Alluvial landscapes in the temperate Balkan Neolithic: transitions to tells. (Notes & News).

Introduction

Relationships among cultivation patterns, settlement location and hydrology are fundamental to the Neolithic. Recent research has expanded Childe's original explanations for transitions to food production based on water location and availability (Childe 1936). Critical are river-system geomorphology and floodplain dynamics. We report on our Romanian fieldwork and its consequences for understanding the 5th-millennium BC shift leading to temperate Europe's first tells.

Context

Sherratt highlighted interplays among cultivation, settlement and hydrology (Sherratt 1980). Early cultivators exploited low-lying, wet, open areas, and practised small-scale, intensive cultivation. Labour investment was low; seasonal flooding replenished soils that supported wide ranges of plants and animals. Cultivation and settlement were spread out and spatially restricted. Later cereal cultivation was more extensive, in drier areas away from water-courses, and relied on rain-fall to feed crops. Labour was more intensive, requiring tree-clearance (first with polished stone and, later, copper axes). New species (e.g. winter-wheats) and increased dependence on barley secured cultivation success; new exploitations of livestock, especially cattle, provided novel labour sources. Sherratt provided a model for cultivation, and suggested social and economic consequences for varying planting regimes, soil-types and water provision. Water-source was critical. However, the model was general, not differentiating between regions nor focusing on specific sites. Complexities of individual river-system hydrologies drew little attention.

Three subsequent papers provide refinement. Halstead (1989) documented differences in climate, vegetation, precipitation, seasonal acuity and economies, to distinguish the northern, temperate, from the southern, Mediterranean, Balkans. Northern subsistence preferred variation, mobility and dispersal of temporary settlement; tell villages do not appear until the 5th millennium BC. To the south, predictability conditioned longer-term habitation; tells appear from 6500 BC. Van Andel (van Andel et al. 1995) detailed river-history, alluviation and settlement in the Trikala basin (Thessaly), showing that early emmer farmers established permanent villages in floodplains and exploited regular floods that freshened silts for cultivation. Davidson (1986) studied site location and geomorphology in the Drama Plain (Macedonia) and demonstrated that, here, Neolithic tells sat both on top of older alluvium and in areas of flooding.

Halstead inserted a north/south divide into Sherratt's model; Van Andel and Davidson's work confirmed Sherratt's conclusions for the south and documented geomorphologies for particular settlements. Our fieldwork in Romania supports and refines Halstead's north/south distinction. We report on the consequences for understanding 5th-millennium BC land-use change and the emergence of tells along lower Danube tributaries and for shaping further work.

Objectives

The Southern Romania Archaeological Project (SRAP) is investigating Neolithic land-use in the Teleorman River (FIGURE 1). Fieldwork follows Howard and Macklin's integrated archaeogeomorphology (Howard & Macklin 1999). A major objective is to understand the middle--late Neolithic shift (from 4500 BC) to permanent tell villages, a transition occurring across the temperate Balkans at the beginning of the 5th millennium BC. Why did people abandon a life-style of short-lived occupations of pit-huts and simple, surface-structures in order to establish permanent villages? Existing explanations are limited to needs for defensive fortification (Morintz 1962) or, unsatisfactorily, interpret change within a culture-historical approach. A second SRAP objective is to understand tell abandonment at the end of the late Neolithic (from 4000 BC)? Again, existing explanations are limited; Gimbutas' (1977) invasion hypotheses linger unhelpfully.

[FIGURE 1 OMITTED]

Answers to these questions rest in the geomorphologies and archaeologies of a preserved alluvial landscape (see recent work in west European alluvial landscapes (Howard & Macklin 1999; Macklin 1999 and references therein)). SRAP research focuses on fluvial geomorphologies of the early and middle 5th-millennium BC Teleorman Valley; i.e. the time of temporary habitation and of the shift to tells.

Archaeo-geomorphologic contexts

Typical of lower Danube tributaries in southern Romania (FIGURE 1), the Teleorman rises in the Carpathians and flows south, joining the Danube west of Giurghiu. The oldest deposits are Permian-Triassic and Jurassic-Cretaceous and are covered by Pleistocene sands, gravels and marls derived from erosion of the uplifting Carpathians (Neumann & Haita 1999: 6-7). These deposits were subsequently blanketed by Loess. Holocene incision and aggradation resulted in valley floor infilling of 2-4 m of coarse gravels overlain by 3-10 m of silts.

The valley contains well-preserved prehistoric landscapes (Spiru 1996); the Neolithic record is typical for the temperate Balkans (Bailey 2000). After 6500 BC, people built new social environments and manipulated local and exotic species in novel ways. Earliest local Neolithic (Cris Culture) sites sit on low terraces overlooking rivers (e.g. Dulceanca on the Burdea River (Comsa 1994) and Magura). Dudesti Culture sites appear in the last quarter of the 6th millennium BC and sit on the higher river terraces. With the 5th millennium BC, Boian Culture sites appear, again, on the higher terraces (e.g. Alexandria--Bucur & Preda 1959). Cris, Dudesti and Boian sites consist of pit-huts or surface-level structures made of clay, mud and wood. Floor-plans are simple, building arrangement is haphazard, occupations short-term. Cereals were cultivated, although not intensively, wild animals were hunted; cattle, pig, sheep, goat and dog were bred. The important distinction of these three Neolithic phenomena from the Gumelnita Culture (from 4700/ 4600 BC) is the permanence of Gumelnita settlement: repeated building and rebuilding of timber-framed wattle-and-daub structures created monumental tells. (At Tangaru (Berciu 1961: 363-496), Vidra (Rosetti 1934) and Cascioarele (Dumitrescu 1986), the lowest levels of Gumelnita tells have traces of Boian occupation.) Gumelnita ends by 4000/3900 BC; the immediate post-Neolithic record (Cernavoda, Celei and Cotofeni cultures) is less clear. An early Bronze Age is present by the mid 4th millennium BC; settlements are fewer, located on high terraces with attention focusing on burial monuments. A more mobile existence may well have prevailed.

The Laceni-Magura reach

Before reaching the Danube, the Teleorman runs between the Laceni and Magura villages. Sporadically, across the valley floor, Boian ceramics erode from the banks of disused drainage channels, providing an exceptional view of material and settlement within alluvial deposits of a Neolithic landscape just prior to the shift to tells. The Laceni-Magura reach also contains a Gumelnita tell (Magura); a second tell is 2 km north (Laceni) and a third 6 km south (Vitanesti). Laceni-Magura, therefore, allows the reconstruction of fluvial conditions for the shift to tell settlement. SRAP has documented patterns of land-use for the early and mid 5th-millennium BC at Laceni-Magura and of tell occupation during the mid to late 5th millennium BC at Vitanesti, Laceni and Magura.

Valley-floor investigations

Fieldwalking, sondage excavation, and channel-bank cleaning uncovered six sites dating to the Giulesti and Spantov phases of Boian (FIGURE 2). An animal bone from a Spantov context at Teleor 008 dates to 4810-4680 cal BC (Beta-148762). These sites are important as Spantov is the final Boian sub-phase and thus provides the critical, final pre-tell context. Before 4500 BC, occupation was short-term. Structures were ephemeral, leaving only disarticulated building materials (burnt and unburnt clay and mud daub). Sediment micromorphology from Teleor 001 revealed a floor sequence within an otherwise poorly defined structure (Haita 2001). Fauna from the early 5th-millennium BC sites (Teleor 008 and 009) (Balasescu 2001) and macro-flora from Teleor 008 (Bogaard 2001) document the Neolithic range of plants and animals: wheats, barleys and legumes; sheep/goat, cattle and pig. Domestic cattle predominate. Though samples are small, the ranges of species from the two different phases of Teleor 008 are distinct. More than twice the species are represented in the earlier Giulesti phase as in the succeeding Spantov phase. Furthermore, more than half the earlier-phase species are wild (horse, cattle, roe deer, fox, marten, otter, cat and beaver); based on MNIs, less than 15% of the later phase species are wild. Patterns are preliminary, resulting from limited excavation. At the very least, they suggest variation in activities during sequential phases of occupation. Durations of site-use also vary. Ceramic styles and fabrics and the building material solidity and technique suggest that Teleor 010 was a seasonal habitation and Teleor 008 a later, longer occupation.

[FIGURE 2 OMITTED]

Valley-floor and river history

The Teleorman has incised through c. 20 m of Pleistocene and earlier materials to form a Holocene valley floor c. 2 km wide. The modern river-channel is slightly sinuous, single-thread, and inset up to 5 m below the majority of the valley floor. Beyond the active channel zone (c. 50 m wide), three older river terrace surfaces have been mapped across the low-relief floodplain surface (range between 1 and 2 m). Along the southwestern valley side, the higher terraces are overlain by alluvial fan cones at the bases of former gullies. Between valleyside and active channel, low-relief sandy river bars, islands and levees are dissected by sinuous palaeochannels (Howard & Macklin 2001). Importantly, the sandy bars, islands and levees match the Boian sites (FIGURE 2). Activity and habitation in the first half of the 5th millennium BC were based on top of these elevated features.

Fluvial sediments, seen in natural, cut-bank exposures of the Teleorman, comprise basal coarse-grained sands and gravels, up to 3 m thick, overlain by up to 2 m of fine-grained silts and clays. The basal unit has a well-developed internal structure comprising low-angle, lateral accretionary surfaces typical of deposition as bar-complexes within a laterally mobile river. The overlying silts and clays were deposited by overbank alluviation and include darker iron-and clay-rich units that can be traced as palaeosols along the bank sections (Neumann & Haita 1999). Across the valley floor, the fine-grained sediments are indurated by carbonate cements, possibly precipitated during the fluctuation of valley-floor water-tables. Radiocarbon dates from the river-banks and the AMS date from Teleor 008 allow a preliminary chronology of development for the Laceni-Magura (TABLE 1): further dating will refine these episodes and expand the river history into and beyond the 5th millennium BC.

Early 5th-millennium BC Laceni-Magura: potential scenarios

Scenario 1. The river ran along a static, single-channel. The valley bottom was a stable, predictable resource base. Locations, densities and ranges of plant, animal and mineral resources were steady, and enabled stability in people's interactions with the valley. A stable single-channel river, providing non-disturbed lands running down to the river would have allowed long-term agricultural commitment to the land.

Scenario 2. The river ran along one channel, the position of which moved across the valley bottom. Valley-floor instability limited peoples' abilities to predict resource locations and densities. A very mobile channel would have created an extremely unstable and unpredictable environment; less frequent channel mobility would have conditioned a less unstable landscape. In any case, instability prevented long-term permanence.

Scenario 3. The river was a multi-thread, anastomosing system with mostly static channels. Stability, predictability and permanence applied. People established long-lasting relationships with the landscape. In this aquatic, perhaps marsh-like, environment the density and, perhaps, variety of plants and animals were high.

Scenario 4. The river was multi-channel and mobile. High densities of aquatic resources were present but the coherence and consistency of the landscape changed often creating an environment not conducive to permanent use of the landscape.

The early and middle 5th millennium BC: summary

Most probably, the river-system was unstable before the mid 5th-millennium BC shift away from temporary uses of the valley floor. The coarse sands and gravels and absence of fine-grained sediment suggest that the river moved across the valley-bottom and that channel(s) moved with frequency (Scenarios 2 or 4). After the shift to tells, the river system probably reflects a landscape with a stable single or multichannel river (Scenarios 1 or 3). If tells were foci for increasingly standardized agriculture (Bailey 2000), then these stable landscapes would have supported long-term cereal cultivation.

Results of settlement-tell investigations

Trenching of the later 5th-millennium BC Laceni and Vitanesti tells and auguring of the latter site inform on their geneses and abandonments. Vitanesti sits on the eastern edge of the valley. Sediments and valley-floor topography resemble those of the central valley floor during the earlier occupations: basal gravel units overlain by interdigitating units of medium-to-fine sands, organic-rich silts and grey-silty clays, all sealed beneath an upper black clay (Neumann & Haita 1999: 11). Earliest settlement at Vitanesti was on a gravel bar. Grey-silty clays overlying the bar-sediments, surrounding the village, suggest that fluvial energy diminished and marshland surrounded the site. As Vitanesti sat at the edge of the valley floor, it is unclear if the reduced fluvial energy reflects climatic change or if channel movement made the area a backwater. Vitanesti dates to the middle and second half of the 5th millennium BC (Gumelnita A2 through B2). Material recovered from the trenches at Tell Laceni is contemporaneous.

Vitanesti and Laceni were abandoned by the end of the 5th millennium BC. Comparison of macro-flora from Teleor 008 and Vitanesti suggests that conditions deteriorated, becoming unfavourable for cultivation (Bogaard 2001). Appearance of rye (Secale cereale) at Vitanesti and its importance as a cultigen at contemporary tells (Carciumaru 1996) suggest that the end of the 5th millennium BC witnessed poorer growing conditions and decreased crop husbandry. If a primary activity at lower Danube tells was cereal agriculture then a deteriorating environment may have led to tell redundancy. Significant episodes of enhanced fluvial activity recorded in Teleorman river-sediments suggest that after the abandonment of Vitanesti (i.e. the mid 4th millennium BC) the frequency of large floods increased significantly, making the landscape less suited to settled agriculture. Work at Podgoritsa (Bulgaria) suggests that towards the end of that tell's life, water-table vacillations rendered the arable landscape unusable (Bailey et al. 1998). Similar conditions may have developed in the Teleorman.

Conclusions

Multi-disciplinary work in an alluvial landscape can expand understanding of transitions in Neolithic land-use in the lower Danube. Future research will extend and refine the following conclusions for 5th-millennium BC Laceni-Magura.

In the first half of the 5th millennium BC, people grazed and bred animals, lived and slept, ate and drank at sites on the valley floor. Similar activities occured on the river's terraces, where rain-fed, small-scale cultivation may have taken place. Apart from ceramic style and decoration, little changed over 1500 years. Boian occupation and land-use were temporary, probably due to an unstable, mobile river-system. This landscape was well-suited for grazing animals, hunting and fishing, gathering plants and other materials (for thatching, building and potting). Cultivation in the valley bottom was small-scale with little labour-investment to clear land. This landscape did not support permanent settlement. Short-term habitations (e.g. Teleor 001, 008, 009, 010, 011) utilized small sand-bars, alongside river channels. Occupation duration and location changed with river position and character. During the first half of the 5th millennium BC, variation and instability define Laceni-Magura.

The shift to tells

In the mid 5th millennium BC, people lived more stable existences on gravel-bar islands at the eastern edges of the valley (e.g. Vitanesti) of no greater permanence, at first, than the habitations out in the valley bottom. Eventually, people settled down in particular places at the valley edge, perhaps where floods were less destructive. At Vitanesti, Magura and Laceni, permanence created tells. Occupation of gravel-bars suggests that people still wanted to live in the valley bottoms but up and away from significant floodings or waterlogging-associated vacillation of the water-table.

Significantly, Vitanesti contains evidence for large-scale cereal cultivation and storage. Tell-location and permanence may be due to changes in the character of the valley floor: the river stabilized into one or several channels. As Sherratt predicted, flooding in this system fertilized valley-bottom soils, facilitating more intensive cultivation. We suggest that increased agricultural activities resulted from exploitation of newly stabilized, fertile valley bottoms; hence tell-location down on valley floors,

but on low gravel-bars at the valley edges.

Land-use models in the Balkan Neolithic

Halstead's Mediterranean/temperate distinction holds; with the detailed work of Davidson, van Andel and that reported here, this distinction can be refined. To the south, in Thessaly, early settlement took advantage of predictable, stable river-systems that early cultivators exploited with new resources and technologies; early agricultural tells appear early (from 6500 BC) in active floodplains. At the same time, in the temperate Balkans, life was less predictable; less stable riverine environments were not conducive to large-scale cultivation. To the north, floodplain exploitation for cultivation (and the related development of tells) only began once alluvial landscapes became more stable. At Laceni-Magura, this is marked by the beginnings of the tells at Vitanesti, Magura and Laceni and the abandonment of temporary valley-bottom occupations.

Sherratt's model of water source, settlement location and cultivation strategies still holds if we refine it with the results of subsequent fieldwork. Most importantly, the relationships are less straightforward than originally proposed. Prehistoric fluvial dynamics have consequences that go beyond the provision of water for plants or nutrients for soils; changes in landscape stability and in river system (e.g. multi-channel versus single-channel) condition human behaviour in complex and vacillating ways.

Most importantly, variation is the rule. If trans-regional models are of use, then local detail must qualify them. Variation within regions, within river valleys and, even, within individual river-reaches argues against any grand generalizing. Results from Laceni-Magura reveal preliminary parameters about 5th-millennium BC life within the specifics of one landscape. More similarly detailed work will reveal whether the patterns at Laceni-Magura have relevance, beyond this particular reach, beyond this river, in other valleys in southern Romania or, indeed, in other parts of the temperate Balkans.

Acknowledgements. Thanks to the Society of Antiquaries, British Academy, Cardiff University, National Historical Museum of Romania, Romanian Ministry of Culture and Teleorman Regional Historical Museum, Alasdair Whittle, Dragomir Popovici, Silvia Marinescu-Bilcu, Mihai Tomescu, Heike Neumann, Costel Haita, Amy Bogaard, Lois Wright, Madame Tantareanu and Christi Mirea.

TABLE 1. Absolute dates and related river activity. (Calibration
database used: Stuiver et al. 1998.)

lab. code     sample context                       material

Historic period
Beta-147288   fine-grain channel fill              wood

Iron Age
AA-38910      fine-grained drape unit              wood
AA-38911      fine-grained drape unit              wood
AA-38912      fine-grained drape unit              wood
Beta-147290   bar core-sands and gravels           bone
Beta-147289   bar core-sands and gravels           wood

end of Early Bronze Age
Beta-147291   bar core-sands and gravels           wood

aftermath of Gumelnita tell villages
Beta-147292   bar core-sands and gravels           wood
Beta-147293   bar core-sands and gravels           wood

semi-mobile Neolithic communities (Boian Culture)
Beta-148762   structure/activity area Teleor 008   bone

lab. code        convent BP          date (cal 2[sigma])

Historic period
Beta-147288   1050 [+ or -] 60     AD 880-1050/1100-1140

Iron Age
AA-38910      3100 [+ or -] 45              1488-1223 BC
AA-38911      3125 [+ or -] 40              1494-1265 BC
AA-38912      3075 [+ or -] 70              1500-1128 BC
Beta-147290    190 [+ or -] 60    1530-1550/1630-1950 BC
Beta-147289   3360 [+ or -] 70    1870-1840/1780-1500 BC

end of Early Bronze Age
Beta-147291   3800 [+ or -] 60              2450-2040 BC

aftermath of Gumelnita tell villages
Beta-147292   4980 [+ or -] 80              3960-3640 BC
Beta-147293   4820 [+ or -] 60   3700-3510/3420--3390 BC

semi-mobile Neolithic communities (Boian Culture)
Beta-148762   5790 [+ or -] 40              4810-4680 BC

References

BAILEY, D.W. 2000. Balkan prehistory: exclusion, incorporation and identity. London: Routledge.

BAILEY, D.W., R. ANDREESCU & S. MILLS (ed.) 1999. Southern Romania Archaeological Project: Preliminary report 1998. Cardiff: HISAR, Cardiff University.

BAILEY, D.W., R. ANDREESCU, S. MILLS & S. TRICK (ed.). 2001. Southern Romania Archaeological Project: Second preliminary report. Cardiff: HISAR, Cardiff University.

BAILEY, D.W., R.E. TRINGHAM, J. BASS, M. HAMILTON, H. NEUMANN, M. STEVANOVIC, I. ANGELOVA & A. RADUNCHEVA. 1998. Expanding the dimensions of early agricultural tells: the Podgoritsa Archaeological Project, Bulgaria, Journal of Field Archaeology 25(4): 373-96.

BALASESCU, A. 2001. Preliminary report on animal bones (the 2000 season), in Bailey et al.: 65-78.

BERCIU, D. 1961. Contributii la problemele Neoliticului in Romania in lumina noilor cercetari. Bucuresti: Academiei Republicii Populare Romine.

BOGAARD, A. 2001. Recovery of macro-faunal remains: technique and results, in Bailey et al.:79-89.

BUCUR, M. & C. PREDA. 1959. Sapaturile de salvare de la Alexandria (reg. Alexandria), Materiale si Cercetari Arheologice 5: 175-6.

CARCIUMARU, M. 1996. Paleoetnobotanica. Studii in preistori si protoistoria Romaniei. Iasi: Glasul Bucovinei Helios.

CHILDE, V.G. 1936. Man makes himself. London: Watts.

COMSA, E. 1994. Asezarea Starcevo-Cris de la Dulceanca, Analele Banatului 3: 14-39.

DAVIDSON, D.A. 1986. Geomorphological studies, in A.C. Renfrew, M. Gimbutas & E. Elster (ed.), Excavations at Sitagroi 1: 25-40. Los Angeles (CA): UCLA.

DUMITRESCU, V. 1986. Stratigrafia asezarii-tell de pe Ostrovelul de la Cascioarele, Culturasi Civilizatie la Dunarea de Jos 2: 73-81.

GIMBUTAS, M. 1977. The first wave of Eurasian steppe pastoralists into Copper Age Europe, Journal of Indo-European Studies 5: 277-331.

HAITA C. 2001. Micromorphology, in Bailey et al.: 56-64.

HALSTEAD, P. 1989. Like rising damp? An ecological approach to the spread of farming in southeast and central Europe, in A. Milles, D. Williams & N. Gardner (ed.), The beginnings of agriculture: 23-53. Oxford: British Archaeological Reports. International series S496.

HOWARD, A.J. & M.G. MACKLIN. 1999. A generic geomorphological approach to archaeological interpretation and prospection in British river valleys: a guide for archaeologists investigating Holocene landscapes, Antiquity 73: 527-41.

2001. Geomorphology, in Bailey et al.: 42-55.

MACKLIN, M.G. 1999. Holocene river environments in prehistoric Britain: human interaction and impact, Journal of Quaternary Science 14: 123-45.

MORINTZ, S. 1962. As ezari si fortificat ii in cultura Gumelnita, Studii si Cercetari de Istorie Veche 2: 273-84.

NEUMANN, H. & H. HAITA. 1999. Soils and geomorphology, in Bailey et al. 1999: 5-25.

ROSETTI, D.V. 1934. Descoperiri paleolitice in preajma Bucurestilor. Sapaturile de la Vidra. Bucuresti: Muzeului Municipiului Bucuresti.

SHERRATT, A.G. 1980. Water, soil and seasonality in early cereal cultivation, World Archaeology 11:313-30.

SPIRU, I. 1996. File de istorie Teleormaneana. Alexandria.

VAN ANDEL, T.H., K. GALLIS & G. TOUFEXIS. 1995. Early Neolithic farming in a Thessalian river landscape, in J. Lewin, M.G. Macklin & J.C. Woodward (ed.), Mediterranean Quaternary river environments: 131-43. Rotterdam: Balkema.

D.W. BAILEY, R. ANDREESCU, A.J. HOWARD, M.G. MACKLIN & S. MILLS *

* Bailey & Mills, School of History and Archaeology, Cardiff University, PO Box 909, Cardiff CF10 3XU, Wales [e-mail] [e-mail] Andreescu, Muzeul National de Istorie a Romaniei, calea Victoriei nr. 12, Bucuresti, Romania. [e-mail] Howard, School of Geography, University of Leeds, Leeds LS2 9JT, England. [e-mail] Macklin, Institute of Geography & Earth Sciences, University of Wales Aberystwyth, Aberystwyth SY23 3DB, Ceredigion, Wales [e-mail]

Received 27 June 2001, accepted 28 November 2001, revised 12 March 2002