Late Pleistocene climate history of the Baranja loess plateau - evidence from the Zmajevac loess-paleosol section (northeastern Croatia)

Headnote

AB strA ct

The Zmajevac loess-palaeosol succession (LPS) of the northeastern Baranja loess plateau is exposed along the southern slope of Bansko Brdo, on the western bank of the Danube River. The investigated 17.5-m-thick section shows 4 palaeosol, 1 loess-like and 6 loess horizons. Their integrative palaeoenvironmental analysis combines quantified data from the mollusc record, magnetic susceptibility, grain-size, calcimetry and mineral abundances to reconstruct the pattern of regional palaeoclimate evolution. This result combined with infrared optically stimulated luminescence age estimates by GALOVIÆ et al. (2009) enabled correlation of the depositional units to Middle to the Late Pleistocene Marine Isotope Stages (MIS) 6 to 2. Magnetic susceptibility measurements show strong peaks in the palaeosol horizons pointing to increased concentrations of pedogenic ferrimagnetic minerals. Sedimentological and mineralogical parameters are in good agreement with other Pannonian Basin LPS. Terrestrial gastropod palaeoecology based on 1705 specimens of 13 species counted from loess and loess-like horizons documents cyclic transitions between cryophilous to cold resistant and mesophilous to thermophilous assemblage types. Whereas Helicopsis striata, Arianta arbustorum and Chondrula tridens are common throughout the succession, the typical loess representatives Pupilla sp., Vallonia tenuilabris and Columella columella are abundant only in certain horizons. Nevertheless, species tolerating open and dry habitats are abundant throughout the succession. The faunal spectra for the samples prove the dominance of transitional palaeoecological assemblage types, whereas uniformly defined types are rare. One of these, the Columella columella assemblage from the base of the section proved to be indicative of the Penultimate Glacial Maximum.

Keywords: mollusc palaeoecology, magnetic susceptibility, grain-size, modal analysis, loess, climate change, Danube, Croatia

1. INtrODUctION

Loess is terrestrial clastic sediment, composed predominantly of silt-size particles, formed by the accumulation of wind-blown dust (PYE, 1995). The Pleistocene loess deposits show two main genetic types: (1) loess formed under periglacial conditions, commonly in front of the ice caps and (2) loess built up by small particles formed in the mountain areas. The second type of loess is present in the study area and also in the wider region (MARKOVIC et al., 2005, 2008, 2009; GALOVIC et al., 2009, 2011; SMALLEY et al., 2011, WACHA & FRECHEN, 2011). Actually, loess covers up to 10% of the world's surface area (PÉCSI, 1968), and is usually inter-bedded with soil horizons. Such successions provide very detailed terrestrial records of Pleistocene climatic fluctuations (KUKLA, 1987; GUO et al., 2002).

Pleistocene sediments are widespread in Croatia covering about 35.7% of its territory (BOGNAR, 1976). In the southern Pannonian Basin (Northern Croatia) they dominantly cover large continuous areas, whereas in the Dinaride Mountains and Adriatic Islands (Southern Croatia), Pleistocene deposits are restricted to small areas. Aeolian silt and sand were blown into the lowland steppe, lakes, marshes and swamps (BAÈANI et al., 1999). The region of Baranja in northeastern Croatia (Fig. 1) is almost completely covered with Pleistocene and Holocene sediments. In particular, some of the thickest loess successions of Croatia are exposed along the Danube River. The total thickness of these loess and loess-like deposits probably exceeds 50 m.

Loess-palaeosol successions of Baranja and northeastern Croatia have long history of investigation (WACHA & FRECHEN, 2011 and references therein). MALEZ (1965) analysed periglacial phenomena in eastern Croatia. RUKAVINA (1983) investigated loess-palaeosol successions and their mollusc fauna, providing an overview on warm periods in the Late Pleistocene of northeastern Croatia. POJE (1982, 1985, 1986) focused on the molluscan fauna from loess-palaeosol sequences at the Vukovar and Ðakovo loess plateaus, south of Baranja. BRONGER (1976, 2003) correlated loess-palaeosol sequences in East and Central Asia with others in south-east Central Europe, which are very close to the Baranja region. GALOVIÆ et al. (2009) focused on the chronostratigraphy of loess deposits of Zmajevac on the southern slopes of the Bansko Brdo in Baranja. From the same section MOLNÁR et al. (2010) investigated mollusc abundance from three palaeosol-related horizons. Finally, WACHA & FRECHEN (2011) provided luminescence dating for Middle to Late Pleistocene sections of Vukovar, while GALOVIÆ et al. (2011) investigated a loess-palaeosol section in Sarengrad.

The aim of this study is the reconstruction of climate changes during the latest Middle and Late Pleistocene based on the mollusc composition of different loess horizons of the Zmajevac PLS. This data will be integrated with magnetic susceptibility measurements, grain size analyses, mineral abundances (modal analysis), and calcimetry to provide a better control on palaeoclimate interpretations. The infrared stimulated luminescence (IRSL) data of Galovic et al. (2009) will be finally compared with present results and integrated into a chronological framework for the Zmajevac LPS whereby a correlation of its depositional units with the Marine Isotope Stages of MARTINSON et al. (1987) will be discussed.

2. GEOLOGIcAL sEttING

The surface geomorphology of Baranja is monotonous consisting largely of Pleistocene and Holocene sediments, with some minor outcrops of Miocene and Pliocene sedimentary and magmatic rocks (VELIC & VLAHOVIC, 2009) (Fig. 1). Tectonically it belongs to the Pannonian Basin, accommodating first Central Paratethys marine deposits during the Middle Miocene, then Lake Pannon brackish deposits in the Late Miocene and finally fresh water sediments of Lake Slavonia in the Pliocene completing the subaquatic deposition (ROYDEN, 1988; PILLER et al., 2007; HARZHAUSER & MANDIC, 2008; MANDIC et al., 2011).

Bansko Brdo is the only morphological high in the area and represents part of a tectonic horst stretching 20 km in a NE-SW direction, reaching the banks of the Danube River to the NE end. Its elevation reaches 244 m (Fig. 1). A combination of active neotectonic uplifting and Danube river erosion exposed large outcrops of a Pleistocene loess-palaeosol succession on Bansko Brdo. The hill is surrounded by a Holocene fluvial, oxbow lake, and marsh deposits.

The oldest exposed rocks at Bansko Brdo belong to the Miocene volcano-sedimentary complex, and include basalt-andesite and pyroclastic rocks comprising volcanic and tuffaceous breccia and conglomerates. K-Ar radiometric, whole-rock measurements indicate an early Middle Miocene age (13.8±0.4 and 14.5±0.4 Ma, PAMIC & PÉCSKAY, 1996). This complex probably originated from partial melting of the heterogeneous lower crust due to continental rifting processes during Pannonian Basin extension (HORVÁTH, 1995, PAMIC, 1997; PAVELIC, 2001; MANDIC et al., 2011). The basalt-andesite intercalates Middle Miocene (Badenian) marine calcareous sand and marl, indicating the synsedimentary character of the volcanic activity (PAMIC & PIKIJA, 1987; LUGOVIC et al., 1990). Furthermore few outcrops of Pliocene fresh-water, coarse-grained clastic deposits occur at Bansko Brdo.

The Pleistocene loess is the most dominant deposit in the geological pattern of Bansko Brdo (VELIC & VLAHOVIC, 2009). The mean thickness of the loess-palaeosol sequence is 28 m for the outcrop section. The oldest loess at the Zmajevac locality produced an infrared optically stimulated luminescence (IRSL) age of 217±22 ka, while the youngest loess has an IRSL age of 16.7±1.8 ka (GALOVIC et al., 2009).

3. sAMPLING AND MEtHODs

Section logging and sampling was carried out in 2009 and 2010. Thereby the depositional units and their internal sedimentary characteristics have been described including photographic and GPS documentation. Data on the vertical changes of magnetic mineral content in the section was gathered from 44 samples collected into 200 ml plastic containers (KUKLA, 1987). Magnetic susceptibility measurements were performed using a Bartington MS 2 laboratory device. Sample material was decanted into, 100 ml plastic containers before fixing to the measuring module with a calibrated standard. Each sample was measured three times for precision. Data was processed with the Bartington software. In general palaeosol horizons show enhanced MS values due to increased concentrations of pedogenic ferrimagnetic minerals (GEISS & ZANNER, 2006), while the loess units are marked by a significantly decreased signal. The variations are related to soil forming processes and reflect differences in composition, concentration, and particle size of the magnetic minerals, between interglacial/interstadial and glacial/ stadial sediments (KUKLA, 1987; EVANS & HELLER, 2001). Therefore magnetic susceptibility may reveal patterns congruent to the SPECMAP marine oxygen-isotope record (MARTINSON et al., 1987), and can provide an excellent stratigraphic correlation tool (MARKOVIC et al., 2008, 2009, 2011).

Thirteen bulk sediment samples (8-10 kg) were collected from loess and loess-like horizons for every 1.5 m of the Zmajevac LPS with at least one sample per defined lithostratigraphic unit. Grain-size analyses combined wet sieving and the pipette method. Classification of the grain size distribution follows WENTWORTH (1922). Mineral abundances (modal analyses) used the 0.063-0.125 mm calcite-free fraction. Heavy and light mineral fraction (HMF and LMF) were separated in bromoform liquid (CHBr3, δ=2.86 gcm-3) by gravity. Qualitative and quantitative analyses of the fractions were based on 300-350 grains per sample and were conducted using a polarizing light microscope (MENGE & MAURER, 1992). The carbonate (CaCO3) content was calculated from the weight difference before and after cold hydrochloric acid (5 %) treatment. Mollusc shells were derived from samples by screen-washing in distilled water, using a 0.7 mm mesh sieve. As proposed by LOZEK (1964), beside completely preserved shells only larger fragments including apices and apertures were picked out and counted. Identifications are based on comparison with collection material at Natural History Museum in Vienna, following the taxonomic concepts of LOZEK (1964) and FRANK (2006). Assemblage analysis follows LOZEK (1964).

4. rEsULts

4.1. Description of the section

The loess-palaeosol succession is situated 450 m south of the Zmajevac section of GALOVIC et al. (2009) and MÓLNAR et al. (2010). It outcrops in the vertical wall beside the road from Beli Manastir to Batina (Fig. 1). GPS coordinates of the bottom point of the section are 45° 48' 38'' N and 18° 49' 7'' E. Its altitude is 92 m. The 17.5-m-thick section exposess 6 loess, 1 loess-like and 4 palaeosol horizons (Fig. 2). The lithostratigraphically defined loess/loess-like horizons are coded from top to the base as L1 to L7, the palaeosol horizons-from P2 to P4. The third palaeosol representing the loess L3 intercalated doublet, received codes P3a and P3b. F1 marks the Holocene soil.

The lowermost unit of the section is the 80 cm-thick light yellowish brown loess L7. In the upper part of the unit carbonate concretions are concentrated. They attain 15-20 cm in diameter, are rounded and somewhat flattened. Mollusc remains and carbonate root casts are also present. The overlying palaeosol horizon P4 is 50 cm thick and light brown to yellowish brown in colour, somewhat lighter in its basal part. L6 follows upsection represented by the 60 cm-thick, light yellowish brown loess. The unit contains relatively abundant mollusc fauna, and a small amount of Fe-Mn concretions and nodules 1-2 mm in diameter.

The light grey to pale yellowish 240 cm-thick loess-like unit L5 follows upsection. Dark brown Fe-Mn nodules and small concretions 1-2 mm in diameter are common. Stems tentatively ascribed to a fossil reed are preserved as limonitized, pale orange to light brown tubes 5-10 cm in length and 0.5-1 cm in diameter. Molluscs are rare in this unit, but several fragments of mammoth tusk (up to 30 cm x 8-10 cm in size) were discovered in its middle part. The surface of the tusks is pale brown to yellowish brown in colour, while the core is white to pale yellow. Laterally, some smaller, 5-6 cm-long tusk fragments were also discovered.

The next loess unit L4 is 40-50 cm thick and light yellowish in colour. It bears abundant white to pale yellowish, carbonate concretions measuring up to 10-15 cm in diameter, with variable, but mostly rounded form. Mollusc remains are rare. A well developed, 180 cm thick pedocomplex follows consisting of two palaeoesol horizons (P3a and P3b) separated by a 40-50 cm-thick light, yellowish brown, loess horizon L3, with a fairly rich mollusc fauna. The top 50 cm of the upper paleosol F3a is brown to light brown, and for the lower 30 cm light brown, to yellowish brown in colour. F3b is brownish in its upper 40 cm and light brown to yellowish brown in its lower 40 cm.

The next horizon L2 is composed of 480 cm light yellowish loess, rich in molluscs underlain by 50 cm horizontally laminated sand, which is fine-grained and light yellow in colour. The stratification is represented by laminae-separated into 1-3 cm-thick sediment packages. L2 is overlain by a 60 cm thick, weakly developed, olive brown to yellowish brown palaeosol horizon P2, with no mollusc remains. Finally, the uppermost 450 cm-thick unit L1 is composed by light yellowish brown silty loess. This unit is rich in mollusc fauna and bears carbonate filled tubes of fossilized root remains.

4.2. Grain-size, carbonate content and mineral distribution

Grain-size analyses indicate silt as the dominant grain-size fraction in all 13 studied loess samples (Fig. 3). Of these, seven (mostly from loess horizons L1 and L2) contain small amounts of clay-sized minerals. The abundance of sandsized particles ranges from 5.5% to 15%. The laminated unit in the base of L2 is composed of 81% sand, 11% silt and 8% clay, with a median grain-size of 0.22 mm. The four lowermost loess samples indicate a higher percentage of coarse silt and sand-sized particles in reference to all loess samples. Median grain-size is largely constant between 0.025-0.03 mm, with exception for 4 lowermost samples showing significant increase, with the peak in L5 horizon (72% large coarse-silt and sand-sized fraction). Skewness is fairly constant in all 13 samples, averaging of 0.78. Sorting is dominantly poor, with an average value around 1.5.

CaCO3 content of the samples ranges between 2.9% and 23.3% (Fig. 3), with an average value of 9.3%. The highest value is measured in the base of the uppermost L1 loess horizon, while the lowest (2.9%) was measured in the L4 loess horizon, bellow the F3b palaeosol. The CaCO3 content is in accordance with typical values for loess deposits, with the exception of 23.3% in the upper part of the L2 loess unit.

Modal analysis pointed out that the light mineral fraction (LMF) is dominant in all samples, whereas the heavy mineral fraction (HMF) range, from 4.15% to 11.21% (Table 1). Quartz is the dominant LMF mineral ranging between 50% and 74% (mean = 59.5%). The feldspar group is the second most abundant ranging between 6% and 32% (mean = 17.1%). The rock particle content is between 9% and 24% (mean = 16.7%). Muscovite is present in all the analysed samples ranging from 1% to 21% (mean = 7.1%). In the HMF, the transparent heavy minerals (THM) are more common than opaque ones, whilst chlorite and dolomite are rare. The most common THM are the mineral groups of epidote, garnet and amphibole. The epidote group is most abundant and ranges from 17% to 37% (mean = 27.5%), garnet group ranges from 16% to 61% (mean = 26.5%), amphibole group from 1% to 35% (mean = 26.1%). Chlorite is present in all samples ranging from 1% to 14% (mean = 4.92%), with mean in L1, L2, and L3 distinctly higher (6.44%), than in L4, L5, L6 and L7 (1.5%).

4.3. Magnetic susceptibility

Magnetic susceptibility (MS) values from loess and loesslike sediment range from 5 to 28.5 x 10-6 SI (Fig. 6). L1 horizon from the upper part of the section ranges from 15 to 20 x 10-6 SI. The uppermost palaeosol horizon (F2) shows the highest measured values within the section (82.5 x 10-6 SI). Values in the loess unit L2 are again much lower, with a mean of 14 x 10-6 SI. A notable peak within this horizon is 28.5 X 10-6 SI. The pedo-complex constituents P3a and P3b are marked by significant peaks that are however lower than in the F2 horizon. Upper palaeosol F3a attains 67.7 x 10-6 SI, the underlying P3b 53.2 x 10-6 SI. The lowermost palaeosol P4 shows again a somewhat higher MS of 58.3 x 10-6 SI.

4.4. Mollusc palaeontology

A total of 1705 terrestrial gastropod specimens classified to 13 species (Tab. 2 and Fig. 4) were obtained from 13 samples. Specimen richness, related to mollusc density within the samples varies upsection (Fig. 5). Hence, the mollusc densities are moderate in horizons L7 (85) and L6 (117), strongly decreased in L5 (5) and L4 (7), and strongly increased in L3 (90), L2 (213), and L1 (136).

Helicopsis striata, H. hungarica and Arianta arbustorum are present in all horizons (Fig. 5), except for L4 and L5. Columella columella is rare in all horizons showing peak abundance in L2 (3.29%, and 4.44%). The mean abundance of Pupilla muscorum and P. loessica is 11.34% with the highest abundance of 14.6% in L2. The latter horizon also produced the greatest number of individuals, representing all 13 species identified in the section.

5. DIscUssION

5.1. sedimentary facies and provenance of loess

The grain-size distribution is in good accordance with other loess localities in the Pannonian Basin (GALOVIC et al., 2009; BOKHORST et al., 2011). The studied locality shows the predominance of fine and medium silt in loess. Three lowermost loess units (L7, L6, L5) display an increase in the median grain-size, and coarse silt and sand exceed 50 %. The horizontally stratified fine-grained sand in the base of the L2 unit may be compared with aeolian sands with plane bed lamination (HUNTER, 1977) and are probably wind ripple deposits (HUNTER, 1977; CLEMMENSEN & ABRAHAMSEN, 1983).

THAMÓ-BOZSÓ & KOVÁCS (2007) published valuable data about the heavy mineral composition of flood plain sediments of the Palaeo-Danube and Tisza. The garnet and epidote mineral groups, amphiboles and chlorite are the most abundant heavy minerals in samples collected near the Danube River. The river Tisza and its tributaries drain abundant volcanic rocks, contrasting with those of the largely metamorphic and granitoid complex of the Danube. Comparing heavy mineral assemblage from Zmajevac LPS, with that data, it is obvious, that the main source area for loess in Baranja is from the Danube flood plain sediments. The main transport direction was from the North or North-West. Nevertheless, the higher concentration of amphiboles in the Zmajevac LPS (if compared with those from the Danube plain in Central Hungary), suggests an additional source area. The Western Carpathians with Neogene calc-alkaline volcanic rocks is the major source for amphiboles (THAMÓ-BOZSÓ & KOVÁCS, 2007). The percentage of amphiboles in the HMF of the Zmajevac LPS is fairly constant, averaging 28.8% in the three uppermost loess horizons and 23.25% in the lowermost four. Alternatively those minerals could also be denudation products from locally exposed volcanic and metamorphic rocks of the southward neighboring Slavonian Mts. (JAMICIC et al., 1987). Mt. Krndija and Mt. Papuk, which are the closest to Baranja of all the Slavonian Mts., consist of amphibolites. Furthermore, Pliocene sands from the northern slopes of Mt. Krndija and Mt. Papuk are of local origin and contain abundant amphiboles (JAMICIC et al., 1987).

5.2. Palaeoclimate, palaeohumidity and palaeovegetation

The identified mollusc species have been classified according to their palaeoenvironmental preferences following KROLOPP & SÜMEGI (1995) and SÜMEGI & KROLOPP (2002) (Tab. 2). The presence and abundance of each species in the sample is considered thereby as a function of available palaeoclimate and palaeohumidity conditions and/or type of palaeovegetation (Fig. 5).

Thermophilous and mesophilous species are dominant in terms of paleotemperature (74-55% abundance, Fig. 5), although none of the warm-climate representatives, typical for interglacial periods, such as Helicigona banatica, Mastus bielzi or Discus perspectivus have been observed. Nevertheless, compared to the Holocene, detected species compositions suggest unfavourable regional climate conditions during loess deposition in the late Pleistocene. Thereby arid climate species are common in all horizons with a mean abundance of 50.49%. Finally open habitat representatives likely dominate the section with mean abundance of 66.81% and peak in L5 loess horizon (71.4%).

The species distribution shows similarities with the Irig section on the southern slope of Fruska Gora in NW Serbia (MARKOVIC et al., 2007). The lowermost palaeosol of the Zmajevac section can be compared with palaeosol horizon V-S1 in Irig, which is barren of terrestrial molluscs representing the Last interglacial (MIS5) climate phase. In the Zmajevac section, corresponding loess horizon may be represented by fossil poor samples 3 and 4. Above this, in both the Irig and Zmajevac sections the dominant Helicopsis striata fauna reflects the Late Pleistocene Central European "warm" loess environmental conditions described by LOZEK (1964). In contrast, the lowermost samples in both sections bear common cooler climate representatives such as Clausilia dubia reflecting the Middle Pleistocene Penultimate Glacial (MIS6) conditions.

The Late Pleistocene assemblages from Petrovaradin show, in contrast, colder and more humid conditions than in either Irig or Zmajevac, probably due to the palaeogeographic position at the northern slope of Fruska Gora (MARKOVIC et al., 2005, 2008). In adition to the Irig locality, the Zmajevac section also represents the more sun-exposed southeastern side of Banovo Brdo.

5.3. Mollusc assemblage analysis

Species compositions of the samples reveal the presence of five glacial type assemblages defined in LOZEK (1964), including Chondrula tridens, Helicopsis striata, Pupilla loessica, Columella collumella and Arianta arbustorum. Therefore, of the six originally introduced assemblage types only those with Bradybaena fruticum marking the warmest glacial episodes (LOZEK, 1964) have not been found in the studied samples of the Zmajevac LPS (Fig. 6).

Horizon L7 displays a Columella columella Assemblage with depleted nominal species but with a high percentage of concurrent Orcula dolium (27.05%) and Clausilia dubia (12.94%). Horizon L6 displays a similar faunal composition (Fig. 5). Thus it is likely classified as Columella columella Assemblage.

Horizons L4 and L5 yielded few molluscs so a determination of mollusc assemblage type is highly speculative. The small number of shells obtained results from strong fragmentation of mollusc shells in that horizons. The L4 horizon is tentatively attributed to transitional Helicopsis striata-Arianta arbustorum Assemblage, whereas L5 is defined as a Helicopsis striata Assemblage.

The L3 horizon is marked by a transitional Helicopsis striata-Pupilla loessica Assemblage. The position of this horizon between two palaeosols of a pedo-complex indicates abrupt cooling during a period of warm and humid climate. In this sample a significant percentage of Cochlicopa lubrica was detected (12.2%), which is often a part of typical loess faunas.

Horizon L2 is characterized by the highest densities of mollusc remains in the section. In its lowermost part, a transitional Pupilla loessica-Helicopsis striata Assemblage with almost equal number of Pupilla sp. and Helicopsis sp. indi-viduals is present. A Helicopsis striata-Pupilla loessica Assemblage follows above. The middle part of the horizon produced a Helicopsis striata Assemblage with common Vallonia tenuilabris (6.59%). Finally the uppermost part bears a transitional Arianta arbustorum-Chondrula tridens Assemblage.

The L1 horizon is characterized by common representatives of a Helicopsis striata Assemblage commonly developed as transitional assemblage types. A sample from the base of L1 contains a typical Helicopsis striata Assemblage, followed by the transitional Helicopsis striata-Columella columella Assemblage. Above, a Helicopsis striata-Pupilla loessica Assemblage was detected. Along with Pupilla sp., Vallonia tenuilabris is also abundant (11%). Finally the uppermost part of horizon L1 belongs to the transitional Helicopsis striata-Chondrula tridens Assemblage.

A Helicopsis striata assemblage is dominant throughout the LPS Zmajevac, with transition to colder and warmer climate conditions, as recorded by the Pupilla loessica-Columella columella and Arianta arbustorum-Chondrula tridens Assemblages. The Helicopsis striata Assemblage indicates climate conditions preceding the last glacial/stadial maximum. Chondrula tridens and Arianta arbustorum are generally abundant species, but their abundance is never high enough for definition of a Chondrula tridens or Arianta arbustorum assemblage. Actually these were defined through abundance of typical concurrent species as defined by LOZEK (1964). Both assemblages mark slightly humid steppe conditions, whereas, the first one is bounded by phases of chernozem formation. The Pupilla loessica and Columella columella assemblages are typical loess faunas and exponents of glacial/stadial maximum. Arid and cold climate conditions are indicated by the Pupilla loessica assemblage, more humid conditions by the presence of the Columella columella assemblage. Pupilla muscorum, Helicopsis striata, Vallonia tenuilabris and Succinella oblonga are concurrent species in Pupilla loessica assemblages, while Orcula dolium, Clausilia dubia and Arianta arbustorum alpicola are additional species in a Columella columella assemblage.

Interestingly the best known NE Croatian terrestrial mollusc assemblage (POJE, 1984) from the well dated (WACHA & FRECHEN, 2011) section of Vukovar reflects a similar cold spell around the MIS 5 lower boundary. Yet, the recorded assemblages seem to reflect somewhat more arid microclimate conditions.

5.4. stratigraphic correlation and palaeoclimate history

Based on IRSL age estimate of 20.2±2.1 ka, GALOVIC et al. (2009) correlated the uppermost part of the loess-palaeosol succession at Zmajevac with the Upper Pleniglacial MIS 2. The base of the succession represented by a loess horizon yielded an IRSL age estimate of 217±22 ka and was tentatively correlated with the Middle Pleistocene MIS 8. However the lowermost 6 m of the Zmajevac succession are not exposed in the present outcrop (Fig. 6) that starts with the loess unit L7. The latter unit yielded the IRSL age estimate of 121±12 ka and was tentatively correlated with the Penultimate Glacial MIS 6 by GALOVIC et al. (2009).

The loess horizon L7 is overlain by palaeosol P4 and followed immediately by the next loess unit (L6). Those two lowermost loess horizons (L6 and L7) are characterized by the presence of a Columella columella assemblage (Fig. 6). Its equivalent is a Pupilla loessica Assemblage indicated in horizons L3 and L2 and likely a typical loess fauna representing however more arid climate conditions. In contrast, the Columella columella Assemblage is an indicator of the glacial maximum (LOZEK, 1964), with dominantly humid climate conditions. Compared with malacofaunal results from Irig and Ruma LPS in Vojvodina (MARKOVIC et al., 2006, 2007), that fauna represents palaeoenvironment of open grassland, and a cold, wet climate. Hence the presence of a Columella columella Assemblage in the basal part of the section strongly supports its correlation with the MIS 6 stage proposed by GALOVIC et al. (2009).

Sedimentological and magnetic susceptibility (MS) data obtained from Zmajevac LPS show similarities with other loess-palaeosol sequences in the Pannonian Basin investigated recently (MARKOVIC et al., 2006, 2007; GALOVIC et al., 2009; MOLNÁR et al., 2010; WACHA & FRECHEN, 2011). MS values are in the expected range, especially in loess horizons. MS values from four palaeosols are comparable with those from Irig in neighboring Vojvodina (MARKOVIC et al., 2007). The lowermost P4 palaeosol, displays significantly weaker signals, than the P2 palaeosol, but stronger than the overlying P3b palaeosol horizon (Fig. 6).

The MS value of 58.3 x 10-6 SI in the P4 palaeosol is lower than expected for an interglacial period and a fully developed soil. Even though it is the oldest palaeosol, a weaker signal than in the youngest F2 may indicate that iron is present in oxidized form in limonitic minerals. It is most likely that the low MS values result from mineral leaching. Such a decrease in the MS signal in clayey horizons was also detected in Germany by ROUSSEAU et al. (2002) and in Hungary by NOVOTHNY et al. (2009). It is very likely that similar processes affected the P4 palaeosol horizon in Zmajevac. In agreement with data by GALOVIC et al. (2009) the P4 horizon is correlated with the MIS 5e interglacial period.

The pedo-complex forming palaeosol horizons P3a and P3b is similar to a pedo-complex from the Vojvodina (MARKOVIC et al., 2007) and Hungary. Reminiscent of synchronous horizon of Hungarian Sütto LPS (NOVOTHNY et al., 2011), the signal from the P3 pedo-comlex is higher than the one measured in Vojvodina. Finally, the strongly increased MS value of 82.5 x 10-6 SI suggests that the uppermost palaeosol P2 represents an interglacial, rather than an interstadial phase. The chronostratigraphic data obtained by GALOVIC et al. (2009) imply the straight forward correlation of this horizon with the MIS 3 stage (Fig. 6).

The integration of palaeontological results with IRSL age estimates fixed the correlation of loess L7 with MIS 6. Through correlation of loess L2 with the Lower Pleniglacial MIS 4, (based likely on integrated palaeontological and chronostratigraphical data (GALOVIC et al., 2009)), the interchanging facies character between units L7 and L2 sug gests climate forcing reflecting the five-folded pattern of MIS 5 (Fig. 6). The loess-like horizon L5 could thereby correlate with the warmer MIS 5c, while L6 and L4 loess units would correspond to MIS 5d and MIS 5b cooling events, respectively. If this correlation proves true the whole studied succession, except for the lowermost loess L7 would have been deposited during the Late Pleistocene (Fig. 6).

6. cONcLUsION

The present investigation of the Zmajevac loess-palaeosol sequence (LPS) established valuable palaeontological data for palaeoclimate and palaeoenvironmental reconstruction of Baranja during the latest Middle and Late Pleistocene. The recorded mollusc compositions display small but important differences to other Pannonian basin LPSs, especially for horizons L7, L6 and L3. Hence the later assemblages indicate still cooler and more humid climate conditions than previously known (POJE, 1984). This difference is probably a consequence of local geomorphology and microclimate conditions. Indeed the mollusc fauna from the rest of the samples is in very good agreement with the previous Pannonian basin record. Additional data obtained from sedimentological and magnetic susceptibility analyses, seem to show fairly good congruence with results from other LPS in the Pannonian basin (MARKOVIC et al., 2006, 2007; NOVO THNY et al., 2011; WACHA & FRECHEN, 2011), although low sampling frequency prevent their use for more comprehensive correlation.

AcKNOWLEDGEMENt

Thanks go to Mathias HARZHAUSER (NHM Vienna) for discussion on gastropod taxonomy, Anita ESCHNER (NHM Vienna) for providing access to malacologicial collection of the Zoological department and Alice SCHUMACHER (NHM Wien) for providing gastropod photographic documentation. We are also grateful to reviewers Manfred FRECHEN and Slobodan B. MARKOVIC for their constructive criticism and valuable improvements of the manuscript. This paper was supported by Projects no. 181-181-1096-1093, 195-1951293-2703 and 195-1951293-0237 of the Ministry of Science, Education and Sports of the Republic of Croatia.

References

rEFErENcEs

AITKEN, M. J. & S. STOKES (1997): Climatostratigraphy.-In: TAYLOR, R.E. & AITKEN, M.J. (eds.): Chronometric Dating in Archaeology. Advances in Archaeological and Museum Science Series, Vol. 2, Plenum, New York.

BACANI, A., SPARICA, M. & VELIC, J. (1999): Quaternary deposits as the hydrogeological system of Eastern Slavonia.-Geol. Croat., 52, 141-152.

BOGNAR, A. (1976): Les i lesu slicni sedimenti i njihovo geografsko znacenje [Loess and loess like deposits and their geographical meaning-in Croatian].-Geogr. horizont, 1-2, 15-31.

BOKHORST, M.P., VANDENBERGHE, J., SÜMEGI, P., LANCZONT, M., GERASIMENKO, N.P., MATVIISHINA, Z.N., MARKOVIC, S.B., FRECHEN, M. (2011): Atmospheric circulation patterns in central and eastern Europe during the Weichselian Pleniglacial inferred from loess grain-size records.-Quat. Int., 234, 62-74. doi: 10.1016/ j.quaint.2010.07.018

BRONGER, A. (1976): Zur quartären Klima und Landschaftsentwicklung des Karpatenbeckens auf (paläo-) pedologischer und bodengeographischer Grundlage.-Kieler Geographische Schriften, 45, 1-268, Kiel.

BRONGER, A. (2003): Correlation of loess-palaeosol sequences in East and Central Asia with SE Central Europe: towards a continental Quaternary pedostratigraphy and paleoclimatic history.-Quat. Int., 106/107, 11-31.

CLEMMENSEN, L.B. & ABRAHAMSEN, K. (1983): Aeolian stratification and facies association in desert sediments, Arran basin (Permian), Scotland.-Sedimentology, 30, 311-339.

CROATIAN GEOLOGICAL SURVEY (2009): Geoloska karta Republike Hrvatske 1:300.000 [Geological Map of the Republic of Croatia 1:300 000-in Croatian].-Zagreb.

EVANS, M.E. & HELLER, F. (2001): Magnetism of loess/palaeosol sequences: Recent developments.-Earth Sci. Rev., 54, 129-144.

FRANK, C. (2006): Plio-pleistozäne und holozäne Molluscen Österreichs.-Verlag der Österreichischen Akademie der Wissenschaften, Wien, 1-797.

GALOVIC, L., FRECHEN, M., HALAMIC, J., DURN, G & ROMIC, M. (2009): Loess chronostratigraphy in Eastern Croatia-A luminescence approach.-Quat. Int., 198, 85-97. doi: 10.1016/j.quaint. 2008.02.004

GALOVIC, L., FRECHEN, M., PEH, Z., DURN, G. & HALAMIC, J. (2011): Loess/palaeosol section in Sarengrad, Croatia-A qualitative discussion on the correlation of the geochemical and magnetic susceptibility data.-Quat. Int., 240, 22-34. doi: 10.1016/j.quaint. 2011.02.003

GEISS, Ch. E. & ZANNER, C.W. (2006): How abundant is pedogenic magnetite? Abundance and grain size estimates for loessic soils based on rock magnetic analyses.-J. Geoph. Res., 111, B12S21. doi: 10.1029/2006JB004564

GUO, Z.T., RUDIMAN, W.F., HAO, Q.Z., WU, H.B., QIAO, Y.S., ZHU, R.X., PENG, S.Z., WEI, J.J., YUAN, B.Y. & LIU, T.S. (2002): Onset of Asian desertification by 22 Myr ago inferred from loess deposits in China.-Nature, 416, 159-163.

HARZHAUSER, M. & MANDIC, O. (2008): Neogene lake systems of Central and South-Eastern Europe: Faunal diversity, gradients and interrelations.-Palaeogeogr. Palaeocl. Palaeoec., 260, 417-434. doi: 10.1016/j.palaeo.2007.12.013

HORVÁTH, F. (1995): Phases of compression during the evolution of the Pannonian Basin and its bearing on hydrocarbon exploration.-Mar. Petrol. Geol., 12, 837-844.

HUNTER, R.E. (1977): Basic types of stratification in small eolian dunes.-Sedimentology, 24, 361-387.

JAMICIC, D., BRKIC, M., CRNKO, J. & VRAGOVIC, M. (1987): Osnovna geoloska karta SFRJ 1:100000. Tumac za list Orahovica L33-96 [Basic Geological Map of SFRY 1:100000, Geology of the Orahovica sheet-in Croatian].-Inst. za geol. istraz., Zagreb, Sav. geol. zavod, Beograd, 1-72.

KOVÁCS, J. (2008): Grain-size analysis of the neogene red clay formation in the Pannonian Basin.-Int. J. Earth. Sci., 97, 171-178.

KROLOPP, E. & SÜMEGI, P. (1995): Palaeoecological reconstruction of the Late Pleistocene, based on loess malacofauna in Hungary.-GeoJournal, 36, 213-222.

KUKLA, G.J. (1987): Loess stratigraphy in central China.-Quat. Sci. Rev., 6, 191-219.

LOZEK, V. (1964): Quartärmolluscen der Tschechoslowakei.-Rozpr. Ust. Geol., 31, 1-374.

LUGOVIC, B., MAJER, V. & STRUMPFL, W.E. (1990): Geochemical Characteristics of Basaltic Andesites from Baranja (Croatia, Yugoslavia).-Geol. vjesnik, 43, 135-142.

MALEZ, M. (1965): O nekim periglacijalnim pojavama u pleistocenu Slavonije [On some periglacial occurrances in the Pleistocene of Slavonia-in Croatian].-Geol. vjesnik, 18, 159-165.

MANDIC, O., DE LEEUW, A., BULIC, J., KUIPER, K., KRIJGSMAN, W., JURISIC-POLSAK, Z. (2011): Paleogegraphic evolution of the Southern Pannonian Basin: 40Ar/39Ar age constrains on the Miocene continental series of northern Croatia.-Int. J. Earth. Sci., 101, 1033-1046. doi: 10, 1007/s00531-011-0695-6.

MARKOVIC, S.B., McCOY, W.D., OCHES, E.A., SAVIC, S., GAUDENYI, T., JOVANOVIC, M., STEVENS, T., WALTHER, R., IVANISEVIC, P. & GALIC, Z. (2005): Palaeoclimate record in the Upper Pleistocene loess-palaeosol sequence at Petrovaradin brickyard (Vojvodina, Serbia).-Geol. Carpath. 56, 545-552.

MARKOVIC, S.B., OCHES, E., SÜMEGI, P., JOVANOVIC, M. & GAUDENYI, T. (2006): An introduction to the Middle and Upper Pleistocene loess-palosol sequence at Ruma brickyard, Vojvodina, Serbia.-Quat. Int., 149, 80-86. doi: 10.1016/j.quaint.2005.11.020

MARKOVIC, S.B., OCHES, E.A., McCOY, W.D., GAUDENYI, T., FRECHEN, M., JOVANOVIC, M. (2007): Malacological and sedimentological evidence for 'warm' climate from the Irig loess sequence (Vojvodina,Serbia).-Geophysics, Geochemistry and Geosystems 8, Q 09008. doi: 10.1029/2006GC001565

MARKOVIC, S.B., BOKHORST, M.P., VANDENBERGHE, J., Mc-COY, W.D., OCHES, E.A., HAMBACH, U., GAUDENYI, T., JOVANOVIC, M., ZÖLLER, L., STEVENS, T. & MACHALETT, B. (2008): Late Pleistocene loess-palaeosol sequences in the Vojvodina region, north Serbia.-J. Quat. Sci., 23, 73-84. doi: 10.1002/ jqs.1124

MARKOVIC, S.B., HAMBACH, U., CATTO, N., JOVANOVIC, M., BUGGLE, B., MACHALETT, B., ZÖLLER, L.,GLASER, B. & FRECHEN, M. (2009): Middle and Late Pleistocene loess sequences at Batajnica, Vojvodina, Serbia.-Quat. Int., 198, 255-266. doi: 10.1016/j.quaint.2008.09.003

MARKOVIC, S.B., HAMBACH, U., STEVENS, T., KUKLA, G.J., HELLER, F., MCCOY, W.D., OCHES, E.A., BUGGLE, B., ZÖLLER, L. (2011): The last million years recorded at the Stari Slankamen (Northern Serbia) loess-palaeosol sequence: revised chronostratigraphy and long-term environmental trends.-Quat. Sci. Rev., 30, 1142-1154.

MARTINSON, D., PISIAS, M.G., HAYS, J.D., IMBRIE, J., MOORE, T.C. & SHACKLETON, M.J. (1987): Age dating and the orbital theory of ice ages: development of a high-resolution 0 to 300 000-year chronostratigraphy.-Quat. Res., 27, 1-30.

MENGE, M.A. & MAURER, H.F.W. (1992): Heavy minerals in Colour.-Chapman and Hall, London, 151 p.

MOLNÁR, D., HUPUCZI, J., GALOVIC, L. & SÜMEGI, P. (2010): Preliminary malacological investigation on the loess profile at Zmajevac, Croatia.-Cent. Eur. J. Geosci., 2, 52-56. doi: 10.2478/v10085-009-0045-0

NOVOTHNY, Á., FRECHEN, M., HORVÁTH, E., BRADAK, B., OCHES, E.A., McCOY, W.D. & STEVENS, T. (2009): The loess profile at Sütto, Hungary.-Quat. Int., 198, 62-76. doi: 10.1016/j. quaint.2008.01.009

NOVOTHNY, Á., FRECHEN, M., HORVÁTH, E., WACHA, L. & ROLF, C. (2011): Investigating the penultimate and last glacial cycles of the Sütto loess section (Hungary) using luminescence dating, high-resolution grain size, and magnetic susceptibility data.-Quat. Int., 234, 75-85. doi: 10.1016/j.quaint.2010.08.002

PAMIC, J. (1997): Vulkanske stijene Savsko-dravskog meðurjecja i Baranje (Hrvatska) [Volcanic rocks between Sava and Drava Rivers-in Croatian].-Nafta Monograph., Zagreb, 192 p.

PAMIC. J. & PÉCSKAY, Z. (1996): Geological and K-Ar ages of Tertiary volcanic formations from the southern part of the Pannonian Basin in Croatia-based on surface and subsurface data.-Nafta, 6, 195-202.

PAMIC, J. & PIKIJA, M. (1987): Badenian andesites and pyroclastics from Baranja in northern Croatia (Yugoslavia).-Acta Geol. Hung., 30/1-2, 111-132.

PAVELIC, D. (2001): Tectonostratigraphic model for the North Croatian and North Bosnian sector of the Miocene Pannonian Basin System.-Basin Research, 13, 359-376.

PÉCSI, M. (1968): Loess.-In: FAIRBRIDGE, R.W. (ed.): The Encyclopedia of Geomorphology. Reinhold, New York, 674-678.

PILLER, W.E., HARZAHAUSER, M. & MANDIC, O. (2007): Miocene Central Paratethys stratigraphy-current status and future directions.-Stratigraphy, 4, 151-168.

POJE, M. (1982): Malakofauna prapornih naslaga istocnog dijela Slavonije (Malacofauna from loess deposits from Eastern Croatia-in Croatian).-Unpublished MSc Thesis, University of Zagreb, Zagreb, 63 p.

POJE, M. (1985): Praporne naslage "vukovarskog profila" i njihova stratigrafska pripadnost [Loess deposits from ''the Vukovar profile'' and their stratigraphic position-in Croatian].-Geol. vjesnik, 38, 45-66.

POJE, M. (1986): Ekoloske promjene na vukovarskom prapornom ravnjaku proteklih cca 500.000 godina [Ecological changes on the Vukovar loess plateau in last 500,000 years-in Croatian].-Geol. vjesnik, 39, 19-42.

PYE, K. (1995): The nature, origin and accumulation of loess.-Quat. Sci. Rev., 14, 653-667.

ROUSSEAU, D.D., ANTOINE, P., HATTE, C., LANG, A., ZÖLLER, L., FONTUGNE, M., OTHMAN, D.B., LUCK, J.M., MOINE, O., LABONNE, M., BENTALEB, I., JOLLY, D. (2002): Abrupt millennial climatic changes from Nussloch (Germany) Upper Weichselian eolian records during Last Glaciation.-Quat. Sci. Rev., 21, 1577-1582.

ROYDEN, L.H. (1988): Late Cenozoic Tectonics of the Pannonian Basin System.-In ROYDEN, L.H. & HORVATH, F. (eds.): The Pannonian Basin. A study in Basin Evolution.-Am. Assoc. Petrol. Geol. Mem., 45, 27-48.

RUKAVINA, D. (1983): O stratigrafiji gornjeg pleistocena s osvrtom na topla razdoblja i njihov odraz u naslagama na podrucju Jugoslavije [On stratigraphy of the Late Pleistocene with review of warm periods and their reflection in the area of Yugoslavia-in Croatian].-Rad JAZU, 19, 199-221.

SMALLEY I.J, MARKOVIC, S.B & SVIRCEV, Z. (2011): Loess is (almost totally formed by) the accumulation of dust.-Quat. Int., 240, 4-11. doi: 10.1016/j.quaint.2010.07.011

SÜMEGI, P. & KROLOPP, E. (2002): Quatermalacological analysis for modeling of the Upper Weichselian palaeoenvironmental changes in the Carpathian Basin.-Quat. Int., 91, 53-63.

THAMÓ-BOZSÓ, E. & KOVÁCS, L. (2007): Evolution of Quaternary modern fluvial network in the Mid-Hungarian plain, indicated by heavy mineral distributions and statistical analysis of heavy mineral data.-Dev. Sedimen., 58, 491-514.

WACHA, L. & FRECHEN, M. (2011): The chronology of the "Gor janovic loess section" in Vukovar, Croatia.-Quat. Int., 240, 87-99. doi: 10.1016/j.quaint.2011.04.010

WENTWORTH, C.K. (1922): A scale of grade and class terms for clastic sediments.-The J. Geol., 30, 377-392.

Manuscript received February 03, 2012

Revised manuscript accepted July 20, 2012

Available online October 30, 2012