Dakotaraptor steini DePalma et al., 2015 is a recently described dromaeosaurid dinosaur from the Upper Cretaceous (Maastrichtian) Hell Creek Formation of South Dakota. The holotype (PBMNH P.10.113.T) is given as an associated skeleton derived from a bonebed that purportedly contains the fossilized remains of other vertebrates including mammals, fish, amphibians, pterosaurs, reptiles, and birds (DePalma, 2010; DePalma et al., 2015). Included within the D. steini hypodigm are three elements that DePalma and colleagues (2015) identify as furculae: one which is part of the holotype specimen and two referred specimens—NCSM 13170 and KUVP 152429 (which was not figured, and which we have not observed directly). The furcula of PBMNH P.10.113.T was intermingled with the other elements assigned to the holotype, KUVP 152429 was found nine metres away from the holotype in the same bonebed, and NCSM 13170 was discovered as an isolated element sixteen miles from the holotype (DePalma et al., 2015). Here we demonstrate that the elements described as D. steini ‘furculae’ are not theropod dinosaur furculae, but are trionychid turtle entoplastra.
The furcula is a median, unpaired element present in extant birds and their non-avian theropod relatives (Nesbitt et al., 2009). Although the furcula is generally thought to have arisen through fusion of the clavicles, recent developmental studies suggest that the furcula is homologous with the interclavicle of early tetrapods (Vickaryous & Hall, 2010). DePalma et al. (2015) themselves noted several differences between the putative ‘furculae’ of Dakotaraptor steini and those of other non-avian theropod dinosaurs. We note that in PBMNH P.10.113.T and NCSM 13170, the ‘furcula’ is extremely craniocaudally compressed, and possesses flattened rami that bulge halfway along the length of the ramus, terminally asymmetrical ‘epicleidia’ with longitudinal striations, a medial juncture bearing a ventral tab (previously identified as the ‘hypocleidium’), and transversely straight, rather than caudally bowed rami. Taken together this suite of characteristics is unknown in other theropod furculae (Nesbitt et al., 2009), yet is consistent with the structure of the entoplastron in trionychid (soft-shelled) turtles.
The entoplastron is a median, unpaired element in the plastron, and, like the furcula of theropods, is a homolog of the interclavicle (Gilbert et al., 2001). In many turtle clades, the entoplastron is a roughly diamond-shaped element; however, in trionychids it takes on a flattened, slender, V-shaped to boomerang-shaped appearance, with lateral projections that diverge at roughly 90°(Hay, 1908; Vitek, 2012; Hutchison, 2013; Vitek & Joyce, 2015), reminiscent of the shape of non-avian theropod furculae. In their description of the associated fauna DePalma et al. (2015) note that multiple turtles, including trionychids (as Trionyx sp.) are preserved at the holotype locality (DePalma, 2010). Trionychids are common elements of Campanian-Maastrichtian North American ecosystems (Brinkman, 2003) and at least five species are represented in the Hell Creek Formation from which D. steini derives (Holroyd & Hutchison, 2002; Holroyd, Wilson & Hutchison, 2014; Vitek & Joyce, 2015). Moreover, several Campanian to modern trionychine trionychids (terminology following Hummel, 1928) have entoplastra that closely match the morphology of NCSM 13170 and the element figured as a ‘furcula’ in PBMNH P.10.113.T. Although KUVP 15249 was not examined by us or figured in the original description, DePalma et al. (2015: p. 6) considered it “virtually indistinguishable” from and “identical” to the holotype ‘furcula’, and therefore it is reasonable to assume that KUVP 15249 may also be a trionychid entoplastron.
Among extinct North American trionychines, the taxonomic identity of the entoplastral elements within the D. steini hypodigm can be refined on the basis of comparative morphology and relative size (Vitek, 2012). Here we follow the trionychid taxonomy of Danilov et al. (2014), but see Vitek & Joyce (2015) for a differing opinion. NCSM 13170 and PBMNH P.10.113.T exhibit an overall gracile morphology (narrow craniocaudally relative to the length of the lateral projections) as in Axestemys splendida (Gardner, Russell & Brinkman, 1995), other species of Axestemys (Vitek, 2012), Aspideretoides allani (Gardner, Russell & Brinkman, 1995), and Apalone and relatives (Vitek, 2011; Danilov et al., 2014).
Several discrete features of NCSM 13170 and PBMNH P.10.113.T are shared with select trionychid species. In NCSM 13170 the craniomedial margin of the rami junction is broad and cranially convex, bearing distinct lateral notches for contact with the epiplastra (Fig. 1). This differs from the condition seen in Axestemys montinsana (Vitek, 2012), yet closely matches the morphology seen in Axestemys splendida and other Late Cretaceous trionychids (Gardner, Russell & Brinkman, 1995). The distalmost one quarter of the ramus in PBMNH P.10.113.T (and in NCSM 13170, although the tip of the ramus is damaged) abruptly tapers asymetrically, representing the end of the contact between the entoplastron and epiplastron (Fig. 1). This morphology is identical to that seen in Axestemys splendida (Gardner, Russell & Brinkman, 1995; Fig. 1), Axestemys montinsana (Vitek, 2012) and possibly Gobiapalone breviplastra (Danilov et al., 2014).
The caudal margins of the rami in NCSM 13170 and PBMNH P.10.113.T bear a notch for the reception of the hyoplastron, which articulates with approximately two-thirds of the entoplastron ramus. The extent of this contact is similar in Axestemys splendida (Gardner, Russell & Brinkman, 1995; Fig. 1), Axestemys montinsana (Vitek, 2012), Axestemys cerevisia (Vitek, 2012), Aspideretoides allani (Gardner, Russell & Brinkman, 1995), and Apalone (Vitek, 2012), yet differs in Aspideretoides foveatus (Gardner, Russell & Brinkman, 1995), Oliveremys uintaensis (Vitek, 2011), Gobiapalone breviplastra, and Gobiapalone orlovi (Danilov et al., 2014). It is noted by Vitek (2012) that this contact in Axestemys is not as extensive as in Apalone and that Axestemys lacks a hyoplastral shoulder locking the entoplastron in place.
Finally, a distinctive longitudinal fluting along the distal third of each ramus for the attachment of connective tissue mars the rami in NCSM 13170 and PBMNH P.10.113.T. This is also present in Axestemys splendida (Campanian-Maastrichtian, Fig. 1), Axestemys montinsana (Paleocene; Vitek, 2012: Fig. 17), and Oliveremys uintaensis (Vitek, 2011).
The largest of the three trionychid entoplastra comprising the D. steini hypodigm (PBMNH P.10.113.T) pertains to a carapace approximately 60 cm in length based on comparisons with comparable materials (Figs. 1A–1D). This is consistent with the size range of Axestemys (Vitek, 2012), and of similar proportions to large trionychid shells known from the Hell Creek Formation (Hutchison & Archibald, 1986).
Taken together, the morphology and size of the PBMNH P.10.113.T “furcula” and NCSM 13170 indicate that they should not be referred to Dakotaraptor steini, and are instead most confidently identified as cf. Axestemys splendida. The holotype material of Axestemys splendida is Campanian in age, yet several specimens from the Late Maastrichtian have been referred to this taxon (Vitek, 2012; Vitek & Joyce, 2015) or are otherwise not identified to species (Holroyd, Wilson & Hutchison, 2014), therefore we refrain from referring these isolated elements beyond cf. Axestemys splendida.