Carnivores from the Borsuka Cave ( southern Poland ) as an example of changes in carnivore assemblages during MIS 2 and MIS 1

Two faunal assemblages were identified in the fossil material of carnivores from the Borsuka Cave. The older one, of Late Pleistocene age, found in layers V-VII and dated as the end of MIS 3 and MIS 2, included nine species: Vulpes lagopus, Vulpes sp., Ursus arctos arctos, Gulo gulo, Meles meles, Martes martes, Mustela erminea, Mustela nivalis and Lynx lynx. The younger assemblage, dated as MIS 1 and found in layers I-IV, comprised six species: Vulpes vulpes, Ursus arctos arctos, Meles meles, Martes martes, Mustela ex. gr. eversmanii-putorius and Felis silvestris. The carnivore material was represented by almost all skeletal elements, with the majority constituted by cranial bones, isolated teeth, vertebrae, metacarpals and metatarsals. Remains of Vulpes vulpes and Meles meles were the most abundant; the presence of young individuals of both species confirmed that they used the cave as a shelter and to raise cubs. The other carnivores were much rarer. Especially the occurrence of an uncommon members of Late Pleistocene paleocommunities: Gulo gulo and Lynx lynx is noteworthy. Some components of the older assemblage: Ursus arctos arctos and Gulo gulo, represented large, robust forms, whose great size according to Bergman’s rule was an adaptation to cool climate conditions. Likewise, the two smallest mustelids, Mustela erminea and Mustela nivalis, were represented by small and gracile specimens which were characteristic of Late Pleistocene cold phases. Metrically and morphologically the animals from the younger period dated as MIS 1 corresponded to the modern European forms.


I. INTRODUCTION
The Borsuka Cave (50º9N53NNN 19º42N12NNE) is located on the northern slope of the Szklarka valley in the southern part of Polish Jura, ca.20 km west of Kraków (Fig. 1).The site was discovered in 2007, during cataloguing and exploration of the Szklarka valley caves (NOWAK 2007).Excavations were carried out in 2008-2010 and a single trench of 4x3 m was dug next to the cave entrance (WILCZYÑSKI et al. 2012a(WILCZYÑSKI et al. , 2012b)).

II. METHODS AND MATERIAL
II.1.METHODS Measurements were taken point to point, with an electronic calliper, to the nearest 0.01 mm.Each value given here is the mean of three measurements.Measurements shown in brackets denote the estimated value of the dimension.Additionally, some measurements were taken using a set for image analysis Olympus (Olympus stereo microscope ZSX 12, camera Olympus DP 71, programme Cell D).This set, coupled with camera Canon EOS 5D, was also used to take photographs.Osteological and dental terminology follows ANDER-SON (1970).The measurements are shown in Figs 2-10.Fig. 2. Scheme of measurements of canid mandible: 1 -total length (infradentale to condyle), 2 -distance: infradentale to angular process, 3 -distance: infradentale to angular indentation, 4 -distance: behind c1 to condyle, 5 -infradentale to anterior margin of masseter fossa, 6 -posterior margin of c1 to posterior margin of m3, 7 -cheek teeth row length (anterior margin of p1 to posterior margin of m3), 8 -premolar row length (anterior margin of p1 to posterior margin of p4), 9 -molar row length (anterior margin of m1 to posterior margin of m3), 10 -distance between mental foramina, 11 -symphysis maximum diameter, 12 -symphysis minimum diameter, 13 -posterior margin of m3 to condyle length, 14 -condyle height, 15 -condyle breadth, 16 -angular process to coronoid process height, 17 -mandible maximum height, 18 -mandible body height between p2 and p3, 19 -mandible body thickness between p2 and p3, 20 -mandible body height between m1 and m2, 21 -mandible body thickness between m1 and m2.

II.2. MATERIAL
The material described here comes from the excavations conducted by the team leader Jaros³aw WILCZYÑSKI in 2008-2010 with cooperations of Micha³ WOJENKA from Institute of Archaeology, Jagiellonian University.The trench of 4x3 m was located near the cave entrance (WILCZYÑSKI et al. 2012a(WILCZYÑSKI et al. , 2016)).All the material is stored at Institute of Systematics and Evolution of Animals, Polish Academy of Sciences in Kraków (Table II).

Table II
Inventory catalogue from the Borsuka Cave (symbol "-" -without the layer, usually heap)

III. TAXONOMIC PART
Twelve species from four carnivore families were found in the sediments of the Borsuka Cave: three 3 canids, one bear, six mustelids and two felids (Table III).The most abundantly represented species were Vulpes vulpes and Meles meles, while remains of other species were scarce.However, even the most abudant species were represented only by

Canis lupus LINNAEUS, 1758
An isolated post-cranial bone belonged to a large canid, of the size and proportions indistinguishable to those of the grey wolf.Although the individual is not small, much larger specimens are known from other sites, also in the Sudetes Mts, for example the NiedŸwiedzia Cave.Unfortunately, no tooth remains which would permit distinction between the different forms of C. lupus were found, and the few skeletal remains are insufficiently diagnostic.For this reason, the large canid was classified as C. lupus.(LINNAEUS, 1758) and Vulpes lagopus LINNAEUS, 1758 Two small and morphologically very similar foxes were found in the sediments of the Borsuka Cave: the red fox Vulpes vulpes mostly in the Holocene period and the polar fox Vulpes lagopus in the Late Pleistocene sediments.Teeth of the red fox are on average slightly larger than those of the polar fox, however the size differences are not so significant (Tables IV-V).I3 of both species was larger than I1 and I2, with an elongated and curved crown.V. vulpes has a larger and stronger I3, with a more massive and curved crown.Also, the lingual cingulum is weaker developed.The upper canine is elongated and narrow, with an only slightly developed lingual cingulum.The canines of V. lagopus are smaller in size, shorter and less curved than those of V. vulpes.P4 of V. lagopus is on average smaller and more robust, with proportionally shorter protocon, which is less oriented mesio-lingually in the anterior direction (BARYSHNIKOV 2015).

Vulpes vulpes
The triangular M1 in V. lagopus has a well-developed metaconulus strongly connected with the posterior part of hypocon, the protoconulus is less developed and more closely associated with the protocon.The triple hypocon is large and smooth, while the mesial cingulum is rather weak.In the general occlusal view M1 of V. lagopus is narrower and smaller, with proportionally longer and narrower talon.The trigon is not so broad as in V. vulpes, with its main cusps lower and almost equal in size (RABEDER 1976).
The differences in M2 morphology between the two fox species are even more pronounced than in M1: the tooth in V. lagopus is much smaller, narrower and more reduced than in V. vulpes.The talon area is narrow and short, separated from the trigon by a deep, V-shaped valley.The metaconulus and the protoconulus are not connected, and both cusplets are more reduced (RABEDER 1976).The mandible body of V. lagopus is shorter and more massive, with a more pronounced difference in massiveness between its anterior and posterior parts.Also, the curvature of the lower mandible margin, which starts at the p3/p4 border is more marked in V. lagopus.(Fig. 11).The lower teeth row is more curved in V. lagopus than in V. vulpes, and the diastemas between the back teeth p4, m1, m2, and m3 are shorter, and the teeth are closer together.Except for the smaller average dimensions, there are no reliable morphological differences between the lower incisors of the red and polar foxes.The lower canines of V. lagopus are metrically smaller than c1 of V. vulpes, with proportionally shorter and more curved crowns.The p1 alveola is more rounded in V. lagopus, while some differences in morphology of the premolars are discernible between both species.However, these imprecisely defined features have not been commonly applied in taxonomical analyses and their diagnostic value has not been widely discussed in the literature.Also, the great intraspecific variation of both species effectively hinders the correct taxonomical assignation.On average, the premolars in V. lagopus are more massive, their posterior part is slightly shorter, and the difference between the narrower anterior and the broader posterior parts is less marked.However, the differences between isolated teeth are so tentative and unreliable, that single premolars are usually classified as undetermined members of the genus Vulpes (RABEDER 1976).The lower carnassial of V. lagopus differs considerably from m1 of V. vulpes in its more compact build, longer and broader trigonid, and a smaller metaconid, and a narrower and shorter talonid.The two largest cusplets in the talonid, the hypoconid and the entoconid, are lower and separated from one another by a deep and broad valley.In V. vulpes both cusplets are quite high and connected by a transverse, well-developed thick ridge.The number of additional small cusplets is usually smaller in V. lagopus than in V. vulpes.In V. lagopus m2 is smaller and more elongated than in V. vulpes.The crown is lower and in occlusal view more oval, with a narrower and proportionally longer talonid, a medium-sized hypoconid, a minute entoconid and no mesoconid (RABEDER 1976;BARYSHNIKOV 2015).
According to the methodology proposed by SZUMA ( 2004), based on the morphological variation of the talonid occlusal surface in m1 of V. vulpes, the two m1 from the Borsuka Cave represented the morphotype P4.It seems that overall the red fox dentition is highly conservative and shows negligible morphological changes through time (BARYSHNIKOV 2006).In foxes, males are somewhat larger than females.The vast number of red fox bones from the Borsuka Cave probably belonged to two individuals, however it is not excluded that a greater number of individuals was represented.The measurements indicate that the remains of V. lagopus belonged to a fully adult, medium-sized individual (Table III).It is impossible to say if it was a large female or a small male.The remains of V. vulpes belonged to a young, adult but rather small animal, with dimensions smaller than the values for the modern Polish red fox (Table IV).The most probable explanation is that they represented a fully grown but young red fox female.
Besides the mandibular material and the isolated teeth listed above, the material included also a variety of post-cranial bones (Figs 12-13): long bones and their fragments, numerous additional bones, metacarpals and metatarsals, and phalanges.Almost all of them were classified as V. vulpes, while only few post-cranial bones were assigned to V. lagopus (Figs 12-13).There are some morphological differences in the proportions of bones and muscle attachments, but they are not significant.On average the long bones, metacarpals, metatarsals and phalanges of V. lagopus are shorter and more gracile (Table VI) (ALTUNA 2004;GERMONPRÉ & SABLIN 2004;BARYSHNIKOV 2006;LIPECKI & WOJTAL 2015).The analysis of the measurements of phalanges, which at first glance have a rather small taxonomic value, provided interesting information.The analysis of a large series of phalanges I and II showed that in V. lagopus they were more gracile than those of V. vulpes (Table VI), even if comparable in total length (Fig. 14) and their proportions are similar to those of the foxes from Kraków Spadzista (LIPECKI & WOJTAL 2015).The scatter plots of the phalanx total length against the smallest diaphyseal breadth show that the fossil sample of both species is rather homogenous.It is also apparent that two phalanges I from the Holocene sediments of Borsuka Cave classified as V. vulpes correspond in size to the recent Polish red foxes (Fig. 14).

Ursus arctos LINNAEUS, 1758
The few small fragments of an ursid belonged to large, arctoid bear.The best preserved were two isolated teeth, with morphology typical of the brown bear.On the tricuspid P4 (MF/6977) there are no traces of metaconulus or metaloph (Fig. 15).The buccal margin is almost straight (only delicately concave in its middle part), whereas the lingual, as well as anterior and posterior margins are much rounded.The paracon is large, round and quite low, the metacone is almost equal to it in size, but lower and more oval in shape, the valley between the two is strongly pronounced.The elongated and oval, low protocon is a well-developed, large cusp, sharply delineated from the remaining part of the crown.Its anterior margin forms a wide, open angle with the posterior margin of the paracon.On the lingual side a delicate crest running from the protocon toward the paracon crosses the valley between the two main cusps.The cingulum is well-developed, but stronger (forming a small crest) in the posterio-buccal part of the crown.It represents morphotype A/B, typical of brown bear (Fig. 15).The tooth measurements (1 -18.45 mm, 2 -10.48 mm, 3 -14.45mm) place it above the range of Ursus arctos arctos, between large Middle and Late Pleistocene forms: Ursus arctos taubachensis and Ursus arctos priscus (Fig. 16).However, its simple structure, without any additional cusplets, excluded such taxonomic placement.P4 in those forms is large and broad, proportionally low-crowned and has smaller cusplets: metaconulus, metaloph and/or protoloph.The only reliable explanation is that these remains, dated at the beginning of the Holocene, represented a large brown bear with priscus-like features still present in its morphology.The same explanation can be applied to the isolated m3 (MF/6979), found in layer 6 in the Borsuka Cave.Because of the heavy wear, no chewing surface details are distinguishable.Only the outline from occlusal view, representing morphotype C can be discerned.This morphotype rarely occurs in Ursus arctos arctos, while it is common in Ursus arctos priscus and Ursus arctos taubachensis.The measurements of the tooth (1 -20.52 mm, 2 -7.16 mm, 3 -15.22mm, 4 -12.94mm) are within the range of those two forms.

Gulo gulo (LINNAEUS, 1758)
The record of Gulo gulo is based on metatarsal II (MF/6976), whose shape and proportions closely resemble other Late Pleistocene remains of the species.Also metrically the bone matches the Late Pleistocene wolverine and exceeds the values for the modern European males (Fig. 17/1a-c, Fig. 18).Large individuals are associated with glacial periods and smaller ones appear during interglacials.The species follows the Bergmann rule (PASITSCHNIAK-ARTS & LARIVIÈRE 1995).The bone also differs from the second metatarsal of Meles meles in its much larger size, especially total length and more gracile build, with an elongated shaft and proportionally slim both epiphyses (Table IX).Besides, the proximal end (distal end missing) and the shaft differ morphologically.The shaft proximal to the middle differs from that of the badger in being less cylindrical and roughly triangular in cross-section.This character of shaft usually permits the medullar foramen to be seen in ventral view in Gulo, whereas in Meles the foramen is usually hidden in this aspect and can be seen best when the metapodial is viewed from the inner side.The rugose tendinal area on the ventral side below the proximal end appears to be more conspicuous in Meles and often extends farther down the shaft than in Gulo.When viewed from the side, the shaft is seen to be slightly more curved in Meles than in Gulo.The ventral arm of the hammershaped facet for the entocuneiform is shorter.In Gulo the proximal end of the bone exhibits a greater extent of bone lying ventral to this facet than in Meles.Frequently in Gulo the dorsal facet for articulation with mtts III reaches back to the proximal end of the bone, and the ventral facet, also for mtts III, does not reach the proximal end.In Meles the situation is reversed for these two surfaces.The border of the median ventral keel of the distal articulation, when viewed from the side, is quite straight in Gulo, as in the specimen from the Borsuka Cave.In Meles the keel usually curves dorsally and the border is convex (Fig. 17/1a-c).
The size and morphology of the preserved dentition of Meles from Borsuka cave are typical of the species (able VII), and the young age of most teeth makes it possible to examine their structure.The I3 is single-rooted and labio-lingually elongated, with most of the crown consisting of a large cusp that is deflected lingually and compressed labio-lingually, resulting in a blade-like form.The blade arches slightly labially in occlusal view.The mesial edge of the main cusp is about half as long as the distal one.The cingulum, quite well-developed mesio-lingually and distally, is continuous with the mesial edge of the main cusp.The distal cingulum is less developed and not so long, bulging out into a minor cusplet at the place of contact with the posterior edge of the main cusp.
The triangular P4 has a slightly convex buccal margin and is slightly concave lingually.The paracon is the largest and the highest cusp, which is compressed bucco-lingually and has three thin, but well-definied ridges running from its top to the base of the cusp.The anterio-lingual ridge is rounded and blunt, while the anterio-buccal and posterior ones, which constitute the anterior part of the P4 blade, are more trenchant.The low metacon is slightly curved buccalo-posteriorly, while the metastyl is somewhat swollen.The buccal margin is little depressed in its median part, and ended with a well-developed, quite thick cingulum.
The trapezoid M1 has a low crown with mostly convex margins.The two main cusps, the paracon and the metacon, are equal in size, but the paracon is higher.The top of the paracon is situated anteriorly to the midpoint of its anterio-posterior length, while in the metacon the peak culminates more posteriorly to the midpoint of its length.Both the main cusps are compressed buccallo-lingually, resulting in a blade-like form, and divided by a well definied, V-shaped deep valley.The paracon blade in its anterior direction passes into a minute parastyle ridge, while the metacone base posteriorly passes into a metastyl.A fine, sharply dinstinguishable cingulum surrounds the anterio-buccal margin of the crown.
The crest on the lingual side is elongated and well-developed, starting from the anterior base of the paracon and the parastyl connection, and posteriorly exceeds considerably the length of the metastyl.The crest is formed by a row of four elevations whose size increases in the anterior-posterior direction.All four elevations are separated from one another by deep, narrow valleys.Near the smallest, anteriormost elevation on the anterio-buccal side there is situated a small, but well-defined paraconul.From its base a short and sharp ridge runs in the posterior direction and disappears into the lingual base of the paracon.The protocon is formed by the largest, posteriormost elevation.
The large metaconul, of the same size as the protocon, is positioned posterio-buccally to the protocon and posterio-lingually to the metacon, and it is clearly separated from both these cusps by deep valleys.The V-shaped valley situated between the metaconul and the metacon runs into the trigon basin, while the U-shaped valley separating the metaconul and the protocon is much more spacious.As a result a very low threshold distinguishes the trigon surface from the talon posterior basin.Four thin and sharp ridges run from the centrally positioned top of the metaconul; they reach the one cusp base so that the whole cusp resembles an isosceles triangle.The most expanded and posterior wall is convex, forming a concave posterior margin in front of and behind the metaconul in occlusal view.Two posterior ridges, demarcating the posterior wall are sharper than the blunter anterior ones.The anterio-buccal ridge reaches the metacon lingual base at the midpoint of the valley length separating the metaconul and the metacon.The posterio-buccal ridge forms a saddle between the metaconul and the metacon, which posteriorly closes the valley separating these two cusps.In the distal direction, after passing this saddle, the ridge vanishes into the posterior base of the metacon and the buccal side of the metaconul.Betwenn these two long ridges, located on the basal surface of the metaconul anterior wall there is a sharp, thin and short ridge, which runs to the bottom of the valley between the metaconul and the metacon.This ridge vanishes into the metacon base.The anterio-lingual metaconul ridge links it with the protocon along the threshold separating the trigon and talon surfaces.
The talon is much broader and more expanded anteriorly and posteriorly than the trigon.In occlusal view it is noticeably convex in the mesial direction.The lingual margin is strongly concave at about the midpoint of its anterio-posterior length, which separates the talon surface in two parts: anterior and posterior.This lingual outline is surrounded by a very thick and high, almost continuous cingulum which starts in the paraconul anterior base and ends in a V-shaped notch, where it connects with the posterior-lingual ridge of the metaconul.This thick cingulum ridge on occlusal side forms a few small and two larger elevations located one behind the other.The lingualmost are the smallest, while the buccalmost are the largest.Among the small elevations, four cone-like ones flank the anterior surface, while the other three flank the posterior surface.A distinct, V-shaped notch borders the buccal and the medial elevations of the posterior row.The other elevations are also more or less isolated by shallower, V-shaped valleys.The grooves separating the small elevations located the lingual-most from the large cingular elevations are thin but well-visible.These large elevations constitute notably large and thick ridges that surround the ligual margin of the anterior and posterior surfaces of the talon.They are separated by a broad depression at the level of the concavity in the talon lingual margin.The posteriorly located hypocon is oriented anterio-lingually and connects with the posterior large elevation, which ends in a sharp bulge, forming a tapered tip.There is a discontinuous, thin cingulum running along the lingual margin from the level of the top of posterior large elevation and ending in the notch between the lingual and medial posterior small elevations.The talon surface is divided into three depressions, where the anterior and the medial one are separated from each other by a low threshold which connects the anterior large elevation of the cingulum with the medial large elevation of the trigon lingual crest.Another low threshold separates the medial and the posterior deppresions, and connects the protocon posterior part with the posterior large cingular elevation.The numerous post-cranial bones of Meles meles are easily distinguished from other mustelid remains based on their relatively large size and massive build, while they differ from the bones of Gulo gulo in their smaller dimensions and stout build (Tables VIII-IX).Especially the well-preserved and numerous metacarpals and metatarsals are proportionally short and massive, with robust both epiphysis .Their measurements fall far below the minimum total length of the wolverine metapodials (Fig. 18).

Table VIII
Measurements of isolated mustelid teeth from the Borsuka Cave (mm).Measurements of teeth according to VON DEN DRIESCH (1976)

Species
Coll Three isolated teeth, a single canine and two p4, represented Martes martes.Metrically and morphologically those remains are indistinguistable from corresponding teeth of the pine marten (Table VIII).Since Martes martes is a classical example of Bergmann's rule, whereby larger individuals are found in colder climates, and those of smaller size are found in warmer regions or during warmer periods, the single canine from the Borsuka Cave belonged to a large-sized specimen.MARCISZAK (2012) reported the following values of c1 length in individuals from recent Polish populations: mean for males 5.83 mm (4.81-6.65 mm, n=163), and mean for females 4.24 mm (3.74-5.14mm, n=108).With its crown length of 7.14 mm the tooth MF/6969 considerably exceeds the dimensions of the modern Polish pine marten and corresponds to the robust Late Pleistocene individuals (5.97 mm, 5.70-7.78mm, n=22).Contrary to c1, the single p4 MF/6968 is quite small and narrow, and indicates a medium-sized marten.Based on an abundant sample from the modern Polish population, MARCISZAK (2012) reported the following values for p4: mean for males 6.14 mm (5.12-6.78mm, n=157), and a mean for females 5.48 mm (4.87-6.06mm, n=108).The teeth described above differ from those of Martes foina in the relatively longer and less curved crowns of c1 and the narrower crown with a less pronounced additional cusplet located posterio-buccally to the protoconid.

Mustela ex. gr. eversmanii-putorius
Besides Martes martes, medium-sized mustelids from the Borsuka cave are also represented by an undetermined polecat classified as Mustela sp.(ex.gr.eversmanii-putorius) (Fig. 19).Two polecat species occur in Poland in the Pleistocene deposits and today: Mustela eversmanii and M. putorius.Although similar in metric and morphological values, they differ in many features, which have been intensively studied by many authors e.g.NEHRING (1904); KOKEN (1912) Most distinctive features are provided by the calvarium and mandible, as well as teeth; especially carnassials and molars differ clearly.Direct measurements are not very useful in polecat identification.M. eversmanii is on average larger, the mean values differ between the two species, but the extreme values are almost the same.Because of the pronounced sexual dimorphism, with adult males twice or more heavier and much larger than females, size differences provide no reliable information besides sex determination.All found differences are taxonomically valuable, when the analysed material is represented by adult specimens.However, the very young age and deciduous dentition of the mandible from the Borsuka Cave make the identification impossible (Fig. 19/1a-c).In the Borsuka Cave also a complete femur of the polecat was found (Fig. 19/2a-b), and findings of long bones of the genera Martes and Mustela are quite common in archeological and paleontological sites.Despite the fact that the bone is adult and complete, as in the case of the above-described mandible it is almost impossible to identify it precisely.The metric and morphological similarity between polecats is so great that post-cranial material is usually referred to as Mustela sp.Although some authors (GALIK 1996(GALIK , 1997) tested metric differences in proportions and morphology of long bones, their attempts were based on a relatively scanty material, not reflecting the large size variation in the two species.Morphological differences pertain mainly to the humerus (KOBY 1951;GALIK 1997;MALLYE and GUÉRIN 2002).The femur from Borsuka Cave is particularly large and robust and belonged probably to an adult male (Fig. 19/2a-b).

Mustela erminea LINNAEUS, 1758 and Mustela nivalis LINNAEUS, 1766
The two smallest mustelids from Borsuka Cave, Mustela erminea and Mustela nivalis, are represented by relatively small material, the best preserved remains being two mandibles of the least weasel (Fig. 20).Distinguishing small members of the genus Mustela in the fossil material is difficult, especially when the material is small and/or incomplete.The most valuable criterion is the larger size of M. erminea, though it is not always reliable because of the pronounced sexual dimorphism (males of smaller species are sometimes larger than females of larger species), the geographical variation, and the effect of altitude and local envinronmental conditions.Late Pleistocene members of Mustela, especially those from cold periods, were on average smaller than the individuals from warmer periods (WÓJCIK 1974;MARCISZAK 2012).One isolated m1 from the Borsuka Cave, with the total crown lengths of m1 of 5.37 mm (MF/6989) (Table VIII), represented M. erminea, while the two much smaller mandibles from the same site, with m1 3.16 mm (MF/6973) and 3.76 mm (MF/6974) long, were determined as M. nivalis.The larger values in both species most probably represented males, the smaller ones -females.Additionally the relatively small size and gracile build suggest that their morphology reflected adaptation to colder envinronmental conditions.These small individuals of both species are metrically and morphologically indistinguishable from modern dwarf individuals which occur in Arctic regions.
Also other metric values proved to be reliable diagnostic criteria to distinguish between the stoat and the least weasel.M. erminea is characterised by a proportionally broader posterior part of the p4 crown, a longer trigonid, a strongly pronounced broadening of the m1 crown situated at the protoconid level, and a narrower and shorter talonid (Fig. 21).The mean length ratio of talonid to trigonid for M. erminea is 23.4,for M. nivalis it is 41.2.This ratio for the specimens from the Borsuka Cave is 28.3 (MF/6989) for M. erminea and 45.0 (MF/6973) and 40.3 (MF/6974) for M. nivalis.At the ame time, the mean breadth ratio of talonid to trigonid is 76.0 for M. erminea and 92.1 for M. nivalis.The values obtained for the specimens from the Borsuka Cave clearly separated the stoat remains from the least weasel remains, with 96.1 (MF/6973) and 96.2 (MF/6974) for M. nivalis.
Because of the uniform morphology and great variation of small felids, it is always difficult to distinguish between the species and forms.Besides, the findings dated as MIS 1 are additionally complicated by the occurrence in the fossil material of the feral cat, a form created by man (KRAJCARZ et al. 2015b).In the past many authors tried to find statistically reliable features which would allow separating the wildcat and the feral cat (SUMIÑSKI 1962;KRATOCHVÍL 1973KRATOCHVÍL , 1976;;HEMMER 1993;KITCHENER et al. 2005).However, detailed metric and morphological analyses of large samples of both forms from archaeological and palaeontological sites showed that virtually no measurements or ratios can be regarded as reliable.Even when mean values differ, the ranges still overlap.On average Felis silvestris is larger and more robust, however the feature is not completely diagnostic.The great intraspecific variation, sexual dimorphism and above all hybridisation between the two forms, especially frequent during the last 2 ka render their identification very difficult.In this context morphological features proposed by KRATOCHVÍL (1973KRATOCHVÍL ( , 1976) ) are more valuable than measurements.The specimen from Borsuka Cave is not very large and rather robust (Fig. 22).The massive mandible with large m1 (length exceeding 8.0-8.5 mm in subfossil finds almost exclusively representing wildcat) is more typical of the wildcat, and the specimen was identified accordingly.

IV. DISCUSSION
The carnivore remains from the Borsuka Cave represent 12 species: Canis lupus, Vulpes vulpes, Vulpes lagopus, Ursus arctos arctos, Gulo gulo, Meles meles, Martes martes, Mustela ex.gr.eversmanii-putorius, Mustela erminea, Mustela nivalis, Lynx lynx and Felis silvestris.They are common components of the Late Pleistocene and Holocene faunal assemblages of Central Europe (KOWALSKI 1959;WOLSAN 1989;WOJTAL 2007;NADACHOWSKI et al. 2015;STEFANIAK 2015;LIPECKI & WOJTAL 2015;WILCZYÑSKI et al. 2016;MARCISZAK et al. 2017).The carnivore remains from the Borsuka Cave represent two different faunal assemblages.The younger one comes from the Holocene layers I-IV, while the older one was found in layers VI-VII of Late Pleistocene age.The material from the younger layers is poorer in species but richer in terms of the number of remains and specimens.Deposition during that period could be associated with both animal and human activities during the last 8 ka.
The older assemblage (layers VI-VII) provided a rich and diverse fauna.Its composition, the traces of carnivore activity (gnawing marks and digested bones) and the lack of traces of human activity (cut marks, burnt bones) indicate natural character of this bone accumulation.Layers I-II contained numerous remains of Vulpes vulpes (adult and young individuals), together with few bones of a wildcat and a badger.The remains of Meles meles and Felis silvestris indicate the presence of deciduous forests in the environs of the cave during the deposition (WILCZYÑSKI et al. 2012a).The layer III was dominated by Meles meles; the accumulation of its remains appears to have been natural.The layer V corresponds with the significant cooling of the climate associated with the Last Glacial Maximum and is almost devoid of archaeological or paleontological material.The most interesting faunal assemblage was found in layer VI, whose accumulation was associated with carnivore activity.Though the fauna included cold-loving species such as Vulpes lagopus, Gulo gulo and Mustela erminea, and those associated with a warmer, forest environment: Meles meles and Martes martes, it was dominated by euryoecious forms which could live in both kinds of habitats.
The euryoecious Late Pleistocene wolf is sometimes regarded as a distinct ecomorph C. lupus spelaeus (GOLDFUSS 1823).This form is mentioned from different European sites, but usually only briefly and almost never described in detail (DIEDRICH 2014).The form/subspecies category tends to be employed rather freely in the canid palaeontology.A preliminary analysis, however, indicates that the form differed somewhat in its morphology from the Holocene and modern specimens of Canis lupus.Apart from the larger average size, it was also characterised by broader teeth and more robust post-cranial bones (BOUDADI-MALIGNE 2012;SANSALONE et al. 2015).These differences require a more detailed analysis (MARCISZAK et al. 2017).
Another euryoecious species, the brown bear, is adapted to a range of climate conditions and habitats, including cold steppe-tundra (MUSIL 1964(MUSIL , 1996;;SABOL 2001aSABOL , 2001b;;PACHER 2007;MARCISZAK et al. 2015;LIPECKI & WOJTAL 2015).It was recorded from various cave and open-air sites of both the Late Pleistocene and the Holocene of Poland (MARCISZAK et al. 2015;NADACHOWSKI et al. 2015;WOJTAL et al. 2015;MARCISZAK et al. 2017).The ursid remains from the Borsuka Cave Belong to the robust brown bear, whose large size reflects its association with the cold periods, even if the date suggested already a Holocene age.
The occurrence of two species, Vulpes lagopus and Gulo gulo, in the layer VI indicates cool conditions and a steppe-tundra and/or tundra-open boreal forest biome near the site during the deposition.The polar fox is a circumpolar species, well adapted to cold conditions.Usually a tundra-dweller, it may also penetrate the boreal forest zone during its winter migrations (SHELDON 1992;KAHLKE 1999;BARYSHNIKOV 2006).During the Late Pleistocene the species was widespread in Europe (KAHLKE 1999), the maximum extent of its distribution range reaching the northern part of the Iberian Peninsula in the southwest (ÁLVAREZ-LAO & GARCIA 2011).Today Vulpes lagopus in Europe is restricted to the tundra regions of Fennoscandia (BARYSHNIKOV 2006).Due to climate warming the red fox wanders much more to the north than previously and aggressive encounters with the polar fox are not uncommon.Usually when they meet, the larger Vulpes vulpes dominates and sometimes even kills a smaller Vulpes lagopus (PAMPERIN et al. 2006).The same scenario is also not excluded for the vicinity of the Borsuka Cave at the time, when the more thermophilic red fox re-colonised after cold period, and when the polar fox was still present (Fig. 23).Its fossil records from Poland are numerous and dated mostly as the Late Pleistocene (WOLSAN 1989;CYREK et al. 2000;WOJTAL 2007;MARCISZAK et al. 2011;NADACHOWSKI et al. 2015;LIPECKI & WOJTAL 2015;WILCZYÑSKI et al. 2016).The wolverine is only defined as a northern element of boreal (taiga-type) forests.In modern times however it also lives in open tundra lowlands as well as at high mountain altitudes, which suggests that it inhabited also the open steppe-tundra (PASITSCHNIAK-ARTS & LARIVIÈRE 1995).Recently the wolverine in Europe is restricted to the areas extending from eastern Europe to the northern parts of the Scandinavian Peninsula, while in the Late Pleistocene its range was much more extensive, reaching as far as the northern part of the Iberian Peninsula in the southwest (DÖPPES 2001).
The oldest remains of G. gulo come from the lowermost layers of the Bioenik Cave (layer 19ad) (MARCISZAK et al. 2011).Based on biochronological data, those lowermost layers are estimated as MIS 10-9 (KRAJCARZ et al. 2014;MARCISZAK 2014).A few isolated teeth from layers 1 and 3 of the Deszczowa Cave are somewhat younger (MIS 6) (KRAJCARZ 2012;KRAJCARZ & MADEYSKA 2012).Gulo gulo from all three above-mentioned, late Middle Pleistocene sites are morphologically distinct from those of Late Pleistocene age.They represent a more primitive form, of a smaller size and slighter build.
The most numerous records come from the Late Pleistocene, when the species was widespread in Europe.Most of them are reliably dated the last glacial (MIS 5d-MIS 2).The remains were found in the Deszczowa Cave (left mandible with p2-m2 and some post-cranial bones of a female, layers 8 and 9, MIS 2) (WOJTAL 2007); the Nietoperzowa Cave (fragment of left mandible with p3-p4, layer 4, MIS 3) (WOJTAL 2007;KRAJCARZ & MADEYSKA 2012); the Mamutowa Cave (layer 5, MIS 3) (WOJTAL 2007); the Ciemna Cave (damaged calcaneus, layer C8, MIS 5) (KRUKOWSKI 1939(KRUKOWSKI -1948;;MARCISZAK et al. 2017).The single mtts II from layer 6 of the Borsuka Cave (WILCZYÑSKI et al. 2012a), from the layer 4, cave no. 4 at Birów Hill and trench F of the Kraków Spadzista open-air site (LIPECKI & WOJTAL 2015) are dated at MIS 2. The species is also found in Sudetic caves, where the most spectacular find is a partially preserved skeleton of very large and robust male dated at the postglacial period (MARCISZAK et al. 2016(MARCISZAK et al. , 2017)).Besides, isolated canines from the Pó³nocna Du¿a Cave and from the layer 21 (trench V) of the NiedŸwiedzia Cave, as well as a dozen well-preserved long bones and metapodials from the Naciekowa Cave of Late Pleistocene age can be added to the inventory.Likewise, the single premolar from the cave in Czarkowa and the fragment of right mandible with p3-p4 from the Mamutowa Cave, both without a stratigraphic context, are most likely of the Late Pleistocene age.Finally, it can be added that before the World War II a complete left foot of a particularly large individual was found in the archaeological site Gródek 2 in the environs of Równe (now Riwne in Ukraine), which was then a part of Poland (LOTH 1933).It seems that the find from Solna Jama is the only reliable postglacial record of the species from Poland.Till now no wolverine remains were found in any archaeological Holocene site in Poland (WYROST 1994;SOMMER & BENECKE 2004).No remains were found during our detailed revision of the material from most of those localities.
Detailed metric and morphological studies showed that late Early and Middle Pleistocene members of the genus Meles did not differ significantly from the modern badger in their teeth structure (MADURELL-MALAPEIRA et al. 2011;MARCISZAK 2012).No differences were found when the material from numerous Late Pleistocene and Holocene sites was compared with stratigraphically older remains.Similarly, no significant differences between skeletal bones were found.Overall, the variation ranges and proportions are com-parable to those of the modern badger.The observed small differences are probably due to interspecific variations.It should be also noted that the badger regularly disturbs the structure of cave deposits, digging extensive systems of burrows.As a result, its remains found in a layer do not have to be of similar or the same age as the sediments and other relics found in the same horizon (WISZNIOWSKA 1980).Taxonomic and morphometrical analysis of such early finds is in many cases rendered very difficult or even impossible by the fragmentation of the material and the presence of teeth without any significant diagnostic features.
The badger is the most common mustelid in the Late Pleistocene assemblages; it occurred in colder and warmer periods (WOLSAN 1989;SOMMER & BENECKE 2004;WOJTAL 2007;MARCISZAK 2012).The badger commonly used caves and rock shelters as a place to rest and raise their young.An example of such behavior is the site Ma³a Cave (Jura Wieluñska).An abundant material dated as Holocene was found only in the layer 22a-b.The badger burrows' systems were found to reach Pliocene layers (WISZNIOWSKA 1980).The species also plays an active role in the accumulation of debris; its waste products accumulate in its long-term latrines which are parts of underground corridors.Besides, various debris accumulate in the deep corridors, sometimes in considerable quantities.Examples of this kind of accumulation are the German sites Pisede 1 and Pisede 2, where postglacial and Holocene material was excavated from badger burrows (PETERS et al. 1973).
The badger is a highly social animal with a developed social structure and attached to the residence, which many generations may use (for more than 100 years).During that time the whole family can create a system of burrows and corridors of several hundred square metres.The presence of juvenile specimens confirms the use of caves as places to rear offspring (KURTÉN 1965;WISZNIOWSKA 1980;MALLYE 2007;MARCISZAK 2012).Abiotic factors, predators and a man also played a role in the accumulation of badger bones.Numerous archaeological sites were found to hold large accumulations of badger bones material with traces of human activity (LÜPS & WANDELER 1993;CHARLES 2000;MALLYE 2007).
Martes martes and, to a lesser extent, the badger are more or less associated with deciduous forests (SOMMER & BENECKE 2004;WILCZYÑSKI et al. 2012a), but the marten lived also in cooler habitats and taiga-type forests (ANDERSON 1970;HERNÁNDEZ-FERNÁNDEZ 2001;MARCISZAK 2012).The two smallest mustelids, Mustela erminea and Mustela nivalis are generally regarded as euryoecious, however the stoat usually prefers cooler conditions compared to the least weasel (ABRAMOV & BARYSHNIKOV 2000;KING & POWELL 2007).
A nearly complete lynx skeleton was also found in the Wierna Cave in Czêstochowa Upland (GA£USZKA 1992;STEFANIAK et al. 2009).Holocene remains of the species in the same upland were found in the layer 1-b of the Bioenik Cave (MARCISZAK et al. 2011).The only record of the lynx from Sudetic caves comes from the trench V of the NiedŸwiedzia Cave, where a maxilla fragment was found in postglacial/Holocene deposits (MARCISZAK et al. 2016).Pre-MIS 1 occurences of Lynx lynx are even less frequent.Well-preserved lynx remains dated at MIS 3-2 were discovered during the last ten years in sediments of the Ciemna Cave (VALDE-NOWAK et al. 2014;MARCISZAK et al. 2017).Recently, as a result of re-examination of the old material, some isolated teeth and postcranial bones were found in late Middle Pleistocene layers of the Bioenik Cave.However, the material has not been described in detail so far and for now it can only be classified as an undetermined member of the genus Lynx.
Felis silvestris is a permanent component of palaeontological assemblages, but usually it is represented by a rather scanty material.Most of the so called "fossil" specimens, whose age was estimated based on biostratigraphical or other indirect evidence, turned out to be postglacial or subfossil.The conjecture was confirmed by a few 14 C AMS dates, which showed a young age of those remains (KRAJCARZ et al. 2015b;MARCISZAK et al. 2017).Except very few bones, most specimens were younger than 12-10 ka.It seems that the earlier literature mentions of Pleistocene occurrence of the species were based on specimens from mixed layers.Examples are the following records: skulls from the Raj Cave (KOWAL-SKI 1967), a humerus and an ulna from the shelter under Tunel Wielki (NADACHOWSKI 1988), a mandible with a phalanx from the Ob³azowa Cave (LIPECKI & WOLSAN 2003), the Ciasna Cave (NADACHOWSKI 1988;NADACHOWSKI et al. 1989), two hemimandibles of the same individual from the Maszycka Cave (OSSOWSKI 1883; KOWALSKI 1951), a humerus from the Mamutowa Cave (NADACHOWSKI et al. 1989) and a calvarium, few mandibles and long bones from Zbójecka, Krakowska and Nietoperzowa caves (RÖMER 1883).
The species was also quite common in Sudetic caves.ZOTZ (1939) mentioned an unidentified cat from the deposits of the Obok Wschodniej Cave.The wildcat was identified based on a few isolated teeth and post-cranial bones from the Holocene layers of rock shelters Ma³gorzata and Trwoga Paleontologa (BIEROÑSKI et al. 2007), as well as the postglacial/Holocene layers of the Solna Jama cave and NiedŸwiedzia Cave (MARCISZAK et al. 2015(MARCISZAK et al. , 2016(MARCISZAK et al. , 2017)).Regarding the record from the Radochowska Cave, it is not even clear if it was the wildcat or domestic cat, because the identification was based on a femur (FRENZEL 1936).The bone was found in the surface strata of the sediments and was not fossilised.F. silvestris is very rarely found in Pleistocene deposits of Central Europe (BARYCKA 2008).So far wildcat remains of truly late Middle and Late Pleistocene age are known i.a. from different layers of the Bioenik Cave (MARCISZAK et al. 2011) and the Nietoperzowa Cave (KOWALSKI 1961;BARYCKA 2008).
Both cats, the lynx and the wildcat are relatively rare in cave deposits, which is typical of most medium and small cats.These forms, unlike e.g.leopard, rarely use caves as shelters to raise cubs.Small cats are solitary, shy, elusive hunters which usually avoid any competition with similar sized and larger carnivores such as wolves or wolverines, which is probably one of the major reasons for their rarity in cave sediments.The scarcity of their remains in cave deposits is due to poor penetration by these predators inside and around the caves.Due to their anatomy (a movable spine, very flexible limb joints, retractable claws) cats are able to survive a fall even from a high altitude and escape from caves which are a kind of trap such as karstic sinkholes.Because of their solitary, secretive lifestyle and relatively large territories, they were only occasional human victims, although their fur was (and is) regarded as valuable (BARYCKA 2008;MARCISZAK et al. 2011).

V. CONCLUSIONS
Twelve carnivore species documented the deposition in the Borsuka Cave during the last 27 ka.The carnivores from the Borsuka Cave form two different assemblages.The younger one, from Holocene layers I-IV, is dominated by remains of Vulpes vulpes, Meles meles, and Felis silvestris, with admixture of the less frequent Ursus arctos arctos, Martes martes and Mustela ex.gr.eversmanii-putorius.The older assemblage from layers V-VII, dated as MIS 2, is a mixture of species adapted to cold climate and steppe-tundra and/or tundra-open boreal forests (Vulpes lagopus, Gulo gulo, Mustela erminea), euryoecious forms (Canis lupus, Mustela nivalis, Ursus arctos arctos) and forest-dwellers (Martes martes and Lynx lynx).
Detailed morphometrical analyses show that many forms were associated with cold periods and cool climate conditions.The large size of Gulo gulo and Martes martes as well as the notably small dimensions of Mustela erminea and Mustela nivalis provide good examples.Likewise, the large size of the brown bear, despite the already simplified teeth morphology characteristic of Ursus arctos arctos, may reflect colder climate conditions.

_____________________________________
A c k n o w l e d g e m e n t s.We are very grateful to Prof. Beata M. POKRYSZKO for linguistic improvements and to the two anonymous reviewers for their helpful critical comments and suggestions.This work was supported by the the National Science Centre of Poland, to Adrian MARCISZAK under Grant "The history of the brown bear Ursus arctos LINNAEUS, 1758 evolution in Central Europe as a key to modern species conservation" no.2015/17/D/ST10/01907; by the Ministry of Science and Higher Education of Poland (project no 1076/S/IBOE/2016); by the Faculty of Biological Sciences, University of Wroc³aw under the internal grant for the young scientists no.0420/1406/16 and by the grant of the Ministry of Science and Higher Education of Poland no.5052/SPUB/IBOE/16.

Fig. 1 .
Fig. 1.Location of the Polish Jura in Europe (A) and in Poland (B), and location of the Borsuka Cave in the Polish Jura (C).Scale bars in km.

Fig. 14 .
Fig. 14.Total length of phalanx I plotted against minimum shaft breadth of phalanx I in European foxes and in foxes from the Borsuka Cave.Measurements in mm, according to LIPECKI and WOJTAL (2015) and own measurements.

Fig. 16 .
Fig. 16.Total length of P4 plotted against posterior breadth of P4 in different European chronoforms of Ursus arctos and in Ursus arctos arctos from the Borsuka Cave.Measurements in mm.

Fig. 18 .
Fig. 18.Total length of metatarsal II plotted against minimum shaft breadth of metatarsal II in European large mustelids Gulo gulo and Meles meles, and in Gulo gulo from the Borsuka Cave (marked as B on the graph).Measurements in mm, according to DÖPPES (2001); LIPECKI & WOJTAL (2015) and own data.

Fig. 22 .
Fig. 22.Total length of m1 plotted against m1 breadth in the wildcat Felis silvestris silvestris and the feral cat Felis silvestris catus from the postglacial and Holocene of Poland and the wildcat from the Borsuka Cave (indicated at the graph by B).Own data, all measurements in mm.

Fig. 23 .
Fig. 23.Aggressive encounter between the red fox Vulpes vulpes (on the left) and the polar fox Vulpes lagopus (on the right) in the vicinity of the Borsuka Cave.In general in canids, the larger species tends to dominate on the smaller one.Drawing by W. GORNIG.

Table V
Measurements of isolated canid teeth from the Borsuka Cave (mm).Measurements of teeth according toVON DEN DRIESCH (1976)

Table VII Mandible
and teeth dimensions of mustelids from the Borsuka Cave (mm).For measurements scheme see Fig.5(for mandible) and Fig.6(for teeth)