Environment and society in the late prehistory of southern Georgia, Caucasus Connor, S.E. and Sagona, A. This is a draft version of a manuscript published in Les Cultures du Caucase (VIe-IIIe millénaires avant notre ère): leurs relations avec le Proche-Orient (edited by B. Lyonnet, 2007, Editions CNRS, Paris, pp. 21-36). Please note that there may be differences between this version and the final published version. The authors will be happy to provide copies on request. ENVIRONMENT AND SOCIETY IN THE LATE PREHISTORY OF SOUTHERN GEORGIA, CAUCASUS Simon CONNOR and Antonio SAGONA INTRODUCTION Recent investigations into the palaeo-vegetation of southern Georgia, specifically on the Tsalka Plateau, have yielded significant results that help us to understand the environmental context of the late prehistoric societies in southern Caucasus. Despite discussions that have inferred ancient socio- economic developments in the South Caucasus, trustworthy data have been limited. Attempts to provide a holistic picture, using multidisciplinary approaches that integrate the results from various fields, have been even fewer. In this paper we draw together data on the palaeo-vegetation and archaeology of southern Georgia from the Mesolithic until the Middle Bronze Age. The present-day landscapes of southern and eastern Georgia are dominated by steppe vegetation, with only a narrow and fragmented forest-belt that divides lowland and highland steppes (Figs. 1 and 2). Evidence from various quarters has suggested, however, that what are nowadays steppes may have been more wooded in the prehistoric past. This evidence includes forest faunal assemblages from archaeological sites in presently treeless landscapes (1), the presence of relict forest soils in steppe areas (2), remnant dwarf stands of trees on the Javakheti Plateau (3), as well as Bronze Age archaeological finds depicting deer-hunting scenes (4) and historical legends, such as the founding of Tbilisi during a pheasant- or deer-hunt (5). All of these point to the intriguing possibility that past landscapes were remarkably different to those we see today. The aim of this paper is to shed much-needed light on the environmental contexts of prehistoric settlements in the South Caucasus. Previous studies in the region have been hampered by a lack of an independent radiocarbon chronology, limiting the possibility of interpreting vegetation change in terms of human impact, and cultural shifts in terms of environmental change. Here we present two new pollen diagrams from Georgia and discuss the results in terms of their regional and archaeological significance. STUDY AREA The Tsalka Plateau lies at the northernmost extension of the vast volcanic upland plateaux of Georgia, Armenia and Eastern Anatolia comprising the Anticaucasus (Lesser Caucasus) Uplands (Fig. 1). The landscape is a striking blend of flat, grassy steppes punctuated by the abrupt peaks of volcanic craters, calderas and scree cones. The Tsalka Plateau is bordered to the north by the Trialeti Range, and forms the headwaters of the Khrami River, a major tributary of the Kura (Mtkvari) River. Geologically, the plateau consists of Upper Pliocene and Quaternary lava flows (mainly dolerite and andesite-basalt), overlain in places by more recent lacustrine and alluvial deposits (6). The average annual temperature in the township of Tsalka is 5.8oC, with cold winters (average -5.3oC in January, absolute minimum -32oC) and mild summers (average 15.8oC). Rainfall averages between 600 and 700 mm per annum. The vegetation of the Tsalka Plateau in the present day is classified as mountain steppe, and consists almost entirely of grasslands, devoid of any forest vegetation. In May 2002, a sediment core was extracted from Lake Imera, which is situated 50 km west of Tbilisi and 10 km northeast of the town of Tsalka (Fig. 2). The site is surrounded by steppic grasslands dominated by Agrostis planifolia, Phleum phleoides, Phleum pratense amongst others, and has a diverse wetland flora. It is a small volcanic maar lake, and, because groundwater levels control its hydrology, it provides a very sensitive gauge of climatic and vegetation changes through time. In September 2003, this record was complemented by another core from Lake Aligol, also a small maar lake, 7 km northwest of Tsalka, located between the village of Santa, a Bronze Age burial mound (‘Kurgan III’) and the Tsalka Reservoir (Fig. 2). Pollen analysis provides the most readily available and interpretable means of determining past vegetation changes in chronological sequence. It is a useful companion to archaeological work because it provides as much information about the vegetation and climate of periods with abundant archaeological evidence as periods without. Pollen was prepared and counted using standard techniques (7). Six accelerator mass spectrometer (AMS) radiocarbon dates were obtained for the Imera record, and ten for the longer Aligol record (Table 1). Summary pollen diagrams are presented in Figures 3 and 4. The curve for each pollen type represents its percentage of the pollen sum. In both pollen diagrams, the percentages have been calculated independently for trees and shrubs (from the arboreal pollen sum), and grasses and herbs (from the non- arboreal pollen sum). The reason for this is that the volcanic lakes of southern Georgia have a high representation of grass pollen, even where forest is the dominant vegetation type (8). In part, this is due to the shallow, drying nature of the lakes; such that as the water level drops, the ‘pollen source area’ (9) is reduced. This is particularly true during early spring, when the Tsalka lakes are often at their lowest level and when many trees are releasing pollen. This explains why low lake-levels in the past are represented in the pollen diagrams by diminished percentages of tree pollen overall. Grasses are also common in the wetland flora of the lakes themselves. POLLEN RECORDS The Lake Imera Diagram The diagram (Fig. 3) spans the last 7000-8000 years and is subdivided into five stratigraphic zones by statistical splitting (10). The lowermost zone indicates that a mixture of subalpine meadows and scattered mixed mesophytic woodlands prevailed around Imeras Tba until 5500 BC. The most important tree species in this zone are hornbeam (Carpinus caucasica), hazel (Corylus), beech (Fagus), ash (Fraxinus) and possibly Zelkova or elm (Ulmus). High proportions of Campanula and other meadow herbs suggest that the landscape was only partially wooded. At the beginning of zone 2, the importance of fir (Abies) and beech (Fagus) increases. The charcoal record indicates that fires were increasingly common between 4000 and 3000 BC, inviting the possibility that humans were using fire slow post-glacial forest advance. Although no definitive answers to this question can be drawn from pollen data, there are some hints of human land- use at this time – alongside the fires was an increase in the pollen of agricultural plants and weeds: Triticum (wheat), Centaurea cyanus (cornflower – asssociated with grain crops), Silene (campion), Linum (flax), Vicia (vetch), Scleranthus (knawel), Trifolium (clover), etc. (11). At 2900 BC, this fire regime abruptly ceased. From 2900 to 1500 BC, the pollen indicates that an oak-dominated savanna with a grassy understorey grew around Lake Imera. This plant community also contained a few hornbeam (Carpinus caucasica), hazel (Corylus) and linden (Tilia), and was probably similar in certain respects to the dry Highland Oak (Quercus macranthera) woodlands that currently grow around the margins of the Tsalka Plateau (12). Between 2000 and 1000 BC, a cooling climate depressed timberlines and led to the replacement of the oak savanna by a coniferous association of pine (Pinus) and fir (Abies nordmanniana). Coniferous forests with fir occur in the study area today only along the northern boundary of the Tsalka Plateau, on the north slope of the Trialeti Range (Fig. 2). The remainder of the pollen diagram falls outside the period of interest of this paper, but it is worth noting that the areas of coniferous forest were burnt down in the 1st millennium BC. After repeated fires, the Tsalka Plateau took on the treeless character it currently displays. There is no doubt that these fires were at least partly anthropogenic (13). The Lake Aligol Diagram The pollen diagram from Lake Aligol (Fig. 4) provides a considerably longer record of vegetation change, covering the past 14,350 years. In zone 1, the prevalence of Ephedra, juniper (Juniperus), wormwood (Artemisia) and goosefoot (Chenopodiaceae) indicates a hyper-arid desert or semidesert- steppe landscape in the late-glacial period, the climate reaching its utmost severity around 10,000 BC – the ‘Younger Dryas’ stadial. After this time, rising precipitation allowed the rapid expansion of pioneer tree species such as hazel (Corylus) and birch (Betula) within an otherwise arid, open steppe. The pollen spectra of Aligol zone 3 are almost identical to those of Imera zone 1, consisting of a mixture of woodland and subalpine meadow species. The most important woodland species are elm, beech, hornbeam and fir, and mesic meadow species include Thalictrum, Dipsacus, Campanula and Polygonum bistorta. Like the Imera diagram, this zone has high charcoal concentrations, suggestive of a high fire frequency between 4000 and 3000 BC. Around 3000 BC, fires ceased and the pollen of oak increases dramatically. Meadow herbs simultaneously decline and grasses multiply, indicating that the surroundings