Chubutemys copelloi is the oldest nonmarine cryptodire from South America represented by a skull. The skull and associated postcranial fragments are from the Aptian Cerro Costaño Member of the Cerro Barcino Formation of Chubut, Argentina. Chubutemys has a processus trochlearis oticum, showing that it is a cryptodire, and an enclosed canalis caroticus internus extending to the posterior margin of the pterygoid, showing that it is a eucryptodire. The skull of Chubutemys is similar to that of other primitive eucryptodires, particularly Dracochelys, but also to Hangaiemys, Judithemys, Sinemys, and Ordosemys. Chubutemys differs from all these, however, in possessing a solidly roofed skull, formed by long, wide parietals, rather than a posterior emargination. Chubutemys also differs from these taxa in having no cheek emargination.
A phylogenetic analysis using PAUP* analyzed 104 parsimony-informative characters resolving into one most parsimonious cladogram of 224 steps, a consistency index of 0.55, and a retention index of 0.74. The phylogenetic analysis weakly joins Chubutemys and meiolaniids on the basis of the prefrontal-postorbital contact. Chubutemys also has a fully roofed skull and slitlike posterior opening of the foramen caroticum laterale (foramen posterius canalis caroticus laterale), features to be expected in a meiolaniid sister taxon. Chubutemys provides further evidence that meiolaniids are related to “basal” eucryptodires (“sinemydids/macrobaenids”), that is, eucryptodires outside the living Cryptodira, the Polycryptodira.
The basicranial morphology of meiolaniids, with an intrapterygoid slit, rather than being a unique feature of the group is instead a modified state of the primitive eucryptodire condition, as seen in such forms as Chubutemys, Dracochelys, Ordosemys, and Sinemys. The intrapterygoid slit of meiolaniids is homologous with the pterygoid flange associated with the foramen caroticum laterale (foramen posterius canalis caroticus laterale of Sukhanov) in non-Polycryptodiran eucryptodires like Ordosemys.
Chubutemys shows that nonmarine eucryptodires were present in South America in the Cretaceous, as they were in North America, central Asia, and Australia.
Introduction
The record of nonmarine cryptodires in South America is very sparse, particularly in comparison to the more abundant pleurodires. There is a record of marine cryptodires from the Jurassic, but the nonmarine record has been restricted to the eucryptodiran Meiolaniidae from the late Cretaceous–Paleogene. Recent vertebrate paleontological field explorations (Rich et al., 1998) in the Early Cretaceous of the Province of Chubut, Argentina, have yielded the oldest skull of a South American cryptodire and the oldest record of a nonmarine cryptodire from South America. The only older cryptodire is the late Jurassic (Tithonian) Neusticemys neuquina (Fernandez and Fuente, 1993), a presumably marine form attributed to the “Plesiochelyidae-Thalassemydidae” by Gasparini and Fernandez (1997) and the “?Plesiochelyidae” by Fernandez and Fuente (1993). Neusticemys neuquina occurs with the pleurodire Notoemys. The skull and cervical vertebrae of Neusticemys neuquina are unknown and thus its more precise relationships are indeterminate at present. 8Broin (1993) listed some Jurassic localities that have yielded turtles, none identified as nonmarine cryptodires. Broin and Fuente (1993b: 182) also listed “chelidae primitifs” from the Rio Limay Formation, thought to be Cenomanian-Turonian, but the basis of the identification is unknown and these must be considered Testudines indet. until described.
The Albian Santana Formation (Maisey, 1990, 1991) has yielded a protostegid marine cryptodire (Hirayama, 1998) that is probably later in time than the Chubut cryptodire. The University of California has a skull and partial skeleton (UCMP 38346) of an undescribed chelonioid from the vicinity of Bogota, Colombia (V 4538), that may be Aptian-Barremian in age (Welles, 1962) and thus as old as or older than the Chubut cryptodire. However, this material has never been described and its age is not substantiated.
Later records of cryptodires in South America that postdate the Aptian Chubut cryptodire consist of meiolaniids:
Los Alamitos Fm., Rio Negro Province, Argentina, Campanian-Maastrichtian, described by Broin (1987).
Salinas de Trapalco, Rio Negro Province, Argentina, Campanian-Maastrichtian, undescribed fragments reported by Broin and Fuente (1993a,b).
Sarmiento Fm., Chubut Province, Argentina, Eocene, tail ring described by Fuente and Fernandez (1992).
Niolamia argentina, Cretaceous or Eocene, Chubut Province, complete skull, see Gaffney (1996) for most recent appraisal.
Another record, identified as a pleurodire: “grand Chelidae primitif” by Broin and Fuente (1993b, see Discussion below), is probably a cryptodire similar to Chubutemys described here.
Although cladistic analysis has been the norm for phylogenetic work (at least in morphology) for decades, recently some have argued that there are limitations in the Linnaean classification system that require its complete abandonment in favor of an alternative, the PhyloCode, incorrectly called by some, “Phylogenetic Taxonomy” (De Queiroz and Gauthier, 1992). Although we do see some problems with the Linnaean system as an exact mirror of a cladogram, with taxon names that can change in content with every addition or subtraction of a basic taxon or character, we do not think that abandonment and creation of a new system is the answer. We take a more traditional view, similar to that expressed by Gaffney et al. (2006; see also Nixon and Carpenter, 2000; Carpenter, 2003; Nixon et al., 2003; Schuh, 2003), that it is better to modify the Linnaean system where convenient, rather than set up a new one, particularly a rankless one. An extreme example of the abandonment of the more traditional system of turtle classification is the Joyce et al. (2004) version of the PhyloCode (q.v.) in which most of the previously named higher taxa of turtles, including the name Cryptodira, have been replaced or redefined with new content. Although this exercise claimed to improve stability in turtle nomenclature, it is obvious that the opposite is the case. In fact, the two proponents of new higher category definitions and names for turtles (Lee, 1995; Joyce et al. 2004) do not themselves agree on the taxon names or definitions. Another classification aspect is the use of geological range in naming groups, the “stem” and “crown” nomenclature, referring to whether taxa include living species. We think that since extinction is not an element of the cladogram, it should not be an element of the classification, which is cumbersome enough without the added burden of geological range. Therefore, we use the higher taxon names for cryptodires explicated in Gaffney (1996) for the taxa that are relevant in this discussion of Chubutemys.
The anatomical terminology used here is developed in Gaffney (1972, 1979). The term “foramen caroticum laterale” as currently used actually refers to both a more anterior foramen opening into the braincase floor and a more posterior foramen opening on the roof of the palate, with the canalis caroticus lateralis (Gaffney, 1979 and others) connecting them. Sukhanov (2000) used the term “foramen posterius canalis caroticus laterale” for the posterior foramen, and we have adopted that usage here.
The Chubut cryptodire is most similar to a series of largely Mesozoic eucryptodires usually grouped as the family Sinemydidae (Gaffney, 1996; Sukhanov, 2000), but which may also be interpreted as errorsuccessive, paraphyletic, sister taxa to the Polycryptodira of Gaffney (1984, 1996). As has been done in previous literature (Brinkman and Peng, 1993a, 1993b; Brinkman and Wu, 1999; Gaffney, 1996; Gaffney and Ye, 1992; Parham and Hutchison, 2003; and Sukhanov, 2000), we refer to these early eucryptodires as “Sinemydidae/Macrobaenidae” for convenience and to indicate the inconsistent nature of phylogenetic analyses of these taxa. Earlier works that have reviewed some of these taxa are Brinkman and Peng (1993a, 1993b), Brinkman and Wu (1999), Gaffney and Ye (1992), Gaffney (1996), Parham and Hutchison (2003), and Sukhanov (2000). Specific taxa that have skulls and/or postcrania particularly relevant to the description of the Chubut cryptodire and that are referred to in this text are:
Dracochelys: Gaffney and Ye (1992), Brinkman (2001).
Ordosemys: Brinkman and Peng (1993a), Brinkman and Wu (1999).
Xinjiangchelys: Peng and Brinkman (1993b), Kaznyshkinet al. (1990).
Hangaiemys: Sukhanov and Narmandakh (1974), Sukhanov (2000), Parham and Hutchison (2003).
Sinemys: Brinkman and Peng (1993b).
Meiolaniidae (including Meiolania, Ninjemys, Warkalania, and Niolamia): Gaffney (1983, 1996).
Otwayemys: Gaffney et al. (1998). See Smith (1999) for another possible Australian eucryptodire from the early Cretaceous.
Judithemys: Gaffney (1996, as TMP 87.2.1), Parham and Hutchison (2003). Skull figured in fig. 13.
Macrobaena: Tatarinov (1959).
Anatomical Abbreviations
bo
basioccipital
bs
basisphenoid
cci
canalis caroticus internus
ex
exoccipital
fcb
foramen caroticum basisphenoidale
fpcci
foramen posterius canalis carotici interni
fpcl
foramen posterius canalis caroticus laterale
fpp
fenestra palatinum posterius
fr
frontal
ju
jugal
mx
maxilla
op
opisthotic
pa
parietal
pal
palatine
pf
prefrontal
po
postorbital
pt
pterygoid
qj
quadratojugal
qu
quadrate
so
supraoccipital
sq
squamosal
vo
vomer
Systematics
Order Testudines Linnaeus, 1758
INFRAORDER CRYPTODIRA COPE, 1864
PARVORDER EUCRYPTODIRA GAFFNEY, 1975
FAMILY INDETERMINATE (“SINEMYDIDAE/MACROBAENIDAE”)
Chubutemys, new genus
Diagnosis
A genus of eucryptodire known only from the skull, similar to Dracochelys, Hangaiemys, and Ordosemys; nasal-frontal contact absent in contrast to Ordosemys; frontal does not enter orbit in contrast to Dracochelys, Ordosemys, Hangaiemys, and Sinemys; temporal roof broadly covered by parietals extending posterior to otic chamber in contrast to the deep emargination seen in Hangaiemys, Judithemys, Ordosemys, and Sinemys (and probably Dracochelys, although the temporal roof is incompletely known); broad parietal-squamosal contact in contrast to Dracochelys, Ordosemys, Hangaiemys, and Sinemys; cheek emargination absent in contrast to Dracochelys, Hangaiemys, Judithemys, Ordosemys, and Sinemys; triturating surface narrow as in Ordosemys and Sinemys, but not very narrow as in Dracochelys and Hangaiemys; foramen palatinum posterius similar in size to that in Ordosemys, not very large as in Dracochelys and Hangaiemys; deep pit on pterygoid absent in contrast to Dracochelys and Ordosemys; foramen posterius canalis caroticus laterale exposed in ventral view, widely separated from foramen caroticum basisphenoidale, and formed in gap between pterygoid and basisphenoid as in Dracochelys and Niolamia, in contrast to more posterior position as in Hangaiemys, Judithemys, Sinemys, and Ordosemys; canalis caroticus internus only partially floored by bone along contact of basisphenoid and pterygoid as in Dracochelys, Hangaiemys, Ordosemys, and Sinemys; foramen caroticum basisphenoidale formed entirely by basisphenoid and open ventrally as in Dracochelys, Ordosemys, Hangaiemys, and Sinemys; basisphenoid triangular in ventral view as in Dracochelys and Niolamia, not rectangular as in Ordosemys and Hangaiemys.
Chubutemys copelloi, new species
Type Specimen
MPEF-PV1236, a partial skull lacking much of the bone from its dorsal surface, but with an internal mold of it present, ventral surfaces preserved intact with stapes in situ, atlas elements disarticulated, and a piece of presumed hyoid in right temporal fossa.
Type Locality
“Turtle Town”, 43°36′S, 68°55′W (figs. 1–Fig. 2.3; Rich et al., 1998: fig. 1), Chubut Province, Argentina.
Horizon
Cerro Castaño Member (fig. 3), Cerro Barcino Formation, Chubut Group, Aptian (Codignotto et al., 1978; Rich et al., 1998; Rauhut et al., 2003) based on the work of two of the authors, T. Rich and A. Constantine.
Referred Material
The locality that yielded the skull of Chubutemys also coughed up enough turtle remains to cause its giddy discoverers to call it “Turtle Town” (figs. 1–Fig. 2.3; Rich et al., 1998: fig. 1). The fossils from the site consist of shell and postcranial fragments of cryptodires and pleurodires, as well as the skull described here. No shell or postcranial material was found directly associated with the type skull of Chubutemys. However, there are postcranial elements that probably belong to Chubutemys. The cryptodire material is identifiable because it consists of a partial carapace with anterior thoracic centra and an articulated eighth cervical, as well as peripherals, including bridge peripherals, that are very similar to the “sinemydid/macrobaenid” postcrania described for Dracochelys, Ordosemys, Hangaiemys, Judithemys, and Otwayemys. The shell fragment with an eighth cervical, MPEF PV1940, is identified as a cryptodire on the basis of the widely spaced zygapophyses and the close agreement of this cervical with those of Ordosemys and Dracochelys (see below). Three sets of peripherals are also identified as cryptodire and probably belong to Chubutemys. These are MPEF PV1941, PV1942, and PV1943. They are very similar to peripherals figured by Brinkman (2001) for Dracochelys. They are relatively flat, show extensive costo-peripheral fontanelles, and have dorsally concave guttering as in Dracochelys.
There is at least one other species of turtle at this locality, a chelid pleurodire much smaller than Chubutemys. Most of these chelid bones differ from the cryptodire in being at least half the size and in having a very differently shaped shell with fused pelvis, no costo-peripheral fontanelles, and finely incised sulci.
Discussion
Discovery of a definitive, skull-based eucryptodire in the Early Cretaceous of Argentina prompts reexamination of the published record, which has generally referred nonmarine turtles to the Pleurodira or Meiolaniidae (8Broin and Fuente, 1993). Much of this material is too fragmentary to reassess with confidence, but two specimens previously identified as pleurodires are probably cryptodires. These were originally identified as “grand Chelidae primitif indetérminé” (Broin and Fuente, 1993b: fig. 5, 18–19) from the Los Alamitos, Rio Negro, Late Cretaceous. Both specimens are in the Museo Argentino de Ciencias Naturales and have been examined by the senior author (E.S.G.). MACN 907 was identified as a fifth cervical, but it is very similar to the undoubted eighth cervical of MPEF PV1940. Both are biconvex with a narrowed centrum and a wide, midway positioned transverse process, distinct from any described chelid. MACN 907 lacks the neural arch as well as the zygapophyses. The principal difference between them is that the anterior central articulation of MPEF PV1940 is slightly larger than the posterior one, whereas in MACN 907 they are closer in size. The other specimen, MACN 908, identified as an eighth cervical (8Broin and Fuente, 1993), probably is an eighth; it agrees closely with MACN 907 but lacks transverse processes and a neural arch and only shows a biconvex centrum, so its identity must remain ambiguous. Nonetheless, it is very likely that MACN 908 is also a eucryptodire, along with MACN 907.
Geology
Location
The “Turtle Town” fossil locality (Rich et al., 1998) is situated in the center of Chubut Province approximately 32 km north of Paso de Indios, on the north side of the Rio Chubut (figs. 1–Fig. 2.3). Most of the best turtle remains were found at a site (site A: 43°35′51″S, 68°54′37″W) approximately 150 m east of the road (Route 40) leading north to the Los Adobes Mine, while a second, less productive site (site B: 43°35′52″S, 68°54′45″W) approximately 400 m to the west of the site A, but on the opposite side of the road, has also yielded a few turtle fragments in addition to a large sauropod phalanx and a well-preserved seed cone.
Stratigraphy
The Early Cretaceous Chubut Group of Argentina is a sequence of fossiliferous nonmarine epiclastic and pyroclastic sediments up to 1500 m thick (Figari and Courtade, 1993) that crop out over a large part of central Patagonia. The sediments unconformably overlie Middle–Late Jurassic volcanics and nonmarine sediments of the Cañadón Asfalto and Marafil Formations, and are unconformably overlain by latest Cretaceous (Campanian–Maastrichtian) continental to marginal marine to marine sediments of the La Colonia, Paso de Sapo, and Lefipan Formations (Figari and Courtade, 1993). The Chubut Group is presently divided into two formations called the Los Adobes Formation and Cerro Barcino Formation, based on sediment composition and depositional process. The older Los Adobes Formation is dominated by epiclastic sediments derived from the underlying Jurassic volcanic basement (unpublished CNEA report: Cerro Solo 1998, Un Proyecto Para La Produccion de Uranio y Molibdeno en la Patagonia. CAE–UAG Geologia Economica y Regional Patagonia). It is thought to be Late Valanginian–Hauterivian in age (Cortiñas, 1996; Page et al., 1999) and is subdivided into two units called the Arroyo del Pajarito Member and Bardas Coloradas Member. The lower unit, the Arroyo del Pajarito Member, is represented by up to 150 m of yellowish-gray to orange-brown, interbedded sandstone, pebbly sandstone, pebble-cobble conglomerate, and minor shale deposited in a high-energy, low-sinuosity fluvial environment. The overlying Bardas Coloradas Member is up to 30 m thick and is composed largely of red to reddish-brown shale and minor fine-grained sand deposited in a lower energy fluvial overbank environment. Both units are well exposed at the Los Adobes Mine (43°21′43S, 68°45′22″W) approximately 32 km north-northeast of “Turtle Town”, while another good exposure of the Bardas Coloradas Member can be found on the road leading up to the La Baritina Mine (43°28′ 35″S, 68°56′09″W) approximately 15 km to the north.
The Cerro Barcino Formation, in contrast, contains large volumes of pyroclastic sediments and is thought to be late Hauterivian to Cenomanian in age (Page et al., 1999). It is currently subdivided into five units called (in ascending stratigraphic order) the La Paloma, Cerro Castaño, Las Plumas, Puesto Manuel Arce, and Bayo Overo Members. Only two of these units crop out in the vicinity of “Turtle Town”: the La Paloma Member and Cerro Castaño Member.
The La Paloma Member (fig. 3) is a sequence of interbedded pale light gray, pale bright green, and pale reddish-brown (occasionally deep red and purple) tuff, siltstone, claystone, and rare, very fine to fine-grained sandstone and limestone up to 35 m thick that conformably overlie the Bardas Coloradas Member of the Los Adobes Formation. The basal contact is well exposed at several localities on the northern (43°12′03″S, 68°52′26″W) and southwestern (43°13′32″S, 68°57′24″W) sides of Cerro Chivos approximately 45 km north of “Turtle Town”.
The Cerro Castaño Member (fig. 3) is a sequence of interbedded reddish-brown, brown, and pale gray (occasionally white, yellowish brown, and pale orange-brown) tuffs, siltstone, and fine to coarse-grained sandstone in excess of 90 m thick that conformably overlie the La Paloma Member. The difference in color between these two units is quite dramatic when both are present in the same outcrop. The contact is often very sharp, and in some places, a thin lenticular coarse-grained sandstone with intraformational clasts can be found at the base of the Cerro Castaño Member.
All of the vertebrate fossils recovered from “Turtle Town” were found lying on the upper surface of outcrops of the La Paloma Member. In this area, Los Adobes Formation is absent and the La Paloma Member rests unconformably on the Jurassic Lonco Trapial Formation. The basal contact is well exposed in a 1.5-km-long section of cliffs up to 30 m in height on the northeastern side of the Rio Chubut (43°35′51″S, 68°56′44″W) approximately 2 km west of “Turtle Town”. Here the La Paloma Member dips very gently (<2°) to the east and infills/blankets the top of the Lonco Trapial Formation, which can vary in relief up to 12 m over a distance of approximately 100 m (fig. 3). The basal unconformity is also exposed in the floor of a broad gully about 1.5 km northwest of “Turtle Town” (43°35′29″S, 68°57′05″W) where several small mounds of Jurassic basement can be seen protruding up to 4 m above the surrounding La Paloma Member sediments in the floor of the depression.
The depositional setting of the La Paloma Member at “Turtle Town” is difficult to assess due to the abundance of clay minerals in the sediments, which upon weathering effectively destroy all sedimentary structures. Consequently, the upper surface of the La Paloma Member on gentle to moderate slopes (< 30°) often has a “popcorn-like” appearance with a deep weathering profile. This problem is further compounded by sediment being washed down the slopes from higher up, which masks the outcrop lower down and necessitates trenching to reveal the true color and lithology of the underlying sediments.
Mapping in the area of site A indicates that the vertebrate remains found at this location are from a thin band of silcrete less than 1 cm thick that contains abundant, randomly oriented, millimeter- to centimeter-sized pieces of silicified plant fragments and stems of branches up to 1 cm in diameter and occasionally short sections of tree trunks up to 20 cm in width. The silcrete band crops out along the top of a ridge and forms small pavement surfaces. The silcrete band occurs above a 1.8-m-thick laminated pale gray calcareous siltstone that in turn overlies an 8-m-thick bright green massive tuff. Both facies are weakly bioturbated in places, which suggests the plant matter in the fossiliferous silcrete band probably accumulated on the floor or edge of a lake prior to cementation.
Morphologic Description
Figure 4 is a partially restored view of the skull, figs. 5–Fig. 6.Fig. 7.Fig. 8.9 are stereophotographs of the skull and fig. 10 is a key to the stereophotographs. The reader should also see tables 1 and 2.
TABLE 1
Comparison of Chubutemys and Other Eucryptodires.
TABLE 2
Measurements of Type Skull of Chubutemys, MPEF-PV-1236 (in mm)
Prefrontal
The external surface of both prefrontals is badly eroded but enough remains on the right side to see a maxilla contact anterolaterally and a frontal contact posteriorly. As preserved, the prefrontal meets the other prefrontal in a long midline suture completely preventing the frontal-nasal contact and anterior frontal contact seen in Ordosemys. The prefrontal of Chubutemys as preserved is consistent with both Dracochelys and Hangaiemys.
Frontal
Most of the right frontal is preserved, but the dorsal surfaces are eroded off. The left frontal is largely represented by a mold in matrix of its internal surface. The frontal of Chubutemys is relatively small and does not seem to reach the orbital margin due to a prefrontal-postorbital contact. Such a contact is present in Meiolania and absent in Dracochelys, Ordosemys, and Hangaiemys, although only a relatively narrow projection of frontal reaches the orbit in those forms. However, the poor preservation of the frontal in Chubutemys must be kept in mind.
Parietal
Most of both parietals has been eroded off and all of their external surface is missing. However, the internal mold of the bone in the matrix retains most of the sutures (figs. 5, 10). The parietal of Chubutemys is a large rectangular element, contacting the frontal anteriorly and the postorbital anterolaterally. Posterolaterally, a suture with the squamosal can be made out on each side, but the postorbital-squamosal (or quadratojugal) suture is too damaged to be seen clearly.
Chubutemys differs significantly from the similar Dracochelys, as well as from the other “sinemydids/macrobaenids”, Judithemys, Sinemys, Ordosemys and Hangaiemys, in having no temporal emargination. The parietal of Chubutemys extends well posterior to the posterior edge of the otic chamber. The supraoccipital is present in part between the parietals, but the sutures are unclear. Meiolania has an extensive temporal roof, but this is the result of a large squamosal and a laterally expanded supraocciptal. The parietal in Meiolania is relatively small, nearly square, and much smaller than the parietal in Chubutemys. Nonetheless, the parietal of Meiolania does agree with Chubutemys in reaching the posterior edge of the otic chamber and broadly contacting the squamosal.
The skull roof in Macrobaena (Tatariniv, 1959) also lacks temporal emargination, but it is too poorly known for the composition of the skull roof to be determined from the literature. For this reason Macrobaena is not included in our data set and has not been included in other authors' data sets.
Near the posterior edge of the parietal in MPEF-PV1236 are the two atlantal neural arches, disarticulated from the atlas (fig. 10).
Jugal
The jugal is present on the right cheek (figs. 8, 10), though not well defined, and is largely missing on the left. There is what appears to be the anterior contact with the maxilla and the posterior contact with the quadratojugal. The postorbital-quadratojugal contact is unclear, and some of the dorsal suture around the jugal is presumably with the postorbital. As interpreted, the jugal is a relatively large element, making up most of the cheek. There is no cheek emargination, and the jugal makes a slight posteroventral extension comparable to the one in Meiolania. Ordosemys, Dracochelys, and Hangaiemys have a cheek emargination and a much smaller jugal than that in Chubutemys.
Quadratojugal
It is likely that the bone area posterior to what we have identified as jugal on the right side of MPEF-PV1236 is the quadratojugal (fig. 10C), in which case it probably contacts the postorbital dorsally, but its relationships to the squamosal or anything else are unknown.
Squamosal
Very little bone representing the squamosal is present on either side, but the mold in the matrix on the right side shows what appears to be the squamosal and its broad contact with the parietal (fig. 10A, C).
Postorbital
Much of the postorbital is present on the right side (figs. 8, 10), and an internal mold of some of it is present on the left. Sutures are probably with the prefrontal anteriorly, the frontal anteromedially, the parietal medially, and presumably the squamosal and/or quadratojugal posteriorly. The lateral and ventral contacts, for example, with the jugal, are unclear.
Premaxilla
The anteriormost end of the skull is broken off and no premaxillary sutures can be seen. The break is about where a maxilla-premaxilla suture would be expected in Ordosemys or one of the other eucryptodires. On the ventral surface, there is a vomer suture at its anterior end and this is most likely with the premaxilla (fig. 10B). This would indicate a premaxilla close in size and shape to that in Ordosemys and Dracochelys.
Maxilla
The maxilla (figs. 6, 8, 10) is best preserved on the right side, but some of the triturating surface is present on the left. Its external surface is present on the right as well as most of its ventral surface.
The labial ridge of Chubutemys is a straight, narrow sheet, similar to that in Dracochelys, Ordosemys, and Hangaiemys. There is no evidence of the cusps seen near the premaxillary contact in Dracochelys. The depth of the ridge is similar to Ordosemys and Dracochelys. The labial ridge is low and parallels the lingual ridge as in Ordosemys, Dracochelys, and Hangaiemys. The width of the triturating surface in Chubutemys is very similar to that in Ordosemys, that is, wider than the very narrow surface of Hangaiemys and Dracochelys, but narrower than that in meiolaniids.
The apertura narium interna in Chubutemys is very similar in position and size to that in Ordosemys, Dracochelys, and Hangaiemys. The foramen palatinum posterius is unusually large in Hangaiemys and Dracochelys, but it is smaller, about as in Chelydra, in Chubutemys and Ordosemys. In Meiolania the foramen is unusually small.
Vomer
The vomer in Chubutemys (figs. 6, 10) is relatively well preserved, broken in the middle but with good contacts. The vomer in Chubutemys is similar to that bone in other non-polycryptodiran eucryptodires. Its contacts are with the premaxilla anteriorly, the prefrontal (questionably) anterolaterally, and the palatine laterally. In contrast to Dracochelys, Ordosemys, Hangaiemys, and Meiolania, there is no posterior contact with the pterygoid. The vomer in Meiolania is usually wide, forming the parasagittal choanal grooves characteristic of this genus. The vomer has a midline, sagittal ridge that appears to be relatively wide in Chubutemys as in Dracochelys, not narrow as in Hangaiemys and Ordosemys, although this could be the result of poor preservation.
Palatine
The palatine in Chubutemys (figs. 6, 10) is present on both sides and has sutures showing most of its limits. It has the usual cryptodiran contacts of vomer medially, pterygoid posteriorly, and maxilla laterally. An anterior prefrontal contact is likely but not clear. Chubutemys differs from Ordosemys, Dracochelys, Hangaiemys, and Meiolania because the palatines meet in the midline and are not separated by the vomer. The shape of the palatine is similar in Chubutemys, Dracochelys, Hangaiemys, and Ordosemys. The bone is much smaller in Meiolania.
Quadrate
The right quadrate in Chubutemys (figs. 6, 8, 10) preserves the cavum tympani, but the processus articularis is broken off. The left quadrate is less complete, lacking much of the cavum tympani as well as the processus articularis. Quadrate sutures are generally unclear, but on the right side one can be interpreted as a quadratojugal and squamosal contact. In ventral view, the quadrate-pterygoid suture is visible on the right side as well as the contact with the squamosal and opisthotic.
As preserved, the cavum tympani in Chubutemys is very deep and has a large dorsal overhang. This condition, however, is probably exaggerated by dorsoventral crushing in this specimen. Nonetheless, the dorsal part of the cavum tympani probably was much deeper than the anterior area. The ventral area is missing. The dorsal area leads into the antrum postoticum, which is mostly missing, but appears to have been comparable in size and shape to that in Ordosemys. The incisura columellae auris is preserved almost completely on the right side and partially on the left. It is narrow but posteroventrally open as in Ordosemys and Dracochelys (unclear in Hangaiemys). In Meiolania it is similar but closed laterally to include the eustachian tube by approximation of the quadratojugal and quadrate. It is possible that this was also the case in Chubutemys, but the lateralmost parts of the quadrate are missing. The processus trochlearis oticum can be seen on both sides. It seems to be comparable in size to that in Ordosemys and Dracochelys.
Both stapes are preserved in place in Chubutemys from the incisura columellae auris to the fenestra ovalis. The stapes is thicker in Chubutemys than in Chelydra with a similar-sized skull. The footplate is conical rather than disc-shaped, a feature of baenids, meiolaniids, and some trionychoids (see Gaffney, 1983: 449). It is unknown in the other eucryptodires compared here.
Epipterygoid
A small section of what appears to be the epipterygoid is preserved on the right side anterodorsal to the cavum epipterygoidei, in its usual eucryptodiran position.
Pterygoid
The pterygoid is preserved almost completely on both sides (figs. 6, 10). Its surface is eroded in some areas and some of the thin web of bone that usually stretches medially from the quadrate ramus has been lost on both sides, revealing the cavum acusticojugulare in ventral view. The pterygoid of Chubutemys meets the palatine anteriorly, the other pterygoid in a short suture anteromedially, the basisphenoid medially, and the quadrate posterolaterally. The pterygoid, along with the maxilla and palatine, forms the foramen palatinum posterius. The processus pterygoideus externus in Dracochelys is characteristically very large, but both processes are mostly missing in Chubutemys, and its size is unknown. In Dracochelys, Hangaiemys, and Ordosemys there is a deep fossa just posterior to the quadrate ramus of the pterygoid. Much of this area in Chubutemys has been eroded but enough is preserved to show that a deep fossa was absent, making Chubutemys more like Chelydra in this feature. Meiolania also lacks a fossa in this position. Ordosemys and Dracochelys also have a distinct ridge paralleling the bone margin laterally, presumed to be for pterygoideus muscle attachment. This ridge is absent in Chubutemys and also in Meiolania.
The medial margin of the pterygoid in Chubutemys contacts the basisphenoid, and on the right side the anterior part of this suture is expanded to form a foramen, just lateral to the midline. On the left side, the foramen is present but damaged by crushing. In Dracochelys this foramen was identified as the foramen caroticum laterale (Gaffney and Ye, 1992), being the entry of the palatine artery into the cavum cranii. Chubutemys appears to have this foramen in the identical position as it is in Dracochelys. In Hangaiemys and Ordosemys the basisphenoid is much shorter than in Chubutemys and Dracochelys, and the foramen is positioned farther posteriorly, close to the foramen caroticum basisphenoidale. The difference is due to a thin, posteromedial sheet of pterygoid that covers the basisphenoid in Hangaiemys and Ordosemys.
The term foramen caroticum laterale as currently used actually refers to both a more anterior foramen opening into the braincase floor and a more posterior foramen opening on the roof of the palate, with the canalis caroticus lateralis (Gaffney, 1979 and others) connecting them. Sukhanov (2000) used the term “foramen posterius canalis caroticus laterale” for the posterior foramen, and we have adopted that usage here.
Supraoccipital
Only a small part of the supraoccipital is preserved (figs. 7, 10). On the dorsal surface some of the bone on the midline portion of the temporal roof is probably supraoccipital. In occipital and ventral view a portion of the crista supraoccipitalis is visible. It is comparable in size and shape with those seen in other non-polycryptodiran eucryptodires. Although very little bone is present on the posterior part of the temporal roof, the internal mold of the ventral surface does show other sutures and there is no indication of a laterally expanded supraoccipital as seen in Meiolania.
Exoccipital
Both exoccipitals are preserved and some sutures are visible (figs. 7, 10). Laterally, the exoccipital contacts the opisthotic and ventrally the basisphenoid. As preserved there is no pterygoid contact as seen in Dracochelys, Ordosemys, and Meiolania. Hangaiemys seems to lack the contact, but it is hard to be sure. It is possible that Chubutemys had a pterygoid-exoccipital contact because clearly some of the posterior portion of the pterygoid is missing due to damage. However, it is difficult to say how much is missing.
The foramen magnum in Chubutemys is about as wide as high, not as narrow as in Dracochelys and Ordosemys, but this could be due to crushing. The two foramen nervi hypoglossi can be seen on the right side and one on the left. The area is not well preserved, but it is possible that the medial margin of the foramen jugulare posterius is preserved on the right side. As preserved the foramen is not enclosed, but its margin could easily be missing. However, neither Dracochelys nor Ordosemys seems to have a definitive foramen jugulare posterius, so the situation is ambiguous. The condylus occipitalis has no sutures preserved on it, but it is presumed to be formed by the exoccipitals and basioccipital as is usual in cryptodires.
Basioccipital
The basioccipital (figs. 6, 10) is nearly complete in MPEF-PV1236, but seems at least slightly eroded along its lateral edges. There is a clear anterior suture with the basisphenoid, but the exoccipital sutures are visible in only a few places. As preserved, there is no pterygoid contact as seen in Dracochelys. The basioccipital in Chubutemys is relatively long as in Hangaiemys and Dracochelys, not short as in Ordosemys, although this comparison is somewhat subjective. The tuberculum basioccipitale in Chubutemys is distinct as in Ordosemys, and much larger than in Dracochelys.
Prootic
Although both prootics are probably complete in MPEF-PV1236, they are mostly covered by matrix. The bone is best seen on the right side where the anterior surface of the otic chamber is exposed, but no sutures are apparent, so the bone is not identified in the figures. The processus trochlearis oticum is comparable in size to that in Dracochelys and Ordosemys. The foramen nervi trigemini, also presumably formed in part by the prootic, is visible on the right but no sutures can be seen. Its position and size are similar to Chelydra.
Opisthotic
Most of both opisthotics are preserved in MPEF-PV1236, but sutures are visible only on the right side (fig. 10). Only the ventral and occipital surfaces of the opisthotic are visible in MPEF-PV1236; the dorsal surfaces are covered with matrix. The opisthotic in Chubutemys is basically similar to that in other cryptodires, as this bone does not vary much within this group. In ventral view it lies between the quadrate and squamosal laterally and the exoccipital medially. As preserved the floor of the cavum acustico-jugulare is largely missing. This reveals the stapes and the fenestra ovalis being formed posteriorly by the processus interfenestralis of the opisthotic. The fenestra ovalis is not visible in ventral view in other eucryptodires. Its visibility in Chubutemys may be due to missing bone in MPEF-PV1236. In MPEF-PV1236 the foramen externum nervi glossopharyngei is visible at the base of the processus interfenestralis as well as the foramen jugulare anterius.
Basisphenoid
The entire basisphenoid is preserved and visible in ventral view in MPEF-PV1236 (fig. 10). Sutures are present for most of its margins. The basisphenoid is triangular and lies between the pterygoids anterolaterally. A posterior transverse suture is with the basioccipital. As preserved in Chubutemys the pterygoid does not reach the basioccipital as in Ordosemys, and the basisphenoid has a free lateral edge just anterior to its contact with the basioccipital. Hangaiemys (Sukhanov and Narmandakh, 1974; Sukhanov, 2000) seems to have a similar condition in which the pterygoid barely reaches the basioccipital. If the pterygoid were only slightly eroded it would produce a free edge for the basisphenoid.
Midway along the length of the basisphenoid is the paired foramen caroticum basisphenoidale (figs. 4, 10). In MPEF-PV1236 this foramen is preserved on both sides with a poorly preserved canalis caroticus internus extending posterolaterally from it. The canalis is only a poorly defined groove on the right, but on the left a small portion of floor is preserved showing that it was a canal (fig. 6). The canal closely follows the pterygoid-basisphenoid suture where that is visible. The foramen posterius canalis carotici interni is interpreted to lie at the posterior end of the groove, formed in or near the pterygoid-basisphenoid suture as in Dracochelys, Ordosemys, and probably Hangaiemys. In Meiolania and Niolamia (Gaffney, 1983) the canalis caroticus internus is deeper within the pterygoid and basisphenoid but still close to the sutural contact of those bones. In Ordosemys, Dracochelys, Hangaiemys, and Chubutemys there is both an open foramen basisphenoidale and ventrally open foramen caroticum laterale, leading Gaffney (1996: 139) to suggest that the branching of the palatine artery from the carotid was unenclosed by bone. Gaffney (1983, 1996) also argued that the condition in Niolamia (and its modified form in Meiolania) was very similar to that in Dracochelys, now known in Ordosemys, Hangaiemys, and Chubutemys. The pterygoid of Niolamia (Simpson, 1938; Gaffney, 1983: fig. 60) is very similar to the pterygoid in Dracochelys and Chubutemys, apparently differing only in the greater thickness of the canalis caroticus internus floor in Niolamia. This is further evidence that the gap in the suture between the basisphenoid and pterygoid in Dracochelys and Chubutemys is the homologue of the intrapterygoid slit of meiolaniids, as suggested by Gaffney (1996).
Carapace
The left anterior fragment of a carapace (fig. 11) is preserved along with postcranial elements in MPEF PV1940. The dorsal surface consists of some bone along with the underlying mold of the missing bone. The medial ends of costals one and two, the posterior edge of the nuchal, neurals one and two, and a presumed preneural are present on the dorsal surface. The nuchal is relatively broad, similar to that in Ordosemys, Dracochelys, and other non-polycryptodire eucryptodires. Costals one and two also vary little among these taxa, and Chubutemys agrees with them. The neurals are not well preserved and the bone surface is eroded off, but sutures are visible. If correctly interpreted, Chubutemys has a preneural between the nuchal and neural one. Ordosemys (Brinkman and Peng, 1993a) is another non-polycryptodire eucryptodire with this element. In Ordosemys it is longer than wide; in Chubutemys it is wider than long.
The only dorsal surface preserved in MPEF PV1940 is in the area of the first costal. This shows sulci of the triple junction between vertebral scale one and two and pleural scale one. The position of this junction in Chubutemys is very similar to that in Ordosemys and the other early eucryptodires. The first vertebral does not seem to be much larger than the second as in these forms and Kirgizemys. The bone surface as preserved is smooth, lacking a textured pattern. The sulci are broadly incised, forming wide troughs.
On the ventral surface of MPEF PV1940 are the first three thoracic vertebrae, the eighth cervical, the left humerus and scapula, and a flat bone fragment that could be a piece of plastron or coracoid. The first thoracic has an anterior central articulation facing anteriorly as in Ordosemys, Xinjiangchelys, Meiolania, Sinemys, and Dracochelys, and it is interpreted as the primitive cryptodiran condition (Gaffney, 1996). The rib head for this centrum is visible, but the distal part of the rib is covered by matrix, thus preventing determination of its length.
Cervical Eight
An eighth cervical vertebra (figs. 11, 12) is in articulation in the partial carapace, MPEF PV1940. As the eighth cervical figures importantly in previous analyses of eucryptodire relationships, it is important to describe that bone in Chubutemys. The eighth cervical is also known in Dracochelys (Brinkman, 2001), Ordosemys (Brinkman and Peng, 1993a), Otwayemys (Gaffney et al., 1998), Sinemys (Brinkman and Peng, 1993b), Xinjiangchelys (Kaznyshkin et al., 1990), and Meiolania (Gaffney, 1985, 1996). Phylogenetic analyses using eighth cervical morphology that include discussion of characters are Gaffney (1996) and Brinkman and Wu (1999). The eighth cervical of MPEF PV1940 is not completely preserved. Figure 12 has been partially restored by assuming bilateral symmetry. The centrum, transverse processes, and right zygapophyses are visible, but nothing of the neural arch or spine can be seen. The ventral edge of the centrum is eroded off and there has been dorsoventral crushing. The crushing has pressed the postzygapophyses so far ventrally that the neural canal is nearly obliterated. Anteriorly, the crushing is much less and may be nonexistent, but this is difficult to determine.
The eighth cervical of MPEF PV1940 as preserved is generally similar to those in other early Cretaceous eucryptodires. The centrum is biconvex as in Dracochelys, Ordosemys, Otwayemys, and Sinemys, in contrast to the procoelous condition of Meiolania and the Polycryptodira. The posterior articulation in Chubutemys is relatively narrow and hemispherical, while the anterior articulation is much wider. This condition can also be seen in Otwayemys, but the wider distribution of this feature is unclear. The transverse processes in Chubutemys are single as in Chelydra and the Polycryptodira, not double as in Dracochelys, Otwayemys, Ordosemys, Sinemys, and Meiolania. The transverse process in Chubutemys is also placed more anteriorly than the middle of the centrum. This condition is also as in the above genera and the Polycryptodira. There is no evidence of cervical ribs. The ventral surface of the centrum is eroded, and the presence or absence of a keel or process is not determinable.
The prezygapophyses in Chubutemys are only represented by the right one, and it is placed much more laterally than in Otwayemys, Dracochelys, Ordosemys, and Meiolania. This may be due to deformation. Its position could have been very similar to those in Ordosemys and Dracochelys, which are farther apart than in Otwayemys. The postzygapophyses in Chubutemys are also widely placed as in Ordosemys, not narrow as in Otwayemys. The eighth cervical in MPEF PV1940 is positioned anterior and lateral with respect to the first dorsal. It also underlies the nuchal. There is no sign on the nuchal of a neural spine pushed through or crushed beneath it. The implication is that Chubutemys does not have a tall neural spine as in Otwayemys and Meiolania, but this is just a guess.
Phylogenetic Analysis
Chubutemys is a cryptodire similar to other cryptodires that have been referred to as “sinemydids” and “macrobaenids” (see Gaffney and Yeh, 1992; Gaffney, 1996; Brinkman and Peng, 1993a; Brinkman and Wu 1999; Brinkman, 2001; Sukhanov, 2000, for references), and it is among these extinct cryptodire groups that its relationships are to be sought. The present analysis is based on an analysis developed in 1992 in a joint effort by Don Brinkman, Gene Gaffney, Howard Hutchison, and Peter Meylan. The major part of this analysis was published in Gaffney (1996). Once again, the senior author is very grateful to his three colleagues for sharing their knowledge. Analyzing relationships of the extinct eucryptodires is hampered by the fact that the terminal taxa being analyzed are something of a moving target. Many named taxa are inadequately described in the literature but are known firsthand by some workers, making corroboration of some published character distributions more difficult than usual.
Although the matrix of 111 characters (appendix 1) is based on the 40 characters in Gaffney (1996), it has obviously been greatly altered by the addition of new ones. In Gaffney (1996) the specific-level taxa in the family Meiolaniidae were analyzed separately from the family Meiolaniidae among turtles, so two data sets were used. We have incorporated some of the characters from the within-group data set here. Brinkman and Wu (1999) made significant additions, and their characters are also added here (except for parsimony uninformative ones due to nonoverlapping terminal taxa). Characters are also added from Rougier et al. (1995) and Gaffney et al. (2006).
Hirayama et al. (2000) proposed a cladogram showing eucryptodires to be nonmonophyletic, with meiolaniids allied with Mongolochelys and outside most other cryptodires. The Hirayama et al. (2000) cladogram is based on a 69 character data set, including the 40 characters from Gaffney (1996), modified in Gaffney et al. (1998), and characters from Parham and Hutchison (2003). We have used at least a version of all the characters in the Hirayama et al. (2000) data set that are parsimony-informative. Because we have deleted some terminal taxa (that in our opinion are not firmly based) from Hirayama et al. (2000), some characters from Hirayama et al. (2000) that were parsimony-informative in that analysis are no longer parsimony-informative in our analysis and have been deleted. A few characters have been altered either in definition or distribution. The altered characters relevant to meiolaniid relationships in particular are as follows:
1. “(10) Prominent enlargement of squamosal reaching supraoccipital.” This character is unchanged here, but we have added another character, supraoccipital with horizontal plate, to reflect the different conditions of the contact. In both Mongolochelys and meiolaniids, there is a large squamosal forming most of a posteriorly directed temporal margin. However, in Meiolania much of this margin is formed by a horizontal plate of the supraoccipital, which is absent in Mongolochelys. In the latter the squamosals meet on the midline, a condition not seen in meiolaniids.
2. Hirayama et al. (2000) united meiolaniids and Mongolochelys on the common presence of “(13) Prominent lingual ridge on maxilla”, or “accessory ridge on the triturating surface” in our usage. Although this ridge is present in Meiolania and Ninjemys, the sister taxon of these, Niolamia, lacks the ridge. In the cladogram and data set for Meiolaniidae developed by Gaffney (1996) and used here, the absence of the ridge is primitive for Meiolaniidae and does not unite Meiolaniidae and Mongolochelys.
3. Another character of more general distribution that breaks up the Eucryptodira in Hirayama et al. (2000) is the “lower cheek emargination”. In Hirayama et al. (2000) the complete absence or presence of a shallow cheek emargination is one state, but we see three states: 0 = absent, shallow, with a large jugal; 1 = absent, shallow, with a large quadratojugal; 2 = moderate to deep. We have used these to reflect what we see as a significant difference when the solid cheek is formed by a large jugal as in Proganochelys, Australochelys, Kayentachelys, as well as Mongolochelys. This condition is common and seems to be primitive at the level of Testudines as well as Cryptodira. Meiolania has a solid cheek formed in part by an unusually large quadratojugal, which seems to be a different character state. This state may be best interpreted as the secondary filling of an emarginate cheek.
4. We have also used four rather than three states for the Hirayama et al. (2000) “Posterior temporal emargination” character. We have added a state, unique to meiolaniids for a fully covered skull roof, but with an unusually small parietal, a condition that is different from the much larger parietal seen in Proganochelys and other non-Casichelydians as well as in the extensively roofed Mongolochelys, Kallokibotion, and Chubutemys. The implication of the unique cheek and temporal roof structure in meiolaniids is that these areas may be secondarily solid from a more emarginated ancestor, similar to the conditions in taxa such as Dracochelys and Ordosemys
5. The Hirayama et al. (2000) character “(34) Cervical ribs of large size” also breaks up the Eucryptodira of Gaffney and other authors. Gaffney and others have used the simple presence or absence of cervical ribs as a character. The problem here is that cervical ribs in turtles show a great deal of variation, much of it described in Gaffney (1985, 1990), Brinkman and Peng (1993a), and Brinkman and Nicholls (1993), with a discussion of the problem in Gaffney (1996: 149). Hirayama et al. (2000) tried to use some of this diversity, but as discussed previously we think that the diversity does not show discrete states. It is true that there is a considerable difference in the relatively large ribs of Proganochelys (Gaffney, 1990) and the small ribs of some later cryptodires, such as baenids. However, the ribs in Meiolania (Gaffney 1985, 1990) do not agree with the morphology of the ribs in Proganochelys and are, in fact, unique among turtles in shape, as well as in their relative size. Therefore, we use only the presence/absence of ribs and not their size or other aspects of morphology.
Corrections to the data matrix in Gaffney et al. (1998) for characters referring to Hangaiemys are from Parham and Hutchison (2003). Parham and Hutchison (2003) also changed some characters for Judithemys (the unnamed TMP 87.2.1 of Gaffney, 1996). However, the senior author has corroborated some of the earlier codings and thinks that the original states for characters 3 (visible on broken edge of prefrontal on right side) and 18 (sutures visible on left side) are correct. Parham and Hutchison (2003) also change characters 4 and 5 in appendix 1, but this may be an error. We use the original codings from Gaffney (1996). Character 25, presence of chevrons, was reported absent by Parham and Hutchison (2003), but chevrons are clearly present on the anterior caudals preserved in TMP 87.2.1 (Parham and Hutchison, 2003: fig. 8D). In order to clarify the senior author's interpretations of the skull of Judithemys (TMP 87.2.1), a set of partially restored views are provided as fig. 13.
The terminal taxa used here are mostly the “usual suspects” from Gaffney (1996), with some additions. As it became apparent that Chubutemys may be related to meiolaniids and that the position of meiolaniids has been called into question by Hirayama et al. (2000), the better known meiolaniid taxa were added to the analysis. The discovery of associated postcranial material of Solnhofia (Joyce, 2000; see also Broin, 1994; Lapparent de Broin et al., 1996) has allowed this taxon to be added. On the other hand, the relationships of the living cryptodire groups are beyond the scope of this analysis, and only a few characters and larger, more inclusive taxa are used. The molecular evidence that the family Trionychidae is the sister taxon to all other living cryptodires is a hypothesis well beyond the scope of this paper, and we continue to use Trionychoidea. We do not think that this lumping of the recent taxa significantly affects cladograms of the extinct cryptodires we have chosen.
The phylogenetic analysis corroborates the eucryptodiran nature of meiolaniids and contradicts the hypothesis of Hirayama et al. (2000). It also supports monophyly of the Centrocryptodira of Gaffney (1996), which has formed central articulations as one of its synapomorphies. The Sinemydidae of Gaffney (1996), supported by the biconvex eighth cervical in the earlier analysis, is not supported in the present analysis; instead, the extinct eucryptodire taxa often referred to as “sinemydids” and/or “macrobaenids” are resolved as a paraphyletic group of successive sister taxa to the Polycryptodira. The biconvex eighth cervical in this resolution becomes a synapomorphy for Centrocryptodira, lost in Meiolaniidae, Judithemys, (?Hangaiemys), and most of the Polycryptodira. The Centrocryptodira is also supported by the biconvex fourth cervical, which does occur in meiolaniids, as well as in the other Centrocryptodira, except Sinemys and trionychoids. The number of unambiguous synapomorphies and the bootstrap test (fig. 14) show Cryptodira and Eucryptodira to be well supported, but the resolution of extinct cryptodires breaks down in bootstrapping. Nonetheless, the most parsimonious tree (fig. 14) shows six unambiguous synapomorphies for the node containing Polycryptodires, Judithemys, Hangaiemys, Dracochelys, Sinemys, and Ordosemys. The particularly interesting result of this data set resolution is that the sister taxon to the above listed group consists of the meiolaniids plus Otwayemys and the subject of the present paper, Chubutemys. This is a little surprising because Chubutemys is similar to Dracochelys in much of its skull. It also does not rely on the fully roofed skull found in both Chubutemys and meiolaniids, because this is interpreted here as separate characters in these two taxa, and is not used as a character uniting Chubutemys and meiolaniids. A previous analysis of the Early Cretaceous Australian cryptodire Otwayemys (Gaffney et al., 1998) allied that form with meiolaniids, and the present analysis also found that relationship, adding Chubutemys as well.
Neither Chubutemys nor Otwayemys is strongly supported as a sister taxon to meiolaniids, but the most problematic element of this resolution is that both taxa have lots of missing data: Chubutemys, 49%; Otwayemys, 68%. PAUP fills in the missing characters consistent with the most parsimonious cladogram (MPC), which probably masks contradictory characters. There is only one unambiguous character uniting Chubutemys with meiolaniids, and that is the prefrontal-postorbital contact that prevents orbital exposure of the frontal. This is actually a rare character among the extinct cryptodire groups; it occurs in Mongolochelys and Polycryptodira, but it is indeterminate in Otwayemys. The sole character uniting Otwayemys and meiolaniids is the transverse process of the cervicals in the middle of the centrum, which is absent in all other Centrocryptodira except Dracochelys.
One could argue that the solid temporal roof of Chubutemys is the primitive condition for meiolaniids. The large parietal of Chubutemys certainly seems to be the primitive condition for cryptodires, although many Paracryptodira and Plesiochelyidae have some degree of emargination. Within Eucryptodira, only meiolanids and Chubutemys (outside a few Polycryptodira) have a fully roofed skull, and the present MPC suggests this may not be a homoplasy. However, if the meiolaniid condition is re-coded as “0” (agreeing with Mongolochelys and Kallokibotion), the same MPC results.
The carotid morphology of Chubutemys makes a good primitive condition for meiolaniids. Jamniczky et al. (2006) gave a summary of carotid canal hypotheses (see also Jamniczkyand Russell, 2004) and proposed an evolutionary progression consistent with our character states. The enclosed canalis caroticus internus posterior to the foramen basisphenoidale has been divided into two character states by Gaffney (1996), Gaffney et al. (1998), and Parham and Hutchison (2003), among others. One state is a thin or partial floor and the other is a thick floor. The senior author now thinks that this might be better changed to present/absent. After looking at a large number of specimens, both recent and fossil, the distinction is beginning to become blurred with a number of specimens within some species showing a fair amount of variation. However, changing the data set to present/absent results in the same MPC. For meiolaniids the carotid is embedded deeply in the pterygoid in Meiolania, but it is just covered by it in Niolamia ( = Crossochelys, see Simpson, 1938: fig. 7), so a thin condition is probably primitive for the group.
The morphology of the canalis caroticus lateralis, the presumed path of the palatine artery, in Chubutemys is also primitive for the condition in meiolaniids. In Chubutemys and Dracochelys there is a step in the basisphenoid-pterygoid suture that opens up on each side very close to the midline to form the foramen posterius canalis caroticus laterale (the posterior opening of the foramen caroticum laterale), the presumed entry point of the palatine artery into the skull. In these two taxa the two openings form a V-shaped suture at the apex of the basisphenoid. In Hangaiemys (Sukhanov, 2000) the pterygoid-basisphenoid shelf forming the ventral margin of the foramen posterius canalis caroticus laterale is transverse rather than V-shaped, but it is otherwise apparently very similar. However, in Sinemys, Judithemys, and Ordosemys there is no shelf or foramina at the apex of the basisphenoid; instead, the foramen posterius canalis caroticus laterale is placed more laterally from the midline and more posteriorly. Although it is still formed by both pterygoid and basisphenoid, the basisphenoid is exposed as a short anterolateral process reaching the foramen posterius canalis caroticus laterale. The foramen is also a more oval foramen rather than the slit seen in Dracochelys and Chubutemys. The disarticulated pterygoid of Niolamia (AMNH 3161) figured by Simpson (1938: fig. 7) is clearly very similar to the pterygoid in Chubutemys and Dracochelys (and probably Hangaiemys, although the senior author has not seen this material). The primitive nature of the Niolamia ( = Crossochelys Simpson) pterygoid was argued by Gaffney (1983: fig. 60). However, even in Meiolania platyceps, the likely homology of the meiolaniid intrapterygoid slit with the foramen posterius canalis caroticus laterale and canalis caroticus lateralis in the extinct eucryptodires is substantiated by comparison of fig. 55 (upper) in Gaffney (1983) showing Meiolania with figure 6A, C in Brinkman and Peng (1993a) showing Ordosemys. These figures show that the intrapterygoid slit in Meiolania actually has paired canals lying laterally formed by basisphenoid and pterygoid, as in Ordosemys and other non-polycryptodiran eucryptodires.
Therefore, Chubutemys provides further evidence that meiolaniids are related to “basal” eucryptodires or “sinemydids/macrobaenids”, that is, eucryptodires outside the Polycryptodira. The basicranial morphology of meiolaniids, rather than being a unique feature of the group, is instead a modified state of the morphology seen in such “sinemydids/macrobaenids” as Chubutemys, Dracochelys, Ordosemys, and Sinemys.
Chubutemys shows that terrestrial eucryptodires were present in South America in the Cretaceous, as they were in North America, Asia, and Australia. At present, the distribution of the (probably paraphyletic) non-polycryptodiran eucryptodires seems to be relatively widespread: North and South America, central Asia, and Australia, but apparently not Africa (although the record is poor) or Europe. Chubutemys itself, if it does prove to be a meiolaniid sister taxon, would represent the Mesozoic distribution of the more restricted monophyletic group in Australia and South America that evolved into the meiolaniids with their characteristic skull roof synapomorphies. Others of the paraphyletic “sinemydids/macrobaenids” were more closely related to the living eucryptodires, the Polycryptodira.
Acknowledgments
The senior author acknowledges the support of the Department of Paleontology, American Museum of Natural History, during the course of this project. We greatly appreciate the help of Juliana Sterli, Guillermo Rougier, Ren Hirayama, and an anonymous reviewer for improving the paper. The figures were done by Steve Morton (figs. 5–Fig. 6.Fig. 7.Fig. 8.9), Aysha Venjara (figs. 10, 11, 12), and Frank Ippolito (figs. 4, 13), with final grooming on all figures by Frank Ippolito. We are very grateful to them all. We thank Judy Galkin for typing and bibliographic help. We also thank Maria Copello, Ruben Cuneo (Director of the Museo Paleontologico Egidio Feruglio), Peter Trusler, Pablo Puerta, Maxi Iberleucha, and Roberto Bianchi (mapping geologist, Comisión Nacional de Energía Atómica). The National Geographic Society (grant 5187–94) and the Australian Research Council provided funding for the fieldwork. The Comisión Nacional de Energía Atómica provided field housing for staff during the fieldwork. This paper was submitted on October 6, 2006, and was accepted September 24, 2007.
References
Appendices
APPENDIX 1
Character List
1. NA, nasals
2. NA, size
3. NA, nares divided
4. NA, nasomaxillary sinus
5. LA, lacrimal bone
6. LA, lacrimal foramen
7. PF, meet on midline
8. PF, pf-po contact
9. PF, pf-vo contact
10. PA, processus inferior parietalis
11. PA, pa-sq contact
12. PA, temporal emargination
13. ST, supratemporal
14. JU, cheek emargination
15. SQ, meet on midline
16. SQ, posterolateral process
17. SQ, sq-so contact
18. SQ, broad sq-qj contact
19. PO, length
20. PO, po-sq contact
21. PM, premaxillary hook
22. MX, accessory ridge on triturating surface
23. MX, length of facial region
24. VO, vomer
25. VO, teeth
26. PAL, teeth
27. PAL, foramen palatinum posterius large
28. PT, vertical flange on proc pter ext
29. PT, interpterygoid vacuity
30. PT, canalis caroticus internus, posterior to foramen basisphenoidale
31. PT, canalis caroticus internus at least partially formed by pt
32. PT, canalis caroticus lateralis vs canalis caroticus internus
33. PT, foramen posterius canalis carotici interni
34. PT, middle ear with ossified floor formed by posteromedial pterygoid process
35. PT, pt-bo contact
36. PT, deep concavity on qu ramus
37. PT, bs-pt space
38. PT, intrapterygoid slit
39. QU, posterior pocket for stapes
40. QU, cavum tympani
41. QU, middle ear with complete lateral wall
42. QU, cavum tympani curved dorsally
43. QU, acute posterior margin
44. QU, incisura columellae auris
45. QU, antrum postoticum
46. QU, covers op laterally
47. QU, cranioquadrate space
48. OP, processus paroccipitalis
49. OP, fenestra perilymphatica
50. PR, anterior surface otic chamber
51. PR, otic chamber thickness
52. SO, crista supraoccipitalis
53. SO, wide occipital plate
54. SO, supraoccipital with horizontal plate
55. EX, recessus scalae tympani
56. BS, basipterygoid articulation
57. BS, foramen anterius canalis carotici interni
58. BS, foramen basisphenoidale
59. BS, foramen posterius canalis carotici interni
60. BS, paired pits
61. BS, ventral bs/bo tubercle
62. CA, columella auris
63. JW, splenial
64. CE, proportions
65. CE, central articulations
66. CE, anterior end of 8th cervical
67. CE, ribs
68. CE, transverse processes
69. CE, doubled transverse processes
70. CE, 4th cervical articulation
71. CE, double articulation between 5th and 6th
72. CE, double articulation between 6th and 7th
73. CE, double articulation between 7th and 8th
74. CE, strong ventral process
75. CE, neural spine on 8th cervical
76. CE, 8th cervical centrum
77. CAU, tail rings
78. CAU, tail club
79. CAU, biconcave caudal near base of tail
80. CAU, central articulations
81. CAU, chevrons
82. SH, coracoid foramen
83. SH, coracoid columnar
84. POSTC, paddles formed by limbs
85. PEL, pelvis sutured to shell
86. PEL, pelvis narrow
87. CAR, first thoracic rib
88. CAR, anterior articulation of first thoracic centrum
89. CAR, knob on ventral surface of nuchal that articulates with 8th cervical
90. CAR, 10th thoracic rib
91. CAR, sacrum with dorsal 10
92. CAR, 9th costal bone
93. CAR, suprapygal
94. CAR, supramarginal scales
95. PLA, plastral fontanelles surrounded by hyo-hypoplastra
96. PLA, carapace-plastron attachment
97. PLA, plastral buttresses
98. PLA, entoplastron separating epiplastra
99. PLA, dorsal process on epiplastron
100. PLA, epiplastron
101. PLA, epiplastral beak
102. PLA, bridge of plastron
103. PLA, mesoplastra
104. PLA, xiphiplastron
105. PLA, xiphiplastron median notch
106. PLA, gular scales
107. PLA, inframarginals
108. PLA, abdominal scute relative to midline
109. PLA, pectoral scute
110. PLA, femoral-anal sulcus
111. PLA, midline sulcus sinuous
APPENDIX 2
Data Matrix
“Synapsida/Diapsida”
00000000000000000000??00000?0000000000000000000000000000000000000000000000000000000000????????????0????????????
Proganochelys
0000000000000000000000000000000?000000000000000000000000100000000000000000000100000000000000?000000000000000000
Australochelys
?000?0??0??0?00000??000???0?000?0000001110000111??010001?0001??0???????????????????????????????????????????????
Palaeochersis
0000?0??0??0?00000??001???0?000?0000001111100111?0010001?00010?00?0?????????0000000000???????0000000000000?0??0
Proterochersis
????????????????????????????????????????????????????????????????????????????????????11000?1??100000000001000?00
Megapleurodira
0010110101(01)11200000(01)(01)(01)0111001201000(01)2011111(01)111100011011(01)3302100(01)(12)101(012)00000(012)0000111011(01)0001121(01)0(01)12000(01)01010?00
Kayentachelys
0010110111001000001000011?01000?000010111110111101111011000021?00000000000?000000110000000001200001000000000?00
Kallokibotion
000?110?1100100000100001110110000110101111111111?1111011?0002??00000000000000?000??0000000?13200112000000000?00
Mongolochelys
00101100110010101010010111011000010010111110111101111011?0002??000000000000?000001?00?001??00210011000000000?00
Pleurosternidae
00101101110(01)1000001010011101100011102011111011111111101101002?000000000000?00000011000000?01(01)2001110000000000(01)0
Baenidae
00101101110(01)1200001000011101100111102011111111111111101101002100(01)000100000000000011000000?01(14)200112000000000010
Xinjiangchelys
????11??????????????????????111001?0(12)011111?1111?11110?102112??0000100000000??10???00?100?01?2?10110011000000(01)1
Solnhofia
0010111111111?00001?0011110111100110201111101111011110111220210000100000000?00?00?1000?????162110????0100??????
Plesiochelyidae
001011111101?200001000011101111001(01)020111110111101111011042021000010000000000000011000000?0102(01)0112000100000?00
Meiolania
01(01)1110011001101110001011101111?01101111111111110111111102202?001100010000011100011000001101?2110110001000?00?0
Ninjemys
??0111?????01?0111??01011101111?01?0111111111111?11111?1????????????????????11?????????????????????????????????
Niolamia
011?1100??001?0111??000111?1111001?010111111111101111111?2102???????????????1??????????????????????????????????
Chubutemys
?01?11?0??0010000??00001110?111101?01011111?1111???11011?21021?012?1????0??2???????????0?????2?????????????????
Sinemys
00101101111212000001000111111110010010111110111101111011021?2???1201020?0012??110??00?000?01?2110121001001?00?0
Otwayemys
???????????????00????0??????????????????1???1??????????????????012?0010?0002???00???0?00?????21?012000100??0?00
Hangaiemys
0010111111021200001010011111111001111011111?1111?111101102212??01?11?100010???????????1????10201012101100100000
Dracochelys
?01?11?1?1121?00001?1001111?11100111101111101111?1111011?22?2??012000?0??112????????0?000?010??1012??110?1?0??0
Ordosemys
0010110?11111200001000011101111001111011111011110111101112112???1201010?0012??110??00?001???02110121?1100??0000
Judithemys
0010111?1102120000100001110111100110101111101111?111101102212??0111111??01010011011000100??102010121011101?0000
Chelydridae
1010111011121000001010(01)111011111011020111110111101111011042021111111111101110011011000110(01)01(01)211012(01)(01)11101(01)1110
Chelonioidea
001011(01)(01)1(01)(01)(12)100000100(01)0111011110011(01)(12)011111011110(01)111011142(01)211111111100(01)1010001111100(01)(01)1001(023)21101210(01)1101(01)0010
Trionychoidea
1010111(01)111(12)120000110(01)01110111100110(12)011111011110(01)1110110(24)2021111211120111110001111000110101(25)2(01)0(01)120001000(01)000(01)
Testudinoidea
1010111(01)111(12)120000010(01)011101111101(01)0(12)01111101111011110110(24)2021111211111111120001111000110001(13)2(01)01120001001(01)0000