Preliminary report on the middle Pleistocene small mammal fauna from Yarimburgaz Cave in Turkish Thrace

[ V o r b e r i c h t ü b e r d i e m i t t e l p l e i s t o z ä n e K l e i n s ä u g e r f a u n a a u s d e r Y a r i m b u r g a z H ö h l e i n T ü r k i s c h T h r a z i e n ] Z u s a m m e n f a s s u n g Aus den Ablageningen der Yarim­ burgaz-Höhle werden zwei Kleinsäugerfaunen vorge­ legt. Die sehr begrenzte Fauna aus dem älteren Sedi­ mentationszyklus I enthält außer Arten, die auch heute noch in der weiteren Region verbreitet sind, vor allem Cricetus cricetus. Die jüngere Fauna aus dem Zyklus III, der einen großen Teil des archäologischen Materials ge­ liefert hat, ist auch faunistisch erheblich reicher. Die Fauna ist gekennzeichnet durch Einwanderer aus dem südrussichen Steppengürtel (Lagurus transiens/lagurus, Cricetus cricetus, Ochotona pusilla, und Sicista subtilis). Hinzu kommt Rattus rattus als Einwanderer von Süden. Die ökologische Interpretation läßt auf eine erhebliche Ausweitung der Steppenregion nach Süden schließen. Arvicola ist im Mittelmeerraum als stratigraphischer Indikator nicht brauchbar. Dagegen kann aus der Übergangsform von Lagurus transiens zu L. lagu­ rus ein mittelpleistozänes Alter abgeleitet werden. Da­ bei kann das frühere wie das spätere Mittelpleistozän ausgeschlossen werden.


Introduction
. The small mammals from their excavations were initi ally investigated by W. v. K. (KOENIGSWALD 1989) and than in more detail by W. S: (SANTEL in press).The large mammals are under study by C. HOWELL, M.C.STINER and E. TSOUCALA (STINER et al. 1996(STINER et al. , 1998)).
The two chambers of the cave contain sediments from the Lower and Middle Pleistocene.The cave was used until Byzantine times.The excavations were carried out in various squares (P,S,T,V,R, and U) in different parts of the cave.The stratigraphic sequences in the various squares could be correlated with each other except for square U.According to the sedimentological investigation of W. FARRAND (1989, and pers. comm. 1990, 1993), nine layers can be distinguished which be long to three different sedimentation cycles (cycle I, II and III).For the interpretation of the small mammal fauna square U is of greatest interest sin ce the layer U-l produced the majority of the small mammal specimens (about 80%).Even though the sediments of square U cannot be cor related with the stratigraphy of the other squares, the layer U-l clearly belongs to the youngest, se dimentation cycle III (HOWELL & ARSEBÜK 1989).Therefore most information derived from the small mammal fauna refers to the time of third se dimentation cycle only.
Two taphonomic factors led to the concentration of small mammals in Yarimburgaz Cave.Most of the small mammal remains were concentrated by owls dropping pellets at their roosting sites.Even though investigations of recent pellet accumulati on document quantitative biases in such accumu lations due to individual preferences of specific owls, their qualitative composition yields signifi cant data for a paleoecological and stratigraphical analyses.We are aware that some difficult to catch species, like the spiny hedgehog (Erinaceus), are often missing in pellets of smaller owls.Such biases can be tolerated, however only, if faunas from pellets are compared to each other.
The second taphonomic process accumulating small mammals in Yarimburgaz Cave only acts on bats.The taxa Myotis myotis, M. blythi, Rhinolophus ferrumequinum, R. mehelyi, and Miniopterus schreibersi were identified so far.Attritional mortality leads to carcass accumulation and buri al because the bats used the cave for hibernation.The bat fauna could not be interpreted in this paper.

The small mammal fauna
The following taxa were identified from Yarim burgaz Cave from the lower cycle I and the upper Cycle III.The letters U-l, X, W, etc. indicate the layers.
Cycle III: U-l, X, and W; Cycle I: S Clethrionomys glareolus (SCHREBER, 1780) Cycle III: U-l and X, not present in extant fauna.

Paleogeographical reconstruction of Cycle m
No identifiable small mammals are available from cycle II.The stratigraphically younger cycle III yielded the major part of the faunal remains.Most of the archeological material belongs to this sedi mentation cycle as well (HOWELL & ARSEBÜK 1989).Most specimen come from Level U-l.Sedimentological analysis indicates that cycle III was deposited in an relatively short period of time (FARRAND pers. comm. 1993) & KRAPP 1978& KRAPP , 1982& KRAPP , 1990)), a similar mixture of species is found today in Turkish Thrace.However, the percentage of steppic elements is distinctly higher in cycle III of Yarimburgaz Cave, indicating a landscape more open and dryer than today.In particular the grea ter steppic influence is indicated by taxa not pre sent in the area any more: Cricetus, Mesocricetus, Cricetulus, Lagurus, Sicista subtilis, and Ochoto na pusilla.Most of these species are characteristic for the Ukraine and the Russian plains north of Turkish Thrace.Although the steppic species oc cur even in semi-deserts, other species such as Sorex, Apodemus sylvaticus and Clethrionomys glareolus signalize the presence of at least restric ted areas with denser vegetation.In contrast to the many immigrants from the north, Rattus rat tus definitely immigrated from the south, most probably from the Levant where the earliest Rat tus is found in Oumm Quatafa (TCHERNOV 1994).Because of the wide range of Rattus, this genus seems not to be a good ecological indicator.In Central Europe, especially Lagurus and Ochotona are regarded as indicators for cool and dry clima tic conditions (HEINRICH 1983, SUTCLIFFE & KOWAL SKI 1976).Among the soricids Sorex reaches fart hest north.Crocidura leucodon and/or suavolens are more sensitive to cold climates and are absent during glacial periods in Central Europe.The pre sence of these taxa and of Mediterranean species like Myomimus roachi provide the lower tempe rature limit for the cycle III fauna.The southern shift of the steppic area to the Bosporus region as documented by the farina of cycle III most proba bly was caused by a general temperature decrea se and a reduction of rainfall.A cold phase with such a significant ecological shifts most probably reflects a glacial period.
One of the voles probably is traditionally des cribed as Microtus rossiaemeridionalis.
The difference of the extant taxon was based on karyological characters.Therefore it can not be examined in the fossil material.Since we want not to indicate the antiquity of the karyotype modi fication we use the traditional term based on mor phological characters.

Stratigraphically significant taxa of cycle III
In addition to the ecologically informative species the fauna of cycle III contains some species often used for stratigraphie purposes.These are Arvico la terrestris, Lagurus transiens/lagurus, Ochoto na pusilla, and Rattus rattus.
Arvicola terrestris (Fig. la) In Arvicola specimens form Central Europe, the relative thickness of the enamel band in leading and trailing edges of the dentin triangles was used as a stratigraphie indicator (KOENIGSWALD 1973, HEINRICH 1982, 1987).The more primitive situati on with a thick trailing edge was regarded to cha racterize the Middle Pleistocene Arvicola cantianus, while the Upper Pleistocene Arvicola terre stris shows a significant reduction of the thickness of trailing edge.In the Mediterranean, especially in Spain and Turkey, the underived condition is still present in the extant fauna (ROETTGER 1987).The Spanish form is identified as Arvicola sapidus (MILLER 1908), the Turkish one as Arvicola terre stris persicus (DE FILIPPI, 1865).Arvicola from Yarimburgaz Cave, represented by 7 molars, shows thickened trailing edges like A.
cantianus but in the Mediterranean this character obviously cannot be used in age assignments.Neither can the absence of the primitive Mimomys-lcM.in two lower Ml (length 3,84 and 3,77 mm), be used as a stratigraphie indicator, sin ce this trait is rare in A. cantianus.
In Turkey it is difficult to distinguish between A. cantianus, and A. terrestris.As KOLFSCHOTEN (1988,1990) accepts both taxa as subspecies of A. terrestris only the Yarimburgaz Arvicola is not identified in more detail.
Since the genus name Arvicola is male the ending of the species name has to be male too therefore Arvicola cantianus is the valid form.
mm with a mean of 2,45) is intermediate between both species if measurements from Central Euro pe are used (CHALINE 1972, HEINRICH 1990, JANOSSY 1962).Lagurus lagurus form Chios described by STORCH (1975) is slightly larger than the Yarim burgaz material.The morphology of the anterior lobe can be studied is seven M.. Two of them (Fig. If the material that exclusively came from level U-1, is regarded as representative, the Yarimburgaz Rattus rattus (Fig. lb) A left upper maxilla with all molars in place mor phologically matches Rattus rattus so well that an identification is beyond questioning.It differs from Rattus norvegicus in several morphological characters of the M.Although the measurements of the molars fall into the range of extant Rattus rattus horn western Anatolia (FELTEN et al. 1971), contermination by recent material is not proba ble, since the maxilla is well fossilized like all other bones from layer W. Rattus is rare in the Me diterranean Middle Pleistocene and missing in Central Europe.Rattus haasi was described from Qumm Qatafa in Israel (TCHERNOV 1968).How ever, as it is represented by lower molars only it is difficult to compare with our material.STORCH (1975) attributes upper and lower molars from Chios to Rattus rattus.These teeth are somewhat larger than the Yarimburgaz material but are still in the range of R. rattus in Anatolia.
Ochotona pusilla (Fig. lc) This small lagomorph is represented by several isolated molars, one P3, and 7 mandibular frag ments.The species O. pusilla can be identified by the morphology of the lower P3, especially by the deep lingual reentrant fold separating an enlarged posterior part from the medial part and by the rhomboid shape of the anterior part (Fig. lc).This shape is known in O. pusilla from the Middle Pleistocene from Central Europe, from Stränskä Skala (TOBIEN 1972) and Hunas (Heller et al. 1983).The length of the alveolar tooth row in the lower jaw is very similar in Middle and Upper Pleistocene material attributed to O. pusilla.
Ochotona is known in the Mediterranean Middle Pleistocene in Central Anatolia (MONTEURE et al 1994) and from Qumm Qatafa in Israel (TCHERNOV 1994) but is missing in Chios (STORCH 1975).Ochotona pusilla which occurs today in the Rus sian plains, did not reach the Eastern Mediterra nean during the cold periods of the Late Pleisto cene.

Age Assignment
The stratigraphie position of the cycle III of Ya rimburgaz Cave is marked best by the transition of Lagurus transiens to L. lagurusin level U-l.In the Ukraine and in the plains of southern Russia, the transition between L. transiens and L. lagurus oc curs in the Gungki-Complex, containing the Oka-Glacial and the Likhvin interglacial period of the Middle Pleistocene (MARKOVA 1990(MARKOVA , 1992)).The lo wer Middle Pleistocene starting at the Brunhes-Matuyama boundary can be excluded because at this time Mimomys savini is still present.Arvicola occurs only in the latest interglacial periods of the Cromerian complex.The upper part of the Midd le Pleistocene in Russia is characterized by the presence of the derived Lagurus lagurus.In Ya rimburgaz Cave cycle III, level U-l, documents the transition from Lagurus transiens to L. lagu rus.Therefore the fauna allows to narrow down the age of cycle III in Yarimburgaz Cave to a cold period in the middle of the Middle Pleistocene.The relative stratigraphie position of Yarimburgaz Cave to some Middle Pleistocene localities of the region is of biogeographic interest.The fissure filling of Emirkaya-2 near Seidisehr in Central Anatolia was recently published on by MONTUIRE et al. (1994).The fauna seems to be de finitely older than Yarimburgaz Cave cycle III if the molar of Mimomys savini belongs to this as semblage.Among the many species this fauna shares with the Yarimburgaz Cave cycle III fauna, Ochotona and three cricetine species are of signi ficance.As an additional element of the steppe environment Allactaga occurs in the Anatolian fauna.Therefore the ecological interpretation of Emirkaya-2 as representing an open and humid environment with a forest in the vicinity has to be questioned.The stratigraphie assignment of this fauna to the Holsteinian, corresponding to the Mindel-Riss interglacial period, cannot be used as a stratigraphie marker since at the time being no vertebrate fauna in Central Europe can be attribu ted with certainty to the Holsteinian.Even the Holsteinian of northern Germany can not be cor related with sufficiency to the glacial sequence of the Alpine region (KOENIGSWALD 1992).
From the Aegean Island of Chios, which was part of the mainland during the Middle Pleistocene, STORCH (1975) described a fauna from the fissure filling Latomi-1.It is assumed to be not older than the middle part of the Middle Pleistocene.STORCH (1975)  This may suggest that the fauna of Qumm Qatafa is older than the cycle III from Yarimburgaz Cave, even if such a correlation bears great uncertain ties.

Biogeographical implications
Even if Yarimburgaz Cave occupies a veiy margi nal position in Europe, the occurrence of several taxa predates the estimated immigration time into this continent, as given by HOSEY (1982).While Mimomys roachi, Apodemus mystacinus, Micro tus guentheri, and Mesocrietus were assumed to have crossed the Bosporus at the lowest sea level stand during the last glaciation, the finds from Ya rimburgaz Cave prove that these species were present west of the Bosporus much earlier.The occurrence of Talpa levantis in the Middle Plei stocene of Yarimburgaz Cave may strengthen the idea of VOHRALIK (199D, who postulated that T. le vantis originated in Asia Minor and penetrated the Balkan region.The small number of Middle Pleistocene localities neither permits the identifi cation of the route of immigration nor of the exact time.Because of the several climatic oscillations combined with sea level fluctuctions, several wa ves of immigration must be postulated.The cli matic oscillations may even have caused partial extinctions between these immigrational phases.It is almost impossible to reconstruct the faunal movement without a large set of densely and well dated localities.The fauna of small mammals from Yarimburgaz Cave is biogeographically very important since it is almost the only fauna in the region from this time period.Together with other localities to be discovered in the future, it will sig nificantly contribute to an improved biogeogra phical recornstruction.

Yarimburgaz
Cave is located in the Bosporus area 25 km west of Istanbul in Turkish Thrace.The cave has been known since last century, but the * Anschriften der Verfasser: Dipl.-Geol.W. SANTEL, Prof. Dr. W. v. KOENIGSWALD, Institut für Paläontologie der Universität Bonn, Nussallee 8, D-53115 Bonn, main scientific excavations were carried out by C.
2 a-b) fall into the range of variability of Lagurus lagurus, but four show the pattern typical of La gurus transiens (Fig. 2 e-h), characterized by the missing or fairly slight reentrant folds of the cap of the anterior lobe.
is in the Russian plains north of the 44°.A detailed stratigraphical assignment can not be given for this faunal assemblage.Its composition indicates a Middle or Upper Pleistocene age be cause of Cricetus cricetus and Cricetulus migra torius.
Most species of this assemblage occur in the extant fauna of the general region.Apodemus my stacinus occurs only on the eastern side of the Bosporus and on the Balkan peninsula.Cricetus cricetus is the only species indicating a somewhat more continental and drier environment for layer S in Yarimburgaz Cave since its nearest.present day occurrence . The following species are present.
TCHERNOV (1968TCHERNOV ( , 1994) )ios, which is more derived than the transitional form found at Yarimburgaz Cave.Therfore we argue that the Chios fauna is somewhat younger than the fauna of cycle III in Yarimburgaz Cave.Form Israel,TCHERNOV (1968TCHERNOV ( , 1994) )described a fauna from Qumm Qatafa in the Judean desert near Bethle hem.So far no distinct stratigraphie position is known for Qumm Qatafa.It is difficult to compa re this fauna with Yarimburgaz Cave since it is si tuated almost 10 s further south.The Israeli fauna contains several African elements.Nevertheless, it shares two important taxa with the fauna of Ya rimburgaz Cave.These are Ochotona as a nort hern immigrant and Rattus as an immigrant from the southeast.On the other hand, instead of the cricetines present in Yarimburgaz Cave, in Qumm Qatafa more primitive cricetines occur.The spe cies at Yarimburgaz Cave occur in Israel much la ter, for instance in Hayonim (TCHERNOV 1994).
(KOENIGSWALD 1973).In the thickness of the enamel band, the trailing edges surpass the leading edges, as is ty pical for a. cantianus in Central Europe.As dis cussed above, this character is of little significan ce in the Mediterranean.Therfore we do not con sider the identification of A. cantianus at Chios to indicate that this fauna may be older than the cy-cle III fauna from Yarimburgaz Cave.Lagurus of fers a better correlation.STORCH (1975) identified