Why were the Neolithic landscapes of Bela krajina and Ljubljana Marshes regions of Slovenia so dissimilar |

This paper aims to address the question of the diversity of the environment of the first farming communities in the region of Bela krajina and the Ljubljana Marshes area of Slovenia (Fig. 1). Differences in the composition of vegetation detected in the pollen record will be analysed in order to estimate whether they were a consequence of specific natural characteristics of the regions studied, dissimilar land-use in the past, or the size of the study sites.


Introduction
This paper aims to address the question of the diversity of the environment of the first farming communities in the region of Bela krajina and the Ljubljana Marshes area of Slovenia (Fig. 1).Differences in the composition of vegetation detected in the pollen record will be analysed in order to estimate whether they were a consequence of specific natural characteristics of the regions studied, dissimilar land-use in the past, or the size of the study sites.
In recent decades an extensive pollen analysis of sedimentary cores and samples collected during archaeological excavations was carried out on Ljubljana Marshes, so the general development of vegetation in the area is very well known (e.g.Culiberg 1991;πercelj 1996 andreferences cited there, Gardner 1999a;1999b) and the results presented in this study (a pollen analysis of 'Na mahu 1' core) accord with previous research.The question of past hydrological conditions in the area was also addressed by several researchers using sedimentological, geomorphological, archaeological and multidisciplinary palaeoecological data (e.g.Melik 1946;πercelj 1966;πifrer 1984;Budja 1995;Velu∏≠ek 2005;Gaspari and Eri≠ 2006;Verbi≠ 2006;Mleku∫ et al. 2006;An-dri≠ et al. in prep., and many others).This led to various interpretations of the complex hydrological conditions in the basin (most complex in the Neolithic; many researchers would probably agree that in the early Holocene the area was covered by a freshwater lake, whereas later it became a floodplain and in some parts a peat bog).In contrast to Ljubljana ABSTRACT -This paper compares the development of Holocene vegetation in Bela krajina and Ljubljana Marshes (Ljubljansko barje) regions of Slovenia.The results of pollen analysis suggest that in Bela krajina the human impact on the environment (forest clearance and burning) was very intensive throughout the Holocene and led to changes in forest composition, increased biodiversity, and the formation of a mosaic landscape.In the Ljubljana Marshes, forest burning and clearance seem less intensive, although changes in forest composition and 'anthropogenic indicator' pollen types were detected.These differences between study regions are presumably a consequence of various climates, hydrology, bedrock and land-use in the past.

KEY WORDSpalynology; Neolithic archaeology; Bela krajina; Ljubljana Marshes
Marshes palynological, research in Bela krajina was less extensive, and both palynologically investigated sites in the area (Mlaka and Griblje, Andri≠ 2001;An-dri≠ in press) are presented in this paper.To date, no studies of past hydrology have been carried out in the area.Therefore, since hydrological conditions on Ljubljana Marshes were very complex, whereas (presumably much simpler) hydrological conditions in Bela krajina have not been investigated at all, more (detailed) research is needed in both regions in the future.Palynological research was much more intensive, and for that reason this paper will focus on only one aspect of both Neolithic landscapes: the composition of vegetation as revealed by palynological studies.
The main reasons for the variety of flora in Slovenia include its geographical position, diverse climate, relief and bedrock (Wraber 1969;Kladnik 1996;Ogrin 1996;Perko 1998).The distinctive vegetation of Slovenia's phyto-geographic regions became apparent after c. 8800 calBP, whereas the onset of an intensive, largescale forest clearance, burning and the formation of the present-day landscape is dated between late prehistory and the medieval period (c.3000 calBP -1000 calBP).Palynological research also suggests that the human impact on the vegetation was important and contributed to increasing biodiversity (An-dri≠ and Willis 2003).Neolithic/Eneolithic farming communities (c.7000-5000 calBP) lived in an environment where differences between regions had already become apparent, but large-scale forest clearances had not yet occurred.In such environments it is very difficult to distinguish the natural from the anthropogenic causes of environmental change.Therefore, in order to better understand the human impact on the vegetation, the vegetation history in the vicinity of archaeological sites was studied in detail.Sedimentary cores for pollen analysis (Fig. 1) were collected at Mlaka (Bela krajina) and 'Na mahu 1' (Ljubljana Marshes) coring locations in order to estimate the similarity/dissimilarity of vegetation development during the Holocene in both study regions.In addition to this, the vegetation history at Griblje was investigated and compared with Mlaka to assess also the intra-regional variability of Bela krajina vegetation.

The Bela krajina study area
Bela krajina (Fig. 1) is located in south-eastern Slovenia, between high Dinaric Plateaus in the west and the Kolpa River and Pannonian Plain in the east.More than half of its territory lies below 400 m a.s.l., on predominantly Mesozoic limestone and dolomite bedrock (Gams 1961;1984;Buser 1984).The climate of Bela krajina is moderate continental-sub-Pannonian, with a sub-Mediterranean precipitation regime (primary precipitation is highest in autumn), and hot, dry summers.The annual amount of precipitation is between 1200 and 1300 mm in western parts.The average temperatures of the warmest month are between 15 °C and 20 °C, and of the coldest month, between -3 °C and 0 °C (Bernot 1984;Ogrin 1996;Plut 1985).
The composition of Bela krajina forests varies according to altitude, land-use and soil type.Predominantly beech forests are characteristic of higher alti- tudes, whereas patchy oak-hornbeam forests grow in the lowlands (Miklav∫i≠ 1965;Wraber 1956;Ma-rin≠ek and ∞arni 2002;∞arni et al. 2003).Meadows, fields, pastures and human-managed birch (Betula pendula) forests ('steljniki'), which were originally used as spring/summer grazing areas, spread into the lowlands due to intensive human impact in the last centuries.

The Ljubljana Marshes study area
Ljubljana Marshes lies in central Slovenia (Fig. 1) at about 289 m a.s.l., on predominantly carbonate bedrock, with Triassic and Jurassic limestones and dolomites in southern and western parts of the basin, whereas Palaeozoic sandstones, conglomerates, shales and limestones prevail in the north and east (Mencej 1989).The bottom of the basin is covered by thick layers of Pleistocene and Holocene alluvial and lacustrine sediments (Mencej 1989;Grim∏i≠ar and Ocepek 1967;Tancik 1965;πercelj 1965;1967).The climate on Ljubljana Marshes is temperate-continental, with a sub-continental precipitation regime (precipitation highest in the summer) and annual precipitation between 1000 and 1300 mm.The average temperatures of the coldest month are between -3 °C and 0 °C, and in the warmest month the average is between 15 °C and 20 °C (Ogrin 1996).
Ljubljana Marsh is currently covered by meadows, fields and patchy woodlands of pine (Pinus), birch (Betula), alder (Alnus) and oak (Quercus robur).Only very small peat-bogs with ombrotrophic species have remained (Martin≠i≠ 1987).Predominantly beech (Fagus) forests grow on the hills surrounding the area (∞arni et al. 2003).In the 18 th century it was much wetter than today, and mostly covered by peatbogs, smaller 'puddles' and springs (Melik 1927) and drier heather lands, with spruce (Picea), pine, alder and birch trees.However, in the second half of the 18 th century, the first drainage works in the area started, and by the end of 19 th century almost all the peat had been burnt or cut in order to obtain dry land needed for agriculture (Melik 1927).

Methodology
The pollen record of three study sites, Mlaka and Griblje (G3) in Bela krajina, and 'Na mahu 1' in Ljubljana Marshes (Fig. 1) is presented and compared in this study.The detailed methodology and results for individual study sites have been published (for Mlaka and Griblje see Andri≠ in press) or are currently in preparation ('Na mahu 1 ', Andri≠ et al. in prep.), therefore only selected data will be presented in this paper.
All sedimentary sequences were collected by Livingstone piston corer, and standard laboratory procedures (Bennett and Willis 2002) were used for pollen analysis.The age was determined by AMS radiocarbon dating of organic carbon extracted from the sediment.The radiocarbon dates at Mlaka and Griblje were calibrated by the BCal program (hosted by the Department of Probability and Statistics at the University of Sheffield, Buck et al. on-line), which incorporates the IntCal 04 calibration dataset (Reimer et al. 2004), and these results were used for age-depth modelling in PSIMPOLL (general linear line-fitting by singular value decomposition for Griblje, and a combination of general linear line-fitting by singular value decomposition and linear interpolation for Mlaka).The conventional radiocarbon dates used for age-depth modelling are marked on each diagram.The age-depth modelling for 'Na mahu 1' core was a linear interpolation between the median values of 14 C dates (the lower two dates were excluded from the age-depth modelling due to an error, presumably caused by the reservoir effect).The percentage pollen diagrams of selected taxa were plotted by PSIMPOLL 3.00, 4.25 and PSCOMB 3.01, C programs (Bennett 1998;Bennett on-line).They were divided into zones using binary splitting by sum of squares, and the number of significant zones was determined by the broken-stick model (Bennett 1996;1998).Dots on the pollen diagram indicate values lower than 0.5 %.Palynological richness (rarefraction analysis) was also calculated by PSIM-POLL.Microscopic charcoal (in two size classes: <40 µm and >40 µm) was counted with the pollen ('Na mahu 1' and Griblje cores) and, in addition to this, the concentration of microscopic charcoal was determined according to Clark's (1982) point count method at both Mlaka and Griblje.

Results
The results are presented on percentage pollen diagrams of selected taxa (Fig. 2) and compared in Table 1.Landscape openness, the microcharcoal record and palynological richness of all three study sites are also compared (Fig. 3).More detailed results are being published in separate publications (Andri≠ in press, Andri≠ et al. in prep.).The mismatch of Mlaka and Griblje pollen diagrams (compare Fagus curves) for levels older than c. 6000 calBP are most probably a consequence of problematic radiocarbon dating at Griblje, where levels between 50 and 61 cm seem to be up to c. 900 years 'too old' (Andri≠ in press).
Later (at c. 9000 -8750 calBP) the forest composition suddenly changed, and thick, mostly beech (Fagus) forest started to grow in Bela krajina and Ljubljana Marshes (Fig. 2, Tab. 1).In Ljubljana Marshes this beech forest persisted until c. 6750 calBP, whereas in Bela krajina the development was much more dynamic.Slight fluctuations in beech pollen curves (some of them coincide with charcoal peaks and even with Cerealia-type pollen grain at Griblje) suggest that the forest in Bela krajina was less dense than in the Ljubljana Marshes region, and occasional smallscale landscape fires probably caused minor openings in the canopy.After c. 7800 calBP beech (Fagus) at the Mlaka site started to decline, and by c. 7300 calBP the percentage of tree pollen had significantly declined (with the exception of lime (Tilia), which increased), whereas herbs and monolete fern spores (Filicales) increased, suggesting the opening of the landscape (Fig. 2a).A similar change in vegetation was also detected at Griblje (Fig. 2b), where beech forest was replaced by a more open landscape, with higher percentages of pine (Pinus) and Trilete spores than at Mlaka.In contrast to Bela krajina, the hills surrounding Ljubljana Marshes basin remained very much forested with beech (Fagus), and after 7600 calBP also fir (Abies) (Fig. 2c).No such increase in fir (Abies) was detected in Bela krajina, although the forest started to regenerate at the beginning of the 7 th millennium calBP.
After c. 7000 calBP, hazel (Corylus) and oak (Quercus) at Mlaka, increased and between c. 6700-6100 calBP hornbeam (Carpinus betulus) woodland was growing around the coring location.The Griblje forest also regenerated, and beech (Fagus), alder (Alnus) and hazel (Corylus) started to increase again, but in contrast to Mlaka, there was no hornbeam (Carpinus b.) phase.The human impact on the environment around both study sites was significant and can be associated with the Neolithic/Eneolithic sites at Pusti gradac and Griblje.Small-scale forest burning (increased micro-charcoal concentration), agriculture and grazing (anthropogenic indicator taxa: Cerealiatype, Centaurea, Plantago l., Chenopodiaceae, Artemisia, Compositae lig.and Compositae tub.) was most intensive at c. 6100 calBP (Fig. 2a, 2b).What happened at the same time on Ljubljana Marshes?A major change in the composition of vegetation occurred between 6750 and 5600 calBP (roughly at the same time as Carpinus b. phase at Mlaka), when beech (Fagus) and fir (Abies) declined, whereas oak (Quercus), alder (Alnus) and hazel (Corylus) increased (Fig. 2c), but an increase in hornbeam (Carpinus betulus, best seen at Podpe∏ko jezero, Gardner 1999b) was less significant than at Mlaka.An increased percentage of herb pollen after 6600 calBP and anthropogenic/grazing indicators (Plantago l.) suggest that the landscape became slightly more open, which can be associated with human activity at the Resnikov prekop settlement, located c. 1.5 km south of the coring location.

Middle Holocene
In the second half of the Holocene, shade-tolerant tree taxa started to increase in both study regions.In Bela krajina, beech (Fagus) forest spread again after 5700 calBP, but forest composition changed at c. 4800 calBP, when beech was replaced by fir (Abies) (Fig. 2a).Similarly, an increase in beech on Ljubljana Marshes is dated after c. 6000 calBP, but this was very soon replaced by fir, which remained the most important tree taxon until c. 4500 calBP, when spruce (Picea) and alder (Alnus) increased (Fig. 2c).The de-

Discussion
The human impact on the environment After comparing the pollen record from both study regions, it appears that the intensity of human impact on the environment in Bela krajina was very considerable.Constant small-scale forest burning and cutting associated with the appearance of anthropogenic indicator taxa and increased biodiversity (palynological richness, Fig. 3) was detected from at least c. 6000 calBP, which led to changes in forest composition, increasing differences between the Mlaka and Griblje study sites and the formation of mosaic landscape (Andri≠ in press).
The human impact on the environment on Ljubljana Marshes seems much less intensive; here, thick beech and beech-fir forest persisted until c. 6750 calBP.At c. 6600 calBP a minor forest clearance (Fig. 2c, 3) and anthropogenic indicator taxa were detected, but the landscape remained very much forested (as already demonstrated by previous research, Gardner 1999a; 1999b) and the development of vegetation much less dynamic than in Bela krajina.Does this mean that in Bela krajina, Neolithic/Eneolithic settlements were more numerous and farming activities more intensive (or earlier) than on Ljubljana Marshes?Not necessarily.The differences between Bela krajina and Ljubljana Marshes could be a consequence of the natural features of the study sites (Fig. 1).Both sites in Bela krajina, Mlaka and Griblje, are small basins with diameters of c. 30 m and without inflowing or outflowing streams.In such small basins, which receive pollen deriving mainly from local vegetation, small-scale local changes in vegetation are very visible (Jacobson and Bradshaw 1981).In bigger basins with complex hydrology, such as Ljubljana Marshes, regional pollen prevails (Jacobson and Bradshaw 1981), so local, small-scale forest clearance is less visible in the pollen record.Considering palynological theory, therefore, these results were expected.
Despite all these differences between Bela krajina and Ljubljana Marshes, there is one similarity: an increase

Fig. 3. Landscape openness, microcharcoal record and palynological richness at Mlaka, Griblje (G3) and 'Na mahu 1' study sites. * hiatus due to peat cutting/burning on Ljubljana Marshes.
in biodiversity (palynological richness, Fig. 3) at the beginning of the 7 th millennium calBP (c.7000 calBP at Mlaka and 6700 calBP at 'Na mahu 1' core), although no major landscape forest clearance has been detected on pollen diagrams.This could probably be associated with small-scale forest clearance and the activities of early farming communities.

Human impact versus climate
Changes in forest composition and differences between the study regions can be a consequence of climatic fluctuations and/or human impact.Today, summers in Bela krajina are warmer and drier than on Ljubljana Marshes, and it is possible that a similar contrast existed in the past.Therefore, less fir (Abies), which needs a humid climate (Ellenberg 1988) was growing in Bela krajina than on Ljubljana Marshes.However, fir is susceptible to fire (Tinner et al. 1999), and it is possible that in Bela krajina it was suppressed by anthropogenic burning of the landscape.The charcoal record suggests that the vegetation around Ljubljana Marshes was not burnt so often.Why notbecause the lake/marshy landscape of Ljubljana Marshes was more difficult to burn, and this landscape was used instead for hunting, fishing and gathering?Due to the complex hydrological conditions, more open land probably already existed (or was regularly formed) near the edge of the lake and/or along floodplains, and very extensive forest clearance for agriculture was not needed (Willis 1995)?
Palynological research also demonstrated that differences in the composition of vegetation occurred not only between the phytogeographic regions of Slovenia, but also within each individual region.This can best be demonstrated by the differences between Mlaka and Griblje, where, although the study sites are located only c. 10 km apart, the vegetation history was very dissimilar.While at Mlaka (located on predominantly limestone bedrock) human pressure on the environment was intensive throughout the Holocene, and the present-day open landscape had formed by the medieval period at c. 1000 calBP, the landscape at Griblje (on sand and clay) remained predominantly forested up to the present.This suggests that areas more suitable for agriculture were probably most intensively used.

Climate
To date no studies of local (regional) climate have been carried out, therefore changes in vegetation (e.g.forest composition) can be only compared with the global (Northern Hemisphere) climate.In both study regions beech (Fagus) establishment at the be-ginning of the 9 th millennium calBP was presumably associated with climatic change -an increase in precipitation.A similar increase in shade-tolerant tree taxa also occurred in other regions of Slovenia (An-dri≠ and Willis 2003) and neighbouring countries (Abies expansion in the lowlands of the southern Alps, Tinner et al. 1999;Gobet et al. 2000;Tinner and Lotter 2006;and Fagus increase in Dalmatia, Schmidt et al. 2000) and it seems that this palaeoenvironmental change was of regional extent.It was limited to the areas south of the Alps, where a continental climate regime had already been replaced by an Atlantic climate regime at about 9100 calBP, whereas north of the Alps, the Atlantic climate is associated with the 8.2 ky BP event (Tinner and Ammann 2001).
At the beginning of the 8 th millennium calBP, beech (Fagus) in Bela krajina started to decline, and by c. 7300 calBP, the beech forest had been replaced by a more open landscape.Fagus decline is limited only to Bela krajina (the Ljubljana Marshes region remained very much forested), so it seems unlikely that it would have been triggered by cold global climatic fluctuations, such as the 8.2 ky BP event (Alley et al. 1993;Meese et al. 1994;Stager and Mayewski 1997;Haas et al. 1998;Alley and Ágústsdóttir 2005).However, the impact of precipitation fluctuations might have been more important than the temperature, especially since the 8.2 ky event is assumed to have been dry in the lowlands (Haas et al. 1998), with lower lake levels north and south of 50°N and 43°N respectively (Magny and Begeot 2004).It is possible that in the 8 th millennium calBP climatic differences between Bela krajina and the Ljubljana Marshes were more pronounced than today, and forest composition in Bela krajina was affected by warm and dry summers, whereas at the same time no major change in forest composition took place on the Ljubljana Marshes.
Is it possible that Fagus decline was caused by the impact of hunter-gatherers and/or farmers on the landscape?Yes.Admittedly, no archaeological sites reliably dated before c. 7000 calBP have been discovered in Bela krajina, but this option cannot be completely ruled out, and further archaeological and multi-proxy palaeo-ecological research of the regional climate is needed.
After c. 6700 calBP, the percentage of beech (Fagus) and fir (Abies) was low in both study regions.This is most probably a consequence of significant human impact (forest clearance and burning), and coincides with a major climatic reversal after 8.2 ka calBP (Stager and Mayewski 1997;Alley and Ágústsdóttir 2005), when the oceanic early Holocene climate with enhanced westerly airflow was presumably replaced by a more meridional flow pattern, with anti-cyclonic summer conditions, and thus a dry climate and lower lake levels in the period between c. 6800-5700 calBP (Seppä and Birks 2001).
After c. 6000-5700 calBP, beech (Fagus), and later also fir (Abies), increase in both regions, with a slightly decreased human impact on the environment.This could have been associated with the cold and wet climate in the 6 th millennium calBP (Mayewski et al. 2004;Haas et al. 1998;Magny 2004;Magny and Haas 2004;Denton and Karlén 1973;Seppä and Birks 2001;O'Brien et al. 1995;Bond et al. 1997).

Conclusions
The above described differences between Bela krajina and the Ljubljana Marshes are presumably a consequence of different climates (wetter/colder in the Ljubljana Marshes region?), hydrology, topography and bedrock, as well as archaeological settlement patterns and land-use (more frequent landscape burning in Bela krajina) in the past.However, different natural characteristics and thus the pollen source area of individual study sites should not be ignored.
While at Bela krajina, where all the study sites are small marshes without inflowing/outflowing streams, the local Neolithic impact on the landscape is very visible, the Ljubljana Marshes study site is much bigger and has a more complex and changeable hydrology, which affected the pollen source area.Here, a weak local human impact on the vegetation is more difficult to detect.
How to proceed: more archaeological research is needed in both regions in order to better understand the economy of past societies and their impact on the vegetation.Especially in Bela krajina, where natural conditions are not very favourable for the preservation of archaeological sites or animal and plant remains, more information about archaeological settlement patterns and the economy during the transition from hunting-gathering to the first farming communities would be very valuable.More multi-proxy palaeo-ecological (palaeo-climatological and palaeohydrological) research is also needed, since the impact of climate fluctuations on the vegetation and hydrology was significant.
This research was funded by the Slovenian Research Agency, project no.Z6-4074-0618-03.I am very grateful to Mateja Belak, who prepared the figures.

Fig. 1 .
Fig. 1.Bela krajina and Ljubljana Marshes study regions and the position of palynological cores.