Chapter Four
Palaeo-environmental reconstructions

Introduction

For a very long time, environmental factors were either under- or over-estimated in Bulgarian prehistoric investigations. If natural aspects were included at all in archeological research, they were most often dominated by a single factor, such as subsistence (soils, vegetation, water springs), possible resources (flint, copper) or landscape (outcrops, self-defence, etc). Very rarely is the physical background considered as complex, thus permitting effective archaeological interpretation (for an exception, see Todorova 1984).

Joint archaeo-environmental investigations are extremely rare (Dennell and Wembley 1975, Bozilova and Ivanov 1985, Bozilova and Atanassova 1989, Lichardus et al. 2001). More often, plant and animal bones remains from archaeological sites were investigated with the interpretative emphasis on subsistence strategies rather then on ecological conditions as a causative factor at all (Dennell 1973, Todorova et al. 1975, Todorova et al. 1983, Raduncheva 1976, Popova 1995, Marinova 1999).

However, the importance of environmental factors was underlined to support two hypotheses – for the secondary Neolithic revolution in the Balkans and for the tragic end of the glamorous Chalcolithic civilizations (Todorova 1986; Todorova and Vajsov 1993: Todorova 2003). In this rather uncertain understanding of the role of physical background, the environment and changes in environmental conditions have yet to find their relevant place in Bulgarian prehistoric investigations. In the current study, it was presumed that rocks, soils, vegetation, etc. have always mattered for humans. However, the use of any of these as resources means that, to find out what their real importance might have been, one has to define their availability. Some resources such as rocks or minerals are the same from the time of the first occupants of the study area but their technological ability to exploit them may well have changed. Others, as climate and the position of river beds, may well have changed over time. The definition of continuities and differences between the present and past landscapes of the study region is the purpose of the following chapter.

Apart from morphological differences (two different catchment basins), the three small river valleys are geographically very similar. They are 1 - 3 km wide and surrounded by low, usually gently sloping hills, which rarely exceed 200m in altitude. Their climate and vegetation are also quite similar, while the geological sequences and contemporary anthropogenic impacts are significantly different. On account of these differences, as well as those pertaining to the availability of environmental data, the physical background of the study area will be presented in two different data sets. The first one seeks to unify the Sokolitsa and Ovcharitsa valleys, since they have a common environmental development. The second geographical data set concerns the Kalnitsa valley microregion, in the territory of the village of Drama, with its relatively low concentration of archaeological sites.

4.1. Geological data

4.1.1.Maritsa Iztok - geological basement and geolithological structure

The geological structure of the Maritsa Iztok basin consists of multiple series of rocks of different ages. Their spread and depth of occurrence as given in Figs. 4.1.1-4.1.2 show a deeply indented palaeo-relief, formed after intensive tectonic movements. On the present surface, these very old rocks are visible on Svetiiliiski visochini (the St. Ilya Uplands), Manastirski vuzvishenia (the Monastery Hills) and the Sakar Mountain and as single spots amongst Neogene deposits. So far, there is no common agreement on the beginning of Tertiary sedimentation in the region. The earliest suggestion is for initial Upper Eocene (Priabonian) infilling (Fig. 4.1.1). The Tertiary sequence as presented in Fig .4.2.1 is accepted in general. Two important differences, however, derive from the two main data sources for the geological development of Maritsa Iztok coal basin. According to the first view (Nedialkov 1985, Manual 1981), coals are Miocene formations, with up to 250 m of strata, while a 15-m thick layer of fine dispersed clays represents the Pliocene. According to the other view (Nam 1995, Kirilova 1985), the Pliocene series is up to 300 m thick and contains the coal layers. Despite these disagreements over the dating of the Tertiary sequence, all the investigators accept the presence of Neogene lacustrine sediments in the study area.

Fig. 4.1.1 Geological map of Maritsa Iztok. Source: Nedialkov 1985

Key: 1-5 Upper Cretaceous Series 1 – Granites; 2 – Diorites; 3 - Dioritic porphyrites; 4 - Contact metamorphic rocks; 5 – Greisen-like alternations in Gradets intrusive; 6 - Brecchia-conglomerates (undifferentiated Cretaceous series); 7 – 8 Upper Cretaceous volcanic-sedimentary rocks; 7 – Andesites; 8 - Andesitic tuffites; 9 - Turounian age -conglomerate reef formation; 10 –11 Triassic System; 10 - Dolomite Formation; 11 – Arkose sandstones and rocks from silicate carbonate formation; 12 – 17 Permian System; 12 - Felsite quartz porphyries; 13 - Spherulitic quartz porphyries; 14 - Quartz porphyry tuffs; 15 - Calc-schists; 16 - Quartz-muscovite schists; 17 – Brecchia-conglomerates (dappled conglomerate unit); 18 – 20 Carboniferous (?) System; 18 – Low-grade schists; 19 - Metaquartz-porphyries; 20 - Metatuff-brecchia; 21 – 22 Undivided Palaeozoic; 21 - South Bulgarian granites; 22 - Low-grade schists (diabasic-phyllitoid complex); 23 - Pre-Cambrian granite-gneisses; 24 - Tectonic zone along Sazliika fault; 25 - Faults 1. Sazliiski 2. Sokolshki 3. Jujen Svetiiliiski 4. Severen Svetiiliiski 5. Gradetski 6. Radevski 7. Pundakliiski 8. Grafitovski thrust; 26 - Boundary of Tertiary distribution; 27 - Boundary of Troyanovo coal level; 28 – thrust.

Fig. 4.1.2 Geological sequence of Maritsa Iztok Source: Nediakov 1985

Key: 1.Soil layer, 2.Fine dispersal calcareous clays, 3.Limestones, 4.Blue/green sandy clays, 5.Sandstones, 6.Grey/black clays, 7.Black clay with coal intrusions, 8.Coal layers, 9.Thin stratum clays, 10.Argillites, 11.Reddish clays, 12.Deluvial-proluvial sandy clays and gravel 13.Upper Cretaceous intrusions, 14.Dolomites and marbles, 15.South Bulgarian granites, 16.Pre-Cambrian metamorphic rocks

4.1.2 Slumps and Volcanoes

An important natural feature of Maritsa Iztok, also utilised in cultural practices, is the phenomenon of mud-volcanoes. These are mound-like hills which can reach up to 8m high but they can be very small too (Figs. 4.1.3 - 4.1.4). They are distributed along the valleys of the Sokolitsa, Ovcharitsa and Eledzjik (a valley West of the study region).

Fig. 4.1.3 Mud-volcano between Mednikarovo and Obrutchishte; Source: Koen 1952
Fig. 4.1.4 Mud-volcano near village of Mednikarovo (known as Atanasivanova mogila); Source: B. Borisov –field documentation

Mud-volcanoes are not met elsewhere in Bulgaria and, in the case of Maritsa Iztok, are connected with coals and the geological substructure. There are several reasons for the appearance of these curious features but the first and most important one are the so-called ancient slumps on palaeo-relief slopes. The latter are the result of active, mainly positive neotectonics, most probably followed by seismological activity, as well as severe fluviatile erosion and intensive rainfall that result in rivers with a high water-level. Their dynamic is very similar to that of the present slumps. During the active period, if the slump’s prism of active pressure reaches the coal layer, it causes swelling of the coals (Fig. 4.1.5).

Fig. 4.1.5 Slump mechanism; Source: Nedialkov 1985 (Key as Fig. 4.1.2)

If this coal swell reaches the surface, it looks like a small, elongated mound. There are several such mounds that can still be seen along the valleys of the Sokolitsa and Gradetska. Ancient landslips are distributed along the valleys of the Sokolitsa, Ovcharitsa and Eledjik because they are related to fluviatile erosion. Most of the surface coal swellings are covered by terrace sediments now making the slumps stable. A few swellings, however, are not yet covered, which leaves the areas vulnerable to further landslips (Nedialkov 1985).

When active, slumps may or may not produce mud volcanoes as a part of the process. The eruption of mud volcanoes is related to the faults and cracks in the swelling zone, when deep-lying inrush waters following the leaks in the sediments reach the surface, sweeping away coals from the swelling and spreading to form mound-like features (pers. comm., P. Karacholov) (Fig. 4.1.6).

Fig. 4.1.6 Mud –volcano forming process; Source: Georgiev 1976

Key: 1-Soil, 2-Terrain deformation, 3-Clays and sands above the coal layers, 4-Deformation zone, 5-Coal complex –clay and coals, 6-clays under the coal layers, 7-Sands under the coal layers that contain inrush waters, 8-Sub-surface weak link and the leak, 9-Mud volcano

Volcanoes could abate or awake, they even could change their place but are always connected with ancient landslips and inrush waters. Usually they could be activated through the renewal of movements of the ancient landslips. The latter might appear as a consequence of coal exploitation. But they also could be activated as a result of fluviatile erosion. Natural eruption is not rapid and devastating but rather long-lasting. Such a burst may need a week to form a real volcanic shape until the fading of the slump movements. After the start of coal basin exploitation, some of the ancient slumps were activated and some new ones appeared of technological origin. Some of them led to the appearance of new, non-natural mud volcanoes (pers. comm., P. Karacholov).

4.1.3 Drama – geological basement and geolithological structure

Two main sources are available for the geology of the Drama microregion. The first one is the investigation results of the long-lasting German expedition. According to their team, the oldest rocks in the Kalnitsa valley are Pre-Cambrian granite-gneiss, biotite-gneiss, two-layered gneiss and amphibolite. A small phyllitoid formation of diabase completes the chart of Pre-Mesozoic sediments of the region. Triassic rocks are represented by quartz, sericite, schists and conglomerate, as well as by marblized limestone and dolomites. Jurassic limestone is in the form of schists with marl clusters. Intrusive rocks of Palaeogene Age in the area comprise gabbro, gabbro-diorite, diorite, quartz-diorite and diorite-porphyry. The detailed sequence and spread of the rocks in the Kalnitsa valley are given in Fig. 4.1.7. Diorite-porphyry surface exposure could be found North, East, South and Southwest of Drama (Fol et al. n. d., Lichardus et al. 2001). Also visible today is some Permian granite in the Northeast edge of the contemporary village. Diorite and gabbro-diorite intrusions in the Mesozoic limestone are believed to lead to the formation of marble, as well as to uplifting of Kalnitsa valley by 100-300m (Kubiniok 1996).

Fig. 4.1.7 Geology of the Drama basin; Source: Fol et al. (n. d.)

The second source for the Drama microregion is Bulgarian geo-survey data, according to which the spread of the rocks and their sequence is slightly different (Fig. 4.1.8). The oldest rocks are formed by the Lower Palaeozoic Sokol formation, containing sericite-chlorite phyllites, argillite-like schists and schistic basic tuffs, situated near the contemporary village. The village itself is on Upper Palaeozoic middle-grained biotite granite (Sakar biotite –2). Triassic rocks are spread Northeast of the study area as an undivided Iskur carbonate group, consisting of dolomites and marbleised dolomite limestone. Among them, spots of Upper Cretaceous diorite-porphyrite are distributed. The other Upper Cretaceous rocks are amphibole-biotite gabbro (manastirski pluton) and quartz-diorite. Neogene sediments are the most commonly represented in the study area and contain sandy clays, sands and coals of the Elhovo formation. Along the Kalnitsa river, Holocene alluvial formations are found - both on river beds and flood terraces as gravels, sands and clays.

Fig. 4.1.8 Geology of the Drama basin Source: Kozuharov et al. 1994

Key: 1. Holocene alluvium (gravel, sand,clay) 2. Miocene/Pliocene Elhovo formation (sandy clays, sands, coals) 3. Upper Cretaceous diotite porphyrite 4. Upper Cretaceous quarz-diorite 5. Upper Cretaceous gabbro 6. Middle/Upper Triasic undevided Iskur carbonate group (dolomites, marbleised dolomite limestone) 7. Upper Paleozoic biotite-granite 8. Lower Paleozoic Sokol formation (sericite-chlor phyllites, argillite-like schists, schistic basic tuffs)

4.2. Geomorphologic data and soils

4.2.1 Geomorphology and pedogenesis in Maritsa Iztok

Quaternary investigations in Maritsa Iztok have always been a part of the common geo-environmental study of the region. There are not special geomorphologic investigations and Pleistocene and Holocene deposits have not yet been differentiated. Generalized Quaternary sediments are represented by alluvium or diluvium clays and alluvium fan deposits (sands and gravels) (Kirilova 1985). Vertical neotectonic activity led to intensive denudation and caused cyclic river erosion. These cycles were synchronous with vertical movements and are traceable in successive down-cutting of the large rivers that formed several erosion-accumulation terraces. The total down-cutting of the river Ovcharitsa amounts to 65m, while 78m is recorded for the river Sokolitsa (Nedialkov 1985).

Quaternary investigations on the broader scale of the overall development of the Maritsa river terraces give some general information for the present development of the rivers in the study area of Maritsa Iztok (Fig. 4.2.1). For the Maritsa river, there are altogether 7 overbay and 3 bay terraces. The latter are the result of positive tectonic movements during the Holocene. For the Sokolitsa river, however, earlier terraces are also common. There are 4 overbay and 2 bay terraces. Around the village of Obrutchishte, the 4th and 5th terraces are of erosion accumulative origin. The alluvium there is 2m thick and contains sands and gravel. Bay terraces are found along the whole length of the river. Towards the lower course, the thickness of alluvium varies between 4m and 7m and consists entirely of sands (Angelova et al. 1993).

Fig. 4.2.1 Geomorphological map of Maritsa Iztok Source: Angelova et al. 1993

Key: 1. Holocene river terraces 2. Upper Pleistocene river terraces 3. Middle Pleistocene river terraces 4. Lower Pleistocene river terraces 6. Eo-pleistocene (Villafrancian) levels and terraces.

Two types of the oldest Bulgarian soils are distributed in the Maritsa Iztok area - smolnitsa and cinnomonic forest soil. Toward the end of the Pleistocene, unconsolidated lacustrine sediments formed low and relatively even relief, that, along with poorly drained geological substratum, caused meadow or meadow-boggy forming process under the conditions of a relatively warm and wet climate. This first stage of smolnitsa development was followed by surface drainage caused by the drying influence of gallery forest and forest steppe vegetation. These conditions are very similar to the present, when smolnitsa is one of the soils with the thickest humus accumulation horizon (Kirilova 1985).

Cinnomonic forest soils are developed on Pliocene deposits, as well as on calcareous or acid substrate. They are formed in more variable hydrothermal conditions, under the influence of sparse and dry deciduous forests and bushes succeeded by treeless areas (Kirilova 1985).

4.2.2 Soil distribution in Maritsa Iztok

The most widespread soils in Maritsa Iztok are leached smolnitsa. They are dark black, with 60-80 cm thick humus accumulation horizon that contains 2.5 – 3.5% humus. The high percentage of clay (up to 50%) in this soil determines its adverse chemical and physical properties. When wet, it is sticky and difficult to form a tilt and, when too dry, cracks up to 10cm wide and 1m deep usually occur. However, it is possible to till smolnitsa in the period following sufficient rain to soften the otherwise hard soil (pers. comm., P. Reynolds, per J. Chapman). A specific feature in Maritsa Iztok is the so-called calcareous cavities in the soil, as well as gypsum-like inclusions of different sizes that appear in the areas of Radnevo and Gledachevo at 100-150 cm in depth (Kirilova 1985).

Meadow smolnitsas have a limited distribution in micro-depressions with relatively high subsoil water level. Often these soils are affected by semi-hydromorphic salinity (Kirilova 1985) which makes them good for animal pasture.

Leached cinnomonic forest soils are the second most widespread soil type in Maritsa Iztok. They develop mainly upon Pleistocene sediments and are moderately loamy, with a soil profile not exceeding 80 cm and low humus content (1.5-2.5%). The humus percentage is slightly higher (2.0-2.5) than in the typical cinnomonic forest soils that developed on calcareous rocks East of the village of Polski Gradets. Upon hill slopes Northeast and Southeast of the same village, strongly leached to slightly podzolized (lessive) cinnomonic forest soils developed on granite or granite-gneiss rocks. These soils have very low humus content – often less than 1% (Kirilova 1985).

Humus calcareous soils (rendzinas) have a limited distribution in the region. Calcareous inclusions in the surface Pliocene sediments formed moderately loamy rendzinas with a relatively thick humus accumulation horizon. Less thick and heavy loamy are the rendzinas developed on calcareous rocks in the Eastern part of Maritsa Iztok. Both soil sub-types have relatively high humus content (Kirilova 1985).

Alluvial meadow and alluvial-diluvial meadow soils are spread along the flood-plains of the Sazliika, Sokolitsa and Ovcharitsa rivers and the valleys of their tributaries, where the level of the sub-soil waters is high. They have a relatively thick humus accumulation horizon and high humus content. There are places where these soils appear in a complex with hydromorphic or semi-hydromorphic soils (Kirilova 1985). Saline soils are found along the river Blatnitsa, one of the tributaries of the Sazliika, and around the town of Radnevo.

4.2.3 Geomorphology and Pedogenesis in the Drama basin

Quaternary sediments of alluvium and diluvial layers of red clay and rubble-intrusive rocks cover almost all the Pliocene deposits of limestone, sand and clay (Lichardus et al. 2001). The thickness of the Tundja Quaternary deposits is up to 40 m but, for the Kalnitsa valley, it is less than 5 m. According to Kubiniok, environmental conditions in the last Glacial did not play any important role in the formation of the relief of the Drama microregion (Kubiniok 1996).

The stratigraphy of the Quaternary deposits in the Tundja lowlands, established through geomorphological surveys in the mid-eighties, shows that, in the study area (the squared area on Fig. 4.2.2), Holocene alluvial deposits with different facies are predominant, together with some eluvial deposits and the pre-Quaternary rocks (Angelova et al. 1991).

Fig. 4.2.2 Quaternary sediments of the Tundja lowlands Source: Angelova et al. 1991

Key: 1. marsh sediments 2. alluvial sediments 3. proluvial sediments 4. alluvial-proluvial sediments 5. diluvial sediments 6.diluvial-proluvial deposits 7. colluvial sediments 8. eluvium 9. infiltrated limestone 10. pre-Quaternary rocks 11. faults.

The identification of pedogenesis in the Drama region was a priority in the palaeo-geographic survey of the area in the German Drama project. The soil formation results, however, are contradictory and in disagreement with Bulgarian data in general. According to the German Expedition’s investigations, black earth (Schwarzerde) started to be formed prior to 4000 BC (towards the end of the Karanovo V period) (Lichardus et al. 2001). Other types of soils developed in the region are rankers of very fine and fine sand developed on acid rock with 2° to 15° dip (profile type C and F). Surface brown soils (flach-mittelgrunige Braunerden) are also believed to have been distributed at some earlier time on the steeper slopes of the Drama area. Weakly developed soils (geringmachtige Boden) are formed on 13° slopes of calcareous porous sediments (kalkhaltigen Lochersedimenten). Humus-poor smolnitsa (vertisolartige Pedosole) develop on carbonate-rich porous sediments (karbonathaltigen Lochersedimenten) over basic rocks (profile type D and G). Along the Kalnitsa river, damp meadow soils (solonetzartige Aueboden) comprising dark-brown – black alluvium has been formed (profile type E). Marblized limestone in the Eastern part of the Kalnitsa valley favoured the development of brown calcareous loam (profile type A and B). All soil types have a high percentage of clay and are all difficult to cultivate by hand. Hence, Neolithic land cultivation was assumed to be meadow horticulture rather than field agriculture (Kubiniok1996).

Fig. 4.2.3 Pedogenesis in Drama microregion; Source: Kubiniok 1996. Profile type A and B Ah - up to 2%; profile type C and F Ah - up to 2-5%, rAp less than 1%; profile type D and G Ap - 2-3%, AhP - 1%; profile type E Ap - 2%, Ah 1%. Source: Kubiniok 1996

The most controversial subject concerns the nature and formation of the black earth. The easiest point to settle is the name of this type of soil. There is a confusion inherited from Bulgarian pedology concerning soil terminology of the chernozem type (Lichardus et al. 2001), since for a long time there was an overlap in uniting chernozem and smolnitsa in one common name – chernozem-smolnitsa (Fig. 4.2.4). Both types share the same mechanical content; they are dark black in colour and very fertile. Recently, however, they were recognized as different soil types, since they have a very different pedogenesis (Soil Atlas 1998, FAO).

Bulgarian chernozem soils are automorphic, e.g., they are formed on acidic loess sediments. They are of eolian origin and are formed after the end of the last Glacial, during the subsequent increase in temperature and decline in precipitation. Their distribution in Bulgaria does not extend South of the Stara Planina mountain range. Some loess sediments (clay and other deposits) are sparsely spread across South Bulgaria but chernozem development is not reported so far in this area (Kenderova pers. comm.).

Smolnitsa is a local, Balkan type of soil. They are distributed in Bulgaria, former Yugoslavia, Albania, Romania and Turkey. They are formed under a transitional continental climate with sub-tropical influence, mainly on Pliocene deposits and old Quaternary terraces. The first stage of their development is hydromorphic – Pliocene mantles with heavy mechanical content in conditions of flat relief. With continuing low water flow, these soils develop into wet meadow and marshy forms. This stage is followed by a period of dryness, after good surface drainage and under the influence of meadow-forest and forest-steppe vegetation. Smolnitsa are developed on different geological substrate from chernozem, as well as on the weathering products of granite and andesite. Recent investigations confirm that smolnitsa were formed under the influence of forest growth, as documented by leaching processes very close to forest conditions. Nevertheless, contemporary distribution of this soil lacks the presence of forests. In Bulgaria, smolnitsa soils are mainly distributed over the lowlands of the Upper Thracian plain, the Tundja district and the Burgas plain and represent cca. 5% of the total soil cover of the country (Georgiev in press).

This relatively detailed description of the genesis and distribution of chernozem and smolnitsa is necessary in order to assess the claim for chernozem formation around 4000 BC (Lichardus et al. 2001). It was mentioned in the publication that the chernozem was of smolnitsa type (Lichardus et al. 2001), so it could be inferred that what was meant in fact was chernozem-smolnitsa. As has become apparent, chernozem-smolnitsa’ is calledsmolnitsa’ in contemporary pedological terminology. Whether or not smolnitsa was formed in Drama around 4000 BC is a difficult question to answer, given the present condition of the data. There is no evidence so far for the specific environmental conditions in this period (4000 BC) that might have favoured the genesis of smolnitsa.

There is one less likely opportunity for the development of chernozem - as a result of meadow-steppe vegetation influence that appeared after forest clearance (Georgiev in press; cf. Kruk 1980 for Southern Poland). The substantiation of such a hypothesis, however, requires specific target-oriented investigations that have not yet been accomplished in the Drama area.

4.2.4 Soil distribution in the Drama basin

One of the important results of pedological investigations in Bulgaria in the last century was soil mapping at different scales. Regional surveys, however, are extremely rare and, for the Drama area, the only available soil map suitable for microregional study is the 1:50,000-scale sheet of the Burgas district, produced in 1961 (Koinov et al. 1961) – reproduced here as Fig. 4.2.4. This early date explains the terminological confusion in naming ’smolnitsa’ as chernozem-smolnitsa. Recent investigation of soils in the Yambol district (regrettably without maps) confirm in general the earlier survey results, adding some new soil types and updating the terminology of soil classification (Baltakova 2001).

According to the soil map (Fig. 4.2.4), leached cinnomonic smolnitsa (chernozem-smolnitsa in the older terminology); typical smolnitsa (formerly chernozem-smolnitsa) and meadow-cinnomonic soils were distributed over the Drama basin, as well as some shallow soils on andesite rocks (Koinov et al. 1961).

Fig. 4.2.4 Soil distribution in the Drama basin, with updated pedological terminology; Source: Koinov et al. 1961

Smolnitsa soils in the Yambol district comprise a 45 to 60cm-thick humus horizon. Their content varies from heavy sandy clay to medium clay with not a very high percentage of humus (230 – 400 tonne per ha in 1m soil layer). They have the same adverse chemical and physical properties as smolnitsa in Maritsa Iztok (Baltakova 2001).

Leached cinnomonic forest soils develop in association with *smolnitsa in the Yambol district. Both soil types share similar evolutionary transitions and their relict traces are still visible in soil profiles in the Elhovo and Thracian lowlands (Baltakova 2001). Due to lack of any more precise information, it might be presumed that the leached cinnomonic smolnitsa distribution shown on the map corresponds with leached cinnomonic forest soils in the recent study (Baltakova 2001). According to the latter, leached cinnomonic forest soils comprise a 20-60cm-thick humus horizon with a humus content in A1 horizon under forest from 2% to 17% and in the Ar horizon from 0.8% to 4.8%. Their hygroscopic capacity depends on their mechanical content but, in general, they suffer from a poor water/air regime (Baltakova 2001).

Meadow-cinnomonic soils share the same characteristics as the cinnomonic forest soils but have a thicker humus horizon, ranging from 40 to 80cm.

The unmapped (Koinov et al. 1961) alluvial-meadow soils are presumed to be spread along the river Kalnitsa. As expected, their distribution was attested on river terraces in the Yambol district (Baltakova 2001). They have a 10 to 70cm-thick humus horizon, below which are river mantles. Their clay content varies between 10 and 60% but lighter soils are prevalent. The humus content varies from 1% to 5%, for the ploughed areas between 1% and 2.5%. Alluvial-meadow soils are crumbly, with a good water/air regime and are not sticky (Baltakova 2001). *

4.2.5 The impact of mining in Maritsa Iztok

The most significant long-term anthropogenic factor in the destruction of the landscape was the gradual de-forestation of the study region. Cleared areas were used for agriculture, leading to the widespread replacement of the natural vegetation by plant cultivation. Along with artificial manuring and irrigation, this caused changes in microclimate and especially in soil texture. Therefore, present soils in Maritsa Iztok differ from their virgin predecessors (Kirilova 1985).

These disturbances, however, were not believed to bring about huge environmental impacts since, in the early 20th century, these areas were mostly small-scale farm lands together with large uncultivated areas with both natural and introduced vegetation (Nam 1995).

There are two major activities that took place during coal exploitation – terrace-like excavation of land and the long-distance transportation of the spoil to enormous spoil heaps. The dual destruction of landscape created both negative shapes – up to 150 m deep - and positive shapes– up to 50 m high. This was accompanied by large-scale infrastructure of special roads, equipment and buildings that have a secondary effect on the landscape.

An additional effect on the hydrology of the region concerns the correction of the rivers’ beds and the numerous artificial lakes and water-tanks that were created for the outflow of the subsurface water. Soils from the exploited areas were stored for future re-cultivation! Pollution is still a problem in the region, despite the long-term experience of addressing the side-effects of mining. Removal of vegetation cores increases soil aridity, not only in the study region but also in the wider area of the Upper Thracian Plain. Last but not least is the almost completely changed native flora and fauna as a result of secondary migration from adjacent areas (Nam 1995).

Natural processes as denudation and erosion cannot follow their original trends in such a devastated landscape and often spill over into neighbouring areas. Radical shifts in the hydrological, gas, thermal and chemical regimes of geological formations could break down the gravity balance and lead to unsuspected changes in the landscape. This is, for instance, the case in the region near the village of Obrutchishte, where a flat zone of several hundred sq. m between two external spoil-heaps has turned into a lowland area (Kirilova 1985).

4.3. Climate and Vegetation according to modern and palaeo-botanical data

4.3.1.Modern data for Maritsa Iztok

Climate

According to the contemporary climatic classification in Bulgaria, Maritsa Iztok falls into the Upper Thracian sub-area of intermediate continental climate, with hot summers and relatively mild winters. Due to the paucity of sharp changes in relief and the low hypsometric fluctuations, its homogeneous climatic conditions are seen in the long-term temperature and rainfall measurements. The average January temperature is 0–1° C, with lower values towards the South East, in the foothills of the Sakar mountain. Roughly the same variability is seen in the mean July temperature that is cca. 24° C. The average monthly and annual temperature regime for the period 1916-1955 in Maritsa Iztok is given in Table 4.3.1 (Kirilova 1985).

Table 4.3.1 Average annual and month temperature for the period 1916-1955; Source: Kirilova 1985 after Climatic Year Book of NRB 1959
I II III IV V VI VII VIII IX X XI XII annual
St. Lubenova
mahala 1921-1955
0.1 2.1 6.4 12 17 22 24 24 20 14 8 2.4 12.6
Mean max. 4.2 7.5 13.1 20 21 29 32 32 28 21 13.6 6.3
Mean min. -4.3 -2.4 1.1 5.4 10 14 17 16 13 7.6 3.1 -1.9
Absolute max 17.6 20.1 30 34 38 40 41 42 39 37 28 19.1
Absolute min -29 -26 -16 -5.3 0.3 4.8 7.5 8.2 1.1 -5.9 -15 -24
Table 4.3.2 Average annual, season and month rainfall for the period 1896-1945; Source: Kirilova 1985 after Rainfall Yearbook of NRB 1962
I II III IV V VI VII VIII IX X XI XII annual
St. Lubenova
mahala 1921-1955
0.1 2.1 6.4 12 17 22 24 24 20 14 8 2.4 12.6
Mean max. 4.2 7.5 13.1 20 21 29 32 32 28 21 13.6 6.3
Mean min. -4.3 -2.4 1.1 5.4 10 14 17 16 13 7.6 3.1 -1.9
Absolute max 17.6 20.1 30 34 38 40 41 42 39 37 28 19.1
Absolute min -29 -26 -16 -5.3 0.3 4.8 7.5 8.2 1.1 -5.9 -15 -24

The data from Table 4.3.2 show that the mean annual temperature is above 12° C, with a relatively high annual amplitude of – 24.1° C. This fact, along with the relatively high diurnal amplitudes – 8.2° C in December and up to 16.2° C in August – is evidence for some continentality in the temperature regime. Temperature variations as shown in rows 4 and 5 are due to the particularities of atmospheric circulation (Kirilova 1985). The number of days with temperatures higher than 10° C varies between 200 and 230 per year (Nam 1995).

Average annual rainfall values for the period from 1896 to 1945 fluctuate between 500-600 mm, which is below the mean annual rainfall for the rest of the country (Table 4.3.2). The summer maximum falls in June, which is typical for a continental climate. However, the secondary minimum rainfall in December that follows the August trough provides strong evidence for Mediterranean influence (Kirilova 1985).

Summer drought is shown in Fig. 4.3.1, when high temperatures accompany lower rainfall in July, August and September. The other two graphs with data from the Stara Zagora and Svilengrad meteorological stations are given for comparison. Mediterranean influence intensifies towards the Southeast part of Bulgaria (Svilengrad), it is not that strong in the middle of the Upper Thracian Plain (Stara Zagora), while the Maritsa Iztok study region lies in an intermediate position (Kirilova 1985).

Fig. 4.3.1 Annual water balance for the Maritsa Iztok and neighbouring regions; Source: Kirilova 1985

Measurement of the duration of snow coverage at the Radnevo station shows an average of 63 days per year, starting on 23rd December and finishing on 2nd March. The mean thickness value of snow coverage is 10-12 cm but in general show great variability. Only 15% of the winters during the period 1931-1970 had stable snow coverage (Kirilova 1985).

As a general trend, the windiness of the region is equable. There are 29 - 31 days per month without strong winds (Nam 1995). Most common is the South wind, that dries the soil in the spring and the strong North wind, that is crucial for the ventilation of Maritsa Iztok power complex area. There are also sporadic Föhn activities.

Surface water

There are several reasons for the relatively low annual flow (0,5-2 l/s/km2) in Maritsa Iztok, that characterizes the region as one with little water supply. The maximum flow is in February, consisting of both rain and snow input. Winter flow exceeds spring flow because of the unstable snow coverage and the loss of spring rainwater through evaporation. Soil water flow is less then 20% of the mean flow value. Increasing cultivation with annual shifting crops and chemical manuring has led to substantial changes in soil texture and agricultural lands have little or no importance for water regulation. Low moistening, intensive evaporation and early exhaustion of the dynamic sub-soil water supplies result in a minimum water flow in August. Summer drought in this region is confirmed by the continuity and frequency of rivers which run dry. Every year, the Sokolitsa runs dry for a period of 75 - 100 days, while the Ovcharitsa runs dry once to three times every 10 years. The pattern of winter flow maximum reveals a strong Mediterranean climatic influence on outflow regimes (Kirilova 1985).

Vegetation

Contemporary geographical indices in Maritsa Iztok as described so far are completed by the distribution of xerophytic and mesoxerophytic vegetation within the study area. Natural vegetation has a limited distribution, represented mostly by relict deciduous Sub-Mediterranean forests of pure and mixed type. They mainly consist of forests of Q. Pubescens and Carpinus orientalis and lower stands of Fraximus ornus and Juniperus, with thorn and sumac in some places. The higher areas support central European species such as Q. frainetto, Q. cerris, Q. sessiliflora Salisb and Ulmus minor (Kirilova 1985).

Carpinus sp. and durmast are found in wetter places, while poplar, willow, elm, ash-tree and Q. robur grow on the floodplains. Bush associations are represented by thorn, sumac, briar and blackthorn. Artificial forests consist of Pinus sp., common locust and Canadian poplar, as well as a few natural species (Nam 1995). The herbaceous assemblage in the region includes meadow species such as Festuca pratensis and Poa silvicola and pasture species such as Poa bulbosa, Festuca pseudovina and Andropogon ischaemum (Kirilova 1985).

The steppe vegetation is not natural but the result of a continuous process of aridisation in the Maritsa Iztok area, which has led to an expansion in xerophytic cover. Secondary associations are widespread all over the study region. Intensive agriculture and inner re-allocation of natural species, as well as the introduction of new ones, have significantly changed the vegetation of the landscape (Nam 1995).

According to a recent study (Bondev 1991), the largest areas of change are the cultivated areas that replaced mixed oak forests comprising Q. cerris, Querceta virgilianae and often Q. pedunculatiflora. Smaller territories were once covered with mixed oak forests of Q. cerris and Q. frainetto or of Q. pubescens and Querceta virgilianae. Smaller areas are covered by the agriculture lands that took over forests of Ulmus campestris L., Fraxinus sp. and Q. pedunculiflora, Andropogone ischaemi, Poaeta bulbosae, Chrysopogoneta grylli and Ephemereta sp., accompanied by thorn bushes and jasmine, currently occupying common pasture areas. Bondev also suggests that, under undisturbed current climatic conditions, 95% of contemporary Bulgarian territory would be covered by forests (Bondev 1991).

Fig. 4.3.2 Contemporary vegetation in Maritsa Iztok; Source: Bondev 1991

Key: 96. Mixed Q. cerris, .Q. pubescentis Wild and Virgilian oak forests; 107. Q.pubescentis Wild and Virgilian oak forests with Mediterranean elements; 121. Shrubs of Christ thorn, mixed with jasmine combined with xerothermal communities replacing xerothermic forest community of Q.pubescentis Wild and Virgilian oak; 129. Xerothermic grass communities with prevalence of Dichanticta ischaemi, Poaeta bulbosae, Poaeta concinnae, Chrysopogeneta grylli and Ephemereta; 130. Mesoxerothermal grass vegetation with prevalence of Poa bulbosa, Lolium perenne and Cynodon dactylon; 133. Farm areas take the place of mixed Q. cerris and Virgilian oak forests, often mixed with Q. pedunculiflora Koch; 148. Mesophytous grass communities (meadows) replacing forests of elm, field ash-tree, Q.roborus and Q. pedunculiflora Koch; 149. Farm areas that replaced forests of field elm, field ash-tree and *Q. pedunculiflora Koch.

Original vegetation is still represented by forests of Q. cerris and Q. pubescens between the rivers Sokolitsa and Ovcharitsa and the forests of Q. pubescens and Querceta virgilianae in the most Eastern parts of the study region, as well as the woods of Q. frainetto and Carpinus orientalis South of the river Sokolitsa. Along the valleys of the Sazliika, Ovcharitsa and in parts of the Sokolitsa, there are still some small areas of native species such as Q. pedunculatiflora, Q. robur, Ulmus minor and Fraximus sp., together with communities of willow, poplar, alder and reeds in the lowest-lying places (Kirilova 1985).

Fauna

The current distribution of fauna is closely related to that of vegetation. Cultivation led to the spread of species that are few in variability but great in number of individuals. The distribution of native animals is considerably reduced and consists of 55% central European species and 25% Mediterranean ones. Among the former are Gricetulus migraterius, Arvicola terrestris (water vole), Phasianus colchicus (pheasant), Passer hispaniolensis (Spanish sparrow), Falco naumanni (lesser kestrel?)and Hippolais olivetorum. Mediterranean influence is found in the distribution of some reptile species, such as Ophisauris apodus, Gymnodactylus cotschyi, Eryx jaculus, Typholos vernicularis, Coluber najadum and Elorphe quatuorlineata. There are also species widespread across present Bulgaria, such as the hare, hedgehog, wild boar, mole, hamster, partridge, pink starling, owlet, thrush, etc (Kirilova 1985). Other species of widespread distribution include Gricetulus migraterius (Grey hamster), Arvicola terrestris (Water vole), Phasianus colchicus (Pheasant), Passer hispaniolensis (Spanish sparrow), Falco naumanni (Lesser kestrel) and Hippolais olivetorum (Olive-tree warbler), Ophisauris apodus (European glass lizard), Gyrtodactylus cotschyi (Kotschy's gecko), Eryx jaculus (Sand boa), Typholos vernicularis (Worm snake), Coluber najadum (Dahl's whip snake) and Elorphe quatuorlineata (Four-lined snake).

Land use

The modern industrialization of Maritsa Iztok did not completely destroy the agriculture of a region once known as the granary of the country. Between the devastated areas, on islands of undisturbed ground, one can surprisingly see strips of sunflower or maize. Within areas scheduled for destruction by mining, some relict agriculture is still practiced.

There is some data on land use in the area prior to industrialization, which concerns not only the study region but the whole county, that generally includes the modern Stara Zagora district.

In 1897, the statistical book of the Bulgarian Principality was published (Atanasov 1897). The part that deals with the study area shows that 147,647 persons were occupied with farming, viticulture, horticulture and forestry; 7,018 with stock or poultry raising, apiculture and sericulture and only 126 with hunting and fishing. In 1888, the total population of the area was 208,396, the ratio of urban/rural population was 1:4 and the mean distance between settlements was 5.27 km. The spatial distribution of land use comprises 200,000 ha of cornfields, gardens and melon-gardens; 9,930 ha of vineyards; 6,500 ha of meadows, 100,600 ha of original and secondary low woodlands; 324,870 ha of pasture, lakes, marshes, rivers and lands unsuitable for cultivation (Atanasov 1897).

The low percentage of the forests – just 4.3% - made some authors conclude that forest clearance of the area was progressive over a period of 1,500 years (Nam 1995).

The Czech traveller and scholar K. Irechek gives some interesting information about the land-use pattern in the last decades of the 19th century in Bulgaria (Irechek 1899). According to his report, simple cultivation without manuring was a recent form of agriculture; it was only after the Liberation in 1878 that medium and small-scale farming replaced large farm enterprises. The species grown were wheat, rye, barley, oats, millet, spelt, maize, legumes, vegetables, melons, pumpkins, tobacco, anise, sesame, cotton, nuts, grapes and fruit-trees (Irechek 1899).

In the Jubilee Book of the Bulgarian Village (Gruev 1931), there are data on 6 villages within the current Maritsa Iztok area. The territories of two of them – Mednikarovo and Mudrets - are included in the current study (Table 4.3.3).

Table 4.3.3 Maritsa Iztok subsistence in the beginning of the AD 20th century; Source: Nam 1995, after Gruev 1931
Village Main subsistence Crop types
Mednikarovo agriculture cereals, vegetables
Drianovo agriculture cereals
Mudrets agriculture tobacco
Pomoshtnik agriculture Rye, tobacco, vineyards
Glavan agriculture,
stock-breeding, sericulture
Rye, tobacco, vineyards
Tianevo agriculture, stock-breeding cereals, tobacco

According to these data, agriculture prevails over stockbreeding. The most important species were different kinds of cereals, followed by tobacco and grapes with a small quantity of vegetables (Gruev 1931).

The relative continuity of land-use was confirmed by the pre-coal-exploitation investigations in the region. An early claim for land use conservation in Bulgaria (Botev and Kovachev 1934) was later supported for the area of Maritsa Iztok in particular (Nam 1995).

4.3.2 Modern data for the Drama basin

Climate

The climatic conditions in the Drama basin require special investigations, which are yet not forthcoming. According to the Bulgarian classification, the Drama microregion and its surroundings lie at the boundary of transitional continental and continental/Mediterranean climates (Jordanova and Donchev 1997).The study area formally belongs to a transitional continental zone but there is some strong evidence for Mediterranean influence (Table 4.3.4).

The closest meteorological stations to the Drama microregion are Yambol, in the heart of the Tundja lowland, and Elhovo, in the foothills of the Strandja mountain (Fig. 4.3.3). The Bulgarian part of this mountain, especially around Elhovo, is believed to display a continental/Mediterranean climate (Jordanova and Donchev 1997). The high rainfall indices for November and December in Elhovo, though, are not a surprise and are indicators of Mediterranean influence (Table 4.3.4). Yambol is some 50km to the North of Elhovo and the Mediterranean influence is not so well documented. Drama lies between these two stations but closer to Elhovo (Fig. 4.3.3). Thus, some Mediterranean influence is not to be excluded, expressed mainly in its mild winters rather than its dry summers.

Table 4.3.4 Average month, seasonal and annual rainfall in the Drama basin; Sources: Jordanov 1937/38, Rainfall Yearbook of NRB 1962, Climatic Yearbook 1990
station source I II III IV V VI VII VIII IX X XI XII Spring Summer Autumn Winter annual
Elhovo Jordanov1
1937/38
43 38 41 38 49 53 46 23 31 38 51 49 128 122 120 130 500
Yearbook 1962 41 36 38 47 52 66 43 23 35 47 58 53 137 132 140 130 539
Yearbook 1990 46 42 35 45 53 58 43 28 36 44 59 56 133 129 139 145 547
Yambol Jordanov 37 36 39 40 54 74 49 32 37 43 54 46 133 155 134 119 541
Yearbook 1962 36 34 37 43 55 75 52 41 35 42 52 50 135 168 129 120 552
Yearbook 1990 37 35 29 46 63 66 52 35 35 40 54 49 138 153 129 121 541
Table 4.3.5 Average month, seasonal and annual temperature in Drama surrounding; Sources: Jordanov 1937/38, Climatic Year Book of NRB 1959
station source I II III IV V VI VII VIII IX X XI XII Spring Summer Autumn Winter annual
Yambol Jordanov
1937/38
1 -1 5 11 16 21 23 23 19 14 8 2 10.4 22.4 13.7 2.5 11.8
Yearbook 1959 1 1.8 6 11 16 20 23 23 19 13 8 3 N/a N/a N/a N/a 12
Elhovo Yearbook 1959 1 2.2 7 12 17 21 24 23 19 14 9 2 N/a N/a N/a N/a 12.5

As Table 4.3.4 shows, there is a general trend towards higher annual rainfall in Elhovo, mainly due to increased autumn/winter rains. The summer maximum remains dominant but a November secondary maximum has become more substantial over the last 60 years. In Yambol, the overall quantity of rainfall has remained constant. There is a decreasing trend of the absolute value of June rains but they still remain the most intensive. The increase in spring rather than autumn rainfall is an indicator of continental climate.

The average annual temperature also shows an increasing trend (Table 4.3.5); as in Elhovo, it is half a degree higher than in Yambol. This mainly due to the slightly higher autumn/winter temperatures, that reveal mild winters under Mediterranean influence.

The complex interrelation between a number of factors (altitude, precipitation, temperature, cloudiness, drought, etc.) for Elhovo are given below (in Fig. 4.3.3), which shows a pattern of hot, moderately dry summers and relatively mild winters.

Fig. 4.3.3 Contemporary climatic indices for Elhovo meteorological station; Source: Atlas na NRB 1973

Vegetation

The natural vegetation is still preserved in some areas within and around the Drama microgerion. This takes the form of forests of Q. pubescentis Wild and Virgilian oak. In the higher parts of the landscape, these forests are accompanied by C.orientalis Mild and Mediterranean elements such as Acer monspessulanum, Juniperus oxycedrus, Jasminum fruticans, etc. There are also some secondary species, such as Phyllirea latifolia, Cistus incanus, Asparagus acutifolius and so on (Fig. 4.3.4; Bondev 1991).

Fig. 4.3.4 Contemporary vegetation in the Drama basin Source: Bondev 1991

Key: 106. Natural forests of Q. pubescentis Wild and Virgilian oak; 107. Q.pubescentis Wild and Virgilian oak forests with Mediterranean elements; 112. Mixed forests of Q.pubescentis Wild, Virgilian oak and C.orientalis with Mediterranean elements partly of secondary origin; 121. Shrubs of Christ thorn, mixed with jasmine combined with xerothermal communities replacing xerothermal forest community of Q. pubescentis Wild and Virgilian oak; 129. Xerothermal grass communities with prevalence of Dichanticta ischaemi, Poaeta bulbosae, Poaeta concinnae, Chrysopogeneta grylli and Ephemereta; 138. Farm areas that replaced forests of Q.pubescentis Wild and Virgilian oak; 149. Farm areas that replaced forests of field elm, field ash-tree and *Q. pedunculiflora Koch.

Much larger areas are, however, covered with secondary vegetation. Immediate near the village and along the river, farmland has replaced forests of field elm, field ash-tree and Q. pedunculiflora Koch. These moisture-loving species have now been replaced by vegetable, fruit and corn cultivation (Bondev 1991).

The largest agricultural areas have replaced forests of Q.pubescentis Wild and Virgilian oak They are found on the cinnomonic forest soils on the basin slopes and low hills. The best tobacco is grown there, as well as some vineyards and cereals (Bondev 1991).

Very close to the contemporary village can be seen a development of xerothermic grass communities dominated by Dichanticta ischaemi, Poaeta bulbosae, Poaeta concinnae, Chrysopogeneta grylli and Ephemereta. This community usually replaces a very wide range of species, such as xerothermic forest species, secondary vegetation and shrubs (oak, hornbeam, thorn, red juniper), as well as some mesophytic forest formations, especially over eroded soils (Bondev 1991).

Natural forests of Q.pubescentis Wild and Virgilian oak are surrounded by shrubs of Christ thorn, mixed with jasmine combined with xerothermic communities replacing xerothermic forest communities of Q.pubescentis Wild and Virgilian oak and rarely replacing Q.cerris or other forests.

Land use

The area around the present village is typical rural agriculture land. The first and second terraces are covered by wheat, while, in the floodplain, garden species or weeds are dominant. Some slopes are also used for cultivation, especially for legumes, vineyards, maize and fruit-trees. The pattern of land use was totally destroyed by communist cooperative farming that is still recognizable in the bulk of uncultivated lands, now covered by weeds and grass. These areas, as well as gently and moderately sloping land, are used for pasture. According to the local farmers, shifting cultivation is practiced, but the species and the rotational cycle have not been determined. The distribution of rankers is believed to supply good current grazing. A relatively recent development, dating to the Ottoman period, is hill-slope cultivation, which resulted in soil erosion and the deposition of colluvium containing Turkish sherds at the foot of the hills (Kubiniok 1996).

4.3.3 Palynological evidence

Introduction

Palynological investigations are a major component of most contemporary interdisciplinary palaeo-environmental studies. Their main task is to reconstruct the past vegetation of the surveyed area and, on a broader scale, the overall vegetational development during the last 15,000 years. The assumption that certain floral taxa can tolerate certain weather conditions made pollen data a primary source for climatic reconstruction – another crucial factor in any environmental investigations. Although vegetational history is generally seen as response to macroclimate changes (Huntley 1990, Wright et al. 1993), nonetheless there are numerous examples suggesting that this oversimplified climate/vegetation interrelation is not uniform and simple (Magny 1982, Joos 1982, Beug 1982, Willis 1994; Magyari 2002).

The last phase of the present geo-chronological sequence – the Holocene - was subdivided into five climatic stages – Preboreal, Boreal, Atlantic, Sub-Boreal, Sub-Atlantic (Roberts 1998:29 citing Blytt-Sernander 1878 - 1906). As a primary source for vegetational history, pollen data justify these stages on a regional level and together with other environmental factors can support the existence of possible local climatic fluctuations (Iversen 1973, Harding 1982, Lamb 1982). In the best cases, broad interdisciplinary studies including lithological, geochemical, molluscan and pollen analyses, together with reliable absolute dating, can provide a good set of data, whose interrelated interpretation might be considered as an appropriate palaeo-environmental reconstruction (e.g., the Ystaad project: Berglund 1991). In most investigations, however, pollen analysis solely has been applied at a broader scale.

The very general explanatory framework of pollen data is that pollen rain is deposited in lakes and peat bogs, as evidence for the surrounding vegetation. Depending on the size of the basin, pollen rain is more or less useful for a broader picture of plant assemblages. If a basin is smaller than 5 ha, the pollen derives from local sources; if the basin is larger, pollen from up to 100 km may have been present (Willis et al. 1997, Willis et al. 1998). It is a key part of the interpretation of pollen data to assess the presence of taxa in the diagrams according to the type of pollen dispersal mechanism, their productivity, the degree of vegetational stability and the quality of pollen preservation.

In archaeology, palynological data has been mainly used for studying the important breakthroughs of cultivation and domestication. The phrases Neolithic Transition" andForest Clearance" became synonymous for human control over natural vegetation and were (and still are) highly debated in their environmental, social, economic, technological and even linguistics aspects (Sherratt 1981, articles in Harding 1982, Ammerman and Cavalli-Sforza 1984, Dennell 1983, Barker 1985, Renfrew 1987, Zohary and Hopf 1988, Mallory 1989, Willis and Bennett 1994, Willis et al. 1998).

Information on the stages, intensity and diversity of the human impact on a particular environment derives from joint archaeo-botanical research. Since the early seventies, when the first interdisciplinary palaeo-environmental reconstructions started to appear, the inter-relations between human communities and vegetation has covered the whole spectrum of possible explanations – from the overwhelmingly cultural importance of the spread of agriculture (Sherratt 1981, Ammerman and Cavalli-Sforza 1984, Dennell 1983, Barker 1985, Renfrew 1987, Zohary and Hopf 1988, Mallory 1989) to the opposite extreme – the dominance of purely environmental factors in vegetational changes until cca. 2500 BC (Huntley 1990, Willis and Bennett 1994, Willis 1994, Magri 1996).

The present situation in Bulgaria – data and interpretation

Over the last 30 years, intensive palynological investigations were carried out in Bulgaria. The establishment of vegetational distribution and variability in the Pleistocene and Holocene was the main goal of the survey of different ecological zones - the Black Sea coast, the full altitudinal range of the Bulgarian mountains, as well as diverse lowlands. The vegetational cover during glacial periods, the presence of refugia for certain taxa and their subsequent migration were the earliest events for consideration in these studies (Filipovich 1981, Bozilova and Tonkov 1985, Filipova 2003 and many others). The further development and diversification of species was a second major task of Bulgarian palynological investigations (Bozilova and Tonkov 1984, Bozilova 1986). Last but not least was the discussion for the type and degree of the human impact on natural vegetational development (Bozilova 1986; Filipova - Marinova and Bozilova 1995).

Some marine palynological surveys of the Bulgarian Black Sea shelf (Komarov et al. 1979, Filipova et al. 1983; Filipova 2003), as well as litho-stratigraphic and bio-stratigfraphic schemes established in marine sedimentological investigations (Khrischev and Shopov 1978, Chepalga 1984), focused joint palaeo-environmental studies on their synchronization with past ecological events (e.g. marine transgressions, climatic changes and vegetational developments). Thus, a Holocene chronostratigraphy was established for Bulgaria (Bozilova 1982), following the generalization of results on a national or European level (van der Hammen and all 1971, Bottema 1974, Beug 1982, van Zeist and Bottema 1982, etc) and according to the climatic-stratigraphical scheme of Blytt-Sernander, the bio-stratigraphical scheme of Firbas (Firbas 1949) and the chronostratigraphy of Mangerud.

Joint archaeological / palynological investigations, however, are still extremely rare2. Only two pollen cores have been taken from places with archaeological sites in the immediate vicinity – both in Northeast Bulgaria:- tell Durankulak and the settlements along the former shoreline of the Varna – Beloslav lakes,. More common are archaeo -botanical studies that resulted in a substantial body of cultivated taxa and weeds of cultivation recovered from archaeological sites (Hopf 1973, Behre 1977, Lisitsina and Filipovich 1980, Chakalova and Bozilova 1981, Yanushevish 1983, Popova 1995 (and references therein), Popova and Bozilova 1998).

The past vegetation cover in Bulgaria and its trends of change or stability has been considered in a broader Balkan and European context (Dennell 1983, Huntley and Birks 1983, Huntley 1990, Willis 1994, Willis and Bennett 1994). Most relevant for archaeological studies are Dennell’s and Willis’ opposing models for substantial (Dennell) and minimal (Willis) environmental impact of the earliest farmers in Southeast Europe. Botanists and archaeologists who have studied plant remains and subsistence strategies at the site level tend to support Dennell’s hypothesis (Hopf 1973, Dennell 1975, Yanushevich 1983, Bozilova 1986, Popova 1995).

The basis for this opposition lies in the data and objectives of the different studies. On one side, there are archaeo-botanical studies searching for cultivated plant remains , with archaeologists trying to incorporate this evidence into a broader socio-economic context. Since the data comes from archaeological sites, in general, these data indicates selective human choice and hence, is not representative for overall vegetational cover. According to this view, the human/vegetation link is seen as the most important factor, which results in interpretations which underlined anthropocentric stress under conditions of, e.g., the adoption of agriculture, with the presence of certain taxa used as indicators of human activity, patterns of land use and crop rotation.

On the other hand, there are global palaeo-environmental studies, which aim to find common features among the scattered pieces of past ecological data and delineate general trends of environmental development. Crucial for these studies on the first place are the similarities and only then the differences that usually appeared at a regional level. The weak point in every general palaeo-environmental study is the regionality of the pollen data. It might be avoided by juxtaposing a series of pollen coring results deriving from one ecological area, as, for example, was done for the Rila Mountain (Bozilova 1977/78).

Both attempts to reconcile these two different approaches to palaeo-vegetational data (Bozilova 1986) and (Willis 1994) failed to provide a relevant palaeo-ecological reconstruction of Holocene Bulgaria according to all available archaeological and ecological data.

In the first case (Bozilova 1986), all the currently known plant remains were mechanically charted, without identification of any trends of human vegetation exploitation or giving any possible explanation of the recurring patterns of such exploitation and hence, reasons for specific cultural practices. It was inferred that there was an anthropogenic impact but no common, regional, chronological or any other human/vegetation interrelation patterns were established.

In the second study (Willis 1994), despite the main focus on human impact in the Neolithic, Copper Age and Bronze Age, evidence from only one Bulgarian archaeological site (Durankulak) is included. Therefore the researcher’s claim for minimal human impact on the natural vegetation is hardly surprising. Willis’ model will be discussed in some detail later on (see below p.149) but here it is worth noting her other claims for Balkan vegetational history. The first one is for the expansion of Pistacia between 9000-8000 BP, the second concerns the change in forest dominance between 8000-7000 BP and the third postulates the increase of hornbeam, fir-tree and beech in the woodlands between 7500-5000 BP (Willis 1994). All of these results are important aspects of the Holocene vegetational succession in the Balkans.

The study regions and the problem of their palaeo-environmental reconstruction

An important part of palaeo-environmental reconstruction of the study regions, then, is the establishment of the past vegetational succession. The first difficulty in any attempts at such a reconstruction is the lack of pollen investigations within the study area3. The second difficulty appeared when data from the two nearest pollen-coring sites (each at a distance of 100 km from the study area) were overlaid in order to test the relevance of such a mechanical approach. The contradictions and similarities in the two pollen indices were so confusing that any interpretation based on this approach would be highly speculative. Therefore, in the situation of lack of any reliable pollen data, an alternative approach – indirect but less speculative – was applied to reconstruct the past vegetation of the study regions. The method called Principal Component Analysis (PCA) is one of the multi-variant data analyses that has been introduced in archaeology as a useful tool for identifying similarities and dissimilarities in complex data (Doran & Hodson 1975; for a PCA approach to Bulgarian palaeo-ecological data, see Ognjanova- Rumenova et al. 1998).

In general, PCA can identify similar interrelated characteristics – positive and negative – among a set of variables, which are different from the characteristics that define each variable. In the case of pollen data, two sites with apparently different vegetational developments, such as high mountain and lowland environments, for example, might appear to be very similar according to other axes. If the sites differ in their physical background and hence, simultaneous development of similar species, they may correspond in diachronic terms and present a consistent and similar balance of taxa, as in the apparent similarity of vegetation in the Preboreal/Boreal Mountain diagrams and those representing the Sub-boreal Plains4. So one has always take care with these analyses and make appropriate comparisons.

As with all statistical methods, PCA does not answer the question why; it is the task of the researcher to make sense of the results of the data analysis. Such an attempt is made for the reconstruction of palaeo-vegetation trends for certain areas in Bulgaria. The second stage is to extrapolate these palaeo-environmental results to the Maritsa Iztok and Drama microregions. A third outcome of the PCA analysis in this case is to test the validity of K.Willis’s model for Balkan vegetational history.

The PCA application

PCA is a part of the SPSS package, in which primary data is stored in tabular form together with variables of the researcher’s choice. Both tables and/or graphs permit the visualization of the results of the analysis. In the current study, 25 pollen samples were used. They derive from eight coring sites - Arkutino (Bozilova and Beug 1992), Durankulak, Kupena and Bezbog (European Pollen Database), Srebarna (Lazarova and Bozilova 2001), Varna (Bozilova and Filipova 1975), Sadovo (Filipovich and Stoyanova 1990) and Shabla (Filipova 1985) - deliberately chosen to represent the widest possible range of landscape forms (for further details see ATable 4.3.1). Most of the samples have exact 14C dates but, in 4 cases, dates are interpolated. Using OxCal v.2.18 (Stuiver & Reimer 1986), calibration of all BP dates back to 6800 was possible. There are at least two determinations from each coring site. Some sites contribute three, others four samples. The choice of variables fell upon 15 species recognised as the most representative and relevant for palaeo-environmental study. The value for the variables comprises the pollen percentage of the species in question as they are given in the publication of the core. Apart from the original pollen diagrams and text-based interpretations in the articles, the European Pollen Database on-line archive was used for 3 sites (Durankulak, Kupena and Bezbog). The types of geographical background, 14C dates (BP and CAL BC) and taxa percentage of all the samples are given in ATables 4.3.1- 4.3.2. The full results of the PCA are presented in Appendix 1.

Interpretation of the PCA is based on only ATables 4.3.6 - 4.3.7 and AFigs. 4.3.2 – 4.3.4 and 4.3.6 - 4.3.7. ATable 4.3.7 shows that, after the extraction method (the initial stage of the PCA), six components appeared to be important for interrelations of this set of variables. The commonest is component 1, which accounts for 22.29% of the total variance; less important is component 2 (17.59% of total variance) and so on, in descending order. Usually not more than 3 components (the first ones) are used in PCA interpretations, since they have been considered as the most significant. The low analytical value of component 4 and downward is confirmed by this case study. In AFigs. 4.3.4 – 4.3.6 and 4.3.9, components 4-6 are shown to characterize extreme situations with no more then 2 samples for each variable; this is therefore not a trend but rather an exception. Extreme situations may be due to the type of the data or to collection and analysis of the primary data, or they may represent certain reality concerning a certain trend, a particular period or some other kind of specific development. For instance, component 5 opposes C. Betulus and Poaceae to Betula and Rumex. An attempt to name the component as a possible environmental one might be very misleading. Reference to AFig. 4.3.10 and ATable 4.3.2 shows that sorrel and birch have their highest distribution at the Srebarna coring site (Sr1 and Sr2), while, at Late Atlantic Shabla (Sh3), C. betulus reaches its peak development according to the other 24 samples. Thus, component 5 for this set of variables represents regional characteristics for three species and therefore is not relevant for the identification of general trends.

Components 1-3 account for 53.70% of the total variance and their characteristics are plotted on AFigs. 4.3.2 – 4.3.5. AFigs. 4.3.6 – 4.3.10 show the inter-relationships between samples and hence coring sites according to these three components. All graphs show that samples tend to cluster according to their place of origin and occasionally to their chronology. This reveals the regionalism of the South Balkan vegetational development and confirms the limitations of pollen data for general palaeo-environmental reconstruction.

Component 1 as displayed on AFig. 4.3.3 could be recognised as an altitudinal component, in which deciduous forests are opposed to coniferous woods. The distribution of samples confirms this opposition (AFig. 4.3.6); thus, high mountains (B1-3) have nothing in common with lowlands at the foothills of a relatively low mountain (A1-4). The Upper Thracian Plain sample (S1) is in the high mountain group because of the extremely high presence of Pinus. According to the publications, this is due to long-distance transport and the stability of Pinus pollen rather then a reflection of real ecological conditions. The last phase of the Varna lake settlement samples (V3) falls in the left side of the scatter because of a sharp decrease in oak, perhaps as a result of intensive forest clearance. The Early Atlantic Danube Plain (Sr1) and Late Atlantic North Black Sea Coast (Sh3: 4580 – 4450 CAL BC) samples used to share the same abundance of mixed oak woods as was found throughout the Late Atlantic-Late Sub-Boreal in the South Black Sea Coast (A1-4: 5210-5050 – 1515-1425 CAL BC). Almost half of the samples lies between the two extremes, indicating that sites are in intermediate locations in relation to the strong altitudinal opposition, as hilly landscapes or even lowlands but with a vegetational dominance of neither mixed oak forests nor pinewoods. A greater degree of taxa diversity, including herbaceous plants, is typical of these places.

Component 2 is harder to interpret, although, in general, it opposes a neutral to moist environment to cold weather conditions (cold nights or cold winters). Its accounts for 17% of total variance and it should be considered together with component 1 when their cumulative value becomes almost 40% (ATable 4.3.6). In AFig. 4.3.6, samples / coring sites are plotted according to component 1 (x-axis) and component 2 (y-axis). A grouping of samples means that sites share broadly similar environments defined by components one and two. The most prominent groupings are A2-A4; Sh1-2 –K2; V1-2-K1-S2 and Sr2-K4. For interpretative purposes, however, broader clusters are important, as well as trends in time at single coring sites according to both components 1 and 2. In this sense, cores from the two archaeological sites (V and D) seem to share similar environments in contrast to widespread current claims for different ecological conditions in these two parts of North Black Sea Coast (Bozilova 1986; Filipova- Marinova 2003). In the last phases of the sites (V3 and D3), the sparse stands of cold-tolerant species (D1, 2 and V1, 2) were replaced by Chenopod-dominated grassland in the case of Durankulak and oak clearance in the case of Varna. In chronological terms, D3 (1420 – 1300 CAL BC) falls at the beginning of the Sub-Atlantic – a period usually related to cooler weather conditions – that favours the expansion of cold-tolerant Chenopodiaceae. The last sample from Varna is from the end of the EBA and may indicate the results of EBA forest clearance. To make any conclusions about the palaeo-environment of Northeast Bulgaria, two more sites should be considered.

The first site – Shabla (Sh1-4) - suggests another refutation of contemporary interpretations of past ecological conditions. Durankulak and Shabla are about 30km apart each other and both are liman lakes. It has always been accepted that they share a similar vegetational history (Filipova 1984; Bozilova 1986). AFig. 4.3.6 is anything but confirmation of such a claim. The first two phases of Shabla are much earlier then the first sample from Durankulak but even the phases that are close in time (D1 (4230 – 4190 CAL BC) and Sh3 (4580 – 4450 CAL BC)) are widely separated on the scatter plot. The same pattern occurs with contemporary samples D3 and Sh4 (both 1420 – 1300 CAL BC). This may be interpreted either that there were wet areas around the lake that favoured the development of moisture-loving species or that weather conditions around Shabla, in general, were not as dry and cool as previously thought. A reference to the species distribution during these phases confirms a different distribution pattern in both areas, with the only similarities of relatively high Chenopods and the presence of Artemisia (ATable 4.3.2).

The second site is Lake Srebarna in the Lower Danube Plain (Sr). The apparent drop of deciduous pollen values from Sr1 to Sr2 is due to a decrease of 50% in hornbeam values for both C. betulus and C. orientalis.

As was mentioned earlier, the samples show regional patterns of past vegetational development. A general trend in Northeast Bulgaria, however, is the decrease in deciduous taxa that had generally started by the middle of the 5th millennium CAL BC. In climatic terms, all the sites tend towards cold weather conditions in their last phases, which generally coincided with the beginning of the Sub-Boreal.

The regionality of vegetational development is not as closely connected to geographical latitude as one may expect, at least on the scale of Bulgaria and with this particular set of sites and variables. Samples from South Bulgaria tend to cluster with samples from North Bulgaria, following the same regional pattern and with the appearance of differences only at the level of diachronic trends.

As was mentioned earlier, the Arkutino marsh, near the Southern Black Sea Coast in the foothills of the Strandja Mountain (A1-4), reveals a long development of mixed oak woods, with fluctuations in the type and density of deciduous species (AFig. 4.3.6). Unfortunately, no diachronic information could be extracted from samples from the Sadovo bog, which is in the heart of the Upper Thracian Plain (S1 (2590-2460 CAL BC) and S2 (1515-1420 CAL BC)). Any interpretation would be biased because of the high percentage of pine in the first sample that, as already mentioned, is not natural for the region. The only valuable information for this bog comes from component 3 (AFigs. 4.3.4 and 4.3.7), which displays the highest densities of Aster-type and Poaceae among all the coring sites. Both bogs are neutral to component 2, which may be interpreted as a stable balance of moisture-neutral and cool/drought-tolerant species.

Before moving to the high mountain samples, it is important to interpret the contribution of component 3 to the overall palaeo-ecological reconstruction (AFigs. 4.3.4 and 4.3.8). Apart from Sadovo (see above p. 145), this component is highly informative about the high presence of grasses at Shabla (Sh1-4), that once again underlies the differences between Durankulak and Shabla. The lower part of the graph shows a higher presence of Corylus in the samples (V3, K2-4), while D3 reveals a dominance of Chenopod over Aster-type and Poaceae pollen.

The only high mountain sample in this set comes from Lake Bezbog in the Pirin Mountain, Southwest Bulgaria (B1-3: 5200 – 2280-2240 CAL BC). It shows a constant development of high-mountain species, with increasing conifers and Fagus towards the later periods (AFigs. 4.3.3 and 4.3.6).

The last coring site is Lake Kupena, located in the low mountain range of the West Rhodopes (K1-4). It was deliberately chosen to be the last in the interpretation, since it has a different pattern of development (AFig. 4.3.6). First of all, Kupena has a surprisingly stable cover of oak and hornbeam from K1 (9288 uncal BP) to K4 (1940-1770 CAL BC), found in no other coring site. The greatest change falls within component 2 and, to lesser extent, in component 3 (AFig. 4.3.8). There is a very intensive development of Ulmus and Tilia and some increase in Corylus. There are two possible explanations for this trend. The first one suggests competition between elm and lime with oak, that favours the development of Quercus throughout the whole sequence. The second hypothesis, that is accepted as more relevant in this particular case, assumes human impact. Together with Tilia and Ulmus, moisture-loving oak taxa were also developed. It has been argued that oak was widely used by the inhabitants of the Western Rhodopes foothills and high hollows (Chakalova & Bozilova 1981, Marinova 1999) since their appearance in the area. Corylus colonized cleared areas as secondary plants after deforestation or as pioneer taxa on open slopes (Bozilova 1977/78); these same areas were later occupied by elm, lime and oak. Oak clearance continued, while Tilia and Ulmus were not so intensively exploited by human communities, leading to a steady presence of oak and an increasing abundance of elm and lime. The decline in all of these species in the last phase (K4) could be either due to increased human exploitation or a climatic change to cooler conditions that allowed the development of competing, more cold-tolerant species such as Fagus and Pinus.

How do the PCA results relate to palaeo-environment of the study region?

The regionality of the Bulgarian pollen data has been underlined several times already. In the absence of a more precise source, however, pollen data from Kupena will be used as reference point for palaeo-environmental reconstruction of the study regions. There are two reasons to consider it as relevant. First, it is the only coring site of the set that shares a similar hilly environment with Maritsa Iztok and Drama, despite its higher altitude of 800 m in the Western Rhodopes. Secondly, successive vegetational processes during the last century in Maritsa Iztok and Drama regions show the presence of species generally present in the Kupena diagrams. Thus, accepting the limitations of the present data, the following hypothesis for the palaeo-environment of the study regions was suggested.

Mixed oak woods occupied the hilly brown forest soils areas. They were not very dense, bearing in mind the relative position of Kupena and Arkutino according to component 1, with the latter diagram indicative of denser oak woods than the former. Among the trees, an under-canopy of different shrubs and bushes would probably have developed. Since smolnitsa soils, with their wide distribution in the each micro-region, tolerate the development of woodland, deciduous trees also covered the low slopes, now under intensive cultivation. Along the rivers, moisture-loving species grew on alluvial meadow soils. Ulmus minor, Tilia and Quercus robur were maybe the first intensively cleared tree taxa in order to open up access to the alluvial soils, which were easy to cultivate. Grassy communities of Chenopods – species that were present in each diagram – must have also developed.

A difficult question to answer is the extent of Mediterranean influence in the Drama microregion. As mentioned earlier (see above, p. 133-137), the area has a strong Mediterranean climatic influence that is confirmed by the contemporary presence of some Mediterranean vegetational elements (Bondev 1991). If we assume that the average annual temperature during the Atlantic was 3-4° higher than nowadays, an even more Mediterranean –like environment could have been prevalent in the Drama microregion. The cooler weather during the later periods – especially the Sub-Atlantic - diminished the Mediterranean elements in the Drama environment, gradually leading to the present sporadic evidence of Mediterranean influence.

How do the PCA results relate to contemporary models of Balkan

vegetational history?

According to Bulgarian palynologists, the vegetational succession of the last 15,000 years in Bulgaria can be summarised in three general points (Bozilova 1986):

(1) in the period 13,000 – 8,000 uncal BC, the lowlands and foothills of Eastern and Southwest Bulgaria were covered mainly by xerothermal grass communities. There was no clear forest boundary and the quantity of deciduous and coniferous species fluctuated according to stadial and interstadial conditions. The existence of refugia claimed for other areas of the Balkans (der Hammen et al. 1971, Bottema 1974) has been confirmed for the Bulgarian uplands as well (Bozilova 1986). The current PCA does not deal with such early periods but general observations made during the study do not contradict this vegetational development.

(2) local environmental factors such as climate, topology, edaphical conditions and the distance of refugia caused several different diachronic trends in forest development during the Holocene (Bozilova 1986). An important conclusion is that coniferous vegetation in the low mountains retreated after the expansion of the beech 2,500 years ago. In contrast, the high mountains show an intensive development of different coniferous species during the last 3,000 years. At the beginning of the Holocene, the low mountains were occupied by xeromesophyllic oak and pine forests, while, during the climatic optimum, fir-tree forests were dominant (Bozilova 1986). Local trends of vegetational development were confirmed by the present study, as well as the competition of beech with other species in the low mountains and its stable position in the high mountains (AFig. 4.3.5).

(3) three trends in vegetational spread and sequence along the Black Sea coast were established. Varna Lake and Arkutino bog are believed to represent a short dynamic period at the beginning of the Holocene, with a rapid replacement of grass communities by woodland. This period was followed by a long-lasting, balanced mesoxerophyllic oak and hornbeam forests (Bozilova 1986). Since the earliest date for Varna is from the Eneolithic (5th millennium CAL BC), it is difficult to justify the first part of this conclusion. The second half, however, is not confirmed here. As mentioned above (see p.143), only Srebarna and Shabla share to some extent the diversity and density of the mixed oak forests of Arkutino. There were woodlands around Varna but they were far from abundant on the South Black Sea coast (AFig. 4.3.6)

Similar environments have also been claimed also for Durankulak and Shabla lakes. Steppe xerothermal grass communities are considered as primary and, shortly after 5500 uncal BC, some xeromesophyllic deciduous species developed there (Bozilova 1986). The differences in the vegetational histories of Durankulak and Shabla have already been argued (see p. 145). The expansion of deciduous species in Shabla happened between 4900 and 4500 CAL BC, while, in Durankulak, woodland was always less dense than at contemporary Shabla (AFig. 4.3.6).

The last claim for Black Sea coastal vegetation for the formation of longoz (hornbeam and rhododendron forests) forests around the mouths of the rivers only in the last 3,000 years cannot be justified due to the chronological and territorial scope of the current study.

The final task of this botanical section is to assess K. Willis’s model for the vegetational history of the Balkans (Willis 1994). The principal aim of such a model is to identify long-term and widespread processes, whose main weaknesses are subjectivity and selectivity. While these shortcomings were more or less successfully overcome in other studies of global vegetational trends (Huntley 1990), Willis’ model seems to suffer from selectivity in site choice (problems with data source) and subjectivity in the method of investigation.

According to the PCA results, some general observations concerning Willis’ four main claims will be given. Detailed objections to some facts in Bulgarian data are presented in ATable 4.3.10.

Since the current data set contains just 3 samples from the period 7000-5000 uncal BC, which is the subject of Willis’ first two claims, additional observations made during the process of this research will be used to make two general points.

First, Pistacia pollen is not present in the diagrams either between 7000-6000 uncal BC or later. Small stands of this species might have been present in Bulgaria but there is no palynological evidence for its existence, even in areas with Mediterranean influence.

Secondly, the suggested change in forest dominance between 6000-5000 uncal BC, especially concerning Corylus (Willis 1994, 781), is not supported by the Bulgarian data. As shown by component 3, with Corylus as one of its characteristics, only one out of all the eight sites is sensitive to Corylus development, particularly after 4950 CAL BC (ATable 4.3.1, AFigs. 4.3.4 and 4.3.8). The claim for dominance change is a strong one but not one that can be supported. Instead, data from Bulgaria would rather suggest diversification of tree taxa in this period, with oak dominant in the lowlands and pine and birch in the uplands.

The third vegetational change broadly defined as an increase of C. orientalis/Ostrya, Abies, C. betulus and Fagus in the woodlands between 6400-3900 CAL BC appears to be based largely on speculation. As AFigs. 4.3.3 – 4.3.5 show, while hornbeam and beech are not inversely related, they did not develop together. Their different vegetational history is confirmed by component 6, which identifies Fagus on its own as an important variable, accounting for 6% of the total variance (AFig. 4.3.5). AFig. 4.3.10 demonstrates a clear increase in beech in only 3 cases (Sr, K, B), all of them however, after 3900 CAL BC. Abies is not among the associated variables but, as argued elsewhere (Bozilova 1986), it shows similarities to the Fagus development as one of its main competitors. The claim for a general increase in hornbeam can be accepted only on a regional level, since component 1 shows only two sites (A and Sh) with an increase in Carpinus levels in the period 6400 – 3900 CAL BC (AFig. 4.3.6).

The last of Willis’ claims will be considered in a little more detail – her claim that anthropogenic disturbance did not start earlier then 3300 CAL BC. This date develops Willis’ previous hypothesis for 5000 CAL BC as a possible beginning for agricultural impact upon the Balkan landscape (Willis and Bennett 1994). While little could be said against the arguments for post-3300 human disturbance in terms of loss of forest cover, it is doubtful that this activity did not appear before 3300 CAL BC. As the cases of Kupena and Shabla demonstrate, human deforestation started at the beginning of the 5th millennium CAL BC. The oak decline at the Varna sites and the remarkable continuity in oak frequencies at Kupena suggests selective tree-cropping, as confirmed by archaeo-botanical evidence from the Varna lake settlements and the Western Rhodopes (Bozilova and Filipova 1975, Chakalova and Bozilova 1981, Marinova 1999).

Whether or not cultivation of plants did or did not affect the surrounding landscape is very hard to generalise for the whole country from the evidence from just one site (the site of Durankulak for Willis). An appropriate approach to this problem is the correlation between archaeo- botanical results from specific archaeological sites and pollen analysis from a basin in the immediate vicinity of the same site that will provide a full picture of natural vegetation sequence prior to and contemporary with human occupation, together with details of the cultivation of certain species and its effect on the surrounding environment. This, I believe, will provide enough evidence to push back the boundary of visible human impact much earlier than 3300 CAL BC.

4.3.4 Other palaeo-environmental sources

There are a few more data sources indirectly related to the past environment of the study regions.

About 20 km South of the Maritsa Iztok area, an Early Chalcolithic settlement near the village of Luda reka was investigated. Pollen samples from archaeological features – a pit and a trench – were taken there. Leaving behind the controversial sampling technique, the results are worth mentioning. Deciduous species such as oak, elm, ordinary and oriental hornbeam, hazel and cornell trees were growing there during the Early Copper Age. Fagus and Salix were also present, as were the herbaceous species Artemisia, Chenopodiaceae, Rumex, Plantago lanceolata and Polygonum aviculare (Lazarova and Stefanova 1997).

While working on other studies, three historians - Irechek, Casson and Venedikov and a Bulgarian economist - Gruev, have provided some interesting environmental information for the historical past of the three study regions.

The Czech scholar and traveller K.Irechek, citing Medieval sources, describes the area around present-day Stara Zagora as very rich, with an abundance of wheat, barley, rye, wine (resp. vineyards), flocks and herds (Irechek 1899).

In the early decades of the 20th century, the Classical scholar Stanley Casson made several scientific trips to Southeast Europe. In 1925, he published a book on Macedonia, Thrace and Illyria, based on classical sources and his own observations (reprinted as Casson 1968). Some of the information concerning natural resources might, again very carefully, be used for retrospective analysis of the past environment.

In this book, Macedonia is mentioned as a supplier of timber to the Greek world, while Thrace5 is described as its granary. Casson suggested that the enclosed plains of Macedonia were less suitable for wheat cultivation than the wider and more open plains of the Nestos and Hebros6 and the downland of eastern Thrace. The high soil fertility of these areas was evidenced by Pliny’s mention of the Hebros valley as producing corn that was reaped in the third or the second month after sowing. An interesting parallel is made with England, where the usual Thracian crop of ten times the amount sown, is achievable after heavy manuring and much labour on the soil (Casson 1968).

According to Casson, a great part of the tobacco-growing areas that he has seen during his trips have replaced ancient vineyards. Casson refers to a 17th century traveller who admired the quality of Maronean wine (Casson 1968).

This intriguing but fairly general information gives a general picture of the South Bulgaria landscape in the late prehistoric (LBA) and in early historic times. During the period before the first historical documents, there was a gradual expansion of wheat cultivation in the Maritsa plain. Whether some forest clearance has taken place in order to expand the cultivation area remains so far an open question. The timber trade did not, however, stimulate intensive deforestation. Vineyards most probably occupied hill-slopes and less favourable areas.

The Maritsa Iztok study area is in the Maritsa basin catchment on the one hand and at the edge of the low foothills of the Sakar Mountain, on the other. One may presume that the increase in wheat cultivation led to some forest clearance. Bare spots on the rolling hills that are in abundance here might have been occupied by small vineyards.

Venedikov (1981) wrote a monograph about agriculture in Bulgaria according to ancient sources. He does not contradict Casson’s observations and enlarges the scope of evidence for the environment of the Classical period (Venedikov 1981). Several authors including Herodotus, Appian and Thucydides mention dense forests spread over Strandja, Sakar , the Rhodopes and Stara Planina mountains. The word venerable used by Venedikov when citing Herodotus may let us conclude that natural vegetation (in this case mixed oak forests) was preserved to a great extent in the mountain areas during the Early Classical period. Nonetheless, wood was widely used as a building and heating material (Venedikov 1981). An interesting aspect of coniferous exploitation was tar production. Theophrastus gives us information about its production in Thrace in the 4th-3rd centuries BC. Tar can be extracted only from pine and it is believed to be crucial for protection from damp (structural beams and uprights, boat timbers etc.) and as a grease (Venedikov 1981).

These ancient sources of information are important in two respects. First, there is an implication of possible pine exploitation in the study regions and, secondly, a trade or exchange network probably existed with the high mountain regions. Coniferous exploitation in later prehistory has been confirmed for Neolithic tell Rakitovo, at 800 masl in the Rhodopes (Bozilova and Chakalova 1981).

Another crucial piece of information coming from Aristotle concerns the use of coal that derives not from mining but from river banks and beds (Venedikov 1981). These data reinforce the above-stated hypothesis for possible prehistoric lignite exploitation in Maritsa Iztok, since coal is still visible in the Sokolitsa river bed and it is claimed that pieces of coal were deposited on the Galabovo tell (Popova 2001).

Several years after Casson’s book, the Jubilee yearbook of the Bulgarian village was published (Gruev 1931). There, one chapter is dedicated to the forests and their devastating destruction. However, the article consists of some curious information that might be of some relevance to the current study. The author describes a picture of vigorous life in the mountain and semi-mountain villages and their surroundings as they appeared in 1930. On the denuded hills around the village, the slightest rainfall caused gullies that spread mud and gravel over the fields, gardens and meadows. Beyond these deforested hills, there were low woodlands with single-species areas of oak, beech or elm. In the most remote surroundings of the villages lay the undisturbed forests (Gruev 1931).

All the three microregions are on the edge of a semi-mountain environment. It is possible that such a pattern of concentric land use zonation may have been found there until recently and on a much smaller scale in later prehistory. The areas closest to the settlements were gradually cleared and most probably cultivated. Beyond the agricultural lands were the pasture areas which may have included sparse woodland. The last land use category comprised dense forests – the source of timber for heating and building materials.

The final palaeo-geographical data that concerns the Drama microregion are the results of inter-disciplinary investigations of the German expedition. According to the results of the survey of the physical environment, few changes have happened since the earliest occupation of the region. The first Neolithic settlement occurred on a terrace of the steep bank of the river Kalnitsa, which is now covered by meadow clays and colluvium. Site abandonment was most probably caused by river overflow, that was followed by millennia of sedimentation, leading to the formation of the current course of the Kalnitsa. The Copper and Bronze Age multilayer settlement known as tell Mezdumekia was formed by a low, even hill overlain by 1.5 m of cultural deposits. The Western part of the hill was formed by an earlier steep bank of the Kalnitsa, which was covered by sediments during or after the site occupation (Kubiniok 1996). The first dwelling on the tell Mezdumekia was founded on a naturally defended place, since the hill was even steeper than the adjacent areas than it is today and was surrounded on three sides by the Kalnitsa and its small (un-named) Northern tributary (Lichardus et al. 2001).

Small-scale geomorphologic investigation in Drama microregion was made within the current fieldwork study. Sediment samples were taken from the locality Ortabozaluk - 1.5 km North of the present village and 2.5 km Northeast of the tell Mezdumekia. Three climatic phases were recognized for the Holocene in the Drama microregion. The first period is characterized by a mild and warm climate and the slope where the sample was taken from was well-drained, most probably indicating active diluvial wash. The subsequent cooling of the climate was inferred from the sample’s characteristic signs of activation of the sedimentation process. More intensive rainfall brought more water, which transported a heavier gravel fraction. It was inferred that this was a process of active gully erosion. The last phase of climatic conditions, according to the content of the third sample, was wet but warm. Formation of the uppermost layer was influenced by pedogenesis and aeolic processes. The high percentage of clay in it indicates long-lasting, mainly chemical weathering (Kenderova et al. n.d.). Unfortunately, these climatic phases are as yet undated. However, the general conclusion is for lack of any drastic changes in Drama microregion relief during the Holocene.

4.3.6 Summary

The differences between the present environment and the palaeo-environment in the three study regions lie mainly in the degree of forest cover, which was much denser in later prehistory. These are areas with deciduous forests, in which the Drama microregion contains more prominent evidence for Mediterranean presence. The PCA of dated pollen assemblages casts doubt on some of the accepted tenets of Bulgarian vegetational history. Anthropogenic impacts on the prehistoric environment by deforestation and cultivation started earlier than has hitherto been claimed but these impacts had no devastating long-term effects and did not cause severe erosion; on the contrary, sustained, successful agro-pastoral strategies continued from the Neolithic up to the modern farming of the 20th century. The soil distribution in the three study regions is similar. I maintain that there was no post-Neolithic chernozem formation in the Drama microregion but rather this soil was a pre-Neolithic smolnitsa. Climatic changes generally followed the established sequence of warmer and drier periods during the Atlantic period, succeeded by cooler and wetter periods during the Sub-Boreal.

The main change in the physical environment was the devastating open-cast mining in the Maritsa Iztok study area, that has dramatically changed the rural environment for ever.


  1. The values are for 28 years’ observations in Elhovo and 30 years in Yambol↩︎

  2. In 2002, sediment coring took place near the tells of Ezero, Galabovo and Djadovo: pollen analysis is currently in progress.↩︎

  3. The sediments from the 2002 core from the small marsh near Galabovo contain very low pollen concentrations (pers. comm., Prof. E. Bozhilova).↩︎

  4. For detailed description of the method see (Doran & Hodson 1975) and Norusis 2000.↩︎

  5. There is different understanding of Thrace as area and population in most of the ancient authors in comparison with an AD 20th century reading. However, in many of the modern interpretations, the study area belongs to ancient Thrace.↩︎

  6. Hebros is the classical name for the river Maritsa↩︎