Two small species of Cremohipparion (Equidae, Mammalia) from Samos, Greece Author Bernor, R. L., Author Tobien, H. text Bayerische Staatssammlung für Paläontologie und historische Geologie 1989 29 206 226 journal article 10.5281/zenodo.1436534 ae518b88-ae03-4875-9174-619da4dc29dc 1436534 Cremohipparion matthewi (A bel , 1926 ) Fig. 1 Hipparion minus, K ormos, 1911 Hipparion matthewi, A bel , 1926 , Fig. 273, pg. 430; pg. 432 Hipparion matthewi, A bel , 1927 , Fig. 110, pg. 120 Hipparion matthewi, SoNDAAR, 1971 , Pl. IV, Fig. a “ Hipparion ” aff.? matthewi, B ernor 1985 , Fig. 18, pg. 220 Ho1otype : Male skull with associated lower jaw, UGR OK/557 originally figured by K or ­ mos (1911, Fig. I and II), later by A bel ( 1926 : 430, Fig. 273, 1927 : 120, Fig. 110) and S ondaar ( 1971 , Pl. IV, fig. a). However, none of these published figures clearly detail the facial and cheek tooth morphology of the type specimen. Type Locality: Island of Samos Greece (no specific locality or horizon given). Age: Turolian, ca. 8.5—5.7 Ma (W eidmann et al. 1984 :487), S en ( 1986 :159) and S en and V a ­ let ( 1986 : 173). Geographic Distribution: Subparatethyan Province (sensu B ernor, 1983, 1984; Greece, Iran, Western U .S.S.R.). Subgeneric characterization (as given by Qiu et al., 1987 : 238 for their nomen Hipparion [ Cremohipparion ]): “This subgenus is very distinctive in the snout and nasal structure. The lateral rims ofthe nasals turn down first, and then incurve into the nasal cavity. The preorbitalfossa is very deep, close to the orbit. The subnasal fossa is also present. The nasal notch is retracted though less than in the subgenus Proboscidipparion. The Samos H. proboscideum seems very close to this subgenus in its facial structure. Unfortunately, no adequate description is available to show whether the nasal bones have incurved lateral rims or not. Under H. garezicum , M el - daze (1967) described some well preserved skulls from Arkneti, Caucasia . The Arkneti skulls are markedly different from the type of the species described by G abunia in 1959 . In our opinion the Arkneti sample may represent a member of the subgenus Cremohipparion in Caucasia .”, Generic Diagnosis (modified from B ernor et al., in press; characters in bold, underlined print are uniquely shared-derived characters of the group) — Hipparionine horses ranging from small to large size with length of tooth row 105—170 mm; lacrimal foramen lacking, although it may be represented by a large depression: preorbital bar short, with lacrimal usually touching or invading posterior limit of POF; when present, POF subtriangular shaped and mostly anteroventrally oriented; posterior pocketing slight to absent, medial depth great to slight, medial internal pits occur only in the most derived species , C. licenti; peripheral outline strong to weakly defined, anterior rim distinct to absent; infraorbital foramen placed inferior to and encroaching upon the anteroventral border of POF; buccinator fossa distinct and unpocketed except in C. licenti; caninus fossa present in some , but not all species; malar fossa lacking except in C. licenti; nasal notch tends to become retracted in this lineage and may or may-not curve inward in species included within the group; no persistent and functional P1; in adult, middle stage-of-wear individuals maxillary cheek teeth moderately curved to straight, with maximum crown height of 40—60 mm, fossette ornamentation complex to simple, posterior wall of postfossette always distinct; pli caballins may be double or single; hypoglyphs dee­ ply to shallowly incised; protocones may show some lingual flattening, but tend to become rounded, are clearly always isolated from protoloph, usually having no noticeable protoconal spur, and are lingually placed relative to the hypocone; P2 anterostyle/paraconid usually elongate, but become shortened in some species; in adult middle stage-of-wear individuals mandibular incisors not grooved, are curved, and I3’s not atrophied; cheek tooth metaconids/metastylids are generally rounded, not angled; ectoflexids do not separate metaconids and metastylids on the premolars, but do so on the molars; pli caballinids generally absent; protostylids may be present and frequent; ectostylids absent; linguaflexids V-shaped to shallow U-shaped; metapodials, when known are elongate and slender. Figure 1: Cremohipparion matthewi , UGR OK/557. A. Skull and lower jaw, left lateral view ; B. Left ma xillary cheek tooth series ; C. Left mandibular cheek tooth series. Emended Species Diagnosis: A small species of Cremohipparion (P2—M3 dimension about 110—120 mm) with preorbital bar and lacrimal as for genus; POF is reduced in medial depth and peripheral outline, no anterior rim, but retains an approximately anteroventral orientation; lacks a caninus fossa; nasal notch retracted only to anterostyle of P2; fossette ornamentation has moderate complexity; pli caballins variably double or single; hypoglyphs moderately deeply incised; P2 anterostyle/paraconid shortened; mandibular incisor and cheek teeth are as for genus. Metapodials are elongate and slender. Description: the type specimen of Cremohipparion matthewi is an old adult male individual, including a nearly complete skull and mandible ( Fig. 1 ). Since its original collection, the skull has deteriorated badly, having lost both posterior orbits, zygomatic arches, incisors and canine teeth. The snout has been broken and remodelled. Likewise, the mandible has been partially destroyed since collection, having lost both rami, left I3 and C. The skull is small for this genus. It preserves a roughly subtriangular, anteroventrally oriented POF, and its anteriormost limit is abruptly interrupted by a large infraorbital foramen. The POF is moderately deep, it is not pocketed posteriorly, but has a large, triangular-shaped lacrimal bone which distinctly invades the posterior 17, 5 mm of the POF. The POF lacks any posterior pocketing, but the preorbital bar remains distinct, and is short (= 21.2 mm). The peripheral border of the POF is moderately well delineated, its strongest border being the ventral one. There is a strong, dorsoventrally oriented ridge at the posterior extent of the POF which is touched posteriorly by the anteriormost lacrimal. There is no caninus fossa (= intermediate fossa of W oodburne and B ernor, 1980 : 1329 and subsequent publications) between the POF and the buccinator fossae. The nasal notch is retracted only to the anterior border of P2. There are a pair of dorsal premaxillary tubercles that are large and very prominent. The I2,s and I3’s have large, mediolaterally elongate infundibula. The canine roots are large and suggest that this individual was a male. The cheek teeth are in an advanced stage-of-wear, but still preserve some important occlusal features. P2 anterostyle is shortened (synapomorphy with “H”. aff. moldavicum; Middle Maragheh, Iran, ca. 8 Ma). Protocones are rounded on the premolars, and slightly more elongate on M1-3; P2 and M1 protocones are connected with the protoloph due to very late wear-stage. When preserved, the pli caballins are short, but distinctly double. The pre- and postfossettes are heavily worn, but still consistently show complex, very finely banded plication on both anterior and posterior surfaces. The hypoglyphs are obliterated on P2“3 and M1-2; on P4 only a shallow hypoglyph is preserved, while M3’s hypoglyphs have moderate depth. The mandible is likewise small and has a highly worn incisor and cheek tooth dentition. I]_2 have highly worn, rounded infundibula; I3’s infundibulum is more mediolaterally elongate. The right canine is preserved and maintains a stout morphology with a strong central lingual pillar, indicative of a male individual. All cheek teeth are heavily worn. The metaconids and metastylids are generally rounded with slightly angular facing borders. Fully encircled protostylids are absent, except for a small one on the left M3. The pre- and postflexids are heavily worn, but generally preserve a short and buccolingually narrow morphology. The linguaflexid is obliterated on P2, distinctly V-shaped on P3_4, and becomes progressively more U-shaped serially from M(_3. The ectoflexid does not separate metaconid/metastylid on the premolars, while it does separate them on the molars. Remarks: Cremobipparion matthewi constitutes perhaps one of the most complex taxonomic muddles of Old World hipparionines. This can be attributed to many past workers using principally size and presumed limb slenderness to distinguish this species . As we discuss below, there are a range of skull, mandibular and postcranial characters which support our assertion that there is certainly more than one species of small horse from the eastern Mediterranean and southwest Asia, necessitating their future complete revision. P avlow (1890) identified a small hipparion MC III fragment from Sebastopol and referred it to a new species Hipparion minus . In his catalogue of Samos fossils, F orsyth M ajor ( 1894 : 5 and 32) referred small, very worn upper right molar from Andriano to Hipparion minus P avlow. S tuder ( 1911 : Fig. 4, pg. 196) figured a fragmentary MT III from Samos, and followed previous authors in applying the nomen Hipparion minus . K ormos ( 1911 : Fig. 1 , plate at end of text; UGR OK 557, Fig. 1 here) figured a complete skull and lower jaw from Samos retaining the nomen Hipparion minus . A ntonius ( 1913 : 244; 1919 :285—287) perpetuated the referral of Samos small hipparions further to the southwest Russian species of Pavlow, and in the 1919 publication remarked that “ Hipparion minus " was morphologically quite similar to the North American horse Neohipparion withneyi G idley, except for the North American form’s more poorly developed “Wangengrube” (POF). Abel (1926) cited the inappropriateness of using P avlow ’s type for “ Hipparion” minus to characterize any hipparion species, and proposed that the name be considered a nomen nudum. A bel ( 1926 : 430 Fig. 273; 432) then citing the skull and mandible figured by K ormos ( 1911 : Fig. 1 ), elected it as the type of a new species, Hipparion matthewi named in honor of W. D. M atthew . A bel ( 1927 : Fig. 110, pg. 120) figured this same specimen once again. W ehrli ( 1941 ) made an extensive review of Samos hipparion systematics. He nominated a new genus for all the Samos hipparions he recognized, raising them to a distinct generic rank, Hemihipparion . W ehrli ’s ( 1941 ) Samos species included: Hemihipparion proboscideum (S tu ­ der), Hemihipparion dietrichi n. sp. and Hemihipparion matthewi (A bel). The nomen Hemihipparion would be a synonym senior to Cremobipparion except W ehrli ( 1941 :374) nominated Hipparion dietrichi as the genotype species . W oodburne and B f.rnor ( 1980 : 1329), B ernor ( 1985 : 205) and B ernor et al. (in press) have cited the specific reasons why H. dietrichi should be considered as belonging to a clade (= a member of the Hipparion s. s. clade; B ernor et al., in press) quite distinct from Cremobipparion (= Group 2 hipparions of earlier work). G romova (1955: 231) recognized A bel ’s nomen Hipparion matthewi from Samos, and gave it an extensive diagnosis. Some of the more important points in her diagnosis were: range of maxillary cheek tooth length = 102.4 —113.5 mm (far less of a range than given by most subsequent authors); short muzzle; single preorbital fossa, short and deep; nasal notch shallowly incised; extremities slender. G romova ( 1952 : 232) cited only K ormos ’s ( 1911 ) skull and mandible specimen in her characterization of facial morphology, and apparently did not have the München specimen of Cremobipparion nikosi available to her for study. She included T obien ’s ( 1938 : Taf. 1) and W ehrli ’s ( 1941 ) measurements in her diagnosis of cheek tooth dimensions, and in so doing mixed the hypodigms of Cremobipparion matthewi and Cremobipparion nikosi . She cited the rarity of well known C. matthewi postcranials, but utilized the collection described by W ehrli ( 1941 ) in her diagnosis of “ Hipparion” matthewi . G abunja ( 1959 :222) and F orsten ( 1968 : 53) essentially followed G romova ’s ( 1952 ) concept of “ Hipparion” matthewi , applying it to small horses from the eastern Mediterranean and southwest Asia. S ondaar ( 1971 : 423) reviewed the American Museum collection of Samos hipparionines. He recognized and refigured the type skull and lower jaw of “ Hipparion” matthewi (UGR OK/ 557; S ondaar, 1974: Plate IV, Fig. a) and subsequently referred all AMNH Samos hipparions of small size (P2—M3 = 100—130 mm; P2—M3 = 110—135 mm), with slender limbs (MT III length = 211—242 mm, proximal width = 27—34 mm), with simple enamel plication, oval to nearly rounded protocone and slightly developed POF, to „ Hipparion“ matthewi . Of this Samos hypodigm, he figured two Quarry 5 specimens: a skull, AMNH 22936 (Plate 1 a, b) and palate AMNH 22888 (Plate 1 c). Sondaar (1971) noted that the type specimen of “H”. matthewi had a shorter muzzle than the AMNH material in his hypodigm. Although AMNH 22936 is a virtually complete adult male specimen, the nasal notch area is destroyed, disallowing any certain referral to either C . matthewi or C . nikosi. However, AMNH 22936’s preorbital fossa morphology, short POB length and P2—M3 length do conform well with either of these two Cremohipparion small species, and may reasonably be considered as being a species of that genus. Of the remaining AMNH Samos skull material referred by S ondaar ( 1971 : Tab. 1 ) to “ Hipparion” matthewi (AMNH 22888, 22936, 20788 (here two distinct specimens), 20787,22907), AMNH 22888 may have a somewhat longer P2—M3 dimension (i. e. 129.4 mm) than would be expected for either C. matthewi or C. nikosi. Likewise, the extensive mandibular series referred by S ondaar ( 1971 , Tab. II) to “ Hipparion” matthewi includes at least two specimens , AMNH 20791 (P2—M3 = 131 mm) and AMNH 22928 (P2—M3 = 134 mm) that may be larger than expected for either C . matthewi or C . nikosi. As a result of our evaluation, it is not presently clear which, if any, of the AMNH material is referrable to either Cremohipparion matthewi or Cremohipparion nikosi . B ernor ( 1985 : 220, Fig. 18) referred a skull fragment, some isolated maxillary and mandibular cheek teeth and postcranial fragments to “ Hipparion ” imatthewi. The skull fragment was the most definitive specimen, being small sized (P2—M3 = 121.9 mm) with POF and POB morphologies agreeing well with the type of Cremohipparion matthewi . Although there is no snout or nasal notch preserved, the skull specimen, GIU P100—1958, is broken at the level of P2 mesostyle, and therefore would not have nasals as retracted as C. nikosi. A referral of this specimen to C . aff. matthewi would presently be acceptable. L ehmann (1984:152 and Taf. VII) briefly described a virtually complete skull (Nr. 3002), an unassociated lower jaw (Nr. 3005) and MT II—IV elements (no numbers given) of “ Hipparion” matthewi . The skull is quite interesting in that L ehmann ’s Taf. 8, Fig. 1 shows the POB as being short, with lacrimal invading posterior rim of the POF, and POF essentially has a morphology consistent with . matthewi and C . nikosi. However, the nasals would appear to be somewhat retracted compared to the holotype of C . matthewi (posterior aspect of P2), but certainly not as far as seen in C. nikosi (mesostyle of P4). Maxillary cheek tooth morphology would appear to be consistent with that found in C. matthewi and C . nikosi, and MT III length (= 223 mm) and diaphysial width (22.2 mm.; L ehmann , 1984: 152) is consistent with the range of variation reported for Samos small horses here. It would appear that this specimen exhibits somewhat intermediate characters between . matthewi and C. nikosi in the degree of its nasal incisure. We believe that despite this intermediate nature, the difference between C . matthewi and C. nikosi is too great (see description for the latter below) to be explained as variability in a single species . K oufos ( 1984 , 1986 , 1987a , b ) has recently cited the occurrence of some small horses from Macedonia. K oufos ( 1984 ) named a new species of small horse, H. macedonicum from Ravin de Pluie 1 (RPL 1). He stated that “ Hipparion” matthewi differs from H. macedonicum in its shorter snout, longer mandibular symphysis, simpler enamel plications, rounded protocone connecting to protoloph (sic., only in later wear), small-rudimentary pli caballin, less developed protostylid, simple enamel in the flexids, as well as having single column ectostylids in the lower milk teeth. The differences in the adult dentition reveal a generally more primitive morphology typical of older European species. It is uncertain however what the juvenile dental morphology would be for Cremohipparion matthewi , since the type specimens of both C . matthewi and C. nikosi are adult individuals. K oufos ( 1984 : 313) cited that the character which most distinguished H. macedonicum from other small Turolian horses ( “H”. matthewi , “H”. gromovae , “H”. periafricanum and “H”. elegans ) was its greater molar row length/premolar length ratio (greater than 100 in H. macedonicum , less than 100 in all others; 84.0 in the type maxilla and 92.9 in the type mandible of Cremohipparion matthewi ; 84.8 in the type maxilla of Cremohipparion nikosi ). He included in this distinction the younger Ravin de Zouaves samples of small hipparion which showed lower (less than 100) length molar series/length premolar series ratios. However, since citing lower molar length/premolar length ratios in the younger Macedonian Ravin des Zouaves (RZO) series, K oufos ( 1987a : 304) referred two of three RZO specimens (RZO 38 and RZO 44, with ratios of 90 and 94.5 respectively) to H. macedonicum . It is presently not clear whether K oufos ( 1987a ) believes that the ratio change within the “late Vallesian — Turolian” aged Macedonian series is insignificant, and macedonicum has members with length molar/length premolar series both above and below 100, or whether it is possible that more than one species of small hipparion is included in the hypodigm of “H”. macedonicum , or if these differences are ontogenetically (= wear) related. If the differences in length molar/length premolar ratio prove to be significant at the species level as K oufos ( 1984 : 313) claimed, then his ( 1987a ) hypodigm would necessarily include more than one species of small horse. K oufos ’ ( 1986 : 71, Tab. 9 ) report on new material of H. macedonicum included discussion of MT III material which he cited as differing from “H”. matthewi in its longer (var. 1 = 239.0 mm), more slender diaphysis (var. 11= 26.6). Of course there are no articulated skeletons of Cremohipparion matthewi , making referrals of postcrania to the type uncertain. Indeed, the measurements of AMNH Samos small horse MT Ill’s that we record here in Table 5 include one Quarry 5 specimen that is quite close to K oufos ’ ( 1986 : Tab. 9 ) dimensions (i. e. F: AM Q 5 22893, var. 1 = 239.4 mm, var. 11 = 25.9 mm). The other Samos individual (F:AM Q 5 ) has quite a shorter length MT III (var. 1 = 213.7 mm), and slightly shorter DAW dimension (var. 11 = 24.2 mm). These individuals are included here in Fig. 3. K oufos ( 1987b : 347, Fig. 8) figured for the first time a skull of H. macedonicum , PXM 20. No raw measurements were given for POF and POB, but the shape of the former is consistent with C. matthewi, and the length of the POB has been reported by the author as being short. The nasal notch can clearly be seen in this figure to be retracted no further than the anteriormost portion of P2, as is seen in C . matthewi.