POLLEN ANAL YTICAL INVESTIGATIO NS IN THESSALY (GREECE)

Sytze Bottema

CONTENTS

1. INTRODUCTION

z. GEOGRAPHY, GEOLOGY, CLIMATE AND VEGETATION OF THESSALY z. r. Geography and geology z.z Climate z.3 Vegetation

3. METHODS

4. RADIOCARBON DATES

5. LAKE XINIAS

5. i. Coring sites 5.z. Coring 5.3. Lithology

6. DISCUSSION OF THE XINIAS DIAGRAJ\'I 6. 1 Plenjglacial 6.z. Late Glacial 6.3. Postglacial

7. LAKE VIVIIS 7. t. Coring and lithology 7. z Pollen sum

8. DISCUSSION OF THE VIVIIS DIAGRAM

9. LOCAL VEGETATION 9. 1. Marsh and water plants of Xinias 9.z. Pediastrum 9.3. Miscellaneous 9-4- Local vegetation of Viviis

to. CORRELATIONS AND COMPARISONS

10. i. Correlation of Greek diagrams ro.z The \'\lurm glacial steppe vegetation of Greece ro+ Comparison with Anatolian and Iranian glacial steppe vegetations

1 i. ARCHAEOLOGICAL IMPLICATIONS OF THE PALYNOLOGICAL INFORMATION

i i. 1. Palaeolithic r 1.z. Neolithic and younger periods

12. SUMMARY

1 3. REFERENCES S. BOTTEi\fJ\

l. INTRODUCTION dis (1975). \X/right (1972) studied the younger Post­ glacial deposits from the Peloponnesos and adjacent

From l 96 5 on, the number of palynological investi­ mainland.

gations in Greece has constantly increased (fig. l ) . Apart from localities which have yielded palyno­ Van der Hammen et al. (1965) and \Xlijmstra (1969) logical information already published (fig. l ) same treated very deep cores from the Plain of Drama. places are mentioned which are still the subject of Co res of Postglacial age from Philippi and Lake Ko­ research. The Lakes of Trikhonis, Vegoritis and

pais were studied by Greig and Turner ( l 974), and Volvi were cored by P. R. Readman c.s. (Geophy­ Turner and Greig ( l 97 5 ). The la te Quaternary period sical Department, University of Edinburgh). These in the northwestern part of Greece was studied by cores were taken primarily for palaeomagnetical ana- the present author (Bottema, 1974). In central 1 ysis. It will be evident in the near future to what Greece, in the area of Katerini and Trikkala Post­ extent palaeomagnetic and palynological research glacial sediments were investigated by Athanasia- supplement or support each other.

In l 96 3 a coring was performed in Lake Viviis as part of a larger expedition (Bottema, l 974) led by \Y./. van Zeist and with the kind help ofR. J. Rodden. Ona secondexpeditionin 1965, Mr. P. Petsas (Ephor of Antiquities for Western JVIacedonia for the Greek Archaeological Service) suggested Lake Xinias as a possible coring si te. Taking part in this expedition were van Zeist, R. ]. Rodden, D. French (now \XI. Director of the British Scbool of Antiquities in An­ kara), Casparie (Biologisch-Archaeologisch \XI. J\. Instituut, Groningen) and tbe present author. In this paper the palynological results ofthe cores from Lake Viviis and Lake Xinias will be discussed. The author is much indebted to Mrs. S. l\'1. van Gelder-Ottway who improved the English and to Mrs. G. Entjes-Nieborg for the preparing of the manuscript. Drawings were prepared by Mr. H. Roe­ link, J. Smit, Jacq. Klein and the present author.

2. GEOGRAPHY, GEOLOGY, CLIMATE AND VEGETATION OF THESSALY

2. 1. Geography and geology

Thessaly consists of the plain of that name, the sur­ rounding hills and mountains and inlets of the Aegean. The area is more or Jess delimited by the 2 . Fig. 1. Palynological investigations in Greece. towns of Larisa, Trikkala, Larniaand V ol os (fig. ) Open circles: published results available; Black circles: results For details on the geology the reader is referred to to be published; Dotted circles: results treated in this publica­ Schneider (1968) and Voliotis (1973). tion. Situated in the east (from south to north) are the 1. Gravouna. 2. Tenagi Philippon. 3. Volvi. 4. Giannitsa. j. Magnisia Peninsula, the Volos Plateau and the Ma­ Edessa. 6. Vegoritis. 7. I'"himaditis. 8. Kastoria. Litochoro. 9. vrovuni l\1Iountains. In the northeast the Ossa l\1oun­ 10. Ioannina. 11. Pertouli. 1 z. Viviis. 1;. Xinias. 14. Voulkaria. ' i 5. Trikhonis. 16. Copais. 17. Kaiafa. 18. Osmanaga. i o. Aghia tains reach an altitude of l 978 m. Across the Tempe­ 20 Galini. valley the Olympus, consisting of a higher part (Ano Pol/e11 a11a61ticali111Jestigatio11s i11 Thessab

Olympus, 2917 m) and a lower part (Kato Olym­ Ossa and Kato Olympus have about the same cli­ pus, 1 5 87 m) can be seen. The border along the north­ mate. Precipitation at the northeastern foot of the west is formed by the Chasia Mountains and that mountains amounts to 600-800 mm. At the summit along the west side by the Pindus chain. Finally, 1 ooo-r 200 mm is measured. Southern, southwestern situated at the southern edge are the Tymfristos and western slopes are much drier. On lower slopes (23 15 m) and the Oeta (highest peak the Pyrgos, next to the Thessalian plain precipitation hardly 2 1 5 2 m). Peaks consisting oflimestone show "Karst" reaches 500 mm, while on the east side of the above­ influence. According to j' vfonopolis (Voliotis, 1 976 a, mentioned peaks it amounts to about 900 mm.

b) traces oftwo glacial periods can be observed above Tbe mean annua! temperature is r r 0C. The differ­ 2000 m. The plain ofThessaly is situated atan altitude ence between summer and winter temperature in­ of 60-1 Go m above sea level. There are two parts, creases from east to west. Prevailing winds come an upper and a lower plain, in the southwest and from the northor northeast over tbe Gulfof Therma. northeast respectively. The former Lake Viviis, at Western slopes thus lie in the rain shadow. Meteoro­ about 50 m altitude the lowest point in the plain, logi ca! stations are present in Trikkala (r r 3 m), has no direct outlet to the sea . .Lviostof the plain is Lamia (69 m), Volos (roo m), Larisa (67 m) (\\"lalther covered with Holocene deposits. The rivers coming and Lieth, 1967). Rainfall is restricted almost com­ from the surrounding mountains partly removed pletel y to the winter months, autumn and spring. sediments forinst ance when the gorge of the Tempe July, August and September average 20 mm. At was formed. that time temperatures are maxi mal, so the ecological One of the most important rivers is the Pinios. drougbt is severe. The average January temperature This river discharged into the plain in Lake Viviis is relatively high, 5 -4°C forTri kkala, 5.5 °Cfor Larisa (L. Karla) but broke through the valley of the Tempe (Greece, 1944) and about 7°C for Volos and Lamia I, to the sea at the end of the \'\/i.irm(Sch neider, 1968). (Walther and Lieth, 1967). The minimum January

At least three periods of a warm and dry climate clur­ temperature ever recorded for Larisa is - 1 3°C. ing the Wi.irm are indicated by Schneider, but he considers it not unlikely that the picture is obscured 2.3. Vegetation by tectonics. \X/hen the Pinios discharged to the sea not all tbe water went there. Tilting of part of the Accorcling to Voliotis' information (1973, r976a, plain formed a depression, the later Lake Viviis. 1 976b) and especially his vegetation map of Greece Schneider estimates that at some time during the (1973) a tentative vegetation map (fig. 2) of the area Postglacial the lake must have been 20 m deep. The has been composed. lake must therefore have been of large extent, which In studies on the vegetation of the j\ifedi terranean in fa et coulclals o be concluded from the distribution and also of the Near and Middle East .a constantly of the Neolithic settlements. It is, however, doubtful repeated complaint is the absence of natura] plant whetber this lake was very deep cluring the period cover due to destruction by man. This problem is

spanned by the pollen diagram taken there. now rapid I y becoming worldwide and is by no means ln the course of time tbe plain driecl up, so finally restricted to these areas. Tbus the vegetation map also the lake was affected (Greece, III, 1945). The is in faet a reconstruction of the natura! vegetation lowering of tbe water-table was not constant and under ideal conditions, disregarding human inter­ fl uctuations occurred. In 1 960 the drainage of the ference. The botanical composition of the forests etc. remnants was started. on the various mountains is rather variable. Details can be fo und in Voliotis' studies. 2.2. Climate Along the coast (fig. 2) an Oleo-Ceratonion is fo und. This vegetation zone remains narrow, never The climate of the Thessalian plain is a modified penetrating more than a few kilometres inland. It rviediterranean climate. The temperature and rain­ requires an average annua] temperature of about 17- fall figures in the area show features ofa continental r 80C and almost 600 mm precipitation. Pi1111s hale­ type. The summers are suffocatingly hot (Greece, I, pemis and Pi1111s 11igra are important constituents of r 944). The winters are cold frost is common. this zone but in practice only a very much degraded 21 S. BOTTEMi\

Fagion moesiacae

1111111 Quercion il i cis

fm Abietion cephalonicae

� j Vacc in i o -Picetal ia

�1-1-1-1 -1-1-1-1 Oleo - Ceratonion 1-1-1-1- Quercion confertae

I I I I Ostryo - Carpinion I I X Coring siles

Surface sample

Fig. z. Vegetation map ofThessaly (after Voliotis, 1973); coring cion ilicis is found. Here the average annua! temper­ sires and surface sample locality. ature is about 14- 15°C. 1v1ost of Thessaly, the plain and bordering foot­ hills, lies in the zone of the Ostryo-Carpinion orien­ stadium is found. This stadium is known as Ph1]1g,a11a. talis. The mean annua! temperature of 12-15 °C and In this zone two subzones are distinguished : the especially winter-minima of - ro°C and even lower, Oleo-Ceratonietum and Oleo-Lentiscetum. Along limit the eu-Mediterranean vegetation. Neverthe­ the coast at the height of the Ossa and Mavrovuni less, the evergreen Q11erc11s coccifera plays an i mpor­ 22 Mountains and sou th of the town of V olos, a Quer- tant part in this vegetation in addition to deciduous Pol/e11 a11a[ytical i11vestigatiom i11 Tbessa61

elements such as Ql!ercm p1!besce11s, OsfJ]'a ca1pi11ifo lia, The Viviis diagram was calculated on the basis of Ca1pi1111s orie11talis, which partly constitute the Coc­ such a pollen sum. cifero-Carpinetum orientalis association. In submontane and montane areas, the Fagion moesiacum is found, subdivided into associations of a Fagetum or combinations of Fagm and Abies (borissi-regis). .l'viediterranean montane forest, dominated by Abies cepba/011ira, is found in the moun­ IOkm tains bordering the southern part of the area studied. A vegetation of Pi11l!s /Jeldreic/Jii belongs to the subalpine zone. A vegetation type which can hardly be ascribed to any zone is the Pinion nigrae. Pi11l!s '·"", 11igra occurs from about 100- 1 800 m altitude in ·"\ Greece. Relicts of Pi11l!s S)'lvestris are found on the

" Olympus. Alpine vegetations in the area are restrict­ ,- ---" " ,'-" ed to the upper part of the Olympus. They belong ·-�:-:, to the Acantholimo-Astragaletalia zone, charac­ terized by spiny, cushion-shapecl species. s""�/?)

__", �-"-\ · - - - ,..·-2____ .-�00-- ';. 3. METHODS

Clay and gyttja samples were prepared using a flota­ tion methocl (Bottema, 1974) after which sometimes a treatment with hydroEluoric acicl fo llowed depend­ ing on whether any minerals were left. Afterwards acetolysis according to Erdtman was applied. Peat samples were boiled in 10% potassium hydroxide followed by acetol ysis. Samples were stained wi th Fig. 3. i\lap of the Xinias area, showing the coring sire. safranine and mounted in silicone oil. The pollen sum on which the percentages in the diagrams are basecl, inclucles all pollen types which are thought to be of non-local origin, viz. trees and 4. RADlOCARBON DA TES herbs, exclucling the pollen types of marsh and water plants. Racliocarbon measurements were carried out by the In the Xinias I diagram (fig. 4) three diffe rent Groningen C 14 laboratory for fo ur levels : pollen sums were tried in order to stucly the effect Xinias (Lamia) I.50- I.60 m: upon the AP/NAP ratio. Pollen sum I includes the (organic fraction) 10,680 ± 90 B.P. (GrI 6889) arboreal pollen types and the series Arte111isia-Cerea­ (calcareous fraction) r l,I 50 ± 130 B.P. (GrN 6888) lia-type. Pollen sum Il is fo rmed by the AP and all The date of the organic fraction is considered to be non-local herb pollen types excluding the Grami­ the actual one, that will be referred to. neae. Pollen sum III comprises those from sum II Xinias (Lamia) 3.70- 3.80 m: 21,390 ± 430 B.P. but including Gramineae pollen. The curves run (GrN 6886) more or Jess parallel, only in zone Y do they show Xinias (Lamia 5.70- 5.80 m: 25,620 400 B.P. ± almost the same value. This is to be expected as the (GrN 6887) types which are excluded, have unimportant values. +5000 Xinias (Lamia) 1 3.40-13.60 m: 46,900 The most complete pollen sum, that including the ----3060 most types, is preferred as a basis for the diagram. (GrN 6882) S. BOTTEMA

5. LAKE XINIAS 6. DISCUSSION OF THE XINIAS I DIAGRAM

5.l. Coring site 6. l. Pleniglacial

The drained lake is si tuated 3 km west of the village To facilitate the discussion the diagram will be of Xinias, 30 km northwest of Lamia (figs. 2 and 3) divided into pollen zones and subzones. The division at an altitude of 500 m. \X/ hen the coring was made, is made on the basis of the most characteristic features in l 96 5, drainage canals appeared to have been dug of the arboreal pollen (AP) curve, the curves of Pi1111s very recently. The smal! plain (the former lake bot­ and Querms and to a !esser extent on the behaviour tom), now agricultural land, is surrounded by hills. of other tree-pollen and herb-pollen curves. Of two cores collected one will be presented. A main division into two parts can be made by means of theAP/NAP ratio. Spectra l-76 show most­ 5 .2. Coring ly high herb pollen percentages whereas spectra 77- 9 l dernonstrate high arbo real pollen values. The coring Xinias I took place at the end of a drain­ The first part can be divided again in to three parts, age canal running from east to west in the centre the lowest part showing still moderate AP values, the of the area, 7 cm above the water level. A Dach­ second part showing lower AP values which fluc­

nowsky ø 3 6 mm was used up to 10.40 m. The cor­ tuate sharply, whereas the third part demonstrates ing was made up to l 3 .80 111 witha Livingstone corer continuously low arboreal pollen percentages. Mi­ ø 28 111m. Another core, 84 cm above the water nor fluctuations make a still more detailed subdivi­ level, was cut out of the ditch bank. s1on necessary. The main zonation is as follows: 5.3. Lithology Zone V (spectra 1-10) Zone ( spectra l 1 -37) \XI (+ )0.80-(+)o.65 m above the water level, brown Zone X (spectra 38-60) soil, disturbed Zone Y (spectra 61-67) (+ )0.65-(+ )0.60 m grey clay, with fi ssures due to Zone Z (spectra 77-9 1). shrinkage; fi lled with brown soil from above (+ )0.60-(-)J.57 m grey gyttja, at the botto111 Zo11e (spectra V 1-10) clay with organic and calcareous remains, a light brown band at 1.19-1.20 m Zone V is characterized by relatively high AP values,

l.57- J.75 111 blue-grey, sandy clay maximally 30-60%, although NAP percentages J.75- I.80 m blue sand generally dominate. Because of these fluctuations

l.80- 2.07 111 blue, sandy clay three subsones are distinguished. 2.07- 2.70 111 blue-green cla y

2.70- 3.06 111 blue-grey clay 511/Jzom V 1 ( sp eclra r-3) demonstrates high Pi1111s 3.06- 5 ·75 111 grey cla y-gyttja values, maximally 34. 1 °lo andQ11ermscerris-type from 5.75- 5.85 111 brown gyttja 10-20%. U/111m, Co1)'!11s, )1111ipems, A/Jies and Fag11s 5 .85- 5 .96 111 grey gyttja are represented with low percentages. 5 .. 96- 6.04 111 brown gyttja The oak pollen mentioned with the nameQ11erms 6.04- 9· l 3 m dark-grey clay-gyttja cerris-type includes all deciduous oak species occur­ 9. 13- 9.48 m yellow-brown gyttja ring in Greece at the present time. This type is iden­ 9.48-12.10 m dark-blue/ grey clay-gyttja tical to 011erms rob11r-type (at 400 X under a light­

12.J0-12.57 111 yellow-brown gyttja microscope), but the name Q11erc11s cerris-type is

12.57-12.60 111 dark-grey clay-gyttja chosen as this species was regularl y met with during l 2.60-1 2 .87 111 brown-yellow gyttje field work in Greece (Bottema, 1974). 1 2.87-12.90 m dark-grey clay-gyttja The herbal vegetation is represented with espe­ 12.90-13.60 m dark-grey clay cially important pollen values forGrarni neae, Che­ nopodiaceae, and Arte111isia. The lowest part of sub- Polle11 a11ab1tical i11vestigatio11s i11 Thessab'

zor.- e V1 is radiocarbon dated 46,900 __ . This in AP values, mainly Qnerms erris-type, Co1]'/11s, c me�11s that the radiocarbon sample yielded an al- Abies, Fag/IS, and Pi1111s. 1110 -=::;t infinite date and thus may be much older than B.P. 11b o11e The picture in this sub­ 46,S::Joo S z W3 ( sp rctm 27-35). zone strongly resernbles subzone \Y h . The bound­ s11b ( sp t ) In this zone Pi1111s values ary of zone with zone X is placed after the last �one V2 er m 4-7 . \Y/ har � dirninished whereasQnerms cerris-type reaches a small peak of the AP curve. ma�imum of 30%. The tree pollen maxirnumof sub­ zorL c: V 1 and that of subzone V2 are separated by Steppe vegetations dominated the Xinias area du ring ah .:i:b-pollen maximum ofabout 70% in spectrurn 3. zone \'!/ . They extended from the level of the lake on-1paring with subzone V1, some tree species at 5 oo m over the slopes of the surrounding rnoun­ apj> e:ar or increase in pollen percentages, for in­ tains and down to the Thessalian plain. stan ce Cfllpi1111s betlli11s, and Cr11pi1111s oriet1talis/Ost1 ]'a. In the Balkans hardly any natural steppe vegeta­ To�;ards the end of subzone V 2 a decrease in AP tions are found at present. The forest steppe of the perc::: entages is visible. This decrease forms the Danube lowlands (Horvat et al., 1974) is still a mat­ botJ �dary with the third and last subzone V 3. ter of discussion and is very difficult to connect with the Greek steppe. Thus it rernains obscure which s11b:;z;.011e sp t J-10). Tree pollen values are plant species are represented by the pollen types in V; ( ec m low·er than in subzone V 2 but the general tendency the glacial spectra. is same. At the beginning of zone V pine forest In loannina on the west side of the Pindus Moun­ tl---i e occ� rred at higher altitudes whereas oak was Jess tains, fo rest was postulated at an altitude of about abu odant, maybe in a forest steppe together with 800 m during a comparable period (Bottema, 1974) Arre111isia. The other trees rnentioned were either lasting from about 40,000- 15,000 B.P. As the east rare or growing at greater distances in the moun­ side of the Pindus catches far Jess rain than the west tairrs in fav ourable locations. The vegetation may side and therefore has a more continental climate, hav e been in a state of unstable equilibrium. As soon forest must have been restricted to levels higher as t }--:i. e climate changed slightly either steppe vege­ than 800 111 . No trees were found close to the lake tatio 11S or forest would have gained terrain depend­ where the herb cover mainly consisted of Artr111isia, ing upon the direction of the climatical change. Chenopodiaceae, various Composi tae and other Prec ipi ration would have been a major factor. species mentioned above. Minor climatical fluctuations probably favoured

( . p lra 11-37) same tree growth on suitable locations closer by, as z011 e W i ec can be concluded from low AP peaks. Such periods Zone resembles zone V but also shows same dif­ lasted 700-1 ooo years, as concluded from the sedi­ W/ fere n- ces. The lower AP values are conspicuous. The mentation rate. san"l.e fluctuations in the AP/NAP ratio, as can be :M any of the AP peaks in zones V and \Y/ show seen in the preceding zone, can be found here. On a certain asymmetry. Forest seems to have spread the b asis of especially the Q111!J"c11s and Pi1111s values rather abruptly, and to have disappeared afterwards zone \Y/ can be divided into three subzones. more gradually. This may point to different condi­ tions depending on whether the forest vegetation s11bzo!le (spectra is characterized by domi­ was being established or maintained. Conditions for IV 11-23) nant lAP values, mainly due to Arte111isia, Cheno­ establishing are more demanding than those for podiaceae, and Gramineae, but to a !esser extent also conservation. \'\lithin the forest a certain micro­ to Ce11/a11rea solstitialis- type, Pla11tago 111ariti111a- type, climate may have prevailed where germination was Tha Cerealia-type, Ligu iflorae, various Tu­ fa voured compared to conditions in an open plain. /i ctr11111, I buli fl orae, and B1111i11111-type. Zone \Y/ would have lasted from about 43,000- 28,000 B.P" as interpolated from the radiocarbon demonstrates an increase dates and assuming a constant sedimentation rate. s11b oJ/f11 72 ( sp ectm 2.;-26) z 25 S. BOTTEMA

Zo11e X ( sp ectra 38-60) a treatment on the Late Glacial in the Eastern Medi­ terranean and the Near East (Bottema, l 978). Zone As mentioned above, the boundary with the preced­ Y forms a transition between a long period with low ing zone is placed after the last (minor) AP peak, tree-pollen values and a contrasting younger period which measured only 25 %· In zone X the AP/NAP with high tree-pollen values. ratio shows no fluctuations, resulting in a smooth The reason fordistin guishing a separate zone Y curve demonstrating low AP and conseguently high is the increase of conifer pollen and Chenopodiaceae NAP percentages. The continuous curve of J-lipp o­ and the decrease of Arte111isir1. Fluctuations within phai:'draws attention, although values are under 1 %· this zone make a fu rther division into subzones This shrub must have been rather abundant as it is necessary. very much under-represented in the pollen rain. According to Freitag (1977) ipp opha would have S11bzo11e sp ectra '61-65). Pi1111s and Ahies percent­ 1-l i:' } '1 ( grown along water courses. Pollen types fo und in ages increase together with Chenopocliaceae; Gra­ zone X are the same as in the preceding zone. How­ mineae show a decline for the first time. ever, tree pollen is consistently low, suggesting that

types such as 011erc11s ce/ '/ 'is-type, Pi11m, and probably S11hzo11e } '2 ( sp ectra 66-69). AP values decrease con­ also Be/11/a can only be produced by long-distance siderabl y. Chenopodiaceae are dominant, at values transport. The pollen distribution of )1111ipems is not of about 90% whereas Arte111isia drops from about very good and the continuous curve with values up 2 5 to about l 0% and Jess. to 1% point to a regular occurrence of some repre­ sentative of this genus. .S11bzo11e '3 sp ec/ra 70-76). AP values increase again, 1 ( The open character of the vegetation is illustrated mainly because of Pi1111s, but Q11erms cerris-type , by a variety of herb-pollen types. Various Umbel­ Co1]'/11S, Abies, Bet11/a, and A/1111.r also demonstrate

liferae, including Femla-type, Caryophyllaceae and higher percentages. J-lelia11!he11111111 are very well represented. In addition Tubuliflorae including the very important Arle111isir1 At the beginning of zone Y coniferous forest started

(mainly herhr1-r1/br1-type), Liguliflorae, Plr111/r1go lc111- to spread over the mountains. Apparentl y conditions reo/r1/r1-type and various Chenopodiaceae reach rela­ were not suitable fo r any increase of deciduous oaks. tively important values during this zone. During zone X temperatures were low but steppe The climate must have been extreme. Temper­ vegetations were favoured especially by drought. atures were low and so was the precipitation. Only The establishment or increase of trees would have towards the end of zone X some slight improvement benefitted most by any increase in precipitation. An occurred, illustrated by the reappearance or increase increase in temperature would have caused even of some pollen types such as Cr1 1pi1111s orimtalis/ more severe drought. At higher altitudes in the Os!J)'rl, U/11111s, and Cot]'l11s. mountains moisture did not play such an important Two radiocarbon da tes are available for zone X: role as at lower al ti tudes. Since only coniferous spe­ 25,620 ± 400 B.P. (G rN 6887) for the level of 5 .70- cies increased and not trees of i ntermediate or lower 5.80 m levels it is reasonable to conclude that there was a

21,390 ± 430 B.P. (GrN 6886) for the level of 3.70- rise in temperature at the beginning of zone Y. At 3 .80 m. lower altitudes the increase in temperature caused Using all available dates to interpolate, zone X the spread of steppe vegetation, in this case Cheno­ Jasted from 28,000-15,000 B.P. pocliaceae. It is not impossible that some of these chenopods grew on the shores ofLake Xinias during periods ofsu mmer drought (\Xlasilikowa, 1 967 ). Sur­ 6.2. Late Glacial face sample studies from Syria point however to high pollen values for Chenopodiaceae where no lakes Zo11e }' (Jpectra 6r-67) are found at all (Bottema and Barkoudah, in press). During tbe next subzone Y 2 conditions became z6 The discussion on this period can also be found in very unsuitable for tree growth. It is probably that Pollen ana61tical investigatiollS in Thessa61 tre s declined in the area and only grew in more 6.3. Postglacial suir::: able parts of the Pindus .Mountains. Steppe vege­ tati.c:in prevailed in the Xinias area, a vegetation in Zo11e Z (spectra 77-91) wh- ch Chenopodiaceae dominated. This period seei::-i1s to be even more unfavourable for plant cover The boundary between zone Y and zone Z is chosen thaC1 the whole Glacial period discussed befare. May­ where a rapid increase of the AP is found, between be �n increase in temperature caused even more spectra 76 and 77. This level yields the date of 10,680 severe drought. Coarse material appears in the sedi­ B.P. mentioned above. ment towards the endof subzone Y z and also during Zone Z covers the period which in northwestern sub.:zone Y3 (see also description of lithology). Un­ Europe is known as the Postglacial. The main charac­ der the extreme climate of subzone Y z plant cover teristic is the high tree-pollen value throughout this must have been very sparse on the slopes around zone. The Postglacial in Greece is studied more often Lalce Xinias. This facilitated erosion when showers, and in more detail than the older periods. Thus more alth. ough very infrequent, transported sand, whereas information is available forcompari son. Zone Z dis­ in 0ther periods the vegetation protected the soil plays characteristics also found in diagrams for Ioan­ bett: er so that mostly clay was deposited. nina, Edessa, Khimaditis (Bottema, 1 974) and

Less extreme conditions were fo und after this Tenaghi Philippon (Wijmstra, r 969) etc. The follow­ peri od during the third subzone Y 3. The conditions ing subzones have been established: Subzone Zr ofsubzone Y r seem to have returned as a comparable (spectra 77-78), Subzone Zz (spectra 79-8 r ), Subzone poll en picture shows up. There is one important dif­ Z3 (spectra 8 2-8 5 ), Subzone Z4 (spectra 86-8 7 ), Sub­ fere nce: Arte111isia does not return to its former zone Z5 (spectra 88-91). valaes but loses even more ground and is maybe repl:aced by Gramineae. Some deciduous species in­ S11bzo11e Z1. During this subzone 011erc11s cerris-type crea. se towards the endof subzone Y 3 demonstrating pollen percentages increase rapidly, together with the development of a different climate. Coniferous same other tree types. From the j1111iper11S curve it species increased again at higher elevations. may be concluded that initially the fo rest vegetation S uch a situation as described must have been was still open. Evergreen oaks started to play a role induced by a change in climate, in this case an in­ together with Pistacia and the submediterranean crea se in precipitation as well as a rise in temper­ species Fraxi1111s om11s. Conifers were not important ature. during this subzone while steppe plants delivering The beginning of zone Y is dated about 15,000 pollen types such as Arte111isia, Ep hedra distacl!)'a­ B.P. by interpolation. This date and the radiocarbon type, and Cmta11reasolstitiali s-type decreased or even date of 10,680 ± 90 B.P. (GrN 6885), at the end of disappeared. At the same time pollen is found from this zone, define zone Y as a parallel of the Late­ Poteri11111/Sa11g11isorba 111i11or. It is not clear \V hether Glacial as we know it from northwestern Europe. this pollen type belongs to the maguis-plant Poteri11111 It must be stressed that the sequence of Ol der Dryas, sp i11os11111 or to Sa11g11isorba 111i11or. This type is found Alleri:id andYo unger Dryas ofnorth western Europe at the beginning ofthe Postglacial, not onlyi n Greece is demonstrated here as an unfavourable period be­ but also in other Mediterranean and Near Eastern tween two periods with a milder climate. This is in countries. Both species are light-demanding but ac­ marked contrast to northwestern Europe where in cording to Horvat et al. ( 1974) Poteri11111 sp i11os11111 faet the reverse is found. A possible explanation for ( Sarco poteri11111 sp i11os11111 ) is restricted to coastal areas these phenomena (see also Bottema, r 978) is the as it is sensitive to frost. Sa11g11isorba 111i11or, among diffe rence in effect of an increase in temperature other species, is fo und in vegetations together with upon two areas so far apart and where the annua] 011erc11s p11besce11S. Sa 11g11isorba 111i11or can probably be distribution of the precipitation is completely dif­ expected in subzone Z I -vegetations. ferent. In the Xinias area, with a pronounced steppe climate, an increase in temperature caused even drier S11bzo11e Z2 shows very high values ofQ11erms cerris­ conditions, resulting in a situation as described for type together with relatively important values of subzone Yz. Pistacia,J1111ipems, and Poteri11111-type. Especially the S. BOTTEM/\

last type measures up to 1 5 %· The boundary with again. Plants producing this type were probably subzone Z3 is placed where Carpi1111s orie11talis/ Ost1J1a favoured by man. The Ericaceae may form part of a and Co1)'!11S increase. The sudden expansion of Car­ maguis resulting from wood-felling activities espe­ pi1111s orie11talis/ Ost1)'a is observed in other Greek dia­ cially in the coastal area. The increase of Abies (5% grams (Bottema, 1974; \Xlijmstra, 1969) and dated and more) also may bethe result ofsecondary growth about 6500 B.P. Interpolation of the radiocarbon of this species where other ( climax) forest had been date obtained for material from a depth of r.5 0-1 .60 felled (Athanasiadis, 1976).

m gives about the same result. Values for 011Cf'C11S coccijera -type increased during subzone Z5 suggesting a spread of evergreen oak S11bzo11e Z; . The AP percentages and the significant where deciduous mixed forest had been destroyed.

value of Poteri11111-type point to rather open forest. In spectra 90 and 91 Olea is foundin relatively im­ The climate responsible for this vegetation would portant numbers (fig. 4). This pollen must have at least have had dry and hot summers. Pistacia and originated from olive groves somewhere in the the possibility of Poteri11111 point to winters with a neighbourhood and are a clear sign of increasing j\ifediterraneancharacter, at least not extremely cold. human occupation. The winters in the plain are too Arboreal pollen still increased during subzone cold for Olea but on the southwest slopes of the

Z3, althoughQ11erc11S cerris-type had to give way to Olympus olives occur up to about looo m. The Ca1pi1111s orie11ta/is/ Ost1J'a. Ahies and Pi1111s show coastal area is important for olive-oil production somewhat higher values as is the case with Fraxi1111s (Geographical Handbook, l 944). excelsior. Pla11tago la11ceo/ata-type formsa closed curve It is difficult to ascertain whether the complete at the beginning of this subzone. As the author Postglacial is covered by the diagram. Spectra 90 knows no other Pla11tago species from the area that and 91 yield Casta11ea pollen ( o. 3 and o. l %) while

match the la11ceo/ata- type, the curve may represent spectra 87 and 91 show J11gla11s. The firstappearance this species. Using Pla11tago la11ceolata as an indicator of these types together with Plata1111s takes place of agriculture, there are signs of such human in­ about 1450 B.C. or la ter (Readman and Bottema, in fluence at about 4500 B.C. (6500 B.P.). For part of prep.). Comparing wi th diagrams ofEdessa and Khi­ this period, beech must have been present some­ maditis (Bottema, 1974), Pertouli and Litochoro where as very low percentages of Fag11s pollen are (Athanasiadis, 1975) spectrum 87 can be dated 3000 found. B.P. or younger (table l ). On the Tymfristos there are large forests of Casta-

S11hzo11e Z4 shows high values for 011erc11S cerris-type 11ea sativa nowadays while on the Oeta ]11glam regia whereas several other arboreal types, including Pis­ is found. AesC11!t1s hippocasta1m1111 is found on both tacia, Fraxi1111s or1111s, U/11111s, Tilia, and Fraxi1111s ex­ mountains. On the Ossa i\ifountains Casta11ea sativa ce/sior guite suddenly decrease or disappear. Erica­ is met with even in pure forests from 300-1 200 m. ceae and Ce11ta11rea so/stitialis-type now start con­ On the northeast exposures of the Kato Olympus tinuous curves. The cause of this is as yet unknown. AesC11!11s hippocasf[JJ!e!flll occurs along small streams j\fanmay have destroyed the trees but the Ericaceae (Voliotis, l 976a, l 976b ). In prehistoric and more for instance may be the sign of human as well as recent times J11glam, Plata1111s and Castama would climatical influence, such as an increase in precipi­ have been limited to the rnountainsarounc l the plain. tation. Amund Lake Viviis sucb species were not to be found. S11hzo11e Z; is characterized by relatively high Fag/IS Aesm/11s is thought to be endemic in the Balkans. values. As is often fo und in diagrams from northern The tree is not abundant at the present time. As tbe Greece, Ericaceae increase at the same time as Fag11s horse chestnut is insect-pollinatecl it is considerably or somewhat earlier. This increase of beech is dated under-represented in tbe pollen min. Even if this at about 4000 B.P. (Turner and Greig, 1975 ; Bot­ tree had been more common at one time it would tema, 1974). At the same time as the Ericaceae show be impossible to demonstrate this by means ofpol len up, Ce11f(ll/J'ea solstitialis-type, abundant during the analysis. In Xinias this tree is representeclonly once, 28 Glacial period and up to subzone Z3, starts to appear in spectrum 87 (table l ) . Pollen anab1tical in1Jestigatiom in Thessab1

For studying the modem pollen rain in Thessaly, lustrative, indicating that these trees, grmving on the �esults of only one surface sample are available. the mountains, are distributed rather well. The faet l\fr. and l\frs. Reinders-de Roever from Lelystad that Castanea and j11gla11s are found for the first time (Necherlands) were so kind as to take a sample during in spectra 90 and 87 of the Xinias diagram shows thei..- archaeological investigations in r 978 near that they did not play a role previously. Volos. The sample comes from a Ph1J'gt1na (a degrad­ ed sc:adium of 1l1aq11is or Pse11do111aq11is) near the coast Co111parison oj the present vegetation JJJiththe 11pp er sa111ples bet\"'1eenHalos and Volos (fig. 2) . AP values in the oj the Xinias diagra111 surface sample spectrum are low, as could be ex­ pect<:d (table 3). Especially the values of Castanea Can '.V e demonstrate a relation between the (partly (o. 5 °J o), )11gla11S (0.5°/o), and Plata1111s (r.7°/o) are il- reconstructed) natural vegetation (fig. 2) and the

TABLE 3

Surface sample (for location see fig. 2); percentages based upon

I: p olien sum including all types except for indeterminata olien sum including all the same types as I, except for Crassulaceae which are thought to be over-represented li: p olien sum as chosen in the diagram Xinias I and Lake Viviis III: p

li III li III

Quer cus coccifera-type 6.2 8.2 9.1 Gramineae 15.9 20.9 23.3 Phil/yrea 0.5 0.6 0.7 Ligulif/orae 1.0 1.3 1.4 Pistacia 0.5 0.6 0.7 Senecio-type 0.2 0.2 0.2 Olea 5.7 7.5 8.4 Matricaria-type 0.8 1.1 1.2 Oleaceae 0.7 0.9 1.0 Xanthium 0.2 0.2 0.2 Quercus cerris-type 2.0 2.6 2.9 other Tubulil/orae 1.3 1.7 1.9 Carp in us orientalis/Ostrya 0.7 0.9 1.0 Ranunculus acer-type 0.7 0.9 1..0 0.2 0.2 Co ry I LIS 0.2 Lotus-type 1.5 1.9 2.2 Cupressaceae 3.6 4.7 5.3 Leguminosae 0.3 0.3 0.4 Ab i es 0.8 1.1 1.2 Campanula 0.2 0.2 0.2 Fagus 0.2 0.2 0.2 Umbelliferae 1.1 1.5 1.7 Pinus 5.6 7.3 8.2 Brassica-type 1.8 2.4 2.6 Alnus 0.5 0.6 0.7 Papaver 0.2 0.2 0.2 cf Salix 0.3 0.4 0.5 cf Anagallis 0.2 0.2 0.2 cl Citrus 0.3 0.4 0.5 Echium-type 1.5 1.9 2.2 Vitis 0.2 0.2 0.2 cl Labiatae 0.2 0.2 0.2 Rharnnaceae 0.2 0.2 0.2 Crassulaceae 23.3 30.5 34.1 Juglans 0.3 0.4 0.5 Lythraceae 0.3 0.4 0.5 Castanea 0.3 0.4 0.5 Galium-type 0.3 0.4 0.5 Plata nus 1.1 1.5 1.7 Rumex acetosa-type 0.8 1.1 1.2 Ericaceae 0.3 0.4 0.5 Urtica 2.8 3.7 4.1 Polygonum aviculare-type 0.8 1.1 1.2 AP 30.2 39.6 44.2 Cyperaceae 2.5 3.2 3.6 Arternisia 0.2 0.2 0.2 Sparganium-type 2.5 3.2 3.6 Chen opodiaceae 4.8 6.2 7.0 lndeterminata 6.2 8.2 9.1 Plantago lanceolata-type 1.8 2.4 2.6 Plantago spec. 1.8 2.4 2.6 Cerealia-type 0.7 0.9 1.0 POLLEN SUM 608 466 417 29 S. BOTTEMt\

upper spectra of the Xinias diagram (or Viviis dia­ 400-437 cm idem light grey gram)? Most of the area around Xinias belongs to 437-447 cm grey clay with many rust-colouted the Ostryo-Carpinion but north of the Othrys ever­ patches and smal! stones. green elements appear in this zone (Coccifero-Car­ pinetum). Thus next to the very common Qmrms The Lake Viviis sediment was difficult to core as p11besce11s from the Ostryo-Carpinion, pollen ofQ11er­ the clay was very hard. The many patches that were C11s coccifera can be expected. Nearer the CQast Qmr­ yellow to rust-coloured may be a sign of oxidation ms coccifera-type pollen is produced by the Quercion and in some samples a very low pollen concentration ilicis. was met with. The faet that no pollen counts could A Quercion confertae inhabits a large part of the be made in several places should be kept in mind Pindus Mountains producing deciduous oak pollen when judging the diagram. \X/henspectrum distances by Qmrmsfraimtto ( c01iferla) ,Q. cerris, andQ. p11bes­ are over 30 cm the pollen content in between was cet1s. The mountain area covered by Fagion and too low to make satisfactory counts. The quality of Abietion is much smaller. this information is deficient compared to the Xinias Although no special zone is ascribed to Pi1111s 11igra diagram but hardly any information is available from one should keep in mind that this tree occurs in many the lowlands in Greece at all. For this reason it is places in wide y divergent altitudes. nevertheless of same use. I Quite understandably the bulk of the pollen is produced by the various deciduous oak species 7.2. Pollen sum whereas the e,tergreen oaks have a more modest share in the pollen precipitation. Cm pi1111s orie11talis/ In preparing the diagram (fig. 5) several pollen sums OstJJ'O pollen is represented fairly well in the zone were tried and provisional diagrams drawn. A dia­ of that name. The share of Fag11s and Abies is smal! gram based upon a pollen sum including only the but so is their share in the vegetation and they are arboreal types could hardly be compared with the growing further away. Xinias diagram. One sum was made including tree Pi1111s, a genus that is often over-represented in types and wind-poliinating herbs and another includ­ the pollen rain, demonstrates unexpectedly low ing tree types and all non-local herb pollen types values. It is quite possible that the upper spectra date (fig. 5). These two diagrams looked very much the back to a subrecent time when deciduous oaks were same (see also Xinias I). The basic difference was still plentiful and by far outnumbered the other trees that the AP/NAP ratio gave, understandably, lower including pine. AP values when more herbs were includecl. Never­ theless both AP curves were ninning parallel at some distance apart. For that reason the sum chosen to 7. LAKE VIVIIS calculate the percentages was that which incluclecl all trees and non-local herbs. Nevertheless some 7. 1. Coring and lithology herbs may have a local effect appearing in consider­ able numbers when the lake dried up, ei ther seasonal­ Coring: In one hole at the west side of the drained ly or for longer periods. According to information (dried up) Lake Viviis; c. 3 km SE of Kata Kata­ from the Geographical J-landbook ( 1 944) this hap­ makion (fig. 2). pened quite often in recent times.

Lithology: 6- 60 cm grey to dark-grey clay with dark 8. DISCUSSION OF THE VIVIIS DIAGRAi\1I patches 60-21 5 cm grey clay with dark and rust-colour- Pollen zones: ed patches

215-276 cm idem dark grey Zom 1 (spectra 1-3) 276-377 cm idem grey 377-400 cm idem dark grey The pollen percentages of deciduous trees like Car- Po/le11 a11a61tical i11vestigatio11s i11 Thessa61

pi1111s orie11/alis/ Ost1]'a, Co1 '! s and Carpi1111s be/11!11S are the pollen rain (Bottema, 974). The preservation ] 1t r relatively low compared with the next zone 2. Maxi­ of the material was not ideal and some grains remain­ mum values for Pi1111s are about 20°/0. The values ed unidentified. Both explanations are, however, not for Liguliflorae are very high as are those of the un­ enough to explain the absence of this group and it identified grains. These two facts point to corrosion is clear that in the lower plain ofThessaly the climax of the pollen content (Bottema, 197 5 ). vegetation was not replaced by maquis when man settled there.

Zom 2 (spectra 4-S) Casta11ea ( o. 7%), Plaf(JJlllS( 0-4%), and Olea ( r .9%) appear in the uppermost spectrum only (table 4).

At the end of zone r, Ost1] 'a-type had already in­ Compared with the diagrams of Pertouli and Lito­

creased, a process going on in zone 2. The increase of choro (Athanasiadis, r 97 5 ), part of the younger sedi­ Ct11pi1111s orie11talis/Ost1J1a is also visible in Xinias I ment must have disappeared or no deposit was form­ (fig. 4). This event is dated about 6500 B.P. (Bot­ ed (see also X i nias I). Even Cerealia-type shows up in tema, 1974). Ericaceae show a closed curve at the spectrum I 5 only (o.6%). In this respect the surface same time, ;;iz. at the beginning of zone 2. This is sample study (table 3) is illustrative. Even on the much earlier than in the Xinias I diagram. A possible coast pollen of]11 gla11s and Cas/a11ea is found. In the explanation could be the difference in elevation be­ ,uea around Lake Viviis they would not have been tween the two lak.es. The vegetation during zone 2 important befare the time of spectrum I 5 (fig. 5). would have been composed mainly of deciduous spe­ The dominance of the non-arboreal pollen types

cies. 0 r scerris orQ. i11jecloria together with OslJJ'tl is striking although this may have been partly caused 11e C11 and/or Cm pi1111s orie11/alis, Co1]'!1ts, and Ct11pi1111s het11- by the drying up of thelake. Still the Viviis area would /11s grew in the plain and the surrounding foothills. have showed open fo rest or treeless vegetations es­ Pine must have retreated to the higher mountains, pecially compared to Xinias, situated at an altitude as indicated by rather low values (10-20%). about 500 m higher. Up to now very few lake sedi­ ments from about sea level in Greece have been Zo11e j (speclm 9-10) studied palynologically. The Drama section (\Xlijm­

stra, r 969) often reflects marshy conditions instead Zone 3 is characterized by a brief increase of Pi1111s of open water so local pollen production is impor­ resulting in the highest AP value of the diagram in tant. Unpublished information from Lake Tri khonis spectrum 9. This increase is quite probabl y the result in Akarnania, at sea level, shows that the lower vege­ of pine spreading over the Volos plateau and the tation zones are represented better than the moun­

Mavrovuni Mountains. A comparable increase is not tain belts (as is fairly predictableI ) . AP values are to be seen in the Xinias I diagram. higher than tbeir counterparts in diagrams originat­ ing from levels of 500 mand higher. Ofcour se human

Zo11e 4 (speclra 11-1 J) influence was also greater upon the lowland vege­ tation (forest) than on the vegetation at higher alti­ Zone 4 shows a constant decline of the AP values. tudes, at least in the early stadia.

Pi1111s and Omrcl/S cerris-type decrease considerably Although the plain of Thessaly was inhabited by and Os!J]'a-type, Ul11111s, Tilia, Co1]'!11S, and C(//pi1111s farming cultures starting in the 7th millennium B.C. hetn/11s disappear. (Halstead, r977) it is difficult to ascertain the in­ This decrease must have been caused by man as fluence of man upon the vegetation from this dia­ J1111iper11S occurs from spectrum r 3 on. Ahies and gram as well as from the Xinias I diagram. Cerealia­ Fagm are still present in low numbers and this is type pollen is onl y represented in spectrum r 5 with an indication that the fo rest destruction took place o.6% (table 2). Pla11tago species, especially P. la11ceo­ at lower and medium altitudes fi rst. lata-type and Ce11ta11rea solstitialis-type as possible in­ It is quite conspicuous that the (sub)Mediterra­ dicators offarrning are present throughout the whole nean types are found either not at all or only in in­ diagram. This i5 not unexpected as the assumed time significant numbers. There are several explanations covered by the diagram falls within the limits of the for this absence. This group is under-represented in Neolithic and later fa rming periods. The beginning S. BOTTHLJ\

of zone 3 shows traces ofa regeneration of pine forest. values. In the case of a fa lling water-table first the Interpolating from the assumed date of 6500 B.P. plants of deeper water would appear, followed by for the increase of Carpi/llts orie11ta/is/ Ost1)'a at spec­ plants of shallow water and finally by marsh plants. trum 3, this possible regeneration dates from about At a depth of 0.5-3 min eutrophic water a Nym­ 3 500-3000 B.P. However, the age of the top sample phaeion (Westhoff and Den Held, 1969) may have

(spectrum I 5) is very difficult to trace, but may be occurred. In moderately eutrophic water to eutro­ a few thousand years, because of the absence of) 11g­ phic water deeper than I m, something like a Pota­ la11s and the very late appearance of Plata/llts and meto-Nupharetum could be found, including i\111-

Casta11ea (in spectrum I 5 ). \'\fith this assumption the phar /11te11111 , ll1j'riopl!J'l/11111 11erticillat11111 or111. sp ica/11111, interpolated date would beabout 3900-4000 B.P. The and PolaJJJoge/011 species. briefincrease ofPi1111s pollen could indicate a decrease In marshy areas, flooded all through the year and in population pressure in the plain and adjoining up to 2-3 m deep, typical representatives of a Phrag­ mountains. mitetalia would have been found like T) pha latifolia, After that period the degradation and destruction a11g11stifolia (included in the Sparga11i11111 pollen­ T. of forestst arted again with renewed fo rce. Accord­ type), Iris psmdacoms, and Cyperaceae, for instance ing to the low AP values forest occurred only on the Sci1p11s laC11stris in deeper water and other represen­ Volos plateau and other mountains, leaving the plain tants of this family in shallower water or in marsh. almost devoid of trees (see also Turrill, I 929). In the vegetation along the edge of the lake Cladi11111 J\n increase in Pleridi11111 is visible after the 1\P JJJarisc11s was present. In the vegetation just above maximum in spectrum 9. Bracken would have water level D1]1op!eris species occurred. The vege­ spread in places where forest was cleared on the tation units or plant species ascribed are represent­ mountains. ed in the pollen diagrams with higher or lower per­ centages. As these plants do not require the same habitat 9. LOCAL VEGETATION but vary in their demands with regard to water depth, water mavement, light, and trophic state, the succes­

9· I. rvfarsh and water plants of Xinias sion should be obvious from the pollen record. How­ ever, such an ideal situation is rare. \'\/ hen the water Half the Xinias I diagram consists of curves of plants depth decreases, ei ther by sedimentation or loweri ng which are considered to be local. Apart from pollen, of the water-table, plants of deeper water will be curves of spores and Pediastn1111 species are present­ followed by others and finall y even marsh plant­ ed. pollen will be met with. Of course the reverse of It is irnmediately seen that this part ofthe diagram such a succession can also be encountered. A marsh not only shows high to very high values but also vegetation will drown when the water-ta ble rises and that most curves demonstrate pronounced fluctu­ be succeeded by for instance a PhragJJJiles vegetation. ations. The pollen and spore dispersal hinders the exact From spectrum 1 -24 pollen curves of Cyperaceae, tracing of vegetation succession as plants gro wing Sp argc111i11111-type, lli)'riopl!J'l/;1111 verticillat11111-type and at same distance may even confuse the production the curve of Di]opteris run very smoothly. Peaks can on the spot. Thus in spectra 24-2 7 Sparga11i11111-type, be seen in the spectra 24-27, p-39, 43-48, and 5 6-64. T)pha latifolia, ll1j1riopl.!)1//;1111 verticil/a/11111-type (also including sp ica/11/JJ) appear together with D!J'OP­ It is always difficult to explain the meaning of such M. peaks. One is inclined to implicate the water level teris, the spore of which represents a large group of of the lake to explain such phenomena. A plant spe­ species. In general Cyperaceae either precede fem cies or a group of species would demonstrate a high or follow an increase in typical water plant-pollen. pollen percentage when the water depth was optimal 111.J'riopl!J'l/11111 11erticillat11111-type generally follows the for their growth at the point where a core was taken same course as Sparga11i11JJJ-type, apart from spectra later. 5 2 and 54 where it is only accompanied by PotaJJJo­ \Xfith a change in water-table one may expect a geto11. Such events are not restricted to the Xinias I change in water-plants and thus in pollen types and diagram but can also be seen in the Ioannina I dia- Pollen a11ab1tica/i11vesti gatio11s in Thessab1

gram on the other side of the Pindus Mountains demonstrates high values for the lower and the (Bottema, 1 974). upper part of the Glacial period presented in the dia­ During Late-glacial times a marked decrease in gram. It is found hardly or not at all during the Post­ pollen types of aguatic origin takes place. Cypera­ glacial. Pediastm111 kawraisk)'i occurs in Greece dur­ ceae are important towards the end of this zone. ing Postglacial times only in the interior lakes of Furthermore Po6igo1111111 aviC11!are, Urtica and D1J1op­ Kastoria and Khimaditis at alti tudes of 6 5 o and

leris indicate marshy conditions instead of open 5 60 m respectively. water. Lake Kastoria lies in between the 2 5.5 °C and 26°C July isotherm (Philippson, I948). Correcting for The Postglacial part of the diagram shows low values altitude the July temperature would be about 23°C. for local pollen. The increased pollen production of The 7°C J anuary isotherm is found close to the lake, the upland vegetation most certainJy infl uences the after correction the January temperature would percentages. amount to 4.9°C. The average J uly temperature for T) 'pha latifolia still has the same values as in the the nearest meteorological station of Ohrid is about previous period. l"V_)'/11phaea is relatively important 2 1 °C, for J anuary c. 3 °C. The corresponding temper­ whereas B11to11111s, A1e101a11thes, and Le11111a, met with atures for Thessaly, as concluded from the meteoro­ before in low numbers, are no longer fo und. Spores logical stations of Volos and Lamia, are 8°C and of !sodes are now regularly encountered. As there 25-27°C for January and July respectively. are many lsoeles species demanding different habitats, In the fo liowing table the average temperatures nothing can be concluded from the presence of these of some Greek sites will be given, together with the spores. presence or absence of Pediastm111 ka1Praisl�)'i. Pteridi11111, a fe m from the upland vegetation and not of local origin, fo rmed part of the undergrowth Site Average Average P. kawraislyi of the deciduous fo rest and appeared early in the January July present or Postglacial. temp. temp. not

9.2. Pediastrum Lake Xinias c. 5 0C c. 250C Lake Viviis c. 8°C c. 270C The various Pediaslm111 species or varieties are also Ioannina 5 .10C 24.0°C counted and calculated as percentages of the pollen Edessa 3 .40C 24.6°C sum. As some types sometimes have very high values Giannitsa .6oC 26.7°C 5 ± they are presented in the diagram on a different Khimaditis r.8°C 230C + scale. Kastoria 3 -4.9°C 2 1-23°C + At fi rst glance the irregular character of the curves is striking. As stated previously (Bottema, 1974), It seems that Pediastm111 ka1/!raisk:)'i is favoured by hardly any information is available on the ecology of low temperatures, possibly an average of about 3 °C Pediastm111. Pediastm111 bo1)'a1111111 is ubiguitous in for January and/oran average ofunder 2 3 °Cfor July. Lake Xinias and this is true for many other (Greek) \Xfhichof these is the Jimiting factor is not known. diagrams. Such Chlorophyceae indicate the presence In addition to temperature other factors must be of water, although of course the presence of water is decisive for the presence or absence of this alga. For also concluded from the faet that sediment was de­ instance in Lake Vegoritis with deep and rather posited. \\!hatelse can be deduced from Pediastm111 ? sterile looking water and bare banks, no Pediastm111 Pedias/m111 bo1J1a1111111 is met with even during the kawraisk)'i has been present during the last few thou­ Late Glacial when all the other Pediastm111 species are sand years. In the nearby Lake Khimaditis, however, absent. Pedias/m111 d11plex var. clathra/11111 as usual this species is very common. shows the same tendency as P. bo1J1a1111111 but with Pediastm111 si111plex is found in large numbers in lower numbers. Pediastm111 d11plex var. arach11oidea some spectra and is completely absent from others. is rarer than the first variety. Apart from spectra 42- No explanation for this behaviour can be given. 4 7 it behaves in the same wa y. Pediastm111 kawraislyi Pediastm111 clathra/11111 is found almost exclusively in 33 S. BOTTE�IA

9. 3. TVIiscellaneous

The curve of the indeterminata includes mostly cor­ rodecl grains which could not be identified. Corro­ sion can be an indication of seasonal drought causing changes in water levels. This is especially the case in spectra 60-76, a period represented by pollen as­ semblages also indicating dry conditions. Only quantitative information is given regarding the presence of three Chlorophyceae, viz. Coe/astm111 retim/a/11111, Sce11edes11111s, and Tetraerlro11. During zone Y they were not fo und indicating different condi­ tions at that time. No fu rther conclusions can be drawn from the presence or absence of these Chloro­ phyceae.

9-4· Local vegetation of Viviis Fig. 6. Pediastrum "kawraiskyi-boryanum". The local marsh flora and lower plants ( Pediastm111) growing in and along the lake show irregular pollen the lower two zones. There it attains very high val­ and spore patterns. This must be due to the changing ues. The species disappears at the beginning of zone water-table of the lake. Cyperaceae, Sp arga11i11111-type, X and appears in low numbers in the first half of !sodes, Pediastm111 bo1J1a1111111 , P. si111plex, and P. cla­ the Postglacial. //;m/11111 are clearly correlated. Their peaks suggest A Pediastm111 species named "kmJ1mis/9i/hotJ'a- higher water levels, whereas low values suggest dry 1111111 " occurred in large numbers in part of the spec­ phases. This is al so concluded from the curve of un­ tra (fig. 6). No matching descriptions could be found identified grains (indeterminata). This group covers in the literature. This type has outer cells which re­ those pollen grains which were tao corroded to per­ semble Perliastm111 katJJmisk)'i. The lobes of the peri­ mit identification. In general low values for indeter­ pheral cells are triangular and not elongated. They rninata (corroded grains) correlate with an abun­

are placed one above the other or they are in one dance of Pediastm111 and pollen of marsh and water plane in which case they slightly overlap. In between plants. The correlation of corroded pollen in relation the cells that form the body, openings are found like to Liguliflora� has been discussed previously (Bot­ in Pedias!m111 clat/Jm/11111 . The cells are shaped more tema, 197 5 ). or Jess as in P. bo1]'a1111111 and are pitted in the same way. For this reason a provisional name of a combi­ nation of P. ka1J1misk:J'i and P. VOt]•a1111111 is given. P. IO. CORREL!\.TIONS A D COl\iIPARISONS kmJJmisk:)'i/ bo1J•a1111111 was found in another Greek dia­ gram before but the specimen was considered to be 1 0.i. Correlation of Greek diagrams an artefact (Bottema, 1 974). Later, however, some specimens were also encountered in samples from Two Greek diagrams demonstrate a period that cor­ Lake Zeribar, Iran (van Zeist and Bottema, 1977). responds with that of Lake Xinias, viz. that ofloan­

It was only when this type was found in such large nina (Bottema, l 974) and that of Tenagi Philippon

numbers as here in samples from Lake Xinias that (\Xlijmstra, l 969). (For a comparison of Xinias, more attention was paid and the type was identifiecl Tenagi and Near Eastern diagrams the reader is as a separate one. The new type correlates to some referred to van Zeist and Bottema, 1977). extent with Pedias!m111 si111plex and not with P. The Ionannina core was taken on the west side of

kawraisk)'i. the Pindus i\fountains (fig. l ) at about the same alti­ 34 tude as that from Xinias. The very deep section of Pollen a11a(J1tica/ i11vestigatio11s in Thessa(J1

IOANNINA SOOm XINIAS SOOm TENAGH I PHILIPPON 40m

Z5

ZI.

Z3

'"" ' :.-:('.7.\'b Z2 ' : ' • ' ' Zl ' ' ' "" ' ' ·/,r�, 5 "tJ ', '

]\!:�Q� - c� )-- 10 c3 . o.ooo -

11

12

10 13

11 1" '' ' ' ' ' ' 12 15

16 13

Fig. 7. Schematical diagrams of Greek sires, showing curves pollen, Arte111isia, Chenopodiaceae and the sum of for Q11erms, Pi1111s, AP/NAP ratio, Arte111isia, Chenopodiaceae the other non-local herbs are schematically drawn. and remaining herbs; for location of the sires see fig. 1. The vegetation at the tbree locations would have differed according to altitude, exposure, distance from the sea, etc. Such diffe rences must be kept in Tenagi Philippon was taken in the Drama Plain at mind when comparing these three diagrams. An at­ about 40 rn above sea level (fig. 1 ). The parts tbat tempt has been made to connect the periods that correspond with the Xinias diagram are shown in could be recognized in the three diagrams on the fig. 7. Tbe curves of 011erms, Pi11m, total arboreal basis of tbe sbape of tbe curves and the radiocarbon 3 5 S. BOTTElvlA dates. The information used is partly derived from below. Such a scheme is more meaningful when the van Zeist and Bottema (r977). palynological information is obtained under compa­ For the Glacial as well as for the Postglacial part rable conditions. clear similarities can be seen. The lower part, the zones Ioannina S4 and T 4, Xinias V r -V 3, and Tenagi Table 4. Tentative correlation of the pollen zones Pr are thought to represent a similar period of time. in diagrams from Ioannina, Xinias and Tenagi. The During this period forest covered that side of the codes used for the three diagrams refer to each dia­ Pindus chain exposed to weather. On the sheltered gram only and cannot be used fo reither ofthe others. side of these mountains fo rest was restricted to suit­ able habitats at a certain altitude. Besides this, steppe Ioannina Xinias Tenagi vegetations dominated. On the southern edge of eastern Macedonia, in the low-lying Drama region, S4 V2 Pr there was only steppe. Any forest remnants must Tr V3 Pz. have occurred further inland at higher altitudes. T3 \':( / r P3 These fo rests or tree stands decrease or even dis­ U2 P5 \XI2. appear with the passage of time, sometimes recover­ U4 3 P7 \XI ing during short fluctuations but on the whole con­ V2 X/\Xl3 Xz. stantly losing terrain. It is evident from the pollen V4 (5 .oo m) X4 X curves that trees hold their own better on the exposed side of the Pindus than on the sheltered side whereas especially oak finally disappears from the Drama 10.2. The \Xlurm Glacial steppe vegetation of region. Greece Up to about r 5 ,ooo B.P. steppe conditions prevail and tree refuges were to be found especially on that Apart from the upper parts the pollen diagrams of side of the Pindus exposed to rainbringing winds. Xinias, Ioannina and Tenagi Philippon demonstrate It is also there that certain minor ameliorations of steppe conditions. The open vegetations provided theclimate favour treegrowth. This isless pronounc­ habita.ts fo r many herbs resulting in a wide range of ecl in Xinias, and in Drama the postulated rise in tem­ herb pollen-types. Only around 40,000 B.P. is the perat.ure would prevent any form of tree growth. Ioanninaarea covered with dense forest thatprevents The level of r 5 ,ooo B.P. is clrawn somewhat lower herb growth. in fi g. 7 than in van Zeist and Bottema ( r 977) . The amount of herb pollen found during the last A little after r 1,000 B.P. trees, especially oak, re­ glacial period varies from site to site. In Ioannina place the steppe vegetation. The sharp decrease of the values fluctuate between 40 and 8oo/0 (apart from AP values in Tenagi-zone Y 3, must be due to local the forest phase). In Xinias NAP percentages are herbs. 80-90%, whereas in Tenagi values exceed 95 %· It The general synchronization ofthe three diagrams will be clear that apart from these values, vegetations is rather obvious. It will be more difficult to corre­ may have differed in composition. late tbe fluctuations in detail. For such a correlation A translation of such NAP percentages in terms of one has to assume that every cbange in climate is vegetation remains difficult. The modem pollen ubiquitous and that such a change affects the vege­ precipitation of much overgrazed and/or destroyed tation and thus the pollen spectra derived from such steppe, fo rest steppe or steppe forestin Syria is quite a vegetation, in a comparable if not exactly the same instructive in this respect (Bottema and Barkoudah, way. It is obvious that vegetations at different loca­ in the press). In such vegetations NAP values amount tions are not the same, the differences being greater to 80-90% of which Artelllisia constitutes z.0-50%. as the abiotic factors diverge. This is clearly evident Vegetation cover averages 5-10% and up to about for locations further apart (van Zeist and Bottema, 30% in patches of vegetation. r977). In the first place there are quanti tative differences For those who like to correlate fluctuations in in the various areas, secondly qualitative differences detail and to give them names, a scheme is given are evident. The quantitative difference applies to Pol/e11 a11ab1tical i11vestigatio11s i11 Thessab'

Arte111isia being better represented at lower al ti tudes in van Zeist and Bottema (1977). In the two remain­ and in a more continental situation. In Tenagi Arte- ing cores they are however grouped together. 111isia measures 30-50%, in Xinias 20-27% and Ioan- A series of types are present on both sides of the rnna. r 0/ Pindus lvfountains·which are not mentioned in the 5-20 10 . Chenopodiaceae are important during Late Gla­ Tenagi core. They include : Cruciferae, Ce11ta11reasol­ cial times in Xinias and Tenagi. Gramineae tend to stilialis-type, Cmta11rea scabiosa-type, K11a11tia, E11- be higher during the Glacial than during the Post­ phorbia, Te11cri11111, Ca111pt11111/a, Verbasc11111-type, Rhi-

glacial in Tenagi and Ioannina. In Xinias grasses are 11a11t!J11s-type, Hjperic11111 pe1forat11111-type, Gerania­ relatively unimportant especially during the Late ceae, Liliaceae. Glacial. Cerealia-type pollen is found during the The sediments of Xinias are very rich in types, whole period covered by the pollen diagrams in same of which are not reported for the other two Ioannina and Xinias. They would have been part sites. These include Scabiosa, Dipsacm, 111alva, and of the steppe vegetation. Co11volvu/11s. For the Postglacial Cj111ocm111be can be The two groups of Ep hedra identified viz. Ep hedm added to this list. In contrast to Ioannina however distt1cl!)'t1-type and Ep hedra jragilis-type behave dif­ Plumbaginaceae were not met with. fe rently. Both are present in the three cores for most Of the local water-plants and marsh flora B11to11111s of the Glacial period but Ep hedra distt1cl!)1t1-type be­ and JV11phar were not found in Tenagi but here Pep/is came more widespread during the Late Glacial. was very common whereas in Xinias only one doubt­ Thalictm111 is found on all locations but in Tenagi ful grain of Pep/is was met with. it is also found at the beginning of the Postglacial \X/hen discussing the similarities and the differ­ indicating relatively dry conditions. Pla11tago behave ences of the herb pollen and especially the steppe more or Jess identicall y at all three loca tions, al though vegetation in the three localities, the origin of the the composition of the various types comprising the cores has to be reckoned with. Tenagi is a large total of Pla11tago may be a little different. In Xinias marshy area with a lot of local pollen, possibly ob­ and Ioannina an increase in Pla11tago is visible towards scuring regional NAP of surrounding steppe vege­ the end of the Glacial period but during the Late tations. Besides the amount of pollen counted is not Glacial in Xinias, Pla11tago did not play any role. known. \'{/henlarge pollen counts are made the rare In Xinias and to a !esser extent in Ioannina a series types tend to show continuous curves. J\.s many herb­ of different types belonging to the Tubuliflorae were pollen types are insect-pollinated one has to consider identified. Comparison is not easy as in Tenagi most the probability of under-representation. of these types are included in the group of Tubuli­ Steppic influences in Ioannina are Jess than in florae, which are not identified to genus level. A Xinias, as can be declucecl from the forest phase striking feature is the presence of Xa11thi11111 in the prevailing there at about 40,000 B.P. Tenagi area in the Glacial period and especially The Xinias area shows more variation in types during the Late Glacial. Not a single grain of Xa11- than the exposed side of the Pindus. This ma y be an thi11111 was found in the other cores. In young Holo­ indication for diversity in biotopes in Thessaly. The cene deposits "Ya11thi11111 occurs in sediments from high Arte111isia values in Tenagi show that at low Lake V olvi in Eas tern i'viacedonia. In Lake Vegoritis altitudes lack of moisture was a Jimiting factor fo r it is much rarer whereas in Lake Trikhonis in South­ other vegetation, resulting in a dominating Arle111i­ western Greece it is altogether lacking. sia-steppe. Cistaceae, especially Helia11the11111111, are common in Xinias and fairly common in Tenagi while in Ioan­ ro. 3 Comparison with Anatolian and Iranian gla­ nina they tend to be more rare. An increase towards cial steppe vegetations the end of the Glacial is visible. Caryophyllaceae are represented fairly well in Ul'esl A11atolia Ioannina and Xinias, but are not represented at all in Tena gi. Umbelliferae are commoninall three cores NAP values from West Anatolian pollen diagrams but in Xinias they attain highest values. There seven (van Zeist et al" r 97 5) amount to 40-75 %, Arte111isia types are distinguished according to the descriptions 20-5 0%, while Chenopodiaceae average about 20%. 37 S. BOTTEMA

A difference with regard to the Greek cores under Table 5. discussion is the presence of some Calligo1111/ll/Ptero­

PJ 'l"/1111 pollen. Despite a difference in altitude of 500- \Xfarm \V'arm to Subarctic Cold steppe 1000 m there is a general resemblance of the Glacial steppe moderate "Wald- climate spectra of Greece and \Xfest Anatolia. climate "Wald- klima" klima" East A11atolia a11d IJl'estem Ira11 Dicerorhim1s Elephas anti- The results of palynological investigations in East heil/i to ech11s q1111s Anatolia (Lake Van) and \,'\festern Iran (Lake Zeri­ Hippopota-

bar) are drawn from van Zeist and Woldring (1978) 111 1/S cf. at/- and van Zeist and Bottema (1977). The authors em­ tiq1111s phasize the different development of the vegetation Eqm1s l! J'- Eq1111s l!J '- there compared with adjacentareas. Thus differences du111ti1111s dm11ti1111s to the Greek/Western Anatolian pollen assemblage Eq1111s cabal- 1l1egaceros ( ildegaceros) iVIeg aceros can be expected. /m cf. ger- In this respect the uali ta tive differences are worth q lllatltCl!S mentioning. The Lake Van and Lake Zeribar dia­ 111 egaceros grams reveal the presence of pollen types not met Cenms ela- Cerv11s ela- ( Cerv11s ela- with in the Greek/Western Anatolia cores. Those phm ph11s phm) types include A11isosciadi11111-type, Atraphaxis, Rhe- (Da/Ila sp.) Da111a sp. 11111, R11111expal ie11tia- type, tatice sp icata/ PS)'ll iostacl!)'S, Capreo/11s Capreo/11s Capreo/m 5 Leo11tice, Bo11gardia, Prosopis, J-laplop!JJ1!l11lll, and S] si­ capreo/11s capreo/11s capreo/11s ri11chi11111-type. Saiga tata- Saiga fctfa- _,,,Ya11!hi11111 was not encountered in these diagrams rtca nca and its presence in Tenagi may be linked with the Bos prillli- Bos prillli- Bos prillli- low altitude. The differences mentioned above sug­ ge1111s ge1111s ge1111s gest a vegetation boundary between \'\festern and B11ha/11s B11ba/11s Eastern Anatolia caused by certain conditions of wbich precipitation was a decisive factor. The foliowing notes are made in this table. Hipp o­ pota11111s indicates winters without freezing. For Ca­ r r. ARCHEOLOGICAL HvIPLICA TIONS OF preo/11s during a warm steppe climate, steppe forest THE PAL YNOLOGICAL INFORTvIATION must be available and for B11ba/m marshes or water must be present at the same. situation.

I I . I . Palaeoli thi c The composition of the species is remarkable and interesting, apart from the dating problems with .Milojcic ( r 9 5 8) considers palaeolithic artefacts such a faunal assemblage. Some ofthese species share found together with fauna! remains fo und at some the same habitat, but others demand a clearly differ­ location along the Pinios River "ais untypisch im ent one. Such different habitats probably would not Sinne der westeuropaischen Klassen Terminolo­ have been available under the same climatical con­ gie". Schneider ( l 968) concludes that these arte­ ditions. Seeing that these species differ so much eco­ facts, compared with other material, should be logically, it is very difficult to put thern together in placed in the Riss/\Xflirm interglacial or Early the same period. \X flirm. Schneider's statement that the bone-bearing layer Which period is represented by the fauna found was fo rmed in a relatively short time is very impor­ together with these artefacts? Schneider gives a tant. Accumulation over a long period would in faet table (table 5) grouping the various species ac­ ma ke it easier to provide an explana tion for the diver­ cording to their climatical amplitude. gent biotopes. What explanation can there be for Polle11 a11a6•ticalinvesti gatio11s i11 Thessa61

this fauna] assemblage brought together within a palynologically by the present author gave promis­ short period? ing resul ts as also the more recent pollen rain is in­ According to the available palaeobotanical infor­ cluded, judging from the presence of Zea 111ais pollen. mation (Wijmstra, 1969), during the Riss/\X!Lirm, However, the beginning of farmingin Greece is not Thessaly should compare with the Postglacial viz. a represented in such cores. closed fo rest must have covered the area. I tis difficult Archaeological evidence from prehistoric Thes­ to place steppe animals like Dicerorhi1111s he111itoerh11s, sal y connecteclwith the vegetation history is scanty.

Eq1111s l!) 1dm11ti1111s or Saiga lalarica in dense forest. Halsteacl( I 976) gives informationon the settlement The period during which Schneider estimates this pattern in Eastern Thessaly, viz. the plain in which

fauna to have existed is not represented in the Xinias Lake Viviis ( = L. Karla) is fo und. In the Early and I diagram. Nevertheless the apparently incongruous j\1[iddle Neolithic many settlements were found, all list of fa una can be connected with the kind of short of them situated in the northwestern part. During fluctuations as found in zone V of the pollen diagram. the Late and Final Neolithic they covereclthe whole Even during the maximum extension of the forest in plain. During the Early and Mi delle Bronze Age the such a pollen assemblage there is still steppe present number of settlements sharply decreased. Halstead in the area. The difference in minimum temperature thinks tbat the population increased again during along the coast and in the plain is relativ el y great and the Late Bronze Age. In his maps no settlements are resulted in a diversity of habitats at that time too. fo und in thevicinity ofLake Vivi is apart from a small Moreover the climatical fluctuation is so rapid that outcrop in the southwest. The reason for this must several fa unas may have followed each other closely. be the floocling that happened from time to time. Comparable periods as demonstrated in zone V Thessaly was a well-known granary in Classical would have occurred also during the early \X liirm, times. Halstead explains the delayed clevelopment in an idea supported by the sequence in the Tenagi dia­ early times because of the isolation incl uding the bar­ gram (Wijrnstra, 1969). Thus the origin of the fauna rier with the coast. Athanasiadis ( 1975) clescribes the as shown in table 3 is more Jikely to have occurred periocl after the Late Bronze Age by means of a dia­ in the early \Xli.irmthan during the Riss/Wurm inter­ gram from Litochoro. The "Micldle Geometric" and glacial. Helenistic cultures (900-200 B.C.) strongly influenc­ eclthe vegetation through fa rming. During Roman r 1 .2. Neolithic and younger periocls and Byzantine times (200 B.C.-J\.D. 900) the extent of cultivated land cleclined, as concluded from evi­ It has been stated previously (Bottema, 1974, p. 165- dence inclucling palynological information. The 166) that the impact of man on the vegetation is re­ flourishing of Byzantium up to the invasion of the flected rather feebly in the Greek diagrams. Only Turks (A.D. 900- 1 400) shows a renewed attack on about 5000 years after the beginning of the Neo­ the natura! vegetation as concluded from the in­ lithic do clear traces appear in the pollen record. Ob­ crease of pollen due to agriculture. viously also indicators have to be lookecl for other The vegetation regenerated to some extent after than l:hosewel l-known from northwesternEur ope. the invasion of the Turks. Large estates (tchifliki) To study the influence of early fa rming upon the developed where Turkish landowners controlled vegetation, suitable sediments are needed. Such suit­ their serfs up to 1881. Part of the Greek population able sediments include those from rather cleep lakes flecl from the area because of uncertain conditions. w here local effects are reduced to a minimum. Co ring Graclually the people settled again and human influ­ in deeper water,-however, causes a lot of difficulties. ences became stronger and stronger. This process The .i'viackereth corer, although very practical for continued after the liberation of Thessaly. coring in deep water, has the disadvantage that its Until very recently farrning was primitive because maximum reach is six metres. In many Greek lakes of the general inability of the people to control the six metres covers only the last few thousand years. water and as a result of the aftereffects of the Turkish Cores from the Lakes of V olvi, Vegoritis and Thrik­ system of land management. honis collected by a team of the Department of Geo­ physics of the U niversity of Edinburgh and stuelied 39 S. BOTTEMA

13. REFERENCES

ATHANASIADIS, A" '975· Zur postglazialen Vegeta- PHIL!PPSON, A" l 948. Das Kli111a Grieche11/a11d.r. Dtimmler, tionsentwicklung von Litochoro Katerinis und Pertouli Bonn.

Trikalon (Griechenland). Florn 164, pp. 99-13z. SCHNEIDER, H. E" 1968. Z11r Q11artii1geologischeu E11t1JJickl!111gsge­

BOTTDIA, s" '974· La!e Q11alema1)' Vegetalio11 His!Ot)' of No rth­ schichte Thessalie11s (Grieche11/a11d) . Bonn. J1Jestem Greece. Thesis, Groningen. TURNER, J · & J· R. A. GREIG, 1975· Same Holocene pollen dia­ BOTTE�IA, s" 197 5. The interpretation of pollen spectra from grams from Greece. Revie111 of Palaeobotm!)' a11d Pa/_)'110/ogy prehistoric settlements (with special attention to Liguli­ 20, pp. 171-204. florae). Palaeohistoria 1 7, pp. 1 7-3 5. TURRILL, w. u" I 929. The pla11tlife of the Balka11 pe11i11s11/a. A BOTTE�IA, s" 1978. The Late Glacial in the eastern Mediterranean pl!J'logeographical st11dJ'. Oxford.

and the Near East. In: \XI.C. Brice (Ed.), The e11viro11111ental VOLIOTIS, D" 1973· Bezieh1111ge11 zJ1Ji.rche11 Kli111a, Bode11 1111d Vege­ histot]' of the J\Tear a11d 111idd/e Eas! si11ce the Last lce Age. tatio11 1111d Vegetatio11s-Zom11 i11 Griechenla11d Sci. Annals, (= London, pp. 1 5-28. Fac. Phys. & Mathem. Univ. Thessaloniki I 3). BOTTUIA, s. Y. BARKOUDAH, in press. Modem pollen pre­ & vouons, o" i976a. Die Geholzvegetation und die Vegeta­ cipitation in Syria and Lebanon. tionszonierung des Nordgriechischen Gebirgzuges Voras­ FREITAG, H" l 977. The Pleniglacial, Late-Glacial and early Post­ Vermion-Pieria-Olymp-Ossa. Bot.Jahrb. S) 'S/. 97, pp. l 20- glacial vegetations of Zeribar and their present-day coun­ 154·

terparts. A supplement to \XI. van Zeist and S. Bottema, vouons, D" 1 976b. Die Geholzvegetation und die Vegeta­ Palynological investigations in western Iran. Palaeohisto­ tionszonierung des Gebirgszuges Tymfristos-Oeta-Par­ ria 19, pp. 87-95. nassos (Griechenland). Ca11do/lea 3 l, pp. 3 7-5 2. GREECE. I. Physical geography, his tory, administration and peo­ WASYLI KOWA, K" l 967. Late Quaternary plant macrofossils from ples ( = Geographical Handbook Series 5 l 6). London, Lake Zeribar, western Iran. Review of Palaeobotai!J' (///dPa6•- 1944· 11ologJ' z, pp. 301-3 11.

GREECE. III. Regional geography ( = Geographical Handbook WESTHOFF, v. & A. J. DEN HELD, l 969. Pla11tmge111ee11schappe11 i11 Series p GB). London, l 94 5. . J\Teder!t111d. Zutphen. 3 24 p. GREIG, J· R. A. & J. TURNER, l 974· Same pollen diagrams from w1p1STRA, T. A" 1969. Palynology of the first 30 metres of a Greece and their archaeological significance. Jo11mal of l 20 metre deep section in northern Greece. Arla Bota11im Archaeological Science I, pp. I 77-194. J\Tærla11rlica 18, PP- 5 11-5 28.

HA LSTEAD, P" 1977· Prehistoric Thessaly: The Submergence WRfGHT_J R, 1-1.E" 1 97" Vegetation history. ln: \XI.A. J\kDonald of Civilisation. In: J. Bintliff (Ed.), il1j,ce11aea11 Geogrnpl!J" and G. Rapp (Eds.), The Jl1i1111esota Il11essc11ia E>.pedi­ Jr. Proceedings of the Cambridge Colloquium, September tio11 : Reco11stmcti11g a Bro11ze Age E1111iro11111e11t. University 1976. Cambridge, pp. 23-28. Minnesota Press, pp. l 88-199.

HAi\li\lEN, T. VAN DER, T. A. \V l i\lSTRA \V . 1-1. VAN DER i\IOLEN, j & ZEIST, \V. VAN, 1-1 . WOLDRING & D. STAPERT, '975· Late Quar­ 1965. Palynological study of a very thick peat section in ternary Vegetation and Climate of Southwestern Turkey. Greece and the \XliirmGlacial vegetation in the Mediter­ Palaeohistoria 17, pp. 53-143. ranean region. Geologie e11 il1ij11bo11111 44, pp. 37-39. ZEIST, w. VAN & s. BOTTDIA, '977· Palynological investiga­ 1-I ORVAT, 1" v. GLAVAC & 1-1 . ELLENBERG, 1974· Vegetatio11 Slid­ tions in western Iran. Palaeohistoria l 9, pp. 19-87. tll/'Opa.r ( Geobotanica selecta IV). Stuttgart. = ZEIST, \V. VAN & 1- 1. WOLDRlNG, 1978. A Postglacial pollen WLOJCrC, v" 195 8. Die neuen mitte!- und altpaliiolitischen diagram from Lake Van in East Anatolia. Re11ie1v Palaeo­ oj Funde von der Balkanhalbinsel. Ger111a11ia 36, pp. 3 l 9-324. hota1!J' a11d Pa6•11ologJ' 26, pp. 249-276.