Valanginian occurrence of Pelomedusoides turtles in northern South America: revision of this hypothesis based on a new fossil remain

Pelomedusoides constitutes the most diverse group of Mesozoic and Cenozoic side-necked turtles. However, when it originated is still being poorly known and controversial. Fossil remains from the Early Cretaceous (Valanginian) Rosa Blanca Formation of Colombia were described almost a decade ago as potentially belonging to Podocnemidoidea (a large subclade inside Pelomedusoides) and representing one of the earliest records of this group of turtles. Here, I revise this hypothesis based on a new fragmentary specimen from the Rosa Blanca Formation, represented by a right portion of the shell bridge, including the mesoplastron and most of peripherals 5 to 7. The equidimensional shape of the mesoplatron allows me to support its attribution as belonging to Pelomedusoides, a group to which the previously podocnemidoid material is also attributed here. Although the Valanginian pelomesudoid material from Colombia is still too fragmentary as to be considered the earliest indisputable record of the Pelomedusoides clade, their occurrence is at least in agreement with current molecular phylogenetic hypotheses that suggest they split from Chelidae during the Jurassic and should occur in the Late Jurassic and Early Cretaceous fossil record.


INTRODUCTION
One of the most diverse clades of Mesozoic and Cenozoic turtles is Pelomedusoides, with fossils worldwide distributed and extant representatives restricted to southern hemisphere (Ferreira et al., 2018;Gaffney et al., 2011;Gaffney, Tong & Meylan, 2006;Hermanson et al., 2020;Vlachos, Randolfe & Sterli, 2018). Recent molecular phylogenetic hypotheses suggest that they split from Chelidae during the Late Jurassic at 161.7 Ma (149.3-168.9 Ma) (Pereira et al., 2017), and total-evidence tip-dating (TE TD) suggest even older date for this splitting during the Early Jurassic at 172.6 Ma (Holley, Sterli & Basso, 2019). However, at present, the earliest indisputable fossil record of Pelomedusoides is the bothremydid Atolchelys lepida (Romano et al., 2014), from the upper Barremian of Brazil; meaning approximately 36 Ma of ghost-lineage.
Almost a decade ago, I described some fragmentary material from the Early Cretaceous (Valanginian) Rosa Blanca Formation of Colombia, which I attributed as potentially belonging to Podocnemidoidea (one of the subclades inside Pelomedusoides) (Cadena, 2011). This occurrence has been questioned and considered dubious by Romano et al. (2014), arguing that the presence of an inguinal buttress that medially extends onto the ventral surface of costal 5 is highly variable within Pelomedusoides, even within all of Testudines. Here, I present new material from a locality nearby to the one from where the material described in 2011 came from; from the same segment of the Rosa Blanca Formation (Fig. 1). This new fossil material allows me to revise the hypothesis proposed back in 2011, and present new evidence and comparisons that support the occurrence of Pelomedusoides during the Valanginian in northern South America. Carapace length estimation. In order to establish an estimation of the total length of the carapace to which the fossil fragment belongs, I measured fifteen specimens of extant podocnemidids that I have examined in recent years, as well as the extinct taxa Francemys gadoufaouaensis from Pérez-García (2019) and Atolchelys lepida from Romano et al. (2014) (Data S1). Using the software Image J2 (Rueden, Schindelin & Hiner, 2017), I set the scale to the one provided in the photos or figures of the specimens and measured the maximum length of both mesoplastra, peripherals 6, 7, and the total length of the carapace (according to their individual preservation). I established the simple linear regression and its equation using Microsoft-Excel (Data S1), and used it to estimate the maximum length of the carapace of UR-CP-0025.  (1996); Dortoka vasconica from Lapparent de Broin & Murelaga (1996); Pelomedusa subrufa CRI-5200, Podocnemis expansa USNM-29476 and Chelus fimbriata MNHN-2581A from personal reference photo gallery; and UR-CP-0025 from this study.

Systematic paleontology
PLEURODIRA Cope, 1864 PELOMEDUSOIDES Cope, 1868 Incertae Sedis respectively. Medially and posteromedially is in contact with the right hyoplastron and right hypoplastron respectively. The most lateral portion of peripherals 5 to 7 is missing (natural breaking). Also in this view, there is evidence of some of the sulci, particularly between marginals and of these with the abdominal scute. There is not indication that the pectoroabdominal sulcus reached the anterolateral corner of mesoplastron. In dorsal and lateral views (Figs. 2C-2F), the peripheral 6 is the most complete of the three preserved, showing a rectangular shape with its most medial margin (contact with costals) missing.
The sulci between marginals are poorly preserved, however there is enough evidence that they were restricted to the peripherals, without reaching the costoperipheral sutural margin.
The sutural contact between peripherals shows a medial indentation (Figs. 2E-2F), however this seems to be due to that the bone is naturally cut and cancellous tissue exposed. In anterior view (Figs. 2G-2H), the peripheral 5 and mesoplastron contact is well defined, and the bridge angle formed between the peripherals and the plastron indicates that the shell was probably low to moderate dome-shaped. Also in this view is evident the considerable thickness of these bones. A close-up of the margin of peripheral 5 (Figs. 2I-2J) shows a very thin external bone cortex and abundance of large pores at the cancellous bone. A large (88.17 cm, carapace length) of the extant Podocnemis expansa USNM-29476 is shown in Fig. 2K for comparison and anatomical location of UR-CP-0025 in a turtle shell.

DISCUSSION
Mesoplastra of Pan-Pleurodira. The mesoplastra bones have exhibited important modifications along turtle evolution. In basal Pan-Testudines as for example Odontochelys semitestacea they were two separate bony plates meeting medially (Li et al., 2008). In basal Testudines as for example Kayentachelys aprix there was a reduction in the number of mesoplastra, being only one pair extended medially reaching the central fontanelle (Joyce, 2007, fig 11). Another transformation of mesoplastra occurred in the both major groups of turtles, with their complete lost in Cryptodires and being one pair but they do not contact one another medially in Pan-Pleurodira (Cadena & Joyce, 2015;Joyce, 2007). Inside Pan-Pleurodira the mesoplastra have exhibited additional transformations from the primitive condition exhibited by Platychelyidae and Cretaceous-Paleocene members of Pan-Chelidae (Figs. 3A-3C, 3G-3I) of being almost triangular in shape, much wider than long to the condition exhibited by almost all Pelomedusoides of having almost equidimensional mesoplastra (Figs. 3E-3F, 3J-3L, 3P). An equidimensional mesoplastron was considered by Gaffney, Tong & Meylan (2006) as characteristic of a Nanorder that they defined as Eupleurodira (Cheloides = Pan-Chelidae plus Pelomedusoides). As I show in Fig. 3, the condition in pan-chelids who have mesoplastra is similar to the one exhibited by platychelids, which allow me to suggest that instead this is a characteristic of Pelomedusoides, shared by UR-CP-0025 described herein (Fig. 3E). Another transformation of mesoplastra inside Pan-Pleurodira is their complete lost in some Dortokidae, some Araripemydidae, crown-Chelidae, and Pelusios spp. (Gaffney, Tong & Meylan, 2006) (Figs. 3M-3O).
Carapace size estimation of UR-CP-0025.Using the simple linear regression equation (y = 6.5362x + 5.5002, x corresponding to maximum mesoplastron length, and y maximum carapace length) obtained from specimens of extant podocnemidids and some fossil pelomedusoids (Data S1). I estimated that the length of the carapace of UR-CP-0025 was of ∼40.59 cm, indicating a much larger size in contrast to the exhibited by Jurassic and Early Cretaceous platychelids, which fluctuated between 20 to 27 cm (Cadena, Jaramillo & Bloch, 2013, , 2017), therefore their fossil record should be expected to occur in Late Jurassic and Early Cretaceous sequences (Fig. 3Q). However, it is important to point out that UR-CP-0025 and the material previously described also from Rosa Blanca Formation (Cadena, 2011) are still too fragmentary to be recognized as the earliest indisputable record of the group, which it is not intention of this study. With this study, I once again showed that the Rosa Blanca Formation is still being a very productive and promising rock sequence in northern South America for future paleontological studies and the understanding of the Early Cretaceous faunas, including the evolution of Pelomedusoides turtles.

CONCLUSIONS
With this study, I once again showed that the Rosa Blanca Formation is still being a very productive and promising rock sequence in northern South America for future paleontological studies and the understanding of the Early Cretaceous faunas, including the evolution of Pelomedusoides turtles. Despite the taxonomic uncertainty that fragmentary material as UR-CP-0025 has, it exhibits the characteristic equidimensional mesoplastron of Pelomedusoides turtles, supporting the occurrence of this group of turtles in northern South America during the Early Cretaceous.

Animal Ethics
The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers): The Ethics Committee of Universidad del Rosario gave aval for the development of this project under document number DVO005 672-Cv1066.

Field Study Permissions
The following information was supplied relating to field study approvals (i.e., approving body and any reference numbers): Local landowner Roberto Serrano verbally authorized the paleontological exploration of the zone and collection of the specimen.

Data Availability
The following information was supplied regarding data availability: The raw measurements are available as a Supplemental File.