First pan-trionychid turtle (Testudines, Pan-Trionychidae) from the Palaeogene of Africa

Pan-trionychids, commonly known as soft-shelled turtles, are an important component of the extant chelonian faunas of Africa (Rhodin et al. 2017). They are characterized by a unique external morphology and cranial and shell anatomy (Vitek & Joyce 2015); most prominently, the characteristic sculpturing pattern on their carapace and/or plastron (Gardner & Russell 1994) has enabled their identification as fossils even on the basis of scarce and fragmentary available remains (Vitek & Joyce 2015; Georgalis & Joyce 2017). Pan-trionychids have a rich fossil record, spanning from the Cretaceous up to the Quaternary, and are distributed throughout large parts of Asia, Europe, Africa, Australia and the Americas (Vitek & Joyce 2015; Georgalis & Joyce 2017). Their Palaeogene record, particularly, is rather abundant, especially in Eurasia and North America, and consists of a diverse array of different taxa, while achieving a wide range of sizes and cranial and postcranial features (Vitek & Joyce 2015; Georgalis & Joyce 2017). In addition, certain Palaeogene forms are known from exceptional, complete skeletons, originating from a number of fossil Lagerstätte localities, thus allowing a comprehensive documentation of their anatomy, variability, ontogeny and functional morphology (Vitek & Joyce 2015; Georgalis & Joyce 2017).

Despite this Palaeogene abundance of the group in other continents, in Africa, the fossil record of pan-trionychids is confined exclusively to the Neogene and Quaternary (Georgalis & Joyce 2017). As such, it has been rationally speculated that Palaeogene pan-trionychids had not yet reached Afro-Arabia (Lapparent de Broin 2000; Joyce & Lyson 2010; Georgalis & Joyce 2017), an area that was isolated from other continents practically throughout the whole Palaeogene. Here, I describe a large shell fragment from the Eocene of Mali, which indicates the presence of pan-trionychids in the Palaeogene of Africa for the first time. The biogeographic origin of the Eocene African form is discussed.

Material and method

The single known specimen (NHMUK PV R 9824) was found during an expedition of the NHMUK and the Kingston Polytechnic Institute in the Samit area of the Gao Region, Mali, in 1981. The specimen is permanently curated at the collections of the NHMUK.

The taxonomy of fossil pan-trionychids follows Vitek & Joyce (2015) and Georgalis & Joyce (2017), and the nomenclature of the turtle clades follows Joyce et al. (2020a, 2020b, 2021).

Geological setting

The single specimen was found in the Samit Oolite in Tamaguélelt (also known in the literature under the spellings Tamaguelelt, Tamaguelalt and Tamaguilelt), 25 km north-northeast of InTasit, Gao Region, Mali (0^o15'E, 17^o36'N). The broader area consists of both Paleocene and Eocene localities (Tapanila et al. 2008; O’Leary et al. 2019; Van Couvering & Delson 2020). A precise age is difficult to estimate, although the oolitic limestone of Samit is considered as part of the Tamaguélelt Formation that spans the Ypresian (early Eocene) (O’Leary et al. 2019). The exact locality cannot be further correlated with the precisely stratigraphically dated sites identified by O’Leary et al. (2019). Nevertheless, it is further noted that fish remains collected from the same oolitic deposits in Tamaguélelt during the same 1981 expedition have been described as no ‘younger than middle or early Eocene’ (Longbottom 1984, p. 3); these, shallow marine–brackish water-living, fish were subsequently treated as ‘possibly early Eocene’ by O’Leary et al. (2019). In addition, during the same 1981 expedition that recovered the pan-trionychid specimen, the same locality also yielded palaeophiid material; the range of this aquatic snake lineage extends up to the start of the late Eocene (see Smith & Georgalis in press), but in the area of Mali it is known only by the giant species Palaeophis colossaeus Rage, 1983, published remains of which have been recovered exclusively from shallow marine to brackish water deposits of early–middle Eocene (Ypresian–Lutetian) age (Rage 1983; McCartney et al. 2018; O’Leary et al. 2019). Notably, this palaeophiid material found in the same 1981 expedition also bears a collection number close to that of the pan-trionychid specimen (NHMUK PV R 9848); my first-hand observation of it confirms that it indeed corresponds to P. colossaeus. Following this somewhat circular rationale, the herein described pan-trionychid find is placed between the early and middle Eocene. For detailed information on the Tamaguélelt Formation, see Tapanila et al. (2008) and O’Leary et al. (2019).

Systematic palaeontology

TESTUDINES Batsch, 1788 (sensu Joyce et al. 2020a)

CRYPTODIRA Cope, 1868 (sensu Joyce et al. 2020b)

PAN-TRIONYCHIDAE Joyce et al., 2004 (sensu Joyce et al. 2021)

PAN-TRIONYCHIDAE indet.

Figure 1

Material

A costal (NHMUK PV R 9824).

Description

NHMUK PV R 9824 represents an incomplete costal, with both its lateral edges being eroded (Fig. 1). The specimen is rather large and wider than long, with a maximum width of 145 mm and a maximum length of 115 mm. Judging from its shape and size, it is most probable that it represents a right costal II or III. The costal is expanded distally. The sculpturing pattern on the dorsal surface of the costal is irregularly netted, and consists of several connected ridges forming a honeycomb shape, as well as certain more disconnected, more prominent, ridges (Fig. 1A). The sculpturing pattern is denser in and around the middle portion of the specimen and slightly fades out towards both its lateral edges (Fig. 1A). On its ventral side, a large and relatively wide costal rib crosses the costal longitudinally (Fig. 1B). The specimen is exceptionally thick, exceeding at certain points 20 mm in thickness (Fig. 1C).

Remarks

NHMUK PV R 9824 can be assigned to Pan-Trionychidae by the presence of sculpturing and the absence of shell scutes (Vitek & Joyce 2015; Georgalis & Joyce 2017). The specimen cannot provide a more precise determination within pan-trionychids and therefore no pan-trionychine or pan-cyclanorbine affinities can be realized. The sculpturing pattern is of no taxonomic utility, given that this feature is rather variable within pan-trionychids (Gardner & Russell 1994; Vitek & Joyce 2015; Georgalis & Joyce 2017). The strong thickness of the carapace is reminiscent of plastomenids, a Cretaceous–Palaeogene North American lineage that may lie within pan-cyclanorbines (Joyce & Lyson 2010), although such topology was not recovered in more recent analyses (e.g. Joyce et al. 2018). In any case, the thickness of the carapace alone cannot be diagnostic for plastomenid affinities because it appears to be more widespread within pan-trionychids than was previously thought (Georgalis & Joyce 2017). A prominent example of this case involves the certain large forms from the Eocene of Asia that were once thought to be plastomenids based on the strong carapacial thickness (e.g. Chkhikvadze 1984), but such affinities have recently been discarded (Georgalis & Joyce 2017).

Discussion

No pan-trionychids had so far been found in the Palaeogene of the Afro-Arabian plate (Georgalis & Joyce 2017). A purported occurrence of a Palaeogene African trionychid had been previously questionably made on the basis of a single carapace fragment from the Eocene of Fayum, Egypt, by Wood (1979), who, nevertheless, highlighted that the Eocene age was far from certain, given that this specimen was dug up by archaeologists and could well represent much younger, subfossil material; the specimen has since been reinterpreted as likely to be of Holocene age (Georgalis & Joyce 2017). I further concur with this interpretation and also highlight that shells of the extant species Trionyx triunguis (Forskål, 1775) (the Nile soft-shelled turtle) have been recovered from ancient Egyptian tombs during archaeological excavations (e.g. de Treville 1975).

Pan-trionychids, therefore, are first recorded in the Afro-Arabian fossil record only during the early Neogene. Of the two major inclusive clades of pan-trionychids, pan-cyclanorbines represent the oldest identified occurrences, and are recorded in the fossil record of the area since the early Miocene (Burdigalian), while pan-trionychines appear several million years later, during the late Miocene (Tortonian) (Georgalis & Joyce 2017; Georgalis et al. 2020b).

The so far total absence of pan-trionychids from the Palaeogene of the continent could be explained by the fact that, at that time, the Afro-Arabian plate was isolated from the Eurasian landmasses (Rögl 1999; Gheerbrandt & Rage 2006; Rage & Gheerbrant 2020). This long isolation, which had already started during the Cretaceous (c. 90 Ma), was terminated during the early Miocene (Burdigalian) with the collision of the Afro-Arabian plate with Eurasia (Rögl 1999; Rage & Gheerbrant 2020). That collision provided a direct land connection with Eurasia, the so-called ‘Gomphotherium Landbridge’ (Rögl 1999), and, in turn, enabled dispersal events of multiple terrestrial lineages of vertebrates into Africa and Eurasia (Koufos et al. 2003; Georgalis et al. 2016, 2019, 2020a). It so remains entirely plausible that pan-cyclanorbines, which suddenly appear in the African fossil record at that time (Burdigalian), are among the lineages that dispersed from the northern landmasses of Eurasia, a view already suggested by Joyce & Lyson (2010). However, the extremely patchy and limited fossil record of pan-cyclanorbines in Asia, along with their total absence from Europe, cannot enable any definite conclusions to be made (Joyce & Lyson 2010; Danilov et al. 2014; Georgalis & Joyce 2017). As for pan-trionychines, their earliest occurrences in the Afro-Arabian plate are marked during the late Miocene, based on material of Trionyx from the Tortonian of Calabria, southern Italy, an area that was part of (or at least close to) Africa at that time (Georgalis et al. 2020b), and another coeval pan-trionychine from the United Arab Emirates (Lapparent de Broin & van Dijk 1999).

The new Eocene pan-trionychid from Mali represents the product of a Palaeogene random and sporadic dispersal event from some other landmass to the then isolated Afro-Arabia. Certain dispersal events have been suggested to occur during the Palaeogene between Africa and Eurasia, facilitated by occasional drops in the sea level of the Tethys Sea that separated these two landmasses (Gheerbrant & Rage 2006; Rage & Gheerbrant 2020). These dispersal events and faunal interchanges seem to have taken place earlier (Paleocene and early Eocene) in the area between southwestern Europe and northwestern Africa (via the occasional emergence of terrestrial platforms in that area) and then, subsequently, during around the middle Eocene via an eastern route, in the area between southwestern Asia and northeastern Africa (Gheerbrant & Rage 2006; Rage & Gheerbrant 2020). The fossil record indeed implies a series of such dispersal events for a wide range of non-marine animals, such as several groups of mammals (Gheerbrant 1990; Solé et al. 2015; Erdal et al. 2016; Borths & Stevens 2019), non-volant birds (Angst et al. 2013; Buffetaut & Angst 2014), and reptiles (Rage 1988; Augé & Brizuela 2020), including turtles as well (Pérez-García et al. 2017), while it has also been suggested for extant reptile lineages based on molecular data (Hipsley et al. 2009; Kelly et al. 2009). The marine dispersal capabilities of pan-trionychids are well-known in the literature (Georgalis & Joyce 2017); this is also in concordance with the identification of fossil forms from marine marginal deposits (e.g. Sánchez-Villagra et al. 2004; Head et al. 2006) and the current documentation of certain extant taxa that show salt water tolerance and which venture into marine environments, far away from the closest coast, and have colonized distant islands (e.g. Taskavak et al. 1999; Fritz et al. 2014). As such, several fossil pan-trionychid taxa have been inferred to have crossed long distances, such as the giant North American Cretaceous–Palaeogene form Axestemys Hay, 1899, which had dispersed also to Europe, probably around the Paleocene–Eocene Thermal Maximum, facilitated by the drop of sealevels and the occasional emergence of terrestrial corridors (Georgalis & Joyce 2017; Pérez-García et al. in press).

Recent geological and palaeontological evidence supports the idea that the area of what is now Mali was covered during the Late Cretaceous to early Eocene by an epicontinental sea, known as the Trans-Saharan Seaway, which bisected the Saharan part of West Africa (O’Leary et al. 2019). The Trans-Saharan Seaway was warm and rather shallow, covering an area of as much as 3000 km² and achieving a depth of only around 50 m (O’Leary et al. 2019). Such an area of shallow water would probably not represent a severe biogeographic barrier to the dispersal of pan-trionychids from the northern parts of the Tethys Sea to its southern counterparts. In particular, on this immediate side of the northern part of the Tethys, was located Central and Western Europe, where, during the Palaeogene, pan-trionychids thrived, achieving a high diversity, with certain taxa also reaching considerably large sizes (Georgalis & Joyce 2017). Accordingly, it could be the case that the new Eocene Malian pan-trionychid was a descendant of one of these European forms that ventured further southwards in the Trans-Saharan Seaway. In favour of such a scenario, other shared congeneric aquatic faunal elements are also present in both Western Europe and the Malian area, such as the large aquatic snake Palaeophis Owen, 1841 (see Smith & Georgalis in press).

An alternative scenario would imply a dispersal from Asia to Africa via the eastern route, the so-called Iranian route, which has also been suggested for a number of Palaeogene vertebrate lineages (Gheerbrant & Rage 2006). However, this dispersal event has been suggested to postdate the age of the Malian pan-trionychid, given that it is thought to have taken place around or after the middle Eocene, coinciding with a progressive closure of the Tethys in the area of the modern Middle East (Gheerbrant & Rage 2006). Nevertheless, Palaeogene faunal connections and interchanges between Asia (and/or the Indian subcontinent) and northern Africa have been envisaged on the basis of terrestrial mammals (Hartenberger et al. 2001; Marivaux et al. 2005; Seiffert 2012; Rana et al. 2015), certain of which could even pre-date a middle Eocene age. In addition, marine-facilitated dispersals of vertebrates between Asia and Africa during the Palaeogene have been suggested for certain reptile groups, such as varanid lizards (Smith et al. 2008; Holmes et al. 2010) and testudinid tortoises (Holroyd & Parham 2003), as well as for colubriform snakes, between northern Africa and the (then isolated) early Palaeogene Indian subcontinent (Zaher et al. in press). The presence of abundant pan-trionychid remains across the Palaeogene of Asia, including large-sized forms (Georgalis & Joyce 2017), along with the abovementioned marine dispersal capabilities of the group, would provide some support for the new African fossil to be instead of Asian origin, but it still looks less likely in comparison with the case of a dispersal from Europe discussed above.

It is also worth mentioning that, during the early Eocene, the Australian herpetofauna included a highly bizarre pan-trionychid, Murgonemys White, 2001 (White 2001; Georgalis & Joyce 2017). Late Mesozoic and early Palaeogene terrestrial connections between Australia, Antarctica and South America have been documented (e.g. Scanlon 1993), and accordingly a marine dispersal from Antarctica to southern Africa could have taken place. However, in the total absence of pan-trionychid remains from Antarctica and the Palaeogene of South America (Vitek & Joyce 2015; Georgalis & Joyce 2017), I choose to discard any such scenario, given that this would directly imply significant ghost ranges in multiple areas, certain of which are relatively well-sampled and explored (e.g. South America).

Finally, a direct marine dispersal from North America to Africa via some kind of oceanic rafting cannot be excluded, given that it has recently been proposed for amphisbaenian squamates (Longrich et al. 2015). In particular, the dispersal over such long oceanic distances by fully terrestrial, burrowing animals such as amphisbaenians was proposed to occur by transport on rafts of trees that float on the water (Longrich et al. 2015). North America during the early Palaeogene boasted a highly diversified array of pan-trionychids (Vitek & Joyce 2015), including also the lineage of plastomenids, which has been inferred to represent stem cyclanorbines (Joyce & Lyson 2010). Therefore, it can only be speculated that if certain squamate lineages found their way from North America to Africa via oceanic rafting, it could be relatively easier for a large-sized pan-trionychid to do the same. Of course, however, it cannot be stated that an amphisbaenian or any other non-aquatic squamate would represent the ideal ecological analogue for a pan-trionychid.

Whatever the case of its exact origin, the new pan-trionychid from the Eocene of Mali described herein indicates that this group was present in the Palaeogene of Africa. This discovery of a single costal in the Palaeogene of Africa is somehow reminiscent of another unexpected single costal, described from the Campanian of Sweden, which documented the presence of pan-trionychids in the Cretaceous of Europe, where they previously lacked any Mesozoic record (Scheyer et al. 2012). The new Malian pan-trionychid further adds an additional lineage to the Palaeogene chelonian fossil record of the continent, besides pleurodires, chelonioids and testudinoids (Lapparent de Broin 2000; Georgalis et al. 2021). It seems apparent that pan-trionychids were only a short-lived group that did not survive long in the ‘splendid isolation’ (sensu Simpson 1980) of the Palaeogene Afro-Arabia. More fossil specimens are certainly needed, along with a more intense sampling at African localities of that age, in order to quantify such an assumption.

Conclusion

Here I describe the first material of pan-trionychid turtles from the early to middle Eocene of Mali. It represents the sole documented occurrence of pan-trionychids in the Palaeogene of Afro-Arabia, a period in which this landmass was fully isolated from other continents. The biogeographic origins of the Eocene pan-trionychid from Mali and potential dispersal events between Afro-Arabia and other landmasses during the early Palaeogene are discussed.