A new cryptodire from the Eocene of the Na Duong Basin (northern Vietnam) sheds new light on Pan-Trionychidae from Southeast Asia

Abstract

Striatochelys baba gen. et sp. nov. is a new pan-trionychid from the middle–upper Eocene (late Bartonian–Priabonian, 39–35 Ma) of the Na Duong Basin in northern Vietnam. It represents one of the best documented and most completely known Palaeogene pan-trionychid species from Asia. Striatochelys baba can be diagnosed by: (1) its relatively small size; (2) the absence of a preneural; (3) the presence of well-developed straight ridges on the costals and neurals in adults; (4) ridges more strongly developed posteriorly than anteriorly; and (5) entoplastron callosity in the shape of a bulge. A phylogenetic analysis recovers S. baba within Pan-Trionychinae in an unresolved polytomy with species of the genus Nilssonia. In addition, comparisons of the new taxon with plastomenids, pan-trionychids from the Palaeogene and extant trionychids from Asia demonstrate a particularly close resemblance to Nilssonia spp. Based on the close resemblance between S. baba and ‘Trionyx’ impressus (a pan-trionychid from the Maoming Basin of southern China), we here assign the latter species to the new genus Striatochelys. The close relationship between S. baba gen. et sp. nov. from the Na Duong Basin and S. impressa comb. nov. from the Maoming Basin further supports the hypothesis that a close connection between both localities existed, as already exemplified by other faunal elements such as pan-geoemydids and crocodylians.

http://zoobank.org/urn:lsid:zoobank.org:pub:E8E5C4DE-E52A-43FB-B9C1-CE3ED2ED66C1

Keywords:

• Eocene
• turtle
• Pan-Trionychinae
• Nilssonia
• Asia

Introduction

Pan-Trionychidae (softshell turtles) represents one of the most diverse aquatic radiations of cryptodiran turtles, with more than 107 fossil and extant species and a fossil record dating back to the late Early Cretaceous (Georgalis & Joyce, 2017; Joyce et al., 2004; Turtle Taxonomy Working Group, 2021; Vitek & Joyce, 2015). The group is characterized by having a reduced carapace lacking peripherals and pygals (Joyce & Lyson, 2017; Meylan, 1987). Softshell turtles range in size from less than 10–15 cm bony disc length (BDL) (e.g. ‘Trionyx’ gobiensis Danilov, Hirayama, Sukhanov, Suzuki, Watabe & Vitek, 2014 from the Late Cretaceous of Mongolia; extant Pelodiscus huangshanensis Y. A. Gong, Peng, Huang, Lin, Huang, Xu, Yang & Nie, 2021; extant Pelodiscus shipian S. Gong, Fritz, Vamberger, Gao & Farkas, 2022 from China) to a BDL of 97 cm (completely preserved carapace of an indeterminate taxon described by Gaffney [1979] from the Eocene Bridger Formation of North America), or even larger (estimated based on a fragmentary specimen described by Head et al. [1999] from the middle Eocene of Pakistan). The largest extant trionychids are slightly smaller than this and have a maximum BDL of 74.5 cm (Pritchard, 2001). Whereas extant trionychids are known from Asia, Africa, North America and Australasia, fossil representatives were more widespread and have also been found in Australia, Europe and South America (Böhme, 1995; Ernst & Barbour, 1989; Georgalis, 2021; Georgalis & Joyce, 2017; Joyce & Lyson, 2017; Vitek & Joyce, 2015). Recently, the in-group relationships between extant species were clarified with the use of molecular data (Le et al., 2014), but comparatively long ghost lineages are currently still reconstructed for several taxa (Danilov et al., 2014; Vitek & Joyce, 2015).

Pan-Trionychidae can be divided into three clades: Pan-Cyclanorbinae, Plastomenidae and Pan-Trionychinae (Joyce et al., 2018, 2021; Lyson et al., 2021). Plastomenidae went extinct during the Eocene (Jasinski et al., 2022; Joyce et al., 2009, 2016, 2018; Lyson et al., 2021), whereas the latter two groups still have extant representatives.

The oldest known pan-trionychids originated in Asia during the Early Cretaceous (‘Trionyx’ kyrgyzensis Nessov, 1995, Perochelys lamadongensis Li, Joyce & Liu, 2015, Perochelys hengshanensis Brinkman, Rabi & Zhao, 2017) (see Brinkman et al., 2017; Danilov & Vitek, 2013). However, the phylogenetic positions of these early members are poorly understood, and they are either recovered as basal pan-trionychids (in accordance with their old geological age) or nested inside Trionychinae, due to their derived set of characters (Brinkman et al., 2017; Georgalis & Joyce, 2017; Joyce et al., 2021). Upper Cretaceous sites in central and east Asia (China, Kazakhstan, Kyrgyzstan and Mongolia) have yielded abundant remains of pan-trionychids (e.g. Brinkman et al., 2017; Danilov et al., 2014; Danilov & Vitek, 2013; Georgalis & Joyce, 2017; Vitek & Danilov, 2010; 2012), but taxa from the Palaeogene are far less common. Kuhnemys palaeocenica (Danilov, Sukhanov, Obraztsova & Vitek, 2015) from the Thanetian (late Paleocene) of Mongolia and Drazinderetes tethyensis Head, Raza & Gingerich, 1999 from the Bartonian (late Eocene) of Pakistan are the only taxa that have not been assigned to the ‘waste-basket’ genus ‘Trionyx’ (Georgalis & Joyce, 2017). Among the latter, ‘Trionyx’ linchuensis (Yeh, 1962), ‘Trionyx’ gregarius (Gilmore, 1934), ‘Trionyx’ johnsoni (Gilmore, 1931) and ‘Trionyx’ impressus (Yeh, 1963) from the Eocene of China and ‘Trionyx’ minusculus (Chkhikvadze, 1973) from the Eocene and ‘Trionyx’ ninae Chkhikvadze, 1971 from the Eocene and Oligocene of Kazakhstan are known. None of these taxa, however, has been included in a phylogenetic analysis, and their relationships are unresolved (Georgalis & Joyce, 2017).

‘Trionyx’ impressus from the Maoming locality of South China represents the only taxon of Bartonian–Priabonian age (late Eocene) from east Asia. Other named species and all other specimens from the late middle Eocene of Myanmar, east China (Zhejiang Province), or east-central China (Henan Province) can only be identified as Pan-Trionychidae indet. (Georgalis & Joyce, 2017).

Recent excavations in the middle–late Eocene deposits of the Na Duong Basin (northern Vietnam) have recovered a diverse vertebrate fauna that includes – besides fishes, pan-geoemydid turtles, crocodylians, birds and mammals – several well-preserved specimens of a pan-trionychid turtle, which has not been studied so far. Here, we describe the pan-trionychid turtle remains from Na Duong, and, based on a set of unique characters, assign it to a new genus and species. Interestingly, the new taxon shows several morphological similarities to ‘Trionyx’ impressus from the late Eocene of southern China, originally described as Aspideretes impressus by Yeh (1963) and subsequently revised by Danilov et al. (2013) and Georgalis & Joyce (2017). Due to the close resemblance of ‘T.’ impressus, we herein assign it to the new genus as well.

Geological setting

The Na Duong Basin is located in northern Vietnam close to the Chinese border (Fig. 1). It represents one of the few areas in East and Southeast Asia with a complete sequence of continental sediments from the middle Eocene to lower Oligocene (Böhme et al., 2013). The basin is part of the Cao Bang-Tien Yen fault system and covers an area of 45 km². The middle–upper Eocene (late Bartonian–Priabonian) Na Duong Formation is 240 m thick with the upper 140 m section being exposed in the Na Duong open cast coal mine. Böhme et al. (2013) biochronologically correlated the fossil mammals from Na Duong (layer 80) with the Chinese Naduan land mammal age (= Ulangochuian stage: Wang et al., 2019), and provided an age estimate of 39–35 Ma. Palaeomagnetic data calibrates the Ulangochuian at 40–37 Ma (Wang et al., 2019).

Figure 1. Map of northern Southeast Asia, showing the Na Duong Basin in north-east Vietnam and the Maoming Basin in southern China. The map was created using the Generic Mapping Tools program (Wessel et al., 2019).

The new taxon, together with most of the other vertebrates, was excavated within the transition zone between the coaly shale of the main seam and the underlying dark-brown claystone (layer 80). The layer 80 sediments represent lacustrine lignitic shales that were deposited at a time of tropical to warm-subtropical climate. The area was also undergoing a transition from shallow pond systems to a large anoxic lake (Böhme et al., 2013; Garbin et al., 2019). This fossil ecosystem has yielded both aquatic and terrestrial faunal elements. The new taxon occurred together with the pan-geoemydid Banhxeochelys trani Garbin, Böhme & Joyce, 2019, two species of crocodiles (Orientalosuchus naduongensis Massonne, Vasilyan, Rabi & Böhme, 2019 and Maomingosuchus acutirostris Massonne, Augustin, Matzke, Weber & Böhme, 2021), a bird (Massonne et al., 2022), anthracotheriids and rhinocerotids (Böhme et al., 2013), a strepsirrhine primate (Chavasseau et al., 2019), and fishes representing the families Amiidae and Cyprinidae (Böhme et al., 2013).

Materials and methods

All of the specimens of the new taxon described herein were found at the base of layer 80 (sensu Böhme et al., 2011) in the Na Duong coal mine. The material consists of two almost complete carapaces with associated plastral elements and seven partial carapaces, some of them with associated plastral elements. Additionally, a highly weathered skull and pectoral girdle, as well as several isolated plastral elements were found, which, however, cannot be associated with any of the carapaces. The best-preserved specimen and designated holotype consists of an almost complete carapace, which was found associated with several plastral elements (entoplastron, both hyoplastra, the right hypoplastron and the left xiphiplastron), as well as additional postcranial material (one cervical vertebra, the left pectoral girdle and humerus, the right radius and ulna, and a phalanx). The holotype material can be confidently referred to a single individual because it was found semi-articulated.

The data set of Joyce et al. (2018) (see Supplemental material File S1) was used for the phylogenetic analysis with taxa and rescorings of Lyson et al. (2021) added. The data set is derived from multiple previous data sets including Brinkman et al. (2017), Joyce and Lyson (2010, 2011, 2017), Joyce et al. (2009) and Meylan (1987). Together with the new taxon, the data set consists of 40 taxa and 95 characters. Adocus lineolatus Cope, 1874 and Carettochelys insculpta Ramsay, 1887 were chosen as outgroup taxa. For the analysis, 11 characters (1, 3, 5, 16, 20, 22, 41, 54, 79, 81, 94) were treated as ordered. A total of 40 characters (42%) could be scored for the new taxon.

The maximum parsimony analysis was conducted as a traditional search in TNT v. 1.5 standard version updated on 13 July 2022 (Goloboff & Catalano, 2016). The multistate characters mentioned above were treated as ordered; the maximum trees were set to 999,999 and the tree replication to 1000. For the branch swapping, tree bisection and reconnection with 10 trees saved per replication was used. Additionally, we applied molecular constraints based on the phylogenetic tree for extant taxa after Le et al. (2014) and an extended implied weighting k = 12 to decrease the impact of variable characters (Goloboff et al., 2018) as in the analysis of Lyson et al. (2021) and Joyce et al. (2018).

Institutional abbreviations

AMNH, American Museum of Natural History, New York, New York, USA; FMNH, Field Museum of Natural History, Zoology Department, Chicago, Illinois, USA; GPIT, Geologisch-Paläontologisches Institut Tübingen, Tübingen, Germany; IVPP, Institute of Vertebrate Paleontology and Paleoanthropology, Chinese Academy of Sciences, Beijing, China; NHMUK, Natural History Museum, London, UK; UCMVZ, University of California Museum of Vertebrate Zoology, California, USA; USNM, National Museum of Natural History, Smithsonian Institution, Washington, DC, USA.

Taxonomic nomenclature

We follow the phylogenetic nomenclature of Joyce et al. (2021) and highlight all clade names defined herein in italics according to the PhyloCode.

Systematic palaeontology

Testudinata Klein, 1760, sensu Joyce et al., 2020

Cryptodira Cope, 1868, sensu Joyce et al., 2021

Pan-Trionychidae Joyce et al., 2004, sensu Joyce et al., 2021

Diagnosis

Members of Pan-Trionychidae can be diagnosed by the following shell features: (1) the presence of sculpturing that covers all metaplastic portions of the shell bones; (2) absence of scutes; (3) absence of peripherals; (4) absence of pygals; (5) absence of suprapygals; (6) boomerang-shaped entoplastron; and (7) a plywood-like microstructure in the metaplastic portions of the shell (modified after Georgalis & Joyce, 2017).

Striatochelys gen. nov.

Diagnosis

Striatochelys can be differentiated from other pan-trionychid genera by the combination of the following characters: (1) relatively small size, with a carapace length reaching an estimated maximum of 27 cm; (2) absence of a preneural; (3) strong straight ridges on the carapace, spanning from costal I to costal VIII; and (4) stronger ridges posteriorly.

Etymology

The genus name is female and derives from the Latin word ‘striatus’ meaning ‘streaked’ due to the multiple anteroposteriorly running ridges across the carapace, resulting in a furrowed surface, and ‘chelys’ meaning turtle in Greek.

Striatochelys baba sp. nov.

(Figs 2–7)

Figure 2. Striatochelys baba, holotype, GPIT-PV-112860-1, Na Duong Formation, middle–upper Eocene, Vietnam. Carapace in A, B, dorsal and C, D, ventral views. Abbreviations: co, costal; cp, costiform process; dr, dorsal rib; ne, neural; nu, nuchal; tv, thoracic vertebra; xi, xiphiplastron. Scale bar equals 5 cm.

Figure 3. Striatochelys baba, GPIT-PV-122867, Na Duong Formation, middle–upper Eocene, Vietnam. Carapace in A, B, dorsal and C, D, ventral views. Abbreviations: alp, anterolateral process; co, costal; dr, dorsal rib; hyo, hyoplastron; hyp, hypoplastron; ne, neural; plp, posterolateral process; tv, thoracic vertebra. Scale bar equals 5 cm.

Figure 4. Striatochelys baba, Na Duong Formation, middle–upper Eocene, Vietnam. Carapace of GPIT-PV-122872 in A, dorsal and B, ventral views. Carapace of GPIT-PV-112862 in C, dorsal and D, ventral views. Carapace of GPIT-PV-112861 in E, dorsal and F, ventral views. Carapace of GPIT-PV-122879 in G, dorsal and H, ventral views. Abbreviations: co, costal; dr, dorsal rib; ent, entoplastron; epi, epiplastron; hyo, hyoplastron; hyp, hypoplastron; ne, neural; nu, nuchal; tv, thoracic vertebra. Scale bar equals 1 cm.

Figure 5. Striatochelys baba, Na Duong Formation, middle–upper Eocene, Vietnam. Plastron of holotype (GPIT-PV-112860-2, GPIT-PV-112860-3, GPIT-PV-112860-4, GPIT-PV-112860-5, GPIT-PV-112860-1) in A, B, dorsal and C, D, ventral views. Missing bones are mirrored and faded. Right medial hypoplastron fragment (GPIT-PV-122873) in E, dorsal and F, ventral views. Left xiphiplastron (GPIT-PV-122866) in G, dorsal and H, ventral views. Abbreviations: alp, anterolateral process; amp, anteromedial process; ent, entoplastron; hyo, hyoplastron; hyp, hypoplastron; plp, posterolateral process; xi, xiphiplastron. Scale bars equal A–D, 5 cm; E–H, 1 cm.

Figure 6. Striatochelys baba, holotype, Na Duong Formation, middle–upper Eocene, Vietnam. Left pectoral girdle (GPIT-PV-112860-7) in A, dorsal and B, ventral views. Cervical vertebra (GPIT-PV-112860-6) in C, dorsal and D, ventral views. Left humerus (GPIT-PV-112860-8) in E, dorsal, F, ventral, G, anterior and H, posterior views. Right radius (GPIT-PV-112960-9) in I, dorsal and J, ventral views. Left ulna (GPIT-PV-112860-10) in K, dorsal and L, ventral views. Phalanx (GPIT-PV-112860-11) in M, dorsal and N, ventral views. Abbreviations: acd, anterior condylus; ap, acromion process; ca, capitellum; cor, coracoid; dp, dorsal process; gf, glenoid fossa; hh, humerus head; if, intertubercular fossa; lp, lateral process; mp, medial process; ph, phalanx; prz, prezygapophysis; pz, postzygapophysis; sc, scapula. Scale bar equals 1 cm.

Figure 7. Skull of a Pan-Trionychidae referred to Striatochelys baba, Na Duong Formation, middle to upper Eocene, Vietnam. Skull of GPIT-PV-122870 in A, dorsal and B, ventral views. Scale bar equals 1 cm.

Diagnosis

Striatochelys baba can be differentiated from Striatochelys impressa comb. nov. by the combination of the following characters: (1) a larger costal VIII, forming the posterolateral margin of the carapace; (2) presence of ridges on the neurals, which are straight anteriorly and sinusoidal posteriorly; and (3) entoplastron callosity in the shape of a bulge (unknown for S. impressa and possibly a synapomorphy for S. baba + S. impressa).

Etymology

The species name is derived from the Vietnamese word ‘Ba ba’, meaning softshell turtle, declared as substantive.

Holotype

GPIT-PV-112860, carapace, xiphiplastron, thoracic vertebrae (GPIT-PV-112860-1) and partial plastron (entoplastron, GPIT-PV-112860-2; hyoplastron right, GPIT-PV-112860-3; hyoplastron left, GPIT-PV-112860-4; hypoplastron right, GPIT-PV-112860-5) and further postcranial material (cervical vertebra, GPIT-PV-112860-6; left pectoral girdle, GPIT-PV-112860-7; left humerus, GPIT-PV-112860-8; right radius, GPIT-PV-112860-9; right ulna, GPIT-PV-112860-10; phalanx, GPIT-PV-112860-11).

Referred material

Additional material consists of: a nearly complete carapace missing only the nuchal with a right hyo- and hypoplastron (GPIT-PV-122867); a fragmentary carapace with a complete row of neurals and a nuchal (GPIT-PV-122879); the anterior part of a carapace associated with an entoplastron (GPIT-PV-122872); fragmented costals belonging to a larger individual (GPIT-PV-122875); fragmented costals and a right hypoplastron (GPIT-PV-122871); a costal fragment (GPIT-PV-122874); a medial part of a hypoplastron (GPIT-PV-122873); an anterior part of the carapace and right hyo- and hypoplastron (GPIT-PV-112861); a part of the carapace and plastron and a first thoracic vertebra (GPIT-PV-112862 and GPIT-PV-112863); multiple isolated costal fragments belonging to a single individual (GPIT-PV-112864); multiple carapace fragments belonging to a single individual (GPIT-PV-112865); left xiphiplastron and a right costal VIII (GPIT-PV-112866); and a single costal fragment (GPIT-PV-112867).

Additionally, an isolated pectoral girdle (GPIT-PV-122869) and skull (GPIT-PV-122870) putatively belonging to a single individual of a pan-trionychid turtle, as well as additional postcranial material consisting of an isolated costal fragment (GPIT-PV-122877), a phalanx (GPT-PV-122878) and a plastron fragment (GPIT-PV-112868) are tentatively referred to S. baba.

Type locality and horizon

The fossils were recovered from the base of layer 80 of the Na Duong coal mine (Böhme et al., 2013) in northern Vietnam (21°42.2′N, 106°58.6′E); Na Duong Formation, Eocene, late Bartonian–Priabonian age (39–35 Ma).

Remarks

The holotype carapace of S. baba was embedded with its ventral part facing upwards. This allowed for a negative relief of the dorsal carapace, especially the ridges, in the sediment below. Based on those ridges, a cast was modelled for the missing posterior most part, but does not cover the correct outline, as seen in the posteriorly complete specimen GPIT-PV-122867.

Additional species

Striatochelys impressa (Yeh, 1963).

Remarks

Striatochelys impressa (IVPP V1036) was originally described as Aspideretes impressus by Yeh (1963) from the late Eocene deposits of Maoming and was recombined as Trionyx impressus by Danilov et al. (2013). Georgalis & Joyce (2017) retained the taxon ‘Trionyx’ impressus as valid but treated its genus-level assignment as uncertain, pending a re-description of the holotype. Based on the high degree of similarity to the new material from Na Duong and a comparison to other relevant taxa, we unite S. impressa and the new species from Na Duong under a new genus name herein.

Description

Overall, most of the material is well preserved. It consists of two nearly complete carapaces with associated plastral elements, as well as vertebrae and appendicular elements. Additional postcranial material of the anterior skeletal region, including partial carapaces, plastral elements, vertebrae and appendicular material can also be referred to S. baba, but are slightly deformed and less well preserved. A single skull referable to a trionychid (here assigned tentatively to S. baba) was found, but it is unfortunately highly weathered and no sutures are visible.

Carapace

Two well-preserved carapaces are available, GPIT-PV-112860 (holotype; Fig. 2) and GPIT-PV-122867 (Fig. 3). Other less well-preserved carapaces are also available (GPIT-PV-122872, GPIT-PV-122879, GPIT-PV-112861, GPIT-PV-112862; Fig. 4), together with multiple costal fragments from at least four individuals. Striatochelys baba is a medium-sized pan-trionychid. The carapace of the holotype (Fig. 2) is from an adult (or near-adult) specimen based on the very short free ends of the ribs and represents the second largest individual, with a carapace length of 20 cm. Another individual is larger than the holotype but only preserves isolated costal fragments. The best preserved costal III? is one-third larger than its holotype equivalent and leads to an estimated maximum carapace length of around 27 cm, likely representing the maximum carapace size for the species. The maximum width is 18.4 cm for the holotype and around 24.5 cm in the largest individual. The carapace is oval in outline with its maximum width at the level of costal III and tapers anteriorly to a width of 35% of the maximum width at the level of the nuchal and posteriorly to a width of 55% of the maximum width at the level of costal VIII. The carapace consists of a single nuchal, seven neurals and eight pairs of costal bones. The dorsal ribs are only slightly visible in larger individuals (Fig. 2), whereas they are clearly pronounced in younger ones (Fig. 3).

The central part of the carapace exhibits up to 20 relatively straight prominent ridges with deep furrows in between. All ridges originate at costal I. The medial ridges extend mainly anteroposteriorly and reach costal VIII posteriorly, whereas the more laterally placed ridges diverge posteriorly and extend posterolaterally. However, even the rather straight medial ridges become more sigmoidal on neurals VI and VII (i.e. posteriorly). Approximately half of the ridges are continuous, while the other half become shallower or are even interrupted by the costal sutures. Generally, the medial ridges are more prominent than the lateral ones and become even stronger posteriorly. Two rows of ridges are further present on all neurals in large individuals (Fig. 2). In smaller individuals (e.g. GPIT-PV-122867; Fig. 3), ridges are only pronounced on the anterior neurals and are not visible on neurals VI and VII. This part (i.e. the anterior medial region) of the carapace becomes slightly domed in larger individuals. The lateral part of the carapace is, in contrast, narrower and flat and seems to develop only later in ontogeny and grows with age. Whereas, for example, in the holotype (Fig. 2) and costal III? of the largest individual this region is clearly defined, smaller individuals like GPIT-PV-122867 (Fig. 3) lack this region completely and the central ridged region fills out the complete size of the carapace.

Aside from the ridges, the overall sculpturing of the carapace comprises small pits, which is typical for pan-trionychids. The sculpturing pattern is sectioned into two regions. Towards the centre of the carapace, the sculpturing consists of multiple unordered pits, whereas towards the lateral margins the pits are ordered in lines following the outline of the carapace in multiple rows. The number of rows depends on the size and thus the age of the individual, with more rows being present in larger individuals.

Nuchal

The nuchal is an unpaired bone forming the anterior part of the carapace (Figs 2, 4). It is trapezoidal and around four times wider than long. In some specimens (e.g. GPIT-PV-122879; Fig. 4) there is a strong nuchal emargination, whereas in others, the anterior margin of the nuchal is only slightly convex. In dorsal view, the nuchal is strongly ornamented with many unordered pits anteriorly and shallow furrows posteriorly at the suture with costal I. Suprascapular fontanelles are not present between the nuchal and costal I of S. baba. As there is no trace of fontanelles even in the smallest preserved individuals (e.g. GPIT-PV-122872), the fontanelles most likely close very early in ontogeny. In ventral view, the anterior and posterior costiform processes are united and situated close to the suture with costal I. A prenuchal is absent.

Neurals

In three individuals of S. baba (GPIT-PV-112860 [holotype], GPIT-PV-122872, GPIT-PV-122879), the complete set of neurals is preserved. In these individuals, seven neurals are present in total (Figs 2–4). The neurals span from the posterior margin of the nuchal to the middle of the costal VII, but do not reach costal VIII. The preneural is either absent or fused with the first neural. The first neural is the largest neural of the row spanning from the posterior margin of the nuchal to the anterior margin of the costal II. The anterior margin of the first neural is rounded (GPIT-PV-112860, holotype) to pointed (GPIT-PV-122879) in contrast to the straight margins of the more posterior neurals. The first four (GPIT-PV-112860 [holotype], GPIT-PV-122872) to five (GPIT-PV-122879) neurals are hexagonal with a long anterior side and a small posterior one. A reversal occurs on the fifth (GPIT-PV-112860 [holotype], GPIT-PV-122872) or sixth (GPIT-PV-122879) neural. The sixth neural in GPIT-PV-112860 (holotype), and GPIT-PV-122872 is again hexagonal with a short anterior side. The seventh neural in all specimens is much shorter than the others and pentagonal with a pointed posterior tip.

Costals

Striatochelys baba has a total of eight paired costals (Figs 2–4). Costals I–VII are similarly-shaped and have a rectangular outline with costal I the anteroposteriorly longest and costal III the lateromedially broadest. The first three costals have a relatively straight, mediolaterally extending suture. More posteriorly, the suture between the costals extends posterolaterally. Costal VIII is trapezoidal and only slightly wider than long. Costals VIII contact each other only anteriorly, whereas posteriorly they are separated by a narrow notch. The posterior margin is only well preserved in GPIT-PV-122872. Medial to the ninth dorsal rib, the costal VIII shows a shallow indentation followed by a convexity. In ventral view, the costals contact two adjacent thoracic vertebrae medially. Depressions for a contact with the ilium are absent on the costal VIII. The tenth thoracic vertebra is only poorly preserved and nothing can be said about its shape. It is also unknown whether a tenth dorsal rib is absent or not preserved.

Plastron

A nearly complete plastron is preserved in GPIT-PV-112860 (holotype; Fig. 5). As for other pan-trionychids the plastron consists of two epiplastra, a single entoplastron, two hyoplastra, two hypoplastra and two xiphiplastra (Figs 2–5). For S. baba, the epiplastra are, however, unknown. A possible epiplastron is present in GPIT-PV-122872, located posteromedially to the entoplastron (Fig. 4B). The plastral bridge is short and only reaches less than half of the hyoplastron width. All plastral elements except for the entoplastron are sculptured with multiple small pits, similar to the carapace.

Entoplastron

The only well preserved entoplastron is GPIT-PV-112860-2 (holotype; Fig. 5). An additional, slightly weathered, entoplastron is preserved in GPIT-PV-122872 (Fig. 4B). The entoplastron is boomerang-shaped with two long posterolateral projecting branches. GPIT-PV-112860-2 (holotype) is asymmetrical, although this appears to be a preservation artefact with the left branch likely showing the correct morphology judging by the comparison with other pan-trionychids. The anterior part of the bone is lateromedially straight. The branches have a straight anterolateral margin and a convex posteromedial one. In lateral view, the entoplastron is slightly curved dorsally with the anterior part being more upturned than the posterior one. The dorsal surface of the bone is smooth, whereas the ventral surface has very prominent oval bulges, which likely represent entoplastral callosities, reaching from the anterior margin to the last third of the branches. The posterior part of the branches ends in a narrow process loosely contacting the hyoplastron.

Hyoplastron

The hyoplastron contacts the entoplastron anteriorly and the hypoplastron posteriorly (Figs 3–5). Four complete hyoplastra are preserved for S. baba (GPIT-PV-112860-3, holotype), GPIT-PV-112860-4 (holotype; Fig. 5), GPIT-PV-122867 (Fig. 3), and GPIT-PV-112861 (Fig. 4F). The hyoplastron is rectangular and three times broader than long. The lateral and anterior margins are straight, whereas the medial margin is convex. Posteriorly, it is sutured to the hypoplastron over its whole width. The connection appears, however, to be rather weak, as the two bones were found articulated in only one individual (GPIT-PV-112861). The ventral surface is sculptured with many small pits which are aligned in lines laterally and more diffusely ordered towards the centre of the bone and at its medial margin. The dorsal surface is smooth with the exception of the raised extension of the paired anterolateral processes. The anterior process is slightly longer than the posterior one. In contrast to the anterolateral processes, the anteromedial process is very short and consists of at least two short spikes, but the surface is abraded, making a correct assessment difficult. However, there is no sign of a strong serration. On the anterior margin, at the area of contact with the entoplastron, a shallow flap is developed in GPIT-PV-112860-3 (holotype) and GPIT-PV-112860-4 (holotype; Fig. 5). In GPIT-PV-122867 (Fig. 3) and GPIT-PV-112861 (Fig. 4F), the anterior margin is, however, completely straight, which might be due to either intraspecific or ontogenetic variation.

Hypoplastron

The hypoplastron contacts the hyoplastron anteriorly and the xiphiplastron posteriorly (Figs 3–5). Three complete hypoplastra are preserved for S. baba (GPIT-PV-112860-5 [holotype; Fig. 5]; GPIT-PV-122867 [Fig. 3]; GPIT-PV-112861 [Fig. 4F]). The hypoplastron can be divided into a small lateral part and a much larger medial part, which are connected by a narrow bridge. Anteriorly, it is sutured to the hyoplastron over the whole width. The connection appears to be weak, however, as only one individual (GPIT-PV-112861) has the two bones articulated. The ventral surface is sculptured with many small pits, which are aligned in lines laterally and more irregularly arranged towards the centre and medially. The dorsal surface is smooth except for the raised extension of the paired posterolateral processes. The posterior process is slightly longer than the anterior one. Posteromedially, the hypoplastron has multiple processes. The lateral-most of these projects into a notch between the two anterolateral processes of the xiphiplastron. Medial to this single process is a shallow notch, followed by a posteromedial fan of four connected processes that direct ventromedially (best visible in GPIT-PV-122867 [Fig. 3]; worn away in GPIT-PV-112860-5 [holotype, Fig. 5]). The fan is followed, again, by a small gap and then a single slightly enlarged and medially projecting anterior process (best visible in GPIT-PV-122867; Fig. 3).

Xiphiplastron

The xiphiplastron contacts the hypoplastron anteriorly (Figs 2–5). Only two xiphiplastra are preserved for S. baba (GPIT-PV-112860-1 [holotype; Figs 2C, D, 5C, D]; GPIT-PV-112866 [Fig. 5G, H]). The xiphiplastron is triangular with a straight anterior and medial margin and a convex lateral margin. Anterolaterally, there are two prominent processes with a deep notch in between, which encompasses the lateral-most medial process of the hypoplastron. It is not discernible, in any of the preserved specimens, whether the left and right xiphiplastra are sutured to each other along the midline. Although the straight medial margin is somewhat indicative of such a contact, GPIT-PV-112866 (Fig. 5G, H) has two separate medial processes, which are seemingly worn away in GPIT-PV-112860-1 (holotype), and that could have prevented a continuous sutural contact between the xiphiplastra. The ventral surface is sculptured with multiple small pits. Although the arrangement is more irregular than that on the hyo- and hypoplastron, weakly developed rows can be distinguished, roughly following the bone outline. In contrast to this, the dorsal surface is smooth (Fig. 5G).

Vertebrae

A single cervical vertebra is preserved (Fig. 6C, D), which is slightly deformed and weathered dorsally. Due to incomplete preservation, the bone cannot be assigned to an exact position in the neck. The anterior condyle is round. The prezygapophysis is larger than the postzygapophysis and closer to the centrum. The dorsal portion of the bone is heavily damaged and nothing can be said about the potential presence of a dorsal process. In ventral view, a midline keel projects from a point slightly posterior of the anterior condyle to the posterior end of the bone.

All thoracic vertebrae are fused to the carapace (Figs 2, 3). The first thoracic vertebra reaches anteriorly the level of the nuchal bone midlength. Its prezygapophyses are very broad. The first dorsal rib is not fused to the carapace and is short and slender (Fig. 4B, D; Supplemental material FIle S2). It originates at the anterolateral part of the first thoracic vertebra and projects posterolaterally until reaching the second dorsal rib. The more posteriorly positioned thoracic vertebrae are rather uniform in size and morphology. No caudal vertebrae of a pan-trionychid have thus far been recovered from Na Duong.

Pectoral girdle

The left pectoral girdle of the holotype GPIT-PV-112860 (Fig. 6A, B) was found in close association with the carapace. As in other pan-trionychids, the scapula and coracoid are tightly sutured at the glenoid fossa with the scapula forming the largest part of the fossa. The scapula has two processes: a dorsal process loosely contacting the carapace and a slightly shorter acromion process. The angle between the acromion process and the main body of the scapula is larger than the angle between the acromion process and the coracoid. The coracoid has a blade-like morphology and is larger than the dorsal process of the scapula. In GPIT-PV-112860-7 (holotype), the blade appears to be anteromedially convex, and thus differs from that of other pan-trionychids, in which the blade is convex posterolaterally. This is, however, likely an artefact stemming from taphonomic distortion.

Appendicular skeleton

Among the bones of the appendicular skeleton, only the left humerus (Fig. 6E–H), the right radius (Fig. 6I, J), the left ulna (Fig. 6K, L) and a single proximal phalanx (Fig. 6M, N) are preserved, which were all found associated with the holotype carapace (GPIT-PV-112860) and are also referred to the holotype individual. The general morphology of the appendicular skeleton does not differ from that of other members of Pan-Trionychidae. The humerus is ‘S’-shaped with a prominent medial process and a smaller lateral one. The humerus head is round to oval and the capitellum is small and round. The radius is slightly larger than the ulna and has an elongated morphology with a sloped proximal and a wide distal articulation surface. The ulna has a broad proximal and a narrower distal articulation surface. The phalanx is dumbbell-shaped with broader proximal and narrower distal articular surfaces.

Skull

Only a single, isolated pan-trionychid skull (GPIT-PV-122870; Fig. 7) has been recovered from Na Duong, which we refer tentatively to Striatochelys baba. It measures approximately 45 mm and, if corrected for the missing posterior part, it would have had a length of approximately 50 mm. The skull is dorsoventrally compressed, and its surface is highly weathered, to the extent that it is impossible to locate any sutures or other diagnostic traits. The material only allows for an assignment to Pan-Trionychidae based on its general slender outline, the anteriorly projecting snout region and the far posteriorly reaching supraoccipital crest. Although it was found isolated it most probably belongs to S. baba, as this is the only pan-trionychid currently known from the Na Duong Basin.

Intraspecific and ontogenetic variation

A high degree of intraspecific variation has been described in many pan-trionychid taxa (Meylan, 1987), and the same is true for Striatochelys baba. The carapace of the holotype (representing the largest nearly complete specimen) is almost planar laterally. In this area, ridges are absent or only weakly developed, although the typical trionychid ornamentation is present (Fig. 2). This lateral area is, however, almost absent in smaller individuals (Fig. 3), in which the ridges nearly reach the lateral margin of the carapace. The presence of this planar area, which is devoid of the otherwise prominent ridges, in isolated costals of another large individual therefore indicates ontogenetic variation in this feature. Interestingly, there is almost no variation in the number of ridges (up to 20) between individuals irrespective of their ontogenetic stage. Overall, the lateral part of the carapace changes much more during ontogeny than the central part, indicating growth from the lateral margin outwards, as in the extant Pelodiscus sinensis (Wiegmann, 1835) (Sánchez-Villagra et al., 2009).

The position of the reversal of the neural orientation in S. baba varies between the fifth and sixth neural. The complete neural row is only visible in three individuals, two of which show a reversal on the fifth neural (Figs 2, 3) and one a reversal on the sixth neural (Fig. 4G). This feature does not seem to correlate with ontogenetic stage, as the individual showing reversal on the sixth neural is intermediate in size.

The hyoplastron has two distinct morphologies in S. baba. In the holotype, the hyoplastron forms an anterior notch, into which the posterior process of the entoplastron projects (Fig. 5). This notch is missing in two smaller individuals (Figs 3, 4F). It is conceivable that the stronger contact between the hyoplastron and the entoplastron in the holotype, formed by the notch and the process, is related to the greater robustness of the plastron generally developing later in ontogeny, as is the case for some other species like P. sinensis or Apalone ferox (Schneider, 1783). To verify this hypothesis, however, more individuals, especially of the same size as the holotype, are needed.

Comparisons

Asian taxa from the Palaeogene

According to Georgalis and Joyce (2017), only a few named Pan-Trionychidae from the Palaeogene of Asia are diagnostic and these were chosen for comparison with Striatochelys baba. These species are Kuhnemys palaeocenica, Drazinderetes tethyensis, ‘Trionyx’ linchuensis, ‘Trionyx’ gregarius, ‘Trionyx’ johnsoni, ‘Trionyx’ minusculus, ‘Trionyx’ ninae and Striatochelys impressa (= ‘Trionyx’ impressus). Whereas most of these species preserve a complete or nearly complete carapace, in ‘T.’ linchuensis only the anterior right part of a carapace is known (Yeh, 1962, fig. 1.1), and ‘T.’ minusculus only preserves a right hyo- and hypoplastron (Chkhikvadze, 1973, pl. 4.2). In addition, comparison with K. palaeocenica is complicated by the juvenile status of this species. Comparison with the plastron is even more challenging as it is only partially preserved in K. palaeocenica (Danilov et al., 2015, fig. 1), ‘T.’ gregarius (Gilmore, 1934, fig. 3), ‘T.’ minusculus (Chkhikvadze, 1973, pl. 4.2) and ‘T.’ ninae (Vitek & Danilov, 2015, figs 2, 3).

Carapace

The carapace of S. baba reaches an estimated size of 27 cm in the largest individual (GPIT-PV-122875), which lies well within the range of ‘T.’ gregarius and ‘T.’ linchuensis (Gilmore, 1934; Yeh, 1962). Drazinderetes tethyensis, in contrast, is much larger, with a carapace length of 80 cm (Head et al., 1999). ‘Trionyx’ johnsoni and ‘T.’ ninae are only slightly larger than S. baba, with a carapace length of around 40 cm (Gilmore, 1931; Vitek & Danilov, 2015). Kuhnemys palaeocenica and S. impressa are each known from only one specimen and reach a carapace length of 12.5 cm and 14 cm, respectively (Danilov et al., 2015, fig. 1; Ye, 1994, fig. 70). The general shape of the carapace is oval in S. baba, which is consistent with most of the other species. It is noteworthy, however, that in D. tethyensis the posterior part of the carapace becomes proportionately narrower than in S. baba (Head et al., 1999, figs 3, 4), whereas in ‘T.’ gregarius, the carapace appears to be more rectangular (Gilmore, 1934, fig. 1). The most striking characteristics of S. baba are the pronounced ridges on the carapace, projecting in multiple, roughly parallel rows from costal I to costal VIII. In most of the species mentioned above, there are no such ridges, but they are present in S. impressa (Yeh, 1963, pl. XX) and, at least posteriorly, in ‘T.’ gregarius (Gilmore, 1934, fig. 1). Importantly, in both taxa, these ridges do not occur on the neurals as in S. baba (Fig. 2).

In S. baba, the nuchal is around four times broader lateromedially than long anteroposteriorly. This is also the case in K. palaeocenica, ‘T.’ gregarius, ‘T.’ ninae and S. impressa. In D. tethyensis, the nuchal is more robust and only three times broader than long (Head et al., 1999, fig. 3). The nuchal is not preserved in ‘T.’ linchuensis and ‘T.’ johnsoni. A shallow nuchal emargination, similar to the one seen in S. baba, is also present in ‘T.’ gregarius, ‘T.’ ninae and S. impressa, whereas there is no such emargination in K. palaeocenica (Danilov et al., 2015, fig. 1) or D. tethyensis (Head et al., 1999, fig. 3). Suprascapular fontanelles are only present in K. palaeocenica (juvenile) (Danilov et al., 2015: fig. 1) and ‘T.’ ninae (Vitek & Danilov, 2015, fig. 3), while they are absent in all other taxa considered here.

Striatochelys baba has seven neurals in total. This also the case for K. palaeocenica, ‘T.’ johnsoni and ‘T.’ ninae. In S. impressa, there are eight neurals (Yeh, 1963, fig. 70) and the number of neurals is unknown in D. tethyensis and ‘T.’ linchuensis. A distinct preneural element is only present in D. tethyensis (Head et al., 1999, fig. 3), whereas in the other Asian species, the preneural is absent. The last neural is always the smallest neural in the species considered here, except for K. palaeocenica, in which the last neural is slightly larger than the penultimate one (Danilov et al., 2015, fig. 1). In S. impressa, the region around the first neural is not preserved, making a statement on the presence or absence of a preneural challenging. The reversal of the neural orientation in S. baba varies between the fifth and sixth neural (Figs 2, 3, 4G), but intraspecific variation in the location of the reversal is common among Pan-Trionychidae (Meylan, 1987). A reversal on the fifth neural is also known for ‘T.’ johnsoni, ‘T.’ ninae and S. impressa, whereas it occurs on the sixth in ‘T.’ gregarius. In K. palaeocenica, the neurals are more uniformly shaped. The last neural does not reach costal VIII in S. baba, which is also true for the other species considered here, except for ‘T.’ gregarius and S. impressa, in which the additional eighth neural reaches costal VIII (Gilmore, 1934, fig. 1; Yeh, 1973, fig. 70).

Costals I–VII are almost uniformly shaped in S. baba. While this is true for most other species as well, in D. tethyensis and ‘T.’ ninae costal I is much shorter lateromedially and costal II much more expanded distally (Head et al., 1999, fig. 3; Vitek & Danilov, 2015, figs 2, 3). In S. baba, the anteroposteriorly longest costal is the first, whereas in K. palaeocenica, D. tethyensis, ‘T.’ gregarius and ‘T.’ ninae the second costal is longer (Danilov et al., 2015, fig. 1; Gilmore, 1934, fig. 1; Head et al., 1999, fig. 3; Vitek & Danilov, 2015, figs 2, 3). In S. baba, costal VIII is triangular and anteroposteriorly longer than costal VI and VII together, which is unknown for the other species from Asia. An overall similar but proportionately slightly smaller costal VIII is, however, present in D. tethyensis, ‘T.’ gregarius and S. impressa, in which it barely reaches the size of costal VI and VII together (Gilmore, 1934, fig. 1; Head et al., 1999, fig. 3; Ye, 1994, fig. 70). In K. palaeocenica and ‘T.’ ninae, costal VIII is much smaller and shorter than costal VII and only forms the posterior margin of the carapace (Danilov et al., 2015, fig. 1; Vitek & Danilov, 2015, figs 2, 3), and in ‘T.’ johnsoni it is additionally also more ‘U’-shaped than triangular (Gilmore, 1931, pl. XI; Ye, 1994, fig. 74).

Plastron

Striatochelys baba has a relatively well-developed and robust plastron. The entoplastron is boomerang-shaped with relatively short branches, which project posteriorly into a notch of the hyoplastron. Based on the width of the carapace, the two hyo- and hypoplastra must have nearly contacted each other at midline; yet the medial processes of the hypoplastra present in S. baba indicate that they did not have a sutural contact. It is difficult to ascertain whether the xiphiplastra were sutured to each other along their midline (Fig. 5). Unfortunately, among the Asian taxa, only K. palaeocenica and ‘T.’ gregarius preserve a complete plastron, which, in both cases, shows a completely different morphology. In these two species, the hyo-, hypo- and xiphiplastra are much farther apart from their counterparts and the entoplastron is slenderer with longer branches, which do not project posteriorly into a notch of the hyoplastron (Danilov et al., 2015, fig. 1; Gilmore, 1934, figs 3, 4). Furthermore, there is no callosity in the shape of a bulging area in the central region of the bone in ventral view in those species, as is the case for S. baba (Fig. 5).

The hyoplastron of S. baba is relatively long anteroposteriorly and has a straight anterior margin and, at least in the holotype (Fig. 5), a well-developed notch anteriorly, in which the branches of the entoplastron fit. Additionally, the anterolateral process consists of two short spikes and the anteromedial processes consist of at least two short spikes, but assessment is difficult due to an eroded surface. In K. palaeocenica, the hyoplastron also possesses a straight anterior margin, but it differs in being very slender, having no notch anteriorly and having very long anterolateral and anteromedial processes (Danilov et al., 2015, fig. 1). The hyoplastron of ‘T.’ ninae is very similar to that of K. palaeocenica, but is slightly more robust and has a bowed anterior margin (Vitek & Danilov, 2015, figs 2, 3). ‘Trionyx’ gregarius, on the other hand, is more similar to S. baba, but the anterior margin is more rounded than straight and there is no notch for encompassing the branches of the entoplastron (Gilmore, 1934, fig. 3). The hyoplastron of ‘T.’ minusculus is, again, quite similar to that of S. baba, but also appears to have a more rounded anterior margin and no notch for encompassing the branches of the entoplastron (Chkhikvadze, 1973, pl. 4.2).

The hypoplastron of S. baba is divided into an anteroposteriorly short lateral part and a longer medial part, which are separated by a marked constriction. The lateral part possesses two small posterolateral processes, and on the medial part, there are more than three short posteromedial processes arranged in a fan-like manner (Figs 3, 5E, F); the exact number of processes on the medial part of the hyoplastron is difficult to assess, as they are broken (Figs 3, 4F) or nearly completely eroded (as in the holotype; Fig. 5). As is the case for the hyoplastron, the hypoplastron of K. palaeocenica is much slenderer and the processes are much longer (Danilov et al., 2015, fig. 1). From ‘T.’ ninae, only fragments of a hypoplastron are preserved, but it is clear that the posterolateral process is much longer than in S. baba (Vitek & Danilov, 2015, fig. 2). In ‘T.’ minusculus, the lateral and medial processes are not preserved, but the lateral part of the hypoplastron is as long anteroposteriorly as the medial part, a condition differing from S. baba (Chkhikvadze, 1973, pl. 4.2). The closest resemblance to S. baba with respect to the hyoplastron is shown again by ‘T.’ gregarius. The only (small) differences are the slightly more robust hypoplastron of ‘T.’ gregarius and the larger paired posterolateral processes in this species (Gilmore, 1934, fig. 3).

The xiphiplastron of S. baba is triangular and robust. The xiphiplastra could have had a long sutural contact with each other along the midline (Fig. 5), although this is difficult to ascertain (see above). As for the other Asian taxa, xiphiplastra are only preserved in K. palaeocenica and ‘T.’ gregarius (Danilov et al., 2015, fig. 1; Gilmore, 1934, fig. 3). The xiphiplastra of these two species are rather similar to each other but differ strongly from those of S. baba. They are much slenderer, rectangular as opposed to triangular, and have four long processes one at each corner.

Gilmoremys from the Cretaceous of North America

Besides S. impressa, the closest resemblance to S. baba among Pan-Trionychidae is shown by some members of Plastomenidae from the Late Cretaceous of North America. The carapace and plastron of Gilmoremys lancensis (Gilmore, 1916) and especially of Gilmoremys gettyspherensis Joyce, Lyson and Sertich, 2018 are very similar in overall morphology to both species of Striatochelys.

Carapace

Gilmoremys lancensis has an overall carapace length of 34 cm and thus is larger than both S. baba and S. impressa (carapace lengths of 27 cm and 14 cm, respectively) as well as its sister taxon G. gettyspherensis (carapace length of 25 cm). The overall carapace shape is similar in G. lancensis and G. gettyspherensis, as well as in S. baba and S. impressa, all having an oval outline and a distinctive sculpturing on the dorsal surface of the carapace. G. lancensis differs from both G. gettyspherensis and the two species of Striatochelys in having sculpturing that consists of grooves (Joyce & Lyson, 2011, figs 8, 10). In G. lancensis, S. baba, and S. impressa, on the other hand, the sculpturing consists of well-developed longitudinal ridges, that reach from costal I anteriorly to costal VIII posteriorly (for S. baba, see Figs 2, 3; for S. impressa see Yeh [1963, pl. XIX, fig. 3, and pl. XX, fig. 1]; for G. gettyspherensis see Joyce et al. [2018, fig. 2]).

Figure 8. Strict consensus tree of 135 equally optimal trees, obtained from the maximum parsimony analysis of 40 taxa and 95 characters. Tree length = 327 steps; consistency index = 0.361; and retention index = 0.605.

The nuchal of G. lancensis is as wide as it is in Striatochelys. It differs from the latter, however, in the absence of a nuchal emargination (Joyce & Lyson, 2011, figs 8, 10). Further differences occur in the number of neurals. Striatochelys baba and G. lancensis both have a total of seven neurals, whereas there are eight neurals present in S. impressa and G. gettyspherensis. Another difference lies in the presence of a preneural and a small first neural in the two species of Gilmoremys (Joyce & Lyson, 2011, figs 8, 10; Joyce et al., 2018, fig. 2). In S. baba, the preneural is absent (Figs 2, 3); in S. impressa, the region is not preserved. In Gilmoremys, costal II is laterally enlarged and projects anteriorly, reducing the lateral margin of costal I (Joyce & Lyson, 2011, figs 8, 10; Joyce et al., 2018, fig. 2). In Striatochelys, by contrast, the distal part of costal II is not enlarged and the distal margin of costal I and II are similarly-sized (Figs 2, 3; Ye, 1994, fig. 70). Costal VIII is, in all four species, triangular and enlarged. It is largest in S. baba, in which it is anteroposteriorly slightly longer than costal VI and VII together, whereas it is barely the size of costal VI and VII together in the other species.

Plastron

In contrast to the carapace, which shows many (small) differences despite its overall resemblance, the plastron is extremely similar in Striatochelys and Gilmoremys. The hyoplastron of Gilmoremys has a well-developed anterior notch for the entoplastron (Joyce & Lyson, 2011, figs 9, 11; Joyce et al., 2018, fig. 3), a feature that is also present in one individual of S. baba (Fig. 5), but absent in two others (Figs 3, 4G, H). The posterolateral processes of the hypoplastron of Gilmoremys are slightly larger than in S. baba, while the fan-like posteromedial processes present in Striatochelys are only weakly developed or absent in Gilmoremys (Joyce & Lyson, 2011, figs 9, 11; Joyce et al., 2018, fig. 3). The xiphiplastron in all three species has the same triangular morphology but is anteroposteriorly slightly longer in Gilmoremys (Joyce & Lyson, 2011, figs 9, 11; Joyce et al., 2018, fig. 3) than in S. baba (Figs 5C, D, G, H).

Extant taxa from Southeast Asia

Southeast Asia has the highest diversity of extant trionychids. According to molecular data, most Asian taxa belong to a single monophyletic group, including five genera, i.e. Pelodiscus, Palea, Dogania, Amyda and Nilssonia, which likely originated between the Eocene and Miocene (Engstrom et al., 2004; Le et al., 2014; Pereira et al., 2017; Thomson et al., 2021). Based on the middle to late Eocene age and Southeast Asian distribution of Striatochelys baba, a comparison with selected members of this extant group is provided below.

Carapace

With a bony disc length (BDL) of 27 cm (GPIT-PV-122875), S. baba is far larger than the smallest Pelodiscus species, i.e. Pelodiscus huangshanensis and Pelodiscus shipian, which barely reach 10 cm BDL, but it is much smaller than the large Nilssonia and Amyda species (e.g. Nilssonia leithi [Gray, 1872], Nilssonia gangetica [Cuvier, 1825], Amyda cartilaginea [Boddaert, 1770]), with a BDL between 38 and 60 cm. In contrast to this, Dogania subplana (Geoffroy Saint-Hilaire, 1809) and Palea steindachneri (Siebenrock, 1906) have a roughly similar size, with 21.7 and 30 cm BDL, respectively (S. Gong et al., 2022; Y. A. Gong et al., 2021; Pritchard, 2001). The overall carapace morphology in different taxa of Pan-Trionychinae is very similar, but one notable difference is the presence or absence of multiple anteroposteriorly extending ridges across the carapace. Such ridges are prominent in S. baba (see above). In extant eastern Asian taxa, similar but weaker ridges are often found in early ontogenetic stages (e.g. in Pelodiscus sinensis; IVPP 525, USNM 539334). If such ridges are also present later in ontogeny (e.g. in A. cartilaginea; FMNH 11088, USNM 22522, 222521), they are much more weakly developed in comparison to S. baba and never appear as double rows on the neurals.

In S. baba, the nuchal is about four times wider than it is long. Among extant Eastern Asian taxa, this is only the case in the genus Pelodiscus (e.g. P. sinensis and Pelodiscus jiangxiensis Hou et al., 2021) and in D. suplana, whereas in other taxa the nuchal is only two to three times wider than long. The absence of a preneural, as in S. baba (Figs 2, 3), is common in extant Eastern Asian taxa. Notable exceptions to this are two species of the genus Nilssonia, i.e. Nilssonia hurum (Gray, 1831) and N. gangetica, where a separated preneural is formed. Striatochelys baba has seven neurals (Figs 2, 3). In extant taxa, the number of neurals is often variable within a given genus or species, ranging from seven to eight. However, some taxa of the eastern Asian group (D. subplana, A. cartilaginea, Nilssonia formosa [Gray, 1869] and N. hurum) invariably have eight neurals. The point of reversal for the neural orientation is less variable in extant eastern Asian taxa and is usually at the fifth or sixth neural. Only N. gangetica and N. hurum show the same intraspecific variability as S. baba (Figs 2, 3, 4G), in which the reversal occurs at either the fifth or the sixth neural. In S. baba, costals I–VII are relatively uniformly shaped (Figs 2, 3). This is also the case for most extant eastern Asian taxa, but in D. subplana (FMNH 224111, USNM 222523, UCMVZ 95937) and N. leithii (FMNH 224231), costal II is distally expanded, similar to the morphology of D. tethyensis and ‘T.’ ninae (see above). Remarkably, S. baba has an enlarged triangular costal VIII, which is longer anteroposteriorly than costal VI and VII combined (Figs 2, 3). In Nilssonia spp. (NHMUK 86.8.26.2, FMNH 223231) and A. cartilaginea (FMNH 11088, USNM 22522), costal VIII is also triangular and thus shaped similarly, but somewhat smaller than in S. baba, while in D. subplana, costal VIII has the same elongated morphology as costals I–VII.

Plastron

As mentioned above, the plastron of S. baba is relatively well developed and robust. In some extant east Asian species, i.e. A. cartilaginea and D. subplana, the plastron is much thinner and the hyo- and hypoplastra are much further apart from their counterparts. In other species, i.e. Nilssonia spp. and P. sinensis, however, the overall plastron morphology is very similar to that of S. baba.

The entoplastron of S. baba is relatively slender, it possesses a callosity in the shape of a bulge and its branches project into a notch of the hyoplastron (Fig. 5). In P. sinensis (IVPP 556, USNM 539335), the entoplastron is much larger and its branches are slenderer. Additionally, callosities are much less developed and the branches do not project into a hyoplastral notch. In A. cartilaginea (FMNH 11088, USNM 22522) and D. subplana (FMNH 224111, USNM 222523), the general shape of the entoplastron resembles the condition in S. baba, but callosities are much less developed and the branches do not extend into a hyoplastral notch. In N. gangetica, the entpoplastron is either similarly shaped (FMNH 260430, USNM 293693) or slightly slenderer (NHMUK 86.8.26.1) than in S. baba, the callosities are much more weakly developed and the branches do not project into a hyoplastral notch. In N. hurum (NHMUK 86.8.26.2), the entoplastron has a shape similar to that in S. baba and also shows a callosity in the shape of a bulge. However, the bulge is much more weakly developed than that of S. baba. Furthermore, the branches do not project into a hyoplastral notch. In N. leithi (FMNH 224231), the shape is again similar to S. baba, but callosities are less developed. In contrast to the extant taxa mentioned above, N. leithi shows well-defined hyoplastral notches for the branches of the entoplastron.

The hyoplastron of S. baba has a moderately short pair of anterolateral processes and at least two short processes anteromedially, which appear to be even shorter than the lateral ones, but their surface is abraded (Figs 3, 5). The morphology of the hyoplastron differs from that found in A. cartilaginea (FMNH 11088, USNM 22522) and D. subplana (NMNH 222523, UCMVZ 95937), in which the anterolateral processes are much longer and the anteromedial processes are both much longer and more numerous. In P. sinensis (IVPP 556, USNM 68476), the hyoplastron is much more similar to that of S. baba. The anterolateral and anteromedial processes are still longer than in S. baba, but much shorter than in the aforementioned two taxa. In N. gangetica (NHMUK 86.8.26.1, USNM 293693), N. hurum (NHMUK 86.8.26.2) and N. leithi (FMNH 224231), the processes are more or less identical to those of S. baba. Possible small differences in the anteromedial processes can easily be explained by the abraded processes of S. baba.

A similar picture emerges in the hypoplastron. The processes in S. baba are short and similar in length to those in the hyoplastron. As in the hyoplastron, A. cartilaginea (FMNH 11088, USNM 22522) and D. subplana (USNM 222523, UCMVZ 95937) show a different morphology with much longer processes, whereas the morphology in N. gangetica (NHMUK 86.8.26.1, NHMUK 293693), N. hurum (NHMUK 86.8.26.2) and N. leithi (FMNH 224231) is almost identical to that found in S. baba. Pelodiscus sinensis (IVPP 556, USNM 68476) is again intermediate in this respect: its processes are shorter than in A. cartilaginea and D. subplana, but still longer than in Nilssonia spp. and S. baba.

The xiphiplastron of S. baba is relatively large, triangular and has short abraded processes (Figs 2, 5). As with the hyo- and hypoplastron the closest similarities are present within Nilssonia spp. and S. baba. In N. gangetica (NHMUK 86.8.26.1, USNM 293693) and N. hurum (NHMUK 86.8.26.2) the xiphiplastron looks almost identical to that found in S. baba. Small differences occur in the medial processes, which are slightly more pronounced and are not covered with callosities in S. baba. In N. leithi (FMNH 224231), the xiphiplastron is more massive than in S. baba and anteroposteriorly almost as long as the hyo- and hypoplastron combined, whereas in S. baba the xiphiplastron reaches only the anteroposterior length of the hypoplastron. In P. sinensis (IVPP 556, USNM 68476), the xiphiplastron is less triangular and more elongated. In addition, there is no direct medial contact between the left and right counterparts as is the case for Nilssonia spp. and probably also for S. baba. In A. cartilaginea (FMNH 11088, USNM 22522) and D. subplana (USNM 222523, UCMVZ 95937), the xiphiplastron is much slenderer and anterolaterally wider than in S. baba.

Discussion of the comparisons

Comparisons with the Palaeogene taxa from Asia underline a general issue concerning Pan-Trionychidae: despite many occurrences, the material in question often consists only of fragmentary shell remains, which are rarely diagnostic at the species level (for discussion of the issue, see also Georgalis & Joyce [2017]). Only eight Palaeogene Asian species were considered valid by Georgalis and Joyce (2017), but their preservation differs greatly. Whereas there are multiple complete individuals preserved for ‘Trionyx’ gregarius, including at least one juvenile, only a single almost complete carapace has been described for Drazinderetes tethyensis, ‘Trionyx’ johnsoni, ‘Trionyx’ ninae (including some additional fragmentary material) and Striatochelys impressa. The only specimen of Kuhnemys palaeocenica is a nearly complete shell of a juvenile individual. On the other hand, ‘Trionyx’ linchuensis only preserves an anterior carapace fragment, and for ‘Trionyx’ minusculus only a hyo- and hypoplastron are known. Pan-Trionychidae are, however, known for their high intraspecific variation (Meylan, 1987), making comparisons based on single individuals especially difficult.

Aside from the aforementioned Asian Pan-Trionychidae, we also provided a comparison with Plastomenidae due to it being the only group with prominent anteroposteriorly extending ridges as adults and a very similar plastron morphology; this is especially true for the Late Cretaceous genus Gilmoremys. However, plastomenids are so far known only from the Cretaceous and Palaeogene of North America (Joyce & Lyson, 2011; Joyce et al., 2018).

Overall, three features of the carapace and plastron are potentially important for a discussion of the close relationship between S. baba and S. impressa, as well as to untangle the potential affinity of Striatochelys within either Pan-Trionychinae or Plastomenidae.

1. Multiple anteroposteriorly projecting ridges on the carapace are rare in Pan-Trionychidae. They occur in some individuals of extant species (e.g. Apalone ferox, AMNH 57384, AMNH 65622; Trionyx triunguis (Forskål, 1775), AMNH 50723, AMNH 50724; Amyda cartilaginea, FMNH 11088, USNM 22522, 222521; Pelodiscus sinensis, IVPP 525, USNM 539334), but are much weaker, usually only present in juvenile/subadult individuals and only rarely present in adult specimens. In fossil taxa, however, they also sometimes occur in the shape of prominent ridges in adult specimens. Besides Striatochelys, they also appear in Plastomenus, e.g. Plastomenus vegetus (Gilmore, 1919), in Gilmoremys gettyspherensis and (in the form of a finer striation present only posteriorly) in ‘T.’ gregarius. All species that have prominent carapacial ridges as adults thus belong to Plastomenidae. Although such ridges are absent in adult pan-trionychines, the appearance of such ridges in juveniles of several (not particularly closely related) trionychine species, indicates that some species might have retained this morphology in later ontogenetic stages as well.

2. The difference in the number of neurals between S. baba and S. impressa is of low significance for separating these taxa. It is known that the number of neurals varies between individuals of a single species in Pan-Trionychidae (Meylan, 1987). Additionally, the number of individuals showing a complete row of neurals is too low in both taxa (three individuals in S. baba, one individual in S. impressa) to establish a clear difference in this regard. The absence of a preneural (first neural fused to the second neural in the character-taxon matrix of Meylan, 1987) is exclusively known for Pan-Trionychinae, with a reversal in Nilssonia gangetica and Nilssonia hurum and basal trionychids. The absence of a preneural in S. baba hence strongly supports a position for Striatochelys inside Pan-Trionychinae. Costals I and II are similarly shaped in S. baba and S. impressa, as in other Asian Pan-Trionychidae, except for D. tethyensis and ‘T.’ ninae. This stands in marked contrast to the condition present in Gilmoremys, in which costal II is strongly bowed anterolaterally, further pointing away from a close relationship between Striatochelys and Gilmoremys.

3. The plastron of S. baba is remarkably similar to that of Gilmoremys, whereas there are many differences to most species of Pan-Trionychinae. The hyo- and hypoplastron are robust and the lateral and medial processes are short. Moreover, a notch is present at the anterior margin of the holotype, as in Gilmoremys (Fig. 5). The triangular xiphiplastra might have been sutured to each other along the midline, based on their straight medial margin; however, only two xiphiplastra are preserved in total and the most complete one (Figs 5G, H) has an anteromedial process that possibly prevented such a suture. In the majority of Pan-Trionychinae, on the other hand, the plastron is much more reduced and the lateral and medial processes are longer. Exceptions to this are several species of Nilssonia spp. In those species, the overall plastron morphology is very similar to that of S. baba, including very short lateral and medial processes on the hyo- and hypoplastron and triangular xiphiplastral. The overall plastron morphology thus indicates a closer relationship of Striatochelys either with Plastomenidae or to extant east Asian trionychines.

In conclusion, the carapace of Striatochelys strongly indicates affinities to Pan-Trionychinae. The absence of a preneural is known only in members of this clade, as well as in basal trionychids, and characters indicating a close relationship with Plastomenidae (like the strongly developed ridges on the carapace) can also be found in some juvenile/subadult individuals of Pan-Trionychinae. Additionally, differences in the shapes of costals I and II further support a distinction from Plastomenidae. The plastron of S. baba is very similar to that of Nilssonia spp. and possibly indicates a closer relationship of the former to members of this genus. The very similarly-shaped plastron of the plastomenid Gilmoremys, on the other hand, probably represents a convergence.

Phylogenetic analysis

For the maximum parsimony analysis, a total of 135 most parsimonious trees with lengths of 327 steps, a consistency index (CI) of 0.361 and a retention index (RI) of 0.605 were recovered (Fig. 8).

Overall, the strict consensus tree is similar to that of most other recent analyses based on the same data set in recovering a monophyletic group, which consists of Plastomenidae and Trionychinae, as the sister group to Pan-Cyclanorbinae (Brinkman et al., 2017; Joyce & Lyson, 2017; Joyce et al., 2018). Alternatively, some analyses find Plastomenidae, Pan-Trionychinae and Pan-Cyclanorbinae in an unresolved polytomy (Joyce et al., 2016; Lyson et al., 2021), or Plastomenidae nested inside Pan-Cyclanorbinae (Evers et al., 2023; Joyce & Lyson, 2011).

In the phylogenetic analysis preformed here, Striatochelys baba is recovered within Pan-Trionychinae in a polytomy with Nilssonia gangetica, Nilssonia hurum and Nilssonia formosa. Pan-Trionychinae is supported by five synapomorphies, four of which are known exclusively for this group, i.e. characters 4(2), 20(2), 38(1) and 79(0) (if Early Cretaceous taxa are considered part of Pan-Trionychinae). A single additional character is found only as an autapomorphy for Pan-Trionychinae, if either Aspideretoides foveatus (Leidy, 1856) or Atoposemys superstes (Russell, 1930) is recovered as the basal-most taxon within Plastomenidae. For a complete list of synapomorphies for major groups and additional tree figures, see Supplemental material File S2.

Only a single autapomorphy supports Nilssonia spp. + S. baba in all trees of our phylogenetic analysis: character 20(1) ‘suprascapular fontanelles closed at hatching’, which is a reversal of one of the autapomorphies outlined above for Pan-Trionychinae, in which the fontanelles only close very late in ontogeny. Two other characters (22[2] and 50[1]) are considered autapomorphies for the group only if S. baba and N. gangetica form a monophyletic sister group with N. hurum + N. formosa or if N. gangetica is the most basal taxon of the group. The relationships between S. baba and Nilssonia spp. are unresolved. A basal position of S. baba is recovered in 33/135 trees, whereas in another 33/135 trees, S. baba is found as the sister taxon to N. gangetica. A basal position for N. gangetica, with S. baba being the sister taxon to N. hurum + N. formosa, is recovered in 34/135 trees. In the remaining 35 trees, S. baba is recovered in a derived position inside Nilssonia spp., either as sister taxon to N. formosa (14/135 trees) or as sister taxon to N. hurum (21/135 trees).

However, not only is the position of S. baba inside Nilssonia spp. ambiguous, but also its position within Pan-Trionychinae in general is not stable. The interpretation of the medial edge of the hyoplastron (character 89) is crucial for the position of S. baba on the tree. The medial edge of the hyoplastron is worn away, but three small projections can be interpreted as part of a serrated anteromedial process. If they are interpreted as such, or if a more conservative approach is taken and the character is scored as questionable (as is the case for our analysis), S. baba is recovered as part of Nilssonia spp. However, if the small projections are interpreted as too small for such a serrated edge, S. baba is placed in a polytomy with Gilmoremys gettyspherensis and Gilmoremys lancensis at the base of Plastomenidae (Supplemental material File S2, Fig. 2).

Another problem with the current phylogeny is in the positions of five Early Cretaceous taxa (Kuhnemys orlovi Khosatzky, 1976, ‘Aspideretes’ maortuensis Yeh, 1965, ‘Trionyx’ kyrgyzensis, Perochelys hengshanensis and Perochelys lamadongensis) recovered deeply nested inside Pan-Trionychinae. If those taxa are forced to the base of Pan-Trionychidae in accordance with their age, S. baba is found at the base of this Early Cretaceous group, which is in strong contrast to its middle–upper Eocene age (Supplemental material File S2, Fig. 3). However, if K. orlovi is allowed to float and thus recovered as a Pan-Trionychinae, S. baba is again nested within Nilssonia spp. (Supplemental material File S2, Fig. 4).

An additional phylogenetic analysis based on the matrix of Evers et al. (2023) yields a completely different result and instead places S. baba as the basal-most taxon of Pan-Cyclanorbinae (Supplemental material File S2, Fig. 5).

As highlighted above, S. baba can be recovered in several different positions on the tree based on only minor changes to the data set. There are two main reasons for this uncertainty. One is related to the problematic position of Early Cretaceous taxa within the crown of Pan-Trionychinae (see also Brinkmann et al., 2017; Evers et al. 2023; Joyce et al., 2021; Vitek et al., 2018) and a re-evaluation of the characters placing these taxa within the crown seems necessary. The second reason is the preservation of S. baba or, more specifically, the poor preservation of its skull, rendering cranial scorings impossible. In the current data set 39/95 characters refer to the skull, while in the data set of Evers et al. (2023) the proportion is 62/116. With more than half of the possible scorings missing and high intraspecific variation in Pan-Trionychinae (Meylan, 1987), well-justified placement of S. baba in the tree is challenging. Until the issues outlined above are properly addressed, the position of S. baba within Nilssonia spp., as suggested here, should be treated with caution.

Palaeobiogeographical implications

Although the results of the phylogenetic analyses should be treated with caution (see above), the position of Striatochelys baba in a polytomy with Nilssonia spp. is unsurprising given its age and its occurrence in Southeast Asia. According to Pereira et al. (2017), the Nilssonia clade originated during the middle–late Oligocene in either East or Southeast Asia, and the split from Palea steindachneri and Amyda cartilaginea occurred sometime during the late Eocene in Southeast Asia. Based on these data, it seems possible that S. baba is more basal than extant Nilssonia spp. However, based on other analyses (Evers et al., 2023; Thomson et al., 2021), Nilssonia spp. originated much later, in the middle Miocene. These results would indicate a more basal position for S. baba within the extant eastern Asian trionychines.

Unfortunately, the majority of taxa from the Paleogene of central and eastern Asia are only incompletely preserved, severely limiting direct comparisons. Consequently, they were also never included in a phylogenetic analysis. Interestingly, those taxa do not show any close resemblances to S. baba based on our comparisons (see above), the only exception being S. impressa from the Maoming locality of southern China. This taxon, as outlined above, appears to be very closely related to S. baba from the Na Duong locality of north-eastern Vietnam, most likely representing its sister taxon.

A high degree of faunal similarity between the two localities is also indicated by the pan-geoemydid turtles. From Maoming, two species are known: Guangdongnemys pingi Claude, Zhang, Li, Mo, Kuang and Tong, 2012 (minimum number of individuals [MNI] = 4) and Isometremys lacuna Chow and Yeh, 1962 (MNI = 6). From Na Duong, Banhxeochelys trani was recently described by Garbin et al. (2019), who concluded that B. trani was most similar to the species from Maoming. The high degree of faunal similarity between the two localities is further supported by the crocodylian fauna: the taxa known from Na Duong, Orientalosuchus naduongensis and Maomingosuchus acutirostris, are closely related to those from Maoming, Dongnanosuchus hsui Shan, Wu, Sato, Cheng and Rufolo, 2021 and Maomingosuchus petrolicus (Yeh, 1958), respectively (Massonne et al., 2019, 2021; Shan et al., 2021).

A third locality, which resembles the Na Duong fossil site in terms of its vertebrate assemblage, is the Krabi Basin of southern Thailand. As in Maoming, two pan-geoemydids have been described: Hardella siamensis Claude, Suteethorn and Tong, 2007 (MNI = 1) and Mauremys thanhinensis Claude, Suteethorn and Tong, 2007 (MNI = 11). Although Garbin et al. (2019) did not find a particularly close relationship between the Krabi taxa and B. trani, their phylogenetic analysis yielded a large polytomy including, among others, all five geoemydid taxa mentioned here. As for the crocodilian fauna, Krabisuchus siamogallicus Martin and Lauprasert, 2010 is closely related to O. naduongensis and D. hsui, while a specimen referred to as the ‘Krabi-Maomingosuchus’ is closely related to Maomingosuchus spp. from Na Duong and Maoming (Martin et al., 2019; Massonne et al., 2019, 2021; Shan et al., 2021). However, no pan-trionychid has so far been described from the Krabi Basin, potentially indicating the prevalence of different environmental conditions that were not suitable for pan-trionychids.

Interestingly, pan-trionychids are rare in the Na Duong Basin compared to other turtles: only nine individuals of S. baba were excavated, which stands in marked contrast to the more than 100 individuals known for B. trani (Garbin et al., 2019). In Maoming, the pan-geoemydids are, in absolute numbers, less frequent than in Na Duong (G. pingi, MNI = 4 and I. lacuna, MNI = 6), but still much more common than S. impressa with only a single known individual. Therefore, in both of these localities, pan-geoemydids are approximately 10 times more abundant than pan-trionychids. In Krabi, the number of pan-geoemydids is comparable to Maoming (H. siamensis, MNI = 11 and M. thanhinensis, MNI = 1). Based on these numbers (and the assumption that pan-trionychids are equally rare in Krabi as they are in Maoming and Na Duong), the absence of a pan-trionychid could also be explained by a sampling bias instead of unsuitable environmental conditions.

Conclusions

Striatochelys baba gen. et sp. nov. is a pan-trionychid from the middle–upper Eocene (late Bartonian–Priabonian, 39–35 Ma) of the Na Duong Basin of northern Vietnam. The taxon is known from extensive shell material of at least nine individuals. The best preserved of these specimens was selected as the holotype, consisting of a virtually complete carapace and plastron. In general, S. baba is a medium-sized species characterized by prominent ridges on the carapace (on both costals and neurals) in adult specimens, a larger costal VIII, forming the posterolateral margin of the carapace, and an entoplastron callosity in the shape of a bulge.

Comparison with plastomenids, other Palaeogene pan-trionychids from Asia and extant taxa from Southeast Asia, reveals a close resemblance of S. baba to the latter group, particularly to members of the genus Nilssonia. Some other characters seen in S. baba are otherwise unique to Plastomenidae. Our phylogenetic analysis supports the results of these comparisons and recovers S. baba in a polytomy with Nilssonia spp.

Striatochelys baba is morphologically very similar to a pan-trionychid from the upper Eocene of the Maoming Basin of southern China, which has previously been referred to as ‘Trionyx’ impressus. Accordingly, we assign ‘T.’ impressus to the new genus Striatochelys, as S. impressa. In the past, representatives of Pan-Geoemydidae among cryptodirans, as well as of Orientalosuchina and Tomistominae among crocodylians, have demonstrated a high degree of faunistic similarity between the Na Duong Basin in Vietnam and the Maoming Basin in China. The close resemblance between the two Southeast Asian pan-trionychids further supports this conclusion and provides additional evidence for the close similarity between the reptile faunas of the Na Duong Basin and Maoming Basin.

Associate Editor: Serjoscha Evers

Acknowledgements

We thank the Editor-in-Chief Paul M. Barrett and Associate Editor Serjoscha Evers, as well as the reviewers Walter G. Joyce and Georgios L. Georgalis, for their comments, which helped us improve the manuscript. We thank our Vietnamese colleagues who facilitated and participated in the Na Duong palaeontological expeditions of 2009, 2011 and 2012: Nguyễn Việt Hung, La Thễ Phúc, Đặng Ngọc Trần, Đồ Đức Quang, and Phan Đồng Pha. We further thank Gabriel Ferreira (SHEP), Márton Rabi and Erich Weber (both University of Tübingen) for stimulating discussions and Walter G. Joyce (University of Fribourg) for stimulating discussions and photographs of comparative material of fossil and extant species. Regina Ellenbracht and Henrik Stöhr (both University of Tübingen) are thanked for preparation and Gabriel Ferreira and Christina Kyriakouli (SHEP) for helping with a CT scan. Erich Weber and Ingmar Werneburg (both University of Tübingen) are thanked for granting access to specimens under their care. The Willi Hennig Society is thanked for providing access to the software TNT v. 1.5.

Supplemental material

Supplemental material for this article can be accessed here: https://doi.org/10.1080/14772019.2023.2217505