TESTUDINES Batsch, 1788

PLEURODIRA Cope, 1864

PELOMEDUSOIDES Cope, 1868

ARARIPEMYDIDAE Price, 1973

TAQUETOCHELYS Broin, 1980

T. decorata Broin, 1980

Fig. 1, Fig. 2 and Fig. 3

Synonymy. L. tenerensis Sereno and ElShafie, 2013.

Type series. The holotype, MNHN.F GDF 847, hypoplastron (Fig. 1P); and ten paratypes: GDF 838, first costal (Fig. 1F); GDF 839, costal (Fig. 2C and D); GDF 840, costal (Fig. 1L); GDF 841, costal (Fig. 2W–X); GDF 842, costal (Fig. 2AG–AH); GDF 843, costal (Fig. 3S and T); GDF 844, peripheral (Fig. 1D); GDF 845, peripheral (Fig. 1E); GDF 846, peripheral (Fig. 1A); GDF 848, xiphiplastron (Fig. 1R).

Other specimens here identified as attributable to this taxon. The peripherals MNHN.F GDF 853 (Fig. 1B), and GDF 854 (Fig. 1C). Several costals corresponding to the first pair: GDF 905 (Fig. 2A and B), GDF 858 (Fig. 2E and F), GDF 859 (Fig. 2 G and H), GDF 851 (Fig. 2I and J), GDF 906 (Fig. 2K and L). Several second to eighth costals: GDF 855 (Fig. 1K), GDF 857 (Fig. 1M), GDF 856 (Fig. 1J), GDF 860 (Fig. 3G and H), GDF 861 (Fig. 2O and P), GDF 862 (Fig. 2U and V), GDF 863 (Fig. 2Q and R), GDF 864 (Fig. 2Y and Z), GDF 865 (Fig. 3E and F), GDF 867 (Fig. 2AA and AB), GDF 868 (Fig. 2AE and AF), GDF 869 (Fig. 2AC and AD), GDF 871 (Fig. 3A and B), GDF 873 (Fig. 2M and N), GDF 877 (Fig. 3O and P), GDF 880 (Fig. 3C and D), GDF 882 (Fig. 1H), GDF 883 (Fig. 1G), GDF 884 (Fig. 3Q and R), GDF 885 (Fig. 3I and J), GDF 886 (Fig. 3M and N), GDF 887 (Fig. 3W and X), GDF 888 (Fig. 1I), GDF 889 (Fig. 2AK and AL), GDF 890 (Fig. 3AA and AB), GDF 891 (Fig. 3U and V), GDF 892 (Fig. 3AC and AD), GDF 893 (Fig. 3Y and Z), GDF 894 (Fig. 2S and T), GDF 895 (Fig. 3I and J), and GDF 1701 (Fig. 2AI and AJ). The hyoplastron GDF 850 (Fig. 1P). The almost complete skeleton MNN GAD28 (see figs. 14.4–14.18 in Sereno and ElShafie, 2013).

Locality and horizon. Late Aptian, Early Cretaceous. GAD 5 level, Elrhaz Formation, Gadoufaoua, Illumeden Basin, Ténéré Desert, central Niger (Broin, 1980, Lapparent de Broin, 2000 and Sereno and ElShafie, 2013).

As indicated (see introduction), the first defined araripemydid was a South American taxon: the Brasilian Aptian–Albian A. barretoi (Price, 1973). This form is known by numerous specimens, including several relatively complete skeletons (Gaffney et al., 2006, Meylan, 1996 and Price, 1973). The description of the Aptian T. decorata from a few disarticulated plates allowed us to recognize the presence of the clade Araripemydidae in Africa (Broin, 1980). The subsequent finding of an almost complete skeleton of an araripemydid in the type area where T. decorata was defined (i.e. the Aptian GAD 5 level of the Elrhaz Formation, in Gadoufaoua, central Niger) allowed me to confirm the presence of this clade on that continent (Sereno and ElShafie, 2013). However, this skeleton was attributed to a new form, ‘L. tenerensis’ (Sereno and ElShafie, 2013). Sereno and ElShafie (2013) justified the designation of a new taxon because the range of differences between the hypoplastra of ‘L. tenerensis’ and the holotype of T. decorata did not allow the attribution of both specimens to the same form. The review of previously published material from Gadoufaoua and the study of unpublished specimens from this locality, as well as the comparative analysis with other members of Pleurodira (including the only other araripemydid currently recognized, A. barretoi), allows me to recognize several shared characters among the holotypes of T. decorata and ‘L. tenerensis’ as exclusive within araripemydidae (e.g., the presence of mesoplastra and of an ornamental pattern composed of tubercles instead of pits). However, all the putative differences recognized by Sereno and ElShafie (2013) between both synchronic and sympatric forms are here identified as part of the intraspecific variability usually recognized in pleurodiran taxa known from several specimens (including, among others, A. barretoi).

Sereno and ElShafie (2013) reported four characters as different between T. decorata and ‘L. tenerensis’. The first one was the morphology of the lateral margin of the hypoplastra. They indicated that the sutural margin of the hypoplastra of ‘L. tenerensis’ with the fifth and sixth peripherals (probably referring to the sixth and seventh ones) is divided discretely into two parts, the suture with the most posterior of these peripherals angling posteromedially at approximately 30° (see fig. 14.14a in Sereno and ElShafie, 2013). They characterized the margin of the holotype of T. decorata as gently convex (see Fig. 1Q). Intraspecific variability is recognized in the araripemydid A. barretoi, specimens with an angulate margin as that present in the holotype of ‘L. tenerensis’ (e.g., Fig. 4D), but also others with a convex margin (e.g., Fig. 4E), being identified here. Intermediate states are also recognized for this taxon (e.g., Fig. 4G). In fact, intraspecific variability for this character is identified in numerous pleurodiran taxa. For example, specimens with well-developed angulate (e.g., Fig. 4O), slightly angulated (e.g., Fig. 4P) or rounded margins (e.g., Fig. 4N and Q), are recognized for the Cenomanian bothremydid Algorachelus peregrinus Pérez-García, 2016.

The second character recognized by Sereno and ElShafie (2013) as different between T. decorata and ‘L. tenerensis’ referred to the contact area of the hypoplastra with the mesoplastra. These authors indicated that the suture between these elements has a discrete angle of approximately 50° in ‘L. tenerensis’, giving the mesoplastra a distinctive pentagonal shape (see fig. 14.14a in Sereno and ElShafie, 2013). They characterized this margin as more irregular in T. decorata (see Fig. 1Q). As a consequence, the ramus of each hypoplastron between the mesoplastron and the posterior embayment was identified as proportionately shorter in ‘L. tenerensis’. The first-hand review of the holotype of T. decorata allows me to recognize that the anterior margin of the hypoplastron was broken, so the ratio between the length of the area in contact with the mesoplastron and that of the region located behind this plate is unknown (Fig. 1Q). Since A. barretoi lacks mesoplastra, the intraspecific variability for that character cannot be evaluated in that taxon. The difference in morphology between the mesoplastron-hypoplastron suture of the holotypes of T. decorata and ‘L. tenerensis’ is much less than that recognized in other species of Pleurodira, as is the case of the aforementioned Algorachelus peregrinus, which shows clearly concave margins in some specimens (e.g., Fig. 4P), being angulated in others (e.g., Fig. 4Q), intermediate states also being recognized (e.g., Fig. 4N and O). The ratio between the length of the area of contact of the hypoplastron with the mesoplastron and that of the region located behind it is recognized as markedly variable in Algorachelus peregrinus, the contact of the mesoplastron with this plate being much shorter in some specimens (e.g., Fig. 4N and Q) that in others (e.g., Fig. 4O and P).

The third difference between both taxa indicated by Sereno and ElShafie (2013) concerned the embayment for the hind limb. They recognized the outline as more deeply arched in ‘L. tenerensis’, with a smooth margin approximately twice the width of that in T. decorata. The study of the holotype of T. decorata allows me to recognize that the postero-lateral development of the embayment is not preserved, because this plate is broken (Fig. 1Q). In addition, the morphology of the embayment of this hypoplastron cannot be well defined, because a significant portion of its medial region is not preserved (see Fig. 1O). Variability in the contour of this structure is identified in other forms of Pleurodira, including the araripemydid A. barretoi (see the specimens of Araripemys in fig. 7 of Fielding et al. (2005), now identified as belonging to the same form sensu Gaffney et al., 2006; and Fig. 4A–C). Sereno and ElShafie (2013) indicated that a weakly developed or absent smooth margin is present in A. barretoi. This condition is recognized in some specimens (e.g., Fig. 4G), but this margin is identified here as well-developed in others (e.g., Fig. 4F), being similar to that of the holotype of ‘L. tenerensis’ (see fig. 14.14a in Sereno and ElShafie, 2013). In fact, the development of this margin is generally subject to intraspecific variability in Pleurodira. Thus, the development of a region equivalent to the area recognized as a smooth margin for the araripemydids, but also developed in a more medial region, is identified as highly variable in the Campanian dortokid Dortoka vasconica (Lapparent de Broin and Murelaga, 1996), being well-developed in some specimens (e.g., Fig. 4J and K), but poorly developed or absent in others (e.g., Fig. 4L and M).

The last character that allowed Sereno and ElShafie (2013) the definition of a new taxon is the identification of differences in the ornamental pattern of the holotype of T. decorata and in that of the hypoplastra of ‘L. tenerensis’. The ornamental pattern of both specimens, as well as that of the other specimens here attributed to T. decorata, is composed of low tubercles, which, in some regions, are grouped to form radiating ridges (Fig. 1S–X). In contrast, the ornamental pattern of A. barretoi is composed of pits (see the outer surface of a costal plate in Fig. 4H, and the detail of an external cast of another costal in Fig. 4I). Sereno and ElShafie (2013) noted differences in the size and prominence of the pattern of tubercles in the hypoplastra of the holotypes of T. decorata and ‘L. tenerensis’, and also in their organization into radiating ridges and sulci, and in its development in the posteromedial portion of these plates. As Broin (1980) indicated, variations in different regions of a single plate, in several zones of the shell of the same individual, and in the same region of the shell of several individuals are present (see Fig. 1, Fig. 2 and Fig. 3, particularly Fig. 1S–X). In fact, these variations can be markedly greater than those observed when those hypoplastra are compared. In addition, relevant variations in all aspects relative to the ornamental pattern considered by Sereno and ElShafie (2013) are here recognized for A. barretoi. Thus, the ornamental pattern is relatively well-developed in the posteromedial portion of the hypoplastra of some specimens (Fig. 4E–G), being absent in this area in others (Fig. 4D). The pits are smaller and more numerous in some specimens (Fig. 4E) than in others (Fig. 4D), a more anastomosed pattern being recognized in some individuals (Fig. 4G).

Therefore, none of the characters recognized by Sereno and ElShafie (2013) as different between the holotypes of T. decorata and ‘L. tenerensis’ allows me to support the identification of both putative synchronous and sympatric species as different forms. In fact, although the intraspecific variability of one species may differ from that of others, the comparative study of several specimens of the other currently recognized araripemydid (i.e. A. barretoi), as well as those of other members of Pleurodira, shows that all characters used by Sereno and ElShafie (2013) for the definition of the second African araripemydid are usually recognized as subject to intraspecific variability, both in closely related taxa and in others belonging to different lineages of Pleurodira. No other character can be established to support the attribution of the holotypes of T. decorata and ‘L. tenerensis’ to two different species. As indicated, the holotype of T. decorata shows characters different from those of the other known member of Arapemydidae, all of them being shared with the holotype of ‘L. tenerensis’ (i.e. the presence of mesoplastra, and an ornamental pattern composed of tubercles instead of pits). Considering all this, the priority of T. decorata is demonstrated, and ‘L. tenerensis’ is recognized as a junior synonym of this form. Consequently, the attribution of T. decorata to Araripemydidae, carried out by Broin (1980), but questioned by Gaffney et al. (2006) awaiting the finding of a more complete specimen, can be confirmed here. Thus, the hypothesis proposed by Broin (1980) regarding the African T. decorata as a form closely related to the synchronic South American A. barretoi is supported.

A description of the complete shell of T. decorata is available in Sereno and ElShafie (2013), based on the holotype of ‘L. tenerensis’. Some additional characters were provided by Broin (1980) from the study of isolated plates. In addition, Broin (1980) recognized intraspecific variability in a character of the carapace, which is supported here: the presence of a continuous neural series from the nuchal to the suprapygal in some specimens, this series being shorter in others, allowing a short to large medial contact of the last pair of costals. This range of variability is also recognized here for A. barretoi. Several shell characters of the differential diagnosis between the South American species and that from Gadoufaoua (based on the holotype of ‘L. tenerensis’, MNN GAD28, now attributed to T. decorata) proposed by Sereno and ElShafie (2013) are here refuted. Thus, these authors characterized the carapace length of the African form (i.e. approximately 15 cm), as smaller than that of A. barretoi (20–30 cm). Several plates of T. decorata corresponding to individuals of larger size than MNN GAD28, compatible with the size recognized for A. barretoi, are identified (e.g., Fig. 1 and Fig. 3). A deeply interdigitating suture between the hypoplastron and the xiphiplastron is not present in all specimens of A. barretoi. Thus, the deeply interdigitating area is limited to the lateral half in some specimens (e.g., Fig. 4B), but absent in others, an arrangement similar to that identified for T. decorata (see fig. 14.13 in Sereno and ElShafie, 2013) being recognized in these last specimens (e.g., Fig. 4A). No evidence of a different arrangement of the contact between the first dorsal vertebra and the nuchal plate can be established between both taxa, because that region is unknown in T. decorata.

Fragmentary remains attributable to araripemydids have been identified in Lower Cretaceous and basal Upper Cretaceous levels (from the Barremian, or possibly before, to the lower Cenomanian) of other African countries, including Morocco, Algeria, Tunisia, and Ethiopia (see Lapparent de Broin, 2000, and references therein). The poor information provided by these remains is insufficient to propose a precise systematic allocation. However, the presence of at least two different ornamental patterns is supported here, one corresponding to that of Taquetochelys and the other being compatible with Araripemys. In fact, the absence of mesoplastra in a hyoplastron with an ornamental pattern similar to that of Araripemys, from the Kem beds of Morocco (see plate XVII I-J in Gmira, 1995), supports the presence of an African form probably closely related to A. barretoi. The confirmation of the presence of Araripemydidae in Gadoufaoua was recognized by Sereno and ElShafie (2013) as clear evidence of the active faunal exchange between South America and Africa immediately prior to the Late Cretaceous. However, the identification of this clade in Africa at least from the Barremian, and in South America at least from the lower Aptian, demonstrates that this lineage was already present in both continental masses before the isolation (see Lapparent de Broin, 2000 and Scotese, 2014).

The recent finding of an almost complete turtle skeleton from the Aptian (Early Cretaceous) of Gadoufaoua (Ténéré Desert, central Niger) confirmed the presence of Araripemydidae in Africa. This clade of pleurodiran turtles had been originally described by a South American species, A. barretoi. The skeleton from Gadoufaoua was attributed to a new taxon, ‘L. tenerensis’. Although a potential araripemydid had been previously described at the same level and area (i.e. T. decorata), its attribution to this clade had been questioned by subsequent authors, and was proposed as a nomen dubium.

The previously recognized putative differences between the holotype of T. decorata and that of ‘L. tenerensis’, which allowed the diagnosis of the latter taxon, are here identified as forming part of the intraspecific variability of the same form. This variability is recognized as compatible with that present in A. barretoi by the personal observation of specimens of this South American taxon, as well as with that of other members of Pleurodira known from several specimens belonging to other lineages. In this way, the African T. decorata is identified as a senior synonym of ‘L. tenerensis’. Thus, T. decorata is the only species of Araripemydidae currently identified in Africa, the Brazilian A. barretoi being the only currently recognized representative of this clade defined outside this continent. The cranial and the postcranial anatomy of both members of this Aptian–Albian clade are well known, which is uncommon for the Cretaceous pleurodiran turtles.