Zaisanamynodon borisovi Belyaeva, 1971

Veine-Tonizzo, Léa, Tissier, Jérémy, Bukhsianidze, Maia, Vasilyan, Davit & Becker, Damien, 2023, Cranial morphology and phylogenetic relationships of Amynodontidae Scott & Osborn, 1883 (Perissodactyla, Rhinocerotoidea), Comptes Rendus Palevol 22 (8), pp. 109-142 : 113-118

Zaisanamynodon borisovi Belyaeva, 1971

( Figs 3; 4; Table 2 )

Zaisanamynodon borisovi Belyaeva, 1971: 43 .

TYPE MATERIAL. — Holotype. ANPIN 2761/1-22 , incomplete skull, lower jaw, and part of the postcranial skeleton, including most of the cervical vertebrae and forelimbs ( Belyaeva 1971).

REFERRED MATERIAL. — ZSN-KKS-28-IPB , fragmentary skull with P2-M 3 in connection with a fragmentary mandible with p4-m3 . MJSN.2020.008.01 and MJSN.2020.008.02 are 3D-printed copies of this skull and mandible, respectively .

TYPE LOCALITY AND HORIZON. — Kiin Kerish locality (48°07’52.2”N, 84°28’52.3”E), north-west of Kiin Kerish Mountain, Kazakhstan. Lower Aksyir Svita, Ergilian ALMA (36.5-33.9 Ma) (i.e., late Eocene) ( Borisov 1963; Belyaeva 1971; Russell & Zhai 1987; Lucas & Emry 1996; Lucas et al. 1996; Emry et al. 1998; Vandenberghe et al. 2012).

EMENDED DIAGNOSIS. — Differs from Zaisanamynodon protheroi in having a relatively short rostrum (about 12% of skull length), an anterior margin of the orbit above M1, incisors relatively large (especially I3/i3), a P2 less complex without anterior and posterior crests connected to the metaloph, a weak paracone fold on M3, lower canine relatively slender and curved, p3 more molariform, and a slight labial groove (cleft) between trigonid and talonid on lower molars.

DIAGNOSIS DISCUSSION. — In the revised diagnosis of Z. borisovi in Lucas (2006), the “lack of a complete cingulum” on P2-4 is considered as a character to distinguish Z. borisovi from Zaisanamynodon protheroi . However, in the original diagnosis of Zaisanamynodon borisovi, Belyaeva (1971) mentioned the presence of a well-developed anteroposterior and lingual collar (i.e., cingulum) and a weak labial collar (i.e., cingulum) on the upper premolars. In the description of Chinese specimens of Z. borisovi, Lucas et al. (1996) indicated the presence of a “complete lingual cingulum” on P2-3, and a “postero-lingual cingulum” on P4, which has been confirmed after direct observation of the specimen AMNH.26034. This feature is also observed on the studied material (ZSN-KKS-28-IPB). Therefore, the lingual cingulum on P2-4 can no longer be considered as a diagnostic character to distinguish Z. protheroi from Z. borisovi .

DESCRIPTION

Skull and mandible

The skull ZSN-KKS-28-IPB ( Fig. 3A, B) is incomplete, deformed and some parts are covered by matrix. The maxilla is incomplete, and only the left anterior part of the jugal is still preserved. The anterior margin of the orbit is above M1. The zygomatic process progressively diverges from the maxilla. It starts above the M2 and separates from it above the M3. The mandible ZSN-KKS-28-IPB ( Fig. 4A, B) is incomplete; the most anterior part is absent. The corpus mandibulae has a straight ventral profile. Only the anterior part of the ramus is present, and it is vertical. The dental formula of the specimen is I?/?, C?/?, P 3/2? M 3/3. The upper and lower anterior dentition are not preserved. The dental wear of the cheek teeth is moderate, the crown height is low, and the enamel is wrinkled and corrugated. The right P2 is very worn. The left P2 is not preserved and the left M3 is covered by matrix. There is no cement, crochet or antecrochet. The left p3 is not preserved but the roots of the right p3 can be observed in occlusal view.

Upper cheek teeth ( Fig. 3 C-G)

The upper premolars are not molarized and are very short compared to the molars (sensu Antoine 2002), with a LP3-4/ LM1-3 ratio ≤0.42. Premolars have a labial and lingual cingulum. The lingual cingulum is strong and continuous. The postfossette is wide. The P2 is very small. It bears a single labial cusp and probably had a single lingual one. The labial cingulum is continuous. On P3-4, the protocone and the hypocone are fused into a single lingual cusp. The protocone is not constricted. The metaloph is directed posterolingually. The paracone fold is very strong, forming a well-developed vertical ridge. On P3, the protoloph is not joined to the ectoloph, whereas they are joined on P4. The hypocone and the metacone of P4 are separated, whereas they are joined for P3. There is a crista on P3 and on the left P4. There is a shallow groove separating the paracone and the metacone on the ectoloph of P4. The metaloph of P4 is very short and weak. P4 has a third loph ( Fig. 3F), posterior to the metaloph and closing the postfossette. This third loph is thin, low, and discontinuous.

The upper molars have a π shape. They have a weak labial cingulum; a weak lingual cingulum is present in anterior and/ or posterior parts of M2 and posterior part of M3 whereas lingual cingulum is lacking on M1. Therefore, character 109 (1) and character 114 (2) in the matrix are scored as “usually present” and “usually absent”, respectively (sensu Antoine 2002). There are no crista and cristella. The protocone is widened. The metaloph is short, continuous, and its width decreases between M1-3. The ectoloph is very long, straight, and smooth. The fold of the paracone is weak, and the parastyle is reduced. There is a postfossette on each molar. The metastyle on M1-2 is long. The M1-2 have a continuous posterior cingulum. The M3 has a quadrangular shape. Its protoloph and metaloph are oblique to its ectoloph. Its metastyle is directed posterolingually. The ectoloph is concave labially.

Lower cheek teeth ( Fig. 4 C-H)

The p4-m3 series have a continuous labial cingulum. The lingual cingulum is only interrupted below the hypoconid. The trigonid is obtuse and is smaller than the talonid. The trigonid and talonid are open on the lingual side. The talonid is equal to the trigonid on the p4-m1, whereas the talonid is longer than the trigonid on the m2-3. The metaconid and entoconid are not constricted. The ectolophid is completely smooth (no labial groove), except on the p4, which has very shallow ectolophid and paralophid grooves. The lingual opening of the posterior valley on the p4 is U-shaped. The lower molar series are long. They bear an oblique hypolophid.

Body mass

The body mass of ZSN-KKS-28-IPB was estimated to be around 4.1 tons with the regression equations of perissodactyls and ungulates of Legendre (1989) based on occlusal surface of m1 ( Table 2 ; Appendix 3). The body mass was also estimated to be around 5.7 tons with the regression equations of Rhinocerotidae of Fortelius & Kappelman (1993), and 4.5 tons with the equations for all ungulates. They are based on the lengths of M2, M3 and the total upper tooth row ( Table 2 ; Appendix 3). In most cases, our estimations are higher than those of Averianov et al. (2016) for Zaisanamynodon borisovi (“2442 ± 257 kg ” see Averianov et al. (2016): supplemental table 2) with regression equation for generalized ungulates of Legendre (1989), except for the TRLU estimate (2.1 tons). This can be explained by the very large dimensions of the cheek teeth of the referred specimen as well as by the much bigger size of M2 and M3 (which are not used by Averianov et al. 2016), compared to the m1.

Furthermore, the body mass estimation obtained through the Rhinocerotidae equations of Fortelius & Kappelman (1993) based on M2 and M3 may also be extremely over-estimated due to the difference of morphology of the teeth of Amynodontidae compared with Rhinocerotidae . Indeed, in Amynodontidae , the M2 and M3 tend to be much larger than all other cheek teeth, which is not the case in Rhinocerotidae . In addition, Amynodontidae retain a metastyle on M3, leading to a quadrangular shape (and thus bigger length), whereas Rhinocerotidae have a triangular M3, without metastyle but an ectometaloph instead. Thus, these values should not be taken at face value but only be used for comparison if one uses the same method, since there are currently non that is specifically made for the rather peculiar amynodontid morphology.

Interestingly, the dimensions of the M2 and M3 of this specimen are comparable to those of the largest land-mammal that ever existed: Paraceratherium (= Indricotherium ). For example, the length of M2 and M3 of Paraceratherium bugtiense is around 74 to 96 mm ( Fortelius & Kappelman 1993), while our specimen measures 81 and 82 mm respectively. This explains the enormous body mass retrieved by our estimations. However, these estimations can be nuanced by the much shorter upper tooth row length (284 mm for ZSN-KKS-28- IPB against 400 mm for “ Indricotherium ” transouralicum ), which is explained by the typical premolar reduction of the Amynodontidae . Therefore, Zaisanamynodon borisovi may still have been one of the largest Amynodontidae , but not as large as Paraceratherium .

REMARKS

ZSN-KKS-28-IPB can be referred to Rhinocerotoidea based on the π shape of upper molars ( Wall 1989). An attribution of ZSN-KKS-28-IPB to Amynodontidae is reliable based on the reduced dental formula, with the absence of the P1/p1-2, which is a derived character typical of Amynodontidae ( Wall 1989; Tissier et al. 2018). ZSN-KKS-28-IPB shares other diagnostic characters of the Amynodontidae : the absence of the crochet; antecrochet; crista; the reduced parastyle on the upper molars; the quadratic M3 with a large metastyle and the elongated lower molars ( Wall 1989; Tissier et al. 2018). Therefore, an attribution to Rhinocerotidae , Eggysodontidae , “ Hyracodontidae ” or Paraceratheriidae can be excluded.

According to our phylogenetic analysis, ZSN-KKS-28-IPB shares two diagnostic characters with Cadurcodontini :a talonid on m3 longer than the trigonid and the constant presence of labial cingulum on upper premolars. It also shares the absence of labial groove separating the trigonid and talonid on lower molars ( Averianov et al. 2016).

Within Cadurcodontini , ZSN-KKS-28-IPB differs from Cadurcotherium in the absence of cement on cheek teeth, an oblique hypolophid on lower molars, and a long metastyle on M1-2. ZSN-KKS-28-IPB differs from Amynodontopsis in the presence of a postfossette on M1. ZSN-KKS-28-IPB differs from Cadurcodon in having a long metastyle on M1-2 and a weak paracone fold on M1-M2.

ZSN-KKS-28-IPB can be referred to Zaisanamynodon by the presence of a third posterior loph on P4, which is a diagnostic character of the genus ( Lucas 2006). Belyaeva (1971), Lucas et al. (1996) andLucas (2006) provided detailed descriptions, illustrations, and measurements of Zaisanamynodon borisovi and Zaisanamynodon protheroi . Following our emended diagnosis of Z. borisovi , ZSN-KKS-28-IPB shares a single diagnostic character of Z. borisovi : the anterior margin of the orbit above M1. However, ZSN-KKS-28-IPB differs from it by the absence of slight labial groove (cleft) between trigonid and talonid on lower molars, as in Z. protheroi . According to Lucas (2006), the complexity of the P2 and the presence of an “anterior and posterior crests connected to metaloph” are considered as differential diagnostic characters of the two species of Zaisanamynodon . This feature cannot be observed on ZSN-KKS-28-IPB because the right P2 is very worn and the left one is missing. For this reason, we cannot use this character to exclude an attribution to Z. protheroi ( Lucas 2006) . In addition, the incisors, canines and p3, which also bear diagnostic characters to differentiate Z. borisovi from Z. protheroi , are not preserved. However, the paracone fold on M3 is weak in ZSN-KKS-28-IPB as in Z. borisovi , and contrary to Z. protheroi ( Belyaeva 1971; Lucas et al. 1996; Lucas 2006), which is the only character that supports the assignment of this specimen to Z. borisovi in our phylogenetic analysis. Therefore, based on these anatomical comparisons of ZSN-KKS-28-IPB with Belyaeva (1971), Lucas et al. (1996) and Lucas (2006), ZSN-KKS-28-IPB can only be tentatively assigned to Zaisanamynodon borisovi .

Moreover, two species of Amynodontidae , Cadurcodon ardynensis (Osborn, 1923) and Zaisanamynodon borisovi are known from the Zaysan basin in north-west Kazakhstan ( Lucas & Emry 1996; Emry et al. 1998). Cadurcodon ardynensis is known from the Kusto Svita at Kiin Kerish and from the Buran Svita at Kalmakpay Mountain ( Lucas & Emry 1996; Emry et al. 1998). The presence of a third loph on P4 excludes an attribution of ZSN-KKS-28-IPB to Cadurcodon . The holotype of Zaisanamynodon borisovi (ANPIN 2761/1- 22; Belyaeva, 1971) and ZSN-KKS-28-IPB both come from the Lower Aksyir Svita at Kiin Kerish. Thus, this supports our identification of ZSN-KKS-28-IPB as Zaisanamynodon borisovi .