(49) The Late Quaternary vegetation history of Western Turkey (original) (raw)

THE ILIPINAR EXCAVATIONS II

edited by
J.J. ROODENBERG and L.C. THISSEN

THE LATE QUATERNARY

VEGETATION HISTORY OF WESTERN TURKEY
S. Bottema, H. Woldring, İ. Kayan

Table of Contents

Introduction J. Roodenberg … 1
1 The Pottery of Ilıpınar, Phases X to VA … 3
L.C. Thissen
2 Technological Study of the Chalcolithic Pottery of Ilıpınar, Phase VB … 155
A. van As, L. Jacobs and M.-H. Wijnen
3 The Potter of Örnekköy. An Ethnoarchaeological Note on the Pottery of Ilıpınar … 169
A. van As and M.-H. Wijnen
4 Stratigraphy and Architecture on the Southwest Flank of Ilıpınar … 177
Frédéric Gérard
5 Miscellaneous … 223
J. Roodenberg, with an Addendum of R. Cappers
6 Radiocarbon Determinations … 257
J. Roodenberg, W. Schier
7 Chipped Stone Assemblages of Ilıpınar, Phases X and IX … 279
Ivan Gatsov
8 Analyse fonctionnelle des assemblages en silex d’Ilıpınar, phases X et IX … 297
Maria Gurova
9 The Late Quaternary Vegetation History of Western Turkey … 327
S. Bottema, H. Woldring and I. Kayan

INTRODUCTION

J. Roodenberg

This second volume on the excavations at Ilipinar does not differ from the design of the first one which appeared in1995, i.e. to present the outcome of studies in the order of completion regardless of their subject. No need to say that such a policy incorporates the risk of incoherence of the contents. In spite of this, reviewers of that volume were courteous enough not to stumble over it, and that circumstance incited the editors of volume number two to follow the same path. Our excuse for this procedure may be explained by practical reasons often outweighing scientific ones. Insofar as the reader does not recognize this situation as his own, his understanding is requested.

A major contribution to this volume is the study by Thissen on the pottery of Ilipinar’s early farming period, e.g. the phases X to VA. This study constitutes the core of a PhD thesis entitled “Early Village Communities in Anatolia and the Balkans, 6500-5500 cal BC”. The bulk of the pottery treated here originates from the so-called ‘Big Square’ and S13 in Ilipinar’s central sector and was complemented with material collected in the southwestern sector. A second pottery study (chapter two), by van As and Wijnen, deals with material collected from phase VB deposits mainly represented at the foot of the western mound slope. These authors also give an account (chapter three) of the last potter in the region, who exploited the same natural clay resources as his prehistoric predecessors. Chapter four comprises the investigation results of habitation remains along the southwestern flanks of the prehistoric mound carried out in four successive seasons by F. Gérard. Form and content of his account inevitably reveal the hand of the editors. In the next chapter Roodenberg deals respectively with the lay-out of the phase VI village, the occupation remains of VB (including a brief report on charred seeds by R. Cappers), and with aspects related to mound formation and settlement size. In chapter six the complete record of radiocarbon determinations is presented according to location, stratigraphy and phase division, whereas W. Schier proposes a statistically simplified chronology. The chipped stone industry from the oldest phases is morphologically and functionally discussed by Ivan Gatsov and Maria Gurova in chapters seven and eight respectively. Finally, the current state of research on Late Quaternary vegetation of western Turkey is presented at the end of the book by Bottema, Woldring and Kayan.

9

THE LATE QUATERNARY VEGETATION HISTORY OF WESTERN TURKEY

S. Bottema 1{ }^{1}, H. Woldring 1{ }^{1} and I. Kayan 2{ }^{2}

Abstract

In this paper the vegetation history of the Marmara district is discussed on the basis of palynology, for the last 12,000 years. Sediments generally represent the younger part of the Holocene. In a core from Yenişehir the Holocene and part of the Late Glacial are present. The palynological results are translated in terms of vegetation and the development of the plant cover is compared with that of other parts of Turkey. The environment of human habitation deduced from archaeological investigations is reconstructed on the basis of the palynological results. Human-induced changes in the vegetation of the area south of Lake Iznik are traced in the pollen curves. Archaeological records suggesting periods with low habitation pressure or even absence of population are compared with evidence from the pollen diagram of Yenişehir.

The behaviour of pollen of the Centaurea solstitialis-type is remarkable and is is compared with diagrams from other parts of Turkey. The discussion focuses among other things on the interpretation of this particular type, treated before by Bottema & Woldring (1984/1986).

1. INTRODUCTION

The Late Quaternary vegetation history of Turkey has been studied in parts, beginning with the southeast (Van Zeist et al. 1968/1970), followed by the southwest (Van Zeist et al. 1975; Bottema & Woldring 1984/1986) and the north (Bottema et al.1995) and here is followed by investigations in one of the remaining parts, the west. The western part of

[1]

1. 1{ }^{1} State University of Groningen, The Netherlands.
   2{ }^{2} University of Izmir, Turkey ↩︎

Asian Turkey differs from the other parts by the fact that it lies at a low altitude compared to the Anatolian Plateau. For this reason a different vegetation history may be expected.

The environmental investigations as referred to above invariably are connected with archaeology in some way or another. In this paper, most of the research was connected with the archaeological investigations by J.J. Roodenberg (Nederlands Historisch-Archaeologisch Instituut (NHAI) in Istanbul and Nederlands Instituut voor het Nabije Oosten (NINO) in Leiden), whereas coring locations more distant from the prehistoric sites still should be regarded as spin-off from Ilıpınar, the key site. Prof. Ilhan Kayan (Izmir University) studied the geomorphology and the soil formation at the NHAI excavations in the Iznik Gölü area. Owing to his help in the field, both in obtaining research permissions and in contributing his field experience, the Groningen team were able to investigate many sedimentary locations and to collect the material treated in this paper.

2. METHODS

During the fieldwork, potential sedimentary areas, for instance large lakes, were identified from the map, while Professor Kayan indicated smaller sedimentary basins in the field. The lakes were of a size allowing the edges to be probed in several places. The samples were collected with a 3.8−cm3.8-\mathrm{cm} Dachnowski corer with an inner tube length of 25 cm . Compared with lake basins in the Anatolian mountains or on the Anatolian Plateau, we often hit sand or gravel in the Holocene deposits of western Turkey, which obstructed the sampling of long sequences. Samples for pollen analysis were treated according to standard procedures, as described in the papers by Van Zeist et al. (1968/1970; 1975).

3. GEOGRAPHY

The area studied forms part of western Turkey (Fig. 1) and more specifically of the southern Marmara Sub-region as described by Erol (1982). It covers Erol’s Bursa and Iznik mountains and basin (no. 122 of TAVO map A VII 2) and the Marmara Lake District (no. 121 of TAVO map A VII 2).

The peninsular projection formed by the Samanlı Dağ contains the large Lake Iznik in the south. The mountains are about 900 m high in the west and rise to 1300 m and more to the east. Iznik Gölü lies at an altitude of 75−150 m75-150 \mathrm{~m} and drains westward to the Gulf of Gemlik.

To the west and south of Lake Iznik lie the archaeological sites of Ilıpınar, Hacılar Tepe, Yeniköy Tepe and Menteşe (Fig. 1). Except Yeniköy these sites have been or are being studied archaeologically by J.J. Roodenberg.

South of Iznik Gölü, the hills of Avdan Dağı (c. 800 m ) separate the Iznik depression from the Yenşehir basin. The Yenişehir basin is drained by the Göksu River, which discharges to the east into the Sakarya River which runs into the Black Sea.

In the southwest, the Yenişehir area is bordered by the Uludağ, which with its 2543 m towers above the surrounding landscape. The lands north of the line BursaMustafakemalpaşa up to the Gulf of Erdek are generally below 150 m above sea level. Only east of Bandırma do we find outcrops up to 800 m . In the lowlands lie the lakes of Apolyont (Ulubat) and Manyas or Kuşcenneti.

4. MODERN CLIMATE

The average January temperature for the Iznik area varies from 8∘C8^{\circ} \mathrm{C} along the coast to 0∘C0^{\circ} \mathrm{C} inland, whereas on the Uludağ the average for this month drops to values below 0∘0^{\circ} C (Yeni Türkiye Atlası 1977). The average temperature for July amounts to 20−24∘C20-24^{\circ} \mathrm{C}. On the coast near Gemlik and more to the southwest of Bursa, temperatures may be somewhat higher.

The average annual precipitation for the Yenişehir area south of the Lake Iznik amounts to 400−500 mm400-500 \mathrm{~mm}. The part towards the Sea of Marmara receives more than 500 mm annually, up to about 800 mm near the sea. During the summer months the area discussed in this paper receives less than 80 mm of rain, apart from the Uludağ where ascending air is responsible for up to 150 mm . Most of the rain falls in the remaining part of the year, with an emphasis on winter and spring.

Figure 2 informs us about the climate of Bursa (Walter 1965).

5. MODERN VEGETATION

The modern vegetation of the Marmara Lake District is much influenced by human activity. A large part has been turned into farmland, either planted with crops or used for orchards. In those parts that are not cultivated, exploitation for timber but more often for grazing has changed the natural vegetation, reducing it to various stages of degraded or secondary plant cover.

The variation in elevation of the area, from sea level up to 2500 m on the Uludağ, produces various habitats. The vegetation map of the Near East (Frey & Kürschner 1989) does not have enough resolution to inform us about details for the Marmara Lake District and offers a more general picture. The higher part of the area around Lake Iznik is thought to carry (under natural conditions) a mixed formation of cold-deciduous montane forest with evergreen broad-leaved trees or shrubs and evergreen needle-leaved forests resistant to cold (in terms used by Frey & Kürschner).

The coastal lowlands may have been covered by a mixed formation of cold-deciduous lowland forest with evergreen broad-leaved trees or shrubs. More inland, around the lakes, scrubland with xeromorphic dwarf shrub is postulated.

The potential vegetation of the Marmara Lake District is also given on the vegetation map of the Mediterranean Region (Quézel & Barbéro 1983). According to these

authors, the lowlands up to about 600 m carry a meso-Mediterranean deciduous forest of Quercus cerris and Q. frainetto on non-calcareous soils and alluvial deposits, while forest of Ostrya carpinifolia and Quercus pubescens grow on limestone.

Above 600 m a zone with Carpinus betulus, C. orientalis and deciduous oak is found, sometimes with Fagus orientalis and Abies bornmuelleriana. On the slopes of the Uludağ, above 1500 m , a zone with Fagus orientalis, Abies bornmuelleriana, and Pinus sylvestris is met. During our visit in 1988, we found sweet chestnut (Castanea sativa) to be common on the slopes of the Uludağ below the beech zone.

The potential vegetation of the Yenişehir area has been drawn from the map of Quézel & Barbéro (1983) and is displayed in figure 3.

6. MODERN POLLEN PRECIPITATION

For the study of the modern pollen precipitation in the research area, seven surface samples were taken. For the location of the surface samples see figure 1. The description of the sample locations follows here.

Sample 1 (TU88-1)

This sample was taken in an olive grove east of the village of Sölöz. In the vicinity the following species were seen: Olea, Populus, Morus, Vitis, Cornus, Salix, Quercus pubescens, Scabiosa, Daucus, Pulicaria, Setaria, Rubus, Portulaca (yellow-flowering), Amaranthus, Xanthium strumarium, Cynodon, Cirsium vulgare, Gramineae.

Sample 2 (TU88-2)
About 3 km south of Sölöz: Mediterranean vegetation with Pistacia lentiscus, Quercus, Arbutus, Pinus, Cistus, Cercis?, Rhus coriaria, Hypericum (androsaemum), Fraxinus ornus, Rosa canina, Paliurus spina-christi, Phillyrea, Spartium. Steep slopes make surveying almost impossible.

Sample 3 (TU88-3)
The road from Sabriz via Bayırköy to Orhangazi/Bursa. Along steep slopes of a deep valley, inaccessible slopes carry less disturbed vegetation. On east-facing exposures, pine is dominant; for west-facing exposures, see the description of sample 2. Towards the southwest, the landscape is more flat, with pasture. Along the road Castanea, Juglans, Platanus, Prunus cerasifera (yellow and red fruit), Pyrus domesticus, Vitis. On a northfacing exposure: Fagus orientalis, large Ulmus, Carpinus betulus, Hedera helix.

Sample 4 (TU88-4)
Near Bayırköy: Fagus orientalis, Quercus cerris, Corylus, Alnus glutinosa, Populus, Salix (along water), Cornus, Mespilus, Prunus (large red fruit), Ulmus cf. scabra.

Sample 5 (TU88-5)
Southern slope of the Uludağ: dominant Castanea sativa (Castanetum), Pinus nigra, Corylus, Rubus, scattered Quercus pubescens, Fagus orientalis (juvenile specimens), altitude 700 m .

Sample 6 (TU88-6)
Fagetum on the Uludağ at 1000 m ; the sample was collected under predominantly Castanea sativa and apple trees. Above this level only Pinus nigra and Pteridium. Here and there cultivated land and sheep grazing.

Sample 7 (TU88-7)
Altitude 100 m . Sample taken about 60 metres downwind of a threshing machine at the edge of Bayırköy, in order to trace the distribution of Cerealia-type pollen. Sample from moss on the bark of a plane tree (Platanus).

6.1. General vegetation description

Between Iznik Gölü and Uludağ a distinctly anthropogenic landscape is present. Bordering the lake a rocky zone is found, whereas more inland weathered clay occurs. Locally an extensive Paliurus vegetation occurs, which is reminiscent of the Kastamonu area in northern Anatolia. Poterium is present, as are most characteristic Mediterranean species. Above the beech zone on the Uludağ, there is a belt with Quercus pubescens, which regenerates from cutting.

The analyses of the modern pollen rain are displayed in figure 4. The pollen percentages are based upon a sum that includes all identified pollen and spores.

6.2. The results of the surface-sample analyses

The transect of surface samples, which starts in the cultivated area near Lake Iznik and ends on the Uludağ, demonstrates an increasing arboreal (AP) pollen percentage. In the first sample (TU88-1), tree pollen is almost exclusively represented by pollen of olive (70%)(70 \%) while twelve other types share the remaining 8%8 \%. These latter taxa were mostly not seen in the vicinity. The absence of Vitis pollen in the sample although grapes were growing in the olive grove is striking. The absence of Vitis (vinifera) in the pollen precipitation stresses the difference in pollen representation of wild climbing dioecious vine, which has high values, and cultivated monoecious vines, whose pollen is mainly not represented or found in very low numbers only. In surface sample 1 we are dealing with the latter subspecies of Vitis.

The herbs are represented by pollen of grasses and Chenopodiaceae. The latter type may, however, have been produced by Amaranthus, which was seen locally. Amaranth has a similar pollen type to many chenopods.

Sample TU88-4 was collected in the forest zone near the village of Bayırköy. The forest had been opened locally and stubble fields and grazed areas were seen. There was clear evidence of lumbering. The presence of a little stream explains the occurrence of Alnus and Salix pollen. Beech pollen measures 2.7%2.7 \%, which suggests under-representation in the pollen precipitation. The high value of 29.4%29.4 \% pine pollen must have been produced by the pine trees that were seen on the opposite side of the stream valley that runs from Bayırköy to Sölöz.

Sample TU88-7 was collected especially for studying the behaviour of Cerealia pollen released by threshing. Moss was sampled from the trunk of a plane tree. Planes (Platanus orientalis) were present along the small stream that runs parallel to the road. The threshing machine was in full operation and a dense cloud of dust and chaff was blown into the air. Large numbers of cereal pollen were freed in the course of the dehusking, because in self-pollinating cereals a lot of pollen stays behind in the chaff. The machine separates the corn from the chaff and blows it away; the wind will do the rest. The dust settled in the little stream valley where the only moss for trapping pollen grew on the trunk of a plane tree, 60 metres from the threshing machine. The latter feature is very pronounced: Platanus pollen amounts to 78.4%78.4 \%. Cerealia-type pollen measures only 0.3%0.3 \%, which means that the threshing did not distribute large numbers of cereal pollen downwind.

The very high plane-tree pollen value in TU88-7 is responsible for a high arborealpollen percentage. Such a high value erroneously suggests extensive forest in the area. The sample displays the same kind of over-representation as is found in the sample from the olive grove (TU88-1). In neither case is there any forest in the area.

Sample TU88-5 presents a pollen image of a sweet-chestnut belt, where other tree species are found in low numbers only. In contrast to the preceding samples 1 and 7 , the representation of several species observed in the field is much better. This applies to Corylus, Castanea, Quercus and Pinus, but not for Malus-type, which as usual is found in very small numbers only.

Sample TU88-6 was gathered among chestnuts and apple trees, another three hundred metres up the slope of the Uludağ. The pollen percentage of sweet chestnut measures 85.5%85.5 \% but Pinus nigra only amounts to 8%8 \%, whereas this tree dominated the forest above the chestnut belt. It is difficult to explain why pollen of oriental beech is not found more frequently, as this species was growing abundantly at a distance of only 100 metres.

On the basis of the lithology, the two samples analysed from the short core taken from the prehistoric mound of Ilipinar were believed to be young enough to indicate the modern pollen precipitation on the mound. The AP value measures about 20%20 \% and a large number of herb types are found, some of them in appreciable amounts, for instance Cyperaceae (sedges), Liguliflorae (dandelion type) and Gramineae (grasses). However, when these spectra are compared with the modern pollen samples, the modest presence of Abies in the two Ilipinar spectra and the absence of Olea, make a recent date unlikely. When correlated with the Yenişehir diagram, they must be older than 3000 years.

Comparison of the surface-sample data with the subfossil spectra of the westTurkish pollen diagrams shows up large differences, which are mostly related to differences in catchment system. The regional pollen precipitation of sedimentary basins is very seldom dominated by one pollen type, whereas surface samples collected among vegetation naturally tend to be dominated by local trees or herbs. The interpretation of the latter category must be in terms of pollen-production and pollen-distribution behaviour of the respective taxa. Palynological absence of a plant species found near the sampling locality can be important information, as well as the presence of pollen of a tree species which was recorded as growing further away.

7. RADIOCARBON DATING

The following depths were radiocarbon-dated at the Groningen Institute of Isotope Research. The organic fraction and the calcareous fraction were dated in order to trace 14C{ }^{14} \mathrm{C} depletion by old carbon from calcium carbonate. The dates of the organic fraction were used for determining the age of the pollen spectra.

• The depths have been adjusted to include the additional 100 cm of the ditch.

8. The Pollen CORES

8.1. Yenişehir

This core was taken west of the town of Yenişehir, south of the road to the villages of Seyman and Erdoğan. Sampling was done from the bottom of a ditch running more or less parallel to the road south of Seyman. The upper samples were collected from the side of the ditch.

Lithology

Further sampling proved impossible because of hard sediment. The pollen diagram of Yenişehir is shown in fig. 5.

Yenişehir pollen zones
To facilitate the discussion of the Yenişehir pollen diagram, the following pollen zones were distinguished:

Zone 1 (spectra 1-4): high non-arboreal pollen values (NAP), mainly Chenopodiaceae and Artemisia. Zone 2 (spectra 5-13): relatively high arboreal-pollen values (AP), mainly Corylus, Fagus, Quercus cerris-type, Tilia, and Ulmus, decreasing in the upper spectra. Zone 3 (spectra 14-27): AP values increasing to about 80%80 \%, due to high Fagus, Quercus cerris-type, and especially Pinus. Centaurea solstitialis-type is low and Gramineae are high. Zone 4 (spectra 28-34): AP values decrease to about 40%40 \%. Centaurea solstitialis-type values are up to 40%40 \%. Zone 5 (spectra 35-50): AP values rising again to 60 80%80 \%, mainly due to Fagus, Quercus, Juglans, and Olea. Centaurea solstitialis-type has low values. Zone 6 (spectra 51 and 52): all tree-pollen values decrease considerably apart from those of Pinus, which amounts to 60%60 \%.

Discussion of the Yenişehir pollen diagram

The Yenişehir diagram covers the longest period of the west-Turkish sediments that were collected for this study. The lowest part might tentatively be dated to around 10,000

BP, if we assume a constant sedimentation rate for that part of the core.
In order to date the lower part of the Yenişehir record, the pollen curves were compared with the nearest radiocarbon-dated evidence, the diagram of Abant Gölü, 140 km to the east. The level of “Yenişehir-680 cm” was thus dated to about 10,000 BP. The lowermost part, down to a depth of 715 cm , thus coincides with the last part of the Late Glacial as it is known from Western Europe. The complete diagram may go back about 11,000 uncalibrated radiocarbon years.

The time range of the Yenişehir diagram qualifies it as a good reference for the Iznik area. The other pollen evidence presented here covers only short sections of the Yenişehir diagram and sometimes even negligible parts.

In Yenişehir zone 1 the sub-fossil pollen precipitation is characterised by 70-95% herb pollen, notably that of Artemisia and Chenopodiaceae. Both types, especially when found together, are highly indicative of steppe conditions and they are present in many Near Eastern and Mediterranean locations (Van Zeist & Bottema 1991).

Surface-sample studies (Bottema & Barkoudah 1979) show that the pollen precipitation of the modern Syrian steppe matches the Late-Glacial samples of Yenişehir very well, be it that in the Late Glacial of western Turkey even lower arboreal-pollen values are found. This suggests that the Iznik/Yenişehir area was completely devoid of trees and that the sparse tree pollen came from the mountainous hinterland on the Anatolian Plateau, where refuges might be expected. The vegetation reconstruction in Iznik/Yenişehir can be used as proxy data for a reconstruction of Late-Glacial climate, which is thought to have been rather cold and above all dry.

Towards the end of zone 1 some tree-pollen types appear, including oak, lime tree and birch. Quite probably such pollen was produced by stands somewhere in the mountains. Birch pollen may well have been produced by trees growing, like those in the Abant region, on the higher part of the mountains (Bottema & Woldring 1995). Around Abant Gölü ( 1300 m ), birch and lime peaked even before 10,000 BP.

Ephedra distachya-type is fairly common in the pollen rain of the Yenişehir basin. In this respect the type behaves in the way it does in many records from all over the Anatolian plateau. In the diagram of Lake Van in eastern Anatolia (Van Zeist & Woldring 1978) significant values are found, and in Abant and Yeniçağa in western Turkey values resemble those of Yenişehir. It is not clear which Ephedra species is responsible for this pollen, but modern surface samples collected by Freytag (1977) in Afghanistan contain large amounts of Ephedra distachya-type pollen.

The Late-Glacial coastal landscape west of the research area differed somewhat from the present situation because the sea level was lower. During the coldest part of the last ice age, around 18,000BP18,000 \mathrm{BP}, the sea level was more than 100 metres lower, and the water had not fully gained its present level by the last part of the Late Glacial.

Towards the end of zone 1, the pollen percentages of herbs show a pronounced decrease, whereas one single tree-pollen type (that of deciduous oak) equally strongly increases. At the beginning of zone 2, oak pollen constitutes 30−40%30-40 \% and this must have been produced by oak stands which developed in the west-Turkish lowlands, as the oak-

pollen values in the Anatolian upland records are much lower. This suggests that in the mountain ranges oak played a minor role compared with other tree species. For Yenişehir the predominance of oak pollen can be translated into a predominance of this tree in the vegetation. The present-day occurrence of deciduous oak in the Central Anatolian foreststeppe illustrates the drought resistance of this tree.

Dry conditions are very likely to have prevailed in the Yenişehir area during the period covered by zone 1. In the lowlands of western Turkey the beginning of the Holocene may initially have been dry, as is the case with the Preboreal of northwestern Europe. It seems that oaks were the first trees to conquer this rolling landscape. Other, more moisturedemanding species occurred at higher altitudes with more precipitation. The only species that show slightly increasing pollen values are Corylus and Alnus. The latter may have profited from the moisture available in and around the basin. Hazel (Corylus) appeared in the mountains later, probably having first spread in the lower altitudes. In this case hazel may have been Corylus avellana as well as the large tree hazel (Corylus columna). The two species show more or less the same distribution pattern (Browicz 1982).

From the fact that sedimentation took place it can be deduced that moisture was available. The basin of Yenişehir filled up with clay that was deposited in water. One may conclude that the basin was fairly deep, i.e. several metres, since pollen and spores of water plants were lacking in zone 1. Evaporation was low before the Late Glacial, but it increased after 10,000BP10,000 \mathrm{BP}. This lowered the water table, which soon left the lake shallow enough to allow the deposition of green algae (Pediastrum) followed by pollen of Myriophyllum, a water plant which flowers under stress conditions such as lowering of the water table (Weeda et al. 1987).

Zone 2 represents a period with a diverse pollen assemblage. To facilitate the discussion, spectra 5-16 will be grouped together. In zone 2, tree pollen reached values of 50−60%50-60 \% and a variety of types contribute to the pattern. Taxa present in moderate to large amounts include Alnus, Carpinus betulus, Corylus, Fagus, Quercus pubescens, Abies, Pinus, Tilia and Ulmus. Apart from the large share of deciduous oak, no pollen taxon really dominates the spectra. Values of conifers are low, especially Pinus. Pine, whether this was Scots pine, Corsican pine or Calabrian pine ( PP. sylvestris, P. nigra or PP. brutia), very probably did not occur in the area, but actually grew somewhere in the mountains. The same is true for fir (Abies). Firs grew in large numbers in the Pontic Mountains, as is deduced from the diagrams of Abant and Yeniçağa in northwestern Turkey, which show continuous pollen curves from 12,000BP12,000 \mathrm{BP} onward. A representative of Abies nordmanniana, the subspecies equi-trojani, may have occurred locally in the Iznik/Yenişehir area.

The vegetation of the Iznik/Yenişehir area naturally differs from that of upland Anatolia, because the area lies at least 800 metres lower than the sampling sites of for instance Abant and Yeniçağa. A different vegetation is to be expected on the lowland hills. Oak, hazel, lime and hornbeam will have formed the forest stands around the basin of Yenişehir. Quite a lot of Fagus pollen is present in zone 2; beeches must have grown on

the slopes of the surrounding mountains, e.g. the Avdan Dağ ( 1287 m ), the Uludağ ( 2493 m ) and the 1600−m1600-\mathrm{m} high mountains east of Iznik.

Originally, pine did not play an important part in the pollen rain of Yenişehir. Towards the end of zone 2, around spectrum 13, pine and fir pollen increased together and especially the 10%10 \% Abies pollen points to conifers growing not far from Yenişehir, for instance on the Uludağ. The somewhat higher conifer-pollen values were preceded by important percentages of Centaurea solstitialis-type, a herb that in terms of pollen replaced most of the Gramineae.

Grass pollen in Anatolian diagrams generally increased at the time when Artemisia and Chenopodiaceae pollen values decreased. In this respect Yenişehir forms no exception. It is no rule that the pollen of Gramineae gave way to that of the Centaurea solstitialistype; there are Anatolian pollen diagrams (e.g. Abant, Yeniçağa and Ladik) where this change is not found (Bottema et al. 1993/1994). Pollen diagrams of the Pisidian area (OroMediterranean/Xero-Euxinian), of Beyşehir, Gölhisar, Avlan and Elmalı demonstrate high Centaurea solstitialis-type pollen percentages. These sites do not show a negative correlation between Centaurea and Gramineae, but a positive correlation of Centaurea solstitialistype with a group of other composites, including Matricaria-type (plants like camomile), Liguliflorae and unidentified Tubuliflorae. In the steppe part of central Anatolia, the plain east of Konya shows relatively low values for various types of composites, including Centaurea solstitialis-type, but grass pollen is present there in great quantity.

The Centaurea solstitialis-type pollen grain has often been a subject of debate. How can we interpret the presence of this pollen type, which is spread by yellow-, white- and pink-flowering annuals of the species Centaurea solstitialis and Centaurea iberica? Both taxa are annuals and like many composites they tend to be insect-pollinated. This characteristic would make it less likely for the pollen grains to appear in large numbers in the pollen counts. Still, from time to time they appear in the pollen precipitation in large quantities, often together with other composites such as Liguliflorae. Liguliflorae, dandelion-type plants, do not release much pollen into the air and natural stands of such plants never account for more than a few percent when the local pollen production is measured. Thus the taxon seems to be under-represented. Still there are situations in which Liguliflorae are found in large numbers in certain sediments or in archaeological deposits. Various causes, possibly even with a cumulative effect, have been suggested to explain the anomalously high percentages (Bottema 1975, Bottema & Woldring 1984/1986). Selective preservation, concentration or transportation by upward air movement in hot summers during the flowering of these species, abundant flowering on flats of lakes that are drying up during springtime, all may cause these high values. The transportation of colluvial matter in which most of the pollen has disappeared by oxidation, but where Centaurea solstitialis-type and other resistant composites remain, may play a role too.

A palynological study in the Sagalassos area, in southwestern Turkey not far from Burdur and Isparta, is instructive in this respect. The classical site of Sagalassos was excavated under the direction of M. Waelkens (Leuven) and the current authors Bottema and Woldring (1995) carried out the environmental investigations. In the alluvial basin of

Çanakl, near Sagalassos, clay has been deposited for a long time and it was observed that inundation took place during winter, whereas the flats dried out in summer. These fluctuations are responsible for a variation in the vegetation from wet spots in the centre (with Eleocharis) to drier ruderal areas at the edges where a dense vegetation of Centaurea solstitialis and Centaurea iberica is found. In surface samples from the edge of the basin, a few metres higher on the plain, Centaurea solstitialis-type pollen plays no role. Evidently little pollen from the nearby stands is brought into the air to precipitate on the higher edges.

In a core collected in the wetter part of the plain of Çanakl, pollen had only been preserved in the upper 30 cm , of which five samples have been analysed (Bottema & Woldring 1995). In these samples the values for Centaurea solstitialis-type are 12−15%12-15 \% at a depth of 30 cm . The values for this type gradually drop to 4%4 \% in the upper sample, the present surface. Values for the Liguliflorae measure 30%30 \% in the oldest sample and 4%4 \% in the youngest. The diversity in pollen types gradually decreases downward, to a minimum in the oldest, deepest sample. Below 30 cm no pollen has been preserved. The oldest sample that contains pollen must have been exposed to corrosion the longest. The high values for composite-pollen types must be explained in terms of survival of the toughest pollen grains. In this respect it is understandable that the deepest sample also contains a very high amount of pollen grains ( 30%30 \% ) that could not be identified as a result of corrosion. The more recent samples were less affected and the uppermost spectrum contained only 5%5 \% unidentified grains.

In Yenişehir, the part of zone 2 that displays high Centaurea solstitialis-type values is dated between 8500 and 6400 BP on the basis of two radiocarbon dates. If we look at other pollen diagrams from Anatolia, especially those from Pisidia, it is not easy to pinpoint this Centaurea phenomenon chronologically. Interpolation of the time covered by the Beyşehir core (Van Zeist et al. 1975, Bottema & Woldring 1984) dates it between about 7200 and about 6200 BP , assuming a constant sedimentation rate. High Centaurea values in Gölhisar and Pinarbaşı can be partly correlated with Beyşehir or Yenişehir and such events may possibly be connected with local phenomena. It is striking that high Centaurea solstitialis-type pollen percentages are never found in organic or peat deposits. The formation of such deposits takes place under conditions that differ from those prevailing in clay deposits. Both sediments may result from the climate, but can also be the effect of local hydrology. It is obvious that the preservation of organics is often connected with the preservation of pollen grains.

Thus, the high values of composite pollen seem not so much a reflection of a particular vegetation type, as indeed a reflection of certain conditions that may be widespread in certain parts of Anatolia in combination with local changes.

The clay deposit displays a much lighter colour in the Centaurea phase than in the phases before and after it, when the clay is almost black. The diminishing organic production may have been caused by increased oxidation due to drying out of the basin. This could have taken place gradually even at the beginning of zone 2 and the high pollen percentages of Myriophyllum may point to a lowering of the water level.

The final conclusion about zone 2 is that during this period a desiccation of the Yenişehir basin took place, during which in winter sedimentation still occurred but in which summer evaporation was greater than the water supply. The basin was marshy with little water, which did not develop into a proper lake. The tree-pollen curves of zone 2 do not hint at desiccation and the mountain forest of beech and fir is clearly expanding! In fact this could be the reason for a certain desiccation of the Yenişehir basin. The explanation is as follows: A strong increase in forest started to retain the water on the mountain slopes, evaporating it through increasing foliage and humic soil, instead of running down the slopes and filling lowland basins. Within zone 2 there is a short period during which tree pollen, apart from pine pollen, suddenly decreases. The dip in AP covers spectra 11-13 and is dated about 7000-6500 BP. Pollen diagrams from western Turkey nowhere indicate a climatic cause to explain this decrease, so one has to conclude that human impact is responsible for this abrupt phenomenon.

After the dip in tree pollen described above, which separates zone 3 from zone 2, the vegetation in zone 3 is composed of the same elements. The main difference is a quantitative one: Pinus pollen gradually increases from a few percent to about 40%40 \%. Zone 3 seems to be the optimal period for tree growth, as is deduced from the tree-pollen percentages, which increase after the dip zone 2 (spectra 11-13), in which the pollen values of light-demanding taxa, such as Chenopodiaceae, Artemisia and grasses, were very important.

The Yenişehir lowlands and the lower part of the slopes around the basin saw a deciduous vegetation of oak, hazel, hornbeam and lime (Tilia), especially on the slopes. In wet places around the basin, alder (Alnus) was found. In higher parts of the mountains, comparable with the present-day belts, forest of beech and Nordmann fir was found. This forest at first may have been somewhat patchy, as suggested by the amount of herb pollen, especially that of grasses. The grasses gradually disappeared and pine must have played a decisive role although it is difficult to locate precisely where it grew. Zone 3 started about 6400 BP and lasted until about 4000 BP .

Zone 4 shows the lowest AP values in the Yenişehir pollen diagram since the Late Glacial. This decrease is caused by a decline of all deciduous types, whereas pine pollen at first demonstrates the same values as in the previous zone, eventually to decrease in the upper spectra. Herb pollen is represented by Chenopodiaceae, Liguliflorae and Tubuliflorae and up to 40%40 \% Centaurea solstitialis-type. Grass pollen does not play a role and Cerealia pollen is only rarely encountered.

The most conspicuous palynological event in zone 4 is the retreat of all deciduous tree species. Percentages of conifers, pine and fir remain constant. Increasing amounts of composites compensate for decreasing deciduous-tree types. The forest on the slopes around the Yenişehir basin must have almost completely disappeared and slope-wash must have been transported to the basin, as is concluded from the Centaurea solstitialis-type pollen grains found in the samples. The coniferous forest, growing on the slopes of the Anatolian Plateau, was not noticeably affected. It is remarkable that no anthropogenic indicators are found in the pollen diagram of zone 4 . There are no significant values of

crops and there is no pollen of weed plants such as Plantago lanceolata. When in some way or other the forests were attacked, there are no signs that a maquis vegetation replaced the dwindling forest. When other pollen diagrams of Turkey are screened for a comparable event, a general depression of many tree-pollen species becomes obvious, although the dating of this phase may be somewhat different. In some cases deciduous pollen types were replaced by pine pollen. Diagrams of northern Turkey (Abant, Yeniçağa and Ladik) show primary and secondary anthropogenic pollen indicators (Behre 1991) appearing as soon as the deciduous tree-pollen types decrease. The pollen diagram of Beyşehir shows the same pattern as the Yenişehir diagram. Conifers display significant values, whereas deciduous tree-types are very much reduced or even absent. Centaurea solstitialis-type and other composite types are abundant (Van Zeist et al. 1975, Bottema & Woldring 1984/1986).

One wonders which prehistoric culture had such impact upon the Turkish forest vegetation as is suggested by the pollen diagram? That the influence of the Ilıpınar settlement on the vegetation extended to the Yenişehir area is not plausible, but there are indications that the region across the low watershed was populated, e.g. the site of Menteşe. Around 4200 BP, the Early Bronze Age (EBA) settlement of Hacılartepe near Ilıpınar was occupied. The numerous mounds in the Iznik-Yenişehir region with EBA occupation demonstrated a higher population density than ever before. Such a dense occupation of the land may have been responsible for the clearing of the lower forest zones that is evident in the pollen diagrams. The degree of deforestation may be explained also by fuel collecting for ore smelting or metalworking (see also section 9).

Towards the end of zone 4, a slight increase in the pollen values of Corylus, Quercus and Fagus is visible, whereas the values for Pinus decrease. The boundary between zone 4 and zone 5 is laid where the arboreal pollen reaches an absolute minimum value and where a very sharp peak of grass pollen is apparent. This boundary at the same time marks a division between two types of sediment. Below 261 cm grey clay with black patches is found and above this level the clay is grey and sticky.

From a depth of 261 cm upwards, pollen of Olea, Fraxinus ornus and Juglans are found, pointing to the beginning of the Beyşehir Occupation Phase, which manifests itself in the palynological picture especially in southwestern Turkey. This phase, first described by Van Zeist et al. (1975), is dated immediately after the volcanic event of Santorini. In uncalibrated radiocarbon years the event is (post)dated 3200 BP. This defines zone 4 as having occurred from about 4000 BP until the Santorini eruption took place. The impact of the ash deposit upon Greek islands and the western part of Turkey is a subject of ecological studies. The Yenişehir sediment shows us that changes in the pollen spectrum already occurred before the zone 4/54 / 5 transition. The Centaurea solstitialis-type curve was already decreasing and several deciduous tree-pollen types had started to recover slightly from the dip in spectra 31 and 32.

Zone 5 shows a regeneration of the vegetation that had been present in zone 3 (from around 6000 to 5000 BP ). Despite a pollen assemblage that looks very similar, some differences can be observed. In the mountain zone, fir (Abies) is almost absent, whereas the group of tree-pollen types appears that is characteristic of the Beyşehir Occupation Phase

(Bottema & Woldring 1984). This group includes Castanea, Juglans, Fraxinus ornus and Olea. While the seemingly natural forest had ample opportunity to establish and maintain itself on the low mountains in the Yenişehir and Iznik area, anthropogenic orchards spread. At the same time a continuous curve of Plantago lanceolata-type pollen developed, testifying a certain form of agriculture. The advanced farming and fruit-growing culture that is deduced from the pollen assemblages did not visibly disturb the vegetation on the slopes of the mountains. There is no distinct impact of colluvial influence in the sediment, which probably would be indicated by various composites, including Centaurea solstitialis-type.

The two spectra that form zone 6 contain high values of Pinus pollen, whereas most deciduous types decrease. Olea pollen is not present in spectrum 51 and scarce in the upper spectrum. High values of pine pollen are also present in the upper parts of diagrams of southwestern Turkey, for instance that of Beyşehir at an elevation of 1300 m and those of Köyceğiz and Ova Gölü.

The most recent millennium is missing from the Yenişehir record. It may have been represented in the half metre thoroughly ploughed topsoil…

8.2. Apolyont Gölü

The sediment of Apolyont Gölü was cored in a water-lily marsh, near the dike that borders the lake. The water level was clearly lower than normal, judging by the water lilies that were in part flowering in dry mud.

Lithology

0−186 cm10-186 \mathrm{~cm} 1 \quad ight grey clay, sometimes with brown or black patches
186−228 cm186-228 \mathrm{~cm} \quad darker clay, somewhat greenish, clouded-like marble or with large black patches in bluish-green clay
228−400 cm228-400 \mathrm{~cm} \quad grey to grey-white clay, sometimes sandy. At 315−333 cm315-333 \mathrm{~cm} somewhat ochre-grey, at 335−350 cm335-350 \mathrm{~cm} a few laminations
400−420 cm400-420 \mathrm{~cm} \quad grey clay with some black bands
420−438 cm420-438 \mathrm{~cm} \quad grey clay with a sand lens at 426 cm
After this quicksand was hit and further coring started at 460 cm :
460−687 cm460-687 \mathrm{~cm} \quad light grey clay, sometimes slightly laminated
687−720 cm687-720 \mathrm{~cm} sand that could not be sampled
The pollen diagram of Apolyont is shown in fig. 6

Apolyont pollen zones

Zone 1 (spectra 1-20) shows tree-pollen values of 70−90%70-90 \%, which contrast with the

next part. Pollen of deciduous oak and Cedrus is relatively well represented, whereas values of Cyperaceae and Gramineae are very low. Zone 2 (spectra 21-28) is characterised by AP values of 40−60%40-60 \%. Various herbs reach significant values: Chenopodiaceae, Gramineae and Cyperaceae. Zone 3 (spectra 29-38): AP values increase to 90%90 \%. The herb types that were prominent in zone 2 have diminished. Zone 4 (spectra 39-41): AP values decrease to less than 50%50 \%. Salix has relatively high values, Gramineae account for over 30%30 \%.

Discussion of the Apolyont diagram

Zone 1 shows a rather uniform part of the pollen diagram; the only change happened in spectra 12 to 17 . In this part deciduous-oak pollen diminishes in favour of pine pollen and many herb types disappear. Although many tree types are found, they invariably show low values. Pine is the only important type. One should keep in mind that the local vegetation in the large lake of Apolyont was rather unimportant and at some distance from the coring location. This could explain the high proportion of pine pollen, which must have been washed in from the mountains.

The differences in tree pollen are not necessarily explained only by the general vegetation composition. The decrease in the deciduous-oak pollen and a number of herb types, together with the increase of pine pollen in spectra 12-17, may be linked to an increase in the size of the lake of Apolyont. Enlargement of the lake forces the local vegetation to retreat and favours long-distance transportation. Below 400 cm there are sand deposits in the sediment, whereas above 400 cm the sediment turns to pure clay. This may indicate that the lake became larger and that sand had already been deposited at the edge, whereas only lighter particles such as clay only reached the spot of the later coring location.

Pine forests growing east of the lake of Apolyont generally are formed by Pinus brutia, whereas further inland extensive stands of Pinus nigra occur.

The modern landscape around Apolyont Gölü is characterised by low rolling hills, flatlands and wide river valleys. This landscape must have offered ample opportunity for forest to develop, predominantly deciduous forest, partly deciduous riverine forest. Still there are not many signs that such forest was present during the time covered by the pollen diagram. Anthropogenic indicators in the pollen diagram include the set of trees well known from the Beyşehir Occupation Phase (Bottema & Woldring 1984), already discussed in this paper in the chapter on Yenişehir. The fact that pollen of sweet chestnut, walnut, manna-ash, plane and olive tree are present throughout the diagram dates the sequence after 3200 BP . It seems that since then the lowlands were exploited for timber and farming to such an extent that not much local forest survived. The question remains to what extent a typical Mediterranean tree such as olive was grown in the coastal area. Olea forms part of the so-called Oleo-Ceratonion, the characteristic plant-sociological association of the Mediterranean coastal area. Still this association does not seem to be purely “natural”, but shows anthropogenic traits (Bottema & Woldring 1984). The classical

maquis, evergreen thorny fire-resistant scrub on rocky slopes up to 300 m , is not found on the low hills and marshy flats around Apolyont.

The most significant member of the Mediterranean association, palynologically speaking, is Quercus coccifera/calliprinos. Pollen of this little oak is included in the same type as Quercus ilex, but this species plays a minor role in Anatolia. Quercus calliprinos grows from sea level up to about 1100 m after the destruction of the sub-Mediterranean deciduous forest zone. The tree produces ample pollen, which can be distinguished from that of the deciduous oaks. In the lower zone, especially pollen of QQ. calliprinos is used to indicate a Eu-Mediterranean vegetation. The fact that this type is absent or unimportant around Apolyont Gölü stresses that we cannot speak of Mediterranean vegetation here. The Anatolian lowland in western Turkey naturally carries the type of vegetation described by Quézel & Barbéro (1985): Quercion frainetto-cerris/Ostryo Carpinion.

Zone 2 shows a decrease of tree-pollen values, mainly pine, and an increase of sedges (Cyperaceae). The absence or reduction of primary and secondary anthropogenic indicators is notable. Only Chenopodiaceae increase, but the cause for this is not clear. However, the change in the diagram may have been caused purely by a change in hydrology rather than by a large-scale change in climate or a major change in anthropogenic pressure.

During the next period, zone 3, pine-pollen values recover and a palynological picture emerges that closely resembles spectra 11-17 of zone 1 . Consequently, the vegetation reconstruction of this zone will be the same.

The upper pollen spectra, assigned to zone 4 , resemble those of zone 2 , especially around spectrum 25 . Deciduous types remain at the same low values, but Pinus decreases from 80 to 20%20 \%. Pollen of grasses and sedges becomes important. Locally willow expands, given an increase of this type to 5%5 \%. Zone 4 must represent modern times, during which the area was diked and the land around the lake was cultivated. Pollen of Xanthium, a persistent weed, points to such cultivation.

No specific details are at hand about the dating of the Apolyont core. It can only be stated that the beginning of the sediment must be 3200 BP or younger because of the presence of the indicators of the Beyşehir Occupation Phase.

8.3. Gölyaka

Coring was done in the floodplain of Gölyaka on the west side of Lake Iznik. The lithology is as follows:

Only three samples produced results, a diagram of which is given in figure 10.

Discussion of the Gölyaka diagram

The core of Gölyaka only informs us on the pollen contents of three samples at a depth of 40 cm,265 cm40 \mathrm{~cm}, 265 \mathrm{~cm} and 275 cm . The spectra are to a large degree identical and have only 15−20%15-20 \% AP. The dominating herb pollen shows high values for composite pollen, including Liguliflorae, Centaurea solstitialis and Matricaria-type. These high values imply selective corrosion of the sediment which suggests rapid deposition, given the fact that the oldest sample resembles the youngest one. Still the presence of fairly high values of Abies in the lowest spectrum points to a date of about 6500 years BP, as can be concluded from the radiocarbon date and pollen content of the diagram of Yenişehir. In the Gölyaka area the forest once was like the forest of the Yenişehir area and it must be stressed that, although Iznik Lake lies not far from the coast, no signs of a Mediterranean maquis are found. Walnut, chestnut and olive are absent, which dates the spectra to before 3200 BP .

8.4. Kuşcenneti

Kuşcenneti I was collected at the edge of the lake near the nature reserve (fig. 1). Lithology: 0−165 cm0-165 \mathrm{~cm}, alternately sand and clay deposits. This core did not yield any pollen.

Kuşcenneti II

The south side of Lake Kuşcenneti carries sandy sediment. The waves from the lake break on this shore. Along the canal in a Scirpus-Typha marsh on the southwestern shore of the lake a coring was made. Locally the site is known as Hamamli-Kocagöl.

Lithology

0−110 cm0-110 \mathrm{~cm} \quad grey clay with Unio fragments at 90 cm
110−135 cm110-135 \mathrm{~cm} \quad sandy grey clay
135−245 cm135-245 \mathrm{~cm} \quad grey clay sometimes with dark patches
245−270 cm245-270 \mathrm{~cm} \quad sand that was not sampled
Three spectra yielded a sufficient amount of pollen, a diagram of which is shown in fig. 9.

Results

The samples from Kuşcenneti, taken at 240 cm,130 cm240 \mathrm{~cm}, 130 \mathrm{~cm} and 60 cm respectively, provide a very sketchy picture. The lower two spectra yield high pine pollen values, which roughly date them to the mid-Holocene when compared with the Yenişehir diagram. The upper spectrum shows a variety of tree-pollen types including Olea, Juglans and Castanea.

This dates the spectrum to around 3200 BP or after. The very low values for Quercus calliprinos-type indicate that no maquis was found in the coastal lake district.

8.5. Çakırca

The Çakırca sample was cored from a large eroded block on the shore of Lake Iznik, 150 metres south of the place where a stream discharges into the lake (fig. 1).

Lithology:

0−47 cm0-47 \mathrm{~cm} \quad dark clay
50−95 cm50-95 \mathrm{~cm} \quad dark clay, upper and lower part sandy

Results

The short core of Çakirca was taken from a large block of gyttja-like clay that had been left on the beach of Iznik Gölü (fig. 1). The pollen is very well preserved, suggesting that it was deposited in the water under favourable conditions. Since then the water level of the lake dropped by at least one metre, leaving the sediment exposed. The deposit was subsequently attacked by the waves and reduced to a big lump.

The sediment of Çakirca was dated by radiocarbon at 90−95 cm90-95 \mathrm{~cm}, yielding an age of 340±80BP340 \pm 80 \mathrm{BP} (GrN-20634). The four pollen samples, which have been analysed, inform us about the pollen precipitation of the last centuries only.

The local surroundings in the northeast corner of Lake Iznik must have been a swamp forest consisting of alder (Alnus), which produced pollen values of 15−60%15-60 \%. Towards modern times trees were cut, but on the northern shore of the lake dense stands of this type of forest are still present. The other deciduous-tree pollen can be divided into two groups: species which grow as remnants of the forest on the slopes or even on the mountains (Carpinus betulus, Corylus, Fagus and Quercus cerris-type), and trees grown for domestic purposes including walnut (Juglans), plane tree (Platanus), some mulberry (Morus) and olive. Although the slopes are intensively cultivated nowadays, with orchards and small plots of arable land, there are not many pollen grains of cereals in the Çakırca samples.

The contents of the young deposits of Çakırca are an indication that earlier centuries are not represented by the spectra from the Gölyaka sediment.

8.6. Ilıpınar

Coring in the downstream of the Ilıpınar spring produced the following lithology:
0−50 cm0-50 \mathrm{~cm} \quad yellow-grey clay
50−68 cm50-68 \mathrm{~cm} \quad yellow clay with gravel

Two samples of this short core could be analysed and are shown in fig. 7.

Results

Only two samples, at 30 cm and 50 cm , produced any pollen in the short core of Ilpinar. The location of the core downstream from the present spring suggests that the sediments are of very recent origin and that the spectra can be used to represent the modern pollen precipitation in the Ilıpınar area.

The two samples have the same pollen content with AP values of about 20%20 \%. The values for tree-pollen types as well as for herb-pollen types are similar in both spectra. The absence of Juglans and Olea pollen is conspicuous and suggests that neither species is present on the Ilıpınar Höyük. Walnut trees were not seen in the vicinity (information J. Roodenberg), but at present olives are planted in large numbers throughout the area.

8.7. Gölbaşı

The core was taken in an Equisetum marsh on soft yellow clay. A river runs into the marsh, which is being turned into a reservoir. The lithology is as follows: 0−155 cm0-155 \mathrm{~cm}, alternating deposits of yellow clay and sand. As this core produced no pollen, this information is reported here as a warning for possible further research in the area.

9. The Prehistoric Occupation Of The Iznik-YenişEHir Area

We are informed about the prehistoric habitation of the Iznik area by the archaeological investigations of J. Roodenberg (1987, 1989/1990, 1992, 1993, 1995, 1999). Roodenberg excavated the site of Ilıpınar, southwest of Lake Iznik. The body of the mound contains habitation levels from about 7000-6500 BP (c. 6000-5400 cal BC) at which time a Late Neolithic/Early Chalcolithic village was present. From about 5400 cal. BC to 3500 cal. BC the mound of Ilpınar lay uninhabited. Around 3700 cal . BC the mound was in use as a Late Chalcolithic burial ground. From c. 2800-2600 cal. BC, Early Bronze Age pithos burials are found. Around 100-200 AD there was an Early Roman farmstead with buildings around the spring. A Byzantine cemetery was used in the late 6th century and early 7th century.

Additional investigations were carried out on the archaeological sites of Hacilartepe and Menteşe for evidence about the period during which the Ilıpınar mound was not inhabited. At the site of Hacilartepe a small village or a large farmstead was found, dating from c. 2800-2600 cal. BC. Excavations at Menteşe revealed a Late Neolithic/Early Chalcolithic village dating from c. 6400-5400 cal. BC, quite contemporary with Ilıpınar Early/Middle Bronze Age burials were unearthed. For the period of 100-200 AD some Roman remains were found (information J. Roodenberg).

The excavated sites cover especially the Late Neolithic/Early Chalcolithic and the Late Chalcolithic/Early Bronze Age. In between, the Marmara district seems to have been deserted. It cannot be excluded that other sites in the neighbourhood will reveal more about the intermediate period, but so far no such sites have been found. It may be assumed that human impact upon the area was most pronounced during the phases for which habitation is evident from the archaeological results.

The periods of human habitation identified by Roodenberg are indicated in the pollen diagram of Yenişehir (Fig. 5). The Late Neolithic/Early Chalcolithic falls towards the end of pollen zone 2 and the Late Chalcolithic and Early Bronze Age coincide with zone 4 .

The Late Neolithic/Early Chalcolithic habitation took place in a landscape primarily clad with deciduous forest, although an increase in the amount of coniferous forest is in evidence. Modern parallels suggest that the species concerned were Pinus sylvestris and/or Pinus nigra. The increase in pine pollen is explained as an increase in pine trees. However, it is not clear whether human activity, for instance lumbering of deciduous trees, caused this increase of pine, or whether people settled in the Iznik area because of some external stimulus that also triggered pine growth. Spectra 11-13 show a pronounced decrease in deciduous-tree pollen, including hazel, deciduous oak, and lime. The decrease started around 7000 BP and lasted a few hundred years. The impact must have been great, as the diagram shows and one wonders whether Middle Chalcolithic habitation (phaseVB), which in Turkey was very dense, must be held responsible for the destruction of the forest, the Late-Neolithic population already having taken the initial steps. After the end of the habitation phase the Yenişehir basin turned from open water into a swamp. It is not clear whether the forest destruction caused a hydrological change in the basin. After a short period of peat deposition, deposits in the basin returned to clay, indicating deeper water.

During the Late Neolithic/Early Chalcolithic, houses in Ilıpınar were first built from wood, but in the course of time house building shifted to mud brick. In the vegetation history, the time covered by spectrum 11 to 13 may have been when wood became scarce. The disappearance of the forest must be only partly the result of logging. A reduction in tree growth of this extent is likely to have been caused especially by large-scale herding and intentional burning for the creation of pasture.

Turkey was densely populated again during the Late Chalcolithic/Early Bronze Age, which in the Yenişehir diagram falls in a larger part of zone 4. Anthropogenic pressure seems to manifest itself in a pronounced drop in tree pollen, from which it is concluded that deciduous forest was fast disappearing. The pine pollen first remained at high percentages, maybe because other tree pollen disappeared, but towards the end of this habitation period pine-pollen percentages also dropped. The low AP values started at 4880 BP , a date calculated from the sedimentation rate between the two radiocarbon dates. This date correlates very well with the isotope date of the beginning of Phase IV, the Late Chalcolithic burial ground.

References

Bottema, S., H. Woldring and B. Aytuğ, 1995 - Late Quaternary vegetation history of northern Turkey. Palaeohistoria 35/36, 1993/1994, 13-72.
Bottema, S. and H. Woldring, 1884/1986 - Late Quaternary vegetation and climate of southwestern Turkey, Part II. Palaeohistoria 26, 123-149.
Browicz, K., 1982 - Chorology of trees and shrubs in south-west Asia and adjacent regions, Vol. 1. Polish Academy of Sciences, Institute of Dendrology. Polish Scientific Publishers.
Roodenberg, J.J., 1987 - Ilıpınar in the prehistory of northwest Anatolia. Palaeohistoria 29, 203210 .
Roodenberg, J.J., L. Thissen and H. Buitenhuis, 1989/1990 - Preliminary report on the excavations at Ilıpınar in NW Anatolia. Anatolica 16, 61-144.
Roodenberg, J.J., 1995 - The Ilıpınar Excavations I, Five seasons of fieldwork in NW Anatolia, 1987-91. Nederlands Historisch-Archaeologisch Instituut te Istanbul 72, 1-180.
Roodenberg, J.J., 1999 - Ilıpınar, an Early Farming Village in the Iznik Lake Basin. In: The Neolithic in Turkey, edited by Mehmet Özdoğan. Istanbul, 193-202.
Walter, H., 1965 - Klimadiagramme der Türkei. Stuttgart-Hohenheim.
Weeda, E.J., R. Westra, Ch. Westra and T. Westra, 1987 - Nederlandse Oecologische Flora, Wilde Planten en hun Relaties 2.
Zeist, W. van, H. Woldring and D. Stapert, 1975 - Late Quaternary vegetation and climate of southwestern Turkey. Palaeohistoria 17, 53-143.

Fig. 1. Map of the area with the locations of the coring sites and the surface samples. 1. Yenişehir. 2. Apolyont Gölü. 3. Gölyaka. 4. Kuşcenneti. 5. Cakırça. 6. Ilıpınar. 7. Gölbaşı. I-VII refer to the surface samples treated in section 6.2 (1-7).

Fig. 2. Climate diagram of Bursa (Walter, 1965).

Fig. 3. Vegetation map of the area after Quézel & Barbéro (1983).
Legend:

1. Deciduous forest of Quercus cerris and Quercus frataretto, locally with Ostrya carpinifolia and Quercus pubescens, more rarely with Castanea sativa, Abies cephalonica and Fagus sylvatica.
2. Often degraded forest of Quercus macranthera, Fagus orientalis, Carpinus orientalis and Carpinus betulus.
3. Forest and shrub vegetation of Quercion ilicis and Pistacio-Rhamnion with sclerophyllous oaks (Quercus ilex, Q. calliprinos), sometimes Q. infectoria and Q. macrolepis.
4. Forest of Carpinus betulus, C. orientalis, deciduous Quercus, locally with Fagus orientalis, Ostrya carpinifolia and Abies bormmuelleriana.
5. Fagus orientalis and Rhododendron ponticum.
6. Pinus pinea.
7. Pinus nigra subsp. palassiana.
8. Fagus orientalis, locally with Abies bormmuelleriana and Pinus sylvestris.
9. Quercion ilicis.
10. Abies bormmuelleriana.
11. Cultivated alluvial plain with remnants of Quercus macrolepis.

Surface samples 1988