Introduction

Paramachaerodus Pilgrim (1913) is a machairodontine felid. That genus and its close relatives were poorly known until the discovery of the Batallones-1 locality (Spain) with its abundant material (Salesa et al. 2003). Until the revision of the genus, stimulated by the discovery of the important material from Batallones-1 (Salesa et al. 2010), two, and eventually three, Paramachaerodus species were accepted in the more recent taxonomic studies as valid for this genus: Pa. ogygia from the Vallesian, Pa. orientalis from the early to middle Turolian and (eventually) Pa. maximiliani from the late Turolian (Beaumont 1975; Morales 1984; Morlo 1997; Ginsburg 1999; Salesa et al. 2003, and references therein). The study of the rich fossil material from Batallones-1 led to the concept of the existence of two genera with close features—Promegantereon Kretzoi known mostly in Vallesian time and the more evolved Paramachaerodus Pilgrim, which replace Promegantereon in Turolian (Salesa et al. 2010). Thus, Promegantereon Kretzoi, considered at the time of this study as a synonym of Paramachaerodus was resurrected. New material from Hezheng district (Gansu Province, China) and Hadjidimovo (South Bulgaria) give the opportunity to develop further ideas about the taxonomy and evolution of the primitive Promegantereon–Paramachaerodus saber-toothed felid lineage.

Materials and methods

The Chinese specimens (a skull and a semimandible from two different individuals) are collected from the Chinese late Miocene site of Shilei locality (Fig. 1a) and housed in the IVPP. The Bulgarian material (two partial mandibular rami, a separate pm3, and a separate upper canine) was excavated by D. Kovachev in the 1980s, came from the Bulgarian Turolian site of Hadjidimovo-1 (Fig. 1b), and is stored in the NMNH-A.

Fig. 1
figure 1

Locations of the localities on the map. a The location of Shilei locality in Hezheng district, Linxia Basin (Gansu Province, China); b The location of Hadjidimovo in southwestern Bulgaria

Full size image

The finds were compared to other similar fossils and in particular to the species of Paramachaerodus and Promegantereon, discussed in the comprehensive work of Salesa et al. (2010). Comparisons with fossil specimens are based on direct examination of the specimens, on figures and descriptions from the literature, and on excellent photos kindly provided by M. Salesa.

The morphological descriptions follow the terminology used by Merriam and Stock (1932), Salesa et al. (2010), and Berta (1987) with some modifications (Fig. 2a). Measurements (in mm) (Fig. 2b; Tables 1, 2, 3) were taken with a digital caliper.

Fig. 2
figure 2

a Nomenclature of teeth used in this work (after Berta 1987 with some modifications); b explanation of cranial and mandibular measurements taken (after Merriam and Stock 1932; see Salesa et al. 2010, with some modifications)

Full size image
Table 1 Comparative skull and mandible dimensions of Hezheng and Hadjidimovo specimens of Paramachaerodus transasiaticus sp. nov. and “Promegantereon” ogygia (Batellones-1, Spain)
Full size table
Table 2 Measurements of upper dentition of Paramachaerodus transasiaticus sp. nov. and other species
Full size table
Table 3 Measurements of lower dentition of Paramachaerodus transasiaticus sp. nov. and other species
Full size table

For the cladistic analysis, 18 cranio-mandibular characters (Appendix) were scored across six taxa (Proailurus lemanensis, “Promegantereon” ogygia, Paramachaerodus orientalis, Paramachaerodus maximiliani, Paramachaerodus transasiaticus sp. nov (Hadjidimovo material), and Pa. transasiaticus sp. nov (Hezheng material), using Proailurus lemanensis as an outgroup. Most of the characters analyzed are modified from the revision of the genus Paramachaerodus by Salesa et al. (2010). Supplemental characters reflect some additional features, which we consider in our taxonomical approach as important in the evolution of the mentioned taxa. The data matrix was compiled in Mesquite version 3.04 (Maddison and Maddison 2015) and the cladistic parsimony analyses were carried out in the program TNT version 1.1 (Tree Analysis using New Technology; Goloboff et al. 2008).

Fossil localities

The skull V 20106 and the mandible V 20107 were collected at Shilei locality in Guanfang village of Guanghe county, Hezheng district in the Linxia Basin, southeastern part of Gansu Province, China (Fig. 1a). This district preserves a continuous and well-developed sedimentary sequence from the Oligocene to the Holocene, which yield abundant mammal fossils (Deng et al. 2004). Thanks to the effort of Z.-X. Qiu and T. Deng from IVPP, who organized a lot of field work and fossil investigation in that district, more than 100 localities have been found, and many new fossil remains of mammals have been reported in Linxia Basin since the 1980s (Deng et al. 2013).

Fossils from Shilei locality were discovered in red clay of the upper part of Liushu Formation, where the thickness of sediments approaches 150 m and is overlain by younger Malan loess. The lithology of the outcrops shows medium brown (5YR4/4 after GSA Rock Color Chart 1991) and dark yellowish orange (10YR6/6) silty mudstone and calcareous argillaceous siltstone with calcium nodules varying in size. Many mammal fossils are discovered in this locality including the felid materials described below, which were collected by T. Deng in 2001. The fossils from that locality are placed in the Yangjiashan Fauna as the result of detailed stratigraphic and associated mammal fauna comparisons, such as Chilotherium wimani, Tapirus hezhengensis, Machairodus palanderi (Deng et al. 2004, 2008, 2013). The Yangjiashan Fauna is placed in the Bahe Stage of the late Miocene (NMU 10), and is a Chinese equivalent of the European Turolian, in particular the MN11 zone of Europe, 8.7–8 Ma (Deng 2006; Deng et al. 2013).

HD 9226, HD 9227, HD 9227-2, and HD 9228 were collected at the Hadjidimovo-1 locality in southwestern Bulgaria (Fig. 1b). Hadjidimovo is situated in the Mesta River valley, close to the eastern end of the Pirin Ridge, not far from the Greek border. Four fossiliferous sites are known in that area, and the most important is Hadjidimovo-1 (Hadjidimovo-Girizite), a fossiliferous horizon with light clayey sands belonging to the Nevrokop Formation (Vatsev 1980). HD-1 has produced more than 20,000 fossils, now housed in NMNH-A), representing approximately 30 species of mammals (Primates, Rodentia, Carnivora, Proboscidea, Perissodactyla, Artiodactyla), including the felid described below, making it one of the richest late Miocene sites of the eastern Mediterranean and peri-Pontic areas (Spassov 2002; Geraads and Spassov 2009). Although the Hadjidomovo fossiliferous area was first mentioned by Nikolov (1973, 1985) (the exact locality discovered by him remains unknown), this huge collection was mostly accumulated and cared mainly between 1985 and 1998 thanks to the efforts of D. Kovachev. Hadjidimovo-1 is older than Pikermi, as indicated by the evolutionary stages of Mesopithecus (Koufos et al. 2003), Hipparion (Cremohipparion) mediterraneum (Hristova and Spassov 2005), and Adcrocuta eximia (pers. obs. NS), as well as by the presence of Tragoportax rugosifrons (Spassov and Geraads 2004). Spassov (2002) suggested that the locality likely could be placed near the MN11/MN12 boundary, and recent faunal analysis demonstrates that the locality must be early Turolian in age (Hristova et al. 2013). It could be included within MN11, and probably in the second half/the end of the zone.

Systematic palaeontology

Order Carnivora Bowdich, 1821

Family Felidae Fischer von Waldheim, 1817

Subfamily Machairodontinae Gill, 1872

Genus Paramachaerodus Pilgrim, 1913

Type species. Felis ogygia Kaup, 1832, by original designation.

Other species included: P. orientalis (Kittl, 1887), P. maximiliani (Zdansky, 1924).

Paramachaerodus transasiaticus sp. nov.

Figures 3–5.

Etymology. From “trans-Asia” (i.e. “across”, “beyond” Asia), in relation to the discovery of this species in the eastern part of Central Asia and eastern Europe.

Holotype. A skull (IVPP V 20106) with its left zygomatic arch missing, occipital and left side of the frontal region damaged, with right I3, right canine broken at the tip, PM3, and right PM4.

Type locality and horizon. Shilei, Guanghe County, Gansu Province (China), Liushu Formation, Turolian (late Miocene, MN11).

Referred material. Chinese material: A slightly damaged skull (IVPP V 20106) and a complete left mandibular ramus (IVPP V 20107) that are not derived from the same individual. Bulgarian material: Two partial mandibular rami from Hadjidimovo-1, HD 9227 (left side, with c1, pm4-m1, and a separate pm3 only preserved as a cast from the same hemi-mandible, HD 9227-2) and HD 9228 (right side, with m1 and a fragment of pm4 from the same individual as HD 9227); and a complete right upper canine crown—HD 9226.

Stratigraphic and geographic distribution. From southern (southeastern) Europe to Central Asia (China) in the time equivalent to the early Turolian (MN11) (second half of the early Turolian?) (after the European Neogene Land Mammal ages).

Diagnosis. Skull with moderately developed sagittal crest and relatively short neurocranium. Teeth with mixed features of Paramachaerodus and Promegantereon (sensu Salesa et al. 2010) such as: presence of crenulations on the upper canines, but not regularly on the lower ones and generally on all upper and lower cheek-teeth; although all crenulations are rather weak. PM3 is very long with elongated and constricted mesial part and strong disto-lingual expansion, there is a small mesial cusp on its antero-lingual part. The buccal border of PM4 is sinusoidal, with strongly expanded mesio-labial part of parastyle, indicating the beginning (a rudiment) of an ectostyle; the protocone of PM4 is relatively large and mesio-distally elongated. PM4 is much longer than PM3. PM3 and PM4 are not aligned in occlusal view. M1 is with two roots. Very short c-pm3 diastema. Possible presence of a strongly reduced, vestigial pm2. Low and elongated cutting m1 with an extremely elongated talonid and a well-marked metaconid. Mandible symphysis rostral surface is moderately verticalized, with incipient mandible flange and moderately marked mental crest on its antero-lateral margin. The coronoid process is high with more vertical anterior edge.

Differential diagnosis. Paramachaerodus transasiaticus sp. nov. differs from Pa. ogygia by its slightly larger size, by the relatively short neurocranium and weaker expressed sagittal and temporal crests, the narrower nasal aperture, the shorter and rounded zygomatic process of the fontal bones, the weaker mastoid process, the regular presence of thin crenulations on nearly all teeth (which are missing as a rule in Pa. ogygia), the more elongated PM3 (in particular the mesial part), the presence of a rudimentary ectostyle, the PM4 more elongate than PM3, the shorter c-pm3 diastema, the elongated talonid-metaconid complex of m1 with a larger metaconid, the incipient flange on the mandibular symphysis, and the more marked mental crest on the antero-lateral margin. The new species differs from Paramachaerodus orientalis by the straight anterior border of the nasal bones (without a distal projection in dorsal view), the narrower nasal aperture, the weaker tooth crenulations, a PM3 with a disto-lingual expansion, a PM4 with a much less developed ectostyle and a sinusoidal rather than straight buccal border, the reduced protocone of PM4, the longer M1 with two roots, the much shorter c-pm3 diastema, the much more developed m1 talonid-metaconid complex (with an elongate talonid and larger metaconid), the mandible coronoid process with more vertical anterior edge. Compared to Pa. maximiliani, Pa. transasiaticus sp. nov. differs by its smaller overall size, the distinctly smaller I3 and upper canine, the markedly weaker tooth crenulations, the longer PM3 (more elongated mesially), the much less developed PM4 ectostyle, the sinusoidal rather than straight PM4 buccal border, the more developed and forwardly situated PM4 protocone, the longer M1 with two roots, and the m1 with much more elongate talonid and metaconid.

Description. Cranium. As Paramachaerodus ogygia (see in: Salesa et al. 2010), the cranium V 20106 is similar in size (Table 1) and proportions to that of an extant puma (Puma concolor), but relatively lower. It is slightly compressed by external lateral stress, its right half part is almost completely preserved (Fig. 3a-A). The nasals are sub-rectangular in shape, with rostral margin wider than the caudal one and with convex caudal border. They are pinched in the middle part and medio-laterally convex in dorsal view. The anterior border of the nasals is rather straight. The nasal aperture is relatively narrow, close to piriform in shape (Fig. 3a-B). The rostral part of the premaxilla is damaged and unclear in shape. The muzzle is short and broad. The frontal region is wide and flat, possibly slightly compressed. The zygomatic processes are short and blunt. The infraorbital foramina are relatively large, higher than wider and placed at the level of the mesial part of the parastyle of PM4 (Fig. 3a-C). The postorbital constriction is well marked. The neurocranium is short and bulged laterally. Spina nasalis caudalis is not marked on the choanae aperture (Fig. 3a-D). The temporal and sagittal crests join at the level above the glenoid fossa. They are relatively weak, especially the temporal ones. The transversal crest is strong. The occipital surface is close to an equilateral triangle. In lateral view, the orbital surface is oval, post-upwards oriented. The zygomatic arch is robust as in Paramachaerodus ogygia. The postorbital process of the zygomatic arch is low and rounded, positioned anteriorly to that of the frontal one. Judging from the right zygomatic arch, this structure has a maximum width near to the caudal part. The posterior palatine foramen is situated antero-ventrally to the sphenopalatine one, and they are similar in size, unlike in the Paramachaerodus ogygia (see also Salesa et al. 2010), where the former is much smaller than the latter. Like Pa. ogygia, the optic foramen and the orbit fissure are placed as in a pantherine cat. The places of foramen rotundum and the oval foramen are placed also as in Pa. ogygia (see also Salesa et al. 2010) (Fig. 3a-E). The tympanic bullae are rather bulged. The mastoid process is relatively weak compared to Pa. ogygia. The jugular process is reduced in size when compared with that of pantherines but close in size and shape to Pa. ogygia. Its apex don’t reach ventrally the level of the occipital condyle border.

Fig. 3
figure 3figure 3

a Paramachaerodus transasiaticus sp. nov., holotype (skull—IVPP (V) 20106) from Shilei, China: A dorsal view; B rostral view; C right lateral view; D ventral view; E left lateral view. b Paramachaerodus transasiaticus sp. nov., left hemi-mandibular ramus IVPP (V) 20107 from Shilei, China: A buccal view; B occlusal view; C lingual view

Full size image

Upper incisor. I3 is large with caniniform crown. Its disto-buccal ridge is more marked than the mesial one. Its detailed morphology is similar to the ones of Pa. ogygia (see also Salesa et al. 2010): the distal surface bears a marked cingulum, which developed into a large mesial, basal cusplet. There is a short diastema, approximately 3 mm long between I3 and the upper canine.

Upper canines (Figs. 3a, 4a, b). The crown is laterally flattened and slightly curves backwards. The buccal surface is smooth and convex (in cross section), the lingual one is slightly concave on its mesial surface and become slightly convex in its middle portion. Both mesial and distal cutting edges are well developed, the mesial edge is lingually deflected at the base and slightly arched laterally after toward the apex; the distal one is sharper. There are very thin crenulations on both edges.

Fig. 4
figure 4

Paramachaerodus transasiaticus sp. nov. from Hadjidimovo-1 (Bulgaria). a, b, HD 9226, right upper canine, in a lingual and b buccal views. c, d HD 9227-2, left p3 (cast), in c lingual, d buccal and e occlusive views; f–h HD 9228, fragment of right hemi-mandible, in f lingual view, g buccal view and h occlusive views; i–k HD 9227, left hemi mandibular ramus, in i lingual view, j buccal view, and k occlusive views

Full size image

PM3. As in Pa. ogygia (see Salesa et al. 2010), the crown is mesio-distally lengthened, with a straight buccal border and a marked disto-lingual expansion, showing a marked constriction at the level of the anterior part of the main cusp. There is a small, but distinct mesial cusp on the mesio-lingual side of the teeth, whereas the distal cusp and the main one are strong. Very thin crenulations can be seen on the anterior edge of the main cusp. There is a cingulum around the distal border, which forms a small distal tubercle straightly behind the distal cusp.

PM4. The upper carnassial has a relatively large protocone, mesio-distally elongated, placed between the parastyle and the paracone. The parastyle is well developed, its height is about the half of the height of the paracone. Thin crenulations can be seen on the anterior edge of the paracone. The buccal border of PM4 is sinusoidal, with strongly expanded mesio-labial part of the parastyle, indicating the beginning of the ectostyle. The metacone-metastyle blade is much shorter than the parastyle-paracone. PM3 and PM4 are not aligned and form a slight angle in occlusive view.

M1. Absent but preserved two alveoli.

Mandibles (Figs. 3b, 4f–k). They are relatively short and stout. V 20107 has three mental foramina, the first one being placed at the level of the postcanine diastema, the main one is at the level between pm3/pm4. The foramina number in HD 9227 is not very clear because of the slightly damaged surface, but the main one is under pm3. Mandible symphysis rostral surface is close to vertical (the angle between the anterior and the ventral border of the horizontal ramus is about 110°), with incipient mandible flange and moderately marked mental crest (but better marked than in Pa. ogygia) on its antero-lateral margin. The masseteric fossa is wide and deep, its anterior border reaches the level of the protoconid of m1, its ventral border has a prominent ridge, extending to the lateral end of the condyle. The coronoid process exceeds in height the level of the lower canine and is with only slightly inclined anterior edge.

Lower canine. Relatively reduced, clearly smaller than the upper one. It is slightly compressed and curved backwards, its lingual face is flat, whereas the buccal face is smooth and convex. The distal ridge is more marked than the mesio-lingual one, thin crenulations on both ridges can be seen clearly on the Hezheng mandible but not on the Hadjidimovo one.

pm2. Practically absent. In the Hezheng and Hadjidimovo mandibles, there is a very small hollow in front of pm3, which could be a vestigial alveolus of pm2, especially in the case of Hezheng fossil.

pm3. The crown is elongated distally and triangularly shaped in occlusal view—the distal part markedly wider than the mesial part. It is slightly concave at the middle part of both the lingual and buccal faces. It dos not have an additional mesial cuspid, and the distal one is reduced. There is a smooth posterior cingulid. Very thin crenulations are observed on the anterior edge of the main cuspid. This tooth is slightly oblique in relation to pm4 in Hezheng mandible.

pm4. The crown is very long and blade like, 4-cusped. The main cuspid is well developed, much higher than the mesial and distal ones and highly positioned in relation to paraconid of m1. It presents very slight crenulations on the mesial edge. The mesial and distal cuspids are almost the same size. The marked posterior cingulid forms a small distal cusplet. A mesio-buccal basal expansion is present in the tooth of the Hezheng mandible. pm4 is not aligned with m1.

m1. The lower carnassial is remarkably elongated and relatively low, the protoconid is significantly longer than the paraconid and separated from it with strongly marked notch. The tanolid-metaconid complex is very well developed (Fig. 5). The metaconid is well expressed and separated from the upper part of the distal edge of the protoconid by a visible horizontal groove. The talonid is crest-like and very elongated. Very thin crenulations are visible on the paraconid-protoconid blade (Hadjidimovo) and on the distal edge of the protoconid.

Fig. 5
figure 5

M1 of HD 9227 showing the very well developed tanolid-metaconid complex

Full size image

Comparisons

The characters of the known mandibles from Hezheng and Hadjidimovo are almost identical, with only minor differences. The mandible V 20107 seems to have more securely a vestigial pm2, which is possibly absent in the Hadjidimovo one, but the range of the variation of this feature (presence/absence of pm2) in “Promegantereon” from the Batallones-1 sample (in which pm2 is absent in some specimens) is not smaller. Furthermore, in the Hezheng specimen V 20107, the lower canine is crenulated (crenulations present in both specimens are very thin and present in the cheek-teeth), and this difference might just be individual variation. In the Hezheng mandible V 20107, pm3 is slightly more mesio-distally elongated, the size of the mandible is slightly larger (Table 1), and the mesio-buccal basal expansion of pm4 is much stronger. All of these characters might be the result of individual, geographic (subspecific), or sexually dimorphic variation. The most important shared features among the material from Hezheng (China: V 20106 and V 20107) and Hadjidimovo (Bulgaria: HD 9226, HD 9227, and HD 9228) include the very elongated m1 with an extremely developed talonid-metaconid complex and the presence of very thin enamel crenulations. Those shared features give us the opportunity to unite the Chinese and Bulgarian specimen into one species.

The skull and mandibles of the new species present typical machairodont characters, including the elongated and mediolaterally compressed upper canines, the crenulation of teeth edges, the existence of a (incipient) mandible flange, and a marked mental crest on the symphysial mesio-buccal margin. With these characters, the fossils described above can be separated clearly from Yoshi Spassov and Geraads (2015), Pristifelis attica (Wagner 1857), Styriofelis Kretzoi (1929), the oldest Pantherinae Felis pamiri Ozansoy 1965, and Panthera blytheae (Tseng et al. 2014). In other words, the new species differs from the conical-toothed cats of the late Miocene of Eurasia (Spassov and Geraads 2015). Given that our new species is half the size of Machairodus/Amphimachairodus (Tables 1, 2), Paramachaerodus transasiaticus will be compared here only with contemporaneous and similar sized taxa of Machairodontinae with crenulated teeth, and excluding the so-called Metailurini (see Salesa et al. 2010; Spassov and Geraads 2015).

Thus, the comparison will focus on the “Promegantereon”–Paramachaerodus group discussed in details by Salesa et al. (2010) in their comprehensive work, where they presented a review of the systematics of the genus Paramachaerodus and a detailed description of the fossils from Batallones-1, attributed by them to Promegatereon ogygia.

Comparison with “Promegantereon” ogygia. Compared to the holotype of Paramachaerodus ogygia (Kaup, 1832), a fragment of right hemi-mandible with c1-pm4 from Eppelsheim, Germany (Beaumont 1975), all measurements of Paramachaerodus transasiaticus are slightly larger, except for the much shorter c-pm3 diastema in our new species (Table 3). Pa. transasiaticus has (weak) crenulations on the upper and lower canines and on all borders of the teeth, and this contrasts to the state in the Eppelsheim material where they are consistently absent. The pm3 of the new species also is rather narrow mesially, and the mandible has a more marked mental crest on the anterolateral margin. Compared with “Promegantereon” ogygia fossils from Batallones-1 (Salesa et al. 2010), most of the general features of the cranium from Hezheng are rather similar to the Batallones specimens, but the new species displays several advanced features. Some of those differences include, a relatively short neurocranium and weaker expressed sagittal and temporal crests, a narrow nasal aperture (assessed by comparison with the photos provided by M. Salesa), the shorter and rounded zygomatic process of the fontal bones, the weaker mastoid process, the thin crenulations on nearly all teeth, the more elongated PM3 (in particular the mesial part), the rudimentary ectostyle formed by a mesio-buccal expansion of the buccal border of PM4, the slightly shorter protocone of PM4, the PM4 more elongate than PM3, the shorter c-pm3 diastema, the elongated talonid-metaconid complex of m1 with a larger metaconid, the incipient flange on the mandibular symphysis, and a more marked mental crest on the antero-lateral margin.

Comparison with Paramachaerodus orientalis. Compared with Pa. orientalis from Maragheh (Iran) and Pikermi (Greece) (Kittl 1887; Salesa et al. 2010), Paramachaerodus transasiaticus exhibits some more primitive features and several other differences, including a straight anterior border of the nasal bones (without a distal projection in dorsal view), a narrower nasal aperture, weaker tooth crenulations (canines included, although some teeth may lack crenulations, see description above), a PM3 with a disto-lingual expansion, a PM4 with a much less developed ectostyle (only a rudimental one), and a sinusoidal rather than straight buccal border, a reduced protocone of PM4, a longer M1 with two roots, a much shorter c-pm3 diastema, a more developed m1 talonid-metaconid complex (with an elongate talonid and larger metaconid), and a coronoid process is with more vertical anterior edge.

Comparison with Paramachaerodus maximiliani. As compared to Pa. maximiliani from Shangyingou (Henan Province, China) (Zdansky 1924) and Venta del Moro (Spain) (Salesa et al. 2010), Pa. transasiaticus shows a number of primitive features and differences including a smaller overall size, a distinctly smaller I3 and upper canine (Table 2), much weaker tooth crenulations, a longer PM3 (more elongated mesially), a much less developed PM4 ectostyle (only a rudimentary one), a sinusoidal rather than straight PM4 buccal border, a more developed and forwardly situated PM4 protocone, a longer M1 with two roots, and a m1 with a much more elongate talonid and a metaconid.

Discussion

Taxonomy of the “Promegantereon”-Paramachaerodus lineage. Paramachaerodus was established by Pilgrim (1913) to encompass several published felid fossils, including Machairodus orientalis Kittl, 1887, Machairodus schlosseri Weithoffer, 1888, and Felis ogygia Kaup, 1832, from the late Miocene of Maragheh (Iran), Pikermi (Greece), and Eppelsheim (Germany). In 1929, Kretzoi created the genus Pontosmilus for a number of taxa (as Po. ogygia, Po. hungaricus, Po. schlosseri, Po. orientalis and Po. indicus). He restricted Paramachaerodus to the single species Pa. pilgrimi which is now excluded from this genus by Salesa et al. (2010). Meanwhile, Pilgrim (1931) considered Pontosmilus as a synonym of Paramachaerodus. Later, Kretzoi (1938) established the new genus Promegantereon to accommodate “Felis” ogygia. More recently, Promegantereon has been considered a junior synonym of Paramachaerodus (Beaumont 1975, 1978), and (as we mentioned above) few species were accepted as valid for this latter genus: Pa. ogygia from the Vallesian, Pa. orientalis from the early to middle Turolian and (eventually) Pa. maximiliani from the late Turolian. Thus, Pa. hungaricus and Pa. shlosseri were considered synonyms of Pa. orientalis (Beaumont 1975, 1978; Morales and Soria 1977; McKenna and Bell 1997; Montoya 1994; Morlo 1997; Ginsburg 1999; Salesa et al. 2003; see more details in Salesa et al. 2010). The taxonomic concept, according to which Promegantereon represents a synonym of Paramachaerodus lasted in a long time until the discovery of a large number of fossils of a Paramachaerodus-like sabre-toothed cat from Batallones-1 (Spain). The abundant Batallones-1 machairodontine material was studied by Salesa et al. (2003, 2005a, b, 2010). Their work led to the concept of the existence of two closely related, but successive in time genera (Promegantereon and Paramachaerodus) in the late Miocene, and to the resurrection of Promegantereon Kretzoi (Salesa et al. 2010).

As we noted above, the fossils from Hezheng and Hadjidimovo are almost identical with some very small differences which we consider as individual/sexual or/and subspecific (geographic) and all this material must be assigned to one and the same species of the “Promegantereon”–Paramachaerodus lineage, i.e. of the genus Paramachaerodus. It is characterized by the low and very elongated m1 with extremely developed talonid-metaconid complex (Fig. 5, 6); the presence of very thin crenulations on teeth (as these crenulation are very thin; they could lack in some cases as the case of c1 in HD); a combination of an elongated PM3 but with strong disto-lingual expansion; PM4 with a combination of a rudimental ectostyle and sinusoidal rather than straight buccal border; the very short c-pm3 diastema; relatively short neurocranium and moderately developed sagittal crest.

Fig. 6
figure 6

Comparative visualization of the ratio length/height of the m1 protoconid (including the talonid-metaconid complex) of Paramachaerodus transasiaticus sp. nov

Full size image

The distribution of the features which differentiate “Promegantereon” ogygia from Paramachaerodus diagnosed by Salesa et al. (2010) are shown in Table 4. Paramachaerodus transasiaticus sp. nov. has mixture of features showing an intermediary situation, thus the difference between “Promegantereon” and Paramachaerodus becomes invisible, as also shown in the cladogram C (Fig. 7). A gradual evolution of the features exists in time from “Promegantereon” ogygia through Pa. transasiaticus to Pa. orientalis and after to Pa. maximiliani, so it is hard to separate the taxa on generic level. It is interesting to note that the gradual evolution of crenulations started here in the same genus from forms without tooth crenulations until forms with well-marked crenulations. The same kind of development of the crenulatuons with time was shown recently for the different stages of the evolution of the genus Metailurus (Li 2015). We consider that the more logical taxonomic decision could be the inclusion of all the previously mentioned forms in one and the same genus, the name of which by priority must be Paramachaerodus Pilgrim. Thus, the genus contains four valid and successive species: Paramachaerodus ogygia, Paramachaerodus transasiaticus sp. nov., Paramachaerodus orientalis, and Pamachaerodus maximiliani. The talonid-metaconid complex of m1 of Paramachaerodus transasiaticus sp. nov. shows in general primitive features, retaining a large metaconid and elongated talonid. In the same time, this extremely enlargement of the talonid-metaconid complex can suggest a secondary evolution of this structure is relation to its cutting function. Such kind of secondary m1 talonid evolution was proposed for the lynx lower carnassial tooth (Werdelin 1981).

Table 4 Comparison of feature diagnosis after Salesa et al. (2010) with some details added by us
Full size table
Fig. 7
figure 7

Cladograms showing the phylogenetic relationships. a based on the matrix of six taxa and 18 characters (Supplementary 1); b based on the same matrix, but the 16th character is deleted; c based on the same matrix but unifying Hezheng and Hadjidimovo. All characters are treated as additive (ordered). The numbers indicate which characters support each clade

Full size image

Cladistic analysis

First, we prepared two analysis methods, using feature 16 (p4 mesio-buccal basal expansion in the crown, which shows some evolutionary trend, but must be strongly related to individual variation) and another lacking this feature. After our opinion, this feature does not have a high taxonomic value and could be present in the population of Paramachaerodus transasiaticus sp. nov. only in some of the individuals of this species [possibly the number of the individuals with this feature in Hezheng population (subspecies?) prevailed]. Using TNT, treating all characters as additive (ordered), a parsimonious tree of each method was obtained, whatever the search options used (Fig. 7). In cladogram A, the Hezheng sample is in a common clade as the sister group of both Paramachaerodus–Pa. orientalis and Pa. maximiliani, and diverge by Hadjidimovo sample by one apomorphy, which is feature 16. In cladogram B, Hezheng and Hadjidimovo are close together, which could be regarded as appurtenance to one and the same species. Feature 16 is the only difference (in fact a very small difference) between Hezheng and Hadjidimovo samples. It is probably an apomorphy (Salesa et al. 2010), but we don’t think only one feature like this could characterize another species; it could be in the case of Hezheng and Hadjidimovo due to individual or subspecific (geographic/geologic age) differences (see above). Thus, we unify Hezheng and Hadjidimovo (in the cladogram A and B both localities are coded separately) to the same new species Paramachaerodus transasiaticus sp. nov. as shown in cladogram C, using all 18 characters (Appendix; Table 5), concluding that there is no distinct apomorphy that can separate Promeganteron from Paramachaerodus.

Table 5 Data matrix used in the phylogenetic analysis
Full size table

Geological and geographical range of Paramachaerodus species. The most primitive form Paramachaerodus ogygia is known from Europe only and is in general a Vallesian form, but possibly reach the beginning of the early Turolian (Crevillente-2) (Salesa et al. 2010). Pa. orientalis is known from the second half of the early Turolian of upper Maragheh (Iran), but in Europe, it is a typical middle Turolian form (Salesa et al. 2010), and we hypothesize that it reached this continent most probably at the beginning of the MN12. The youngest and most evolved form Pa. maximiliani probably had a very wide distribution from Central Asia (China) to southwestern Europe (Spain) (Salesa et al. 2010). After the biochronology of Hadjidimovo (second half of the early Turolian) and Hezheng (cf. MN11) (see above), Paramachaerodus transasiaticus sp. nov. must be related to the early Turolian after the European land mammal ages. It was also largely spread, as Pa. maximiliani, from Central Asia to southern (southeastern) Europe. In southern Europe, it lived in the time span between the time of the existence of Pa. ogygia and the arrival of Pa. orientalis. Keeping in mind the development of tooth crenulation in Pa. transasiaticus sp. nov., we can assume that the Puente Minero (Spain, MN11) upper canines (Alcalá et al. 1991) are related to this species. In Asia, the new species probably coexisted (but geographically separated) with the more evolved Pa. orientalis. The strongly reduced, but high positioned metaconid of m1 of Pa. maximiliani give us the opportunity to suggest that the new, described here species could be an ancestral for Pa. maximiliani form. Pa. orientalis shows a more evolved condition than Pa. transasiaticus, but the very specialized talonid-metaconid complex of m1 (see above) of the latter do not suggest an evolution from this form in direction to Pa. orientalis.

The paleohabitat of Paramachaerodus transasiaticus. The paleoenvironment of Hadjidimovo-1 is discussed by Spassov (2002), who interpreted the landscape of this locality (based on the study of the rich mammal fauna), as a forest-steppe mosaic, resembling a forest-savanna mix. This opinion also was supported by Merceron et al. (2006) and Clavel et al. (2012), who both indicated that the Turolian paleohabitats of southwestern Bulgaria were dominated by open wooded landscapes and a relatively dry climate, with a continuum of habitats ranging from slightly wooded areas to relatively open landscapes. Other fossils found at the Shilei locality (Hezheng) are Ictitherium sp., Hyaenictitherium hyaenoides, Machairodus palanderi, Felis sp., Hipparion sp., Acerorhinus hezhengensis, Chilotherium wimani, Chleuastochoerus stehlini, Palaeotragus microdon, Gazella sp., Tapirus hezhengensis (Deng et al. 2008), and Struthio linxiaensis (Hou et al. 2005). All of those taxa are typical of the Hipparion Fauna found in the late Miocene of northern China, and they point to a hot and semi-arid savanna-like paleoenvironment (Deng 2004). This inferred paleoenvironmental difference likely was not an obstacle for Paramachaerodus because it is well known that extant large felids occupy a wide variety of environs. For example, the puma (Puma concolor) occupies the different habitats from Argentina north to Alaska (USA). The hunting behavior of this new Paramachaerodus species (Fig. 8) might have been similar to the behavior of the ambush predators such as the extant solitary pantherines (jaguars or leopards), as indicated for Pa. ogygia by Salesa et al. (2006).

Fig. 8
figure 8

Life appearance of Paramachaerodus transasiaticus sp. nov. Reconstruction of the head with the position of the skull (after the skull—IVPP (V) 20106 and the mandibles—IVPP (V) 20107 and HD 9227) (reconstruction: Velizar Simeonovski, Field Museum of Natural History, Chicago & National Museum of Natural History, Sofia)

Full size image

Conclusions

The analysis shows that Hezheng and Hadjidimovo machairodontine samples are almost identical. They show only some small differences which is reasonable to consider as individual, sexual, and geographic/subspecific differences, thus it is logical to assign all these geographically separated fossils as remains of one and the same species.

Both samples from Hezheng and Hadjidimovo represent a new species of the “Promegantereon”–Paramachaerodus lineage–Paramachaerodus transasiaticus sp. nov., characterized by: the presence of very thin crenulations on teeth (as these crenulation are very thin, they could lack in some cases as the case of c1 in HD); the low and very elongated m1 with extremely developed talonid-metaconid complex; a combination of an elongated PM3, but with strong disto-lingual expansion; PM4 with a combination of a rudimental ectostyle and sinusoidal rather than straight buccal border; the very short c-pm3 diastema; relatively short neurocranium.

The combination of all this mixture of features between “Promegantereon” and Paramachaerodus, as well as the age of the Hezheng and Hadjidimovo (early Turolian, after the European Land Mammal Ages) put Paramachaerodus transasiaticus sp. nov in intermediary position between the representatives of Promegantereon (sensu Salesa et al. 2010) and Paramachaerodus. Thus, it is hard to separate “Promegantereon” ogygia in generic level from the representatives of Paramachaerodus and all species of “Promegantereon”–Paramachaerodus lineage could be regarded as species of one and the same genus—Paramachaerodus Pilgrim.