INTRODUCTION
Living lizards includes ~7000 species distrib uted in all continents, except Antarctica (Uetz et al., 2020). The term lizard refers to a paraphylet ic assemblage of mostly terrestrial limbed squa mates (e.g., iguanians, lacertoids, gekkotans), thus excluding snakes and amphisbaenians. All these extant limbed and limbless taxa plus some completely extinct lineages (e.g., mosasaurs) con stitute the clade Squamata (e.g., Gauthier & de Queiroz, 1998; Pyron, 2017; Simões et al., 2018; Sues, 2019; Simões & Pyron, 2021). Squamates have a patchy Mesozoic fossil record, especially during the Triassic and Jurassic periods, when they probably originated and had their first di versification event (e.g., Evans & Jones, 2010; Simões et al., 2017, 2018). The lizard record is also unevenly distributed in Mesozoic rocks, with considerably abundant and taxonomically di verse occurrences in the Cretaceous of Laurasia (e.g., Borsuk-Białynicka & Moody, 1984; Gao & Norell, 1998, 2000; Nydam et al., 2000; Nydam & Cifelli, 2002; Conrad & Norell, 2007; Evans & Wang, 2010; DeMar Jr et al., 2017; Herrera- Flores et al., 2021). Conversely, the record in the southern continents is still scanty and the sam pling bias coupled with the incompleteness of most hitherto known lizard specimens hamper solid interpretations on early lizard distribution and diversification (Benson et al., 2013; Cleary et al., 2018). For instance, the most diverse Mesozoic record of Gondwanan lizards is from the Cretaceous of Brazil, including a few relative ly complete specimens (e.g., Simões et al., 2017; Bittencourt et al., 2020), namely Tijubina pon tei, Olindalacerta brasiliensis, and Calanguban alamoi from the Aptian-Albian Crato Formation (Araripe Basin; Bonfim-Júnior & Marques, 1997; Evans & Yabumoto, 1998; Bonfim-Júnior & Rocha-Barbosa, 2006; Simões, 2012; Simões et al., 2015a, 2017), Neokotus sanfrancis canus from the Valanginian Quiricó Formation (Sanfranciscana Basin; Bittencourt et al., 2020), Gueragama sulamericana from the Turonian- Campanian Goio-Erê Formation (Bauru Basin; Simões et al., 2015b), Brasiliguana prudentis (Nava & Martinelli, 2011) and a lizard-like squa mate (Candeiro et al., 2009) from the Campanian Adamantina Formation (Bauru Basin), and Pristiguana brasiliensis from the Maastrichtian Serra da Galga Formation (Bauru Basin; Estes & Price, 1973) (Fig. 1). In contrast, Mesozoic lizards from Argentina include a few scattered remains collected in Upper Cretaceous units (Candeleros, Anacleto, and Los Alamitos forma tions) and identified as unnamed iguanians and scincomorphans (Apesteguía et al., 2005; Albino, 2002, 2007, 2011; Brizuela & Albino, 2011; Albino & Brizuela, 2014) (Fig. 1). Although in the last years the number of named lizard species and records grew considerably in South America, their phylogenetic and taxonomic affinities re main unclear for most of these occurrences, es pecially those based on fragmentary and/or iso lated material (Albino & Brizuela, 2014; Simões et al., 2015a, 2017). For the rest of Gondwana, lizard or lizard-like remains are also sparse, in cluding iguanians and scincomorphs that come from the Middle Jurassic of India (Evans et al., 2002), the Upper Jurassic of Tanzania (Zils et al., 1995), the Lower Cretaceous of Morocco (Richter, 1994; Broschinski & Sigogneau-Russell, 1996) and South Africa (Ross et al., 1999), and the latest Cretaceous of Madagascar (the par tially complete skeleton of the possible cordylid Konkasaurus mahalana; Krause et al., 2003).
The aim of this contribution is to describe a new genus and species of fossil lizard based on a partial skull (MACN-Pv-N 120) from the Upper Cretaceous Bajo de la Carpa Formation (Neuquén Group, Neuquén Basin). This spec imen was collected approximately 35 years ago in the historical locality “Boca del Sapo”, cor responding to outcrops located north of the Neuquén City, now the campus and vicinity of the Universidad Nacional del Comahue (Woodward, 1896; Gasparini, 1971; Bonaparte, 1991), Neuquén Province (Patagonia, Argentina). MACN-Pv-N 120 remained unnoticed in a small box together with bone fragments of notosuchian mesoeucrocodylians and a tooth of an abelisau rid theropod collected and donated by Oscar de Ferrariis (former Director of the Museum of the Universidad Nacional del Comahue) in the 1980s to the Colección Paleovertebrados of the Museo Argentino de Ciencias Naturales “Bernardino Rivadavia”. The new lizard specimen remained unnoticed in the collection possibly due to its small size and fairly weathered surface. It was a surprising finding during a revision of this collec tion and, thus, was further prepared. Although the skull is incomplete, it is currently by far the most informative squamate lizard from the Mesozoic of Patagonia and the first named Mesozoic lizard from Argentina.
MATERIALS AND METHODS
Geological and Paleontological Settings
MACN-Pv-N 120 was collected from the cam pus of the Universidad Nacional del Comahue, in Neuquén city, Neuquén Province, Argentina. This outcrop of the campus of the Universidad Nacional del Comahue together with a near by series of cliffs in the north of Neuquén City are referred as the “Boca del Sapo” locality and has yielded several fossil remains since the late XIX century (Woodward, 1896; Gasparini, 1971; Bonaparte, 1991). MACN-Pv-N 120 comes from the Bajo de la Carpa Formation, Río Colorado Subgroup, Neuquén Group, which extends along the west and north of Neuquén, west of Río Negro, and south of Mendoza provinces (e.g., Ramos, 1981; Leanza & Hugo, 2001; Garrido, 2010). The Bajo de la Carpa Formation conform ably overlies the Plottier Formation and is con formably covered by the Anacleto Formation. The Bajo de la Carpa Formation reaches up to ~145 m of thickness in some areas and is main ly composed of coarse-grained, light red to pink sandstones intercalated by thin reddish silt stones and claystones (Leanza & Hugo, 1997; Garrido, 2010). Paleosols, siliceous geodes, chemical nodules, and rain drops marks are very abundant throughout the unit (Rodríguez et al., 2007). Most sediments of this unit were depos ited in low-sinuosity fluvial paleoenvironments (Garrido, 2010); however, some layers at the “Boca del Sapo” locality were interpreted as ae olian in origin (Sánchez et al., 2006; Rodríguez et al., 2007; Filippi et al., 2015). The Bajo de la Carpa Formation was deposited under warm and semiarid climatic conditions (Garrido, 2010).
The fossil tetrapod content of this unit is highly diverse, but with conspicuous differenc es along the fossil-bearing outcrops distributed throughout the Neuquén Basin (e.g., Bonaparte, 1991; Garrido, 2010; Porfiri et al., 2018). The “Boca del Sapo” locality in Neuquén Province, where MACN-Pv-N 120 was found, and the “Paso Córdoba (or Paso Córdova)” locality, near General Roca City, at Río Negro Province, share a similar fauna, with similar facies of possibly coeval age. Both localities yielded no tosuchian mesoeucrocodylians (Notosuchus terrestris, Comahuesuchus brachybucalis, and baurusuchids-Wargosuchus australis in “Boca del Sapo” and Baurusuchidae indet. in “Paso Córdoba”-; Woodward, 1896; Gasparini, 1971; Bonaparte, 1991; Martinelli, 2003; Pol, 2005; Martinelli & Pais, 2008; Leardi et al., 2018), snakes (Dinilysia patagonica; Woodward, 1901; Bonaparte, 1991), and small-sized theropod dino saurs (the abelisauroid Velocisaurus unicus and alvarezsaurids-Alvarezsaurus calvoi in “Boca del Sapo” and Achillesaurus manazzonei in “Paso Córdoba”-; Bonaparte, 1991; Martinelli & Vera, 2007; Brissón Egli et al., 2016). The “Boca del Sapo” locality also includes the mesoeucrocody lians Microsuchus schilleri (Woodward, 1896; Leardi et al., 2015), Neuquensuchus universitas (Fiorelli & Calvo, 2007; Lio et al., 2018), and peirosaurids (Fiorelli, 2010), the enantiornithine Neuquenornis volans (Chiappe & Calvo, 1994) and egg-clutches with embryos referred to this clade (Schweitzer et al., 2002), and the ornithu romorph Patagopteryx deferrariisi (Alvarenga & Bonaparte, 1992). The mesoeucrocodylian Cynodontosuchus rothi might have been found at this locality (Woodward, 1896), but a precise location is unknown. A caudal vertebra of a me dium-sized indeterminate abelisaurid (Ezcurra & Méndez, 2009) was reported as possibly recov ered at the “Paso Córdoba” locality. This spec imen aside, the faunal content of the “Boca del Sapo” and “Paso Córdoba” localities is almost entirely composed of small-sized animals.
In other outcrops of the Bajo de la Carpa Formation, the aforementioned species are not yet recovered but instead its tetrapod content includes titanosaurs (Bonitasaura salgadoi, Rinconsaurus caudamirus, Traukutitan eo caudata, Overosaurus paradasorum; Calvo & González Riga, 2003; Apesteguía, 2004; Juárez Valieri & Calvo, 2011; Coria et al., 2013), meg araptoran (Tratayenia rosalesi; Porfiri et al., 2018) and abelisaurid (Viavenator exxoni and Llukalkan aliocranianus; Filippi et al., 2016; Méndez et al., 2018; Gianechini et al., 2021) thero pods, ornithopods (Mahuidacursor lipanglef; Jiménez-Gomis et al., 2018; Cruzado-Caballero et al., 2019), dinosaur eggshells (Calvo et al., 1997; Garrido & Calvo, 2004; Cruzado-Caballero et al., 2016), mesoeucrocodylians (the peirosaurids Gasparinisuchus peirosauroides, Kinesuchus overoi, and Barrosasuchus neuquenianus; Martinelli et al., 2012; Fillippi et al., 2018; Coria et al., 2019), and turtles (Lomalatachelys neuqui na; Lapparent de Broin & de la Fuente, 2001). The conspicuous faunal differences between lo calities can be related to different depositional settings along the basin and/or different ages of deposition (Sánchez et al., 2006; Rodríguez et al., 2007; Garrido, 2010; Filippi et al., 2015). The age of the Bajo de la Carpa Formation is constrained to the Santonian Stage of the Upper Cretaceous (Bonaparte, 1991; Hugo & Leanza, 2001; Leanza et al., 2004; Rodríguez et al., 2007; Garrido, 2010).
Computed Tomography
MACN-Pv-N 120 was scanned at the Servicio de Microtomografía of the Facultad de Odontología of the Universidad de Buenos Aires, using a CT scanner Sky Scan 1272. The micro-computed tomography (μCT) dataset has a total of 1502 slices with a voxel size of 10.67038 μm; it was acquired with a voltage of 90 kV and current of 111 μA. The visualization and 3D ren derings were performed using 3DSlicer software.
Phylogenetic analysis
The phylogenetic relationships of MACN-Pv-N 120 within Squamata were tested using the phylogenetic dataset published by Gauthier et al. (2012) and subsequently modified by Simões et al. (2015a, b). MACN-Pv-N 120 was added to this data set, resulting in a data matrix com posed of 610 morphological characters scored across 197 species-level terminals (Supporting Information). This matrix was analyzed with a fully unconstrained topology and alternatively using a constrained topology following a total evidence-based backbone (hereafter called un constrained and constrained analysis/trees, re spectively). The total evidence phylogenetic hy pothesis follows that of Reeder et al. (2015: fig. 1), in which the extant major clades of Squamata have the following interrelationships: ((Gekkota + Dibamidae) + (Scincoidea + (Lacertoidea + (Serpentes + (Anguimorpha + (Pleurodonta + Acrodonta)))))). We constrained the monophyly of each of these clades, as well as the interrela tionships among them.
In both cases, the data matrix was analyzed under equally weighted parsimony using TNT 1.5 (Goloboff et al., 2008; Goloboff & Catalano, 2016). The search strategies started using a com bination of the tree-search algorithms Wagner trees, TBR branch swapping, sectorial searches, Ratchet and Tree Fusing, until 100 hits of the same minimum tree length were achieved. The best trees obtained were subjected to a final round of TBR branch swapping. This tree search strategy was designed to find the maximum num ber of most parsimonious trees (MPTs) spread across the widest possible swathe of tree space (see discussion in Langer et al., 2017), ensuring the finding of optimal topologies 100 times. By contrast, other heuristic tree search strategies using a determined number of replications may result in a considerably low number of hits of global optimal trees, not ensuring an optimal hit per replicate, and, thus, the risk of non-find ing all possible MPTs. Zero length branches in any of the recovered most parsimonious trees were collapsed. The non-squamate lepidosaur Gephyrosaurus bridensis was used to root the trees and the following multistate characters were ordered (=additive) based on previous anal yses (Simões et al., 2015b): 3, 7, 10, 12, 18, 25, 33, 38, 39, 41, 43, 45, 48, 49, 51, 56, 58, 63, 65, 66, 67, 70, 80, 82, 83, 84, 90, 93, 97, 99, 101, 102, 105, 106, 108, 111, 114, 120, 126, 128, 129, 130, 132, 133, 140, 141, 143, 149, 152, 155, 167, 168, 178, 182, 184, 185, 187, 188, 189, 203, 204, 208, 216, 217, 220, 223, 231, 238, 242, 248, 250, 251, 256, 258, 260, 263, 268, 271, 276, 277, 283, 285, 288, 300, 301, 302, 303, 306, 309, 311, 312, 316, 326, 328, 337, 340, 343, 346, 347, 349, 350, 360, 361, 364, 368, 369, 372, 375, 382, 388, 389, 390, 392, 394, 396, 414, 415, 418, 419, 420, 421, 435, 440, 454, 455, 456, 457, 458, 459, 460, 463, 468, 475, 477, 483, 486, 487, 488, 518, 529, 535, 570, 572, 584, 588, 589, 590, 593, and 602.
As a measure of branch support, decay indi ces (= Bremer support) were calculated (Bremer, 1994), and as a measure of branch stability, a bootstrap resampling analysis (Felsenstein, 1985) was conducted, performing 1,000 pseu doreplications. Both absolute and GC (i.e., differ ence between the frequency whereby the original group and the most frequent contradictory group are recovered in the pseudoreplications; Goloboff et al., 2003) bootstrap frequencies were reported. In order to analyze the effect that a few topo logically unstable terminals may have on Bremer supports, this index was recalculated after the a posteriori pruning of such terminals, which were previously detected in the subsample of subop timal trees with the iterPCR protocol (Pol & Escapa, 2009). Finally, analyses forcing topologi cal constraints were conducted to find the mini mum number of steps necessary to force alterna tive suboptimal positions for MACN-Pv-N 120.
RESULTS
Systematic Paleontology
Reptilia Linnaeus, 1758 [Laurin & Reisz, 2020]
Diapsida Osborn, 1903 [Gauthier & de Queiroz, 2020]
Lepidosauria Haeckel, 1866 [de Queiroz & Gauthier, 2020a]
Squamata Oppel, 1811 [de Queiroz & Gauthier, 2020b]
?Polyglyphanodontia Alifanov, 2000
Paleochelco gen. nov.
LSID. urn:lsid:zoobank.org:act:F8CC7556-9033- 4DD8-93F3-08CB9B64F18
Etymology. “Paleo” in reference to old, antique, from the ancient Greek παλαιός (palaiós), and “chelco”, popular name used for lizards in some places of Central and South America.
Diagnosis. As for the type and only known spe cies.
Type species. Paleochelco occultato.
Paleochelco occultato sp. nov.
LSID. urn:lsid:zoobank.org:act:71A120B7-9B97- 4427-AEF7-296817C9E9F5
Holotype. MACN-Pv-N 120, incomplete anteri or half of skull that includes partial nasals, par tial maxillae, prefrontals, frontals, vomers, left septomaxilla in transverse section, left lacrimal and anterior tip of jugal, and anterior portion of left palatine. The specimen was embedded in a sandstone block with abundant quartz clasts. The nasals and right maxilla are mostly pre served as natural molds (Figs. 2-6).
Type locality and horizon. Campus of the Universidad Nacional del Comahue, north of Neuquén City (Neuquén Province, Argentina), included within the series of outcrops that have been historically called “Boca del Sapo locali ty”; Bajo de la Carpa Formation (Río Colorado Subgroup; Neuquén Group, Neuquén Basin), Late Cretaceous, Santonian (Garrido, 2010).
Etymology. Occultato, from the Latin occul tatum, meaning hidden, because the holotype remained unnoticed in the collections where it is housed for more than 35 years.
Diagnosis. Small lizard (estimated skull length between 2.0-2.5 cm) that differs from other squamates on the basis of the following combi nation of character states: rostrum lacking orna mentation on the external surface of bones; lac rimal bone present; large anterodorsal process of prefrontal with a convex external surface; trans versely wide vomer/palatine contact, forming an almost flat, ventrolaterally facing surface; ante rior tip of jugal reaching the level of the maxillary tooth positions and fully exposed laterally below the orbit; broad and relatively long anteromedi al process of palatine that runs along the lateral margin of the vomer; palatine with a small and pointed lateral process that articulates with the 62
maxilla; teeth with pleurodont implantation and absence of heavy deposits of cementum around the tooth bases.
Description
MACN-Pv-N 120 is represented by a dam aged partial skull with generally weathered bone surfaces (Figs. 2-4). As a result of the latter, it is difficult to discern the limits between bones. The premaxillae are not preserved, thus the tip of the rostrum is unknown. However, the presence of the anterior region of the left maxilla, including part of the posterior edge of the external naris (Fig. 2), allows determining that the rostrum was relatively short anteroposteriorly, strongly tapering anteriorly, and probably shorter than the post-rostral region of the skull. The pre served external surface of the bones is smooth, lacking any sign of ornamentation.
Maxilla. The left maxilla is nearly complete, whereas the right one is strongly abraded, lack ing the alveolar margin and all teeth. In later al view, the maxilla is subtriangular in contour, with distinct anterior, dorsal, and posterior api ces. The external surface of the bone is dorsoven trally convex and slightly anteroposteriorly con cave on its ventral two-thirds (Fig. 2). The shape of the maxilla results in a continuous transverse broadening of the rostrum towards the post-ros tral region in dorsal and ventral views. The max illa is dorsoventrally tall at its anterior half and strongly decreases in height posteriorly (Fig. 2). It is anteroposteriorly long, at least twice its max imum height. The anteroventral and posteroven tral edges of the left maxilla are broken and, as a result, the exact tooth count is unknown, but at least 12 tooth positions are preserved. Small, subcircular foramina of sub-equal size open on the lateral surface of the maxilla approximately 0.9-1.00 mm dorsal to the alveolar margin of the bone and they are generally aligned to each alve olus. There is a larger foramen positioned more anteriorly and dorsally than the others, match ing the size and position of the anterior maxil lary foramen present in several other saurians (e.g., Dilkes, 1998). The facial (= dorsal) process of the maxilla possesses a short contact with the frontal, anteriorly to the nasal-prefrontal suture (Figs. 2,3). The distal portion of the facial process bows medially and, as a result, its external sur face faces dorsolaterally. The contact between the maxilla and the prefrontal is extensive, antero dorsally to posteroventrally oriented, and appar ently slightly interdigitated. The suture with the lacrimal is longer than the latter, straight, and possesses a similar orientation, being inclined ~30 degrees from the coronal plane (Fig. 2). The anterior process of the maxilla is only partially preserved, but it is possible to recognize a small portion of a rounded, well-notched posteroven tral border of the external narial opening. In left lateral view, part of the pulp cavity of the teeth is exposed because of the eroded surface of the alveolar margin (Fig. 2). The posterior process of the maxilla tapers posteriorly and extends dis tinctly beyond the level of the anterior border of the orbit. The medial surface of the maxilla is covered with matrix, but the contact with the palatine is exposed (Fig. 4). In ventral view, the maxilla forms a transversely narrow supradental shelf (=palatal shelf) (Fig. 6C). The μCT images show that the vertical thickness of the shelf is variable, tapering anteriorly and posteriorly, and relatively thick at the area where it contacts the palatine. Also, above the supradental shelf there is a longitudinal groove, which is interpreted as the superior alveolar canal (Fig. 6C).
Septomaxilla. A laminar bone is exposed in cross-section on the broken anterior region of the skull (Figs. 3, 5A). This bone is positioned on the left half of the rostrum, dorsal to the vomers, and ventral to the mold of the nasals. As a result, it is interpreted as a partial left sep tomaxilla. The main portion of the septomaxilla is mainly transverse, with a slight ventromedial to dorsolateral orientation. This region is most ly straight (Fig. 5), contrasting with the strong ly dorsally expanded and vaulted septomaxilla typical of scleroglossans. Thus, the shape of the septomaxilla indicates the presence of a rela tively small vomeronasal organ in Paleochelco occultato, resembling the condition of iguanians and Cretaceous polyglyphanodontians (Gauthier et al., 2012). A short, dorsomedially oriented lamina indicates that the pair of septomaxillae formed a median partial septum. The morphol ogy of the contact between the septomaxilla and other bones cannot be assessed either because of their damage or the region is covered with ma trix.
Nasal. The nasals are poorly preserved. It can not be determined whether they were paired or fused to each other (Fig. 3). The nasal is most ly eroded on the right side, whereas on the left side it is represented by a crystal mold. The na sal seems to have been considerably shorter than the frontal and has a posterior process with a rounded “V”-shaped edge that subdivided the anterior end of the frontal.
Lacrimal. The lacrimal is a splint-like and elon gated bone with an anterodorsal to posteroven tral orientation in lateral view (Fig. 2). This bone forms an extensive oblique suture with the max illa and the external anteroventral border of the orbit. The orbital margin is not laterally raised. The dorsal tip of the lacrimal laterally overlaps the ventralmost region of the prefrontal. At this point, the suture defines a sharp process pro truding to the orbital cavity (Fig. 2). The open ing of the nasolacrimal duct is not preserved or exposed.
Jugal. Only the anteriormost tip of the anteri or process of the left jugal is preserved (Figs. 2, 4). This small portion of the jugal indicates that this bone likely formed most of the ventral bor der of the orbit and its anterior tip reaches the level of the last three preserved maxillary tooth positions. The anterior process of the jugal over laps laterally the posteriormost region of the lac rimal and is well exposed laterally ventral to the orbit, contrasting with the condition in spheno donts and deeply nested polyglyphanodontians (Gauthier et al., 2012). In ventral view, the jugal also has a large contact with the maxilla, extend ing anteriorly up to the level of the tenth pre served tooth (Fig. 4).
Prefrontal. The prefrontal is a relatively large bone that forms most of the anterior border of the orbit (Figs. 2, 3, 5C). It is subtriangular in lateral view and has an anteroposteriorly and dorsoventrally convex external surface (Fig. 2). The prefrontal possesses a tongue-shaped ante rior process that extensively contacts the maxilla ventrally and anteriorly, and the frontal dorsally. The quality of the μCT scan does not allow de termining the degree of overlapping between the prefrontal and adjacent bones. The posterodorsal corner of the prefrontal possesses a small depres sion pierced by a tiny foramen. The posterodor sal region of the prefrontal tapers posteriorly and ends before the level of minimum width of the paired frontals, indicating that the latter bones participated on the external border of the orbit. There is a thin ventral process of the prefron tal that has an extensive anterodorsally to pos teroventrally oriented suture with the lacrimal (Fig. 2). The posterior margin of the prefrontal is widely concave and does not raise laterally.
Frontal. The frontals are also poorly preserved, with most of their external surface eroded (Figs. 2, 3). Nonetheless, they are clearly paired el ements, with a longitudinal suture extended along their entire preserved length (Figs. 3, 5C). They are anteroposteriorly long bones and only slightly transversely constricted around the level of mid-length of the orbit, thus lacking a strong interorbital constriction or supraorbital shelf (Fig. 3). The frontal possesses an anteroposteri orly and a transversely convex external surface. Each frontal finishes anteriorly on a ‘W’-shaped suture, with an anteromedial process that sep arates both nasals along the median line and a transversely broader anterolateral process that it is subdivided by the tapering dorsal process of the maxilla (Fig. 3). The ventral surface of the frontals is completely covered with matrix, but the μCT data shows that the olfactory tract is only incipiently laterally delimited, thus lacking subolfactory processes, and there is no median pillar. There is no evidence of a pineal foramen in the preserved region of the frontals, but this opening was probably present more posteriorly on the skull roof.
Vomer. The vomers are the most extensive ele ments in the preserved region of the palate (Fig. 4). These bones are paired and separated by a narrow longitudinal slit anteriorly and a median suture posteriorly. However, it cannot be ruled out that the apparent lack of contact between vomers on the anterior region of the palate is a result of taphonomic distortion. The anterior half of the vomer is relatively narrow and wid ens posteriorly, which results in a transversely narrow internal choana (Figs. 4, 5D). The area of the opening of the vomeronasal organ is covered with matrix, but the μCt shows that the vomer contacted the palatal shelf of the maxilla. A small foramen, opened posteriorly, is placed at the level of the third preserved tooth, which is interpret ed as the foramen for the medial palatine nerve (Figs. 4, 6A). The eroded anterior surface of the left vomer shows the anterior portion of the ca nal of the medial palatine nerve that is filled with matrix (Fig. 5D). The posterior end of the vomer possesses a V-shaped contact with the palatine, in which the apex is anteriorly oriented and lat erally displaced from the level of mid-width of the hemipalate (Figs. 4, 5D). The right vomer is poorly preserved and the foramen for the medi al palatine nerve seems to be more posteriorly placed than in the left side.
Palatine. The palatine is represented by the anterior portion of the left side element and a damage portion of the right one. It shows a rel atively long and wide anteromedial process that forms an oblique, anterolaterally to posteromedi ally oriented, suture with the vomer (Fig. 4). The palatine possesses a small and pointed lateral process that articulates with the maxilla (Figs. 4, 5D). This lateral process has a well-developed an terior projection that results in a palatine-max illa articulation that extends along the level of three tooth positions. Although incompletely preserved, a small posterolateral projection also exists. The lateral process of the palatine is sep arated from the medial one by a strongly con cave margin that forms the posterior border of the internal choana. The ventral surface of the bone on the intersection between the lateral and anteromedial processes is distinctly depressed into a choanal fossa (Fig. 4B). Unfortunately, the degree of posterior extension of this fos sa cannot be determined because of damage.
Dentition. The dentition is partially preserved on the left maxilla, lacking the tips of all tooth crowns. The teeth have a pleurodont implanta tion and are ankylosed to the lingual side of the bone (Figs. 4, 6). No enamel is preserved and the pulp cavity is exposed in several teeth because of weathering (Figs. 2, 4, 6). The teeth increase in size distally until the ninth-tenth preserved position, and then decrease in size in the last two preserved teeth. The lingual side of each tooth is strongly apicobasally concave, with a labiolin gually wide base about four times labiolingually broader than mesiodistally long (Fig. 6D). This part of the teeth is ankylosed to the maxillary bone. Each tooth is positioned close one to an other, forming a fusiform notch in labiolingual view between successive teeth (Fig. 6). A replace ment pit occurs at the base of some teeth on the lingual side, and they have different degrees of development (Fig. 6). These pits are clearly ob served in the preserved third, fourth, sixth, sev enth, eighth, ninth, and eleventh tooth positions, and the largest of them occurs in the eighth po sition. The preserved fifth tooth position is emp ty and lacks evidence of a functional tooth or a replacement. No further details can be provided about the tooth crown morphology. There is no evidence of palatal teeth in the preserved por tions of vomers and palatines.
Phylogenetic analysis
Unconstrained analysis. The analysis found 10,032 most parsimonious trees (MPTs) of 5,306 steps with a consistency index (CI) of 0.18338 and a retention index (RI) of 0.79249. The topol ogy of the strict consensus tree is consistent to those found by Gauthier et al. (2012) and Simões et al. (2015a, b), but with a massive polytomy at the base of Iguania and unresolved relationships among the main clades of non-iguanian non- “krypteian” squamates. The iterPCR protocol found that the Brazilian species Tijubina pontei and Olindalacerta brasiliensis are highly unsta ble taxa and their a posteriori pruning allowed recovering Polyglyphanodontia as the sister tax on to Scleroglossa + Mosasauria. The new taxon Paleochelco occultato is alternatively recovered at the base of Polyglyphanodontia, as the sister taxon to Polyglyphanodontia + Scleroglossa, or as one of the sister taxa to the Mosasauria + Scleroglossa clade (Fig. 7).
Constrained analysis. The phylogenetic anal ysis using the total evidence-based topological constrain found 19,008 MPTs of 5,464 steps with a consistency index (CI) of 0.17807 and a retention index (RI) of 0.78492. The strict con sensus tree is generally well resolved, but it has a massive polytomy composed of Lacertoidea, Polyglyphanodontia, Acrodonta, Anguimorpha, Pleurodonta, and Serpentes (Fig. 8). Paleochelco occultato is recovered within Polyglyphanodontia, in a trichotomy with Tchingisaurus multivagus and more deeply nested members of the clade. The iterPCR protocol found that the alternative positions of Olindalacerta brasiliensis as an an guimorph or an iguanian among the MPTs are the responsible of the massive polytomy. The strict reduced consensus tree with the a posteri ori pruning of Olindalacerta brasiliensis is well resolved. Tijubina pontei, which was one of the rogue species in the unconstrained analysis, is recovered in a trichotomy with Anguimorpha and the Polyglyphanodontia + Iguania clade (Fig. 8).
Comparisons of Paleochelco occultato with main squamate clades
Paleochelco occultato is easily distinguishable from Scincoidea, Lacertoidea, Anguimorpha, and Paramacellodidae by several features, including the absence of skull ornamentation and palpe bral ossification (see Estes et al., 1988; Evans & Chure, 1998; Conrad, 2008). Paleochelco occulta to differs from iguanians by the absence of pre frontal tuberosity, strong medial process of max illa, fused frontals that are constricted between orbits, and a frontal shelf (Estes et al., 1988; Gao & Norell, 2000; Conrad & Norell, 2007; Conrad, 2008). The new species differs from teioids and resembles polyglyphanodontians in having unfused frontals and lacking retracted external nares and maxillary sculpturing (Estes, 1983a; Denton & O’Neill, 1995; Conrad, 2008). Paleochelco occultato also resembles polyglypha nodontians in lacking sculpturing on prefrontal, linear interorbital margins of frontals (Conrad, 2008), and maxillary tooth row beginning anteri or to mid-orbit level (Gauthier et al., 2012). Thus, a phylogenetic placement of Paleochelco occulta to close to or among polyglyphanodontians is also bolstered by these comparisons.
DISCUSSION
The phylogenetic relationships of Paleochelco occultato
Unconstrained analysis. Paleochelco oc cultato as a non-iguanian non-scleroglossan squa mate is the best supported hypothesis, being re covered as an early Polyglyphanodontia, the sis ter taxon to Polyglyphanodontida + Scleroglossa, or as one of the sister taxa to the Scleroglossa + Mosasauria clade (Fig. 7). When Paleochelco oc cultato is recovered within Polyglyphanodontia, this taxon is found as its earliest branching spe cies, the sister taxon to the Asian Tchingisaurus multivagus, or the sister taxon to all the other members of the clade (i.e., the Asiatic and North American Gobinatus arenosus, Adamisaurus magnidentatus, Polyglyphanodon sternbergi and Gilmoreteius). The placement of Paleochelco oc cultato within Polyglyphanodontia is support ed by the presence of a frontal-maxilla suture, separating nasal from prefrontal (character 37: 0→1), and vomer extending backwards beyond anteriormost contact of palatine with maxilla (character 213: 0→1). The alternative positions of Paleochelco occultato and Tchingisaurus mul tivagus around the base of Polyglyphanodontia are not supported by apomorphies, but they are a result of the lack of overlapping phylogenetically informative characters between these two taxa for this part of the tree. Paleochelco occultato is excluded from more deeply nested polyglyphan odontians because of the absence of a jugal over lapping medially the lacrimal (character 145: 0→1) and a jugal failing to be laterally exposed below orbit (character 149: 2→0).
When Paleochelco occultato is found outside Polyglyphanodontia, its position as the sister taxon of the clade that includes polyglypha nodontians, mosasaurs and scleroglossans is supported by the presence of paired frontals (character 36: 1→0), contact between frontal and maxilla, separating nasal from prefrontal (character 37: 0→1), and septomaxilla with me dial flange (character 205: 0→1). By contrast, the exclusion of Paleochelco occultato from the Polyglyphanodontia + Scleroglossa clade is not supported by any apomorphy. The placement of Paleochelco occultato closer to scleroglossans than to polyglyphanodontians is not supported by any apomorphy in our analysis, but its exclu sion from the Mosasauria + Scleroglossa clade is a result of the absence of a septomaxilla dorsally expanded and convex, reflecting a large size of the vomeronasal organ (character 200: 0→1). In some other MPTs, the presence of a clade com posed of Paleochelco accultato, Tijubina pontei (Early Cretaceous of Brazil), and (in some oth er trees) with Olindalacerta brasiliensis (Early Cretaceous of Brazil) is not supported by any apo morphy and the position of this South American clade as the sister taxon to the Mosasauria + Scleroglossa clade is based on the presence of an ectopterygoid with an obtuse angulation in dor sal view (character 272: 0→1) and ectopterygoid posterior process developed as a small lateral knob (character 283: 0→1) -but both character states are unknown in Paleochelco occultato-. In some MPTs, Paleochelco occultato is recov ered closer to the Mosasauria + Scleroglossa clade than to Tijubina pontei and Olindalacerta brasiliensis, or alternatively the latter two taxa are closer to that clade. These positions are not supported by apomorphies in our analysis.
The branch supports around the alternative phylogenetic positions that Paleochelco occulta to acquires among the MPTs are very low, with a minimum Bremer support and bootstrap fre quencies below 20% for the clade that includes all squamates with the exception of iguanians. The a posteriori exclusion of Tijubina pontei and Olindalacerta brasiliensis, as suggested by the iterPCR protocol, slightly increases the re sampling frequencies of this branch, but still ≤30%. In this strict reduced consensus tree, the branch supports of Polyglyphanodontia, without Paleochelco occultato and Tchingisaurus mul tivagus, are high, with a Bremer support of 3 and bootstrap frequencies ≥70%. By contrast, the clade that includes mosasaurs and scleroglossans has very low branch supports.
Under constrained suboptimal topologies, three additional steps are necessary to place Paleochelco occultato at the base of Iguania (with out synapomorphies supporting this hypothesis), one step to place it at the base of Mosasauria (sup ported by the absence of a nasal-prefrontal con tact; character 19: 0→1), and two steps to be the sister taxon to or to be placed within Scleroglossa (without synapomorphies supporting the first hypothesis and the presence of a vomer extend ing backwards beyond anteriormost contact of palatine with maxilla as a synapomorphy of Paleochelco occultato + Autarchoglossa; char acter 213: 0→1). In particular, four extra steps are required to force a sister taxon relationship between Paleochelco occultato and the iguanian Gueragama sulamericana (mid-Cretaceous of Brazil) and five steps to find the new taxon as the sister species to Paramacellodus oweini (from the Jurassic-Cretaceous boundary of North America and Europe and very closely related to Neokotus sanfranciscanus from the late Early Cretaceous of Brazil). Both positions are not supported by synapomorphies, and it is very interesting to re port that this constrain generates Gueragama sulamaricana to be positioned as a sister taxon to the Mosasauria + Scleroglossa clade, thus out side Iguania.
Constrained analysis. The position of Paleochelco occultato in this analysis is congruent with that in some of the unconstrained MPTs; i.e., as one of the earliest branching polyglyphan odontians (Fig. 8). The placement of Paleochelco occultato within Polyglyphanodontia and outside the clade composed of more deeply nested poly glyphanodontians is supported by the same syn apomorphies as in the unconstrained analysis. After the a posteriori pruning of Olindalacerta brasiliensis, the calculated branch supports of Polyglyphanodontia, including Paleochelco occul tato and Laurasian forms, are relatively low, with a Bremer support of 2 and bootstrap frequencies below 30%. By contrast, the clade that includes polyglyphanodontians more deeply nested than Paleochelco occultato and Tchingisaurus mul tivagus is better supported, with a Bremer sup port of 3 and absolute and GC bootstrap frequen cies of 89% and 87%, respectively.
Under constrained suboptimal topologies (in addition to the original backbone), five addition al steps are necessary to place Paleochelco occul tato as a pleurodontan iguanian (supported by the presence of a maxillary facial process with a dorsolaterally facing apical surface, character 116: 0→1; and a frontal contacting maxilla, sepa rating nasal from prefrontal, character 37: 0→1), six steps to place it as an acrodontan iguanian (without synapomorphies supporting this hy pothesis), four steps to place it as an anguimorph or as the sister taxon to the clade composed of Mosasauria + Serpentes (without synapomor phies supporting these hypotheses), two steps to place it as a lacertoid (without synapomorphies supporting this hypothesis), three steps to place it as a gekkotan, closer to scincoidean than to other squamates, or as a paramacellodid (with out synapomorphies supporting the first two hypotheses and the presence of a jugal entirely exposed laterally above orbital margin of max illa, character 149: 0/1→2, supporting its place ment as the sister taxon to the paramacellodid Parmeosaurus scutatus).
Conclusions about the phylogenetic re lationships of Paleochelco occultato. The positions of Paleochelco occultato as a non-igua nian non-scleroglossan squamate in the uncon strained analysis and as a polyglyphanodontian in the constrained analysis are moderately well supported, respectively (Figs. 7-8). The affinities of Paleochelco occultato with polyglyphanodon tians remain ambiguous in the unconstrained analysis, as well as the possible presence of a South American clade composed of the new taxon and the Brazilian Tijubina pontei and Olindalacerta brasiliensis. Both possible hy potheses have very interesting biogeographic implications. If the affinities of Paleochelco oc cultato with polyglyphanodontians are bolstered in the future, it would imply the presence of a Gondwanan representative of an otherwise Laurasian clade, as it has been previously sug gested with the possible polyglyphanodontian affinities of the Brazilian Tijubina pontei and Olindalacerta brasiliensis (Simões et al., 2015a). Alternatively, if Paleochelco occultato, Tijubina pontei, and Olindalacerta brasiliensis form a clade to the exclusion of other squamates (only recovered in some MPTs of the unconstrained analysis), it would represent a South American or Gondwanan endemism of a Mesozoic squa mate group with a completely unknown previous evolutionary history.
Paleochelco occultato and its implications on the fossil record of lizards
As indicated by previous authors (e.g., Evans, 2003; Benson et al., 2013; Cleary et al., 2018), the main obstacle for understanding the early evo lution and origins of different squamate lizard clades is probably the poorly documented fossil record in the Southern Hemisphere. Despite the paucity of this record, the growing fossil record of lepidosaurian taxa highlights a hidden history of the group in Gondwana (e.g., Bonaparte et al., 2010; Simões et al., 2015a, b; Bittencourt et al., 2020; Romo de Vivar et al., 2020).
Early to Middle Jurassic lizards are known from India (Evans et al., 2002) and a paramacel lodid scincomorphan has been described from the Late Jurassic of Tanzania (Broschinski, 1999). The Early Cretaceous record of Gondwanan liz ards is restricted to Brazil, with three different taxa of uncertain phylogenetic position, prob ably representing stem or early scleroglossans (Simões et al., 2015a, 2017). The Late Cretaceous fossil record of lizards in Gondwana counts with the occurrence of acrodontan iguanians in Africa (Apesteguía et al., 2016), scincoids in Madagascar (probably cordylids; Krause et al., 2003), and some remains from South America, including paramacellodids and acrodontan iguanians from Brazil (Simões et al., 2015b; Bittencourt et al., 2020). From the Santonian- Campanian of Brazil, Nava & Martinelli (2011) described Brasiliguana prudentis based on a maxilla as belonging to Iguania because of the weakly inclined anterior margin of the maxillary nasal process and the pleurodont tooth implanta tion. Further, Albino & Brizuela (2014:351) men tioned that the maxilla of Brasiliguana pruden tis does not present a strong/large palatine pro cess, similar to the condition of Corytophaninae, Polychrotinae, Iguaninae, and Hoplocercinae. However, Brasiliguana prudentis shares with Polyglyphanodontidae the presence of waist ed teeth with replacement pits at its base, and shares with chamopsiids a subpleurodont tooth attachment, a barrel-like tooth crown, and the subcircular cross-section of the teeth (Woolley et al., 2020). Considering these features, non-igua nian affinities can be proposed for Brasiliguana prudentis. If affinities with chamopsiids are bol stered in the future (or with polyglyphanodon tians considering their phylogenetic proximity), Brasiliguana prudentis and Paleochelco occulta to could be nested among non-iguanian lizards, highlighting the diversity of still poorly known clades in the Late Cretaceous of South America.
Estes & Price (1973) erected Pristiguana brasiliensis based on some cranial and postcra nial remains from the Maastrichtian of Brazil. They remarked features similar to teiioid lizards but interpreted them as plesiomorphies. Because of that, Borsuk-Białynicka & Moody (1984; see also Benson et al., 2013) considered that the ar guments for the iguanid assignment were not convincing and that Pristiguana brasiliensis could be assigned to the Teiidae. However, the combination of character states of Pristiguana brasiliensis sustains its iguanian affinities, as supported by other studies (e.g., Apesteguía et al., 2005; Nava & Martinelli, 2011; Daza et al., 2012; Albino & Brizuela, 2014; Martinelli & Teixeira, 2015; Simões et al., 2017). In sum, the fossil record of Brazil is composed of early forms (scincomorphans or stem or very early sclero glossans), scincomorphs (paramacellodids), and both acrodontan and pleurodontan iguanians.
In contrast, the record of Cretaceous lizards in Argentina is restricted to a couple of poorly informative specimens. An isolated iguanian frontal was reported from the Cenomanian of Río Negro Province (Apesteguía et al., 2005). Daza et al. (2012) indicate that the charac ter state combination listed by Apesteguía et al. (2005) is not exclusive to any lizard group. However, as sustained by Albino & Brizuela (2014), the simultaneous presence of frontals with an hourglass-shaped contour determined by concave lateral margins, dermal sculpturing, transversely concave dorsal surface, and exten sive supraorbital flanges are exclusively present in iguanians (see Apesteguía et al., 2005; Smith, 2009). In addition to this specimen, still unpub lished remains from the same locality indicate that lizards were diverse and include not only iguanians, but probably other different clades (Apesteguía et al., 2019).
A poorly preserved dentary from the Campanian of northern Patagonia was referred to the Scincomorpha by Brizuela & Albino (2011); they noted special affinities with scin coids. Together with the early paramacellodids from Brazil (Bittencourt et al., 2020), this den tary constitutes the only unambiguous sclero glossan (if it is actually a monophyletic clade) specimen from the Mesozoic of South America (Albino & Brizuela, 2014). To this meager fossil record, we add the stem-scleroglossan or proba bly polyglyphanodontian (depending on the phy logenetic framework used) Paleochelco occultato.
Because squamates have a long fossil histo ry that extends into the Triassic (Simões et al., 2018) and most of our knowledge of their early evolutionary history is derived from Laurasian records, several authors proposed that most lizard clades originated in the Northern Hemisphere and then arrived to South America during a possible biotic interchange with North America during the Campanian-Maastrichtian span (Borsuk-Białynicka & Alifanov, 1991; Alifanov, 1993, 1994, 2016; Gao & Hou, 1996). In this sense, the Cretaceous occurrence of polyglyphanodontians in North America (which were related to teiioids in morphological analy ses; Estes, 1969, 1983b; Gao & Fox, 1996; Gao & Norell, 2000; Nydam et al., 2007) contributed to the hypothesis that South American teiioids were late surviving members of a Cretaceous teiioid migration from North into South America (Savage, 1966; Presch, 1974; Estes, 1983b; Estes & Báez, 1985; Nydam et al., 2007; Giugliano et al., 2007). However, the alternative phylogenetic relationships of Paleochelco occultato, together with recent findings in other Cretaceous fos sil sites from diverse Gondwanan landmasses, suggest that the early squamate faunas of the Southern Hemisphere were well integrated with faunas from other parts of the world. In this sense, phylogenetic data suggests that Toxicofera or Scleroglossa (depending on the phylogenetic hypothesis followed) separated before the break up of Pangea, reinforcing the idea of an early radiation and wide distribution for most lizard clades (Simões, 2012; Bittencourt et al., 2020), as recently advocated for acrodontan iguanians (Simões et al., 2015b; Apesteguía et al., 2016). The recent findings of scincoids, acrodontans, pleurodontans, and stem scincomorphans also sustains that Gondwana was probably a major theater for the evolution of several lizard clades.
In contrast to a Mesozoic fossil record main ly composed of worldwide distributed taxa, the early Neogene to present squamate diversity in South America is characterized by taxa that are almost exclusive to that continent, including tei ids, gymnophthalmids, tropidurids, liolaemids, leiosaurids, amphisbaenids, anomalepidids, and micrurids (e.g., Bittencourt et al., 2020). The absence of such clades in the Mesozoic record in the continent may be indicative of a post-Creta ceous explosive radiation and diversification that followed the mass extinction event (Longrich et al., 2012), as postulated for birds and mam mals (e.g., Feduccia, 2003; O’Leary et al., 2013). Alternatively, this scenario would be the result of a still extremely poor Cretaceous squamate re cord in South America (see below).
Paleochelco occultato and the extinction of sphenodonts
Traditionally, the tuataras (Sphenodon punc tatus) have been seen as a “living fossil” or rel ic of ancient times because sphenodonts had a highly diverse evolutionary history with an ex quisite fossil record during the Mesozoic (e.g., Sues, 2019). However, the geographical range of sphenodonts seems to become restricted after the Middle Jurassic, first in Laurasia and later in Gondwana, being especially well-represented in Cretaceous beds of South America (Evans et al., 2001; Apesteguía & Novas, 2003; Apesteguía, 2005; Jones, 2006; Apesteguía & Rougier, 2007; Apesteguía & Jones, 2012). This contraction was considered by some authors as the result of com petition with squamate lizards (Carroll, 1988; Milner et al., 2000; Zug et al., 2001). Particularly for South America, it was proposed that the arriv al of modern lizards coming from North America through Central America resulted in the compet itive displacement of sphenodonts (Apesteguía & Novas, 2003; Apesteguía & Rougier, 2007; Apesteguía & Jones, 2012). However, the dis covery of “Laurasian advanced lizards” in South America since Early Cretaceous times (Simões, 2012; Nava & Martinelli, 2011; Simões et al., 2015a,b, 2017; Bittencourt et al., 2020) indicates that sphenodonts and “modern” lizard clades co existed for over 50 million years, thus making un likely the “competitive displacement hypothesis” for explaining the disappearance of sphenodonts around the world. In this scheme, the presence of Paleochelco occultato in the Late Cretaceous of Patagonia reinforces the idea of Simões et al. (2015b) that sphenodontians were not necessari ly dominant, at least in terms of taxonomic diver sity, over squamates in South America as previ ously hypothesized. Perhaps the extremely poor Mesozoic lizard record in Gondwana is a result of biases in sampling because sphenodontians usually have more robust-built skulls, favoring its preservation in some fossil assemblages.
Supplementary Information. Phylogenetic data matrices in NEXUS and TNT formats availa ble at http://morphobank.org/permalink/?P4059.