Interpreting character variation in turtles: Araripemys barretoi (Pleurodira: Pelomedusoides) from the Araripe Basin, Early Cretaceous of Northeastern Brazil

Introduction

Turtles are remarkable reptiles that possess a bony shell alongside other unique morphological characters, such as a shoulder girdle enclosed by the ribcage. The order Testudines is nowadays split in two monophyletics groups: Pleurodire (side-necked) and Cryptodire (hidden-necked) turtles (Gaffney, 1975; Joyce & Bell, 2004).

Pleurodires are split in two clades: Cheloides and Pelomedusoides (Gaffney, Tong & Meylan, 2006; Sereno & ElShafie, 2013). The Cheloides comprise a single lineage, Pan-Chelidae, ranging from Cretaceous to present day (Lapparent de Broin & De la Fuente, 2001; De la Fuente, 2003; Bona & De la Fuente, 2005; Gaffney, Tong & Meylan, 2006; Maniel & De la Fuente, 2016; De la Fuente et al., 2017), while the Pelomedusoides comprise six families, the extant Pelomedusidae and Podocnemididae, and the extinct Bothremydidae, Euraxemydidae, Peiropemydidae and Araripemydidae (Ferreira et al., 2018).

Araripemydidae was first erected by Price (1973) to accommodate Araripemys barretoi from the Romualdo Formation of the Araripe Basin. Later on, Lapparent de Broin (1980) described Taquetochelys decorata, from the Elharz Formation, based on shell fragments. This taxon was assigned to the clade Araripemydidae based mainly on shell texture, which was deemed similar to the distinctive pitted ornamentation pattern of A. barretoi. The genus was considered as incertae sedis within Pelomedusoides for lacking diagnostic features due to the fragmentary state of the material (De la Fuente & Lapparent de Broin, 1997; Gaffney, Tong & Meylan, 2006) and was regarded as a nomen dubium by Sereno & ElShafie (2013). Fielding, Martill & Naish (2005) described a new species for the genus Araripemys (A. ‘arturi’) based on fragmentary material from the Crato Formation (Araripe Basin), but the taxon was deemed poorly established and ended synonymised with A. barretoi in subsequent reviews (Gaffney, Tong & Meylan, 2006; Oliveira & Kellner, 2007). Araripemydidae was thus a redundant taxon until the description of Laganemys tenerensis Sereno & ElShafie (2013), based on an almost complete skeleton from the Elharz Formation of Niger, which was allocated within this family. Recently, Pérez-García (2018) has proposed a synonymisation between T. decorata and L. tenerensis, retaining the species within Araripemydidae.

The genus Araripemys is represented by the single species A. barretoi. The holotype and most of the specimens known come from the Romualdo Formation (Early Albian) of the Araripe basin (Price, 1973; Kischlat & de Almeida Campos, 1990; Schleich, 1990; Meylan, 1996; Gaffney, Tong & Meylan, 2006), but there are records also for the Crato Formation (Late Aptian) of the same basin (e.g. Fielding, Martill & Naish, 2005; Oliveira & Kellner, 2005, 2017; Romano et al., 2013) and for the Itapecuru Formation (Albian) of the Parnaiba Basin (Batista & Carvalho, 2007). There is a large number of specimens retrieved from the Romualdo Formation nodules displaying the characteristic 3D preservation of this formation, with several of them exhibiting both cranial and post-cranial elements. However, for the Crato Formation the only known specimens are, so far, a fragment of a shell along with a hindlimb which was the holotype of A. ‘arturi’ (Fielding, Martill & Naish, 2005); an isolated plastron (Oliveira & Kellner, 2005); two partially complete juvenile specimens displaying incomplete skulls assigned to Araripemys cf. A. barretoi (Oliveira & Kellner, 2017); and an almost complete post-cranial skeleton (SMNK, no collection number provided) which was figured by Naish (2007), but remains undescribed. No complete specimen has ever been described for the Crato Formation so far.

Morphological variation remains the keystone over which most of the natural sciences are built. Following the very didactic paper by Grande (2004), there are three main categories of morphological variation: taxonomic, ontogenetic and individual.

Pritchard (1988) has already drawn attention to how uneven are the studies on intraspecific variation in turtles, with a ‘frustratingly incomplete’ literature concerning extant species. Adding to the problem, palaeontologists usually have to study shell-only material, as it is the most common form of preservation: complete specimens retaining the skull are a rare find. Using A. barretoi as an example, for which dozens of shell-only specimens are known (the holotype, for instance, is a shell-only material), there are only four nearly complete skulls described (Gaffney, Tong & Meylan, 2006).

The identification of morphological variations in fossil turtles has important implications for phylogenetic analysis. The diagnostic features presented by Fielding, Martill & Naish (2005) to erect a new Araripemys species, A. ‘arturi’, were invalidated, those being: (1) ovoid-shaped carapace, whereas A. barretoi would display a posterolateral angulation at the carapace lateral margin (Price, 1973; Meylan, 1996); (2) peripherals 9 and 10 equally long as wide; and (3) lack of arrow-shaped pedal unguals, supposedly characteristic of A. barretoi after Meylan (1996). The first has been proposed to be taphonomical or dubious due to the material incompleteness; while the second could represent ontogenetic, sexual or individual variation; and the third, in turn, has been proposed to represent individual variation (Gaffney, Tong & Meylan, 2006; Oliveira & Kellner, 2007; Pérez-García, 2018). Although the new species were synonymised with A. barretoi, the very nature of the features claimed were never elucidated. The knowledge of morphological variation due to sexual dimorphism as well as individual and ontogenetic variation will allow a refinement of the characters used for phylogenetic or systematic purposes, at least at the species taxonomic level.

The turtle shell presents a limited number of characters and shows a wide spectrum of individual variation, the most usual variation occurring on the neural bones of the carapace (Romano et al., 2013). The supposed ancestral condition of eight predominantly hexagonal neurals can be modified either by proliferation or reduction of elements in the series; however, it was suggested that neomorphic elements occur rarely, while elimination and subsequent reduction in neural numbers would be far more frequent (Pritchard, 1988). Changes occur usually at the end of the series, by loss of exposed neurals, by obliteration (when the neurals are present but covered by the surrounding bones), and by changes in shape (Pritchard, 2008). Intraspecific variation of neural bones is widespread among living taxa, but very unevenly documented (Pritchard, 1988, 2008, and references therein). It was first reported for A. barretoi by Meylan (1996) concerning the contacts of the last neural. The plastron shows less variation in its number of components than the carapace, but displays as much or even more variation in shape (Pritchard, 2008).

Another important source of intraspecific variation, this one well established in living forms but sometimes not so easy to determine in the fossil record, is sexual dimorphism. It can be observed in turtles usually relative to shell and plastron features. On sexual dimorphism in recent and fossil taxa see for example Lagarde et al. (2001), Lapparent de Broin, De la Fuente & Fernandez (2007), Pritchard (2008), Cadena, Jaramillo & Bloch (2013), and Cadena (2015), among others.

The goal of this work is to describe 14 new specimens of A. barretoi (three from the Crato Formation and 11 from the Romualdo Formation) and reexamine other five (one from the Crato Formation and four from the Romualdo Formation), and to report all morphological variation, establishing patterns of intra-specific variation for this species. Such approach has never been done before for A. barretoi, despite the high number of known specimens. This new data will add to the current knowledge on the morphology of A. barretoi and will allow a refinement of the characters used in systematic analyses.

Geological Setting

The Araripe Basin is a sedimentary basin whose origin is related to the breakage of Gondwana and the opening of the South Atlantic Ocean during the Early Cretaceous. This intracratonic basin is located on the borders of the states of Ceará, Pernambuco and Piauí in Northeastern Brazil. It has an area of approximately 9,000 km² encompassing not only the Araripe Plateau but also the Cariri Valley (Valença, Neumann & Mabessone, 2003).

The Santana Group is the most fossiliferous unit, and includes the Crato and Romualdo Formations. These are two konservat lagerstätten famous for their taxonomic diversity and extraordinary preservation of organic structures, including remarkable soft tissues (e.g. Martill, 1988; Martill, Bechly & Loveridge, 2007; Campos & Kellner, 1997; Kellner, 1996a, 1996b; Pinheiro et al., 2012; Maldanis et al., 2016). However, there is not much agreement concerning its stratigraphy or age, which is usually estimated as Aptian–Albian; the reader is referred to Martill (2007) for a review. In this work we use the stratigraphic framework of Valença, Neumann & Mabessone (2003), with the Santana Group being constituted, from the bottom to the top, by the Rio Batateira, Crato, Ipubi, Romualdo and Arajara Formations.

Materials and Methods

Results

Systematic Palaeontology

Testudines Batsch, 1788

Pleurodira Cope, 1864

Pelomedusoides Cope, 1868

Araripemydidae Price, 1973

Araripemys Price, 1973 Araripemys barretoi Price, 1973

Synonymy. Araripemys arturi Fielding, Martill & Naish, 2005

Holotype. DGM 756-R, incomplete shell (cast MN 6945-V).

Locality and Horizon. Outcrops of the Crato (Aptian–Albian) and Romualdo Formations (Albian), in the Santana Group of Araripe Basin, Northeastern Brazil.

Revised diagnosis. Pelomedusoid with extensive temporal and cheek emargination; basisphenoid very long nearly reaching palatines; narrow triturating surface without accessory ridges; thin labial ridge; incisura columela auris not enclosed by bone; fossa precolumellaris deep; quadrate-basioccipital contact absent; ventral exposure of the occipital extensive; prootic widely exposed ventrally; processus interfenestralis of opisthotic widely exposed ventrally; processus paraoccipitalis of the opisthotic surpassing by far the supraoccipital crest; very long neck; postatlantal cervical vertebrae with completely fused postzygapophyses; first thoracic strongly sutured to the nuchal; medial and lateral centralia absent; flattened shell; fine ornamentation consisting variably of pits and/or ridge-and-sulcus configurations; first costals separating nuchal from first peripherals; small and subtriangular peripheral 1; neural series including 8–10 neurals; persistence of costal fontanelles in mature forms; reduced cruciform plastron with well-developed axillary and inguinal buttresses; inverted V-shaped entoplastron; J-shaped epiplastra forming a sharp point anteriorly; absent mesoplastra; absent gular scutes; three midplastral fontanelles.

Description

The full, detailed and individual descriptions of all newly reported and redescribed specimens can be found in the Supplemental Material. In this section we combine and compile data on all specimens (mentioning them wherever appropriate) focusing on new morphological data and on observed variations. These data comprise features of the skull, shell, and unguals. Variations are also compiled in Table S1 of Supplemental Material 2.

We present here data on 14 new specimens, as well as revisit five previously reported specimens. The new specimens are, from the Crato Formation: LP-UFC 722 (almost complete specimen exposed in ventral view, with a skull in dorsal view; Figs. 1 and 2), MPSC R 2107 (almost complete plastron in ventral view; Fig. S1), UFRPE 5302 (posterior half of carapace in dorsal view; Fig. S1); and from the Romualdo Formation: MN 6949-V (almost complete shell and skull, plus hyoids, two cervical vertebrae, and a complete manus; Figs. 3 and 4), MPSC R 2308 (entire shell; Fig. S2), DGM 346-LE (almost complete shell; Fig. S2), MN 6744-V (almost complete shell, two cervical vertebrae; Fig. S2), MN 6743-V (partial shell; Fig. S3), DGM 1449-R (carapace; Fig. S4), MPSC R 010 (entire shell), MPSC R 874 (almost complete carapace), MPSC R 134 (carapace; Fig. S4), MPSC R 137 (carapace; Fig. S4), SMNK PAL no number (carapace, cervicals and skull; Fig. S4). We further revisit specimens DGM 756-R (partial shell and femora, holotype; Fig. S2), MN 6637-V (carapace impression, plastron, several post-cranial elements including an ungual; Fig. S3), SMNK PAL 3979 (partial shell and hindlimb, holotype of A. “arturi”; Fig. S3), BSP 1977 I 1 (almost complete shell with a few postcranial elements, including an undescribed ungual; Fig. S3) and BSP 1981 I 38 (almost complete shell; Fig. S3).

Skull

The skull of A. barretoi has already been described in detail by Meylan (1996) and Gaffney, Tong & Meylan (2006). We present here two new well-preserved skulls, providing further data on its morphology and character variation.

The skull of LP-UFC 722, from the Crato Formation, is almost complete and exposed in dorsal view, but some details from the ventral view can be assessed from the CT scan-generated 3-D model (Fig. 2). Overall, the pattern of bone elements and contacts is similar to what has already been described for A. barretoi. This skull exhibits an ‘extra foramen’ in the opisthotic–exoccipital suture (Fig. 2D), similar to AMNH 24454 (Gaffney, Tong & Meylan, 2006) and unlike the other known skulls. The foramen jugulare posterius is clearly open as seen in the CT scan-generated 3-D model (Fig. 2D). Both hyoids are preserved, the right one being exposed and the left one hidden ventral to the skull. It can be seen in the CT scan-generated 3-D model. The files for the slices of the CT scan can be assessed through a link available in the Supplemental Material 1.

MN 6949-V includes a three-dimensional, well-preserved skull from the Romualdo Formation (Fig. 3). Its general morphology is also consistent with that previously described (Meylan, 1996; Gaffney, Tong & Meylan, 2006). Differently from LP-UFC 722, its foramen jugulare posterius is completely enclosed by a bony bridge and the ‘extra foramen’ is absent. This specimen further exhibits the stapes preserved in natural position (Fig. 3), conforming the previous identification of the incisura collumela auris by Gaffney, Tong & Meylan (2006). A case of asymmetry can be seen regarding this bone. In MN 6949-V, the right coronoid is quite small and forms the posterior half of the coronoid process. The anterior half is formed almost entirely by the dentary, with only a slender dorsal projection of the coronoid, similarly to what is seen in AMNH 24454 and THUg 1907 (Gaffney, Tong & Meylan, 2006). The left coronoid, on the other hand, extends onto and occupies entirely the anterior half of the process. Again, the same morphology is seen in AMNH 24454 and in THUg 1907 the coronoid occupies most of the anterior half of the coronoid process (Gaffney, Tong & Meylan, 2006).

These two skulls are the most complete up to now, especially MN 6949-V. This specimen shows quite clearly the presence of a considerable premaxillary prognathism in A. barretoi (Fig. 3). They both further reveal the presence of an extensive foraminization in the surfaces of the premaxillae, maxillae and dentaries, presumably neurovascular and related to the horny beak.

Carapace

Elements and contacts. The composition of the carapace in A. barretoi varied in the observed specimens mainly in respect of the number of neurals, varying from eight to 10 (see Table S1; Fig. 5). Aside from that, all observed specimens exhibit a nuchal, eight pairs of costals, 11 pairs of peripherals, a pygal and a suprapygal. In all specimens where the region could be observed, the nuchal was separated from the first pair of peripherals by the first pair of costals. Concerning also the contacts between the neurals and surrounding elements, the neural series shows great variation (Table S1). In specimens MN 6744-V, UFRPE 5302 and probably DGM 1449-R there was no contact between the last neural and the suprapygal, with the last pair of costals slightly meeting medially. Finally, the pygal also showed some variation. In specimen UFRPE 5302 the pygal has a unique triangular shape and there is no contact between the pygal and suprapygal, with the last pair of peripherals meeting dorsally to the pygal.

Costal fontanelles. The specimens showed variation concerning the relative sizes of the costal fontanelles. The smallest of all specimens herein described (MPSC R 137) shows quite large fontanelles (Fig. S4E). Larger individuals tend to exhibit relatively larger costals and smaller fontanelles. The two largest specimens (DGM 1449-R; MPSC R 134) only show diminutive fontanelles (Fig. S4A–S4C).

Shape. The shape of the carapace in A. barretoi is known to vary from the squared morphology described by Price (1973) and Meylan (1996) to the oval morphology seen in SMNK PAL 3979 (Fielding, Martill & Naish, 2005). The new specimens herein presented exhibit a variety of shapes, including the oval shape and distinct forms of squared carapaces (Fig. 4; Fig. S2). Oval-shaped carapaces can be seen in specimens LP-UFC 722, DGM 346-LE, MN 6744-V and MPSC R 134, while specimens MN 6949-V, DGM 1449-R, MPSC R 010, MPSC R 874, MPSC R 135 and UFRPE 5302 exhibit all squared-carapaces. In the holotype, the marked angulation that gives the carapace a squared outline is formed between peripherals 8 and 9, as in specimens MN 6949-V and MPSC R 134. In specimen DGM 1449-R, this angulation is formed between peripherals 9 and 10, and in UFRPE 5302, MPSC R 137 and MPSC R 010, between peripherals 7 and 8.

Ornamentation. The ornamentation of nuchals, neurals and costals surface did not vary between the observed specimens. In the carapace, only peripherals exhibited variation in, with specimens DGM 1449-R and LP-UFC 722 exhibiting a combination of pits and ridge-and-sulcus pattern in such elements.

Plastron

Anal notch. Schleich (1990) described two specimens attributed to A. barretoi which display distinct anal notch morphologies (Fig. 6), one which is U-shaped (BSP 1977 I 1) and another V-shaped (BSP 1981 I 38). The V-shaped anal notch can be further seen in specimens MPSC R 010 and MPSC R 874, while the U-shape can be seen in LP-UFC 722, MPSC R 2107, MPSC R 2308, MN 6949-V, DGM 756-R and MN 6637-V.

Ornamentation. All observed epi- and entoplastra exhibited a pitted ornamentation. On the other hand, there was variation in the pattern seen in hyo-, hypo- and xiphiplastra (Fig. 7). Specimens DGM 756-R, SMNK-PAL 3979, MN 6949-V, DGM 364-LE, MPSC R 010 and MPSC R 2308 exhibited a pitted ornamentation in these elements, while the same elements showed a combination of pits and ridge-and-sulcus ornamentation in specimens LP-UFC 722 and MN 6743-V. Specimen MPSC R 2107 exhibited exclusively ridge-and-sulcus ornamentation in these elements, with no signs of pits.

Unguals. Variation in the unguals of specimens attributed to A. barretoi has already been reported, varying from arrow-head shaped in AMNH numbers 24453, 24454 and 24456 (Meylan, 1996) to simple unguals in SMNK PAL 3979 (Fielding, Martill & Naish, 2005) and MN 6949-V (Oliveira & Kellner, 2007). Here, we report on the unguals of MN 6637-V and LP-UFC 722 (see Supplemental Material), which are arrow-head shaped, and of MPSC R 010 and BSP 1977 I 1 (see Supplemental Material), which are simple.

Discussion

We have presented above a plethora of morphological variation within specimens of A. barretoi. Some of these variations are easily recognisable as intraspecific (e.g. ontogenetic changes, or individual variations, like polymorphisms and sexual dimorphism), while others pose challenges over their interpretations. Accurate discrimination of intra- and interspecific variations in the fossil record can be difficult, especially when control provided by extant analogues is inexistent or insufficient.

Understanding of variation seen in extant species should serve as a control for interpreting variation in the fossil record. Sometimes, osteological variation in extant turtles can be extensive, as what is seen in the orbital margin of Chelonia mydas (Cryptodira: Cheloniidae). Two specimens of C. mydas, a female and a male (respectively UFRPE 5383 and 5384) are here illustrated, and topographic variation can be observed on some of the skull bones (Fig. S5). On the female skull, the frontal composes a significant part of the orbital margin, standing between the pre-frontal and the post-orbital. However, on the skull of the male specimen, the frontal does not take part on the orbital margin; instead, the contact between the pre-frontal and post-orbital precludes the frontal from reaching the orbit.

In this section, we discuss the diversity and plausible explanation for all variations reported. One must bear in mind that not always the condition found/described in a holotype or figured specimen turns out to be the typical condition observed in a larger sample; see, for instance, the case of Amia calva and the single vs paired parietals (Grande, 2004).

Conclusions

Araripemys barretoi is a remarkable fossil taxon in bearing so many individuals. Including the specimens described herein, there have been a total of 45 specimens referred to A. barretoi in the literature, to our knowledge (Price, 1973; Kischlat & de Almeida Campos, 1990; Schleich, 1990; Meylan, 1996; Fielding, Martill & Naish, 2005; Oliveira & Kellner, 2005; Gaffney, Tong & Meylan, 2006; Batista & Carvalho, 2007; Carvalho, Ghilardi & Barreto, 2016; Oliveira & Kellner, 2017; this work). This has allowed us to compile a particularly detailed account of the morphological variation seen in a fossil turtle.

Our identification of two morphotypes characterised by variations unreported as intraspecific variations within any known species provides evidence for the possibility that what is currently known as A. barretoi is actually composed of two distinct species. In such a scenario, morphotype 1 would represent a new species, while morphotype 2 would correspond to the type species A. barretoi and A. ‘arturi’ as well as a junior synonym.

However, given our relatively low skull sample (which is still rather high for a fossil species), we refrain from affirming with confidence the existence of two species of Araripemys before more material can corroborate this hypothesis. A geometric morphometric analysis of the skull is also planned in the future development of this investigation, as a mean of further testing this hypothesis.

We finish by quoting Gaffney, Tong & Meylan (2006) who, in a memorable line in their paper, when trying to explain the meaning of the complex variation exhibited by the foramen jugulare posterius of A. barretoi, put plainly (but apparently not without some exasperation): ‘It all goes to show that some of this crap makes no sense at all’ (Gaffney, Tong & Meylan, 2006; p. 148, line 53). We hope we have helped to disentangle this issue a bit.

Supplemental Information