus with Sokotosuchus and Phosphatosaurus has not been recov- ered in previous phylogenetic analyses. In this analysis, ?01?2010???12???0??1 00000000011111111112222222222 333333333342 33333333334 12345678901234567890123456789 012345678909 01234567890 11010100100?11???1?110100A11BB001??00010BB001??00010 44444444455555555556666666666 777777777788 8 12345678901234567890123456789 012345678901 2 the only unambiguous synapomorphy supporting this clade is character 45, a reversal to state 0, which is the

gen. gen. presence of deep pits on the dorsal surface of the parietal. Character 47, a pronounced anterolateral process of the postorbital, unites all dyrosaurids except sp. nov. sp. nov. Table 3. Morphologic character states of Anthracosuchus balrogus et Anthracosuchus balrogus et Notes: Multistate characters are treated as uncertain. For nexus file, see Electronic Supplement. A Chenanisuchus. 16 A. K. Hastings et al.

Figure 11. Cladograms resulting from a phylogenetic analysis of Dyrosauridae. Left cladogram is a strict consensus with all named species known from more than dentary fragments. Right cladogram represents a single cladogram resulting from analysis with the two wildcard taxa removed (Acherontisuchus guajiraensis and Congosaurus bequaerti). In the right cladogram, dotted lines for the wildcard taxa represent approximate placement based on cladistic analysis of Hastings et al. (2011). The right cladogram is placed in stratigraphic and geographic context. Dates were obtained from Gradstein et al. (2004). CI, consistency index; RI, retention index; RC, rescaled consistency index; HI, homplasy index.

Relationship of Dyrosauridae to other crocodyliforms between Dyrosauridae and . This crocodyli- Downloaded by [University of Florida] at 08:16 26 May 2014 In many phylogenetic studies of crocodyliforms, long- form analysis utilised a cladistic data-set that included irostrine taxa group together despite instances of clear representatives from all major lineages of Crocodyli- convergence (Clark 1994). Characters largely tied to formes (Jouve et al. 2006) and added to it the new short- Anthracosuchus balrogus Cerrejonisuchus longirostry most often yield a close phylogenetic snouted and improcerus. The morphological matrix of Jouve et al. relationship (Jouve et al. 2006). A long-standing problem (2006) was chosen because 234 characters were coded has been the largely -aged thalattosuchians pairing across a large sampling of Crocodyliformes (n ¼ 47) that with the Cretaceous pholidosaurids and Cretaceous– included the dyrosaurids Dyrosaurus and Chenanisuchus. Eocene dyrosaurids (Jouve et al. 2006; Pol and Gasparini Although more recent studies have also investigated this 2009; de Andrade et al. 2011). problem thoroughly (e.g. Pol and Gasparini 2009;de The most primitive dyrosaurids in the analysis above Adrade et al. 2011), these studies included only long- were also the shortest snouted. We carried out an irostrine dyrosaurids, excluding the short-snouted Chena- additional phylogenetic analysis to test the hypothesis nisuchus. that longirostrine thalattosuchians are much more The two Cerrejo´n dyrosaurids were coded and added to primitive within Crocodyliformes than other analyses the matrix (Table 4) that was then run with a heuristic suggest and that inclusion of the new brevirostrine search with 10,000 repetitions and the random seed dyrosaurids will result in a more disparate relationship function in PAUP version 4.0b10 (Swofford 2003). The Downloaded by [University of Florida] at 08:16 26 May 2014

Table 4. Cladistic data for Anthracosuchus balrogus gen. et sp. nov. and Cerrejonisuchus improcerus used in the analysis by Jouve et al. (2006). 111111111122222222223 123456789012345678901234567890 Anthracosuchus balrogus 10?02/30 0 113010???????????110113 Cerrejonisuchus improcerus 10?0200113011011???????0110113 333333333444444444455555555556 123456789012345678901234567890 Anthracosuchus balrogus 01000200?0000012011011??0??1?? Cerrejonisuchus improcerus 01000200?0000012011011??0??1?? 666666666777777777788888888889 123456789012345678901234567890 Anthracosuchus balrogus ?110 2 ? – 01?0??0????0???01010100 Cerrejonisuchus improcerus ?110 2 ? – 01?01?0??00010?01011100 111111111111111111111 999999999000000000011111111112 123456789012345678901234567890 Anthracosuchus balrogus 01?1 1 ? ? 0?10??1???1122?0000??0? Cerrejonisuchus improcerus 01111010010??10??111200000???? 111111111111111111111111111111 222222222333333333344444444445 123456789012345678901234567890 Anthracosuchus balrogus 011?00100?1???????0?0011???1?? Cerrejonisuchus improcerus ????????0?11???21000001??????1 111111111111111111111111111111 555555555666666666677777777778 123456789012345678901234567890 Anthracosuchus balrogus ??0? 1 ? 0 0?????????????????????? Cerrejonisuchus improcerus 0101 1 ? ? ???1?01???????2???????? 111111111111111111122222222222 888888888999999999900000000001 123456789012345678901234567890 Anthracosuchus balrogus ?11? 0 00/10010000000??1?????????1 Cerrejonisuchus improcerus ?1000010100000000?11?????????1 222222222222222222222222 111111111222222222233333 itrclBiology Historical 123456789012345678901234 Anthracosuchus balrogus ????????????0??10???1??? Cerrejonisuchus improcerus ???? 0 0 ? ?????0??11???1???

Note: A total of 234 characters were coded and the other 47 taxa used in the previous analysis remained unchanged in the current analysis. For nexus file, see Electronic Supplement. 17 18 A. K. Hastings et al.

result was 30 equally most parsimonious cladograms position within Crocodyliformes, basal to nearly all of (Figure 12), as opposed to 124 in the heuristic search of Mesoeucrocodylia (Figure 12). and Jouve et al. (2006). The overall results were similar to Dyrosauridae were united in a node, but different from those of Jouve et al. (2006). Thalattosuchians shared a the previous analysis in that Elosuchus had an unresolved close relationship with dyrosaurids and pholidosaurids. All relationship in a polytomy including Dyrosauridae and a four dyrosaurids were monophyletic, although the only clade that includes all other pholidosaurids. Jouve et al. internal resolution had Cerrejonisuchus and Anthracosu- (2006) found Elosuchus as primitive to both Pholidosaur- chus as sister taxa. Despite the short-snouted dyrosaurids idae and Dyrosauridae. Pol and Gasparini (2009) studied being present in the analysis, the resultant topology the correlation between longirostrine characters and the was very similar to other studies addressing the thalattosuchian phylogenetic issue. They concluded from thalattosuchian issue (e.g. Pol and Gasparini 2009;de their analyses that because these characters are not Adrade et al. 2011). uniformly found in longirostrine taxa, they are in fact Following Jouve et al. (2006), the same data-set was informative of the phylogenetic evolution of the taxa. In used for another heuristic search with 10,000 replicates the present study, the addition of short-snouted dyrosaurids with the 15 characters suggested as associated with did not resolve the problem of convergence driving the longirostry removed (characters 5, 7, 8, 12, 15, 30, 46, 47, topology between thalattosuchians, dyrosaurids and 68, 83, 103, 150, 161, 172 and 189). The result of this pholidosaurids. These results suggest that these convergent analysis was very similar to the heuristic search of Jouve characters are not limited to longirostry and are present et al. (2006), with Thalattosuchia having a more basal even in non-longirostrine taxa. Downloaded by [University of Florida] at 08:16 26 May 2014

Figure 12. Two cladograms from analyses utilising a matrix with representatives across Crocodylomorpha. (A) Strict consensus cladogram resulting from analysis using all 234 characters of Jouve et al. (2006). (B) Strict consensus cladogram resulting from analysis that omitted the 15 characters associated with longirostry. Note the very different placement of Thalattosuchia with respect to Dyrosauridae in the two analyses. CI, consistency index; RI, retention index; RC, rescaled consistency index; HI, homplasy index. Historical Biology 19

Biogeography taxa recognised in the earlier analysis were not included. In order to discern possible methods of dispersal within The third limitation is due to S-DIVA utilising phylogeny Dyrosauridae, a biogeographic analysis was run on the to determine dispersal; taxonomically uninformative dyrosaurid matrix (Table 3) using the program S-DIVA fossils cannot be included. Thus, the present analysis and (Yu et al. 2010a), which utilises a tree file as generated by discussion are limited to identifiable taxa known from PAUP (Swofford 2003). We assigned each taxon to a nearly complete fossil skulls and do not take into account region: Africa, North America and South America, regions where dyrosaurid fossils were found that could not assigned as regions A, B and C, respectively (Figure 13). be diagnosed to genus. Finally, the program cannot The program generates a consensus cladogram from the incorporate temporal data, and timing of dispersal must be tree file, incorporating the biogeographic region of the taxa determined using ages of the fossil taxa. included (Yu et al. 2010b). The program generates likely All past phylogenetic studies of Dyrosauridae have regions of origin for the ancestors of the taxa included in recovered an African origin for the family, and the current the analysis. The program creates a pie chart at each of analysis supports this as well. The biogeographic analysis these nodes with per cent probability by region for the recovered five optimised dispersal histories, with a ancestor. The program can be set to allow for multiple minimum of three dispersal events from Africa to the regions to be occupied simultaneously. New World within Dyrosauridae. The first of these Four limitations of the software should be kept in occurred with the ancestor of Anthracosuchus to South mind. The first is that the program from which it was America (Figure 14). The second dispersal occurred with derived (DIVA; Ronquist 1996) treats dispersal between the ancestor of Guarinisuchus munizi reaching eastern different regions as equally possible, with no preference Brazil. The third independent trans-Atlantic dispersal for a certain direction or between certain regions would have been the ancestor of H. rogersii immigrating to (Kodandaramaiah 2010). The second is that the program North America. Had Acherontisuchus guajiraensis been must assign a topology with no polytomies. As a result, included, it would likely have represented a fourth taxa with unresolved relationships in non-biogeographic dispersal, as its relationships were most often close to studies should be interpreted with caution when using S- Arambourgisuchus and Dyrosaurus of Africa and did not DIVA (Yu et al. 2010b). For this reason, both wildcard have a direct relationship with Guarinisuchus or any other Downloaded by [University of Florida] at 08:16 26 May 2014

Figure 13. Cladogram resulting from biogeographic analysis of Dyrosauridae using the program S-DIVA (Yu et al. 2010a). Solid arrows indicate dispersal events. The dotted line indicates a possible back dispersal from South America to Africa. 20 A. K. Hastings et al.

known from fragmentary fossils in Asia (e.g. Buffetaut 1978b, 1978c) due to the limitations of the analysis. The dispersal into central South America is indicated by Palaeocene fossil occurrences in Bolivia (Buffetaut 1991) that likely occurred during the Late Cretaceous while a shallow continental seaway was still present in western South America (Riccardi 1991; Hoorn et al. 2012).

Discussion . The new taxon can be referred to Dyrosauridae based on the following diagnostic characters for the family, following the diagnosis of Jouve et al. (2006): (1) lateral margins of external nares lower than anterior or posterior margins; (2) supratemporal fenestrae at least twice longer than wide; (3) occipital tuberosities present, directed posteriorly and formed by exoccipitals; (4) exoccipitals participate greatly in occipital condyle; (5) neural spines extremely elongate dorsally in caudal Figure 14. (Colour online) Map showing dispersal pattern and timing for Dyrosauridae. Map is of the Late Cretaceous, colourised vertebrae; (6) dorsal margin of neural spines narrow, not from Scotese (2001). There are at least three independent dispersals thickened; (7) caudal centra quadrangular in cross section; from Africa to the New World (see text for explanation). (8) haemal arches extremely elongate ventrally. For a list of characters discerning Anthracosuchus from all other New World dyrosaurid. Most of these dispersal events dyrosaurid diagnoses, see Electronic Supplement. would have occurred during the Late Cretaceous. This is Palaeobiology. The orbits of Anthracosuchus are very supported by the one Late Cretaceous report of a dyrosaurid widely spaced for Dyrosauridae (Figures 2–4). Typical from Colombia, although the six isolated vertebral centra dyrosaurids (e.g. Dyrosaurus, Rhabdognathus and Hypo- could only be identified as belonging to Dyrosauridae saurus) have centrally placed orbits, close to the midline of (Langston 1965). Congosaurus bequaerti shared a close the skull (Jouve, Bouya, et al. 2005). The only other relationship with other African taxa in the analysis of dyrosaurid with widely set orbits is Chenanisuchus (Jouve, Hastings et al. (2011), and it likely only required dispersal Bouya, et al. 2005). The only extant crocodylian with within Africa to account for its occurrence in . This notably wide-set orbits is Gavialis gangeticus. A bivariate interpretation excludes the occurrence of dyrosaurids plot of width of skull measured at orbits and interorbital Downloaded by [University of Florida] at 08:16 26 May 2014

Figure 15. (Colour online) Bivariate plot with natural log transform of skull width as measured at orbit (x-axis) and interorbital width of dyrosaurids and extant crocodylians. Historical Biology 21

width (with natural log transform) revealed that Anthra- cosuchus had greater similarity in terms of orbital spacing with Gavialis than with typical dyrosaurids, Alligator mississippiensis, Caiman crocodilus or nilo- ticus (Figure 15). Extant Gavialis typically acquire their prey, almost exclusively fish, below the water surface (Thorbjarnarson 1990; Grenard 1991). The three extant species with more narrowly placed orbits more commonly acquire prey from a position at the surface of the water (Grenard 1991; Thorbjarnarson 1993). More widely set orbits may be tied to a non-surficial prey acquisition feeding strategy and likely indicate a deviation by Anthracosuchus and Chenanisuchus from the typical behaviour of dyrosaurids. This subsurface predation method may also be inferred in the highly marine metriorhynchid crocodyliforms, which have laterally placed orbits (Pol and Gasparini 2009). The distal pubis is rarely recovered for New World dyrosaurids, with the only other figured specimen from H. rogersii from the Late Cretaceous marine deposits of New Jersey (Troxell 1925). Both have a more elongate suture between left and right pubes as compared to extant crocodylians. That of Anthracosuchus (Figure 8)is proportionally longer than that of H. rogersii. The pubic cartilage attaches to the anterior portions of these bones in extant crocodylians, and the symphyseal region is where the ischiopubis and ischiotruncus muscles attach (Uriona and Farmer 2008; Schwarz-Wings et al. 2009). A non- enlarged distal ischium, also preserved with UF/IGM 68, indicates highly reduced attachment surfaces for the ischiopubis, ischiotruncus and ischiocaudalis muscles (Figure 8). The ischiopubis muscle in particular is used primarily in buoyancy control (Uriona and Farmer 2008). Much like the implications for Acherontisuchus (Hastings et al. 2011), it would appear Anthracosuchus also relied less on these muscles for pitch control within the water. The pubis is heavily utilised in extant crocodylian respiration and the bone is pulled ventrally and rotated Figure 16. (Colour online) Portions of pelomedosoid turtle Downloaded by [University of Florida] at 08:16 26 May 2014 carapace, UF/IGM 71, with crocodyliform bite marks. (A, C) Costal slightly by the rectus abdominus and diaphragmaticus fragment in dorsal view; (B, D) same fragment in ventral view; (E, muscles, bringing air into the lungs (Claessons 2004). The G) neural fragment in dorsal view; (F, H) peripheral in dorsal view; pubic cartilage is what dominantly attaches these two bones (I, K) m1 tooth from holotype of Anthracosuchus balrogus gen. et in extant crocodylians (Uriona and Farmer 2008). The longer sp. nov., UF/IGM 67, placed within the impact shell scar featured in and more rigid suture marks in Anthracosuchus may have (I); (J, L) peripheral indorsal view. bm, bite mark; bm?, possible bite mark; bm&p, bite mark and regrown bone plug; dm, drag marks; reduced the flexibility of this joint, but provided a firmer m1, first maxillary tooth; pr, proximal rib. Scale bar for A–H, J and basis from which to depress the pubes and take in air. This L equals 5 cm; scale bar for I and K equals 1 cm. firmer suture may have also meant that Anthracosuchus was less capable of pubic rotation as in extant crocodylians much more undulating fashion than extant crocodylians, (Claessons 2004). The Eocene Dyrosaurus maghribensis of which are nearly entirely tail-based swimmers (MacDonald Morocco possessed a pubis that is spatulate and much more 2005; Schwarz-Wings et al. 2009). The osteoderms of similar to that of extant crocodylia (Jouve et al. 2006). hyposaurine dyrosaurids are known to imbricate dorsally, The preserved ribs of Anthracosuchus indicate similar providing bracing for the vertebral column. vertical positions for muscle attachments, as indicated for Osteoderms are not as extensive along the lateral Hyposaurinae (Schwarz-Wings et al. 2009)andAcheronti- portion of the dyrosaurid body as in some extant suchus (Hastings et al. 2011). The resultant angle for axial crocodylians, likely affording dyrosaurids greater flexi- muscles was found to imply dyrosaurids likely swam in a bility at the sacrifice of reduced stability and armoured 22 A. K. Hastings et al.

protection. This flexibility would have been necessary for (Florida Museum of Natural History), Gary S. Morgan Student the more undulating locomotion of dyrosaurids (Schwarz- Research Award and the Cerrejo´n Coal mine. Wings et al. 2009). The osteoderms of Anthracosuchus likely did not imbricate at all, implying an even greater degree of flexibility, at the sacrifice of reduced stability, References particularly for rostroterminal torsion. Barbosa JA, Kellner WA, Viana MSS. 2008. New dyrosaurid crocodylomorph and evidences for faunal turnover at the K–P Pitting in osteoderms has been associated with thermal transition in Brazil. Proc R Soc B. 275(1641):1385–1391. regulation in extant crocodyliforms, with the increased Benson RBJ, Butler RJ, Lindgren J, Smith AS. 2010. Mesozoic marine surface area and veination allowing a greater amount of tetrapod diversity: mass and temporal heterogeneity in geological megabiases affecting vertebrates. Proc R Soc B. 277 heat transfer to take place along the ’s broadly (1683):829–834. exposed dorsal surface (Seidel 1979). Bones of the skull Brochu CA. 1996. Closure of neurocentral sutures during crocodilian that are unpitted are separated from the dermis by muscle, ontogeny: implications for maturity assessment in fossil . J Vert Paleontol. 16(1):49–62. as opposed to pitted surfaces that are very near to the Brochu CA, Bouare´ ML, Sissoko F, Roberts EM, O’Leary MA. 2002. dermis (Seidel 1979). The osteoderms of Anthracosuchus A dyrosaurid crocodyliform braincase from . J Paleontol. 76(6): are unpitted and likely resided deeper in the dorsal 1060–1071. musculature than in extant crocodylians, and were likely Buffetaut E. 1978a. Les Dyrosauridae (Crocodylia, ) des phosphates de l’Eocene infe´rieur de Tunisie: Dyrosaurus, Rhabdog- not utilised as much for thermoregulation. The small nathus, Phosphatosaurus.Ge´ol Mediter. 5(2):237–256. rivulets and thin crevices radiating from the centre of the Buffetaut E. 1978b. A dyrosaurid (, Mesosuchia) from the dorsal surface are instead likely indicative of keratinisa- upper Eocene of Burma. Neues Jahrb Geol Pala¨ontol Monatshefte. 5:273–281. tion, which has been proposed for similar features on the Buffetaut E. 1978c. Crocodilian remains from the Eocene of . cranium as a display feature in the extinct crocodyliform Neues Jahrb Geol Pala¨ontol Abh. 156:262–283. Aegisuchus (Holliday and Gardner 2012). Buffetaut E. 1979. Sokotosuchus ianwilsoni and the evolution of the dyrosaurid . Field Monogr. 1(6):31–41. Diet. The teeth of Anthracosuchus are blunt and robust, Buffetaut E. 1991. Fossil crocodilians from Tiupampa, (Santa Lucia which may be indicative of a durophagous diet with a Formation, Early ) Bolivia: a preliminary report. Rev Tec strong crushing bite (Massare 1987). A large turtle shell Yacim Pet Fisc Boliv. 12(3–4):541–544. Burns ME. 2000. Taxonomic utility of ankylosaur (Dinosauria, (UF/IGM 71) has preserved several dents, scores and Ornithischia) osteoderms: Glyptodontopelta mimus Ford, 2000: a scratch marks on dorsal and ventral surfaces that resemble test case. J Vert Paleontol. 28(4):1102–1109. predation attempts by a crocodyliform (Noto et al. 2012). Cadena E, Bloch JI, Jaramillo C. 2010. New podocnemidid turtle (Testudines: Pleurodira) from the middle-upper Paleocene of South There are straight puncture marks, tooth slide marks and at America. J Vert Paleontol. 30(2):367–382. least one rehealed puncture mark (Figure 16). The largest Cadena EA, Bloch JI, Jaramillo C. 2012. New bothremydid turtle of these puncture marks fits very well with the largest (Testudines, Pleurodira) from the Paleocene of Northeastern Colombia. J Paleontol. 86(4):689–699. preserved tooth (m1) of UF/IGM 67 (Figure 2). Cadena EA, Ksepka DT, Jaramillo CA, Bloch JI. 2012. New pelomedusoid turtles from the late Palaeocene Cerrejo´n Formation of Colombia and implications for phylogeny and body size evolution. Acknowledgements J Syst Palaeontol. 10(2):313–331. Many thanks to L. Teicher, F. Chavez, G. Hernandez, C. Montes Claessons LPAM. 2004. Archosaurian respiration and the pelvic girdle and the rest of the crew at the Cerrejo´n coal mine for access to the aspiration breathing of crocodyliforms. Proc R Soc Lond B. site and housing during field work. Thanks to Ecopetrol S.A.-ICP 271(1547):1461–1465. Clark JM. 1994. Patterns of evolution in Mesozoic Crocodyliformes. for logistic support. Thanks to H. Garcia, A. Rincon, E. Cadena, Downloaded by [University of Florida] at 08:16 26 May 2014 In: In the shadow of the . New York, NY: Cambridge F. Herrera, S. Wing and M. Ramirez for help collecting fossils at University Press; p. 84–97. Cerrejo´n. J. Bourque provided useful guidance for specimen de Andrade MB, Edmonds R, Benton MJ, Schouten R. 2011. A new preparation and conservation. Thanks to E. Whiting for taking species of (Mesoeucrocodylia, ) additional measurements during manuscript revision. Thanks to from , and a review of the genus. Zool J Linn Soc. 163(s1): R. Pellegrini for access to and assistance with the collections at the S66–S108. NJSM. Thanks to J. Head for assistance and guidance during the de Stefano G. 1903. Nuovi rettili degli strati a fosfato della . Boll development of this manuscript. Lastly, thanks to the Colombian Soc Geol Ital. 22(1):51–80. Geological Survey (SGC) in Bogota´, Colombia, for access to Denton RK, Dobie JL, Parris DC. 1997. The marine crocodilian specimens and facilitating ongoing research of the remarkable Hyposaurus in North America. In: Ancient marine reptiles. New York, NY: Academic Press; p. 375–397. vertebrate fossils of the Palaeocene–Eocene of Colombia. Dodson PD, Krause W, Forster CA, Sampson SD, Ravoavy F. 1998. Titanosaurid (Sauropoda) osteoderms from the Late Cretaceous of Madagascar. J Vert Paleontol. 18(3):563–568. Gasparini Z, Pol D, Spalletti LA. 2006. An unusual marine crocodyliform Funding from the Jurassic–Cretaceous boundary of Patagonia. Science. Funding for research and fieldwork was made possible by the 311(70):70–73. Gradstein F, Ogg J, Smith A. 2004. A geologic time scale 2004. National Science Foundation [grant numbers DEB-0733725 and Cambridge: Cambridge University Press. DSGC 640179], funds from the Florida Museum of Natural Grenard S. 1991. Handbook of alligators and crocodiles. Malabar, FL: History, the Smithsonian Tropical Research Institute Paleobiol- Krieger Publishing Company. ogy Fund, the Geological Society of America Graduate Student Hastings AK, Bloch JI, Jaramillo CA. 2011. A new longirostrine Research Grant, the R. Jerry Britt, Jr Paleobiology Award dyrosaurid (Crocodylomorpha, Mesoeucrocodylia) from the Paleo- Historical Biology 23

cene of north-eastern Colombia: biogeographic and behavioural Marinho TS, Carvalho IS. 2009. An -like sphagesaurid implications for new-world Dyrosauridae. Palaeontology. 54(5): crocodyliform from the Late Cretaceous of Brazil. J S Am Earth 1095–1116. Sci. 27(2009):36–41. Hastings AK, Bloch JI, Cadena EA, Jaramillo CA. 2010. A new small Massare JA. 1987. Tooth morphology and prey preference of Mesozoic short-snouted dyrosaurid (Crocodylomorpha, Mesoeucrocodylia) marine reptiles. J Vert Paleontol. 7(2):121–137. from the Paleocene of Northeastern Colombia. J Vert Paleontol. McGregor J. 2005. Crocodile crimes: people versus wildlife and the 30(1):139–162. politics of postcolonial conservation on Lake Kariba. Zimbabwe. Hay OP. 1930. Second bibliography and catalogue of the fossil Vertebrata Geoforum. 36(2005):353–369. of North America. Vol. 2. Carnegie Inst Washington Publ. Noto CR, Main DJ, Drumheller SK. 2012. Feeding traces and 390:1–1074. paleobiology of a Cretaceous () crocodyliform: example Head JJ, Bloch JI, Hastings AK, Bourque JR, Cadena EA, Herrera FA, from the Woodbine Formation of Texas. Palaios. 27(2):105–115. Polly PD, Jaramillo CA. 2009a. Giant boid snake from the Paleocene Pol D, Gasparini Z. 2009. Skull Anatomy of Dakosaurus andiniensis neotropics reveals hotter past equatorial temperatures. Nature. (Thalattosuchia: Crocodylomorpha) and the phylogentic position of 457:715–717. Thalattosuchia. J Syst Palaeontol. 7:165–197. Head JJ, Bloch JI, Hastings AK, Bourque JR, Cadena EA, Herrera FA, Riccardi AC. 1991. Jurassic and Cretaceous marine connections between Polly PD, Jaramillo CA. 2009b. Brief communication arising (reply). the Southeast Pacific and Tethys. Palaeogeogr Palaeoecol. Nature. 460:E4–E5. 87:155–189. Hill RV. 2006. Comparative anatomy and histology of xenarthran Ronquist F. 1996. DIVA, version 1.1. Computer program and manual osteoderms. J Morphol. 267(12):1441–1460. available by anonymous FTP from Uppsala University, Uppsala, Holliday CM, Gardner NM. 2012. A new eusuchian crocodyliform with . Available at: ftp.uu.se or ftp.systbot.uu.se. novel cranial integument and its significance for the origin and Ross CA. 1989. Crocodiles and alligators. New York, NY: Facts on File. evolution of Crocodylia. PLoS ONE. 7:e30471. Schwarz D, Frey E, Martin T. 2006. The postcranial skeleton of the Hoorn C, Wesselingh FP, ter Steege H, Bermudez MA, Mora A, Sevink J, Hyposaurinae (Dyrosauridae; Crocodyliformes). Palaeontology. Sanmartı´n I, Sanchez-Meseguer, Anderson CL, Figueirido JP. 2012. 49(4):695–718. Amazonia through time: andean uplift, climate change, landscape Schwarz-Wings D, Frey E, Martin T. 2009. Reconstruction of the bracing evolution, and biodiversity. Science. 330(927):927–931. system of the trunk and tail in hyposaurine dyrosaurids Jaramillo C, Pardo-Trujillo A, Rueda M, Harrington G, Bayona G, Torres (Crocodylomorpha; Mesoeucrocodylia). J Vert Paleontol. 29(2): V, Mora G. 2007. Palynology of the Upper Paleocene Cerrejo´n 453–472. Formation. North Colombia Palynol. 31(1):153–189. Scotese CR. 2001. Digital paleogeographic map archive on CD-ROM. Jouve S. 2005. A new description of Dyrosaurus phosphaticus (Thomas, Arlington, TX: PALEOMAP Project. 1893) (Mesoeucrocodylia: Dyrosauridae) from the Lower Eocene of Seidel MR. 1979. The osteoderms of the and their North Africa. Can J Earth Sci. 42(3):323–337. functional significance. Herpetologica. 35(4):375–380. Jouve S. 2007. Taxonomic revision of the dyrosaurid assemblage Sereno PC, Larsson HC, Sidor CA, Gado B. 2001. The giant (Crocodyliformes: Mesoeucrocodylia) from the Paleocene of the crocodyliform Sarcosuchus from the Cretaceous of Africa. Science. , West Africa. J Paleontol. 81(1):163–175. 294(5546):1516–1519. Jouve S, Bardet N, Jalil N-E, Suberbiola X, Bouya B, Amaghzaz M. 2008. Swofford D. 2003. PAUP* 4.0b10. Sunderland, MA: Sinauer Associates. The oldest African crocodylian: phylogeny, paleobiogeography, and Thorbjarnarson JB. 1990. Notes on the feeding behavior of the differential survivorship of marine reptiles through the Cretaceous- (Gavialis gangeticus) under semi-natural conditions. J Herpetol. Tertiary boundary. J Vert Paleontol. 28(2):409–421. 24(1):99–100. Jouve S, Bouya B, Amaghzaz M. 2005. A short-snouted dyrosaurid Thorbjarnarson JB. 1993. Fishing behavior of spectacled caiman in the (Crocodyliformes, Mesoeucrocodylia) from the Palaeocene of Venezuelan Llanos. Copeia. 1993(4):1166–1171. Morocco. Palaeontolgy. 48(2):359–369. Troxell EL. 1925. Hyposaurus, a marine crocodilian. Am J Sci Ser 5. Jouve S, Bouya B, Amaghzaz M. 2008. A long-snouted dyrosaurid 9(54):489–514. (Crocodyliformes, Mesoeucrocodylia) from the Paleocene of Tolkien JRR. 1954. The fellowship of the ring. London: Unwin Books. Morocco: phylogenetic and palaeobiogeographic implications. Uriona TJ, Farmer CG. 2008. Recruitment of the diaphragmaticus, Palaeontology. 51(2):281–294. ischiopubis and other respiratory muscles to control pitch and roll in Jouve S, Iaroche`ne M, Bouya B, Amaghzaz M. 2005. A new the American alligator (Alligator mississippiensis). J Exp Biol. crocodyliform dyrosaurid from the Palaeocene of Morocco and a 211(7):1141–1147. phylogenetic analysis of Dyrosauridae. Acta Palaeontol Pol. 50(3): Walker AD. 1970. A revision of the Jurassic victor 581–594. (Marsh), with remarks on the classification of crocodiles. Philos

Downloaded by [University of Florida] at 08:16 26 May 2014 Jouve S, Iaroche`ne M, Bouya B, Amaghzaz M. 2006. A new species of Trans R Soc Lond Biol Sci. 257(816):323–372. Dyrosaurus (Crocodylomorpha, Dyrosauridae) from the early Walker AD. 1990. A revision of Sphenosuchus acutus Haughton, Eocene of Morocco: phylogenetic implications. Zool J Linn Soc. a crocodylomorph reptile from the Elliot Formation (Late or 148(4):603–656. ) of South Africa. Philos Trans R Soc Lond Biol Sci. Jouve S, Schwarz D. 2004. Congosaurus bequaerti, a Paleocene 330(1256):1–120. dyrosaurid (Crocodyliformes; Mesoeucrocodylia) from Landana Wheatley PV. 2010. Understanding saltwater tolerance and marine (Angola). Bull Inst R Sci Nat Belg Sci Terre. 74:129–146. resource use in the Crocodylia: a stable isotope approach Kellner AWA, Campos DA, Riff D, de Andrade MB. 2011. A new [dissertation]. Santa Cruz, CA: University of California. crocodylomorph (, ) with horn-like Whetstone KN, Whybrow PJ. 1983. A ‘cursorial’ crocodilian from the tubercles from Brazil. Zool J Linn Soc. 163(s1):S57–S65. Triassic of Lesotho (Basutoland), southern Africa. Occas Pap Univ Kodandaramaiah U. 2010. Use of dispersal-vicariance analysis in Kans Mus Nat Hist. 106:1–37. biogeography – a critique. J Biogeogr. 37(1):3–11. Wing S, Herrera F, Jaramillo C, Gomez-Navarro C, Wilf P, Labandeira C. Langston WL. 1965. Fossil crocodilians from Colombia and the Cenozoic 2009. Late Paleocene fossils from the Cerrejo´n Formation, history of the Crocodilia in South America. U Calif Publ Geol Sci. Colombia, are the earliest record of Neotropical rainforest. Proc 52:1–168. Natl Acad Sci. 106(44):18627–18632. Langston WL. 1995. Dyrosaurs (Crocodilia, Mesosuchia) from the Yu Y, Harris AJ, He XJ. 2010a. S-DIVA (Statistical Dispersal– Paleocene Umm Himar Formation, Kingdom of . US Vicariance Analysis) 1.5. Accessed from http://mnh.scu.edu.cn/soft/ Geol Surv Bull. 2093:F1–F36. blog/sdiva on September 2, 2011. MacDonald S. 2005. Activation of the axial muscles during aquatic Yu Y, Harris AJ, He XJ. 2010b. S-DIVA (statistical dispersal-vicariance locomotion in the estuarine crocodile (Crocodylus porosus) [honours analysis): a tool for inferring biogeographic histories. Mol thesis]. Brisbane: University of Queensland. Phylogenet Evol. 56(2):848–850.