1. Climatic and human impact on the environment?: a question of scale.

: The title of this Special Issue of QI - ‘The Neolithic of Northern Greece and the Balkans. The environmental context of cultural transformation’ - frames the central issue of this paper – how were Neolithic and Chalcolithic landscapes in the Aegean, Balkan and Carpathian (ABC) zones shaped and transformed by climatic and anthropogenic impacts? The difficulties in interpreting proxy records for the middle, transitional stage of the Holocene aridification sequence, falling between the early wet stage and the late arid stage, have been created by the conjoint influence of two kinds of impact – climatic and anthropogenic. An unhelpful influence in this debate stems from Willis and Bennett’s (1994) hypothesis of minimal human impact on the pre-Bronze Age landscapes of South East Europe. In this paper, two questions are posed: (1) what were the effects of the claimed global changes in Holocene climate at the regional and local scale in the ABC zones?; and (2) can we recognise human impact in these proxy records prior to the Bronze Age of our study regions? Following a discussion of general long-term climatic trends and RCCs (episodes of rapid climatic change), I base a discussion of the so-called 8200BP ‘event’ and pre-Bronze Age human impacts on a suite of 24 well-dated proxy records – mostly pollen sequences. The principal findings are that there is little evidence for impact from the 8200BP ‘event’ in these records, while there is substantial evidence for pre-Bronze Age human impacts on the landscapes of the Aegean, Balkan and Carpathian regions.


Introduction
The middle Holocene period witnessed one of the most dramatic periods of change in European prehistorythe spread of a farming way of life from Western and North West Anatolia to the Aegean zone (zone A), the Balkans (zone B), the Carpathians (zone C) and beyond into North-Central Europe. This period of change in three contiguous and inter-related zonesthe ABC of South East Europewas the focus of the Saloniki Conference from which this Special Issue of QI has been distilled. The title of 'The Neolithic of Northern Greece and the Balkans. The environmental context of cultural transformation' frames the central issue of this paperhow were Neolithic and Chalcolithic landscapes in the ABC zones shaped and transformed by climatic and anthropogenic impacts?
The general climatic framework for this discussion is the three-stage division of the Holocene into an Early Holocene wetter stage, a transitional stage and a Mid -Late Holocene aridification stage. Several commentators (e.g., Roberts, N. et al., 2011) have suggested that the difficulties in interpreting proxy records for the transitional period (7000 -4000 cal BC: Brayshaw et al., 2011; or 5000 -3500 cal BC: Galop et al., 2009) relate to the unknown strength of anthropogenic influences on local and regional ecologies in relation to climateforced changes to vegetation history. There is still no agreement on the causes of the aridification trend or the ways in which this was materialized in proxy records. In a discussion of both the climatic and the anthropogenic impacts on middle Holocene landscapes, I seek to answer two questions: (1) what were the effects of the claimed global changes in Holocene climate at the regional and local scale in the ABC zones?; and (2) can we recognise human impact in these proxy records prior to the Bronze Age of our study regions? It is inevitable that I confront issues of scale in these questions, from global trends to local events. An example of scalar issues is the way that annual changes in grain-sowing or hunting strategies represented much more fine-tuned practices than the colossus of global climatic changes such as the 8200BP event. I approach this topic from the viewpoint of an environmentally-aware prehistorian with a scepticism to environmental determinism, on the grounds that human communities must have been flexible enough to react to, if not to predict, the directions in regional environmental changes and make thoughtful choices about where to live and what cuisine (food and drink) to select. Here, I shall make "the default assumption … that the humans and non-humans are mutually implicated -that they co-constituted the world" (Head, 2008: 376). It is clear that this approach relates closely to the approach to cultural entanglements made by Hodder (2012). This approach contrasts with what most palaeo-environmental scientists working in the Holocene of the Balkans and the Carpathian Basin take as a 'normal' research goal -the identification of 'human impacts' on the 'natural' vegetation (e.g., Willis & Bennett, 1994;Magyari et al., 2012;Connor et al., 2013).

Holocene climatic trends
A broad perspective on climatic trends in the study region depends increasingly on the results of the analysis of large data sets, consisting of as many pollen diagrams or other multi-proxy sequences as are available, often at a millennial time-scale. As Giesecke et al (2011Giesecke et al ( : 2809 observe, "a large number of Holocene climate shifts or short-lived excursions are reported in the literature so that it seems almost possible to find one within the uncertainty of any standard radiocarbon date." Because of forest resilience, a high-amplitude or long-lasting shift in climate parameters is needed to produce vegetational change recognizable at a millennial timescale. In this account, I do not focus on the effects of the Younger Dryas phase (10950 -9750 cal BC: Straus & Goebel, 2012), since it is earlier than the periods discussed in this Special Issue. However, the impact of the '8200BP event' is an important part of the general climatic story in later Balkan prehistory (8000 -4000 cal BC).
The quantity and quality of palaeo-environmental research in the Balkans and the Carpathian Basin has improved enormously over the last decade, particularly in respect of well-dated proxy sequences. A key publication milestone was volume 21 of the journal 'The Holocene' (2011), devoted to Holocene climate change. We can identify two cross-cutting trends in this research narrativelong-term trends in European climate and episodes of rapid climate change (henceforth 'RCC', following Denton & Karlén, 1973).
In the former, the stadial terms used since the Blytt-Sernander system (Pre-Boreal, Boreal, Atlantic and Sub-Boreal) have been replaced by a three-stage division of the Holocene into an Early Holocene wetter stage, a transitional stage and a Mid -Late Holocene aridification stage. The Early Holocene stage was a period dated 9500 BC to 5000 BC by some (Galop et al., 2009), while others suggest aridification began earlier, in the early 7 th millennium BC Sadori et al., 2011) or cca. 6000 cal BC . N. Roberts et al. (2011) identify a stable Early Holocene boundary in the Adriatic Sea between a wetter Eastern Mediterranean and a West Mediterranean zone where warm, wet westerlies had less impact. It is important to recall that major glaciers continued to exist until cca. 4800 cal BC, cooling the global climate mainly through the introduction of melt-water into oceans (Wanner et al., 2008). The Late Holocene aridification phase marks a period of decreasing precipitation in the east Mediterranean, beginning at some point in the 4 th millennium cal BC (Galop et al., 2009;Brayshaw et al., 2011) and continuing until the present day. The effects of these East Mediterranean climatic trends are important for our understanding of climatic change in the ABC zones, as much as human dwelling in these regions.
The identification of synchronous RCC episodes has been attempted by, inter alia, Majewski et al. (2005), Magny (2006) and Giesecke et al. (2011). The last-named underline that such efforts are based upon the acceptance of one of the two main climatic hypothesesthe dynamic equilibrium hypothesis (Prentice et al., 1991), by which directional changes in climate can produce changes in the spatial patterning of species distributions 1 . Since Majewski et al. (2005) work at a millennial time-scale, it is hard to imagine the effects of RCCs on local communities because of the fuzziness of their temporal definition. Thus, Majewski's 7 th millennium cal BC RCC -termed the 'Glacial Aftermath' phase -may well be tied into the '8200BP event', while it is possible that his 4 th millennium cal BC RCC, marked by Alpine glacial re-advances and increases in the tree-line, is related to the inception of the Mediterranean aridification stage. Magny's research into long-term West-Central European lake levels correlates higher-than-average lake-levels with higher annual precipitation, lower summer temperatures and a shorter growing season. The greater chronological precision of his regional database enables Magny to identify three RSS episodesat 7600 -7200 cal BC, 6350 -6150 cal BC (equivalent to the '8200BP event') and 4400 -3950 cal BC (perhaps coeval with Majewski et al.'s later RCC episode). Thirdly, in a wide-ranging test of the dynamic equilibrium hypothesis conducted at a centenarial timescale, Giesecke et al. (2011) highlight three peaks of RCC in the majority of the 59 proxy records under study: the start of the Holocene period, the '8200BP event' and an episode in the late 5 th millennium BC (perhaps related to Magny's third RSS episode). In vegetational terms, Giesecke makes a strong case for synchronous expansion of Corylus avellana from a variety of refugia in the Early Holocene.
What we can infer from these studies is that the timescale of the analysis is critical in producing useful results for regional and local social practices. There does appear, however, to be agreement on an earlier RSS episode (the '8200BP event') and a later episode (dated somewhat less precisely in the late 5 th millennium cal BC). How was the '8200BP event' caused and what results did it produce? Peyron et al. (2011: 141) have published a succinct account of the 8200BP 'event', which they maintain was triggered by a weakening of the thermohaline circulation in the North Atlantic, in turn leading to more ice-cover in Baltic Seas, stopping the penetration of mild, moist Atlantic air into Europe and allowing greater penetration of the Eurasian/Siberian high, which led to cooler, drier winters and springs (cf. Pross et al. 2009). If summer cooling is invoked as a major explanatory factor for the 'event' (Heiri et al., 2003), climatic modeling would produce the biggest fall in the Alpine tree-line in the whole of the Holocene (Heiri et al., 2006). While the mechanism for the 'event' is clear, however, the chronological resolution of the 'event' is by no means well-defined. According to Giesecke et al.'s (2011) results, the 'event' had effects on two sets of pollen proxy records: one group dated 7600/7450 BC -7100 BC and the other group dated 7400/7100 BC -6800/6600 BC. They suggest an initial change in vegetation, with progressive destabilizing effects on other vegetation cover, such that the 'event' acted as a large-scale disturbance. However, the actual timescale of the initial trigger in what is called an 'event' (but is clearly nothing like an event!) remains mysteriousperhaps centuries -with the most comprehensive investigation (Giesecke et al., 2011) characterizing the effects of the 'event' over 700 years and Weninger et al. (2014) embedding the '8200BP event' into a 600-year rapid climate change interval (6600 -6000 cal BC). It is possible to conceive of an 'event of such major significance that its effects were profound and long-lasting but I await a chronological definition of the trigger 'event'. Weninger et al. (2009) have been the strongest advocate of major cultural impacts from the rapid climate change events identified by Majewski et al. (2004). The 2006 paper (Weninger et al. 2006, Figs. 3 -5) exploring the archaeological effects of the '8200BP event' presents cumulative 14C diagrams to check whether there were declining regional settlement at the time of the 'event'. Despite their own data, which show that Greece, Cyprus and Bulgaria, as well as a majority of sites (e.g., Nea Nikomedeia) showed cumulative 14C peaks at the time of the 'event', Weninger et al. continue to claim (p. 401) that this aridity event triggered the spread of early farming in the East Mediterranean! This is as sloppy as Ghilardi et al.'s (2012) claim that a marine transgression in the NW Аеgean preceded the first occupation of Early Neolithic Nea Nikomedeia by a few centuries, when, in fact, the recent re-dating of this site (REF.) makes the interval more like a millennium. Again, Ghilardi et al's claim that the onset of a lagoon environment near Nea Nikomedeia was related to the '8200BP' event ignores their own dating of the lagoon formation to 5826/5659 calBC -centuries after the 'event'.
In the second paper (Weninger et al. 2009), by an inversion of the usual invasion route (Clark 1965), Weninger's team proposed that the major disturbance to Greek and Balkan vegetation was a series of extremely cold air masses that swept over the Balkans and down into Greecepart of the Eurasian/Siberian High proposed for the '8200BP event'  also striking the region between 4000 and 3200 BC. While, on this occasion, Weninger's team does not comment in detail on the relationship between the '8200BP event and the spread of farming through the Balkans, the team proposes that the 4 th millennium rapid climate change episode had a deleterious effect on Climax Copper Age tell settlement structures, whether in Greece, Bulgaria and Romania The weakness of this proposal is that the dates of the episode do not correlate well with the supposed changes in settlement pattern in any of the three areas. The alleged Siberian High pre-dates the dwindling site numbers in the Final Neolithic); reliance on Todorova's (2002a) reconstruction of catastrophic environmental change in Bulgaria is hardly wise since this is based upon a very outdated sealevel reconstruction (Fairbridge 1961); and the abandonment of Gumelniţa tells post-dates the alleged 'event' (while Pietrele may have been abandoned c. 4250 cal BC, it is by no means the latest occupation of tells in the Lower Danube valley, which go on into the early 4th millennium calBC. Even if there was a closer chronological correlation between the alleged climatic cause and the settlement effect, no flexibility is afforded the prehistoric communities in finding ways to adapt to fluctuations in climate. The latest paper from Weninger's team (Weninger et al. 2014) presents an extraordinary mass of 14C data, coring results and site-based archaeology but completely fails to link all of these data sets to support their main proposal: "We demonstrate a precise temporal coincidence (within given error limits) and strong social impact of rapid climate changes on Neolithic dispersal processes" (2014, 2). An important change in Weninger et al.'s approach (2014, 8) is to propose that the '8200BP event' (viz., the Hudson Bay outflow) was embedded within one of several Holocene rapid climate change intervals lasting 600 years (6600 -6000 cal BC). It is argued that these intervals were defined by more pronounced bouts of Siberian High pressure originating in Central Asia and moving over the North Pontic, South East Europe, the Aegean and the Adriatic. However, although Weninger et al. produce summaries of the latest dates from early sites in Turkey, Greece and North Bulgaria, there is no explanation of how the alleged Siberian High affected these sites, whether to cut short their occupations or to initiate supposed demographic shifts to new areas. The absence of any serious discussion of settlement archaeology, not to mention subsistence studies or social arguments, makes this paper of limited value.
A fourth example of exaggerated claims for the '8200BP event' comes from the re-analysis of the Tenaghi Philippon pollen core, Northern Greece (Pross et al. 2009). Here, a massive disturbance in terrestrial ecosystems, with greater aridity and lower winter temperatures, is In summary, the traditional climate narrative of a cooler and wetter Anathermal, or Pre-Boreal and Boreal, a 'Thermal Maximum' termed the 'Altithermal', or 'Atlantic' period (5500 -2500 cal BC), and a cooler and wetter Medithermal , or Sub-Boreal and Sub-Atlantic, has been replaced by a parallel three-stage narrative based upon an Early Holocene wetter stage, a transitional period, and a Mid -Late Holocene aridification stage. There is still disagreement among the palaeoclimatologists about the causes of the changes in their proxy records, especially in the transitional stage. The hitherto imprecise timescale of Rapid Climate Change episodes in the Holocene proxy records limits their utility in discussions of climatic impact on cultural development. While there is widespread agreement about the significance of the 8200BP 'event' in general terms, it is hard to date the duration of the trigger event within two to four centuries, and the dates of its effects can extend to over 700 years. It remains difficult to relate the effects of the 8200BP 'event' to any particular cultural development or collapse, let alone any settlement dislocation. Can the impact of the 8200BP 'event' be seen in mid-Holocene vegetational records?

Mid-Holocene vegetation records and the 8200BP 'event'
This summary of trends in mid-Holocene vegetational histories is based upon 25 proxy records, for the most part pollen sequences, all of which dated by a minimum of six AMS calibrated BC dates ( Fig. 1 & Tables 1 -2). The sites range in altitude from sea-level (Mljet, Black Sea cores, Durankulak Lake, Mount Athos Basin) to 2190masl, at Lake Ribno Banderishko in the Pirin Mountains, SW Bulgaria. For this review, sites are classed as 'lowland' (< 300 masl); 'upland' (300 -1000 masl) and 'mountain' (>1000 masl). The East -West transect of sites ranges from Griblje (Slovenia) in the West to the Heraklea Peninsula in the Crimea in the East.  (Table 1). Both the East -West and the North -South transects will be discussed so as to reach a general picture of vegetational change for the Mesolithic, Neolithic and Chalcolithic of the Balkans and the Carpathian Basin. Although the effects of mixed farming practices are often difficult to separate from climatically-forced ecological changes (e.g., Magyari et al. 2012), the former is considered in a later section, with a focus on a smaller number of proxy records.
The proxy records with sections dated between 6500 and 5000 cal BC were examined with a view to detecting the effects of the 8200BP 'event'. The overall increased winter precipitation record at Taul But there is an overall continuity, if not expansion, of deciduous oak forests in much of the region.
The variety of dominant species (Corylus at Avrig and Kiri-tó, Fraxinus at Mohos, Corylus at Turbata and Ulmus at Preluca) indicates considerable diversity of deciduous forest cover at the inter-regional as well as the local level. Two species which later came to prominence over much of temperate Europe -Carpinus and Fagusare recognized for the first time in several areas. Carpinus develops at all altitudes -in the lowlands (Lake Stiucii), the upland zone (Prespa) and the mountain zone (Pirin), while the expansion of Fagus occurred in the North-Westernmost site of Griblje and the SE Hungarian lowland site of Kiri-tó, with special prominence at Prokoško Jezero. A general trend in the late 7 th -6 th millennia BC would appear to be a greater vertical differentiation in the zonation of woodland, offering wider opportunities to communities which could integrate their (seasonal) movements with this differentiation.
It is significant that there are signs of vegetational change that could be attributed to the impact of the so-called 8200BP 'event' in far fewer than half of the proxy records. There is a fall in rainfall in the Sofular Cave record, while there are drops in overall Arboreal Pollen combined with increases in Artemisia, Chenopodium and/or Poaceae peaks at varying altitudes in the Southern Balkans (Maliq, Prespa and Tenaghi Philippon). It is also possible that this 'event' influenced the development of a true coniferous altitudinal zone in the mountains (Pirin). However, the vast majority of proxy records dominated by mixed deciduous forest have shown no or minimal impact from the 8200BP 'event'.
The most likely interpretation of these data is that the climatic impact of the 8200BP 'event' must have been small to minimal for the communities living in such a wide range of environments as were present in our study region. An alternative interpretation is that the 8200BP 'event' had serious consequences for people and landscapes but that these impacts were so short-term and rapidly reversible that no impact could be detected in the vast majority of the proxy records. Indeed, the Mount Athos Basin core showed five centennial-scale climatic perturbations between 7300 calBC and 4500 calBC, marked by the abrupt decline of Quercus and a concomitant rise in non-steppe grasses (especially Cichorioideae), suggesting that human communities could well have learned from their previous experience and built in a certain resistance to these changes.

The interaction of human practices and changing environments
The proximity of the Balkans and the Carpathian Basin to the steppe zones of the Near East, Anatolia and the North Pontic meant a considerable diversity of plant associations in the Early Holocene period. In contrast to the closed deciduous oak woodlands of the temperate zone (e.g., the Lower Oder valley diagrams, Northern Germany: Jahns, 2000), there were parts of the lowland landscape which were naturally open, with normal steppe components such as large-grained grasses (Poaceae, Artemisia, Chenopods, Plantago, Rumex and Polygonum). For this reason, we cannot automatically interpret such species as indicators of 'human' activities, as would have been the case in Northern Germany, at least without the presence of additional plant species (Connor et al., 2013). Such open areas, juxtaposed with various woodland associations, would have been attractive to the earliest farmers, whose cropping and pasturing was on such a scale that required little, if any, deforestation. Equally, the natural forest fires which remained at a high level from 8700 -5100 BC (Feurdean et al., 2013) would have helped to maintain natural clearings in the woodland canopy, especially near rivers and streams. These open woodlands would have made attractive places for dwelling for the earliest farmers, whose tool-kit did not, for the most part, include large stone felling axes for extensive forest clearance (Chapman, in prep.).
Two decades ago, Willis & Bennett (1994) proposed an influential two-part hypothesis in which they argued (1) against Neolithic 'human impacts' on the Balkan -Carpathian landscape and (2)  Turning to the pollen proxy records which registered indicators of human farming practices (Table 2), there are five pollen sites within 1km of a prehistoric site; five diagrams within 5 -10km of a settlement; and six pollen sites with regional evidence for prehistoric settlement, often 25 -50km from the coring site. The overall expectation would be that the strongest signals for farming would occur on coring sites closest to settlements. This is the case for the 5 th millennium BC sections of the Sárlo-hát core, the 5 th -4 th millennium BC sections of the Durankulak and Nebelivka cores (Albert et al., in press) but not for the Kiri-tó core, close to Phase 2 Ecsegfalva 23 (Whittle, 2007). While the higher level of local burning at Ecsegfalva was attributed to local settlement practices, there were low frequencies of cereal-type pollen and secondary indicators (Willis, 2007). The small size of the Körös settlement, with its permanent garden cultivation, is the most likely reason that relatively few traces of farming were found in the nearby lake corea result supporting the Willis and Bennett thesis.
However, there is indubitable evidence from both Sárlo-hát and Durankulak for intensive agriculture and/or extensive pastoralism prior to 4000 BC, as well as from Nebelivka in the first centuries of the 4 th millennium BC -causing the rejection of the Willis & Bennett thesis.
The most intriguing example is the Lake Maliq core, where the only traces of human impact produced by the long-term occupation at the lake-side site of Maliq (dated 6200 -3000 BC) came in the earliest dwelling phase in the form of a short-lived forest clearance (Denèfle et al. 2000;Bordon et al. 2009).
In the next group of pollen coring sites, each of the authors emphasizes the small-scale nature of the farming practices attested in their cores. The Straldzha core, located in the Yambol Basin in SE Bulgaria, lies close to several 6 th -5 th millennium BC tells, whose farming practices were marked by a continuous low level of Triticum-type pollen (Connor et al., 2013). Comparable findings, with the addition of charcoal peaks, were made at Lake Prespa despite the absence as yet of any Phase 3 lakeside settlement (Panagiotopoulos et al. 2013).
The Slovenian diagrams of Mlaka and Griblje (Andrić, 2007) show the creation of a mosaic of fields, pastures and meadows under the influence of local farmers. An intriguing occurrence at both Griblje and Lake Prespa is the short-lived occurrence of cereal-type pollen in the late 7 th millennium BCwell before the local beginning of farming. A final pattern concerns the intensification of farming, in conjunction with charcoal peaks, cca. 4000 BC and in the succeeding centuriesfound in both Mljet and the Slovenian cores. Both of these areas lie on the periphery of the core distribution of major Chalcolithic settlement, with a delayedaction farming expansion perhaps related to cultural changes in the core zone. The general pattern of these vegetation storiesthe small-scale interventions of Neolithic and Copper Age farmerscan offer only limited support to the Willis -Bennett model because there is no clear definition of the source of the indicator pollens. It could equally well be argued that an apparently small-scale intervention created by a settlement 10km distant from the pollen coring site would represent a much bigger 'impact' than if the same indicator species had been found on a pollen site next to a Neolithic settlement.
In the absence of precise archaeological mapping in the micro-region of the third group of five pollen coring locations, it becomes even harder to evaluate the data in the light of the Willis -Bennett notion. The most remarkable case is the suite of farming indicators in the Pirin Mountains diagram, in a lake at an altitude of 2190 masl! There is evidence of a continuous curve of secondary indicators from 5500 -3800 BC, with cereal-type pollen in the earlier part of this period (Marinova et al., 2012). While this period coincides with an intensification of Neolithic settlement in the Struma valley to the West, it is still hard to identify the mechanisms by which these pollen species were incorporated into such a highaltitude lake. Dörfler (2013: 352) proposes that the Plantago lanceolata curve at Prokoško Jezeroover 1,000m higher than the Neolithic sites in the Visoko Basin -could result from either lowland pollen blown upslope or moderate human disturbance round the lake, perhaps indicating small-scale cattle herding. All of the pollen coring sites related to prehistoric settlements at a regional scale could be used to support an argument for a much bigger 'impact' than if the same indicator species had been found next to a Neolithic settlement.
In summary, if the Willis -Bennett hypothesis for minimal 'human impact' on the Balkan environment during the Neolithic and Copper Age has not yet been falsified on a broad scale, it has certainly been challenged at a series of individual sites. The most general result is that, in the period 6200 -3000 BC, approximately a half of the proxy records showed signs of human impact at varying levels of intensity. The principle can be proposed of a general relationship between the scale of the 'impact' and the distance of a pollen coring site from the prehistoric settlement(s) whose inhabitants were supposed to have caused that 'impact'. In the final part of this section, I shall examine in more detail a set of four pollen cores whose relationship with a nearby site has been well established.

Proxy records next to prehistoric settlements
The Dikili Tash cores were collected close to the 17m-high Neolithic tell, with the Dik4 core 1.75km from the tell and the Dik12 core only 150m from the tell (Glais et al. 2015). While the Dik12 core shows perennial pasture plants from the mid-7 th millennium BC (the local Early Neolithic) and anthropozoogenous taxa from the mid-6 th millennium BC (the local Middle Neolithic) (2015, Fig. 4), there was a sudden appearance of Cerealia pollen in the mid-7 th millennium BC in the Dik4 core, with Cerealia values rising to 10% -a clear sign of human impact (2015, Fig. 6). The human impact from the mid-6 th millennium BC onwards was masked by the dramatic increase in alder-carr, showing a local closing of the environment. However, once alder values had declined to c. 30%, there were renewed signs of human impact, especially pastoral practices, in the late 6 th millennium BC. These data are significantly early in the settlement of North Greece, showing local human impact on what was then 'a pristine forested environment ' (2015, 246), representing a far earlier series of human impacts than was detected in the regional pollen diagram of Tenaghi Philippon.
At Sárlo-hát, North East Hungary, a series of changes in the type and scale of land-use can be detected in pollen cores taken less than 100m from a Middle Neolithic and a Copper Age site (Magyari et al., 2012, esp. 294 -6). Woodland clearance and burning was practiced by the Middle Neolithic Alföld Linearbandkeramik groups to produce wetland pasture, presumably for cattle as much as caprines, rather more than for arable farming. Low-intensity farming is proposed for this group, in which settlement dispersion reaches its Neolithic maximum. In the following Late Neolithic, there is an increase in open grassland indicators of up to 40%, suggesting large open areas for cattle husbandry. However, mixed farming is indicated by the shared cereal species and crop weeds found in both the Sárlo-hát core and the plant macrofossil assemblage at the Late Neolithic tell site of Polgár-Csőszhalom, 11km to the South (Fairbairn, 1992: 1993: Raczky et al., 2011. Forest clearances through burning continued on into the Early Copper Age, but with fewer pastoral and arable indicators, suggesting a dispersion of both cattle-keeping and cereal cultivation in smaller homesteads rather than a decrease in the number of cattle kept in comparison with the Late Neolithic. This vegetation record shows that Neolithic and Copper Age environments could be strongly modified in this region, in which Corylus-dominated broad-leaved forests have been extensively cleared round the lake predominantly for animal-keeping rather than arable farming. The scale of clearances are also consistent with the fluctuations in the settlement record in Eastern Hungary, from Middle Neolithic dispersion to Late Neolithic tell-based nucleation to Copper Age dispersion. At Durankulak, a series of pollen cores from the lake is located at 100m from the Late Neolithic settlement on the shore and from the Chalcolithic tell on the 'Big Island' (Todorova, 2002;Bozhilova & Tonkov, 1985;Marinova, 2003;Marinova & Atanassova, 2006;Tonkov et al., 2014). The extent of clearance required for any farming practices is uncertain, since the dominant vegetation around the Durankulak Lake continues to be an open Chenopod -Artemisia steppe with stands of trees in moister areas until 4000 BC, whereupon a forest-steppe developed lasting well into the 2 nd millennium BC. The low level of cerealtype pollen in the late Neolithic and even earlier, at the turn of the 7 th millennium BC, (Tonkov et al., 2014: 280) suggests small-scale farming in the area, mostly in the open areas with better-quality soils. It is only in the Chalcolithic that we see increased signs of Triticum and Hordeum pollen, in combination with secondary indicators and lower percentages of arboreal pollen. These data point to not only larger-scale arable and pastoral practices but also clearance of some of the remaining forests. In the 4 th millennium BC (transition from the Final Chalcolithic to the Early Bronze Age), the absence of primary and secondary indicators of farming suggests the abandonment of fields near Durankulak Lake, until an increase in pastoral indicators is noted after 2700 BC. As in North East Hungary, it is in the 5 th millennium BC -well before the Bronze Age! -that local communities increased the openness of their local environment for the intensification of arable cultivation and pastoralism.
One of the surprises of the Nebelivka core, located 250m from the 236-ha. Trypillia megasite, was not that there was early 4 th millennium BC human impact but that that there was not an overwhelming human impact. The human impact showed up as a series of deforestation episodes, a suite of nine fire events, including a massive fire event dated to c. 4190 BC, and a continuous, if not particularly high, Cerealia curve (Albert et al., submitted). However, these impact events can be dated to before and after the 150-year duration of the Nebelivka megasite (3950 -3800 BC: Millard et al., in prep) as well as during the occupation. Moreover, there was remarkably little soil erosion in the whole sequence, as confirmed by a very slow sedimentation rate. All of these findings create a problem for the interpretation of Trypillia mega-sites as long-term, permanently occupied settlements with massive populations; the Nebelivka group is modelling various alternative scenarios which involve either seasonal aggregations or far lower populations (Gaydarska, 2016;Gaydarska & Chapman, in press).
In the Slovenian karst areas, where farming began much later than in the East Balkans, the Griblje and Mlaka diagrams (Andrić, 2007)  The results of the comparisons of proxy vegetational records with dwelling sequences at settlements close to the pollen cores show clear evidence for 'human impact' in the Neolithic and Chalcolithic periods, dated to one or two millennia before the Bronze Age when impact began according to the Willis -Bennett model. The key factor which Willis and Bennett overlooked was the scale of agricultural practice (Bogaard 2004), which remained small-scale in the Early Neolithic but increased in scale in the Later Neolithic and Chalcolithic.

Conclusions
In this paper, two questions have been investigated: the effects of claimed global changes in Holocene climate at the regional and local scale and the extent of human impact in these proxy records prior to the Bronze Age of the ABC zones. I selected several claims for environmentally generated settlement changes, including what has become the most notorious of the climatic forcing mechanismsthe 8200BP 'event'. The worst correlation of supposed environmental changes with settlement changes were Weninger et al.'s (2009 and2014) studies, when, in the former, the supposed climatic forcer coincided with the biggest expansion of Moldavian -Ukrainian Neolithic settlementthe Cucuteni A -Trypillia B phasesthroughout the 4 th millennium BC, while, in the latter, there were several major peaks in the cumulative 14C record coinciding with the '8200BP event' In the numerous studies of the alleged impact of the 8200BP 'event', it was difficult to pin down the chronology of the 'event' to within 500 yearsmeaning that it has not been possible to examine the question of supposed correlations with changes in settlement patterns or subsistence strategies at local or regional level. At a more detailed level, the investigation of the impact of the 8200BP 'event' on proxy records such as well-dated pollen sequences revealed that changes in regional or local vegetation could be related, even with relative chronological imprecision, to the 8200BP 'event' in only a minority of cases. The rarity of climatic impact on 25 well-dated pollen sequences shows that either there was a minimal impact from the 8200BP 'event' (more probable) or the impact was short-lived and rapidly reversible (less probable). Given the failure of the environmentally determinist model, the message to Aegean -Balkan -Carpathian prehistorians is that alternative causes of settlement and/or cultural change in the late 7 th millennium BC should be urgently sought.
We could make a start by noting that local communities made thoughtful choices about where to live, how to cope with slow or minor changes in vegetation cover or lake / river levels, as well as what to eat and drink. The current dating of proxy records provides at best a centennial record, equating to three or four human generations. The choices made by communities, households and persons meant varying pressures on local landscapes, leading to different perceptions as well as varied actual environments (Head, 2008)

. The increase in
Corylus meant greater potential for nut-crops, while Cornus mas increases led to the wider consumption of Cornelian cherry. Both developments may have led to renewed communal focus on these harvested 'crops', with the potential for woodland management. There was also a significant development of medicinal plants from the Late Glacial onwards .
Even more important was the way households valued timber for firewood and building. The Neolithic has been described as 'the age of building' (Borić, 2008: Chapman, 2014, with dozens of houses built on the nucleated settlements of the Aegean, the Balkans and the Carpathians. Advanced time planning was necessary for the coppicing of hazel for wattleand-daub construction (a 5-year cycle), as well as planting new oak trees for the next generation of houses (a 10-year cycle). One of the key findings of an experimental programme of 'Neolithic' house-building in the Ukraine was the realisation that the production of an individual house can be viewed as a symbolic fusion of the different elements that made up the Trypillia landscape, including clay from the earth, straw from the steppe or cultivated fields, wood from the forest and reeds and water from rivers and lakes (Johnston et al., in press). There was a multitude of ways in which environments and human communities co-created themselves. farming or started to burn the local forest for their clearances (Bogaard 2004). It seems likely that, for the Early Neolithic of the South Balkans and for the Hungarian Plain, only slight modifications were made to the forests, which tended to increased patchiness and species diversity. This led especially to increased forest-edge zones and secondary forests, which were ideal for farming practices such as coppicing and pollarding and the forest pasture of small ruminants (Marinova et al., 2012). The combination of population nucleation and an increased scale of farming, both arable and pastoral, from 5300 BC onwards led to the need for forest clearance, often through burning, with long-term effects on the diversity and structure of the broad-leaved forests. An increased tendency to altitudinal differentiation of vegetation zones led to a greater variety of forest resources, which may have been an important factor in strategies of vertical movement, such as transhumance. There was a widespread re-afforestation in many areas of the study region in the 4 th millennium BC, which combined with a greater tendency for settlement dispersion, the expansion of woody taxa such as Carpinus and Fagus and a lower incidence of natural fires to produce less disturbed, more closed forest landscapes. In a new cycle of farming expansion, Early Bronze Age communities reached a different balance with their trees and grasses, often leading to the signs of an apparently greater removal of trees than had happened before.