Hipparion moldavicum Gromova, 1952

Koufos, George D. & Vlachou, Theodora D., 2005, Equidae (Mammalia, Perissodactyla) from the late Miocene of Akkaşdağı, Turkey, Geodiversitas 27 (4), pp. 633-705 : 666-682

Hipparion moldavicum Gromova, 1952

LOCALITY. — Akkaşdağı, Keskin, Turkey.

AGE. — Middle Turolian, MN 12 (late Miocene).

MATERIAL. — Complete skull associated with the mandible, AK7-155; complete skull, AK2-501; 2 partial skulls, AK2-500, AK6-226; maxillary fragment with P3-M3 sin, AK5-509; maxillary fragment with P2-M3 dex, AK2-174; maxillary fragment with P3- M3 sin, AK2-504; maxillary fragment with P2-M2 sin, AK12-1; maxillary fragment with P4-M3 dex, AK6-237; maxillary fragment with P3-M2 dex, AK6- 325; 2 mandibles with p2-m3 dex and sin, AK2-178, AK5-644; 2 mandibular fragments with p2-m3 dex, AK6-187, AK2-53a; mandibular fragment with p3- m3 sin, AK4-229; mandibular fragment with p2-p4 dex, AK11-86; mandibular fragment with p2-m2 dex, AK2-52; mandibular fragment with p3-m3 dex and p4-m3 sin, AK2-53b; mandibular fragment with p3- m2 dex, AK5-287; mandibular fragment with p3-m1 sin, AK5-300; mandibular fragment with p4-m1 sin, AK5-298; mandibular fragment with m1-m3 dex, AK5-288; mandibular fragment with m1-m3 sin, AK2-171a-b; mandibular fragment with p4-m3 sin, AK2-n.n.; distal part of humerus+radiocubitus +carpals+McIII, AK6-261; 6 distal parts of humerus, AK2-233, AK3-217, AK4-146, AK5-539, AK6-20, AK12-30; 4 distal parts of radius, AK3-97, AK6-230, AK11-113, AK11/12-9; 2 McII+McIII+McIV, AK6- 248a-c, 259a-c; 13 third metacarpals, AK2-5, 243, AK4-96, AK5-6a-c (with laterals), AK5-76, 155, 543, n.n., AK6-195, 278, AK7-44, 43, AK12-35; 5 distal parts of McIII, AK2-8, AK3-144, AK5-179, AK6-9, 278; 5 proximal parts of McIII, AK3-262, AK4-97, 152, AK5-155, 546; distal part of tibia+astragalus+ tarsals+MtIII, AK6-179; distal part of tibia+astragalus+calcaneum+tarsals+MtIII, AK6-267a-i; 4 tibiae, AK3-8, AK5-551, AK6-16, 184; 26 distal parts of tibia, AK2-17, 18, 21, 117, 261, AK3-147, 154, AK4- 104, 220, 250, AK5-161, 187, 399, 401, 469, 560, 562, 561, AK6-114, 161, 186, AK11-100a, 126, AK12-42, 47, AK14-6; 7 astragali, AK2-415, AK4- 155, AK5-167, AK6-11, 53, 79, AK7-53; 4 partial astragali, AK2-263, AK10-13, AK11-129, AKK-275; 9 calcanei, AK2-31, 265, AK4-100, AK5-11, AK6- 225, AK7-93, 111, 159, AK11-55; 5 partial calcanei, AK2-266, AK5-101, 224, AK5-104, AK11-132; 2 MtIII+tarsals, AK5-154, AK6-180; MtII+MtIII+ MtIV, AK11-117; 16 third metatarsals, AK2-2, 4, 269, 399, AK3-n.n., 142, AK4-162, 163, AK5-18, 565, 545, AK6-7, AK7-45, 46, AK12-48, 49; 11 proximal parts of MtIII, AK2-271, AK3-5, AK5-17, 80, 79, 150, 483, 485, AK6-185, 217, AK11-40; 6 distal parts of MtIII, AK4-164, 168, AK5-77, AK6-14, 246, AK11-39; distal part of MtIII+PhI+PhII+PhIII, AK6- 262a-d; 4 first phalanges, AK5-114, AK6-198, AK7- 133, AKK-279; 4 second phalanges, AK5-16, 120, 338, 339; 3 third phalanges, AK3-291, AK4-179, AK7-114; part of a third phalanx, AK5-122.

DESCRIPTION

Skull and dentition

The description of the skull morphology of the small-sized hipparion from Akkaşdagwı is mainly based on the specimen AK2-501, while addition-

Miocene Equidae from Akkaşdagwı, Turkey al information is taken from the specimens AK2- 500 and AK6-226. The skull AK2-501 is almost complete, well preserved but slightly compressed laterally ( Fig. 32 ). It lacks the left zygomatic arch, part of the left orbit and the incisors. The other two specimens AK2-500 and AK6-226 are partial skulls lacking the muzzle and the posterior part of the skull. The first one is slightly deformed tilting to the right, and the second one is crushed ( Fig. 33A, B ).

Koufos G. D. F Vlachou T. D.

The small-sized hipparion from Akkaşdagwı has a narrow and elongated skull ( Fig. 32 ). Its length is 375 mm and the breadth between the anterior borders of the facial crests is approximately 130 mm (AK2-501). The muzzle is narrow and short (the index m15 × 100/m 1 in AK2-501 is 318.2). The narial opening is narrow and elliptical-shaped.The nasal notch is situated above or in front of P2 and it is placed far from the orbit (138 mm in AK2- 501). The anterior border of the nasals is situated above the canines. The preorbital fossa is large, oval, anteroposteriorly oriented ( Figs 32A ; 33A ); its maximum depth is observed medially and varies between 15 and 18 mm (AK2-501, AK2-500 and AK2-226). Both anterior and posterior rims are well expressed and situated above P3 and M2 or M3 respectively. The posterior rim is slightly pocketed, especially in the specimen AK6-226 ( Fig. 33B ). The peripheral rim is also well expressed but, in its lower part, the continuity is interrupted by a pit variable in size ( Figs32A ; 33A ). The upper part of the fossa is rugose with pits more or less expressed ( Fig. 32A ). The distance orbitpreorbital fossa is short and varies between 16 and 23.5 mm. The lacrymal suture approaches but does not touch the posterior rim of the fossa. The infraorbital foramen is situated above the posterior half of P2 and just inferior to the anteroventral rim (AK2-501) or above the anterior border of P2 and just superior to the anteroventral rim (AK6-226). The zygomatic arch is developed very close to the parietal (AK2-501). The crista facialis is strongly projected. Its anterior border is situated above the end of P4 or at the middle of M1 and close to the maxilla (distance crista facialis-alveolar level = 17.8-23.3mm). The palate is long, narrow and deep. The index m2 × 100/m13 varies from 176.5 (AK2-500) to 196 (AK2-501). The choanae are relatively wide and their anterior border is situated at the contact between M2 and M3.

The toothrow is almost straight and short comparatively to the skull length ( Fig. 34 ). The length of the toothrow varies from 127.3 mm in the old individuals to 140.2 mm in the young adults. The toothrow AK2-500 retains the dP1 ( Fig. 33A ). The teeth are small ( Fig. 34 ). The anterostyle of P2 is rounded, short and projected mesially. The length of the anterostyle reduces as the wear increases. The fossettes are closed with moderate enamel plication at their borders. The plis are narrow and shallow depending on the wear stage. The plication number is 10-29 in the premolars and 9-23 in the molars. The protocone is large comparatively to the tooth size and, in the more worn dentitions, it is oval to rounded and connected to the protoloph, while in the young to adult individuals it is more elliptical and always free. The pli caballin is well developed in the premolars but weak in the molars. It is multiple in the less worn teeth (AK2-500) while in the more worn dentitions it is double to single in the premolars and single to absent in the molars. The hypocone is elliptical-rounded with deep distal hypoconal groove while a lingual hypoconal groove is rarely present in M3 (AK2-501).

Among the cranial remains of the small-sized hipparion there is a complete skull in association with the mandible, AK7-155 ( Fig. 35 ), that shows several similarities with the above described skulls; however, it differs mainly in the shape of the preorbital fossa, the morphology of the occipital and possibly in the skull width. The preorbital fossa of AK7-155 is triangular-shaped, moderately pocketed posteriorly while the anterior rim is not well expressed ( Fig. 35A ). The maximum depth (> 15 mm) is observed at its posterior part, and the pits are absent. Comparatively to AK2-501, the occipital condyles are narrower, less curved, their backside is less flat and they are strongly projected backwards ( Figs 36 ; 35 ; 32 ). In spite of the above mentioned differences, the skull AK7-155 is similar in size and dental morphology (length, tooth size, enamel morphology), the position of the nasal notch and the proportions of the muzzle. Both skulls must be old male individuals as their canines are strong. Therefore, the differences we noted can be considered neither as sexual dimorphism nor as ontogenetic growth differences. Among the Taraklia (type locality) material of H. moldavicum there are two skulls, nos. 1256/2963 and 1256/3639, that, despite their different fossa morphology, are ascribed to the same species ( Gromova 1952: pls I, III). Unfortunately, no information is available about the width of the Taraklia skulls and their occipital morphology. In comparison with the Taraklia material, we believe that AK7-155 must be incorporated in the small hipparions group together with AK2-501, AK2-500 and AK6-226.

Mandible

The mandible has a relatively long and narrow snout with shallow and narrow cup ( Figs 37 ; 38 ). The index m2 × 100/m7 is 208 (186 for AK7- 155). The horizontal ramus has a medium depth, and the symphysis is narrow and close to the premolars (AK2-178, AK5-644). The symphysis in AK7-155 is narrow but more robust, and it is placed more distant to the premolars. The toothrow is short with small teeth ( Figs 37C ; 38D ). The parastylid is well developed in the P3, 4. The metaconid and metastylid are triangular to elliptical in the premolars and rounded in the molars. The ectoflexid is deep reaching the preflexid in the premolars and very deep in the molars where it touchs the linguaflexid. The linguaflexid is shallow and open in the premolars while it is V-shaped in the molars.

Postcranials

The metapodials are relatively long and slender ( Fig. 39 ). The index McIII or MtIII Length × 100/Radius or Tibia Length is on average 85.0 and 76.5 respectively indicating long metapodials. The robusticity index (m11 × 100/m1) is 14.4 for McIII and 13.4 for MtIII. The keel index (m12 × 100/m13) is 123.8 for McIII and 125.7 for MtIII. In the third metatarsal the articular facet for cuneiforme II is usually absent.

COMPARISONS

The species H. moldavicum was erected by Gromova (1952) based on the material from Taraklia ( Moldavia). According to her diagnosis, it is a medium-sized hipparion with long muzzle, high frontal, short toothrow, single elongated and deep preorbital fossa situated close to the orbit,

Miocene Equidae from Akkaşdagwı, Turkey moderately developed nasal opening whose posterior border is situated above the anterior border of P2. The upper teeth have an oval (“court et large”) protocone and moderate enamel plication, while the lower teeth have a deep ectoflexid. The metapodials have a moderate length and they are slender. Later on, H. moldavicum was synonymized with H. mediterraneum ( Forstén 1978) but recently it was recognized as an independent species ( Bernor 1985; Krakhmalnaya 1996; Forstén F Krakhmalnaya 1997). Bernor (1985) in an emended diagnosis for H. moldavicum , in

Miocene Equidae from Akkaşdagwı, Turkey addition to the characters given by Gromova (1952), mentioned that the preorbital fossa is subtriangular, anteroposteriorly oriented, dorsoventrally deep, slightly posterior pocketing, with distinct anterior rim, well expressed peripheral outline and short preorbital bar. The upper teeth have moderately plicated enamel, the protocone is generally rounded, and the P2 anterostyle is usually long but it can be short and rounded.

The small-sized skulls from Akkaşdagwı fit morphologically to the description that Gromova (1952) gave for H. moldavicum from Taraklia and, in general terms, share the characters that Bernor et al. (1996a) attributed to “ Cremohipparion group”. However, contrary to “ Cremohipparion group”, the studied skulls retain the dp1, and the lacrymal suture approaches but seems to not invade the posterior part of the preorbital fossa.

The logarithmic ratio diagram in Figure 40A compares the skull proportions of the Akkaşdagwı sample with H. moldavicum from various localities. On the whole, the Akkaşdagwı skull proportions track quite close those of the type sample of H. moldavicum from Taraklia. The samples have similar skull length, but they strongly differ in the position of the vomerine notch that seems to be placed closer to choanae in the Akkaşdagwı skulls. This

Koufos G. D. F Vlachou T. D.

causes the increase of the basion-vomerine length against choanae-vomerine length (measurements 3, 4 in Fig. 40A ). The position of the vomerine notch is somewhat questionable especially if the skull is crushed or deformed, such damages may change the proportion of the choanae-vomerine and vomerine-basion lengths. Consequently, we consider that the significance of these measurements is somewhat doubtful. However, the Akkaşdagwı sample differs from H. moldavicum from Taraklia in having slightly shorter and narrower muzzle (measurements 1, 15), greater skull height in front of P2 (measurement 25), larger anteroposterior orbit diameter, apparently shorter distance between preorbital fossa and orbit (measurement 32), and the infraforamen placed closer to the posterior edge of the preorbital fossa (measurement 34 in Fig. 40A ).

Miocene Equidae from Akkaşdagwı, Turkey

Some cranial remains from Maragha have been originally described by Bernor (1985) as H. aff. moldavicum . Later, Watabe F Nakaya (1991b) assigned the small-sized hipparion from Maragha to H. moldavicum studying some new material stored in the University of Kyoto ( Japan) and part of the old Maragha collection stored in the BMNH. Apart from the diagnostic characters that Gromova (1952) and Bernor (1985) gave for the species, they additionally noted that the lacrymal suture does not invade, but touch the posterior edge of the preorbital fossa, the preorbital fossa is anteroventraly oriented, the anterior rim is weakly to clearly expressed, the enamel plication number in M1 varies from 12 to 21, and the pli caballin is multiple to single depending on the stage of wear. The Akkaşdagwı sample also fits with this description, and the cranial dimensions of the two samples are more or less similar. The differences are mainly detected in the facial region ( Fig. 40A ). Comparatively to the Akkaşdagwı sample, H. moldavicum from Maragha has significantly shorter cheek teeth (m31),longer preorbital bar (m32) and smaller preorbital fossa (m33, m35), while the infraorbital foramen is placed further from the alveoli (measurement 37). Moreover, in the Maragha sample, the muzzle is longer and the teeth larger. The Figure 41 focuses our study, with a set of box and whiskers plots, on these crucial variables. The upper tooth dimensions (m7-9) are ploted in Figure 41 A-C. The medial quartile of the Akkaşdagw ı sample (1) in all related plots is virtually identical to that of the Maragha sample(2). Hence, the teeth differences that appeared in the ratio diagram (measurements 7, 8, 9 in Fig. 40A ) are not certain differences but they are due to the small size of the sample. In fact, the Akkaşdagw ı sample overlaps in its interquartile with the Maragha sample. Concerning the length of the preorbital bar (m32) ( Fig. 41D ), the Akkaşdagwı sample (1) has a slightly lower medial and lower quartile boundary than the Maragha (2) and Taraklia (4) samples. The size of the preorbital fossa at Akkaşdagwı (m33, m35) does not really differ from that of the Maragha sample. In Figure 41E and G , the Akkaşdagwı (1) interquartile range overlaps with Maragha (2). Similar observations are also true for the m34 and m36 ( Fig. 41F, H ), while the plot for the m37 shows that Maragha (2) compared to Akkaşdagwı has a higher median and interquartile range.

H. matthewi is another small-sized hipparion known from various localities in Greece, Turkey, Bulgaria and former Yugoslavia. The type specimen has been found in an unknown locality of Samos while, according to Sondaar (1971), the species is very common in Q5 (Samos). The Akkaşdagwı skulls are quite similar morphologically to the type of H. matthewi . The small size, the structure of the preorbital fossa, the muzzle morphology and the short narial opening are some of the common features between the two samples. However, H. matthewi from Q5 and especially the type skull are significantly smaller than the Akkaşdagwı skulls, as shown in the ratio diagram ( Fig. 40B ). Box and Wiskhers plots ( Fig. 41 ) confirm the size difference of the two samples (1, 4) and show that besides the similarities in the preorbital fossa morphology, the interquartile ranges of the corresponding measurements of H. matthewi (sample 4) are below the Akkaşdagwı ones (sample 1).

Hipparion elegans is closed to H. matthewi , H. moldavicum and to the studied form. It was erected by Gromova (1952) as a new species from Pavlodar ( Kazakhstan), morphologically similar to H. moldavicum . Hipparion elegans differs from H. moldavicum in having a smaller size, and quite deep preorbital fossa with very pronounced peripheral rim ( Fig. 40B ). All these characters also distinguish it from the Akkaşdagwı sample.

Hipparion mediterraneum , H. proboscideum and H. forstenae are also clearly different from the Akkaşdagwı sample firstly because of their differences in size and secondly because of their facial morphology. They are all larger in size and characterized by a deep, anteroventrally oriented preorbital fossa, which is separated from the buccinator fossa by a canine fossa.

The mandible of the studied form is very close to H. moldavicum from Taraklia ( Fig. 42 ), confirming the similarity of the two samples. Hipparion matthewi is smaller, while H. elegans is closed to H. moldavicum and to the Akkaşdagwı form.

Concerning the metapodials, the Akkaşdagwı sample is also more similar to H. moldavicum than to H. matthewi and to H. elegans . Figure 43A shows that the metacarpals from Akkaşdagwı have rather larger dimensions compared to H. matthewi from various localities and H. elegans . The proportions of the Akkaşdagwı sample follow closely those of H. moldavicum from Taraklia, Tudorovo and Novaya Elisavetovka even though the metacarpal in the former sample is longer than the later ones ( Fig. 43B ). Hipparion moldavicum from Novaya Emetovka 2 is the exception of the previous ascertainment since its metacarpal is as long as that of Akkaşdagwı sample, but the rest of the dimensions are larger and closer to H. mediterraneum from Pikermi ( Fig. 43B ). The mean robusticity index (m11 × 100/m1) is 14.4 in the Akkaşdagwı material versus 14.5-15.3 in H. moldavicum from the various localities, 12.8 for H. matthewi from Samos Q5, 12.3 for H. matthewi from Kemiklitepe A-B ( Turkey), 16.0 for H. matthewi from Dytiko (Axios Valley, Greece) and 14.0 for H. elegans confirming in general terms the previous assumptions.

The metatarsals from Akkaşdagwı are also very close to those of H. moldavicum from Taraklia, Tudorovo and Novaya Elisavetovka ( Fig. 44A ). The robusticity index (m11 × 100/m1) is 13.4 for the Akkaşdagw ı sample versus 12.8-13.6 for H. moldavicum from various localities, 11.2- 12.4 for H. matthewi and 12.0 for H. elegans . The logarithmic ratio diagram ( Fig. 44A ) shows that the Novaya Emetovka 2 form has slightly longer metatarsal with deeper mid-shaft diaphysis (measurement 4), wider proximal articular surface (measurement 5) and larger suprarticular width (measurement 10) in comparison to the Akkaşdagwı form and the other H. moldavicum samples. On the other hand, the metatarsals of H. m o l d a v i c u m f r o m T a r a k l i a a n d N o v a y a Elisavetovka appear to be shorter with strongly reduced distal articular dimensions, shallower proximal articular surface and especially, when the Akkaşdagwı sample compared to the Taraklia sample, they further differ in having more robust diaphysis. The metatarsal of H. mediterraneum from Pikermi is clearly distinguished from that of the studied sample in being significantly shorter with a quite robust diaphysis, larger proximal articular width and larger distal maximal depth of the medial condyle (measurements 3, 4, 5, 14 in Fig. 44B ). Contrary to H. mediterraneum , H. matthewi and H. elegans h a v e l o n g a n d s l e n d e r m e t a t a r s a l a s i n t h e Akkaşdagwı sample. They differ, although, from the Akkaşdagw ı metatarsal in having strongly smaller dimensions ( Fig. 44B ).

The astragalus and calcaneum from Akkaşdagwı are also very close to H. moldavicum from Taraklia, despite the smaller height for astragalus ( Fig. 45 ). In comparison to H. mediterraneum from Pikermi, the astragalus is larger, but the calcaneum is similar in size. However, H. moldavicum and the Akkaşdagwı form have calcaneum with significantly broader distal articular surface than in H. mediterraneum (measurement C 6 in Fig. 45 ). Similarly to the metapodials, H. matthewi and H. elegans differ from the Akkaşdagwı sample by their smaller tarsal bones ( Fig. 45 ).

The comparison of the phalanges adds nothing more to the previous observations. The first and second phalanx of the Akkaşdagwı small-sized form have more or less similar dimensions to those of H. moldavicum , and they are larger compared to H. matthewi and H. elegans ( Fig. 46 ). Both phalanges from Akkaşdagwı differ from that of H. mediterraneum in being longer, more slender but with larger distal articular dimensions (measurements 1, 3, 6 in Fig. 46 ).

In summary, the morphological and metrical comparisons suggest that the small-sized Akkaşdagwı hipparion is closely related to H. moldavicum from Taraklia while it is larger than H. matthewi and H. elegans and smaller than H. mediterraneum from Pikermi ( Fig. 47 ).

Figures

FIG. 32. — Hipparion moldavicum, Akkaşdağı, Turkey, skull AK2-501; A, lateral view; B, dorsal view; C, ventral view. Scale bar: 2 cm.

FIG. 33. — Hipparion moldavicum, Akkaşdağı, Turkey, skull AK6-226; A, lateral view;B, left toothrow, occlusal view. Scale bars: 2 cm.

FIG. 34. — Hipparion moldavicum, Akkaşdağı, Turkey; A, right toothrow, AK2-501; B, right toothrow, AK7-155. Scale bar: 1 cm.

FIG. 35. — Hipparion moldavicum, Akkaşdağı, Turkey, skull AK7-155; A, lateral view; B, dorsal view; C, ventral view. Scale bar: 2 cm.

FIG. 36. — Hipparion moldavicum, Akkaşdağı, Turkey, occipital region; A, skull AK2-501; B, skull, AK7-155. Scale bars: 2 cm.

FIG. 37. — Hipparion moldavicum, Akkaşdağı, Turkey, mandible AK2-178; A, lateral view; B, dorsal view; C, right toothrow, occlusal view. Scale bars: 2 cm.

FIG. 38. — Hipparion moldavicum, Akkaşdağı, Turkey, mandible AK7-155; A, lateral view; B, dorsal view; C, left toothrow, occlusal view. Scale bars: 2 cm.

FIG. 39. — Hipparion moldavicum, Akkaşdağı, Turkey; A, third metacarpals, AK2-243, AK6-259b, AK6-248b; B, third metatarsals, AK6-7, AK12-48. Scale bars: 2 cm.

FIG. 40. — Logarithmic ratio diagram comparing the skull of Hipparion moldavicum from Akkaşdağı with H. matthewi, H. elegans (B) and H. moldavicum (A) from various localities. Standard H. mediterraneum, Pikermi, n = 1-9 (Koufos 1987a).

FIG. 41. — Box and Whiskers plots on skull variables; A, premolar length; B, molar length; C, length of the upper maxillary teeth; D, length of the preorbital bar; E, maximum length of the preorbital fossa; F, distance between the back of the preorbital fossa and the infraorbital foramen; G, height of the preorbital fossa: perpendicular to its maximal length; H, distance between the preorbital fossa and the facial crest; I, height of back of the infraorbital foramen above the alveolar border. Samples: 1, Hipparion moldavicum, Akkaşdağı; 2, H. moldavicum, Maragha; 3, H. matthewi, Samos; 4, H. moldavicum, Taraklia.

FIG. 42. — Logarithmic ratio diagram comparing the mandible of Hipparion moldavicum with other similar forms from various localities. Standard H. mediterraneum, Pikermi, n = 2-24 (Koufos 1987a).

FIG. 43. — Logarithmic ratio diagram comparing the third metacarpal of Hipparion moldavicum from Akkaşdağı with other similar forms (A), and with H. moldavicum from various localities (B). Standard H. mediterraneum, Pikermi, n = 12-39 (Koufos 1987a).

FIG. 44. — Logarithmic ratio diagram comparing the third metatarsal of Hipparion moldavicum from Akkaşdağı with H. moldavicum (A) and other similar forms (B) from various localities. Standard H. mediterraneum, Pikermi, n = 11-65 (Koufos 1987a).

FIG. 45. — Logarithmic ratio diagram comparing the astragalus and calcaneum of Hipparion moldavicum from Akkaşdağı with other forms from various localities. Standard H. mediterraneum, Pikermi, n = 6-27 (Koufos 1987a).

FIG. 46. — Logarithmic ratio diagram comparing the first and second phalanges of Hipparion moldavicum from Akkaşdağı with other forms from various localities. Standard H. mediterraneum, Pikermi, n = 2-8 (Koufos 1987a).

FIG. 47. — Logarithmic ratio diagram comparing the front and hind legs of Hipparion moldavicum from Akkaşdağı with other forms from various localities. Standard H. mediterraneum, Pikermi, n = 3-61 (Koufos 1987a).