We are sincerely grateful to Dr. S.V. Kruskop and Dr. V. Lebedev (Zoological Museum of Moscow State University, Moscow, Russia) for an opportunity and kind assistance in our work with the collection of extant Chiroptera housed in their museum. Special gratitude goes to Dr. Lutz Christian Maul (Senckenberg Research Institute and Natural History Museum Frankfurt, Germany) and to Prof. Dr. M. van Kolfschoten (Leiden University, Netherlands) for the careful reviewing of the manuscript and useful comments. We are also grateful to Dr. G. S. Rautian (PIN, Russia) and Mr. Colin Matchett (USA) for helpful corrections of the English text. Dr. Nicole Limondin-Lozouet (Laboratoire de Géographie Physique, Environnements quaternaires et actuels, Meudon, France) kindly translated several sections into French. This study was supported by the Russian Foundation for Basic Research (project no. 13-06-12015-OFI-m-2013), a grant from the Russian Science Foundation (project No. 14-50-00036) and the project of the Department of General Biology of the Russian Academy of Sciences “Historical Dynamics of Bioresource Mammal Species in the Pleistocene on a Background of Climatic Changes and Anthropogenic Effect.”

1 - Introduction

1The Balkan Peninsula is a unique ecological and biogeographical area in Europe because of its remarkably high biodiversity. The abundance and diversity of plant and animal taxa, ecosystems and landscapes are a result of the long and complex paleogeographic history of this region (Rage & Rocek, 2003; Parmakelis et al., 2006; Kuhlemann, 2007) as well as of particular climatic, geomorphological, petrographic, orographic, and hydrographic features (Savić, 2008; Murienne et al., 2009). Diversity of biotopes and a large number of refuges throughout the territory also explain an exceptionally high number of relict and endemo-relict species (paleostenoendemics) and communities dating back in some cases to the Pleistocene or even to the Tertiary. It is generally accepted that among the three European peninsulas - Iberian, Italian and Balkan - that have acted as refuges during the last glaciations (Petit et al., 2003; Tzedakis, 2004), the Balkan Peninsula is a center of biodiversity in Europe (Gaston & David, 1994). Central and Northern Europe appear to have been repeatedly populated by expansions from Balkan refuges during the Pleistocene interglacials (Storch, 2004; Savić, 2008).

2The Pleistocene was characterized by important changes of climatic conditions and human distribution throughout the Eurasian continent, reflecting global paleoecological transformation. Studies of fossil land mammal communities contribute to our knowledge of paleoenvironmental conditions and their effect on the global spread of hominids (Kahlke et al., 2011). Such studies are well-known from some Balkan regions, in particular, from Croatia and Slovenia (Jánossy, 1986; Aguilar et al., 1998, Kostopoulos et al., 2002; Forstén & Dmitrijevic, 2004; Etren et al., 2005; Bogićević & Nenadić, 2008; see also references therein). However, relevant data from present-day Serbia and Montenegro are still poorly known. The rich fossil material from Trlica in Montenegro allows starting filling this gap. The presence of small mammal remains from Trlica was already established in preliminary papers (Dimitrijević, 1990; Crégut-Bonnoure & Dimitrijević, 2006; Argant & Dimitrijević, 2007), but the first detailed study of the rodents from this site was performed by Bogićević & Nenadić (2008). The composition of the rodent fauna was used for stratigraphic considerations suggesting a lower Early Pleistocene age.

3Although the Plio-Pleistocene chiropteran records from Europe have steadily increased during the last decades, our knowledge about fossil bats of the western part of the Balkans is still insufficient. Pliocene, Pleistocene and Holocene bats are known from the Balkans, particularly from Hungary (e.g., Kormos, 1934; Kretzoi, 1956; Topál, 1974, 1979, 1983, 1989), Bulgaria (e.g., Popov, 2000, 2004), Slovenia (Aguilar et al., 1998; Sigé et al., 2003; Toškan & Kryštufek, 2007; Toškan, 2009; Obuch, 2012), Romania (Kormos, 1930a, b; Topál, 1963a, b), Serbia (Dmitrijević, 1991, 1996, 1997), Croatia (Pičuljan, 2012), Greece (Kretzoi, 1977; Chatzpoulou et al., 2001) and Turkey (Lindenau, 2005; Suata Alpaslan, 2011). However, in many sites bat remains have been identified only to order rank (e.g., Pazonyi & Kordos, 2004; Pičuljan 2012), from others only lists of species were mentioned without morphological descriptions (e.g., Jánossy et al., 1968; Jánossy, 1976). Nevertheless, in some studies, the Late Pleistocene and Holocene bat records were used for paleoenvironmental reconstructions (Dmitrijević, 1997; Popov 2000, 2004; Sigé et al., 2003; Lindenau, 2005).

4Fossil bat remains are extremely rare in Montenegro. Some fragments of limb bones of bats have been mentioned from the Upper Pleistocene Mališina stijen rock-shelter (village of Ljutići near Pljevlja, northern Montenegro), but they were identified as Chiroptera indet. (Bogićević & Dmitrijević, 2004). In the present study, we document for the first time bat remains from the Paleolithic site of Trlica in Montenegro and provide new data on rodents and other small mammals from this locality. The results are used for biostratigraphical and paleoecological conclusions.

2 - The trlica locality

5Trlica is a hill of Triassic limestone situated north of the city of Pljevlja near the borders between Montenegro, Serbia and Bosnia-Herzegovina (Forstén & Dimitrijević, 2004: fig. 1, p. 56). A karstic cavern in the limestone is filled with clastic sediments enclosing fossil bones and teeth. It was accidentally opened in the 1960s during the construction of the road Pljevlja-Bjelo Polje (fig. 1A; Bogićević & Nenadić, 2008). The deposits containing mammal bones and teeth have an intense red color (fig. 1B) clearly different from the surrounding white Triassic limestones. The cross section of the Quaternary sediments, excavated in 1988, 1990 (Dimitrijević, 1990) and in 2001 (Crégut-Bonnoure & Dimitrijević, 2006; Bogićević & Nenadić, 2008), was about 3 meters high. Five beds of Pleistocene sediments have previously been distinguished in the cave profile (Crégut-Bonnouire & Dimitrijević, 2006). Remains of large herbivores, particularly ruminants, were most abundant in all layers, while carnivores and rodents were significantly rarer (Bogićević & Nenadić, 2008). The mammal fauna includes Dinaromys dalmatinus (Kormos, 1931), Allophaiomys cf. pliocaenicus Kormos, 1933, Mimomys ex gr. reidi Hinton, 1910, Glis sackdilligensis Heller, 1930, Hystrix cf. refossa Gervais, 1852, Canis etruscus Major, 1877, Canis (Xenocyon) falconeri (Forsyth Major, 1877), Vulpes sp., Ursus etruscus Cuvier, 1823, Gulo schlosseri Kormos, 1914, Mustelidae gen. et sp. indet., Homotherium crenatidens (Fabrini, 1890), Panthera cf. gombaszoegensis (Kretzoi, 1983), Lynx sp., Pachycrocuta brevirostris (Aymard, 1846), Palaeoloxodon antiquus (Falconer & Cautley, 1847), Stephanorhinus cf. hundsheimensis (Toula, 1902), Equus stenonis Cocchi 1867, Equus cf. major Depéret, 1983 ex. Boule, Alces cf. carnutorum (Laugel, 1862), Eucladoceros giuli Kahlke 1997, Cervus elaphus Linnaeus, 1758, Megalovis balcanicus Crégut-Bonnoure & Dimitrijević, 2006, Soergelia intermedia Crégut-Bonnoure & Dimitrijević, 2006, Bison (Eobison) sp. nov. (Dimitrijević, 1990; Codrea & Dimitrijević, 1997; Forstén & Dimitrijević, 2004; Crégut-Bonnoure & Dimitrijević, 2006; Bogićević & Nenadić, 2008). The composition of the rodent fauna and the presence of Stephanorhinus cf. hundsheimensis and a small-sized Bison certainly indicate a stratigraphic age of the Trlica fauna correlating with biozones 20-22 (Guérin, 1980) and MNQ2 or MNQ3 (late Early Pleistocene, after Agusti et al., 1987; fig. 2) (Codrea & Dimitrijević, 1997; Bogićević & Nenadić, 2008). The discovery of Hystrix is indicative of the Mediterranean type of climate and environment (Maul, 1994).

6Archeological excavations at Trlica site in 2009 led to the discovery of the first Paleolithic artifact (Derevianko et al., 2010). Recently, new data were published on the facies composition of the deposits, their mineralogical and petrographic structure, and their palynological and paleontological data (Derevianko et al., 2012a, b). According to new archeological excavations conducted in 2010-2012, the deposits of the Trlica site are about 5 m thick and include 12 beds (see fig. 1B). The upper layers (beds 1-4) are about 1.0 - 1.40 m thick ranging in composition from loam to clayey loam, which were accumulated mostly in subaerial, weakly hydrated conditions. They were formed during rock weathering processes and probably accumulated in a paleokarstic site by redeposition of lateritic soil via vertical fissures or small karstic channels. This model of sediment genesis is supported by the lack of stratification, the absence of sandy lenses and the rarity of fossil bones within the upper layers.

7Beds 5-6 are about 1.0 - 1.50 m thick and apparently of colluvial origin. Their lower part was probably formed as colluvial fan deposits. The layers yielded bones and teeth of small and large mammals as well as coprolites of hyenas and fossil burrows of large animals, most probably of badger.

8Bed 7 is 0.15 - 0.20 m thick and displays a strong travertine cementation of deposits, which is indicative of changes in the sedimentary environment and probably a significant break in sedimentation between beds 5 - 6 and 8 - 10. Bed 7 contains bones of large mammals and considerably fewer bones and teeth of small mammals.

9Beds 8 - 10 are 1.0 - 1.30 m thick and were formed as colluvial deposits. A high density of bones of large and small mammals and coprolites of hyenas is characteristic of bed 10, which is 0.1 0.60 m thick and contains an erosive break rich in bone fragments of large herbivores and coprolites of hyenas.

10Beds 11 - 12 were formed mostly in subaquatic conditions and composed of loam, well-sorted loamy sands and sands with gravel inclusions, granules and small pebbles. In bed 11, which is about 0.60 m thick, bones and teeth of both large and small mammals were found. Bed 12 lacks fossils.

11Thus, the cave deposits belong to different facies, they are characterized by variable provenance and several, possibly long, sedimentation gaps. Most of the infill was brought from outside, while the proportion of deposits of cave origin is relatively low.

3 – Material and methods

12All fossil small mammals (a total of 362 specimens, tab. 1) come from beds 3 - 11. The fossil bat remains (8 specimens) originate exclusively from bed 10.

13The bat tribe taxonomy follows Simmons (2005). The dental terminology of bats follows Miller (1907) and that of the upper canine follows Rosina (2015: fig. 1, p. 2) with some modification because in the upper canine of Rhinolophus the lingual crest is absent and the lingual concavity extends between the anterior ridge and posterior crest. Both the lengths of single teeth and of tooth rows are measured as the maximum distance between the posterior and anterior crown edges of respective teeth. The widths of the upper molars are the maximum distance between the lingual and buccal crown edges measured in parallel to the mesial side (Sigé, 1968: fig. 4, p. 72). The following measurements were taken: length (L) × width (W) × height (H) of C sup., p4 (in the latter case - the maximum crown width); length × width of the molar trigonid (Wtr) × width of the molar talonid (Wtl) of m1, 2, 3; the lengths of the tooth row: Lm1-2, Lm1-3; Hmdm1 is the height of the mandibular body measured on the lingual side below m1; Hmdm3 is the height of the mandibular body measured on the lingual side below m3. The specimens were measured under a binocular microscope (MBS-10) with ocular micrometer. All measurements are given in millimeters with a precision of 0.01 mm. Photographs were taken by a scanning electron microscope located at the Borissiak Paleontological Institute of the Russian Academy of Sciences (PIN, Moscow, Russia). The fossil material from Trlica, listed in the current paper, is presently housed in the PIN (coll. TRL/Ch; Moscow, Russia). The osteological collections of extant Chiroptera housed in the Zoological Museum of Moscow State University (ZMMU) were used for comparison of the fossil specimens with living bats.

4 - Results

4.1 - Rodents and other small mammals

14A total of 354 fossil mammal specimens (except from bats) were examined. The distribution of specimens over the Trlica sequence is shown in tab. 1. Most fossils come from beds 5 - 11, whereas beds 3 - 4 yielded only a few fragments of rodent incisors and several rootless cheek teeth of Microtus and Lagurus.

15Bed 5 has yielded 176 small mammal specimens dominated by vole remains, 19.3 % of which are rootless and 9.6 % rooted molars. The vole with rooted molars Mimomys ex gr. savini (Newton), an archaic vole with rootless molars Terricola hintoni (Kretzoi) and the more advanced vole T. arvaloides (Hinton) as well as the steppe lemmings (Lagurini) with molars lacking cement, are common in bed 5. Among the lemming remains, a lower m1 of Lagurus ex gr. transiens Janossy was discovered. Also a single rootless molar of Microtus nivalinus Hinton with crown cement, like in Terricola, has been found here. Moreover, the dormouse Glis sackdillingensis is abundant. Bed 5 also yielded fossil remains of both the large-sized hamster Cricetus praeglacialis Schaub, 1930 and the small-sized hamster of the genus Cricetulus, together with a single tooth fragment of Hystrix sp. Bed 6, which is relatively thin and contained few fossils, has yielded a single tooth of G. sackdillingensis and an upper third molar of Mimomys sp. (possibly M. ex gr. pliocaenicus F. Major) with roots, cement and small enamel island. Bed 7 probably corresponds to a significant break in sedimentation and includes only a few tooth fragments of Hystrix sp. The fossil record of small mammals in bed 8 is also sparse and includes only two molar fragments of Mimomys sp.

16The small mammal assemblage from the lower layers (beds 9 - 11) is appreciably different. Bed 9 has yielded two rooted molars of Mimomys sp. (ex gr. pliocaenicus), some fragments of rooted molars of Mimomys sp. and two teeth of voles definitely belonging to the tribe Microtini. Bed 10 contains the richest fauna of small mammals (tab. 1). Remains of voles with rooted molars, including Mimomys pliocaenicus, are most abundant, representing 32.6 %  of the assemblage. Vole species with rootless molars, including the steppe lemmings Lagurini, represent 9.6 %. Remains of Hystrix refossa and the hare Pliolagus (=Hypolagus?) are rather abundant in bed 10: their proportion being 3.7 %. The single bone fragments of Martes sp. and a small-sized Vulpes sp. are also found here. Bed 11 yielded only 14 tooth fragments of small mammals, among which M. pliocaenicus is most abundant. A single tooth fragment of a microtine rodent and several specimens of Hystrix refossa are also found here.

4.2 - Chiroptera

17Fossil bat remains are very rare at the site of Trlica and compose 2.81 % of the whole assemblage (tab. 1). They are represented by a single dentary fragment, two isolated teeth and several unidentified postcranial fragments (three bones of the autopodium, femur and tibia). Only two bat taxa, Rhinolophus cf. macrorhinus anomalidens (family Rhinolophidae) and Myotis blythii longocaninus (family Vespertilionidae), were identified from Trlica deposits.

Systematic palaeontology

Order Chiroptera Blumenbach, 1779

Superfamily Rhinolophoidea Weber, 1928

Family Rhinolophidae Gray, 1825

Genus Rhinolophus Lacépède, 1799

Rhinolophus macrorhinus Topál, 1963

Rhinolophus cf. macrorhinus anomalidens Topál, 1979 (fig. 3A, B)

18Material (for measurements, see tab 2): TRL/Ch-1, left C sup., the tooth has a heavily worn crown, TRL/Ch-2, left C sup.

19Description: The upper canines are very massive with strongly distally curved apices. The crowns are semilunar in occlusal view, with a well-developed cingulum and have a flat lingual concavity and cingulum depression at the posterolingual margins (fig. 3A, B).

20Comparison: The semilunar shape of the canines and their strongly distally curved apices indicate that the specimens belong to the genus Rhinolophus. They differ from Hipposideros in having a more roundish outline of the crown, which is not compressed buccolingually and has a broad talon on the posterolingual side surrounded by a weakly developed cingulum (fig. 3A, B). In contrast, the posterolingual talon of C sup. in Hipposideros is surrounded by an elevated well-developed cingulum (e.g., Álvarez Sierra et al., 2006: fig. 6 a, c, p. 419). Moreover, the anterior ridge of the upper canine of Rhinolophus is more curved and hooked than in Hipposideros (fig. 3A, B; Martinez, 2010: Figs. 3.8 and 3.9, pp. 58-59 and fig. 3.49, p. 135; Sevilla, 1990: Figs. 7-1 b, c, p. 177). The buccal crest of the upper canine in Hipposideros is poorly developed (e.g., Sevilla, 1990: Figs. 7-1 a, b, p. 177; Figs. 9-5 a, b, p. 178; Álvarez Sierra et al., 2006: fig. 6 c, p. 419); contrariwise, the anterobuccal ridge and buccal crest of Rhinolophus are well developed (fig. 3A, B; Wołoszyn, 1987: fig. 6, p. 236; Popov, 2004: fig. 7 C, p. 414). The fossil canines share the distinctive morphology with extant R. ferrumequinum (e.g., specimen ZMMU S-83976, Ukraine) and R. fumigatus (e.g., specimen ZMMU S-168839, Gvineya). However, the fossil canines from Trlica are larger and longer than those in comparative material of living R. ferrumequinum from Europe and R. fumigatus from Gvineya (index LC/WC × 100 % is equal to, or less than, 125, see tab. 2) and also differ from the latter in having less developed depression on the posterolingual crown margin.

21Fossil upper canines from Trlica also differ from Late Pliocene R. kowalskii Topál, 1979 (Podlesice, MN14, Poland), R. wenzensis Wołoszyn, 1988 (Podlesice, MN14, Poland; Muselievo, MN15b, Bulgaria) and R. estramontis Topál, 1979 (Osztramos 1f, MN 14, Hungary) in the larger size (tab. 2; Popov, 2004: tab. 1, p. 413). Furthermore, they both have an evident posterolingual notch of the lingual concavity and a small depression on the posterobuccal face of the crown (fig. 3A, B), which means as well an expected extrusion of P2 buccally from the tooth-row. The last morphological feature is absent in Pliocene species of Rhinolophus (Wołoszyn, 1987: fig. 6, p. 236 and fig. 10, p. 250; Topál, 1979: Pl. 2, figs. 1-7, p. 70; Pl. 4, figs. 1-2, p. 71; Popov, 2004: fig. 7, p. 414) belonging to the “ferrumequinum” group (Topál, 1979). The above-discussed morphological features evidently appear for the first time in the upper canine of Early Pleistocene R. macrorhinus anomalidens from Osztramos 8 (Hungary; Topál, 1979: Pl. 5, fig. 1, p. 73). The upper canine of the nominative taxon R. macrorhinus from Beremend Loc. 4 (MN16, Hungary) has a visible depression on the posterobuccal face of the crown (Topál, 1963b: fig. 1, p. 221), but the posterolingual notch of the lingual concavity is missing (Topál, 1979: p. 81). Since Middle Pleistocene time, the upper canines of fossil species of Rhinolophus of the “ferrumequinum” group acquired the complete set of morphological features of extant R. ferrumequinum (Topál, 1979: p. 64). From the Middle Pleistocene R. f. tarkoensis (Tarkő, Hungary) and R. f. cf. topáli (Uppony 1 L, Hungary), which are only somewhat larger than the extant R. ferrumequinum (Topál, 1979: p. 92), the fossil canines from Trlica differ in the larger size and the more elongated tooth crown (index LC/WC × 100 % is less than, or equal to, 125; tab. 2; also see Topál, 1979: fig. 5, p. 95). In general, fossil canines from Trlica are more similar to R. m. anomalidens; however, the specimens were identified as R. cf. m. anomalidens because of their scarcity and poor preservation.

22Remarks. Some finds of R. cf. macrorhinus, including the isolated upper canines, are known from the Polish Early Pleistocene sites Kadzielnia 1 (MN17/Q1, Wołoszyn, 1987), Kamyk (Q1) and Kielniki (Q1/Q2). In the Early Pleistocene Hungarian site of Kövesvárad, R. cf. m. anomalidens was recorded, in particular, with upper canines (Topál, 1963a, b; 1976). Measurements of all these fossil forms are also included in tab. 2.

Family Vespertilionidae Gray, 1821

Tribe Myotini Tate, 1941

Genus Myotis Kaup, 1829

Myotis blythii (Tomes, 1857)

Myotis blythii longocaninus Popov, 2004 (fig. 3C)

23Material (for measurements, see tab. 3): TRL/Ch-3, right mandible with m1-m3 and alveoli of p4.

24Description: The dentary is broken anteriorly in the area of premolars and, posteriorly, in line with the ascending ramus. However, the alveoli of p4, which are rounded and similar in size, are preserved. The lower molars are myotodont. The length ratios of the lower molars are m1 ≈ m2 > m3. In m1, the angle between the paraconid and metaconid is wider and the paralophid is more curved than in m2, while the trigonid of m2 is wider. The para- and metalophids of the lower molars are somewhat curved and the metalophids are convex posteriorly, so that the tips of the metaconids are directed anteriorly. The talonid of m3 is reduced in size and has a distinct hypoconulid.

25Comparison: On the basis of the slightly curved para- and metalophids and anteriorly directed metaconids of the lower molars, the fossil specimen from Trlica is determined as Myotis. Moreover, the roundish and not reduced alveoli of p4 suggest that the tooth crown was elongated as is also typical of Myotis.

26Fossil Myotis from Trlica is larger than most of the Late Miocene, Pliocene and Early Pleistocene Myotis species from Europe (tab. 3). It differs from the Early Pleistocene M. ghardalamensis Storch, 1974 from the Ghar Dalam cave of Malta (Storch, 1974) in the slightly more slender dentition, the more elongated m3 and narrower m2 and m3. The fossil specimen from Trlica is larger than the extant M. blythii (Tomes, 1857) from Europe, but smaller than the living M. myotis (Borkhausen, 1797). Moreover, it differs from both these large Myotis species in the less reduced talonid of m3, similar to that of extant M. bech-steinii Kuhl, 1817, which is considerably smaller in size (tab. 3). Morphologically it is most similar to the Pliocene M. blythii longocaninus from the Muselievo site of Bulgaria (MN15 b, Popov, 2004) and M. cf. blythii from Osztramos Loc. 9 of Hungary (MN15, Topál, 1983).

27Remarks. The fossil subspecies M. b. longocaninus is very similar to Malta’s M. ghardalamensis (Storch, 1974), though Popov (2004: p. 426) found that M. b. longocaninus “differs from M. ghardalamensis Storch, 1974 in having more elongated p4 and m1-m2, and less reduced M3”. However, the measurements of these two species are very close (tab. 3, see also Popov, 2004: Table 4, p. 427) and future more detailed comparisons are apparently needed to clarify their taxonomic relation. This is beyond the scope of this paper. In any case, M. b. longocaninus is apparently close phylogenetically to extant M. blythii; however, it is more primitive in having a less reduced talonid of m3, less reduced first premolars and more gracile molars (Popov, 2004: Table 4, p. 427). According to Popov (2004), the nominative taxon M. b. longocaninus is also close morphologically to M. cf. blythii from Middle Pliocene sites Osztramos 9 and 13 (MN15b, Hungary). The latter form is morphologically more advanced and, thus, closer to extant M. blythii. The dimensions of all of these Myotis species and fossil Myotis from Trlica are included in tab. 3.

5 - Discussion

5.1 - Faunal and biostratigraphic consideration

28The Balkans are usually regarded as an important refuge for European mammalian faunas during the Pleistocene glacial episodes (Storch, 2004). It is generally accepted that the mountains of Serbia and Montenegro are one of the most important Ice Age refuge regions of Europe, where species that were much more widely distributed at lower elevations during periods of colder climate could still find a suitable habitat when temperature ameliorated (Schulze, 2002). Moreover, the territory of Serbia and Montenegro is considered as an important center of endemic speciation not only of the Balkan Peninsula, but also of the whole eastern part of the Mediterranean Region (Savić, 2008). This paleogeographic history caused the peculiarities of both modern and fossil small mammal faunas of this region.

29According to Bogićević & Nenadić (2008) the small mammal assemblage from Trlica includes Dinaromys dalmatinus, Allophaiomys cf. pliocaenicus, Mimomys ex gr. reidi-pusillus, Glis sackdilligensis, and Hystrix cf. refossa. Our study enlarges the previous small mammal lists of Trlica (Dimitrijević, 1990; Crégut-Bonnoure & Dimitrijević, 2006; Argant & Dimitrijević, 2007) by the following taxa: Cricetus praeglacialis, Cricetulus sp., Mimomys pliocaenicus, M. ex gr. savini, Terricola sp., Lagurus ex gr. transiens, Microtus nivaloides, Pliolagus sp. and Martes sp. (tab. 1). At the same time, our material does not include D. dalmatinus, which was recorded in both the upper (beds Ia and Ib) and lower parts (bed IV) of the Trlica site (Bogićević & Nenadić, 2008), corresponding approximately to beds 5-6 and 10, respectively, of our geological section. The vole D. bogdanovi Martino and Martino, 1922 is the only living member of the Tertiary genus Dinaromys and probably the only surviving member of the Pliomys lineage dating back to the Ruscinian (Pevzner et al., 2001). The modern geographical range of this paleoendemic rodent is rather small and covers the mountainous landscape (mostly in the Dinaric Alps) of the western Balkans, where it is closely associated with underground rocky habitats, especially in accumulations of rocks and boulders, caves and crevices in rocky substrate (Gromov & Polyakov, 1977; Shenbrot & Krasnov, 2005; see also references in Kryštufek & Bužan, 2008). As follows from the paleontological data, the vole D. bogdanovi evolved in the Upper Pliocene along the northeastern Adriatic coast and western Balkans (Kowalski, 2001). Its Pleistocene range exceeded the recent one, at least in the northwest of the Balkans and its extinction progressed southwards during the Early Holocene (Kryštufek & Bužan, 2008). Species closely related to extant D. bogdanovi are known from northeastern Italy (since the Late Pliocene), the western Balkans (since the Early Pleistocene), Hungarian Carpathians (since the Middle Pleistocene) and from the Ukraine (Early Pleistocene; Kretzoi, 1955; Kowalski, 2001, see also references in Kryštufek, 2004). Given that the ancient vole genus Dinaromys evolved mostly in the Dinaric Alps, this vole should be considered an expected component of Trlica taphocenosis. Probably, lack of Dinaromys in our materials is caused by taphonomical reasons or by a distinct rarity of the remains of this vole in Trlica tapho-cenosis. Bogićević & Nenadić (2008) additionally recorded Glis sackdilligensis in bed Ib which is in agreement with our results (tab. 1).

30Early studies on the faunal composition of large herbivores and small mammals have correlated all beds of the Trlica locality with biozones 20 - 22 (Guérin, 1980) and MQ2 or MQ3 (Agusti et al., 1987; Codrea & Dimitrijević, 1997; Bogićević & Nenadić, 2008). However, our new studies of small mammals indicate that deposits of Trlica belong at least to two facies which differ in age and biostratigraphical position. The composition of the rodent fauna (see tab. 1) confirms that the lower layers (beds 9 - 11) are significantly older than the upper layers.

31In Western and Central Europe Mimomys pliocaenicus with rooted molars has been recorded in the Villafranchian at St. Vallier (lower part of MNQ17) and at Olivola (lower part of MNQ18), afterwards it has no longer been recorded during Jaramillo event. Mimomys pusillus (Mehely) existed from Farneta Faunal Unit (the lower part of MNQ18) to Pirro Faunal Unit (upper part of MNQ18) (Gliozzi et al., 1997). In the Russian Plain, M. pliocaenicus is known from MN17 to the upper part of MNQ18, and M. pusillus is recorded from the upper part of MNQ18 to the lower part of MNQ20 (see fig. 2; Agadjanian, 2002). At the Villanian-Biharian boundary (sensu Fejfar & Heinrich, 1983) after the paleomagnetic Olduvai episode in Central Europe, the biostratigraphical zone of M. pliocaenicus was replaced by the M. savini-M. pusillus zone (Agadjanian, 2002, 2009). During the Cromerian, M. pusillus gradually disappeared and was replaced by M. savini (Hinton, 1910). In Eastern Europe, M. reidi and M. pusillus, along with M. pliocaenicus and Allophaiomys, occurred in the Odessaian Faunal assemblage, and, later, they occurred with M. savini and A. pliocaenicus (fig. 2) in the Tamanian Faunal assemblage of the stratigraphical scale of continental deposits of southern Russia (Gromov, 1948). In the Tiraspolian Faunal assemblage, M. savini and M. pusillus existed from the late Early Pleistocene until the beginning of the Middle Pleistocene (Agadjanian, 1992, 2009; Markova, 2007). Moreover, during the Tiraspol Time (approximately correlated with the Cromerian), the relative proportion of these two vole species gradually changed, so that the proportion of M. pusillus steadily decreased ().

32Thus, based on the co-occurrence of M. ex gr. pliocaenicus and A. pliocaenicus in the lower layers (beds 9 - 11) of the Trlica site (see tab. 1), the faunal assemblage from these beds is dated to the second half of the Early Pleistocene (Late Villafranchian; see fig. 2). It is older than the Jaramillo event and therefore older than 1 Ma. Besides, coexistence of M. pliocaenicus and Allophaiomys in bed 10 is typical of the Late Villanium and Early Biharium time, and correlates the age of the bed with MNQ17-MNQ18 (according to Guérin, 2007). Moreover, fossil remains of Pliolagus were also recorded in this layer, which is typical for Early Pleistocene faunal assemblages of Europe (Gureev, 1964) and occurs in most of the Pliocene and Pleistocene sites of the Don and Volga river basins (Sukhov, 1970, 1975; Agadjanian, 2009). A remarkable fact is the relatively high number of the porcupine Hystrix refossa records in the lower part of the deposits. It represents 21 % in bed 11 and 6.7 % in bed 10 (tab. 1). This large porcupine with hypsodont (high-crown) cheek teeth was described for the first time from the Late Pliocene site of Les Etouaires (MN16, France; Chaline, 1972). Recently, it has been recorded in several Pliocene and Early Pleistocene sites across Europe (e.g., Weers, 1994, 2005; Weers & Rook, 2003). On the other hand, occurrences of Microtini and Lagurini in bed 10 indicate that its lower age limit is the paleomagnetic Olduvai episode (Olduvai Subchron, 1.95 - 1.77 Ma; fig. 2) of the Matuyama Chronozone, because voles with rootless molars did not exist before this boundary (Agadjanian, 2009). Thus, based on the rodent assemblage, the lower part of the Trlica section is dated from 1.8 to 1.0 Ma (fig. 2). The study of large mammals from Trlica performed by Vislobokova & Agadjanian (2015) corroborates this conclusion. Similar faunal assemblages are known in Italy from the sites Monte la Mesa, Pietrafitta (Farneta), Cava Pirro, and Soave Cava Sud (Masini & Sala, 2007) and in Southern Spain from the sites Venta Micena and Orce 7 of the basin of Guadix-Baza, and site Gruta 1, and also in the North of the Iberian Peninsula from the site Sima del Elefante (Agustí et al., 2010; Cuenca-Bescos et al., 2010). The faunas with Allophaiomys are also common in Central and Eastern Europe (fig. 2, see also Maul & Markova, 2007: Agadjanian, 2009; Markova, 2014)

33All bats found in bed 10 at Trlica are known only from the Pliocene or Pleistocene sites of Europe. Myotis blythii longocaninus was described for the first time from the Middle Pliocene Muselievo site of Bulgaria (MN15b, Popov, 2004). The specimen of M. b. longocaninus from Trlica is also similar morphologically to M. cf. blythii from Osztramos Loc. 9 site (MN15, Hungary). This could indicate a maximum age of the Middle Pliocene for the Trlica site, but, according to the other arguments (see below); the site is younger in age. Thus, so far the record of M. b. longocaninus in Trlica is the youngest for the taxon.

34Two upper canines from bed 10 at Trlica belong to Rhinolophus cf. macrorhinus anomalidens, which is also known from the Early Pleistocene Osztramos 8 site of Hungary (Topál, 1979) and belongs to the “ferrumequinum” group. According to the evolutionary model proposed by Topál (1979), in the Middle Pleistocene, the upper canines of fossil rhinolophids of the “ferrumequinum” group have acquired the modern morphology of extant R. ferrumequinum (Topál, 1979: p. 64). Essentially, the Early and Middle Pleistocene species were considerably larger than the descendent modern taxon. If Topáls’ model is correct, the presence of R. cf. m. anomalidens in the lower part of the Trlica site confirms the Early Pleistocene age of the faunal assemblage including it, because R. cf. m. anomalidens from Trlica is considerably larger than both Middle Pleistocene R. f. tarkoensis and especially extant R. ferrumequinum (tab. 2). The facies composition of the Trlica site and, primarily, the significant breaks in sedimentation suggest that there was a time span about 400 ky between the formation of the taphocenoses of beds 5 - 6 and 9 - 11.

35Glis sackdillingensis, Terricola hintoni, Mimomys ex gr. savini and Lagurus ex gr. transiens recorded in bed 5 are typical for the first half of the Middle Pleistocene (Janossy, 1969; Sutcliffe & Kowalski, 1976; Chaline, 1977a, b; Stuart, 1977; Preece & Parfitt, 2000; or the age of the Tiraspol Faunistic Complex, according to Aleksandrova, 1971; Agadjanian, 1977, 2009). The co-occurrence of G. sackdillingensis and M. pusillus in bed 6 also indicates that the age of the faunal assemblage from the upper unit of Trlica site dates to the first half of the Middle Pleistocene. It might be that beds 1-4 should be dated to the second part of the Middle Pleistocene or the beginning of the Late Pleistocene, as it is supported by the absence of hiatuses between beds 1-6. The presence of advanced voles (Microtini tribe, e.g., Microtus arvalinus) in beds 3 - 4 (tab. 1) additionally corroborates this dating.

5.2. - Zoogeographical peculiarities of the mammalian fauna of Montenegro with an emphasis on the Trlica faunal assemblage

36Montenegro shows a high biological diversity related to the small territory of the country, which is attributable to its geographical position, heterogenic distribution of habitats, topographic variations, geological history and climatic conditions (Stevanović et al., 1996; Kapa, 2010) and is included in the Mediterranean biodiversity hotspot (e.g., Médail & Quézel, 1999). There is a zonation of the flora and fauna from the cold mountainous north and south to the warm Mediterranean coast in the west part of the country. Thus the territory can be divided into two main bio-geographical regions, Mediterranean and Alpine. There is a considerable endemism in the Montenegro flora and fauna with dominant Central European, Illirian, Alpine and Mediterranean elements (Kapa 2010). Mountain forests occupy 54 % of the territory, making Montenegro one of the most forested countries in Europe with the highest mammal diversity in the mountainous and forested north part of the country.

37The extant mammalian fauna of Montenegro is rich (65 mammal species are recorded, Kryštufek, 2004) and includes terrestrial mammals, such as carnivores (e.g., Canis lupus L., Ursus arctos L., Vulpes vulpes L., Lynx lynx L., Lutra lutra L.), ungulates (e.g., Sus scrofa L., Cervus elaphus L., Capreolus capreolus L., Rupicapra rupicapra L.), rodents (about 26 species), insectivores (about 10 species) and bats. For the most complete lists of rodents and insectivores, see Kryštufek (2004: p. 107-106; and also Tvrtković et al., 1985; Kryštufek, 1991; Petrov, 1992).

38The modern bat fauna of Montenegro is very rich and includes 28 species: Rhinolophus ferrumequinum (Schreber, 1774), R. hipposideros Bechstein, 1800, R. euryale Blasius, 1853, R. blasii Peters, 1866, Myotis myotis, M. blythii, M. nattereri Kuhl, 1817, M. emarginatus E. Geoffroy, 1806, M. mystacinus Kuhl, 1817, M. alcathoe von Helversen and Heller, 2001, M. brandtii Eversmann, 1845, M. daubentonii Kuhl, 1817, M. capaccinii Bonaparte, 1837, Vespertilio murinus Linnaeus, 1758, Eptesicus serotinus Schreber, 1774, Hypsugo savii Bonaparte, 1837, Pipistrellus pipistrellus Schreber, 1774, P. pygmaeus Leach, 1825, P. kuhlii Kuhl, 1817, P. nathusii Keyserling and Blasius, 1839, Nyctalus noctula Schreber, 1774, N. leisleri Kuhl, 1817, Plecotus auritus Linnaeus, 1758, P. macrobullaris Kuzyakin, 1965, P. austriacus J. Fischer, 1829, Barbastella barbastellus Schreber, 1774, Miniopterus schreibersii Kuhl, 1817 and Tadarida teniotis Rafinesque, 1814 (Presetnik et al., 2014). The highest diversity of bat species is observed along the Adriatic coast and in the southern part of the Balkan Peninsula (Kryštufek, 2004). The bat fauna of the lower altitudes is characterized by typical (sub-) Mediterranean species, such as R. blasii, R. euryale, Myotis capaccinii, Miniopterus schreibersii, and T. teniotis, whereas in the forest habitats and/or higher altitudes of the Dinaric Mountains, there are Myotis alcathoe, M. brandtii, N. leisleri, P. auritus, P. macrobullaris, B. barbastellus, and V. murinus. Accordingly, the composition of mammals from the Trlica taphocenosis is diverse. Myotis blythii longocaninus is very closely related to living M. blythii (Popov, 2004), which is a widespread and common species in the country recorded from the sea level up to 1.700 m altitude (Presetnik et al., 2014). According to Topál (1979) and Wołoszyn (1987), another fossil bat from Trilca, R. cf. macrorhinus anomalidens, belongs to the “ferrumequinum” group of large-sized species of horseshoe bats. Extant R. ferrumequinum is also widespread and common in Montenegro (Presetnik et al., 2014). Both fossil bats are typical Mediterranean species and quite naturally present in the Pleistocene fauna of Trlica.

39The lower layers (beds 9 - 11) have yielded not only southern European mammals, such as, e.g., Ursus cf. etruscus, Pachycrocuta brevirostris, Stephanorhinus etruscus, Equus stenonis, Cervus acoronatus, Hystrix refossa, Mimomys pliocaenicus, but also some Siberian and Central Asian elements, i.e., Capreolus and Lagurus sp. African species, e.g., Lycaon lycanoides are also present here. The endemics of the Balkan Peninsula, Megalovis balcanicus and Soergelia intermedia, are common in the Trlica taphocenosis (Vislobokova & Agadjanian, 2015, 2016).

40The upper layers (beds 6 - 5) have yielded U. deningeri, E. cf. suessenbornensis, S. hundsheimensis, C. cf. acoronatus, Bison cf. shoetensacki, which are widespread in Europe from the beginning of the Middle Pleistocene. Among rodents such species are Glis sackdillingensis, Mimomys ex gr. savini, Terricola hintoni. A typical Asiatic faunal element from the upper beds of Trlica is the steppe lemming of the tribe Lagurini.

41It is generally accepted that various bio-geographical bridges repeatedly connected the Balkans with European, African, and Asian lands during their history (Savić, 2008). Thus, the Balkans are considered to be the crossroads of three continents and play a great role in mammal migration and distribution (Kostopoulos et al., 2002; Koufos et al., 2005). The mosaic composition of the Trlica fauna indicate that, during the Pleistocene, the area of modern Montenegro, the Balkans, the Danube River Basin and South Germany have provided migration pathways from the south to the north, from Africa to Europe, and from the east to the west, from Asia to the Atlantic coast. In particular, it is known that the steppe lemmings (Lagurus lagurus) migrated westerly up to France via northern Central Europe and even to the islands of Great Britain (Chaline, 1972; Sutcliffe & Kowalski, 1976). On the other hand, it also represented the main gate to Western Europe for the faunal elements coming from Asia and America through Beringia (Spassov, 2003).

5.3 - Paleoecological and taphonomic aspects

42As discussed above, the older faunal complex of mammals from the Trlica taphocenosis corresponds to the Early Pleistocene, while the younger faunal complex is typical for the early Middle Pleistocene. This last was the time of the distribution of the ancient Homo in the Mediterranean along the southern edges of the European subcontinent (Kahlke et al., 2011). The high number of Ovibovini among bovids (here and below see Vislobokova & Agadjanian, 2015, 2016) and the lack of Gliridae among rodents in the lower part of the Trlica deposits indicate that there was a relatively cool climate and there were a lot of open land habitats in the studied area in the Early Pleistocene. At the beginning of the Middle Pleistocene the increased abundance of the dormouse Glis and the decreased number of Ovibovini indicate that the climate became milder and the forest areas increased, including broad-leaved forest. The presence of Lagurini and the equids in all layers of Trlica strata is evidence of the continued existence of the local steppe areas at the North of the Montenegro territory during the Early and early Middle Pleistocene. However, abundance and diversity of the cervids indicate that large areas were covered with shrub (Vislobokova & Agadjanian 2015, 2016).

43The presence of several episodes of sediment accumulation at the Trlica site is connected with different taphonomic processes in the karstic cavity and paleogeographic stages of the development of the surrounding environments.

44Beds 11 and 12 are deposited by a stream with an unstable hydrological mode. Alluvial material of these beds was apparently accumulated at the level of a cave, when the channel of the Cheotina River was approximately 150 m higher than at the present time. The absence of large mammal bones or hyena coprolites in bed 12 probably suggests that during accumulation of these strata, subaquatic conditions in the cave occurred only incidentally. When beds 8-10 accumulated, the cave was inhabited at least by bats and visited by hyenas.

45The taphocenosis of Trlica includes both large and small mammals; accumulation of bone fragments of particular animal groups differed in mechanisms and sources. Predators used the karstic cavity as a refuge or den for raising young, as follows from the presence of abundant layers of coprolites and teeth of young animals, mostly hyenas and bears. Bones of large herbivores, deer and bovids were apparently brought by large predators, as evidenced by considerable fragmentation and bite traces on the bones (Crégut-Bonnoure & Dimitrijević, 2006; original data). In addition, many fragments of large bones show traces of treatment by gastric juice. Skeletal remains of predators are also frequent in the taphocenosis; the proportions of bear, hyena, and large cat bones are 4.9, 2.8, and 1.95 %, respectively, and wolves and African hunting dogs are 1.2 % (Vislobokova & Agadjanian, 2015).

46Most bones of small mammals to all appearances come from pellets of birds of prey and possibly from excrement of terrestrial predators. It is also possible that the remains of some animals accumulated as the result of natural deaths, for example, during hibernation (e.g., bats, Hystrix).

47Bones of dormice found in beds 5 and 6 obviously come from pellets of owls. All extant dormice are nocturnal and crepuscular animals inhabiting trees (Airapet’yants, 1983) and, hence, nocturnal birds of prey, owls, apparently produced pellets with their remains. At the same time, owls successfully prey on other terrestrial rodents, such as voles, which also occur in the Trlica taphocenosis. The statement that bone fragments of small mammals come from pellets is supported by the fact that some specimens display traces of effects of gastric juice and apparently passed through the digestive system.

48Since chiropteran remains are rather rare, it is difficult to propose with certainty a taphonomic model of their accumulation. However, it is hardly probable that they come from pellets of birds of prey, because bat bones and teeth lack traces of digestive juice and bed 10 has yielded very thin postcranial bones. Bats were apparently resident inhabitants of the cave; however, they had never been abundant, since their bones are rather scarce. In our opinion, chiropters only sporadically occupied the cave cavity which was likely a grotto. Nevertheless, the room between the floor and top was sufficient (at the least during the time of the deposition of the lower part of the cave sequence) for flights of horseshoe bats, which dwelt there. Because of certain structural skeletal features, most of the extant Rhinolophidae show very poor ability of moving along a hard surface (Kuzyakin, 1950). They usually pass to a suspended position just from flight, attaching from below to a vault of a refuge, and can move upside down along it by means of hind legs for only a few steps (Hall, 1989). Under a hanging animal, there must be a sufficiently large space, because its flight is preceded by a passive falling, which is followed by rapid expansion of wings required for active flight (Kuzyakin, 1950).

6 - Conclusions

Our new data have revealed the presence of units of different age in the sequence of the Trlica site and several, probably large, gaps in sedimentation. From the taxonomic composition of the mammal fauna, deposits of the Trlica locality are divided into older lower units (beds 9 - 12) and younger upper units (beds 8 - 1), which are separated by a significant time interval. The fauna of older beds corresponds to the second half of the Early Pleistocene (Late Villafranchian), it is dated MNQ17 - MNQ18 and therefore is older than 1.2 Ma. The fauna from the upper unit of the Trlica site is evidently of basal Middle Pleistocene age.

49According to our study, the list of previously known small mammals from Trlica at Montenegro is enlarged by Cricetus praeglacialis, Cricetulus sp., Mimomys pliocaenicus, Mimomys ex gr. savini, Terricola sp., Lagurus ex gr. transiens, Microtus nivaloides, Pliolagus sp., Martes sp. and small species of Vulpes. The fossil remains of two bats, Rhinolophus cf. macrorhinus anamalidens and Myotis blythii longocaninus, are recorded for the first time.

50Against a background of a high modern biological diversity of Montenegro, which includes the Central European, Illirian, Alpine and Mediterranean floral and faunal elements, Pleistocene taphocenosis of the Trlica site includes some mammals of southern Europe (e.g., Stephanorhinus etruscus, Equus stenonis, Hystrix refossa, Mimomys pliocaenicus etc.), Africa (e.g., Lycaon lycanoides) and also elements of Siberia and Central Asia (e.g., Capreolus sp. and Lagurus sp.). Taphocenoses of the upper beds (6 - 5) of the Trlica site includes species that were widespread in Europe from the beginning of the Middle Pleistocene (e.g., Ursus deningeri, Equus cf. suessenbornensis, Bison cf. shoetensacki, Glis sackdillingensis, Mimomys ex gr. savini and Terricola hintoni etc.). This supports the hypothesis that the Balkans and Central Europe were the pathways and shortest crossroads between Europe, Africa and Asia and played a great role in mammal migration and distribution.

51The majority of remains of small mammals of Trlica apparently come from pellets of birds of prey and excrements of terrestrial beasts of prey. It is most likely that bats were resident inhabitants of the cave and their bones probably accumulated as a result of the natural deaths of animals.

52According to paleoreconstructions, the climate in the studied area during the Early Pleistocene was relatively cool, but it became milder at the beginning of the Middle Pleistocene and the forest areas increased, including broad-leaved forest. However, the local steppe areas covered with shrub have continued to exist in the North of the Montenegro territory.