Neolithisation of Sava-Drava-Danube interfluve at the end of the 6600 – 6000 BC period of Rapid Climate Change > a new solution to an old problem

The Early Neolithic of Sava-Drava-Danube interfluve, situated on the southern edge of the Carpathian basin, was marked by Star≠evo culture. It first appeared in western Syrmia and eastern Slavonia, spreading through Baranja and western Slavonia and further west along the Drava River sometime around or slightly after 6000 BC. Over 100 Star≠evo culture sites are known today, but only a handful have been excavated, and even fewer properly published. Data from four sites are used in this paper: Sopot, Zadubravlje – Du∫ine, Slavonski Brod – Galovo and Virovitica – Brekinja (Fig. 1). The earliest radiocarbon dates can be attributed to features exhibiting the full so-called Neolithic package. The south-east to north-west direction of spread of Neolithisation can be observed through the available radiocarbon dates.


Introduction
The Early Neolithic of Sava-Drava-Danube interfluve, situated on the southern edge of the Carpathian basin, was marked by Star≠evo culture.It first appeared in western Syrmia and eastern Slavonia, spreading through Baranja and western Slavonia and further west along the Drava River sometime around or slightly after 6000 BC.Over 100 Star≠evo culture sites are known today, but only a handful have been excavated, and even fewer properly published.Data from four sites are used in this paper: Sopot, Zadubravlje -Du∫ine, Slavonski Brod -Galovo and Virovitica -Brekinja (Fig. 1).The earli-est radiocarbon dates can be attributed to features exhibiting the full so-called Neolithic package.The south-east to north-west direction of spread of Neolithisation can be observed through the available radiocarbon dates.
The first relative chronology of Star≠evo culture for the region in question was determined by Stojan Di-mitrijevi≤ (1969.40;1979.237-238).This chronology is still in use, and no attempts have been made to challenge it or to attribute absolute dating to its phases1 1 .
ABSTRACT -The idea of the Neolithisation of the Sava-Drava-Danube interfluve has undergone very little change since S. Dimitrijevi≤'s time.Despite their many shortcomings, new archaeological excavations and radiocarbon dates of Early Neolithic sites have provided us with new insight into the process of Neolihisation of this region.Using the recently published work by B. Weninger and L. Clare (Clare, Weninger 2010;Weninger et al. 2009;Weninger et al. 2014) as a starting point, the available radiocarbon and archaeological data are used to build up a time frame comparable to the wider region of Southeast Europe and climate conditions for specific period.The results fit the model of Neolithisation well (Weninger et al. 2014.9, Fig. 4), filling in the geographical gaps.

Relative chronology of Star≠evo culture in northern Croatia
The basic typological periodisation and relative chronology of Star≠evo culture in northern Croatia was determined by Dimitrijevi≤ (1969.40;1979.237-238)with very few later additions (Minichreiter 1992. 54-55;Markovi≤ 1994.62-63).Dimitrijevi≤ separated Star≠evo culture into seven phases: Monochrome, Linear A, Linear B, Girlandoid, Spiraloid A, Spiraloid B and the Final Ωdralovi phase (1969.40; 1979.237-238).Kornelija Minichreiter later added Linear C phase (Minichreiter 1992.54-55),which she considers parallel to Spiraloid A phase in eastern Slavonia and Baranja.Zorko Markovi≤'s proposed chronological system was never accepted (Markovi≤ 1994. 62-63) 2 2 .No recent attempts have been made to further study the problem of the relative chronology of Star≠evo culture, except Minichreiter's internal chronologies of the Zadubravlje -Du∫ine, Slavonski Brod -Galovo, and Pepelana sites (Minichrieter 1992;2007a;2007b;2007c), although the problem of chronological sequences of these phases and their apparent parallel appearance in different regions still waits to be addressed.

An old problem
The process of Neolithisation of the Sava-Drava-Danube interfluve has not been properly addressed since Dimitrijevi≤'s time.The development of early Neolithic Star≠evo culture was observed by Minichreiter mostly in the light of new excavations carried out at the Pepelana, Zadubravlje and Galovo sites in terms of internal chronology (Krajcar Bro-ni≤, Minichreiter 2007;2011;Minichreiter 2001;2007a;2007b;2007c;2010;Minichreiter, Boti≤ 2010;Minichreiter, Krajcar Broni≤ 2006) but the process itself was not the focus of Minichreiter's research.
The last campaign at the Sopot site also provided some fragments of white painted pottery, although some fragments were found earlier scattered over the site out of context 4 4 .How and where the so-called Neolithic package reached northern Croatia and how that affected the Mesolithic way of life remain unanswered question.
There is another problem concerning most of the new and some of the old archaeological excavations: most of the finds were never published.The same holds for the radiocarbon dates obtained from these excavations.They have not been published or are published without any actual context, or perhaps with just a short description of the feature they were collected from.Thus, the view of how and where the neolithisation started in this region is limited.Nevertheless, data available, although limited, can provide a basis for a new interpretation of the beginning of Neolithic in northern Croatia.

C sequences of four sites
From over hundred Star≠evo culture sites only six have been radiocarbon dated 5 5 and only four of these sites have usable data 6 6 .From Sopot near Vin-kovci (Fig. 1.1), Zadubravlje -Du∫ine (Fig. 1.2), Slavonski .and Virovitica -Brekinja sites (Fig. 14) 37 radiocarbon dates are available, most of them from Galovo (Tab. 1) 7 7 .Reconstructing a time frame from a limited number of radiocarbon dates per site, the obvious difference in quantity from site to site, the limited information about the context from which the samples were taken, etc., is methodologically questionable, but it is the only information available at the moment.
Sopot near Vinkovci (Fig. 1.1)The Sopot site was systematically excavated from 1996 until 2008 in 13 campaigns (Krznari≤ πkrivanko 2015.372)although there were several previous excavations during the 20 th century (Krzna-ri≤ πkrivanko 2015.371-372).The excavations focused on a late Neolithic Sopot culture tell settlement, but during the last campaign in 2008: "Dug into the pre-sterile layer (prehistoric humus), for the first time at Sopot, a structure was excavated which can be attributed to the Star≠evo culture.Pit dwelling SU 519 was dated to the period between 6060 and 5890 BC." (Krznari≤ πkrivanko 2015.378) (Fig. 2) 8 8 .The pit was found on the easternmost side of the excavated area, very close to the summit of the settlement plateau (Krznari≤ πkrivanko 2009.90).The pottery from this pit was coarse, but according to the excavator, fragments of white painted pottery were discovered before, scattered over the site out of context 9 9 .These finds have not been published yet.The area excavated was 375m 2 , several metres in depth (Krznari≤ πkrivanko 2011.212; 2015.375).
Three dates are available, from pit SU 519 (Beta 251910) and two pre-virgin soil layers SU 80 (Beta 251909) and 143 (Beta 251911) (Fig. 3; Tab. 1).All three dates group around 6000 BC or just slightly earlier.These are the oldest dates for Star≠evo culture in northern Croatia so far.The existence of two layers into which the Star≠evo feature was dug in may suggest one or more episodes of flooding sealing off the Star≠evo layer, rather than being layers 3 Pits 9, 205, and 207. 4 See note 9. 5 Only the published data were taken into consideration, with the exception of one unpublished date from Galovo provided by Dr Kornelija Minichreiter, for which we are grateful.6 Dates from Toma∏anci -Pala≠a site were published as a 5660-5300 BC range without details (Balen, Gerometta 2011.84, note 2) and for Beli∏≤e -Staro Valpovo site the date was published as 6400-5450 BC, a sample being taken from a male skeleton, but with no other information (πimi≤ 2007a.34, note 4; 2007b.10; 2012.27).Both sites were excluded from this paper.7 Sopot 3 dates; Zadubravlje 8 dates; Brekinja 2 dates; Galovo 24 dates.8 Krznari≤ πkrivanko (2011.219)mentions two layers (80 and 143) through which this structure was dug in.These two layers were detected over other areas at the site and contained sporadic Star≠evo finds (Krznari≤ πkrivanko 2009.90).9 M. Krznari≤ πkrivanko, personal communication. of prehistoric humus.The next occupation level at this site appears almost 1000 years later during the late Neolithic Sopot culture 1 10 0 .
Zadubravlje -Du∫ine site (Fig. 1.2) Rescue excavations on the site were carried out in 1989 and 1990.An area of 6200m 2 was excavated (Minichreiter 1992.31;1993.93;2001.203, Fig. 4).Several large and small pits were explored, comprising working and living areas, many post holes, several kilns and a well (Minichreiter 1992;1993a;2001).No white painted pottery was reported from this excavation, but dark painted linear motifs were found together with coarse pottery characteristic of Linear A phase (Minichreiter 1992.35, 103, Tab. 21;2007c.173).
Eight radiocarbon dates are available from this site (Fig. 4; Tab. 1).The oldest date (Z-2924) comes from a sample taken from the well.It was probably taken from a larger beam or a trunk and may be considered old wood; it was not taken into further consideration in this paper.Most of the dates group between 5900 and 5600 BC (Fig. 4; Tab. 1) with one exception (Z-2925).
Slavonski  The site is situated about 9 km west of the Zadubravlje -Du∫ine site.The excavations started in 1997 and are continuing.An area of around 3000m 2 yielded 11 large pits which are considered burial pits, pitdwellings and working pits, one above ground struc-ture (Minichreiter 2010;Minichreiter, Boti≤ 2010. 107), several large fences, a large number of post holes and smaller pits 1 11 1 .White painted pottery was found in pits 9, 205, and 207 (Fig. 5) although most of the painted pottery in all three pits exhibit dark motifs (Minichreiter 2007b).Here, as at Zadubravlje site, the coarse pottery corresponds to Dimitrije-vi≤'s definition of Linear A phase (Dimitrijevi≤ 1979. 243).
At the end of its life, the site was covered with an 80cm layer of sediment indicating no later occupation except a late Bronze Age necropolis (Minichreiter 2007a.33-34).
Twenty-four dates are available from this site, but some should be excluded.Some samples may have been contaminated by water from flooding episodes 1 12 2 , while others (Z-3586, Z-3584) may be consi-

Fig. 3. Sopot 14 C dates.
10 The oldest house SU 23 is dated to 6020±100 BP/5050-4780 BC (68.2%) (Z-3139) and 6010±100 BP/5040-4770 BC (66.35%) (Z-3140) (Krznari≤ πkrivanko 2011.211, Tab. 1).11 The last section excavated since 2010 was only published in short reports (Minichreiter 2011;2012;2013;2014;2015). 12 Most probably samples Z-5043, Z-5044, Z-4879 and Z-4880.The site was damaged by clay exctraction for a brick factory now closed, and there is a large hole to its north-east side.After heavy rains, water sometimes reaches the level of the site.Damage could have been done to unexcavated features with 80 cm of sediment already removed from their top.However, the oldest date from pit 323 (Z-4357) comes from the nearby southern area of the site, which could not have been affected by flood episodes.Further detailed analysis of the material from the southern pits is required.
dered old wood.Most of the dates group between 5800 and 5550 BC although a younger group of dates can be observed between 5500 and probably 5150 BC (Fig. 6).This group of dates belongs to the part of the feature (Z-3925, the youngest hearth from a pit 9 containing burials) or features (Z-3583, pit-dwelling 37; Z-2935 pit 15 containing male burial) situated on the north side of the site.However, if we consider the possible validity of the three oldest dates (Z-4357, Z-3586, Z-3584), then a third phase of the settlement can be added which would represent slightly younger phase than at the Sopot site and an almost parallel phase at Zadubravlje site.
Virovitica -Brekinja site (Fig. 1.4) The site was excavated in 2005 on the future western Virovitica by-pass road.An area of 9000m 2 was explored, but Star≠evo culture features were found only on 5400m 2 .Merely the border area of the settlement was excavated, yielding a few partially explored features interpreted as working and residential areas.Most of the finds were fragments of coarse pottery with a barbotine surface.No painted pottery was found.The excavators date this site to the Spiraloid B phase (Sekelj-Ivan≠an, Balen 2007).
Only two dates are available from this site (Fig. 7), but unfortunately no description of the context from which the samples were taken has been published.The dates confirm the late existence of this site, probably between 5500 and 5300 BC. ❸ Slavonski Brod -Galovo: The oldest date is from a small pit 323 from the south side of the excavated area (Z-4357) while the next two dates (Z-3586, Z-3584) are from the two wooden fences.These three samples group just after 6000 BC.The next group of dates is the largest, and groups between 5800 and 5550 BC.The three youngest dates (Z-3925, Z-3583, Z-2935) range from about 5500 to 5100 BC; these samples were taken from pits 9, 15 and 37 situated on the northern side of the site.❹ Virovitica -Brekinja: Two dates from this site belong to the period between 5500 and 5300 BC 113 3 .
It is worth noting that three of the sites, Sopot, Galovo, and Zadubravlje 114 4 , were at some point covered by one or several layers of sediment.These layers sealed off the sites for a considerable time: at Sopot, for about a thousand years, and at Galovo until the end of the Bronze Age.Life at Zadubravlje site was never renewed.
The sum of all the usable dates from all four sites (Fig. 8) shows the existence of Star≠evo culture in this area roughly between 6000 BC (or slightly earlier) and 5300 BC.Obvious problems with some dates make it impossible to firmly establish the boundaries of phases, but it is possible to establish a time frame for further discussion in this paper.
The overview of 14 C dates provided by Bernhard Weninger et alii (2014) from central Anatolia to north-east Hungary (Fig. 9) can be complemented with the dates available for north Croatia, filling in the geographical gap.It is also clear that Neolithisation reached northern Croatia just after 6600-6000 BC, following after the Bulgarian and Romanian dates 115 5 , but before the Hungarian dates (Hertelendi et al. 1995.242, Tab. 1).On the territory of Serbia the oldest dates come from southern Serbia 116 6 , while the Iron Gates 117 7 and Donja Branjevina site (Budja 2009.127, Tab. 1, 129, Tab. 2;2013.42.Fig. 1) 118 8 show very similar dates to those from northern Croatia.

New solution -RCC 6600-6000 as terminus post quem
For a new interpretation of the beginning of the Neolithic in northern Croatia, it is necessary to have a closer look at investigations of the Holocene climate fluctuations of the northern hemisphere.Recent palaeoclimate research has discovered the existence of a distinctly repetitive series of cooling ano- malies during the Holocene (Weninger et al. 2009;Weninger, Clare 2011.11)termed 'Rapid Climate Change' (RCC) events (Rohling et al. 2002;Mayewski et al. 1997;2004;Clare et al. 2008;Weninger, Clare 2011.11;Budja 2015).The RCC intervals are: 9000-8000, 6000-5000, 4200-3800, 3500-2500, 1200-1000 and 600-150 BP, the most recent corresponding to the LIA (Mayewski et al. 2004;Budja 2007;Weninger et al. 2009;2011;Clare, Weninger 2010).The main cause of these events, in addition to solar intensity weakening (Perry, Hsu 2000;Bond et al. 2001;Mayewski et al. 2004.244;Marino et al. 2009.3246)(Thomas et al. 2007.75;Weninger et al. 2009. 11).This cooling event was amplified between 8.2 and 8.0 ka calBP by the collapse of a remnant Laurentide ice-dome and subsequent drainage of large amounts of melt-water from the Hudson Bay into the North Atlantic (Bauer et al. 2004;Budja 2007. 191; Marino et al. 2009.3246;Weninger et al. 2009. 11-12;2014.8;Weninger, Clare 2011.17)resulting in one of the most extreme climate anomalies of the entire Holocene (Weninger et al. 2014.10).Marine Core LC21 from the southeast Aegean to the east of Crete showed rapid sea-surface temperature variations resulting from the rapid movement of extremely cold air masses over the surface of the Aegean Sea over a distance of some 700km over short periods of time during winter and early spring (Weninger et al. 2009.10;2014.10;Weninger, Harper 2015).Cold and dry air flowing rapidly over a warm ocean surface caused evaporation (Weninger et al. 2009.11),which in turn could have provoked intensive precipitation (Weninger et al. 2009.33).Similar cooling events were documented for the Adriatic Sea (Siani et al. 2013;Budja 2015.172).However, in Eastern Mediterranean periods of extreme drought were documented, such as the low levels of Dead Sea (Migowski et al. 2006;Budja 2007.194;Weninger et al. 2009.15-16)(Fig. 10.H), during which 'flash-flood' events would have had devastating results on the environment and human occupation (Weninger et al. 2009.33).Weninger et alii (2014. 14) distinguish two phases: an earlier phase at 6600-6200 cal BC (RCC only) and a later phase at 6200-6000 cal BC (RCC amplified by Hudson Bay impact).
Northern Africa was arid until 8500 BC, when an abrupt arrival of monsoon rains created savannahlike environmental conditions, which prompted the swift population of the area by 7000 BC; conditions changed back to arid after 5300 BC (Kuper, Kröpelin 2006.806.Fig 3).Nile River sediments record an enhanced riverine contribution between 8600 and 5500 calBP, indicating enhanced precipitation in north-east Africa (Magny, Combourieu Nebout 2013. 1449).The aridity in North Africa, southern Near East and Middle East is due to the reduced northward migration of the Intertropical Convergence Zone (ITCZ) in summer (Berger, Guilaine 2009.41).
Southern Europe shows similar records, indicating that it should be integrated into the same climatic zone during the 8200 calBP event (Berger, Guilaine 2009.41).
In southern-central Europe, pollen spectra show a sudden disappearance of Corylus avellana (hazel) and rapid expansion of Pinus (pine), Betula (birch) and Tilia (lime) and an invasion of Fagus sylvatica (beech) and Abies alba (fir).This change in vegetation is thought to relate to annual temperature decreasing by about 2-3°C and increased moisture availability.The rapid retreat of drought-adapted Corylus was probably caused by taller and longerlived trees (Pinus, Betula, Tilia, Quercus, Ulmus, Frayinus excelsior, etc.) (Tinner, Lotter 2001;2006;Budja 2007.194;Nafradi et al. 2015. 14, Dörfler 2013.322)(Dörfler 2013.329).In north-west Romania (Preluca Tiganului and Steregoiu) pollen-based analysis showed a drop of 1.5-2°C in mean annual temperature, 2-4°C in mean temperature of the coldest month, and an increase in mean temperature of the warmest month, while precipitation shows a de- crease of about 200mm.This suggests an intensification of seasonality and continentality (Feurdean et al. 2008.500).

.510).
There is also evidence for a major period of flooding in several parts of Europe, i.e. central Europe (fluvial episode associated with a massive accumulation of tree trunks in the Rhine Valley between 8500 and 8000 cal BP -Berger, Guilaine 2009.37-38),Britain, the Danube delta and specifically the French Alps between 8250 and 7950 BP (Bonsall et al. 2002.4).A recent study by Gerardo Benito et alii (2015) combines more than 2000 14 C and OSL dated flood units from 12 regions in Europe and North Africa (Benito et al. 2015.2,Fig. 1), but unfortunately this study does not include data from regions between the Alps and the Black Sea including the entire Carpathian basin 1 19 9 .Studies have shown that intensive runoffs of the Po, coastal Italian, Apennine and Albanian rivers may have had an impact on lowering salinity during the middle part of the Holocene in the south Adriatic Sea (Magny, Combourieu Nebout 2013.1449;Siani et al. 2013.505).19 It is, however, interesting to note an increase in flooding after the 8.2 ka event, first in temperate European regions and slightly after in Mediterranean regions (Benito et al. 2015.4, Fig. 3).These flood episodes in both regions correspond to IRD 5b, i.e. to the period between 5700 and 5100 cal BC (Gronenborn 2009.100;Benito et al. 2015.4.Fig. 3), adding to our understanding of climatological conditions during the early Neolithic in northern Croatia and especially during its end.
The end date for the 8.2 ka event (Weninger et al. 2014) falls at 6000 BC, or slightly before.The available radiocarbon dates from the early Neolithic period from the Sava-Drava-Danube interfluves fall within the same time range.Thus, the end of the RCC 6600-6000 BC can be taken as a terminus post quem for the appearance of the early Neolithic in this region.It is important to stress that the full 'Neolithic package' is present at all the above-mentioned sites 2 20 0 .
Eszter Bánffy and Pál Sümegi (2012) discuss the role of the Central European-Balcanic Agro-Ecological Barrier (CEBAEB) 221 1 in the Carpathian Basin as a zone of interaction between immigrant farming communities and the local forager population, although the reasons for the existence of the barrier were manifold: the climatic conditions on the southern side of the boundary were favourable, but the soil was sandy and unsuitable for long-term settlement and crop cultivation (Bánffy, Sümegi 2012.59),especially given the limited agricultural knowledge of immigrant communities adapted to different conditions in south-eastern Europe; the specific climatic conditions in Carpathian Basin due to the basin effect (Bánffy, Sümegi 2012.59) which could trigger a unique transformation in vegetation and soil types; wet habitats and sandy ridges more suitable for a Mesolitic population based on hunting and foraging (Bánffy, Sümegi 2012.60)etc.The German oak tree-ring record 222 2 (Fig. 11) shows another low between 6000 and 5900 BC, possibly indicating unfavourable climate conditions such as excess of precipitation that could have caused large-scale and unpredictable floods (Bonsall et al. 2002;Bonsall 2007) representing impenetrable barriers, although their intensity could have been less pronounced than in the previous 8.2 ka period.However, some of the barriers recognised, such as the one in the Kalocsa area and Tolna Sárköz region (the southern Danube region in Hungary), are difficult to explain using ecological/environmental arguments (Bánffy 2013b).Bori≤ (2011) does not see the gap in the settlement of the area between the late Mezolithic and early Neolithic in the Iron Gates gorge; gaps can be seen at certain sites, while others were settled during that time.Bori≤ also does not agree with Bonsall (Bonsal et al. 2002;Bonsall 2007) that largescale floods were the reason for discontinuity of occupation of sites, but considers different economic strategies to be the reason for that.It is also clear in the Iron Gates gorge that Early/Middle Neolithic (5950/5900-5500 BC) settlements are at least to a point situated in different environments and on different soils (such as Aria Babi) (Bori≤ 2011.183).
The question is whether some kind of ecological barrier existed in the Sava-Drava-Danube interfluve, influencing the emergence and spread of Neolithisation in this region or whether this spread was dictated by some other reasons, such as in the Tolna Sárköz region.So far, specific environmental analysis compared with archaeological data has not been done for the Sava-Drava-Danube region, and only circumstantial data are available which might give an insight into the climatic conditions of this region during the Neolithic.In a recently published paper (Pearson et al. 2014) two sub-fossil wood samples are of interest for the region discussed here.Samples 25 from O∏tra Luka near Ora∏je and 151 from the mouth of Krapina River near Zagreb yielded very interesting dates (Fig. 12-13).The date for sample 25 from O∏tra Luka is almost identical to three dates from Sopot (Fig. 3) and the date for sample 151 from the Krapina River coincides with the beginning of flood episodes (Benito et al. 2015.4, Fig. 3c), which in turn coincide with IRD 5b (Bond et al. 2001.2131, Fig. 2;Gronenborn 2009.99, Fig. 2;Benito et al. 2015.4, Fig. 3b).As the extreme flooding in spring 2014 showed, extremely high precipitation in the central Bosnian mountain region can provoke extreme episodes of flooding of the Vrbas, Bosna and lower flow of the Sava rivers.This scenario may have happened around 6000 BC or slightly after; debris could have sealed off the Sopot site for the next 1000 years 223 3,224 4 .
Interestingly, in the Proko∏ko jezero pollen record, a significant drop in the Corylus record and a rapid increase of Fagus are noted around 6000 BC (Dörfler 2013.322, Fig. 11).This episode could have been provoked by increased precipitation.
The early Neolithic in Sava-Drava-Danube interfluve shows all the characteristics of full Neolithic package.White painted motifs on some fragments from the Galovo and Sopot sites suggest the earliest Linear A phase, although these fragments are accompanied by fragments with black painted motifs.The percentage of both white and black painted pottery is very low in the overall amount of pottery from all sites.There is a difference of opinion as to whether a Monochrome phase of early Neolithic existed throughout south-east Europe preceding the 'developed' Neolithic (Stojanovski et al. 2014.24):Lolita Nikolova (2007.91)puts Hoca Çesme 4-3, Krajnitsi 1, Divostin 1, Koprivets 1, Dzhulyunitsa, and Smurdesh 1 sites in this phase, dating it to ?6300-6200/6100 cal BC, while Krauß (2011) argues that the amount of painted pottery during the whole early Neolithic in south-eastern Europe is notably below 10%, and that painted pottery was present from the very beginning in this region 225 5 .This latter can be applied to Sava-Drava-Danube region: the amount of painted pottery is well below 10% at observed sites and it is already present in the earliest dated sites.So far, no traces of Star≠evo Monochrome phase have been found in our observed region (Minichreiter 1992.54;2007b.23).Nevertheless, some sites in the Balkans were attributed to that earliest Monochrome phase: Krajnici, Koprivec and Poljanica-platoto in Bulgaria; Divostin, Donja Branjevina and Grivac in Serbia; Pe∏terica, Rudnik and Grn≠arica in the Republic of Macedonia (Krauß 2010.41;Stojanovski et al. 2014), Foeni-Sa ˘las and Gura Baciului in Romania (Drasovean 2007.70, Fig.2-4;Luca, Suciu 2007.79;Krauß 2010.36).
In southern Banat, a part of the Iron Gates region, only two sites are known: Hotu Cave, near Steierdorf (Anina) and Ba ˘ile Herculane-Hotilor Cave (Krauß 2014.195).These sites are situated a few kilometres north of the Iron Gates in the hilly parts of the southern Carpathians, while in the northern part of the Banat, the Mesolithic is as yet unknown (Krauß 2014.195-196).In Ba≠ka, Mesolithic sites are documented at Hajdukovo close to Subotica and at Ba≠ka Palanka, west of Novi Sad; the sites are located exclusively on the banks of the alluvial plains of the rivers (Krauß 2014.196).In Transdanubia, the Regöly 2 site was recently explored, but there are some older finds from the vicinity of Győr, from the Sárrét bog, Csór-Merítőpuszra, next to the village of Nádasdladány and from a sand dune on the flood plain of the Kapos River at Kapshomok (Krauß 2014.196).In Alföld, excavations were carried out at two sites in the Sződliged/Vác area, together with intensive survey in the Jászság area.The sites at Sződliged are situated close to the Danube and are compared to Iron Gates sites by the excavators (Krauß 2014.201).Extensive surveys were carried out in Jászág Basin, the large flood plain between the upper Tisza and Danube rivers, and two Mesolithic sites were found: Jászberény I and Jásztelek I ( Krauß 2014.201-202).Five more sites from the Alföld have been mentioned in the literature, some of them situated on the sandy dunes of the floodplains or close to major rivers (Krauß 2014.212).In the Sathmar district of north-west Romania Ciumesti II site is also situated on a dune (Krauß 2014.212).
Maria Gurova and Clive Bonsall (2014), comparing known data for Upper Palaeolithic and Late Mesolithic sites in the Balkans, show the peripheral distribution of latter sites within the Balkans.In their assessment, most of the sites are located within 50km of the sea or the Danube.One of the reasons for Mesolithc peripheral site distribution could have been the change in vegetation cover -from semi-desert steppe and forest-steppe ecosystems to a major expansion of temperate forest during the early Holocene.Dense forests protected animals, thus reducing hunting productivity and they have also "posed significant challenges for inter-group communication and participation in viable mating networks" (Gurova, Bonsall 2014.98).Edible plants in temperate forest ecosystems are found on the forest margins, such as upland tree lines, recently burned areas or along sea, lake and river shores.Aquatic resources became an essential substitute for the lack of animal biomass (Gurova, Bonsall 2014.97-98).
Bearing in mind the specific site positions on dunes near large rivers or at elevations near floodplains, the change in resources due to the change in the environment and appearance of dense temperate forests, it is clear that the lack of Mesolithic sites in the Sava-Drava-Danube region, specifically eastern Slavonia, was partially due to environmental change at the beginning of the Holocene.The only large river in this area is the Danube, while the Sava and Drava plains experienced regular seasonal flooding, but were also covered by dense forests.Mesolithic settlements can nevertheless be expected on the high right margin of the Danube and ∑akovo-Vinkovci loess plateau.However, intensive occupation of these regions from the Late Neolithic to modern times reduces the hope of finding surviving Mesolithic sites.

Conclusion
The appearance of the early Neolithic, i.e.Star≠evo culture in the Sava-Drava-Danube interfluve can be placed around 6000 BC or just slightly before, as current research and radiocarbon dates show.The earliest dates and finds come from the Sopot site situated in the south-eastern part of Slavonia.After a short period of settlement during this initial Neolithic phase, an episode of flooding probably sealed off the site for the next 1000 years.Shortly after the abandonment of Sopot site, the Zadubravlje and Galovo sites situated in the Sava valley to the west were settled.Both sites show traces of flooding, which was probably the reason why both sites were abandoned and never resettled (except for the late Bronze Age necropolis at the Galovo site).Most of the dates from Zadubravlje and Galovo group after 5850 BC, although two older dates, slightly younger than the Sopot dates, come from the Zadubravlje site and four dates from the Galovo site.The youngest group of dates can be observed at the Galovo and Virovitica -Brekinja sites (with one date from Zadubravlje) starting after 5500 BC.The end of Star≠evo culture can be expected before 5100 BC.Radiocarbon dates also indicate the north-western spread of Star≠evo culture to the Drava valley around 5750 BC and later.
Taking climatological/ecological data into consideration, the beginning of Star≠evo culture can be linked to the end of the 6600-6000 BC period of Rapid Climate Change (8.2 ka cal BP) but it cannot be expected earlier than 6100 BC.Similarly, the end of Star≠evo culture can be linked to the 7.1 ka calBP event (5300-5100 BC).
There is circumstantial evidence of climatic instability in this region, i.e. excess of precipitation that can probably be linked to flooding responsible for sealing off the Sopot site and for the unfavourable conditions prior to the Drava region settlement around 5750 BC.
The white painted pottery found at Sopot and Galovo sites shows the connection to other regions of south-east Europe in the Neolithisation process.This connection is also visible in available radiocarbon dates.Weninger et alii (2014.6)give the best summary of the process: "The moment the Neolithic left the Aegean basin, which appears to have occurred not earlier than 6100 cal BC, it apparently took little more than 100 years to become established at sites in Serbia, Bulgaria, and Romania, and little more than around 200 years even to have reached the Pannonian Basin." Although the climate was not the primary reason for cultural change, and in some cases played almost no role in it (Budja 2015;Flohr et al. 2015), it is worth noting that radiocarbon dates from early Neolithic sites of the Sava-Drava-Danube interfluve fit well between documented RCC periods; further archaeo-logical, geoarchaeological and palaeoenvironmental research is needed to firmly establish a cultural, climatological and environmental connection.The term archaeological climatology or archaeoclimatology, introduced by Fabian Welc (2016) describing a subdiscipline within geoarchaeological science, can be applied to this future interdisciplinary research.

Epilogue
During spring 2016, a small scale geoarchaeological survey in eastern Slavonia was conducted by Zagreb Institute of Archaeology and the Institute of Archaeology, Cardinal Stefan Wyszynski University in Warsaw with the cooperation of the Faculty of Geology, University of Warsaw.The survey consisted of geological sampling and a geophysical survey and was conducted at, among other places, at Slavonski Brod -Galovo and Sopot sites.At the Galovo site, a core of 9m was taken, while at the Sopot site the depth of the core was 3.6m.Magnetic susceptibility preliminary results show an episode of drought followed by an episode of extreme wetter conditions just at the end of life at the Galovo site, while the cultural layer at Sopot site was sealed off by a pronounced wet episode.The interchange of dry and extremely wet conditions can be compared to the eastern Mediterranean periods of extreme drought during which 'flash-flood' events occurred (Weninger et al. 2009. 33).The full results of this survey will be published in the future.
, appears to have been a strengthening of atmospheric pressure gradients between Siberian (High), Iceland (Low) and the Azores (High), conditions supporting an influx of extremely cold air from the polar regions into Europe(Clare et al. 2008;Weninger et al. 2009;Clare, Weninger 2010;Weninger, Clare 2011;Weninger et al. 2014; Weninger,  Harper 2015.478,Fig. 2; Benito et al. 2015.5).RCC events correlate with ice-rafted debris (IRD) events(Bond et al. 1997) although not entirely.Further-more, the North Atlantic Oscillation (NAO) positive phases may have been connected to these cold events(Bout-Roumazeilles et al. 2007.3212).The most severe of these events was the 8.2 ka event (6600-6000 BC)(Alley et al. 1997;Magny et al. 2003; Alley, Ágústsdóttir 2005;Thomas et al. 2007;Clare et al. 2008;Gronenborn 2009;Marino et al. 2009;Weninger et al. 2009;Weninger et al. 2014) during which temperatures in the North Atlantic region dropped over the course of the subsequent 160 years
take the archaeological context more into consideration but recognise the possible climatic influence on Neolithisation process from the Middle East to central Europe or at least deal with ecological/environmental conditions that could have enabled or prevented the spread of Neolithisation at specific point in time.These studies seem to indicate a halt in Neolithisation advances caused by the 8.2 ka event, which led to unfavourable conditions across certain European regions, mainly the Carpathian basin and other regions of temperate Europe.

Fig. 11 .
Fig. 11.Timing and structure of the 8.2 ka calBP event on a global scale.The relevant time frame for the beginning and duration of early Neolithic in northern Croatia is marked in darker green (modified after Weininger et al. 2014.15.Fig.9).
This paper presents partial results of my doctoral research in the framework of Zagreb Institute of Archaeology's Archaeological landscapes and identities (A1 and A2) research programme.I would also like to thank my colleagues Dr K.Minichreiter (Zagreb), M. Krznari≤ πkrivanko (City Museum Vinkovci) and Dr I.Krajcar Broni≤ (Rudjer Bo∏- kovi≤ Institute, Zagreb)for all the information provided.