PAL YNOLOGICAL INVESTIGATIONS IN GREECE WITH SPECIAL REFERENCE TO POLLEN AS AN INDICATOR OF HUMAN ACTIVITY

S. Bottema

CONTENTS

I. INTRODUCTION

2. PALYNOLOGICAL INVESTIGATIONS OF THE LAKES TRIKHONIS, VEGORITIS AND VOLVI 2.1. Introduction 2.2. Trikhonis 2.2. I. Geography, climate and vegetation 2.2.2. The diagram Trikhonis 5 2.2.3. Zonation 2.2.4. General vegetation reconstruction 2.3. Vegoritis 2.3. 1. Geography, climate and vegetation 2.3.2. The diagram Vegoritis 8 2.3.3. Zonation 2.3.4. General vegetation reconstruction 2.4. Lake Volvi 2.4. I. Geography, climate and vegetation 2.4.2. The diagram Volvi 6 2.4.3. Zonation 2.4.4. General vegetation reconstruction 2.5. The da ting of the cores of Trikhonis, Vegoritis and Volvi 2.5.1. General 2.5.2. Lake Volvi 2.5.3. Lake Trikhonis 2.5.4. Lake Vegoritis

3. THE REPRESENTATION OF POLLEN TYPES IN VARIOUS SOURCES 3. 1. Some rema rks on the practical limits of pollen analysis with emphasis on the Greek situation 3.2. Origin of pollen

4. DIVERSITY OF POLLEN ASSEMBLAGES COMPARED WITH DIVERSITY OF VEGETATION 4. 1. General 4.2. Diversity versus pollen sum and origin of sa mples 4.3. The diversity of pollen spectra compared to the diversity of the vegetation: An example 4.4. Diversity of subfossil assemblages 4.4. 1. Lake Vegoritis 4.4.2. Lake Trikhonis 4.4.3. Lake Volvi 4.4.4. Conclusion

5. DISCUSSION OF SOME SELECTED TYPES 5. 1. Platanus and Juglans 5.2. Vitis

6. POSSIBLE INDICATORS OF EARLY FA RMING IN GREECE

7. PALYNOLOGICAL EVIDENCE FOR DEMOGRAPHIC EVENTS IN (PRE)HISTORIC GREECE

8. ACKNOWLEDGEMENTS

9. SUM MARY

10. REFERENCES 257 258 S. BOTTEMA

I. INTRODUCTION 2. PALYNOLOGICAL INVESTIGATIONS OF THE LAKES OF TRIKHONIS, In this paper an attempt is made to analyse the VEGORITIS AND VOLVI possibilities of the use of pollen as indicators of human activity in Greek history. Several approaches will be made in this respect. New 2. 1. Introduction palynological information is brought forward The cores from these three lakes were taken by in the form of diagrams from three lake Professor K. M. Creer (Department of Geo­ sedi ments covering part of the Holocene. physics, University of Edinburgh) et al. for These sediments include the youngest periods the study of palaeomagnetism (Creer, 1981). that are often lacking in peat bogs or drained The cores were taken using a Mackereth marshes where the upper parts have been corer that is able to sample 6 m in one stretch ploughed or where organic remains have been in deep water. A disadvantage of the Mack­ oxidized. ereth corer is that it cannot sample more The use of pollen as an indicator of human than 6 m of sediment. The sediments were activity in Greece up to now has not been very studied palynologically to provide dates for successful in tracing the onset of fa rming. the geomagnetic variation curves. It was Compared with Iversen's (194 1) study on the expected furthermore, because of the origin of Landnam in Denmark such sharp indicators the material, that the samples would provide as Iversen was able to demonstrate do not good information on the Holocene vegetation seem to be present in early Neolithic Greece. development. Pollen of possible weed plants does occur in As all three cores contained clay, samples levels older than those ascribed to Neolithic were treated with a specific weight separation times. An attempt is made to evaluate the method (Bottema, 1974). Further processing statement 'weed' plant by comparing between was done accord ing to standard methods. In a plant-sociological study of Greek fo rests and chapter 2.5. the dating problems will be information on the vegetation history of discussed (Bottema in Creer et al., 1981). Greece. Only a short, general reconstruction of the In circles of nature conservancy and nature vegetation will be given in the discussion of management the idea is favoured that up to the diagrams. In chapters 5, 6 and 7 a the 20th century agriculture increased the discussion of the possible relations of diversity of the landscape and the diversity in vegetation history and (pre)history will be numbers of plant species in Western Europe. presented. In this study attention is paid to diversity of the vegetation as well as to diversity of pollen 2.2. Trikhonis assemblages and possible relations between the two. It is supposed that an increase in 2.2. 1. Geography, climate and vegetation pollen diversity on some occasions might ind icate human activity. Lake Trikhonis is situated in Southwestern The history of a few selected species and a Greece (38° 36'N, 21° 30'E) in the province of larger group of weeds is treated in detail. For Arkanania (fig. I) at an elevation of about 20 this purpose ten pollen diagrams from Greece m above sea-level. It forms part of a complex were selected. depression originating from late Tertiary and In younger periods, the influence of man Quaternary times when the whole of this area upon the vegetation in Greece is clearly was severely fractured. The main depression visible. Such an impact was not the result of a contains lakes Amvrakia and Ozeros; a constantly increasing pressure upon the en­ subsidiary trough lies to the north of the vironment but a phenomenon that could Arakinthos Mountains (up to 1000 m) and vary in time and in space. An attempt will be holds the lakes Angelokastron and Trikhonis. made to link the changes in vegetation derived Overflow of. these lakes runs to the Acheloos from the pollen evidence with (pre)historic that discharges into the Gulf of Patras. South events. of Lake Trikhonis, sandstones and flysch are found , in the north alluvial deposits occur (Gree ce, 1944). The climate of the area is Mediterranean. In Palynological investigations in Greece 259

, -- \( , and Juglans are not present in contrast to the \ I I \ , , I \ , upper spectra . Values of the deciduous oak ) I \ , , \ ' \ " type rise steeply after spectrum 14. Zone I can ,,,.r---,...;- --"" - - - ( ....1,, ,-...... - / / ,") , I ) 2 • ..... I be subdivided into a subzone la (spectra 1-4) "'/- _ ..... • ,./"'- ) \0 ..... I and a subzone Ib (spectra 5- 14) on account of , 6 5 � 7 •• � . L, the higher vaues in the latter of Artemisia, 0--",,// 8 • 4 ../ I- • 3 I Plantago lanceolala-type and Cenlaurea sols­ I . titialis-type. 9. \ ��\(""-..,� I J .,) "- • 10 •11 Zo.ne Il (spectra 15-19). Compared with the K ('\' preceding zone, high deciduous oak pollen values and lower herb pollen percentages. � ...... 1 Zone III (spectra 20 and 21). Lower arboreal � �I'"3,..---..;: ...A.�" .14 ,,- ,/ pollen values, mainly due to lower Quercus. I.., � " -----< Zone IV (spectra 22-24). Increasing arboreal D'-'\ ( pollen values.

Zone V (spectra 25-29). Again declining AP ��� values. Fig. I. Map of Greece indicating the core locations from which pollen diagrams have been used in this study. I. 2.2.4. General vegetation reconstruction Gravouna, 2. Tenagi Philippon, 3. Volvi, 4. Giannitsa, 5. Edessa, 6. Vegoritis, 7. Khimaditis (I & 111),8. Kastoria, The geographical setting of the area is very 9. Litochoro, 10. Ioannina 11, 11. Pertouli, 12. Xinias (I & diverse. Vegetation zonation varies from local 11), 13. Trikhonis, 14. Copais. sea-level lowlands, marshes and saline habitats to the high mountain belt further inland. The pollen precipitation caught by the lake represents these successive vegetation zones, January the mean temperature is about 10"C, quite probably with an over-representation of while in July it is over 25° C. Precipitation the lower zones. This last statement of course averages 750- 1 000 mm of which about two­ is valid inasfar as the lower zones have not thirds fa lls during the winter (Polunin, 1980). been degraded too much. The lake lies in a transitional zone of Pollen types from the mountain belt display Mediterranean and deciduous forest. For a low values in the diagram. This must have description of such vegetations, see i.a. been because the pollen-producing forest lays Polunin, 1980; Horvat et al., 1974. at quite a distance. That beech has low values is understandable as it is also hardly found in 2.2.2. The diagram Trikhonis 5 (fig. 2) the area nowadays. Abies and Ca rp in us From the core (code number 5) a diagram has orientalis/ Ostrya may be influenced by the been prepared. The pollen percentages for the long distance to the sampling location and various types have been calculated on the they are thus outnumbered by the pollen basis of a pollen sum including all types of production of the foothills and lowland areas. trees and herbs apart from typical marsh and When comparing the lowest spectra and the water plants. uppermost spectra of the Trikhonis diagram, hardly any conspicuous difference can be seen. That would be a reason to believe that 2.2.3. Zonation the general vegetation picture of the beginning The Trikhonis 5 diagram is divided into five of zone I-time was the same as that visible pollen zones on the basis of the curves of today. Such a suggestion may hold for the various typical pollen types. general vegetation cover or species compo­ sition but some details are worth mentioning. Zone I (spectra 1-14). These spectra are Compared with the lowermost spectra an considered a separate zone because Platanus increase in Mediterranean elements can be 260 S. BOTTEMA observed, a feature common in Greek dia­ the deciduous oak forest sets in and this grams. Today riverbanks are lined with process lasts up to the present (or sub-recent) plane trees, with their fresh green and time. Mediterranean xerophytic vegetation characteristic bark, a sight unknown during must have increased either absolutely or the time of zone I. The same is true for the relatively. walnut which nuts are now a common local product. 2.3. Vegoritis Dating of the diagram will be discussed in chapter 2.5. To give a rough outline, lower 2.3. 1. Geography, climate and vegetation samples date from 5000-6000 B.P. (Bottema in Creer et al., 1981). Lake Vegoritis is situated in Northern Greece During zone I-time plant cover around (40° 45'N, 21°45'E) at an elevation of about Lake Trikhonis was generally Mediterranean 570 m above sea-level in the mountains west of vegetation. Deciduous or semi-deciduous the Plain of Macedonia (fig. I). The lake oaks were present in the Arakinthos moun­ forms part of the closed basin of Ptolemais. tains and the mountains north of the lake. Only recently a drainage channel towards There also Carpinus orientalis or Ostrya car­ Edessa has been made. pinifolia and Corylus could be expected. The Voras Mountains on the west side and During subzone Ib-time fa rming and the Vermion Mountains on the east side of the grazing will have affected the forest and lake are mainly composed of limestone. provided open space for herbs. This is Towards Ptolemais in the south alluvial concluded from the appearance or increase of deposits predominate. a group of pollen types discussed in chapter 6. Details on the climate can be fo und i.a. in At the beginning of zone II a sudden change the study of the cores from Lake Khimaditis takes place. Pollen of farming indicators (Bottema, 1974) at about 20 km to the suddenly decrease. Mediterranean xerophytic southwest. Accord ing to Polunin (1980) the elements demonstrate much lower values, a rea belongs to the Central European region whereas deciduous or semi-evergreen oaks as to climatical division. The average January markedly increase (the deciduous oak pollen temperature is between -5 and ('f C and the type includes semi-evergreen species which average for July measures 20 to 25° C. occur in Southern Greece as well). As to the vegetation the reader is also At the same time pollen of Juglans and referred to the description of the vegetation of Platanus appears (the behaviour of these two the Khimaditis area (Bottema, 1974). At types. will be treated in detail in chapter 5). present the vegetation around the lake is very The level of this palynological change is much degraded, mainly due to overgrazing. visible in the sediment where a narrow band of On slopes shiblyak-type shrub can be found. volcanic ash is found (fig. 2). A discussion of Species observed north of the lake include: the age of zone 11, the volcanic band, the Quercus macedonica, Q. pubescens, Juniperus radiocarbon dates and the palaeobotanical cf. oxycedrus, Carpinus orientalis, CQf'nus, results is presented in chapter 2.5. 1. (see Amygdalus, Secale montanum, Aegilops, Bottema, 1980 and Bottema in Creer et al., Stipa, Bromus and about twenty other herb 198 1). species. The Voras and the Vermion Moun­ After zone II-time herb types increase. tains have in some parts reasonably well pre­ From the tree types only deciduous types served sweet chestnut, beech and pine for est. demonstrate lower pollen values. The other tree pollen types remain the same or show a 2.3.2. The diagram Vegoritis 8 (fig. 3) slight increase. Changes in the vegetation must have taken place in the deciduous forest The diagram prepared for the Vegoritis 8 core zone, where Quercus was common. is calculated on the basis of a pollen sum Zone III is of relatively short duration and including all pollen types apart from those of marsh and water plants. soon AP percentages start to increase in the next zone IV. Again some form of forest 2.3.3. Zonation regeneration takes place that comes to a definite end in uppermost zone V. The Vegoritis 8 diagram IS divided into two puring zone V-time the final degradation of pollen zones. Palynological investigations in Greece 261

Zone Y (spectra 1-9). The boundary between arguments for dividing this zone into two this zone and the next zone Z is laid where a subzones. The increase of beech pollen is sharp decline in coniferous pollen can be dated to 3995±60 B.P. (GrN-6596) in observed. Zone Y is subdivided into subzone Khimaditis I and 4080±55 B.P. (GrN-6600) in Y I (spectra 1--4) and subzone Y2 (spectra Kastoria. Together with Fagus, pollen of 5-9). The distinctive features of subzone Y I herbs like Artemisia, Cerealia-type, Liguli­ compared to Y2 are 'very low values for Fagus florae, Polygonum aviculare-type increase. and the absence of culture indicators. In The increase of Fagus is not restricted subzone Y2 pollen of Cerealia-type, Cheno­ to this part of Greek Macedonia but is also podiaceae, Artemisia, Sanguisorba minor/ found southwest of the Pindus (Bottema, Poterium and Polygonum aviculare-type is 1974) and in Tenagi Philippon, Eastern Mace­ met with. donia (Turner & Greig, 1975). In other parts of Greece, for instance Thessaly (Athanasi­ Zone Z (spectra 10-27). Characteristics that adis, 1975; Bottema, 1979), Boeotia (Turner lead to the establishing of this zone are, apart & Greig, 1975) and Akarnania (this paper), from the sharp decline of coniferous pollen beech pollen is hardly present also after 4000 types, the appearance or increase of Juglans, B.P. Platanus, Vitis, Plantago lanceolata-type, It was stated previously (Bottema, 1974) Cerealia-type, Gramineae, Rumex acetosa­ that the increase of Fagus pollen percentages type, Po lygonum aviculare-type and higher pointed to an increase in precipitation. In the values for Pteridium spores. The following Vegoritis diagram this hypothesis is supported subzones have been established: Subzone Zl by a slight increase in values of Ericaceae. (spectra 10-15) displays low AP values. In This is more clearly seen in the other diagrams subzone Z2 (spectra 16-20) an increase in tree from Macedonia. It may sound plausible to pollen is visible, i.a. caused by Pinus and connect an increase of beech with an increase Abies. During the next subzone Z3, AP values in precipitation. At the same time an increase are again lower. Whereas a decrease in in fa rming activity is concluded from the Quercus percentages can be seen, Juniperus pollen curves of a group of herb types. Is there demonstrates a slight increase. a connection between increasing precipitation and increasing farming activities? The ex­ planation of the pollen curves discussed here General vegetation reconstruction 2.3.4. remains problematical. The pollen curves obtained for the Vegoritis 8 The boundary between zones Y and Z is core inform us about the vegetation devel­ laid where a marked decrease in AP values opment of the area around the lake for about takes place, caused by a decline of Pinus the last 6000 years (see chapter 2.5.). Inform­ pollen percentages from 50% to less than 5%. ation on the late Quaternary vegetation Abies displays the same tendency, about 20% development of this part of Greece can be ob­ at 340 cm and a complete absence at 300 cm. tained from the diagrams from the lakes and It took about 200 years (assuming a marshes of nearby Khimaditis, Edessa and constant sedimentation rate) for the conifer Kastoria (Bottema, 1974). forests to be destroyed. At the same time the During the time of zone Y I, forest was the herb types discussed in the part on zone Y2 dominant vegetation around Lake Vegoritis. show an .important increase. Whereas the Most of this forest was formed by pine and fir conifers seem to have disappeared from Greek that probably grew on the higher ground. Macedonia to a large extent, some other tree Deciduous forest was dominated by deci­ pollen taxa increase. Did they only relatively duous oaks with an admixture of oriental increase because coniferous pollen dis­ hornbeam and/ or Ostrya carpin(folia, Ulmus, appeared, or do they represent an increase Tilia, Corylus and Carpinus betulus . Low in the number of trees? The increase of herb pollen percentages point to forest with a Quel:cus coccifera-type and Olea seems to be generally closed canopy. connected with the increase of certain herb Beech is present during subzone Y I-time in pollen types. Deciduous oaks and juniper very low numbers, the pollen percentages seem to profit from the decrease of the two measuring about 0.3-0.8%. The change in the coniferous types. Fagus pollen curve in spectrum 5 is one of the Replacement of Pinus by Quercus is a very 262 S. BOTTEMA

common feature in Turkey (van Zeist et al., bescens at lower elevations. This type does 1975). Cores from SagUt and Bey�ehir not seem to be affected by the events. Calpi­ demonstrate that after the destruction of pine nus orientalis can withstand grazing reason­ forest, abandoned areas regenerated towards ably well and regenerates quickly (Turrill, oak forest. The increase of Quel"CLIS pollen 1929). percentages can be either a relative increase, The increase in herb pollen values is caused caused by lowered coniferous values or an by types that represent different habitats. increase caused by an expansion of oak Some of them point to agriculture and/ or invading cleared areas. The increase of waste lands, others indicate grazing. Juniperus pollen points to open vegetation. Not much can be concluded from the Light-demand ing juniper profits from forest presence of the Gramineae pollen. They can clearing followed by extensive grazing. be strictly local, but if this is the case then The dominating tree pollen types in the generally they fluctuate more widely. They Vegoritis diagram are accompanied by other may originate from the grazed areas in the tree types with much lower values, but with mountains or from arable fields or fa llow land the same palynological behaviour. The in the valleys. In practice they originate from appearance of Quercus cocci/era-type (most various sources. The increase in Gramineae is likely representing Q. cocci/era as Q. ilex is at least an illustration of the landscape restricted to Eu-Mediterranean conditions) becoming more open. points to an increase in xerophytic shrub Some herb pollen types increase from one developing from over-exploitation by grazing tenth of a percent to a few percent. These etc. Such pollen must originate from relatively mostly belong to insect-pollinating taxa. This fa r away south of the Vermion Mountains as is an indication of an increase in the amount even today Quercus cocci/era does not occur of open space. The increase of Sparganium­ in the area. type is not easily explained as it has no clear The same is true for Olea. The nearest olives connection with the other pollen types dis­ occur in small numbers in the foothills of the cussed. There is also no parallel found in Vermion Mountains towards the Plain of the nearby Khimaditis cores. Macedonia, where they seem to be restricted In the group of spores, Pteridium draws by the cold 'Vardaris' winds. Hegi (1926) sug­ attention. Soon after the decrease of the gests the 4° January isotherm as the limit coniferous values, Pleridium reaches values of for Olea. Lake Vegoritis has an average up to 20%. Bracken must have been growing January temperature of about 3° C, calculated on clearances, including places where pine and after Philippson (1948). It is very unlikely that fir had disappeared as was observed in higher winter temperatures were higher dur ing zone parts of the Pindus. ZI-time, as the upper spectra show about the The important change in the Vegoritis 8 same values for Olea and they represent the diagram that leads to the division into zones Y link with the modernclimate. The Olea values and Z is stressed yet again by the behaviour of together with those of Quercus cocci/era must the green algae Pediastrum boryanum and P. be ascribed to long-distance transport. duplex. These are found in zone Y but are The changing situation also affected some almost completely lacking in zone Z. The tree species that grew in the deciduous forest decrease may have been caused by a change in zone. In contrast to the increasing oak pollen water depth, but a change of the mineral there is a gradual decrease of Ulmus, Tilia and composition of the influx of the lake cannot Calpinus betulus which soon disappeared be excluded. The increase in indeterminata in from subzone Z I altogether. These three this case does not point to corrosion as the pollen types are produced by species that preservation of the pollen grains is good . The form part of the deciduous forest at higher increase must be caused by a greater diversity elevations. They must have formed a small resulting in more unidentifiable grains. part of a mixed oak forest. Considering the It is of course not impossible that two pollen curve of Calpinus orientalis/ Ostrya, factors, viz. a climatic change and human there seems to be no change. Especially exploitation are responsible for the important Calpinus orientalis will have occurred in changes in the vegetation. Could human combination with for instance Quercus pu- exploitation have been initiated or triggered Pa[yno[ogica[ investigations in Greece 263 by a climatic change? It is suppose that at t e 2.4. Lake Volvi beginning of subzone Y2 an ?Incr ease �In precipitation took place. The course of events 2.4. 1. Geography, climate and vegetation that led to the establishing of the zone Y jZ Lake Volvi is situated in the eastern part of boundary will be discussed in chapter 6. Greek Macedonia (40° 45'N, 23° 30'E), north Subzone Z2 is a period during which forest of the Chalkidiki peninsula (fig. I). The regenera ted t·o some ex tent. A P percentages elevation is about 100 m above sea-level. Lake increase, the group of possible indicators of Volvi lies in a valley that leads to the east to fa rming activity decrease. the Gulf of Orfanou. The southern ridge, Forest from the mountain belt regenerated mainly Tertiary conglomerates and crystalline but only slightly compared to the previous rock , measures 1165 m in Mount Kholomon. zone Y. Compared with that pcriod moun­ In the north the valley is bordered by tain forests are mai nly composed of beech mountains including Mount Kerdhillion (1092 and the number of pine and fir trees must m) that mainly consists of crystalline rock. have been low. A deciduous species that Precipitation in the Volvi area is about 5 0 will have played a role in forests on higher mm, of which about the same amount fa lls 0In elevations is the hornbeam. Carpinus betulus winter as during the summer (Philippson, has even the same values as it had during zone 1948; Polunin, 1980). This makes the Volvi Y. area even drier than Thessaly. The aver­ During subzone Z2 Platanus, that was still age January temperature is about (p C, the rare during su bzone ZI, beca me very well average for July measures over 25 C. The established. The percentage of a bout 1-20/£ modern vegetation is very much distu� rbed . point to a normal occurrence of the pla ne tr e � Remnants show elements of a shiblyak that is under-represented in the pollen rain (Turrill, 1929). Paliurus spina-christi is v r (Bottema, 1974). For a detailed discussion see . � � common. On ridges towards the Chalkldlkl chapter 5. 1. peninsula , Quercus cocci/era shrub is fo und. During subzone Z3 man launched a new In 1963 W. van Ze ist and the present author attack on the vegetation. This was especially visited Lake Volvi but could not carry out a fe lt at lower elevations where deciduous oaks coring on the edges of the lake. In a dense and Carpinus orientalis/ Ostrya suffe red very much. high marsh vegetation on the east side only 10 pollen remained more or less at the same lev l. � cm of clay was collect-::d before coarse sand Maybe it was the oriental hornb am that stlll � . was hit. survived gr azing or even profited from It (Turrill, 1929). An increase in grazing pressure can also be concluded from the increasing 2.4.2. 171e diagram Volvi 6 (fig. 4) Juniperus values. From the cores the one with code number 6 When one considers the treeless, ba rren was analysed . The pollen percentages for the mountains of the present-day Vegoritisj Khimad­ va rious types have been calculated on the itis area, the level of about 50% tree pollen ba sis of a pollen sum that includes all types found in the upper spectra is very high. This apart· from those of aquatics and marsh pollen must originate from the relatively fe w plants. tree stands and scattered trees which are found in the area . The erosion and overgrazing are 2.4.3. Zonation responsible for the re latively low pollen values of the herb species, that in fa ct do not profit The Volvi 6 diagram is divided.into five pollen much from the available open space. zones: When the same pollen sum is used for the sa mples of Khimaditis Ill, from the middle of Zone A (spectra 1-5). This zone is defined by that la ke, about the same values are found. In relatively low AP values. Some herb types are the case of Khimad itis I much lower AP found in relatively important numbers, name­ values would result from such a calculation ly ·Artemisia, Plantago spec., Humulus/ because of the high local production of i.a. Cannabis, Cerealia-type Sanguisorba mi­ Gramineae which also played a part in the nor/ Po terium, Rumex-acetosa-type, Rumex peat formation. hydrolapathum-type and spores of Pteridium. 264 S. BOTIEMA

Of the trees Olea pollen IS very well repre­ again to dominate over Mediterranea n sented. xerophytic elements. Pinus and Fagus in­ creased, the latter on higher elevations, Zone B (spectra 6- 1 1). This zone is charac­ further away. The increasing shade caused terized by an increase of tree pollen, especially Juniperus to retreat. The role of Vitis and deciduous oa k, Carpinus orientalis/ Ostrya, Juglans during zone B-time is less important Pinus and Fagus. Pollen percentages of Ouer­ than before and afterwards. An explanation cus cocci/era-type are declining. for such a regeneration of natural vegetation will be given in chapters 6 and 7. Zone C (spectra 12-15). Herb pollen values, For a comparison of the Lake Volvi especially those of Gramineae show a clear diagram the nearest palynological inform­ increase. ation are the diagrams from the Plain of Macedonia and adjacent mountains in the Zone D (spectra 16-18). This zone is west (Bottema, 1974) and the Tenagi established for the same reasons as zone B. Philippon diagrams (Wijmstra, 1969; Turner & Greig, 1975) in the east. The study by Zone E (spectra 19-23). AP values, especial­ Turner & Greig covers the period present in ly Quercus cerris-type demonstrate a pro­ the Volvi core. Their study has the advantage nounced decrease. of three radiocarbon dates. Palynological comparison does not shed much light on the dating of the Volvi diagram. In Turner & 2.4.4. General vegetation reconstruction Greig's Tenagi Philippon diagram the indi­ The Volvi diagram shows a series of stadia cator type Juglans is hardly present whereas during which the vegetation, especially the Platanus is totally absent. forest vegetation, was affected by man, whereas in other episodes some regeneration 2.5. The oating of the cores of Trikhonis, took place. Vegoritis and Volvi At the beginning of the diagram, zone A­ time, human influence upon the vegetation in 2.5. 1. General the hills around Lake Volvi is clear and indi­ cated by pollen of Vitis, Juglans, Platanus, Dating formed an important part of the study Artemisia, Chenopodiaceae, various Plan­ on the geomagnetic variations in Greece by tago types includinglanceolata, Humulus/Can­ Creer et af. (198 1). In addition to dates nabis, Cerealia-type, Rumex acetosa-type, obtained by comparison with geomagnetic Po lygonum aviculare-type, Xanthium, Urtica studies mainly from the U.K., radiocarbon and spores of Pteridium. Such influence dating was done on nearby cores in the three must date back from before zone A-time lakes. To supply more information on the already, but there is no information from possible ages of the sediments, the cores were older deposits. studied palynologically. The dates obtained The influence of man during zone A-time by the three dating methods are plotted in fig. must have been considerable. Maquis with 5. Quercus cocci/era and Ericaceae was already The radiocarbon dates, inferred dates from common. Relatively high percentages of Olea nearby cores, are invariably too old compared pollen point to olive yards. Deciduous oaks with the information obtained from the two and Carpinus orientalis/ Ostrya had suffered other methods. Lack of organic material in from cutting and/ or grazing and probably so the sediments and the effects of hard water, had Pinus. In degraded areas Juniperus must common in Greek lakes with water supply have spread as is concluded from values of from limestone regions, may give unreliable about 5%. Trees like Platanus and Juglans as results. well as Vitis have been favoured by man. The The palaeomagnetic dates have been various herb types mentioned above for zone obtained by comparing with dated sequences A point to intensive fa rming. from ' the U.K. and some information from In the following period, zone B-time, the Southeastern Europe. Creer thinks that the vegetation in the Lake Voivi area regenerated Greek records are not more than 150 years to some extent. Deciduous elements increased older than the corresponding features in the PalynologicaZ investigations in Greece 265

LAKE VOLVI found at about 550 cm. As in Vegoritis the . "

;1t 1 J---( dQlu E C-IL assumption that the tie-line represents the gol _ pollendQhs Pompeii eruption of A.D. 79 could be right. /lI'lngnelic.nge-depthturvu For one metre of sediment pollen of Jug/ans 5 d'pthinm and Platanus is already present and could even have been present below that level, but LAKE TRI)(HONIS. LAKE VEGORITIS that cannot be checked here. In Trikhonis, however, the tie-line almost � I completely synchronizes with the increase of Platanus and Jug/ans pollen curves. The archaeomagnetic choice would be A.D. 79, the palynological choice would be 3000-3200 B. P. Two possibilities remain, either the volcanic ash layers are not from the same eruption or the appearance of Platanus and Jug/ans did 5 5 d.pthinm d.plhinm not occur at the same time in mainland Fig. 5. Palaeomagnetic, palynological and ra diocarbon Greece. age-depth curves (after Creer et al. , 1981) for the lakes of Palynological dating was done by com­ Vegoritis; Volvi and Trikhonis. paring the three diagrams with- other pollen diagrams from the area which were radio­ carbon dated. Such dates were obtained mainly from peat and gyttja samples. The U.K. record. The palynological dates are number of clear palynological events in the either in agreement with the archaeomagnetic younger Holocene of Greece is very small and dates or they are older. Creer et al. mention thus possibilities for accurate dating are sharp peaks in the intensity and susceptibility restricted . logs for all three lakes. They used one of these For the complete Holocene six pronounced sharp peaks as a correlation tie-line. These tie­ palynological zones can be distinguished for lines are associated with a band of volcanic the area north-east of the Pindus and five material. Creer et al. think that the eruption zones for the Epirotic side of that mountain that caused this volcanic band is likely to be chain. From such evidence it is concluded that the A.D. 79 eruption of the Somma-Vesuvius the three cores only cover part of the connected with Pompeii and not the Santorini Holocene from about 6000 B.P. onward or as eruption (dated 3370± lOO and 3527±44 B.P.). is the case in Lake Volvi even much later. They concluded this on the basis of the The palynological ages of the three cores preliminary time-depth curves and i.a. the have been discussed before (Bottema in Creer prevailing wind. The present author ascribed et a/., 198 1). In fig. 5 the estimated ages are the volcanic band in the Trikhonis 5 core to given. The palynological argumentation is as the Santorini eruption on the basis of follows: palynological evidence (Bottema, 1980). The archaeomagnetic tie-lines are to be 2.5.2. Lake Vo /vi found as follows : Vegoritis 8 at 270 cm; Volvi 6 at 550 cm; Trikhonis 5 at 300 cm. It can be The oldest part of this core must be younger read from the diagrams (figs. 2, 3 and 4) that than 4000 B.P. as Fagus is already present in the P/atanus and Jug/ans curves do not start the lowermost spectrum. In the nearest site, at the same time with regard to the Tenagi Philippon (Turner & Greig, 1975), at archaeomagnetic tie-line. The appearance and about 80 km to the north-east, the appearance spread of Platanus and Jug/ans is thought to of beech pollen is dated 4193±120 B.P. have happened at about the same time in (BLN-955). The presence of Jug/ans at that Greece around 3200 B.P. (Bottema, 1974). lev�l in Lake Volvi dates the sediment even In Vegoritis 8 the beginning of the P/atanus younger than 3000 B.P. The rather high Olea and Jug/ans curves can be dated about 3000 values, found at 400-600 cm, together with B.P. when the magnetic age-depth curve IS Ce realia percentages, may point to the end of used. the 'Classical' period and the beginning of In Volvi 6 the archaeomagnetic tie-line IS the 'Roman' period at about 200 B.C. 266 S. BOTTEMA

(Athanasiadis, 1975). No specific indications (GrN-6596) and in Kastoria at 4080±55 B.P. on the basis of pollen behaviour for the last (GrN-6600). A date from Edessa of 5260±65 2000 years in Greece are known at the B.P. (GrN-6 185) fo r this event may be moment. For that reason the changes in the unreliable as a hiatus occurs just above the AP/N AP ratio as suggested by Athanasiadis location of the rad ioca rbon sample. (1975) have been used as a possible time-scale. The level of 300-340 cm shows the appear­ Thus, the decrease in tree pollen and the ance of walnut and plane tree. According increase of Cerealia-type pollen at about 350 to the dates of 3135±170 B.P. (GrN-6182) cm may represent the end of the Byzantine and 3280±55 B.P. (GrN-6184) for two cores period and the beginning of the Turkish of nearby Khimaditis, this event must have occupation. The level of about 250 cm which taken place in Vegoritis about 3100-3300 B.P. demonstrates again an increase of AP values Finally, at 30 cm, the sediment must be may point to the return of Christian refugees. younger than c. 250 B. P. as Zea mays was According to Athanasiadis, the mountain found there. forests regenerated to some extent when the refugees returned to the low lands. Finally, the level of 120 cm must be younger than c. 3. THE REPRESENTATION OF POLLEN 250 years as can be concluded from the TYPES IN VARIOUS SOURCES presence of Zea mays pollen. 3. 1. Some remarks on the practical limits of pollen analysis with emphasis on the 2.5.3. Lake Trikhonis Greek situation The lower part of core Tr5 must be dated younger than 6500 B.P. as is indicated by the In the Balkans about 6500 seed plant species presence of Carpinus orientalis/ Ostrya. A occur (Tun'ill, 1929; Polunin, 1980). That comparison with two diagrams from loannina means that the number of plant species in (Bottema, 1974), 180 km to the north, suggests Greece amounts to a few thousand at least. an age between 5000-6000 B. P. based i.a. on What is the palynological translation of a the behaviour of the Cory/us curve. Fagus few thousand Greek plant species'� From pollen cannot be used here for dating as the practica l experience it is known that the beech is and was hardly present in this area . number of pollen types ident ified with a light The level of 300-320 cm is dated at 3 100-3300 microscope is lower than the number of plant B.P. accord ing to the curves of Jug/ami and species producing such types. On many P/atanus (Bot tema, 1974). The presence of occasions pollen grains are identified above Zea mays dates the level of 110 cm younger the species level. than about 250 yea rs. The information of this As an example the number of plant species diagram will not form part of the discussion in from a plant-sociologica l study of forest this study as there is no agreement on the vegetations in Centra l and Southern Greece dating. (Barbero & Quezel, 1976) is considered in terms of the number of pollen types 2.5.4. Lake Vegoritis represented . Barbero and Quezel mention 479 species. A number of these species share the Diagrams from Khimaditis, Edessa and Kas­ same pollen type, thus lowering the number of toria situated at 20, 15 and 50 km respect­ pollen types that can be identified . Besides, ively, are used to date the Vegoritis diagram. the number of pollen types identified depends The lowest sample must be younger than to a great extent on the technical expertise of 6500 B. P. as CGlpinus orienlalis/Ostrya is the palynologist. H is knowledge and personal already present. The Ostrya-type increase in limits to absorb morphological characteristics, Edessa is radiocarbon-dated to 6385±55 the presence of a good reference collection, B.P. (GrN-61 86). Values of Ti/ia and V/mus ident ification tables, key ca rd systems and suggest that the lowest part is not much picture collections are also a decisive fa ctor. younger than 6000 B. P. Tne present author was able to identify From 400-4 1 0 cm an increase of Fagus about 200 pollen types in the Greek sediments. pollen percentages is visible. At Khimadi­ Some types remained unidentified . These tis, this event is dated at 3995±60 B.P. types were discussed with colleagues in the Palynological investigations in Greece 267

150 department, but such discussion was never I very fruitful as the team members share more or less the same knowledge. The identification limits make the method not as accurate as one - would like. The diversity of a pollen spectrum : remains hidden when a large number of Fig. 6. Number of pollen types in three sources of origin. species share · the same pollen type. I. A series of plant relevees studied by Barbero & Quezel Under the given conditions the present (1976), 2. The lakes of Vegoritis, Volvi and Trikhonis, 3. author would recognize 125 pollen types in 110 Greek surface samples. Barbero and Quezel's 479 plant species. Such types represent .taxa, varying from species to family level. samples, 36 types found in the former In the pollen diagrams of Vegoritis, Tri­ category not being present in the latter, whereas khonis and Volvi (chapter 2) a fraction of 31 types in this latter category were not met the 125 pollen types from Barbero and with in the subfossil samples. The difference in Quezel's list were found . Compared with the composition of the three sources has several 125 pollen types recognizable in the plant­ causes. The re levee-list gives exclusively sociological relevee-list, only 70 types were forest species, translated into pollen types. found in the samples of Lake Vegoritis. For The relevee-list includes for instance insect­ Lake Trikhonis and Lake Volvi these numbers pollinated species which pollen types were are 71 and 51, respectively. never met with in subfossil or modern surface Another source of information on the samples. The lake-sediment samples represent diversity is a series of surface-sample spectra the pollen precipitation from a large area as from Northwestern and Central Greece well as local water plants. In contrast surface (Bottema, 1974). The 110 surface samples samples catch a lot of local pollen but have 73 pollen types in common with the 125 generally no water plants. derived from the list of Barbe ro and Quezel. Conservation of pollen grains in surface This number is of the same order as the samples is never as good as that from number of types found in the sediments of the sediments cored in the centre of a lake or three lakes. It should be stressed that the a marsh that never dried up during the qualitative aspect, viz. the number of pollen accumulation of organic material. On the types from three completely diffe rent sources other hand, surface-sample spectra are is discussed here. First, a plant-sociological generally accompanied by a vegetation relevee-list of 479 names of species of which description that can be of important help in the palynologist maintains that he could the identification of little known pollen grains. group 125 pollen types representing these 479 Finally modern imports are found hardly or plant species. Then, 152 pollen types identified not at all in subfossil samples whereas in in sediment samples, giving an idea about the surface samples they may show up. pollen rain produced by past vegetations. Finally, the modern pollen precipitation from 3.2. Origin of pollen a large part of Greece derived from samples taken from amidst the vegetation and deli­ Above it is explained that differences in vering 144 pollen types, as far as the palyno­ composition .exist between the pollen types logist was able to identify. derived from a series of plant-sociological As mentioned above the three sources do relevees, subfossil pollen spectra or modern share a number of types, but there are also surface-sample pollen spectra. Any specific types which are characteristic for a special source includes or excludes certain pollen source only. Slightly more than 70 pollen types. types from lake sediment samples do not have In the following, three clearly defined counterparts in the re levees of Barbero and groups are brought forward. The third group Quezel and this also holds for the Greek includes the crops and weeds that will form surface samples. In fig. 6 the presence of part of the theme of this study. pollen types in the three sources is shown schematically. It can be seen that there is also Pollen derived fr om long-distance transport a difference between subfossil and surface (I) .. The pollen types derived from Barbero 268 S. BOTTEMA

and Quezel's relevee list represent herb or tree Table 1. Category HI pollen types, found in a series of Greek species mentioned for the various locations surface samples and three Holocene lake sediments. Plant and listed in the relevees. In su bfossil sediment species representing these pollen types were not present in a samples and surface-sample spectra, pollen relevee-list compiled by Barbero & Quezel (1976).

from local and more distant sources is present. Present in four sources Present in three sources Pollen of Betula and Picea is found with (surface samples and three (either lakes or surface about the same values in the modern pollen lake cores) samples) rain in Greece as in subfossil samples. Apart Artemisia Aq uilegia-ty pe from occurrence of Picea abies in the Greek Asphodelus Centaurea scabiosa-type Rhodope Mountains (Zoller et al., 1977) there Centaurea cyanus-type Echium-type are also few indications that Betula occurs in Centaurea solstitialis-type Heliotropium Greece (Tutin et al., 1964). Generally, pollen Cerealia-type Nigella finds of Picea and Betula must be ascribed to Chenopodiaceae Plantago major-type Delphinium-type Solanum nigrum long-distance transport. Aumulus/Cannabis Spergula(ria)-type Papaver Urtica Pollen derived from water and marsh plants Plantago lanceolata-type Urtica pilulifera (ll). It is self-evident that this group of pollen Plantago coronopus-type types is very well represented in samples from Plantago maritima-type lake sediments and a bsent from relevee-lists Plantago media-type Polygonum aviculare-type from forests. In surface samples this group is Polyg(;mum hydrolapathum-type rare and highly dependent upon the selection Vitis of the surface-sample locality which is Xanthium preferably in or near undisturbed forest. Zea mays

Pollen derived fr om crop or weed plants (Ill). To study possible indicators of prehistoric farming, taxonomically narrow groups, prefer­ they are included with the group of crop and ably species, have to be traced. It is clear weed plants. An advantage with these three that such species, and pollen derived from types is that in the Mediterranean no relatives them, do not occur in the relevee-lists with the same pollen type occur. compiled by Barbero and Quezel, that repre­ This selected group of pollen types, as far as sent natural or fairly natural forest. a more or less closed curve is present, is The last category, of crop and weed plants, depicted in a scheme (fig. 22). The presence of has been looked at more closely. It is assumed the pollen types is given schematically in that forest relevees do not include crop and percentages for the last 11,000 years. A curve weed species. If pollen originating from long­ of Gramineae is added, demonstrating open distance transport and pollen types from landscape versus forest. This group of pollen water and marsh plants are excluded , a group types will be treated in chapter 6. of pollen types remains that is ascribed to the last category. These pollen types, present in all three lake sediments and in the surface 4. DIVERS ITY OF POLLEN samples but not in the forest relevees, are ASSEMBLAGES COMPARED WITH shown in table I. DIVERSITY OF VEGETATION

The method followed above clearly misses 4. I. General one group of plants or, as the case may be, pollen types. These are anthropogenic In the discussion of pollen diagrams or elements that escaped to or were brought to palynological information that is presented in forest. Thus Barbero and Quezel listed some some other way, the pollen types found are species in their re levees that in fact are discussed for instance according to their elements of the forest at the present time, but mutual relations in the form of percentages. that were origi nally absent from Greece. Diversity of assemblages is seldom a matter From the ' palynological evidence Juglans, for discussion. Platanus and Castanea all appear about 3000 In this study diversity simply means: the B.P. As possible indicators of human activity number of pollen types identified in a sample Palynological investigations in Greece 269

ar the number af plant species identified in 4.2. Diversity versus pallen sum and arigin a r�levee. Diversity as an aspect af planteca­ af samples laglcal studies is a mare camplex phena­ If the vegetatian were to.re main canstant aver menan than in the sense in which this term a lang periad af time, the number af pallen is used here (Odum, types identified would mostly depend upan It is widely knawn19 69)that. a natural clased . the size af the pallen sum. The larger the farest is less · diverse than a farest i which pallen sum the mare pallen types are faund. clearings are made. When apenings ar�e made Hawever, after the pallen sum has reached caus d by waad cutting ar far farming ar � certain evels the number af types will hardly grazmg, passibilities are created far new ! ar nat mcrease any mare. species that did nat accur ar were scarce in the On cauld avaid the first prablem by farest befare. An impartant element that is � countmg a canstant number af pallen grains added by apening up farest is penetratian af and then study the diversity af the spectra. light. There are practical difficulties invalved in Ta what extent are these changes in the field c u tin a canstant pallen sum. The visible in the pallen rain? When the attack ? � p distributIOn af the pallen grains under the n the natural vegetatian praceeds, an ��� . caverglass af the caunting slide is nat even. In mltlal enrichment in species is later fallawed general small g ains mave to.th e edges mare by degradatian, a lawering af the number af � . than large grams. In preparing a caunting species. F arested si apes are fi rst apened then slide pressure is aften exerted upan the caver­ defarested, g azed, avergrazed, denuded by . � glass to.spr ead the liquid (far instance silicane erasIOn and fmally anly a limited number af ail) in which the sample is embedded and unpalatable ar prickly herbs is left . this will alter an even distributian. Such At a certain stage the number af pallen xperience is a warning nat to. stap at far types, the diversity, will be maximal. The � mstance caunted grains but to. count the caurse af the diversity in time can be complete500sl ide. ith caunted grains the campared with the diversity af a regenerating . number mcluded m� the pallen sum is meant. farest an abandaned fa rmland as described by . Thearetlcally ane cauld caunt half a slide but Odum There the number af species is this half will also. rarely cantain grains. l w at (lfirs?69t,). then steadily increasing until ? When passible a pallen sum af abaut500 ar fmally, t wards a mare stable situatian, again ? mare is counted in aur studies. 1000 lawer diversity is faund. In the situatian It is stated that the pollen sum includes discussed far Greece the arder af events differs anl a certain number af types: thase types fram rege eratian as described by Odum, but � � derived fram the upland vegetatian. Lacal the trend IS the same. water and marsh plants wauld influence and As to. the arigin af the diversity in a pallen isturb t e picture very much if they were spectrum the fallawing shauld be remarked. ? � mcluded m the sum, especially because they The qualitative campasitian af a pallen can and aften do. fl uctuate independently s ect um daes nat anly depend upan the lacal p � af the pallen praductlOn af the regianal SituatIOn, as far instance pallen ariginating vegetatian. Such water and marsh plants do. fram lang-distance transpart may be present. . nat farm a pra l m when they have pallen Insect-pallinated trees and herbs are in . � � that can be Id�ntlfled easily. When they share practice nat widely dispersed and thus under­ . the type wI h p ant species fram the upland represe ted. There will be practical and . � ! � vegetatIOn It will be very difficult to. trace thearetlcal botanical aspects af this appraach this effect. Lacal influences upan a pallen but the questIOn asked here is rather simple; spectrum can be vaided at least to. a large are there clea r changes in the number af � extent by the chOice af the sediment. Cares P llen types in time, viz. in palynalagical ? . fram the middle af deeper lakes are very dl efSlty? an we campare pallen spectra in . � far thiS purpase as they reduce the thiS way, wlthaut� meticulaus carrectians etc.? gaad quantitative effect af large numbers af lacal Ta get an idea af the diversity as it is delimited pallen grains. Nevertheless, in the case af here, the number af pallen types is caunted far . Grar:'meae and Cyperaceae, far instance, the the ?epth af the samples nnd thus in relatian to. time. qualitative effect remains. Campared with surface-sample spectra it 270 S. BOTTEMA must be said that this last group cannot escape local influence due to the manner of sampling METSOVON moss cushions, often in the middle of the vegetation. In such samples the identifiability of the pollen types is increased by the 1000 15 100 description of the vegetation surrounding the • !I sample location and lowered due to I \ I \ corrosion of the pollen by exposure to I \ I ------­ \ oxygen, a process that would not occur under 1500 65 15 f I water. I " I " I Detailed vegetation description is of great " help with the identification of rare or cons­ picuous pollen types. The middle of a lake 1000 55 50 lacks this helpful detail and such pollen types found there are often included in a larger ---- AP% taxon...... numbr. ol pollfn lJPts Such factors increase or decrease the --poHrnsum ------number of identifiable pollen types in the 500 \5 15 - Fig. 7. Number of pollen types counts but they do not change the real number per sa mple, pollen sum and of pollen types or the vegetation that produces AP/ NAP ratio in a short surface­ them. The method followed thus has its limits sa mple transect near Metsovon in considering the diversity in pollen types and o 35 0 • (Pindus). hrn�e the diversity of the vegetation translated from these pollen types.

from five locations taken from various levels 4.3. The diversity of pollen spectra (fig. 7): compared to the diversity of the Cultivated area with orchards down in vegetation: An example I. the valley; Is there any relation between the qualitative 2. Herb-rich meadows with bracken and aspect-pollen types-and vegetation type? A forest at some distance; short series of surface-sample spectra from a 3. Boundary of meadow and fi r forest; study on the relation between modern 4. Within fir/ beech forest; vegetation and pollen precipitation is ver) 5. Within beech fo rest (some fir seedlings). illustrative (Bottema, 1974). When collecting In fig. 7 the number of pollen types per surface samples it is often very difficult to find surface sample, the pollen sum of these suitable transects presenting well preserved samples, and the AP/NAP ratio are shown. vegetations with clear gradients or zonations. The pollen sum of these five samples varies A short transect from the Pindus Mountains from c. 700 to 1800. The number of identified near Metsovon is very useful in this respect as pollen types amounts to 51 for the cultivated it fu lfils the cond itions mentioned above (fig. area; 49 fo r the meadow with bracken; 42 for 7). the boundary of meadow and forest; 20 for Near Metsovon one finds at high altitude a both the forest samples. It should be stressed montane forest. At the lower edge of this zone that in the spectra with the lowest number of the forest has been cut down and meadows pollen types, the pollen sum is twice that of have been developed that are maintained by the spectrum with the highest number of grazing. These meadows are partly covered pollen types. with bracken (Pferidium aquilinum) especially The diffe rence in the number of pollen types with increasing elevation towards the forest. is caused by the fact that the open vegetations When one climbs the mountain the lower are not represented palynologically in the forest zone is formed by Abies cephalonica. fores,t whereas the trees do precipitate "their At higher elevations Fagus sy/vafica appears pollen in the open fields. It is not necessarily and soon almost pure beech forest is found the case that the meadow vegetation is so with some occasional fir seedlings. much more diverse in plant species. The dense Summarizing, pollen spectra are available forest is simply less diverse than the same Palynological investigations in Greece 271

------1000 15 100 - -

1500 i5 15

� 1 \ 1\ I \ I \

1000 55 50

500 \5 15

: ; o 35 0 cultivated aru lorestrd pinr forut :. cultivated aru foreslrd�shrub delOfutrd :;

Fig. 8. Number of pollen types, pollen sum and AP/NAP ra tio in a long surface-sample transect through the Greek mainland.

forest in which clearings are made. It is clear are identified; a rise in AP percentages is that the transition, forest to open field, is paralleled by lowered diversity in any characterized by an increase in identified particular spectrum. pollen types. As the AP/NAP ratio is generally given in The number of pollen types is not only the kind of pollen diagrams shown here, correlated negatively with the degree of counting of the numbers of pollen types may forestation, it is also correlated with the not even be necessary. When it is postulated AP/N AP ratio. This is quite logical as there that ea rly fa rmers made clearings in a natural will normally be a relation between the degree forest and thus promoted diversity, it would of forestation and the AP values. be possible theoretically to demonstrate this The Metsovon example is quite clear as to cutting activity etc. by counting the number of the diversity aspect but other surface-sample pollen types. Does the information support transects show that the usefulness of this the idea that one might just as well follow the method is open to some doubt. AP/NAP ratio to get information on diversity When a long transect through the Greek and thus possibly find indications of human mainland is considered (fig. 8), it can be seen activity? that in some spectra the number of pollen In fig. 8 it can be read that this idea holds types is still correlated positively with the size for some situations but fa ils in others. When of the pollen sum. Unfortunately, botanical the course of the AP / NAP ratio is compared diversity is not known for many surface­ with notes on the basis of fig. 8 that roughly sample locations as not all plants were ind icate the state of the vegetation, dis­ identified at all locations. In fig. 8 also the crepancies show up. Cultivated areas may AP/N AP ratio is given. Also here this ratio is have the same high AP value as nearby forests correlated negatively with the number of whereas sometimes also the reverse can be pollen types, thus with diversity. The lower found. the AP percentages are, the more pollen types In figs. 9, 10 and 11 from the lakes 272 S. BOTTEMA

- LAKE VEGORITIS "ii.

.. "ii. :;; 1000 115 100 /'- / ' / ,- - -', f', //,...... , I ,-/ J I

1500 65 ; 15 ----:'------\.------�------1 \ ; :I \ . '. /', I ... / '· 1 \ :. I \:': / 1000 55 50 ,::\ ... ..//\, \/\\\ lA ': .. / \I : "

500 15 15 � - -� ------:- -\ - ---

Fig. 9. Number of pollen types, pollen sum and APj NAP ratio of the Lake '" '" )00 deplh incm '" ::soo , ! , I ! , -:, Vegoritis diagram. o 35 0 1 1 I

- � LAKE TRIKHONIS g � .. "ii. � 1000 115 1

" '1 1 '-/\ A � :'. : : 1500 65 15 --- -i - -/r - ; ------� ------?'. .------!- - -;: 0--\- - - \ - - - i\ - -/ -- --- �;- �\- �- I : ... 1 1 \\ :: . . ::r \ I \: '. \ ' ::1 \ �_...J I \ : • " ... / \f \ : ,,, ' :t \ . . / ./ :1 /.'. V \ : I: \ \ \ 1 \. .... j: \ : /" 1000 55 50 'C "\,\. . .,...... ", ...... \<»:L . . . \

500 15 15

Fig. 10. Number of pollen types, pollen sum and APj NAP ra tio of the 100 100 lOO 400 SOD drplh incm iDO --'-I o 3S 0 _IL-",---,---,--,---,-I--'---'-��-'- ' --,-�-,---,---,I--,---,---,-�---,-I ----'-' _',---,--11 Lake Trikhonis diagram.

Vegoritis, Trikhonis and Volvi, the number of the younger part of the three figures, however, pollen types, the AP / NAP ratio and the the number of pollen types is increasing pollen sum are plotted against depth and thus whereas pollen sums are reasonably constant. against time. In these three curves from time Does the curve of the number of pollen to time a positive relation of the pollen types types (the diversity) demonstrate a behaviour with the pollen sum can be seen. Especially in that fits into the conclusions from the pollen Palynological investigations in Greece 273

LAKE VOLVI

------1Il00 75 100 ------

r-__ , ....., / � \ I \ I \ A ------"' ------1500 65 75 ------:---:' - - - - - 7 ____• -\______-".c_ � _ \ / , / \ / \ / --_-J' \ /,_./ \ / \ / / / "----../ \ / \ ...... \ / / .. , . ------./- -- -' 1000 55 50 -- ,,:" ------�\­ : �

...... < 500 15 15 - -- -

Fig. 11. Number of pollen types, pollen sum and AP/NAP ratio of the Lake Volvi diagram.

curves as they are explained in chapter 2? Is the AP/NAP ratio and diversity/ number of there some connection with the (pre)historical pollen types. The size of the pollen sum is events? fa irly constant. Some remarks on the history of man in Greece are made in chapter 7. Human impact 4.4.2. Lake Trikhonis upon the vegetation is shown sche,matically in fig. 23. On the basis of the course of the In fig. 10 it ca n be seen that the diversity curve AP/NAP ratio and the grouping of the types does not correlate so negatively with the from category III an attempt is made to AP/NAP ratio as is found in Lake Vegoritis. demonstrate anthropogenic pressure upon the Maximum diversity is found in the lower part vegetation (fig. 22). of the core up to about 360 cm although here initially it is low. The number of pollen types in the lower spectra is 42 minimally and 74 4.4. Diversity of subfossil assemblages maximally with an average of 55 types. Diversity has dropped during zone 11 (fig. 2) 4.4. I. Lake Vegoritis where the volcanic ash layer is fo und. Dur ing The lower part of the Vegoritis diagram those five spectra average diversity is 48. The demonstrates little influence of man upon the AP percentages have increased during this vegetation (fig. 3). At the same time the part of the diagram. diversity is low, measuring between 33 and 55 From zone 11 on, above 200 cm a constant pollen types, averaging 42 types (fig. 9). increase of the diversity is visible. The average Around 3100 B.P. (according to uncalibrated value compared to the lower part, below 340 radioca rbon dates) the impact of man upon cm, is 54. The averages of the various zones do the vegetation rapidly increased , as did also not differ but the course of the curves does. In the diversity. During the steady decline of the the lower part important fluctuations are not AP/N AP ratio, diversity numbers range from always found to coincide with AP values, 47 to 59, averaging 52 types. whereas in the upper part a regularly After this phase (subzone 11-1, fig. 3) the increasing curve can be seen, clearly linked remaining period shows fluctuating curves with the AP/ N AP ratio. In the lower part demonstrating a negative correlation between diversity shows maxima where AP values are 274 S. BOTIEMA low. Such maxima of 65 types and more do a group of weeds is based on information not occur in the upper part. A possible from various sources (Athanasiadis, 1975; explanation for this behaviour (see fig. 10) is Turner & Greig, 1975; Bottema, 1974; 1979; that during the time that the lower part was 1980; Filipovitch, 1977; Wijmstra, 1969). deposited more natural elements were still In the study presented here the reader is present in a vegetation that was already referred to figs. 12-2 1 which give a selection of strongly influenced by man. pollen curves relevant to this matter. At the end of zone I diversity in the area dropped because of the formation of relative­ 5. 1. P/atanus and Juglans ly monotonous oak forest (deciduous/ semi­ In this chapter attention is devoted to these evergreen). In the course of time this two species which develop parallel pollen type of forest was replaced by maquis, and curves in the younger part of the Holocene thus the diversity of the vegetation had again (figs. 12-2 1). increased. It is a matter of speculation as to where 4.4.3. Lake Vo/vi plane tree and walnut originally grew at the end of the last Ice Age. There are no The general trend in fig. I1 is that the diversity indications that there were refugia in Greece curve correlates, as is fo und above, negatively and there are no indications that these trees with the AP/NAP ratio. Only a spectrum at were found anywhere in Greece before 3500 420 cm is an exception to this trend, with a B.P. This could mean that plane tree and pollen sum that is twice as high as in the other walnut were extremely rare in Greece and the spectra. The lower part (zone A-B) averages find of some stray pollen grains of Pia/anus in 45 types (min. 35, max. 55); the upper part Greece should be mentioned . These grains (zone C-E) averages 52 types (min. 47, max. could, however, be due to lo ng-distance 59). Although two main zones are established tra nsport. Pia/anus pollen could have been some short periods can be observed. These incorrectly identified, but in the case of the short fluctuations are mainly caused by easily ident ifiable pollen of Juglans this is out changes in the balance between Medi­ of the question. terranean maquis and Ce ntral European forest A pollen diagram from fo rmer Lake Xinias from higher elevations around Lake Volvi . (Bottema, 1979) on the southern edge of the During zone E the diversity decreases al­ Thessalian plain shows the presence of some though the forest does not show an Platanus pollen grains in sediments dating expansion in the form of higher AP values. It back to the first half of the Holocene. The is possible that in this case the vegetation was possibility cannot be excluded that P/atanus disturbed to such an extent that barren hills, occurred in a few localities in Thessaly dur­ like those we see nowadays, had developed, ing that time. If Platanus did occur before with an accompanying loss in diversity. the fourth millennium, it must have been ex­ tremely rare. 4.4.4. Conclusion If both species did not occur in Greece, The translation of the number of pollen types when did they appear in Greece and how? in terms of diversity seems reasonable. The Were they imported by prehistoric man or did practical use of this feature is, however, they spread over the Balkans at a time when limited. Diversity often turns out to correlate suitable conditions had developed, or was it a with the AP percentages. Thus it is easier to combination of such factors? There are no read the AP curve, that is usually available. indications that these species came via Thrace. No sharp decline in diversity could be traced Was there any reason for prehistoric man for those periods during which vegetation had to import both species? One can easily almost disappeared , a widespread occurrence understand that planting Jug/ans was ad­ in many parts of Greece. vantageous for man. Furthermore Juglans and Pia/anus have one striking point in common that may link their history. Both tree 5. DISCUSSION OF SOME SELECTED species favour riversides (Hegi, 1957). Did the TYPES available biotope, in this case riverine forests, drastically increase? Did the newcomers simply The discussion of P/atanus, Jug/ans, Vitis and replace riverine species that were there before Palynological investigations in Greece 275

: r- , .. : : - -ZEA � t- i ! r-­ : i t­ : ! t­ : i i ; i ! r­ r-- � ! JOO ! - i � � : ' ! ! ! : ! 1 I ; i , , ! ... ! !J.I :: 1: : : : : : : :

500 i : i : : !: i : : , ' :, I

600 j • 9.0 • Bp. '.0 SP.sp . '.0 . '¥l . 'P . 'P Fig. 12. Selection of pollen curves from the Lake Volvi diagram indicating antropogenic activity.

them and that had a weak competitive destruction in the mountainous hinterland. position? Erosion of the valleys and a shift of water In many Greek diagrams, especially from distribution over the seasons may have caused Macedonia but also from the Epirotic side of Fraxinus excelsior to retreat and Platanus to the Pindus near Ioannina, one species clearly spread . decreases and is palynologically succeeded by Platanus is rfIentioned in relation to a Platanus or luglans, namely Fraxinus ex­ possi ble succession of Fraxinus excelsior but celsior. One may observe this fact, but how the walnut may be a more likely successor as can it be explained? Fraxinus excelsior grows luglans prefers a deeper soil and not a pebble in riverine forests with deep soil and avoids bed . Under (semi-) na.tural conditions luglans i.a. pebble beds. Platanus orientalis can often regia is found in the Balkans in ravine forests be found along and in rivers with a pebble at the present time. bed, a situation that is very common in Greece If the hypothesis holds that river valleys nowadays. The water regime of such a river is changed due to a change in water regime, in a high water transport during the winter turn caused by deforestation, then the cause months and little or no transport during the should eventually be visible in the pollen Mediterranean summer. Such a regime is the diagrams. In part of the Greek diagrams result of climate combined with ruthless forest studied such a deforestation is indeed visible 276 S. BOTTEMA

EDESSA

IO'

.. ,

'"

Fig. 14. Se!ection of pollen curves from the Edessa diagram indicating antropogenic activity.

(Bottema, 1974; 1979) and in this study for diagrams is anthropogenic activity. instance in Yegoritis' zone Z. This pheno­ If climate did not cause a change in river menon is only found in the diagrams for beds then the influence of man might have led Macedonia. An increase in NAP values, to the forming of pebble beds. When caused by pollen types derived from culture prehistoric man destroyed all the mountain indicators, accompanies the beginning of the forests, erosion will have taken place (as one P/atanus/ Jug/ans phase. When climate, with can see nowadays). Heavy rain would have increased precipitation, causing increased washed the soil down the slopes into the discharge , became established, such a change valleys and finally to the sea. As there was no must have influenced the natural vegetation. vegetation of importance that held the water, Instead the only clear picture in the pollen providing a regular water supply to the river Palynological investigations in Greece 277

VEGORITlS

w � >

<

�w " w u

'" '0 30 '0 50 60 io ' 80 '90 ! : : , , , i ! i :

i I I� : I i : : : I : I : : I I I , , ! : : : , i : : : � !, I : : , � : i : : : : : ! : : i : i ! : i I : : 300 I I i : : ! i lIE : , , : , : : : ; : : , : i : : I : , i 1 ! : i i i :, ! , I i I , i , : ! , i ! ! ! : 500 l: j ! : : ! i : : �c'1' ',0 'I' ',0 , ',0 , 'P ',0, , 'P I 1 I I Fig. 15. Selection of pollen curves from the Lake Vegoritis diagram ind icating antropogenic activity.

system, such streams would almost be very well as they shed their ba rk every yea r. It torrential after heavy rain, leaving behind is, however, doubtful whether air pollution only bigger elements, for instance rounded played a role during the Greek Bronze Age. pebbles. This picture would not appear Planting plane trees to provide shade seems a suddenly, but it would develop gradually as better suggestion, as can be appreciated In fa rming practices continued. There are no Mediterranean countries nowadays. signs that these activities occurred on a wide scale already before the appearance of Jug/ans The geomorphological history of river valleys and P/atanus. does not indicate any very important changes The combination and the values of pollen of around 3000 B.P. The so-called 'Younger fi ll' types which indicate fa rming activities include is dated by Vita-Finzi (1969) to about A.D. P/atanus and Jug/ans palynologically, but to 400- 1 800 (Bintliff, 1982). Pollen diagrams what extent are they also linked practically? that cover the period of A.D. 400- 1 800 do not Although we may call Jug/ans a culture show special characteristic changes in the plant/tree, it is more difficult to consider pollen curves of species that are linked with P/atanus as a weed . Today plane trees are of river valleys. The phenomenon and the doubtful origin. They are often considered as delimitation of the 'Younger fill' in time, are hybrids, and grow along many roadsides in seriously doubted by i.a. Wagstaff (198 1) and South European countries. They are planted Eisma (1978). i.a. because they can withstand air pollution The coul�se of the pollen curves in the pollen 278 S. BOTIEMA diagrams does not point to a climatic change 5.2. Vilis that stimulated the spread of walnut and plane Vilis tree. The reverse is to conclude a climatic Th er.e are obvious reasons fo r using as change from the appea rance of Juglans and an 111dlcator of human activity. Modern Plalanus. The combination of pollen types occu.rrence of wild vine (Vilis sylveslris) is and their specific significance, however, restncted to the northern half of Greece north of the line Larissa-Trikkala-Ioannina Logo­ strongly points in the direct ion of the ( influence of prehistoric man. In chapters 6 thetls 111 J. Renfrew, 1973). Van Zeist and and 7 we will discuss this matter. the present author fo und wild Vilis in rather de se shrub and small trees of Carpinus The appearance of Juglans and Plalanus is � . not an isolated occurrence for Greece, but ofTenlahs, Cercis, COlinus, Quercus cf. pubes­ is found in other countries too. For North­ cens, elc., 17 km southwest of Lake loan­ nina. owever, there is some uncertainty western Iran, in the area of Lake Urmia � (Rezayah), the same pollen picture can be concern111g Its occurrence in the wild. On the found at about the same time, although pol­ distribution maps stands of wild vine in len assemblages from both regions diffe r �he southern Vojvodina are absent, although It was considerably (Bottema, in prep.). In the frequently seen in riverine forests (Bottema & Ott away, 1982). It is probable mountainous part of Turkey Juglans and that wild vine occurred in many places in Plalanus do not behave in the same way . Greece where it seems to be absent now and palynologlcally. Juglans is present in low ' that it disappeared due to human inter­ percentages during the H olocene, whereas vention. Plalanus pollen is generally absent apart from Vilis a late appearance in coastal sites (van Zeist el The palynological presence of during the Holocene diffe rs from place to place. In al., 1975). Altitude may be the limiting fa ctor Macedonia this type is often met with in the in Anatolia. Plalanus/ Juglami phase, after about 3000 A place of natural occurrence of Plalanus is B.P. and scarcely before. The fi nds from the Ghab Vaney in Northwestern Syria before (Niklewski & van Zeist, 1970), where plane 3000 B.P. may originate from wild plants as well as from cultivated grape. tree is fo und continuously during the last . VU IS sylveslris grows at heights up to 400 m 50,000 years. abO\e sea-level and occasionally up to 800 m. Juglans pollen appea rs in sediment cores � CultIvated vine \-v ould have a more restricted from Italy and France in Roman time mainly. distribution (Suva lesai in J. Renfrew, 1973). A dIagram from Lagoa Comprida, Portuga l Greek Macedonia, including Kastoria and (van den Brink & Janssen, in press) shows the Khimaditis/ Vegoritis area, lies above 500 stray grains of Juglami and a continuous curve for Caslanea long before the fifth millennium m. ab�ve sea-level. The distribution map of B.P. An early Juglans grain in loannina 11 WIld v111.e by Logothetis (in J. Renfrew, 1973) contradIcts Suvalesai's note on the elevation. (fig. 19) is ascribed to long-distance transport. E peci lly in the higher part of Macedonia Pollen of Juglans and Plalanus does not � � WIld vme occurs. That would explain the s ow the same distribution and percentages presence of the early (pre-Neolithic) appear­ (f�Igs. 1 2-2 1 ). For Macedonia, values are very ance of Vilis pollen. The increase during the much the same. Cores from Khimad itis ' Juglans/ Plalanus phase was undoubted ly Edessa and Kastoria show the two tree polle caused by man. types at about 3200 B. P. � We have discussed the situation on higher In the Plain of Macedonia , at about sea­ elevatIons, about 500 m above sea-level, but level, both Planlanus and Juglans pollen is what about the behaviour of Vilis pollen ve ry scarce. This is remarkable as part of the below 500 m? In the valley west of Edessa at pollen transport could be effected . by rivers. about 400 m above sea-level Vilis was present The same is true for loannina 11 (Epirus). at least at about 9500 B. P. In the low lying Nevertheless, at the present time on the Plam of Macedonia, Vilis pollen is more eastern slopes of the Pindus Mountains along numer0US than in other Greek cores, sug­ the road from Kalambaka to loannina numerous plane trees can be seen bordering gestll1g that in the riverine forest in the delta vine was abundant. In the Macedonian the Acheloos river. lowlands the Vilis curve is the reverse of the Palynological investigations in Greece 279

Vitis curve in the highlands. Up to 10% Vitis indicate early fa rming in Greece. One could pollen is found in the lower part of the imagine that in the same way as Iversen (194 1) Giannitsa diagram but during the period of used Planlago lanceolala as an indicator for the Juglansl Platanus phase Vitis pollen the Landnam in Denmark, one or more pollen percentages are considerably lower. As the types could be indicative fo r the onset of sediments in Lake Volvi are not old enough Neolithic fa rming in Greece. we ca nnot say how' the vine behaved in that Of course, the situation in Greece diffe rs low part of Thrace. Turnerand Greig (1975) consi'derably from that in Denmark or do not mention Vitis for the Drama area. Northwestern Europe as a whole. Because of In the Plain of Macedonia a clear nega­ va riation in elevation, substratum, exposure tive correlation exists between the culture and climate, vegetation types developed which indicators and Vitis (fig. 13), whereas in the undoubtedly demonstrate different reactions, higher parts a definitely positive correlation when they are attacked by man and his exists. Also in somewhat lower areas like domestic animals or replaced by his crops, Edessa and La ke Volvi one ca n speak of than those in Northwestern Europe. These a positive correlation. How ca n this be supposed vegetational changes may be visible explained? When fa rmers entered the Plain of in the pollen assemblages. Do Greek vege­ Macedonia wild vine was very common. Quite. tat ions show ind icators of this process and probably it grew in riverine forests with if so, are they identical all over the area Fraxinus excelsior etc. This type of forest was studied or are they diffe rent? soon destroyed resulting in a decrease of Vilis In addition to the three new diagrams dealt (the decrease of pollen of Fraxinus excelsior with in this paper, other available information and Vilis lead to the conclusion that the has been treated in the same way. A selection riverine forest was dest royed). The assumed of ten diagrams from Northwestern Greece destruction of wild vine was not fo llowed by (Bottema, 1974) together covers the Holocene, domestic viticulture. so also pre-farming periods are present. The Wild vine was never very common at higher diagrams represent an altitudinal range from elevations as a stuitable biotope was relatively sea-level to 650 m above sea-level (figs. 12-21). scarce there compared with the Plain of The most important types discussed above are Macedonia. On such higher locations Vilis given schematically in fig. 22. pollen percentages increased after about La ke sediment as well as peat is present as 3000-3500 B.P. due to viticulture. basic material from which the samples were On the Epirotic side of the Pindus Vilis taken, although this variation is advantageous occurred in the loannina area for most of the only to u certain extent. As will be known, Holocene but as is evident from the pollen pollen curves derived from clay or gyttja tend percentages it was not abundant. Sediment to demonstrate smooth curves compared with from the coastal area of Akarnania that curves from peat sediment or comparable represents about the last 6000 years shows a deposits with a lot of orga nic detritus. Cores rather constant presence of Vilis . According from the centre of lakes are fa r less influenced to the distribution map by Logothetis this by loca l vegetation than cores from marshes would be cultivated grape. It is however or cores from locations close to (marshy) lake doubtful to what extent this map gives shores. indications of past distribution of Vilis Thus the diagram Khimaditis III (fig. 17) sylveslris or its domesticates. taken in the lake of that name, compares well The same can be said for the Thessalian with the three new cores under di�cussion. It is part. Vitis occurs in the Xinias area (Bottema, more difficult to compare between Khimaditis 1979) soo n after the beginning of the III and Khimaditis I (fig. 16) taken from the Holocene. It is clear that such fi nds must be nearby marsh. The remaining diagrams (figs. attributed to Vi/is sylveslris. 13, 14, 18 and 19) from Giannitsa, Edessa, Kastoria, Xinias anct loannina have been 11 11 POSSIBLE INDICATORS OF EARLY taken al so from loc ations where local 6. FARMING IN GREECE vegetation will often have been present very nea r to the coring site. In chapter 3.2. the method is described for The pollen types from the so-called group selecting a group of pollen types that may III '(chapter 3.2.) are selected from the ten 280 S, BOTTEMA

KHIMADITIS III

,-J J I Itnu ,! I io 30 ' .0 �'80 ' 90 ,i ! 10' . ,b . . ,� , r- , i : I- i 1 ,, 1 i : J I-

'" ::1 i -

:

'" -----, - : i :

'" -10,,,.'.2 i -! ! l Fig. 17. Selection of : ! pollen curves from i : the Lake Khimaditis ; III diagram ind icating , lO. H , =:f ,..H. r antropogenic '\" 'I' 1,o ,6,o 'P T activity.

diagrams and shown in figs. 12-21. In figs. that cover the complete Holocene present a 12-18 diagrams from Macedonia are shown. picture that, starting below, shows a dense They originate from Vegoritis, Khimaditis patterncause d by the presence of many types, (two), Edessa, Kastoria, Giannitsa and Volvi, This is followed by a less dense part (absence or Two diagrams originate from southwest of the low values of many types) and then again a Pindus, Trikhonis and loannina 11 (figs, 19 dense pattern when the types returnto higher and 20), The remaining diagram (fig, 21) is values. In those diagrams where the younger from Xinias (Thessaly), part of the Holocene is pre sent, it is self­ The figures are drawn in such a way that the evident that only part of that pattern can be A P / NAP ra tio is shown first, followed by found. the curves of P/atanus, Jug/ans, Vilis and The black-white-black pattern is possible, Castanea, Then follow the types present in beca use a series of types could be grouped group III (chapter 3.2,), at least when such together which show the same property. They types are present with a curve, Only in figs. 18 are all found during the early Holocene, after and 21 are the curves calculated from a basic which they disappear because they cannot sum including tree types only. tolerate the shade of forest and fi nally they The various pollen types are grouped return when man opens up the forest. according to the shape of their pollen curve. One of the results of the grouping is that Finally the curves of the Gramine\le and types that were thought to be of local origin, Pteridium are shown. At least part of the grass growing somewhere on the edge of the marsh, pollen doubtless has connections with man turned out to have an identical behaviour and his animals, to well-known weed and crop types. For Especially in the diagrams made for the lake insta nce, Po/ygonum avicu/are-type pollen sediments the grouping of the various curves was for a long time considered 'a s mostly local demonstrates a visual pattern, The diagrams (Bottema, 1974), but it turns out that it is Palynological investigations in Greece 281

KASTORIA

lO.

Fig. 18. ·Selection of pollen curves from the Lake Kastoria diagram indicating antropogenic activity.

closely linked with types indicating fa rming. and Chenopodiaceae. They are present during Pollen types that appear in the early as well late Pleistocene and early Holocene times. as in the late Holocene are: Artemisia, Then they become rare but they do not

Chenopodiaceae, Polygonum a viculare-type, disappea r altogether. In younger Holocene Rumex acetosa-type, Centaurea solstitialis­ times they expanded again. The fa ct that they type and Sanguisorba minor/ Poteriul11. All do not disappear makes them unsuitable as these types represent more than one plant indicators for the onset of Neolithic fa rming species. Only in the case of Sanguisorba in Greece. They may indicate more wide-scale minor/ Poterium the number of species is fa rhling, but they are no subtle ind icators. known with some certainty to be limited to A series of typica l culture indicators have two. relatively important values in the upper parts Some of these types can be listed as of the diagrams representing the late Holo­ predominantly steppic, fo r instance Artemisia cene period. This group includes Cerealia- 282 S. BOTTEMA

IOANNINA "

'00

lOO

30'

'" Fig. 19. Selection of pollen curves from the Lake Ioannina Il diagram indicating antropogenic activity.

type, Plantago lanceolata-type , other Plantago problem remains whether there were any types and Humulus/ Cannabis. If we trace Neolithic farmers in the mountains. There is these curves back to periods during which proof of early Neolithic habitation in the early farming may have taken place, the Plain of Macedonia where they inhabited the corresponding values have diminished to site of Nea Nikomedeia (Rodden, 1962; 1965). values of often under I %. Sometimes a The Giannitsa diagram (fig. 13) may give particular type is gone altogether. information on the Nea Nikomedeia farmers. Cerealia-type indicates cereal crops only This diagram covers about 8000 years accord­ partly as the pollen type includes some wild ing to the radiocarbon dates. It is thus impossi­

Gramineae, which are difficult or impossible ble to see whether Plantago lanceolata- type to separate from the Cerealia proper (van Zeist pollen was present before that time. In et al., 1975). In Greece, for instance, Hordeum Thessaly Plantago lanceolala-type pollen is spontaneum, Triticum aegilopoides and found in low percentages from early Holocene Secale montanum occur in the wild . Such wild times onward in the diagrams of Xinias I cereals and wild grasses as mentioned by van (Bottema, 1979) and Xinias II (fig. 21). Zeist et al. explain the fact that the curve In Northern Greece Plantago lanceolala representing Cerealia-type pollen runs back pollen may be an indication of early Neolithic into the Late Pleistocene. fa rming, but in Thessaly and more to the Plantago lanceolata-type (defined as a type south no such indicative value is apparent. because the reference collection for Greece The same can be said for Humulus/ Cannabis. of the Biologisch-Archaeologisch Instituut is This pollen type clearly indicates human not complete as regards Plantago species) activity. Humulus is, on the other hand, a generally does not occur in the lower part of natura'l constituent of the vegetation as is the diagrams. The plantain curves start about indicated by the presence of Humulus/ 8000 B.P. in the diagrams of Edessa, Cannabis throughout the first half of the Khimaditis I and III (figs. 14, 16 and 17). The Holocene, be it in low numbers. It is obvious Palynological investigations in Greece 283

� TRIKHONIS > � . � u . is � � � � �6

i 11, iU i !, i I, !,!, !, I, I, -"AI' _I J I ,b '0 jo <0 50 60 70 80 90 � 1 j I, : .-- '00 'ZEA I- i :i ii ! i i - 0 i i : i j ! ! 200 : i I j ! ! i i : 0 0 : : ! 300 : : , : : : : i : ! : i c- o : : '00 i : 0 i 0 ! : i : j : i I 0 , 1 � : : I : : : ! 0 0 soo : i : I i 0 i : ! I : i i : i: i j ! '\0 . "!' ,0 . '.0 5.0 '.a 'P, '. 0. 'po ! i Ii

Fig. 20. Selection of pollen curves from the Lake Trikhonis diagram indicating antropogenic activity.

that also from this gradually increasing type sion Delphinium-type and Centaurea cyanus­ no proof can be obtained of early Neolithic type undoubtedly indicate farming activity, farming. but they do not mark the beginning of such The third group consists of the remaining activities. types. The types in this group never show Urtica is found in low percentages, and with important values and their appearance is some very high values. The high values irregular. Representatives are Xanthium, Aspho­ suggest that Urtica sometimes fo rmed part of delus, Me rcurialis annua and Urtica. Xan­ the local vegetation. Rumex hydrolapathitm­ thium is restricted to the Thracian/ Mace­ type may be a local marsh plant that comes donian part. It does not show a closed curve into this group also because it did not occur in for any period but pollen is present in early the relevees of Barbero and Quezel. Holocene times already. The Gramineae are included in the figures, Asphodelus may ind icate open, bare because they may ind icate a trend especially vegetation, or over-grazing. This type is only wh�n the core has been taken from deeper present in cores from lower levels or in the water. Gramineae pollen from marshes and south (Ioannina, Trikhonis). An exception is upland species may be found together, the Vegoritis (fig. 15), where a few Asphodelus more local group causing the irregular pattern. grains are met with. Lake depos its show smoother curves than For the three lake diagrams under discus- organic deposits. 284 S. BOTIEMA

XINIAS "

"

'OD

IS66O.t.250BP Fig. 21. Selection of pollen curves from the Lake Xinia 11 diagram indicating antropogenic activity.

Thus, the method of using the pollen types mentioned in group III fails in attempting to trace the onset of fa rming in Greece. It is, however, a useful method for studying the intensity of farming and the occurrence of various fa rming phases.

7. PALYNOLO GlCAL EVIDENCE FOR : DEMOGRAPHIC EVENTS IN (PRE)­ ,� HISTORIC GREECE 5, , : 6'. ! In this chapter it will be seen whether there 7� � i are connections between demographical fl uc­ -� :: e ?: tuations, especially invasions, and palyno­ : ., I logical information. Such con nections were already made by Athanasiadis (1975) when 1 0 -: he discussed pollen diagrams for Litochoro lJ: and Pertouli in Thessaly. It is thought that ill.�. ��Lilh lli: � Fig. 22. Schematic pollen diagram s owing a selection of in Greek (pre)history many events occurred pollen curves indicating human influence. involving displacement or invasion of smaller or larger groups of people. Some of these movements, especially in prehistoric times, Finally one spore is added, Pteridium. The are merely hypothetical or await more con­ Pteridium curve closely resembles the Plan­ vincing proof. Others that are more recent tago lanceolata type curve, the first finds have been traced in written records from occurring before- the onset of human farm ing that time. and herding activity. Alt hough bracken and The impact of such events upon the plantain must have been constituents of a environment, or more specially upon the natural vegetation, they seem to be sensitive vegetation, is very difficult to predict or to indicators for an early stage of farming, make visible. We will see what information although not decisive for the recognition of can be gathered from the archaeological the first Neolithic fa rming activities. record and what can be learned from written Palynological investigations in Greece 285

,all.nl.tu t.1end., ,.�" hit' up.. u"lIi" .h' C14'''n BP

o ------. ------l EXCHANGE OF 6DO.OOO TURKS AGAINST 1.)00.000 GREEKS � TURKISH OCCUPATION

f IHYASlOH. " VI At" 1000 - - . -- - I BYZANTINE EMPIRf \ INVASlOH. " SlAYS f

ROMAH IHFlUE.CE GO I I VA lOH 2000 ------I : TH C N S . MACEOONIANS I CITY STATES

DO RI AN INVASIONS 3000 ------}

4000 ------2500 INVASIONS OF EURASIAN PASTO- (RAUST$ IACC. TO GIMBUTASJ)

- . •• ••••- .. • •••••••• . •• .• . 5000 ------. ------.-

NEOllTHIC .•• ----•• -- 6000 -- . - -.------.------. ------. ------• HUMAN INflUENCE UPON THE VEGHATIOH

Fig. 23. Chronological table showing historical events and a scheme of human impact upon the vegetation derived fr�m palynological evidence.

information from later periods. When read­ the author is much indebted to Mr. A.E. ing the general outline of Greek history in . Lanting. For comparison with the palyno­ the Geographical Handbook Series-Greece logical record it is necessary that the pre­ (1944), it turns out that quiet periods hardly historic cultures are fi rmly delimited in occurred (see a general scheme taken from the time and space. In West and Central Greek Geographical Handbook Series-Greece in fig. Macedonia only very fe w modern excavations 23). It should be stressed that the quantitative have taken place. The general distribution of assessments of movements or migrations are the various cultures and their dating are based not always equally reliable. The number of upon surveys, from the study of surface finds. people involved is (has to be) based often The cultural succession in Thessaly is well upon ge neral assumptions; especially the known but absolute dates of the Dimini, invasions in prehistoric times need further Larisa and Rachmani periods are few proof. (Hauptmann, 1981). In East Macedonia, east Apart from the number of people involved, of the Strymon, a better situation is found the impact upon the environment remains bec'ause of the excavations of Sitagroi highly questionable. Good studies of demo­ (Renfrew, 1970; 1971; 1973) and Dikilitash graphic history of the Greek interior on the (Theocharis & Deshayes, 1962; Deshayes, basis of excavations are lacking. For the 1968; 1970a; 1970b; 1971; 1973) on the Plain following information on the prehistoric part, of Drama, A series of radiocarbon dates from 286 S. BOTTEMA

Sitagroi points to a hiatus between about 3400 factors may have been of influence upon the and 2600 B.C. Accord ing to Turner & Greig environment. For instance when the Roman (1975) in their study on the vegetation history empire collapsed some economic effects in the Drama area, a slight decrease in the would have been ahead of this event and some amount of forest can be detected, indicating may have been delayed. Some economic that, although no habitation was present on events can be translated into damage to Sitagroi, prehistoric man was still present in vegetation and others cannot. the vicinity. In the Byzantine Empire invasions of Slavs Gimbutas (1974; 1977) thinks that an took place from about A.D. 500 to 675. invasion of Eurasian steppe pastoralists Historical reports mention numbers of about halted in Rumania around 3500 B.C. They 100,000. were driven southward by another invasion About A.D. 1000-1100, invasions of Vlachs around 2700 B.C. to Macedonia. took place. They were less important than the I n the period from c. 2300 to c. 1900 B. C. Slavs. Still, as they took to the mountains, covering part of the Early Bronze Age, many they could have exerted pressure upon the settlements were destroyed. An invasion, mountain forests. Accord ing to the sensus of according to Gimbutas, came from the East 1928 c. 20,000 Vlach-speaking people lived in European steppes. Hanschmann and Miloj cic Greece. It is likely that the actual number was (1976) think that the invasion came from higher. Many Vlachs had already left for Macedonia. Best (1973) postulates Thracian Rumania after the First World War. Still the tribes coming from the north. However, he number of invading Vlachs has to be counted confounds Greek calendar year dates with in tens of thousands and not in hundreds of Bulgarian uncalibrated radiocarbon dates. thousands. The fall of the Byzantine Empire Similarly, confounding of radiocarbon years in 1453 when the Ottoman Turks took over is and calendar years is also evident in also not an important limit for economic Hammond (1978). environmental events. In 1204 for instance a Many investigators ascribe an invasion united fleet of Venetians and Crusaders around 2300 B.C. to the ancestors of the conquered Greece and adjacent areas. The Greeks who spoke an Indo-European lan­ French Boniface de Montferrat became king guage from which the Greek dialects devel­ of Macedonia, Thessaly and much of Central oped in Greece. Others think that the Proto­ Greece. So in fact the area was free of Greeks invaded around 1600- 1550 B.C. Proof Byzantine domination already almost two of the presence of Greek-speaking people hundred years before the definitive fall of the dates from 1200-1 100 B.C. empire. This is of importance as for instance Astour (1965) suggests that Semitic tribes Athanasiadis (1975) links the pollen zones came from East Cilicia to Greece immediately with the main political events. after the Santorini eruption. Palestinian/ early Apa rt from the supposed impact of man Phoenician names would have been abundant upon his natural surroundings, some palyno­ locally at that time in Greece. An invasion of logical information may be derived from the Dorians is postulated c. 1200/ 1100 B.C. the method of fa rming. Athanasiadis points coming from Northwestern Greece. In view of out the difference between Turkish agri­ the many cities destroyed this must have been culture and Greek fa rming. Migration, dur­ a turbulent time. ing the Turkish occupation occurs in the The cultural gap between the Mycenaean form of Christian refugees escaping to the period and the period after it seems to be mountains. The Turkish occupation did not much smaller than was thought for a long end at the same time all over Greece; in Greek time . Proof that the people who burned the Macedonia and part of Thrace it occurred as settlements came from Northwestern Greece late as in 1913. is not very convincing. This made some The final demographic shifts include the investigators (i.a. Thomas, 1978) think that return of the refugees from the mountains and the 'Dorians' were the lower classes in the exchange in 1923 of I. 3 million Gr'eeks Mycenaean Greece who revolted against the from Turkey with 390,000 Turks/ moslems ruling class and chased them away. from Greece. The political and economic effects were not From the foregoing it is clear that many necessarily the same. Especially economic proven or supposed demographic shifts have Palynological investigations in Greece 287 occurred In Greece. There is a lot of around 3100-3200 B.P. (fig. 23), especially disagreement about these shifts especially west of the Plain of Macedonia, but not over those concerning invasions, apart from those the watershed in Lake Kastoria. In the Plain that have happened since Roman times. of Macedonia the pollen record does not show In fig. 23 a chronological table is given any sudden increase of human impact. The demonstrating a series of (pre)historic events sudden disappearance of coniferous forest and a tentative scheme indicating the impact (chapter 2.3.) is contemporaneous with pos­ of man upon the vegetation derived from tulated invasions of the Dorians. If we pollen diagrams. These diagrams originate assume that Dorian tribes moved in from mainly from Greek Macedonia and one from elsewhere, then they took a rather difficult the Epirotic side of the Pindus Mountains. road . If they came from the north they The scheme for Lake Trikhonis is not given apparently must have avoided the Plain of here, because there are too many dating Macedonia, at least the pollen record shows problems with this diagram. The vegetation no reaction here. A reason for avoiding the schemes differ from site to site considerably Plain of Macedonia may have been the and in this respect they unfort unately match inaccessibility of extensive marshy areas in the the archaeological record very well. centre. The edges, however, would have When vegetation history is concerned it is offered reasonable opportunities for travel­ appealing to look at some historical invasions, ling. Did they meet opposition there? because they were at least recorded. Of special The effe cts upon the vegetation of the importance is also the object of such Khi mad itisj Vegoritis area must have been movements. People who invaded fertile ca used by a relatively large number of people, lowlands and towns will have a different kind mainly herdsmen with sheep and goats, partly of impact upon the vegetation than herdsman settlers who took to the valleys. The palyno­ with large flocks spreading over the logical information supports the idea of a mountains. A disadvantage in studying the Dorian invasion. effect of subrecent migration may be the Although this is not a subject treated in advanced state of deterioration of Greek this study, it is worth mentioning that in forests. As the human influence has already South\ovestern Turkey a comparable event been exerted for millennia the invasions of takes place at the same time (Bottema & Goths, Slavs or Vlachs hardly add to the Woldring, in prep.). disturbance of nature. The other fluctuations in human impact In the diagrams of Volvi and Vegoritis such concluded from the pollen record cannot be invasions are not visible from the course of the brought into line with each other or with pollen curves. One may conclude that these information on migration that we know to invasions did not exert pressure upon the have happened. For the post-Roman period Greek vegetation to any marked extent. the rather detailed palynologica l information As we do not learn much by looking at the from the lakes of Vegoritis and Volvi agrees to most recent invasions, we turn to the lowest some extent with Athanasiadis' (1975) scheme part of the scheme of fig. 23. The first fo r the North Thessalian diagrams of indication of events that are synchronous to Litochoro and Pertouli. some extent is fo und at about 4000 B. P. in The general pattern proposed by Athan­ Kastoria and the Khimaditis area, thus in asiadis can be found in the two Macedoni­ Northwestern Greek Macedonia. In Vegoritis, an cores mentioned above. No signs·are found Giannitsa and loannina comparable changes of any impact made by invading Slavs and are fo und a little earlier. In Edessa the Vlachs. surroundings seem to be less affected. To resume, one may state that invasions are In calendar years this time would be about not necessarily reflected in the pollen record . 2500 B.C. Accord ing to Gimbutas (1974), It is of importance if human pressure is during that time (2700 B.c.) Eurasian pasto­ exe.rted upon an untouched natural forest or a ralists invaded Northern Greece. It cannot long-since degraded shiblyak. Post-Roman be excluded that the invasion postulated invasions in Northern Greece are not by Gimbutas is connected with the vegetation detectable in the pollen record . Increase in ' change mentioned above. anthropogenic pressure upon the vegetation in The next important change takes place NorthernGr eece around 4500 and 3100 B.P. may 288 S. BOTTEMA

be re lated to successive invasions of Eurasian BINTLlFF, J.L., 1982. Palaeoclimatic modelling of pastoralists (Gimbutas, 1974) and Dorian environmental changes in the East Mediterranean tri bes. region since the Last Glaciation. In: J.L. Bintliff, & W. va n Zeist (eds.), Palaeoclimates, palaeo­ environments and human communities in the Eastern 8. ACKNOWLEDGEMENTS Mediterranean region in la ter prehistory (B.A. R. International Series 133). Oxford , pp. 485-527. BOTTEMA, S., 1974. Late Quaternary vegetation history I am very thankful to my colleagues W. van of NorthlVestern Greece. Thesis, Groningen. Zeist and H. Woldring for stimulating discus­ BOHEMA, S .. 1979. Pollen analytical investigations in sions and practical help, to A.E. Lanting for Thessa ly (Greece). Palaeohistoria 21. pp. 20-40. his information on prehistoric events, to H. BOHE MA, S., 1980. On the history of the walnut (Juglans Roelink and J. Klein for their fine drawings, regia L.) in southeastern Europe. Acta Botanica Neerlandica 29, pp. 343-349. to G. Entjes-Nieborg for typing the manuscript, BOHEMA, S. & B.S. OHAIVAY, 1982. Botanical, to W. Haaima for the photography ' and malacological and archaeological zonation of to S. van Gelder-Ottway for the correction of settlement deposits at Gomolava. Journal of the English. Archaeological Science 9, pp. 22 1-246. BRINK, L.M. VAN DEN & CR. JANSSEN, in press. Holocene vegetation development in the Serra da 9. SUMMARY Estrela, Portu gal. CREER, K.M., P.W. READMAN & S. PAPAMARINO­ New palynological information from Greece is POULOS, 1981. Geomagnetic secular variations in Greece through the last 6000 years obtained from presented . The information of Holocene lake lake sediment stud ies. Geophysical Jo urnal of the sediments is used to compare the repre­ Royal Astron. Society 66, pp. 193-2 1 9. (With an sentation of pollen types with those from appendix by S. Bottema, Palynological dating of modern surface samples and with the pollen some young Holocene sed iments from three Greek types derived from the herb and tree species lakes, pp. 217-2 19). listed in plant-sociological relevees from (semi-) DESHAYES, J., 1968. Dikili Tash. Bulletin de Correspondance Helleniqlle. pp. 1062- 1077. natural forest. Differences in the qualitative DESHAYES, J., 1970a. Dikili Tash. Bulletin de composition of pollen assemblages from these Correspondance Hellenique, pp. 799-808. sources are used to trace types that indicate DESHAYES, J., 1970b. Les fouilles de Dikili Tash et human activity and influence on the vege­ I'a rcheologie Y ougosla ve. Zbornik Narodnog tation in prehistoric and historical times. Muzeja u Beogradu 6, pp. 21-43. The concept of diversity in vegetation is DESHAYES, J., 1971. La sequence des cultures a Dikili compared with that in pollen assemblages. Tash. Actes du Vllle Congres International des Sciences Preh istoriques et Protohistoriques. Some pollen types from trees, herbs or weeds Beograd 1971. Beograd, pp. 492-496. connected with the activity of man are dealt DESHAYES, J. 1973. Dikili Tash. Bulletin de with in particular. It is investigated in how far Co rrespondence Hellenique, pp. 464-473. such pollen types could mark the first fa rming EISMA, D., 1978. Stream deposition and erosion by the activities in Greece. Attention is paid to the eastern shore of the Aegean. In: W.C Brice (ed.), various demographic events that are thought The environmental history of the Near and Middle to have happened in Greece and their East since Last Ice Age. London, pp. 67-8 1. FILl POVITCH, L., 1977. Palynological data for the (possible) impact upon the environment. Postglacial distribution of Juglans in the compo­ sition of Bulga rian nora. Phytology (Bulga rian Academy of Sciences) 6, pp. 32-37. 10. REFERENCES GIM BUTAS, M., 1974. The destruction of Aegean and East

Mediterranean urban civilisation around 2300 RC. ASTOUR, M.C, 1965. Hellenosemitica. Leiden. In: Bronze Age Migrations in the Aegean ATHANASIADIS, N., 1975. Zur postglazialen Vegeta­ (Proceedings of the First International Colloquium). tionsentwicklung von Litochoro Katerinis und Per­ London, pp. 129- 139. touli Trika lon (Griechenla nd). Flora 164, pp. 99- GIMBUTAS, M., 1977. The first wave of Eurasian steppe 132. pastoralists into Copper Age Europe. The Journalof BARBERO, M. & P. Q UEZEL, 1976. Les groupements Indo-European Studies 5, pp. 277-338. forestiers de Grece Centro-Meridionale. Ecologia G REECF., vol. I. Physical geography, history, adminis­ Mediterranea 2, pp. 3-86. t rat ion and peoples (= Geographical handbook

BEST, J.G.P., 1973. An Outline. In: Th e arrival of the ser. B . R . 516). [London] Nava l Intelligence Greeks (Publication of the Henri Frankfort Division, 1977. Founda tion, Vol. I). Amsterdam, pp. 12-53. GREIG, J.R.A. &J. TU RNER, 1974. Some pollen diagrams