Documenta Praehistorica XL (2013) Pietrele in the Lower Danube region: integrating archaeological, faunal and environmental investigations Norbert Benecke1, Svend Hansen2, Dirk Nowacki3, Agathe Reingruber2, Kenneth Ritchie4 and Jürgen Wunderlich3 1 Scientific Department of the Head Office, German Archaeological Institute, Berlin, DE 2 Eurasia Department, German Archaeological Institute, Berlin, DE agathe.reingruber@dainst.de 3 Institute for Physical Geography, University J. W. Goethe, Frankfurt, DE 4 Center for Baltic and Scandinavian Archaeology, Schleswig, DE ABSTRACT - The c. 9m high tell-settlement of Pietrele-Magura Gorgana, situated close to the Danube river, is one of the westernmost sites of the Kodžadermen-Gumelnifa-Karanovo VI cultural complex that spread over the whole Westpontic region during the 5th millennium BC. Until recently tells were equated with the site when, in fact, they represent only the outstanding part of a far more complex settlement system as we now know from Pietrele thanks to geomagnetic prospections and subsequent excavations. People living on the tell, together with the inhabitants from the flat area around it, formed a vast community that must have had a strong impact on its habitat and, vice-versa, was strongly affected by the immediate surroundings. During the settlement period a lake covered huge parts of the floodplain. It provided not only a considerable part of the diet, but ensured, through the direct access to the main river, continuous and extensive over-regional exchange. IZVLEČEK - Okoli 9 metrov visoka naselbina tipa tell Pietrele-Mägura Gorgana se nahaja blizu reke Donave in je ena najbolj zahodnih naselbin kulturnega kompleksa Kodžadermen-Gumelnifa-Kara-novo VI, ki se je širil čez celotno zahodno pontsko regijo v času 5. tisočletja BC. V preteklosti so tell enačili z naseljem, danes pa zahvaljujoč geomagnetskim preiskavam in kasnejšim izkopavanjem na najdišču Pietrele vemo, da sama naselbina tell predstavlja le najbolj izjemen del veliko bolj kompleksnega naselja. Tisti ljudje, ki so živeli na tellu, in tisti, ki so živeli v ravnici okrog tella, so bili skupaj del velike skupnosti, ki je morala imeti močan vpliv na okolje, in na katero je tudi neposredna okolica morala imeti pomemben učinek. V času obstoja te naselbine se je v tej ravnici nahajalo veliko jezero. Le-to je predstavljajo pomemben vir za prehrano takratnih ljudi, prek neposrednega dostopa do glavne reke pa je omogočalo tudi stalno in obsežno menjavo na nadregionalni ravni. KEY WORDS - Copper Age; Gumelnifa-culture; Pietrele; geomorphology; palaeolake; fauna; hunting; fishing; symbols on pottery Introduction In archaeological investigations, bodies of flowing water are considered the main arteries by means of which groups of human populations disseminated, and with them knowledge and goods also spread. In this regard, the Danube played a special role, as first pointed out by Vere Gordon Childe (1929). He dedicated (1925) one of his first books to this water- course, although even earlier in his pioneer work The Dawn of European Civilization one chapter carries the title 'Danubian Civilization'. In the German literature, the Early Neolithic has been designated as Donauländischer Kulturkreis (Buttler 1938), the first farmers being conceived as population groups that immigrated to Southeast Europe from the Near East. Following the course of the Danube, they spread west, thereby driving out the post Ice Age hunter-gatherers of Central Europe. Standing bodies of water in Southeast Europe, unlike those in the circum-Alpine sphere, have received far less attention in research. Exceptions are the excavations led by Giorgos Chourmouziades at Lake Orestida in Kastoria, Greece (Chourmouziades 2002) and at Ezerovo, near the Lake Varna led by H. Todo-rova (Todorova, Tonceva 1975.30-46). Lakeside settlements or wetland settlements in Switzerland and south-western Germany have been the focus of investigations since 1854 (Schlichtherle 1997). Between c. 4200 and 850 calBC, Neolithic and Bronze Age communities built their houses close to the water, in some cases on posts or piles in waterlogged soils which excluded oxygen and thus remained preserved through time (Wolf 1998.27-35). In the course of the first measures to regulate the Jura from 1868 to 1891, the decline in the water level of over 2m revealed sites with so-called pile dwellings (e.g., Cortaillod, Lake Neuenburger; Schlichtherle 1997). In contrast, during land drainage measure along the Lower Danube in the 1960s, no traces of prehistoric settlements were noted, although deep channels must have been dug into the meadow ground. Yet, during the drainage of the Danube flood plains in 1961-1967, mainly already known tell settlements were investigated anew by means of soundings by Vladimir Dumitrescu (Gumelnita and Cäscioarele: Du-mitrescu 1965.215-234; Dumitrescu 1966.51-99). Additionally, Eugen Comga explored the settlement at Radovanu between 1960 and 1970 (Com§a 1972. 44-45). Although land drainage activities had a dramatic impact on the hydrological balance of the entire Lower Danube, thereby causing extensive changes to the line of adjoining terraces and intrusions into the soils, none of these consequences were noted in the archaeological literature. The Danube flood plain prior to drainage As late as the 1960s, a number of lakes extended between Giurgiu and Cälära gi, a lacustrine landscape, which was fed by the Danube and abundant tributaries to the north (Fig. 1). Some of the standing bodies of water west of the Argeg river valley were interconnected; to the west, they ended at about the site of Pietrele (at the edge of the terrace) and Go-stinu (in the flood plain). East of the Argeg and as far as the separation of the Borcea canal at Cälära §i, the lake plateau was supplied by the Mostigtea River. Once a bountiful aqueous biotope, today it is traversed by drainage canals. Excavations at some tell settlements along the Danube were conducted during or even before World War I. After beginning with a trial trench in 1904, by 1953 the tell near Ruse had been almost completely excavated (Černakov 2009.30-31); C. Schuch-hardt excavated at Cernavodäuntil 1917 (Schuch-hardt 1924.9-27); whereas, unfortunately, the work of Leo Frobenius at Cunegti during World War I remained unpublished (Popescu 1935-36.109). In the 1920s, members of the Institute of Archaeology in Bucharest undertook investigations at the tell settlements of Gumelnita (Dumitrescu 1924; 1925), Sultana (Andriesescu 1924.51-107), Cäscioarele (Stefan 1925.138-197) and the prehistoric settlements on what was then the peninsula in Lake Boian (Christescu 1925). The expeditions were immediately published in the first two volumes of the newly-founded journal Dacia in the form of detailed preliminary reports. Between 1940 and 1960, during and after World War II and before the Danube mea- Fig. 1. Map of the Lower Danube Region (U. S. Army, Corps of Engineers, 1955) with some of the main Neolithic and Chalcolithic sites (A. Reingruber). dows were drained, excavations were conducted - as far as verifiable - only in Pietrele and Span|ov; smaller investigations were undertaken in Sultana by local museums, but never published. Dumitru Berciu excavated in Pietrele for a maximum of six weeks in 1943 and 1948 (Berciu 1956.504); after that, in 1956-1957 he was occupied with work in Tangaru, another tell settlement northwest of Pietrele (Berciu 1961). Sebastian Morintz excavated in Spantov in 1952 and 1956 (Santierul Spanfov 1953; Morintz, Preda 1959.163). The aforementioned settlements of the Neolithic Boian and Copper Age Gumelnita cultures appear at regular intervals, at a distance of approx. 20-40km apart. Multi-layered or flat settlements, scarcely visible today, can surely be presumed in the terrain in between. Those sites that have been recognised and excavated, however, suffered the same fate as the tells: the finds recovered from Prundu, Chiselet, Bo§-neagu, Värä§ti and Alexandru Odobescu remain more or less unpublished. When plotting these Neolithic and Eneolithic settlements on a map, their relation to the former lake landscape becomes quite obvious; investigated settlements are particularly numerous in areas in which flowing water meets standing water (near the present-day localities Oltenita und Dorobantu). The absence of higher tells to the south of the present course of the Danube cannot be attributed solely to a gap in research. No larger rivers flow there, except for the Lom, at whose confluence with the Danube the Ruse tell is located (Fig. 2). Moreover, although numerous rivulets drain the Ludgorie plateau east of the Lom, they do not carry water all year round. So, apparently this southern area - with a more limited amount of not only water but also sun in comparison to the northern banks - did not attract prehistoric settlers as the north did. Similarly, no tell-settlements are known from the Neolithic or Copper Age in the area extending from the bend east of Cunesti near Cälära§i as far as Cernavodä. By contrast, high tells are located along the Da- nube flowing south-north, such as Cernavodä(at the mouth of the Carasu River into the Danube, that is, with the Danube-Black Sea canal built into the riverbed, 1949-1987), Bordusani and Harsova (near the confluence with the Ialomita). Settlements are even more numerous in the area around the northern knee of the Danube, where smaller lakes are found to the south of the river and large lakes to the north (Fig. 2). Undoubtedly, this ecological situation cannot be applied directly to the 5th millennium BC. Likewise, the morphology of the Danube meadowscape between Giurgiu and Cäalaärasi, much less as far as the Danube delta, cannot be reconstructed within the time limits of a single project. However, based on more than 160 core drillings in the stretch between the tell sites of Pietrele and Gumelnita, new results regarding the pa-laeo-landscape of the Danube valley were obtained. Reconstruction of a palaeolake in the Lower Danube valley Until now, studies of the fluvial history of the Lower Danube and of landscape development along that stretch of the Danube are rare. An overall view of the geomorphological and geological setting within the study area was produced by Institutul de Geologie si Geografie al Academiei Republicii Socialiste Romania (1969). More detailed geoarchaeological research within the framework of the joint Romanian-British Southern Romanian Archaeological Pro- Fig. 2. Map of the Lower Danube Region with sites of the Gumelnifa-Kodzadermen culture (A. Reingruber). Fig. 3. Topographic map of the study area at the Lower Danube showing the locations of corings, positions of former lakes and the supposed palaeolake with an area of approx. 60km that might have developed in the 7th millennium BC and existed for more than 5000 years (D. Nowacki). ject (SRAP) project (Howard et al. 2004; Macklin et al. 2011) focused on the Teleorman Valley, about 70km west of Pietrele. Furthermore, Alexandra M. F. Tomescu (2000) and Maria Lazarova and Elissaveta Bozilova (2001) provided information on the regional vegetation history based on palynological studies in southern Romania and northern Bulgaria. Much more research has been conducted on the Holocene and late Pleistocene evolution of the Danube delta and changes in the level of the Black Sea (e.g., Pa-nin 2003; Giosan et al. 2006; Yanko-Hombach et al. 2007). A discussion of the chronology of sea-level changes and the impact on the archaeological sites along the Black Sea coast is still underway. To have a closer view of the development of the Ho-locene flood plain within the study area between Giurgiu and Oltenita, a multi-proxy approach based on a variety of methods (e.g., the evaluation of historical and recent topographic maps and satellite images, corings as much as 17m in depth, geoelectric profiling, sedimentological, geochemical, microfaunal and pollen analyses) have been applied (Wunderlich et al. 2012; Nowacki, Wunderlich 2012). A comparison of historical topographic maps and satellite images allowed the detection of recent geo-morphological changes, as natural or anthropogenic modifications to channel beds and the silting up of ancient lakes. Furthermore, these sources provided the geographical background for taking sediment cores, and enabled the regionalisation of data attained from various analyses of the sediments. More than 160 sediment cores were taken by vibra-coring (Fig. 3), using open and closed sections. In addition to grain-size analyses, chemical elements such as iron, aluminium, titanium, strontium, magnesium, calcium, copper, manganese, sulphur, nitrogen, phosphate, total organic and inorganic carbon (TOC, TIC) were identified. Furthermore, conductivity was measured and the remains of microfossils and pollen analysed. The chronological framework of the stratigraphy is based on AMS 14C and OSL dating. The corings and following analyses allowed the identification of distinct sediment layers, which can be assigned to different sedimentary environments. The time bar in Figure 4 shows the different litholo- gical units, starting with fluvial sediments at the base, overlain by limnic sediments, which are mostly covered by channel and flood plain deposits. In detail, the following facies units from base to top could be distinguished: Fluvial deposits These sediments are characterised by olive-coloured sand and partly gravel. OSL-dates vary from 32.7 ± 1.5 ka BP (HDS-1570) to 15.9 ± 0.9 ka BP (HDS-1573) indicating that the deposition of this layer occurred at least until the late Pleistocene. It can be attributed to a braided river system. Limnic deposits - grey These grey deposits are characterised by a fining upward sequence, but with slightly alternating grain-size distribution in every analysed core. The only OSL date available from the lower part of this layer gives an age of 9.03 ± 0.4 ka BP (HDS-1574; cf. Fig. 4). Due to the results of geochemical and microfossil analyses - for example, comparatively low C/N ratios (e.g., Meyers, Ishiwatari 1993) - these sediments can be attributed to a limnic environment. It can be assumed that vast parts of the Lower Danube valley were inundated as a result of the rapid rise in the level of the Black Sea. The extent of the palaeo-lake as reconstructed from the corings is depicted in Figure 3. After a period of 2000 years, the environmental conditions changed and the following sequence was deposited. This was when the Neolithic settlement was founded at Pietrele. Limnic deposits - very dark grey/black A limnic environment continued to prevail in the study area. However, the sediments of this sequence are characterised by a very dark grey or even black colour. This dark layer (DL I), which could be identified in nearly every core, was dated by several AMS-14C and one OSL dating to the time span of the Copper Age settlement period (Fig. 4). Near the centre of the palaeolake, Dl I is predominantly characterised by clay, whereas next to the littoral, the layer mainly consists of coarser material. This is evidence that the sediments originated from the adjacent slopes, where they were possibly eroded due to increasing human impact. Limnic deposits - dark grey The lake sediments above DL I are characterised by a dark greyish colour and consist predominantly of clay and silt, respectively. Further dark layers are intercalated, varying in number and thickness from core to core. For example, the dark layer DL II is not age stratigraphic units I960 - floodplain drainage T aggradation channel facies * 1290- 1630 col AD (Piet 73 / KIA-41258 toKIA-41261) Danube river denudation I youngest deposited material 255 -395 cat AD (Piet 80/KIA-41264) estimated age of DL III based on palynoiogy data 800 till 300 BC . dating DL II I 2129- 19i4calBC(Piet IO/KIA-31592) * 2135- 1956 cal BC (Piet 41 /KIA-38744) dating DL I 4221 -3979 cal BC (Piet 10/KIA-31593) 4688-4271 cal BC (Piet 63/KIA-41256) dating (OSL) 6880 BP +/- 370 (Piet 83Ö/HDS-1575) dating (OSL) 9030 BP+/-410 (Piet 83b/HDS-1574) maximum age range of available dating settlement period (fiat area) settlement period (tell) alluvial deposits/anhropogenically reworked aggradation channel deposits limnic deposits - dark grey limnic deposits - very dark grey (dark layer) limnic deposits - grey fluvial deposits (pre paEaeolake) facies change BC Fig. 4. Time bar showing the different stratigraphic units in the sedimentary record of the Danube floodplain derivedfrom more than 160 corings, as well as AMS 14C and OSL ages providing the chronological framework (D. Nowacki). Fig. 5. The tell site of Pietrele, Mägura Gorgana and its immediate surroundings with trenches from 2002-2012 (D. Nowacki). as thick as DL I and occurs only in the western part of the palaeolake. This layer was dated to approximately 2000 calBC (Fig. 4). The whole stratigraphic layer is interpreted as the main palaeolake period, which ended with the silting up of the lake. Aggradation channel deposits The period of aggradation was characterised by a prograding channel system, which fragmented the whole lake area into smaller basins. This channel system and the small lakes in between existed until the flood plain was drained in the 1960s; the remnants thereof can still be identified in the landscape on recent satellite images and topographic maps. The channel and levee deposits of the 'aggradation channel deposits' are separated from the 'limnic deposits' below by a sharp unconformity, indicating that the limnic sediments were locally eroded during the aggradation phase. In the small basins between the channels, lake sediments reach the surface. At these locations, lakes prevailed, since they were drained some 50 years ago. Alluvial deposits/anthropogenically reworked The uppermost stratigraphic layer close to the surface contains reworked material that came up dur- ing the digging of drainage canals. Furthermore, it consists of alluvial sediments that were affected by pedogenesis after drainage and severe changes in the groundwater table. This stratigraphic layer alternates in grain-size and colour from core to core. The results presented here of our geomorphological research in the area close to the site provide evidence that during the Neolithic and Copper Age settlement at Pietrele people lived close to a vast lake extending over about 500km2. The western boundary of the palaeolake could be located about 10km west of Pietrele. There, the Danube probably entered the lake. To the east, limnic deposits were found as far as Gumelnita, defining the minimum extent of the lake, and even in Bulgaria south of the Danube comparable lake deposits were found. Possibly, the lake extended even further to the east, connecting the settlements of the 5th millennium BC. Along the lakeshore, swampy wetlands constituted suitable habitats for wild animals such as wild boar, which was proven to have been a major source of food for the people at tell Pietrele (see contribution by N. Benecke). The finds of fish bones within the settlement layers also corroborate the existence of a lake, as the majority could be assigned to species preferring still waters (see contribution by K. Ritchie). The changing properties of the sediments within the record of limnic deposit, for example, the dark layer DL I that was partly accumulated during the main settlement period at Pietrele and neighbouring settlements from 4600 to 4250 calBC, might have been due to human impact. Intensified agriculture causing soil erosion as well as the input of organic matter from the settlements around the lake possibly changed the ecological conditions of the palaeolake. Pietrele - 1000 years of settlement history Still towering some 9m above the surroundings today, the Magura Gorgana settlement mound near Pie-trele was without doubt an imposing feature in the 5th millennium BC (Fig. 5). Upon a relatively steep projecting socle of settlement debris, heightened with additional deposits, stood large two-storied structures in close alignment. Settlement mounds were an architectural innovation on the Lower Danube, for never before had such a form of settlement appeared in this region, an otherwise characteristic appearance during the late 7th and 6th millennium BC in Southeast Europe. However, in the middle of the 5th millennium bC, when the Mägura Gorgana tell began to be built, settlement mounds were already an anachronism: this form had been given up elsewhere in Southeast Europe. According to the available data, it seems presumable that the erection of this tell was an organised process that occurred prior to 4600 calBC on the Lower Danube. However, we can establish the beginnings of the settlement hill in Pietrele with certainty only when the corresponding settlement layers have been reached. It has been a long-held view that the mound represented the settlement as a whole. Our investigations could show for the first time that the tell was only a part of a substantially larger settlement. According to findings made thus far, the surrounding flat-land settlement existed long before the mound was built: it existed further during the entire duration of habitation atop the mound and ended c. 4250 calBC with the end of the tell. Excavations conducted in the flatland settlement since 2009 have brought forth astonishing new per- spectives. The hitherto oldest recorded habitation there can be dated to the last two centuries of the 6th millennium BC (Hansen et al. in press.Fig. 59). In the north-eastern area of the flatland settlement, habitation layers were uncovered that held pottery from an early phase of the southern Romanian Late Neolithic (Hansen et al. in press.Fig. 2-3, 28). Radiocarbon dates of the layers from which the pottery derives are still being processed, but stylistically the vessels can be dated to the beginning of the 5th millennium BC. At present, no further radiocarbon dates for the various Late Neolithic find contexts are available, yet all stylistic phases of Boian culture are probably represented in the flat settlement around the tell. The research on the Late Neolithic period in southern Romania is full of gaps (e.g., Neagu 1999); hence, it can be anticipated that future excavation in Pietrele will lead to a basic revision of the actual temporal sequence in the cultural development of the Late Neolithic period. There are signs that the settlement existed for some 1000 years, which is a significantly longer time than could have been assumed until now. With the discovery of Late Neolithic layers in Pietrele, there is now the promising expectation that the development of the settlement until the forma- species Danube meadow floodplain forest slope area water bodies high terrace mammals wild pig beaver roe deer aurochs red deer otter wolf wild horse polecat fox badger hare red deer roe deer bear lynx wild cat wolf pine marten hare birds ducks geese swans herons, storks birds of prey birds of prey cormorant pelican crane fish all species reptiles pond turtle mollusca river mussels Fig. 6. The ecotope in the surroundings of the settlement at Magu-ra Gorgana and the wild animals occurring inside as derived from their preference for a specific habitat. tion of the settlement mound can be followed more precisely, a potential that is of profound importance for explaining the genesis of the 'tell phenomenon' on the Lower Danube. Furthermore, the economic strategies and their changes in the course of a lengthy settlement history can be determined, thanks to abundant plant and animal remains obtained by hand-collecting as well as by dry sieving and flotation. Animal husbandry and hunting Excavations at the settlement mound of Magura Gor-gana in the community of Pietrele have yielded comprehensive collections of animal remains, thanks to the favourable conditions for preservation. The remains enable insight into the economic basis of the settlement site, as well as aspects of local environmental conditions (Benecke 2004; 2006). From a chronological point of view, the finds can be roughly assigned to two phases in time, namely the Late Neolithic Boian culture and the Copper Age Gumelnitta culture. Assemblages of animal bones from the time of the Boian settlement phase are available from the outer settlement, although the size of these collections is relatively small. By contrast, excavations on the tell yielded exclusively animal remains of the Gumelnitta settlement phase, encompassing very comprehensive inventories. A comparison of finds from these two temporal phases reveals differences in the use or management of animal resources. The Boian settlement in Pietrele is typical mainly of an agrarian oriented food economy. Animal husbandry focused primarily on cattle. Thus, among the bones of economic domestic animals, c. 70% are cattle, while pig and small ruminants (sheep and goat) are each represented by c. 15%. The small inventories do not enable any deductions about aspects of secondary use. Compared to animal husbandry, hunting played only a minor role in the food supply. The proportion of game among the bones of mammals amounts to only 8%. Wild boar and red deer were among the most frequently hunted wild animals. The number of finds of other, potentially easily obtainable, natural animal resources in the surroundings of the settlement, such as birds, fish, mussels and turtles, suggests that exploitation was very low. Animal remains from the subsequent Gumelnitta settlement attest to the extensive use of natural animal resources as nourishment for the inhabitants and a source of raw materials. Only half of the meat documented in waste deposits derives from domesticated Fig. 7. Relative frequency of the principal game species (ungulates only) from the Gumelnipa layers in the tell (according to total number of finds). animals. In contrast with the Boian settlement, pig is the most frequent among domestic animals, amounting to 56%. The cattle bones amount to 33% and those of sheep/goat are 11% of the bone material. Available data on stock composition (age, sex) show that pig were kept mainly for meat, while cattle, sheep and goat were probably also raised for milk. Wild mammals comprise a numerically significant group, rich in species, among the finds of the Gumelnitta settlement. They include various ungulates such as wild boar, red deer, roe deer, aurochs and wild horse, predators, and beaver and hare. They comprise between 45% and 55% of the mammalian bone material a proportion that allows the conclusion that hunting was widely practiced. Wild boar is by far the predominant species among game animals (Fig. 6). The surrounding deciduous and mixed forests, alder swamps and aggradation zones of lakes and old coniferous groves with abundant underbrush offered an optimal environment for this species (Herre 1986.51). Considering this beneficial ecotope in the environs of the settlement site, animals probably took advantage of these ideal conditions, especially in the Danube meadows, and, consequently, were hunted. The second most frequent game species is red deer. This species is a typical inhabitant of sparse woods and wooded river valleys and meadows. However, it is also found in open grassland and heather landscapes if they are expansive enough to offer sufficient possibilities of escape (Bützler 1986.125). Accordingly, red deer could be hunted in the surrounding Danube meadows as well as on the high terraces near Magura Gor-gana. Unlike red deer, roe deer is characterised rather by its preference for the periphery of woods (Lehmann, Sägesser 1986.254), where it can seek refuge. However, it also prefers vast open spaces, a landscape that can be presumed to have existed on high terraces. Roe deer could also appear occasionally in slope areas. The presence of wild horse is relatively clear. As specialised grass eaters and a typical element of the steppe fauna, these animals are primarily assumed to have been inhabitants of high terraces along the Danube. Presumably, sparse wooded cover and park-like landscapes and the steppe were the preferred habitat of auroch (Requate 1957. 325). Groups of these large wild cattle probably also inhabited the high terraces. Figure 6 shows potential habitats, including areas of other species of wild game attested in the bone material from Mägura Go-rgana. In addition to wild mammals, bone material found in the Gumelnita settlement phase attests the hunting of various species of birds as well, above all ducks, geese, swans, herons and cranes. Fishing was another important branch of the food economy (see contribution by K. Ritchie). This was augmented by the regular exploitation of mussels and turtles for food. The shells of river mussels (Unio spec.) were present in almost every unit of finds, and sometimes whole storage units were discovered, such as in contexts P10L241 and L254, where 5121 mussels were counted. Assuming two shells per complete bivalve, 2560 mussels were disposed. Considering that the shells were disposed of in middens after the organic matter had been consumed, their accumulation cannot indicate the storage of fresh food, as shown by the almost complete absence of mussels that were still closed, i.e. complete bivalves. Furthermore, in areas B and F on the tell, species of Unio appear mainly in the alleyways. Hence, river mussels must have been part of the common diet. In addition K. Ritchie identified a minimum number of individuals (MNI) of nine turtle shells in area B on the tell in one single unit of finds (feature P12B 246). kinds of animals or groups also stem from the meadows. Possibly the local animal husbandry made use of the natural sphere as well, in which case, the domestic pig could easily have found food. Fish bones in and off of the tell The analysis of fish bones from Pietrele is in its beginnings. The results presented here are based primarily on sieved samples and do not yet include the hand-collected material from these contexts. The samples included have been recovered mainly by wet-sieving using mesh-sizes of 0.5mm (P10F345 and F346, P11 B024, F811 and F821) and 4mm (P12B270). On the other hand, the fish materials included in shell middens P10L241 and L254 (see contribution by N. Benecke) were obtained by manual sorting. Some elements (e.g., loose teeth, ribs, spines) are not included in the totals, because they would bias the results in an unrepresentative manner. It should be noted that it is possible, for example, that Siluridae (catfish) are under-represented and Aci-penseridae (sturgeon) are missing, as most were large individuals, and the bones from these fish were noticed in the field and set aside with the other bone material where they await identification. The taxa listed in Figure 8 are fish typically classified as limnophillic (preferring still waters), but they also adapt to a variety of habitats (Dinu 2010; Froese, Pauly 2013; Kottelat, Freyhof2007). The one exception to this is Alosa sp., which keeps to the main channel of the Danube during its migration into freshwater (Kottelat 1997; Whitehead 1985). Although not present in any of these samples, the identification of Acipenseridae (sturgeon) in other contexts by N. Benecke (2004) also supports at least oc- All in all, this demonstrates that all of the ecotopes in the surroundings of the settlement site - the Danube meadows and the bodies of water there, the slopes as well as the high terrace - were integrated in the exploitation of natural animal resources. As can be recognised in the compilation in Figure 6, the focus of these activities was evidently on the Danube meadows, where the most frequently hunted animal was presumably wild boar (Fig. 7), but the other attested and quite frequent Family Species English Romanian German Clupeidae Alosa sp. shad scrumbie Maifisch Cyprinidae Abramis brama bream platica Brasse Cyprinus carpio carp crap Karpfen Aspius sp. asp peste lup Rapfen Rutilus sp. Scardinius sp. rudd ro^ioara Rotflosser cf. Alburnus sp. bleak albisoara Ukelei cf. Phoxinus sp. minnow craiete Elritze Esocidae Esox lucius pike stiuca Hecht Percidae Gymnocephalus sp. Perca fluviatilis perch biban Barsch Sander lucioperca pike-perch salau Zander Siluridae Siluris glanis wels catfish somn Wels Fig. 8. Species offish analysed in the units listed in Figure 9. casional fishing in the main channel of the river, where there are deep, strong currents. Chronologically, six of the eight collections analysed in this paper are from an early period of the Gumel-nita culture: four units derived from the lowest levels excavated in trenches B and F on the tell so far, and two from a coeval level revealed in trench L in the outer settlement. The units belonging to P10F were dated radiometrically to 4500- 4450 calBC; no 14C dates are available for the two units in trench B, but they are from features that can be typologically dated to the same period as P10F. The features from off the tell in trench L yielded pottery that also date them to an early phase. The other two units in trench F (P11F811 and F821) are considerable younger, dating from the end of the tell-sequence at around 4250 calBC. The data show that, internally, the samples from trenches B and L are relatively homogenous, although the contexts from trench F display considerable variation (Fig. 9). Because of the small sample sizes and differences in recovery methods, caution is needed when interpreting differences between the samples, and any conclusions must be considered preliminary. However, the available data do show variability between the trenches that is of some interest. Fish bones from the shell middens in trench L are overwhelmingly cyprinids, with a small number of other species included. In trench B, the cypri-nids are also the most important species, but here there are larger numbers of pike and fish of the perch family, and it is in this trench that catfish are most significant. The results from trench F 2011 (P11F) are similar to trench B, although catfish are less common. Trench F 2010 (P10F) has an even larger percentage of pike and perch family remains, although cyprinids still comprise almost half of the identified bones (though note that P10F346 is very similar to trench B). The difference between the absolute dominance of cyprinids from contexts off the tell and the more balanced representation of fish from the two trenches in the tell indicates that the processes responsible for creating these deposits were not the same. Further work is necessary to clarify whether this is the result of contextual, cultural, temporal, or other factors. One conclusion that can be drawn from these preliminary investigations regards fishing techniques. The presence of many specimens of very small fish, together with Alosa sp. (an anadromous fish that does not feed during its migration into freshwater - meaning that it will not respond to fishing methods relying on the use of bait), strongly point to the use of nets or traps in addition to the spears/harpoons and/ or fishhooks that were probably employed to catch some of the impressively large specimens present in the assemblage. Fishing in Pietrele The Danube River and the lakes were important food sources for the inhabitants of Pietrele and neighbouring settlements in prehistory and well into modern Ф sp ÖS Ф Л Ф -0 öS Ф Л "O Ф sp ös Ф Ф sj> öS Ф Ф sp öS Ф C C "O "O -0 -0 "O -0 u u u u 3 3 л П П