Anthracotherium, CUVIER, 1822

GENUS ANTHRACOTHERIUM CUVIER, 1822

The attribution of anciently described material of Anthracotherium tothenewgenus Paenanthracotherium led us to emend Anthracotherium as follows.

Emended diagnosis

Medium to large Anthracotheriinae s.l., with bunodont cheek teeth. Lower molars with fully developed postectoprotocristid (171), postprotofossid (181) and posthypofossid (351), a prehypocristid divided into two mesial arms (290), m3 with an entoconulid on its talonid (421), the lack of an accessory cristulid issued from the hypoconulid and mesially directed between the pre- and the posthypocristulid on m3 (461). Upper molars with parastyle smaller or equal to mesostyle (681) and with the presence of an endometacrista (601).

Differential diagnosis

Anthracotherium differs from all other Anthracotheriinae by the presence of an entoconulid on m3 (421). It differs from Heptacodon , Paenanthracotherium gen. nov. and Prominatherium by a smaller or equal parastyle compared with the mesostyle (681). It further differs from Heptacodon and Paenanthracotherium gen. nov. by a fully developed postectoprotocristid on the lower molars (171), the presence of a postprotofossid (181), a prehypocristid dividing into two mesial arms on the lower molars (290) and the presence of a posthypofossid on the lower molars (351). Anthracotherium differs from Paenanthracotherium gen. nov. in having no entostylid on p3 (60), the main arm of the prehypocristid connecting the trigonid distal wall (320), a mesiodistal development of the ribs of labial cusps on the upper molars of almost half of the molar length (480), an endometacristule or secondary cristids labial to the metaconule (561), and a well-developed cingular style on the distolingual cingulum of P3 (911). It differs from Heptacodon by a strong premetacristid on lower molars (150), the presence of an ectoentocristid (270), a premetacristule divided into two mesial arms (531), a postparacristule extending towards the transverse valley (591), a diastema between p1 and p2 (861), a postparafossa on P3 (950) and no postprotofossa on P4 (1060). Anthracotherium differs from Heptacodon and Anthracokeryx in having a distolingual cingulid on p4 forming a continuous wall (70), an inflated prehypocristid in the transverse valley of the lower molars (301), an incomplete ectohypocristulid on m3 (391; complete in Anthracokeryx and absent in Heptacodon ), the lack of an accessory cristulid mesial to the hypoconulid on m3 (461) and the lack of cristids on lower canine (800). It differs from Anthracokeryx and Prominatherium by a distostyle that is at the level of the metacone and not the metaconule on the upper molars (571), and further differs from Anthracokeryx by a postentocristid on m1 and m2 (281), the presence of a postectoprotocrista (491), a mesiolingual style (= protostyle) on the mesial cingulum of the upper molars (661), a fully developped metastyle (701), a postprotocrista on P4 that joins the distostyle (991) and a concave to straight mesial margin on P4 (1020).

Type species

Anthracotherium magnum Cuvier, 1822 from Cadibona ( Italy), early Late Oligocene ( MP25 ).

Other included species

Anthracotherium monsvialense De Zigno, 1888 ; A. bugtiense Pilgrim, 1907 ; A. kwablianicum ( Gabounia, 1964) ; A. chaimanei Ducrocq, 1999 .

Doubtful species

Anthracotherium alsaticum Cuvier, 1822 ; A. valdense Kowalevsky, 1873 ; A. seckbachense Kinkelin, 1884 ; A. illyricum Teller, 1886 ; A. bumbachense Stehlin, 1910 ; A. bimonsvialense-magnum Golpe-Posse, 1972 .

Distribution

In Asia, late Eocene of Thailand and Oligocene of Pakistan and in Europe, Oligocene (MP21-29) of France, Georgia, Germany, Italy, Spain and Switzerland (e.g. Kowalevsky, 1873; Teller, 1886; Stehlin, 1910; Gabounia, 1964; Ducrocq, 1999; Scherler, 2011; Scherler et al., 2013; see Supporting Information, Appendix S5 for a more exhaustive list of localities and references).

Remarks on the collection from Mouillac

The specimens from Mouillac ( France) differ only very slightly from the type species, A. magnum from Cadibona, in having a slightly larger size ( Fig. 8; Supporting Information, Appendix S6), a smaller hypoconid on p4, a smaller endohypocristid on the lower molars, and a posthypocristid reaching the postentocristid on m1 and m2 instead of directly joining the distostylid. As inferred from the present phylogeny and taking into account the weak variation, we conclude that the specimens from Mouillac should be assigned to A. magnum .

Remarks on the doubtful species

Anthracotherium alsaticum was described based on the fragmentary mandible of a juvenile individual. Although its assignment to Anthracotherium is clear, its attribution to a species different from A. magnum remains doubtful, because this species displays a large intraspecific variability concerning the size and tooth morphology (e.g. Squinabol, 1890).

Anthracotherium valdense was subject to many synonymies with A. magnum since its first description (e.g. De La Harpe, 1854; Rütimeyer, 1857; Stehlin, 1910; Scherler, 2011). It differs from the type species only by its well-developed distostylid, or distal cingulid, on m3. It ranges within the size variability of A. magnum ( Scherler, 2011) , and its assignment to a proper species is, indeed, questionable.

Regarding the species A.? illyricum and A.? bumbachense , their attribution to Anthracotherium cannot be confirmed owing to the scarcity of the discovered fossils and/or their poor preservation.

Regarding A. bimonsvialense-magnum , represented by only a fragmentary right maxilla with P3–M3 (holotype:IPS1738, collectionoftheMuseumofSabadell, Spain), the genus attribution is certain, but the specific assignment remains doubtful. Some characters confirm its attribution to Anthracotherium : the prominent distolingual style and the well-developed ecto- and postprotofossae on P3, the presence of enamel knobs on the labial side of the metaconule on upper molars, and the general quadratic shape of upper molars, with a parastyle slightly smaller than the mesostyle. However, the size differences proposed by Golpe-Posse (1972) are concordant with the important size variability within the genus and notably within A. monsvialense (Ghezzo & Giusberti, 2016; Fig. 8; Supporting Information, Appendix S6). Therefore, we suggest that it should be synonymized with A. monsvialense .

Remarks on the size variability of Anthracotherium Figure 8 shows the measurements of upper and lower molars and p4/P4 of each species of Anthracotherium mentioned here. The measurements of each studied species are in the Supporting Information (Appendix S6). Size may help to discriminate the species of Paenanthracotherium but cannot be used for Anthracotherium . Hence, the differentiation of a small specimen of A. magnum from a large A. monsvialense is not easy. Furthermore, these measurements confirm earlier assumptions (e.g. Squinabol, 1890) concerning the large variability of A. magnum .

PALAEOBIOGEOGRAPHICAL IMPLICATIONS

Figure 9 illustrates the distribution of anthracotheriines in Eurasia and North America from their origin in the Eocene until their extinction in the end of the Oligocene. Our phylogenetic study corroborates that anthracotheres originated in Asia and began their early diversification from the Bartonian. The Pondaung formation in Myanmar revealed a high diversity and abundance of Hippopotamoidea (see Tsubamoto et al., 2002; Soe, 2008) and could be dated to the middle Bartonian (40.2 ± 0.5 Mya) after U–Pb age analyses on zircons by Zaw et al. (2014). According to our phylogenetic results, representatives of the three main clades are present in this fossil formation (stem Hippopotamoidea , Microbunodontinae + Bothriodontinae and Anthracotheriinae with A. pangan , Anthracokeryx tenuis and Anthracokeryx birmanicus , respectively). From the Texan deposits, supposed to be close in age (early Duchesnean = middle Bartonian; Vandenberghe et al., 2012), Holroyd (2002) described a representative of Anthracotheriinae , Heptacodon yeguaensis , that demonstrates the first dispersal event of the family. The exact origin of anthracotheres might thus have taken place earlier in the beginning of the Bartonian or in the late Lutetian of Southeast Asia.

The diversity of anthracotheres is related to the dispersal events recorded in each subfamily (see Lihoreau et al., 2004 for Microbunodontinae; Lihoreau et al., 2015 for Bothriodontinae ). We observed the same phenomena within Anthracotheriinae , but our revised phylogeny and systematics of the subfamily suggest a more complex sequence of events than previously stated. Given that Prominatherium is known by few remains and its position in our phylogeny can be ambiguously related to Heptacodon , we suggest that it represents the first expansion from Asia to Europe during the Priabonian. This is confirmed by corresponding deposits from Croatia ( Hellmund & Heissig, 1994), Romania ( Patrulius, 1954) and Italy ( Grandi & Bona, 2017) ( Fig. 9). This first arrival in Southeastern Europe earlier than the Grande Coupure Event ( Stehlin, 1909) suggests a maritime path, because no terrestrial connection existed. Accordingly, it is interesting to note that all this material was found in coastal marine deposits, suggesting that this species was dispersing via an archipelago, as Southern Europe was at this time ( Grandi & Bona, 2017). This is in agreement with the alleged aquatic habits of some representatives of Hippopotamoidea (e.g. Lihoreau et al., 2015), but anthracotheres had probably not developed such habits enough to allow them to cross the Perialpine and Paratethys seas and settle in Western Europe. According to our results, Prominatherium is not related to subsequent European Anthracotheriinae , suggesting the extinction of this genus before the Oligocene. Hence, two new dispersal events can be recorded, with the arrival of Anthracotherium during the Grande Coupure Event (MP21) and the genus Paenanthracotherium with a possible slight delay (MP22 with Céreste and MP23 with Briatexte; Supporting Information, Appendix S5). Neither genus crossed the Oligocene–Miocene boundary, and they even seemed to disappear shortly before (at least in Europe, last occurrences of P. bergeri and A. magnum are dated to the European Mammal Level MP29; Scherler et al., 2013; Supporting Information, Appendix S5). Similar cases can be observed in other large mammal taxa. For instance, three genera ( Bakalovia , Amynodontopsis and Bachitherium ) seem to show an early dispersal of bothriodontines, rhinocerotoids and ruminants, respectively, towards the South Alpine and Carpathian Europe, whereas new representatives of the same taxa, such as Bothriodon , Elomeryx , Epiaceratherium , Ronzotherium and Bachitherium , hint at the new arrivals in Western Europe after the Grande Coupure Event ( Lihoreau & Ducrocq, 2007; Boehme et al., 2013; Mennecart et al., 2018; Tissier et al., 2018). It is worth noting that Elomeryx also occurs in the Late Eocene of Western Europe, with E. crispus from La Débruge (MP18, France) (BiochroM’97, 1997). This occurrence is still incompletely understood, because Elomeryx is considered paraphyletic and might be the stem group of several lineages of anthracotheres on different landmasses ( Lihoreau & Ducrocq, 2007). For a better understanding of its evolutionary history, a detailed review of this genus is needed, as stated by the same authors.

Finally, the geographical repartition of the large selenodont Paenanthracotherium and the robust bunodont Anthracotherium seems congruent with niche partitioning. Indeed, even if the presence of both genera might not be evident in all localities (evidence in the Bugti Hills, Pakistan and, possibly, in Le Garouillas, France), they are both present in the same regions ( France, Germany, Switzerland and Pakistan). This distribution may also represent a difference in habitat preference, potentially linked to aquatic environments ( Tütken & Absolon, 2015) or food preferences, i.e. plant consumption ( Sieber, 1936).

CONCLUSION

To review anthracotheriine systematics and understand the early phylogenetic divergence within the superfamily, it is crucial to describe the diversity of hippopotamoids through time and to describe the main dispersal events of large mammals during the Late Palaeogene. The phylogenetical hypothesis presented here differs from previous scenarios by proposing at least three dispersions of anthracotheriines from Asia towards North America and Europe. This new scenario is indicated by the presence of Anthracotherium in the late Eocene in Thailand ( Ducrocq, 1999). A similar distribution is observed for the genus Epiaceratherium ( Rhinocerotidae ), with the record of Epiaceratherium naduongense Boehme et al., 2013 from the Late Eocene in Northern Vietnam and the first European occurrence of this genus ( Epiaceratherium bolcense Abel, 1910 ) in the early Oligocene in Northern Italy ( Becker, 2009; Boehme et al., 2013). However, early Asian species of Paenanthracotherium need to be recovered to gain a better understanding of the palaeogeographical distribution of Anthracotheriinae . We suggest a systematic revision of Chinese anthracotheriines and profound analysis of the origin of Prominatherium and Heptacodon . Consequently, this paper is the first step in a complete review of the subfamily. Additionally, this study reveals the presence of two large anthracotheriines during the Oligocene in Europe and Pakistan; hence, it is crucial to study the palaeoecology of both taxa to understand the niche partitioning that occurred within the family and to compare it with that of the other contemporaneous large mammals (e.g. Rhinocerotidae and Entelodontidae ).