First record of the root vole Microtus oeconomus ( PALLAS , 1776 ) ( Arvicolidae , Rodentia , Mammalia ) in Croatia – fossil remains from the Late Pleistocene of Vindija Cave

Late Pleistocene remains of Microtus oeconomus (PALLAS, 1776), a species that is not any more a member of the extant fauna of Croatia, have been recently found in Vindija Cave sediments. The sample of 21 teeth shows a slight increase of M1 length and decrease of A/L ratio within the cave sequence from older (G complex; 32 to 42 ka) towards younger strata (E+F complex; layer E is dated at ca. 18.5 ka). The most frequent morphotype is type B sensu NADACHOWSKI (1982). The new record does not allow further detailed ecological implications than what is already known about the Late Pleistocene habitats in this region.

Microtus oeconomus is the only extant Microtus species with a holarctic distribution and ranges from northwest Europe, eastward through north and central Asia to Alaska and northwest Canada (BRUNHOFF et al., 2003).Four phylogeographical groups have been identifi ed and traced back to several glacial periods.They refl ect late glacial and postglacial range expansions and shifts as well as fragmentation of populations (idem).

GEOLOGICAL SETTINGS
The geographic position of the cave is 46° 18' 12'' N, 16° 14' 38'' E, and the entrance is 275 m a.s.l.(PAUNOVIĆ et al., 2001;WILD et al., 2001, and references therein).The cave consists of a single, funnel-shaped chamber over 50 m long, 28 m wide and 20 m high, although prior to exca vation, the cave roof was less than 3 m above the surface (MIRA-CLE et al., 2010).
Thirteen layers (from the youngest A /Holocene/ to the oldest M /MIS 6 or older/) show different sedimentation cycles, and comprise faunas of different composition.Microtus oeconomus teeth have been uncovered from stratigraphic units I, G lower, G upper (inclusive signature G2), complex E+F and layer E (after original signature; see Fig. 2).

Measurements:
The overall tooth length (L), the length of anteroconid complex (a) and width (W) have been measured and A/L index was calculated (according to VAN DER MEULEN, 1973).All measurements are in millimetres (mm).The parameters have been statistically calculated with STATISTICA 6.0    StatSoft.Inc., 1994-2001 for Windows).On graphs standard deviation (SD), standard error of the mean (SE) and mean values are indicated.
Morphotype D: with BSA4 very well developed, connection between T5 and T6 is very narrow (not confl uent).
The fossil material is stored at the Institute for Quaternary Palaeontology and Geology of the Croatian Academy of Sciences and Arts in Zagreb.

Measurements:
First lower molars (M 1 ): L, a, W and A/L values are given in Figs. 4 and 5, and Tab. 1. Measured parameters were grouped and compared for two complexes -E+F and G. Mean values for L, a and W are greater for the samples from the E+F complex than for the G complex (Figs. 4 and 5), while for the A/L index is opposite.

Second lower molars (M 2 ):
Only two M 2 (Tab.2) with values of L and W enable no morphometrical analysis.

Morphotypes:
On 19 M 1 of Microtus oeconomus, (12 of them are from the G complex), morphotypes B, C and D sensu NADACHOW-SKI (1982) have been identifi ed (Fig. 3; Tab.3), while morphotype A is completely absent.In the small sample, the most abundant morphotype is B (10 teeth or 52.6 %).

DISCUSSION
Based on M 1 morphology alone, a specifi c determination between Microtus oeconomus and Chionomys nivalis might be in some cases problematic, since (as shown by ANGER- MANN, 1971) that both species share common M 1 morphotypes.In his paper about Late Quaternary rodents of Poland, NADACHOWSKI (1982) elaborated this problem in fossil material.The "oeconomus« morphotype is the most frequent in the Polish Late Pleistocene samples, but decreases during the Late Glacial.NADACHOWSKI (1982) assumes that at the margin of the species range, non-typical morphological features occur more frequently, which parallels the situation during the last glaciation when under the infl uence of ameliorating climate the root vole retreat to the north, and the frequency of the "alien« morphotypes increased to 36 %.Also the extreme frequency of the nivalid morphotype (35 %) is observed in recent populations inhabiting the tundra zone (NADACHOWSKI, 1982).The comparison between the frequency of Microtus oeconomus of various Polish sites shows even greater differences (from 2-5 % to 46 %), but up from the Upper Pleniglacial a gradual decrease is observed: from 10-30 % in the Upper Pleniglacial, to 4-17 % in the Late Glacial and eventually to 1-8 % during the Holocene (NADACHOWSKI, 1982).However, analysing the vole dentition from Bacho Kiro cave (Bulgaria) NADA-CHOWSKI (1984) corrected the previous determination of M 1 of Microtus cf.oeconomus to extreme variants (morphotype A) of the snow vole Chionomys nivalis (MARTINS,  1842) based on data of fossil and recent samples from Switzerland, the Tyrol, Murańska cave and other localities in the Tatras, Greece, Romania, Bulgaria etc.
In the M 1 of Vindija Cave, the most frequent morphotype is B (52.6 %; four closed triangles), morphotype C is less abundant, morphotype D is represented by one tooth only, and morphotype A is completely lacking.These data would fi t roughly between the values of the Late Glacial and recent condition in Polish samples, documented by NADA-CHOWSKI (1982): Lower Pleniglacial (LP) -61.7 %, Upper Pleniglacial (UP) -60.9 %, Late Glacial (LG) -46.8 %, recent (R) -59.2 %.
The metrical analysis of M 1 of Microtus oeconomus shows an increase of the length and a decrease of the A/L index from the older G to the younger E+F complex.Also NADACHOWSKI (1982) observed a distinctive increase of M 1 length during the Late Quaternary of Poland (with the smallest specimens in the Lower Pleniglacial, and the biggest in the extant fauna -idem).The mean A/L values from Poland fl uctuate between 48 and 49 but the differences between the groups are not statistically signifi cant.In the Vindi ja sample, the maximal values are somewhat smaller, and minimal values are slightly bigger compared to the Polish samples, whereas the mean values remain more or less similar.However, L and A/L are smaller in the Croatian samples compared to the Polish ones (Tab.4) (NADACHOWSKI, 1982;MAUCH LENARDIĆ, 2005).Comparison of M 1 length gives similar mean values for the Vindija E+F complex sample and recent German specimens (Tab. 4;TAST, 1982).
According to SPITZENBERGER & BAUER (2001), a morphological rather similar ancestor of Microtus oeconomus fi rst appeared in the Lower Pleistocene, and specimens of the present shape during the Middle Pleistocene respectively.They quote the oldest 'true' M. oeconomus as being from the volcanic ash in the open-site Miesenheim near Koblenz (Rheinland-Pfalz, Germany), dated at 612 ka.Microtus oeconomus belongs to the Holarctic Pleistocene tundrasteppe fauna, which existed during the LG (more than 55 ka), and therefore also occurred on the North American continent.The species also occurs in some Late Pleistocene Austrian localities (Gamssultzenhöhle, Grossen Ofenbergerhöhle: Upper Würmian, Nixloch: Dryas -10.55 ka until Holocene) (SPITZENBERGER & BAUER, 2001).Within the Late Pleistocene and Holocene, several populations west and south of the recent main distribution of M. oeconomus became isolated.In southern Germany, the Czech Republic and Hungary, the species was still recorded in the Holocene, in Moldova until the Neolithic, and in southern France, as the last boreal element of Ice Age fauna, the species disappeared ca.7000 years ago (SPITZENBERGER & BAUER, 2001).Recent investigations in Hungary have confi rmed the presence of some isolated (sub)populations, one near Balaton lake (NE), and a second (Barcs), about 100 km east of the Vindija site (BIHARI et al., 2007).
The distribution is confi rmed by NADACHOWSKI (1982) who mentions a Late Pleistocene expansion of the range of Microtus oeconomus further west than today: in Central Europe, England, France and Switzerland as well as further south in Hungary and Romania.The species survived here longer than many other tundra adapted species; e.g., in Hungary it still occurred during the Subboreal (Bükk Phase) in the Bükk mountains, and in the caves of the Aggtelek Karst (NADACHOWSKI, 1982).At the beginning of the Holocene, the distribution of this species was divided into two parts: a tundra range and a forest-tundra range of Europe, parts of Asia and North America, with isolated populations in the Netherlands, Austria and the Small Hungarian Lowland (NADACHOWSKI, 1982).
Microtus oeconomus is an indicator of a cold climate and tundra-forest environments with abundant water, respectively (NADACHOWSKI, 1982).Tundra (root) voles appear to have an extremely wide tolerance to different habitats, but may be locally restricted by competition with other microtines (GUTHRIE, 1968).In some regions, Microtus oeconomus could be restricted to low wetland communities, unforested areas, wet meadows, around lakes, streams or mar sh es (GUBÁNYI et al., 2007).
Microtus oeconomus has been determined for the fi rst time in Croatia in the Late Pleistocene sediments of the Vindija Cave (MAUCH LENARDIĆ, 2004, 2005), and it is not part of the extant fauna in the vicinity of the cave (PETROV, 1992).Thus the fi ndings from Vindija Cave imply a somewhat colder and more wet climate in this region during some phases of the Late Pleistocene than today.The youngest strata from which this species is determined in Vindija is the E+F complex (layer E is dated as ca.18.5 ka).There are no records of this species in younger layers, neither are there in the faunal remains from other Croatian localities with contemporaneous strata, such as Marlera I (southern Istria), the Mu jina pećina cave (middle Dalmatia; MAUCH LEN AR-DIĆ, 2005, 2008) or Vela spila on Korčula island, respectively (MAUCH LENARDIĆ, in press).

Figure 4 :
Figure 4: Graph of mean values, standard error of the mean values (SE) and standard deviation (SD) for length L (a) and length of anteroconid complex a (b) of Microtus oeconomus M 1 for the Vindija Cave E+F and G complexes.

Figure 5 :
Figure 5: Graph of mean values, standard error of the mean values (SE) and standard deviation (SD) for width W (a) and A/L index (b) of Microtus oeconomus M 1 for the Vindija Cave E+F and G complexes.

Table 1 :
Minimum (min.), mean and maximum (max.)values of L and A/L for Microtus oeconomus M 1 from Vindija Cave.

Table 2 :
Measurements (L and W)for Microtus oeconomus M 2 from Vindija Cave.