landscape and land use in postglacial greece

Landscape and Land Usein Postglacial Greece

Sheffield Studies in Aegean Archaeology, 3

SHEFFIELD STUDIES IN AEGEAN ARCHAEOLOGY

ADVISORY EDITORIAL PANEL

Dr Stelios ANDREOU, University of Thessaloniki, GreeceProfessor Susan ALCOCK, University of Michigan, USADr John BENNET, University of Oxford, EnglandProfessor Keith BRANIGAN, University of Sheffield, EnglandDr William CAVANAGH, University of Nottingham, EnglandProfessor Jack DAVIS, University of Cincinnati, USADr Peter DAY, University of Sheffield, EnglandDr Charles FREDERICK, University of Sheffield, EnglandDr Paul HALSTEAD, University of Sheffield, England

SHEFFIELD STUDIES INAEGEAN ARCHAEOLOGY

Landscape and Land Usein

Postglacial GreeceEdited by

Paul Halstead and Charles Frederick

(SheffieldAcademic Press

www.Sheff ie ldAcademicPress.com

www.SheffieldAcademicPress.com

Copyright © 2000 Sheffield Academic Press

Published by Sheffield Academic Press LtdMansion House

19 Kingfield RoadSheffield Sll 9AS

England


Printed on acid-free paper in Great Britainby Bell and Bain, Glasgow

British Library Publication Data

A catalogue record for this book is availablefrom the British Library

ISBN 1 84127 184 5


Contents

Abbreviations

Preface Charles Frederick and Paul Halstead

List of Contributors

1. ATHANASIA KRAHTOPOULOUHolocene Alluvial History of Northern Pieria, Macedonia, Greece

2. ACHILLES GERASIMIDISPalynological Evidence for Human Influence on the Vegetationof Mountain Regions in Northern Greece: The Case of Lailias, Serres

3. MARIA NTINOU AND ERNESTINA BADALLocal Vegetation and Charcoal Analysis: An Example from Two LateNeolithic Sites in Northern Greece

4. JENNIFER MOODYHolocene Climate Change in Crete: An Archaeologist's View

5. MARGARET ATHERDENHuman Impact on the Vegetation of Southern Greece and Problems ofPalynological Interpretation: A Case Study from Crete

6. CHARLES D. FREDERICK AND ATHANASIA KRAHTOPOULOUDeconstructing Agricultural Terraces: Examining the Influence ofConstruction Method on Stratigraphy, Dating and Archaeological Visibility

7. HAMISH FORBESLandscape Exploitation via Pastoralism: Examining the 'Landscape Degradation'versus Sustainable Economy Debate in the Post-Mediaeval Southern Argolid

8. PAUL HALSTEADLand Use in Postglacial Greece: Cultural Causes and Environmental Effects

9. AMY BOGAARD, MICHAEL CHARLES, PAUL HALSTEAD AND GLYNIS JONESThe Scale and Intensity of Cultivation: Evidence from Weed Ecology

10. TODDWHITELAW

Settlement Instability and Landscape Degradation in the Southern Aegeanin the Third Millennium BC

11. CHRISTOPHER MEE AND PETER JAMESSoils and Site Function: The Laconia Rural Sites Project

7

9

13

15

28

38

52

62

79

95

110

129

135

162

This page intentionally left blank

Abbreviations

Athens Annals of Archaeology

Arkhaiologiko Deltion

Arkhaiologiki Efimeris

To Arkhaiologiko Ergo sti Makedonia kai Thraki

American Journal of Archaeology

Mitteilungen des Deutschen Archaologischen Instituts, Athenische Abteilung

Archaeological Reports

Annuario della Scuola Archeologica di Atene

Beitrage zur Ur- und Fruhgeschichtlichen Archaologie des Mittelmeer-

Kulturraumes

British Archaeological Reports

Bulletin de Correspondance Hellenicjue

Bulletin of the Institute of Classical Studies of the University of London

Annual of the British School at Athens

Cambridge Archaeological Journal

Corpus der minoischen und mykenischen Siegeln

Classical Philology

Classical Quarterly

To Ergon tis Arkhaiologikis Etairias

Greece and Rome

Journal of Archaeological Science

Journal of Hellenic Studies

Journal of Mediterranean Archaeology

Kritika Chronika

Oxford Journal of Archaeology

Praktika tis en Athinais Arkhaiologikis Etairias

Proceedings of the British Academy

Proceedings of the Cambridge Philological Society

Studies in Mediterranean Archaeology

Studi Micenei ed Egeo-Anatolici

AAA

ADelt

AE

AEMTh

AJA

AM

AR

ASA

BAM

BAR

BCH

BICS

BSA

CAJ

CMS

CP

CQErgon

G&R

JAS

JHS

}MA

Kr Chron

OJA

PAE

PBA

PCPS

SIMA

SMEA

PrefaceCharles Frederick and Paul Halstead

The Round Table on Aegean archaeology heldin Sheffield on 22-24 January 1999 was devot-ed to the subject of 'Postglacial EnvironmentalChange in Greece'. As investigations of envi-ronmental change in postglacial Greece haveproliferated over the last three decades, thecompeting claims of climate change andhuman land use, as agents of transformationof the landscape, have been vigorously cham-pioned by different groups of scholars. Palaeo-ecologists have sought to understand how andwhy the landscape of Greece has changed overthe last ten thousand years. For archaeologists,the reconstruction of past landscapes is criticalto an understanding of the problems andopportunities encountered by ancient humansocieties. Archaeologists have also been tanta-lized by the prospect that the palaeoecologicalrecord may bear the imprint of human activityand so reveal something of the nature andscale of past land use.

In recent years, however, there has beengrowing awareness of the difficulty of disen-tangling natural and anthropogenic agents oflandscape change. In part, this results from thecoarse chronological resolution of both thepalaeoecological and archaeological records,issues stressed by a number of contributors tothis volume. In part, it reflects the need for bet-ter understanding of the relationships in thepresent between landscape and land use andbetween land use and human settlement. Toexplore some of these issues, the meetingbrought together palaeoecologists and archae-ologists with an interest in land use andhuman settlement.

This volume presents revised versions ofpapers from the meeting. Participants were

invited to attempt both a critical review ofwork to date and a consideration of avenuesfor future work and both of these objectivesare reflected in this volume. For similar rea-sons, the papers which follow include bothcase studies of particular regions and contri-butions with a broader geographical focus.Several contributors emphasize the role ofstudies of the present and recent past as thebasis for reconstruction and understanding ofthe more distant past. The papers are orderedalong the continuum landscape-land use-set-tlement, although the breadth of many contri-butions ensures a degree of arbitrariness here.No determination or dominance is covertlyimplied by this ordering and the volume canbe read with equal plausibility from front toback or from back to front.

Nancy Krahtopoulou presents a geoarchae-ological study of landscape change in coastalPieria, in northern Greece. Significantly, des-pite the relatively fine chronological resolutionof her study, she encounters difficulties in dif-ferentiating between natural and anthro-pogenic causation on the basis of temporalpatterning in alluvial deposits. AchillesGerasimidis discusses palynological evidencefor human impact on the montane forests ofnorthern Greece. His argument that anthro-pogenic impact is clearest over the last twomillennia is consistent with Krahtopoulou'sobservation that the frequency of alluvialepisodes increases significantly over the sametimespan. Maria Ntinou and Tina Badal com-pare the off-site palynological record fromnorthern Greece with the on-site charcoalrecord from two Late Neolithic settlements inthe same region. The charcoal evidence offers

10 Charles Frederick and Paul Halstead

support for the argument of palynologists thatearly farmers in northern Greece had limitedimpact on regional vegetation.

A second group of papers focuses on south-ern Greece. Jenny Moody documents geoar-chaeological evidence for widespread floodevents during the mediaeval to early modernperiod in Crete, which she relates to the 'LittleIce Age' climatic phenomenon. An earlierLittle Ice Age, during the Middle to LateBronze Age, may have had a dramatic effect onland use and social change at the time of theMinoan 'palaces7. Margaret Atherden exam-ines a pollen core, spanning the last millenni-um, from the White Mountains in Crete. Usingobservations on the present-day ecology ofMediterranean vegetation, she thoughtfullytackles the appropriately thorny issue of rec-ognizing the effects of grazing/browsing andburning in the palynological record. CharlesFrederick and Nancy Krahtopoulou explorethe relationship between terracing and thegeoarchaeological record. As well as offering aresearch design for exploring the (pre)historyof terracing in the Aegean, they expose thecomplex ways in which surface archaeologicaltraces might be concealed from, or revealed to,intensive survey projects by different forms ofterrace construction.

Hamish Forbes addresses the contentiousissue of livestock and overgrazing, taking avariety of documentary sources to argue thatstocking densities have remained remarkablystable in the southern Argolid over the lastthree centuries, despite dramatic changes inhuman demography and economy. Recentstocking densities have been constrained by avariety of informal, density-dependent mecha-nisms. Whether similar mechanisms were inplay in the distant past is an open question,with profound implications not only foranthropogenic impact on the landscape, butalso for the nature and sustainability of earlyanimal husbandry. Paul Halstead explores the

likely impact on landscape of intensive andextensive forms of land use, and revisits earli-er work on prehistoric land use in the light ofpalaeoecological hints of mid-Holocene desta-bilization of the landscape. Amy Bogaard,Mike Charles, Paul Halstead and Glynis Jonesreport on studies of modern weed ecology inGreece and other parts of the Mediterranean.These studies are developing a methodologyfor distinguishing different forms of intensiveand extensive crop husbandry regime on thebasis of archaeological assemblages of weedseeds.

Todd Whitelaw's paper critically reviewsthe evidence for changing patterns of settle-ment in the fourth and third millennia BC in theAegean, arguing that apparently similar trajec-tories of dispersal and nucleation conceal sig-nificant regional and local divergences. Giventhe tendency for some palaeoecologists tosqueeze poorly dated local records into asupra-regional chronological straitjacket,attempts to understand landscape degradationin terms of correlation with changing patternsof settlement have been undermined by bothgroups of practitioners. With closer collabora-tion between palaeoecologists and archaeolo-gists, however, and with finer chronologicalresolution in both data sets, he is optimisticthat a more reflexive and subtle understandingcan be achieved of the relationship betweenhuman marginal colonization and landscapedegradation. In the final chapter, the methodsof geoarchaeology are applied to small ruralsites discovered by surface survey in Laconia.Such sites have been found in large numbersby intensive survey projects and an understan-ding of their permanence, longevity and func-tion is critical to attempts to use surface data asevidence for regional patterns of demography,land use and social organization. To this end,Chris Mee and Peter James show how the on-site analysis of soils complements the tradi-tional evidence of surface artefact scatters.

Preface 11

Bill Cavanagh, Adrian Lane, Mark Macklin,Oliver Rackham, Kathy Willis, Jamie Wood-ward and Eberhard Zangger, presented papersat the Round Table that they could not, for var-ious reasons, contribute to this volume. We arealso grateful to Paul Buckland and KevinEdwards, who chaired sessions at the RoundTable, and to all other participants, whohelped to make the meeting a success. Thanksare due to the Institute for Aegean Prehistoryand the London Hellenic Society, for financialhelp towards the costs of holding the Round

Table and of publishing this volume, respec-tively. 'Domestic7 arrangements during themeeting were managed by Mike Lane, EleniNodarou, and Giorgos Vavouranakis, whilethe task of feeding participants was heroicallyundertaken by Ellie Hitsiou, Valasia Isaaki-dou, Vangelio Kiriatzi and Sevi Trianta-phyllou, with specialist assistance from NancyKrahtopoulou. Mike Charles, Rob Craigie andRocky Hyacinth kindly retrieved parts of thecontributions which follow from the ether.

CONTRIBUTORS

MARGARET ATHERDEN

ERNESTINABADAL

AMY BOGAARD

MICHAEL CHARLES

CHARLES FREDERICK

HAMISH FORBES

ACHILLES GERASIMIDIS

PAULHALSTEAD

PETER JAMES

GLYNIS JONES

University College of Ripon and York St John, Lord Mayor'sWalk, York Y037EX, UK.

Departamento de Prehistoria y Arqueologia, Universitat deValencia, Avenida Blasco Ibanez 28, 46010 Valencia, Spain.

Department of Archaeology and Prehistory, University ofSheffield, Northgate House, West St., Sheffield SI 4ET, UK.



Department of Archaeology, University of Nottingham,University Park, Nottingham, NG7 2RD, UK.

Department of Forestry and Natural Environment, AristotleUniversity of Thessaloniki, PO Box 270, 54006 Thessaloniki,Greece.


Department of Geography, University of Liverpool,Liverpool L69 3BX, UK.


ATHANASIA KRAHTOPOULOU Department of Archaeology and Prehistory, University ofSheffield, Northgate House, West St., Sheffield SI 4ET, UK.

CHRISTOPHER MEE

JENNIFER MOODY

MARIA NTINOU

TODD WHITELAW

SACOS, University of Liverpool, 14 Abercromby Square,Liverpool L69 3BX, UK.

Anthropology, Archaeology and Environmental Studies,Baylor University, Waco, Texas, USA.

Departamento de Prehistoria y Arqueologia, Universitat deValencia, Avenida Blasco Ibanez 28, 46010 Valencia, Spain.

Institute of Archaeology, University College London, 31-34Gordon Square, London WC1H OPY, UK.

Holocene Alluvial History of Northern Pieria,Macedonia, Greece

Aihanasia Krahtopoulou

Introduction

During the last two decades, archaeologicalstudies in Northern Pieria (Fig. 1.1) have indi-cated that the landscape has changed dramat-ically over time and that geomorphic agentsof soil erosion and valley alluviation haveseriously affected the preservation and visibil-ity of the archaeological record in the area (foran overview of the history of archaeologicalresearch in Northern Pieria, see Besios andPappa 1995). A detailed geoarchaeologicalstudy was undertaken, therefore, in order toinvestigate the co-evolution of landscape andhuman settlement in the area and to assess thepotential of the surface and sub-surfacearchaeological records (Krahtopoulou 2001).A recent critique of geoarchaeological researchin Greece has exposed the need for fine tem-poral and spatial resolution as the basis bothfor reconstructing landscape change and forevaluating alternative agents of change(Endfield 1997). For this reason, the presentcase study focuses on the intensive investiga-tion of a relatively small area.

Research Design and Methodology

Two small-order streams, the Gerakaris andthe Agios Dimitrios, which originate in the

Pierian hills and discharge into the ThermaicGulf, were selected for detailed investigationof late Quaternary alluvial history (Fig. 1.1).The streams drain an area of 60 km2 and 18km2, respectively, and are underlain by highlyerodible, unconsolidated Upper Miocene-Lower Pliocene sands, sandy loams, loamysands and clays (IGME 1986). Surface findsfrom non-systematic, extensive archaeologicalfield reconnaissance and preliminary reportsfrom small-scale rescue excavations indicate ahuman presence in the Gerakaris and AgiosDimitrios catchments from, at least, the LateNeolithic onwards (5400/5300-4700/4500 BC)(Besios 1993; Besios and Pappa 1995; Gram-menos et al 1997; Pappa 1999).

Field observations in the lower reaches ofthese streams, immediately above the presentcoastal plain, revealed a series of alluvialdeposits and intercalated palaeosols, contain-ing in many cases in situ preserved archaeo-logical sites and structures. Selection ofresearch localities was determined by accessto property and/or presence of suitable natur-al and artificial exposures. In the AgiosDimitrios valley, lack of adequate exposureswas compensated by mechanical excavation.In all, 17 vertical sections were cut back,cleaned, examined and described, while 220soil and sediment samples were collected andanalysed for mean particle size, loss-on-igni-

1

Figure 1.1 Location of the study area.

Holocene Alluvial History of Northern Pieria, Macedonia, Greece 17

tion, calcium carbonate content and magneticsusceptibility.

Temporal control for the Gerakaris andAgios Dimitrios sequences is provided by C14

(AMS) dating of 17 charcoal and soil samples,collected from buried A soil horizons. In situburied archaeological sites and structuresassociated with specific palaeosols offer addi-tional chronological information and in manycases refinement of the chronology estab-lished by C14 dating.

The Gerakaris Alluvial Sequence

Field observations, as well as absolute (seeTable 1.1) and relative dating of the depositsidentified in the Gerakaris valley, resulted inidentification of four late Pleistocene andeight Holocene alluvial units, separated bypalaeosols. Two informal alloformations havebeen recognized. The late Pleistocene alluvialfills are named the Nea Trapezous alloformationand the Holocene deposits the Kato AgiosGiannis alloformation, which are abbreviated asNT and KAG respectively. This paper focusesprimarily on the Holocene alluvial record.

Four geomorphic surfaces (TO, Tl, T2 andT3) were recognized in the lower reach of theGerakaris (Fig. 1.2). The usually narrow mod-ern floodplain (TO) lies at an elevation of 2 mabove the stream bed. It is inset into the firstterrace (Tl), a relatively narrow, gently slop-ing surface, which rises to an elevation of4.50-5.50 m above the channel. The secondterrace (T2) is a wide, flat surface, which dom-inates the Gerakaris valley and lies at an ele-vation of 9 m above the active channel. Thehighest terrace (T3), which has been identifiedonly near the Gerakaris valley walls, is slop-ing and lies 16-19 m above the present riverbed. As the stream enters the alluvial plain ofKaterini and flows eastwards towards thecoast, these geomorphic surfaces 'disappear'

under the flat, featureless surface of the allu-vial plain.

The Nea Trapezous Alloformation

The four units belonging to the Nea Trapezousalloformation are designated NT1 to NT4 inorder of decreasing age. Generally, these arerarely exposed in the lower reaches of thestream and the sequence remains poorlydated. Two C14 dates, archaeological inclu-sions, stratigraphic position and degree ofpedogenic development place deposition ofthe Nea Trapezous alloformation in the latePleistocene between 38.8 ka bp and, perhaps,10.0 ka bp (Krahtopoulou 2001).

The Kato Agios Giannis Alloformation

Some time after the deposition of unit NT4,but before 8.5 ka bp, the Gerakaris incised andcut a new valley, which was subsequentlyfilled in with 8.5 m of alluvial deposits ofHolocene age. The Holocene sequence (theKato Agios Giannis alloformation) is well datedby absolute (C14) (Table 1.1) and relative (insitu buried archaeology and historical infor-mation) means and eight discrete alluvialunits are presently recognized: KAG1 toKAG8 from oldest to youngest. The lowerboundary of the oldest alluvial fill presentlyrecognized in the Gerakaris valley (KAG1) isnowhere exposed and it is possible that more,as yet undetected, early Holocene alluvialunits are buried in the valley axis.

Deposition of KAG1 predates 8,470±75 bp(7600-7351 BC) and seems to be confined with-in the newly cut valley floor of the Gerakaris.After a period of landscape stability andpedogenesis, the second fill (KAG2) wasdeposited across the Holocene valley, sometime before 5,360±60 bp (4339-3999 BC; Final

Table 1.1 List of radiocarbon dates from the Gerakaris stream.

Lab. no.

AA31853AA31852AA32856AA32859AA32855AA31861AA32858AA32854AA32857AA34066AA31860AA31859

Depth(cm)

13525086044574551036532421521012082

Materialdated

CharcoalCharcoal

SoilSoil

CharcoalCharcoalCharcoal

SoilSoilSoil

CharcoalCharcoal

Stratigraphicunit

NT1?NT2?KAG1KAG2KAG2KAG3KAG3KAG4KAG4KAG5KAG6KAG7

813C

-24.3-24.3-11.9-21.0-25.7-23.3-21.7-24.7-22.1

Post-1950-24.3-29.0

513Ccorrected age

38,800±110024,930±2408,470±756,900±655,360±605,H5±603,765±852,715±501,610±50Rejected580±70420±45

Calibrated ageAD/BC

Not calibratedNot calibrated7600-7351 BC5966-5643 BC4339-3999 BC4039-3775 BC2464-1940 BC973-799 BC264-566 AD

1284-1443 AD1418-1627 AD

Archaeological phase

PalaeolithicPalaeolithicMesolithicEarly-Middle NeolithicFinal NeolithicFinal NeolithicEarly-Middle Bronze AgeEarly Iron AgeRoman-Early Christian

Late ByzantineLate Byzantine-Post Byzantine

Figure 1.2 Schematic cross-section of the Gerakaris stream, depicting stratigraphic relationships and chronological data.


Neolithic). According to a radiocarbon date onthe middle part of a cumulic soil preservednear the valley walls, aggradation was inprogress by 6,900±65 bp (5966-5643 BC;Early-Middle Neolithic). Unlike the two old-est members of the Kato Agios Giannis allofor-mation, which are poorly exposed due to deepburial under the modern channel, KAG3 isobserved valley-wide. Aggradation of the unitis closely bracketed between 5,360±60 bp(4339-3999 BC; Final Neolithic) and 5,115±60bp (4039-3775 BC; Final Neolithic) and is fol-lowed by at least 1350 years of landscape sta-bility, sub-aerial exposure and pedogenesis.During that period, the Gerakaris floodplainwas inhabited, as indicated by an EarlyBronze Age site, identified during fieldworknear the valley axis. The stream aggradedagain sometime after 3,765±85 bp (2464-1940BC; Early-Middle Bronze Age) and KAG4 allu-vium covered the valley floor. By 2715±50 bp(973-799 BC; Early Iron Age), however, and forat least 1100 years, according to interbeddedarchaeological features across the lower reach-es of the stream, the landscape was stableagain. Sedimentation resumed soon after1,610±50 bp (264-566 AD; Roman-EarlyChristian period) and possibly around thethird or fourth century AD. Deposition of unitKAG5 was accompanied by lateral migrationof the stream to the north-east, closer to thevalley walls. As a result, the low Pierian hills,west of the modern village of Korinos, startedto be covered with fluvial deposits. The lastthree units (KAG6, 7 and 8) were deposited inan episodic fashion during the last 1300 years.According to archaeological inclusions in theKAG5 soil and a radiocarbon date on the soilformed within KAG6, this unit was depositedbetween the sixth-seventh century AD and580±70 bp (1284-1443 AD), and was followedby a period of stability Deposition of the nextyounger unit, KAG7, began after 580±70 bp(1284-1443 AD) but ended by 420±45 bp

(1418-1627 AD) when the landscape was stableonce again. The last unit, KAG8, was formedsome time between 420±45 bp (1418-1627 AD)and the late eighteenth century AD, when theGerakaris aggraded again, migrated laterallyto the north-east and its present position, andthen incised, abandoning the T2 surface andforming a new terrace (Tl). Deposits below Tland TO have not been examined in detail.

Architecturally, the Kato Agios Giannis allo-formation represents a sequence of alluvialdeposits, that accumulated mainly by verticalaccretion during valley-wide flooding events.A single stream incision event occurred fairlyrecently, between the fifteenth-seventeenthand the late eighteenth century AD. Differ-ences in depositional environment betweendifferent units indicate that the Gerakaris wasmoving across its floodplain throughout theHolocene. All members of the Kato AgiosGiannis alloformation are relatively fine-grained and they range in mean particle sizefrom loamy sands to silts, reflecting theirsource (the Upper Miocene-Lower Plioceneformations). The discrete sedimentary pack-ages are separated by very weakly to moder-ately well developed Inceptisols (KAG1,KAG2, KAG3, KAG4, KAG5 and KAG6palaeosols) and Entisols (KAG7 and KAG8soils) (Birkeland 1999), that show no evidenceof erosion during sub-aerial exposure andweathering. The degree of pedogenic devel-opment in the Holocene sequence mainlyreflects the time available for pedogenesisbefore renewed sedimentation. Overall, oldersoils appear to be more pedogenicallyadvanced than younger soils. There are inter-nal differences in sedimentary character andpedogenic expression and maturity betweentemporally similar deposits and palaeosols indifferent parts of the alluvial system.Palaeosols formed in a floodplain environ-ment are slightly more developed than con-temporary channel proximal soils. Some of

20 Aihanasia Krahtopoulou

the lower soils in the sequence (KAG1, KAG2and KAG3 soils) are mildly gleyed and/orexhibit manganese stains, indicating periodicsaturation by a fluctuating water table(Birkeland 1999).

The Agios Dimitrios Alluvial Sequence

The Agios Dimitrios sequence proved to beless complete than the Gerakaris, mainlybecause of limitations on fieldwork imposedby the relatively shallow incision of the stream,its heavily overgrown banks and the intensivecultivation of the surrounding landscape.

Three discrete geomorphic surfaces (TO, Tland T2) have been recognized in the lowerreaches of the stream, immediately above thecoastal plain of Korinos (Fig. 1.3). The relativ-ely narrow active floodplain (TO) rises 2-3 mabove the channel. It is inset into the first ter-race (Tl), which is a flat to gently sloping sur-face, up to 80 m wide and 5-6 m above thestream bed. The vestiges of T2 have been obser-ved north and mainly east, and approximately10 m above, the modern channel. Judgingfrom their relative position and pedogenicfeatures, the deposits beneath T2 representLate Pleistocene alluvial units, but they werenot examined in any detail during fieldwork.

Stratigraphic investigations focused on theHolocene alluvial sequence, which is infor-mally named the KHros alloformation andabbreviated KTR. It embraces at least sevensedimentary units (KTR1-7, from oldest toyoungest), which are separated by palaeosols.Radiocarbon dates (Table 1.2) and archaeolog-ical inclusions indicate that these depositsspan the last 7500 years at least.

The Kitros Alloformation

At some, as yet unknown, time the Agios

Dimitrios incised and abandoned the latePleistocene surface (T2) and formed the newHolocene valley, where the seven members ofthe Kitros alloformation started to accumulatein a punctuated fashion. Unit KTR1 is the old-est deposit identified so far in the AgiosDimitrios valley. It is poorly exposed and hasbeen identified only near the Holocene valleywalls at the edge of Tl. The lower boundary ofthe unit is not exposed and it is uncertainwhether or not other, older Holocene depositsare buried below KTR1. Absolute dates fordeposition of this fill are not presently avail-able. The soil formed within the unit is associ-ated with a Late/Final Neolithic archaeologi-cal site (5400/5300-3300/3100 BC), which pro-vides a minimum age for its deposition. Thereis no good maximum age for the fill. On thebasis of Stratigraphic position and soil mor-phology, however, it is certain that unit KTR1aggraded during the Holocene. After a periodof landscape stability and pedogenesis, sedi-mentation resumed and unit KTR2 was depo-sited. The A horizon of the soil formed withinKTR2 is cut by a hearth, associated with pot-tery dated to the sixth century AD. Further-more, the KTR2 soil is the most stronglydeveloped of the Holocene age observed todate in the Agios Dimitrios valley and exhibitsa cambic and stage II (Gile et al. 1966) calcichorizon. A palaeosol with similar pedogenicexpression, but with a slightly less stronglydeveloped calcic horizon, has been observedin the Gerakaris Holocene alluvial sequence(the soil formed within KAG4). This KAG4soil had a minimum of 1100 years in which toform. This suggests that KTR2 was depositedby the Agios Dimitrios some time before theClassical period (fifth century BC), after whichthe channel incised, leaving the surface uponwhich this strong soil developed free fromflooding until deposition of KTR5 alluvium.

The next unit described is KTR3, which waslaid down before 2,100±100 bp (390 BC-123 AD;

Figure 1.3 Schematic cross-section of the Agios Dimitrios stream, depicting stratigraphic relationships and chronological data.

Table 1.2 List of radiocarbon dates from the Agios Dimitrios stream.

Lab. no.

AA31854AA31858AA31855AA31857AA31856

Depth(cm)

33037025020060

Materialdated

CharcoalCharcoalCharcoalCharcoalCharcoal

Stratigraphicunit

KTR3KTR3KTR4KTR4KTR6

513C

-25.5

-26.0-24.1-24.3

513Ccorrected age

2,100±1002,100±1701,675±501,540±45530±45

Calibratedage AD/BC

390BC-123 AD517BC-318 AD243-530 AD419-637 AD

1308-1453 AD

Archaeological phase

Classical-RomanClassical-RomanRoman-Early ChristianEarly Christian-Early ByzantineLate Byzantine-Post-Byzantine

22 Athanasia Kmhtopoulou

Classical-Roman period). There is no goodmaximum age for deposition of this fill, and itis likely that deposits predating KTR3 areburied in the Agios Dimitrios valley axis. In-complete exposure of the Holocene sequencein this stream, however, inhibits any conclu-sive argument on the occurrence, number andnature of these deposits, and additional field-work is necessary to clarify stratigraphic rela-tionships and establish chronologies. The nextunit detected, KTR4, was deposited in theHolocene valley between 2,100±100 bp (390BC-123 AD; Classical-Roman period) and1,675±50 bp (243-530 AD; Late Roman-EarlyChristian period) and was followed by a peri-od of stability and pedogenesis. The AgiosDimitrios aggraded again at some pointbetween 1,540±45 bp (419-637 AD; EarlyChristian period) and the tenth-twelfth centu-ry AD, according to an interbedded Byzantinesite. KTR5 alluvium now covered the entirevalley, overlying unit KTR4 in the centre of thevalley and KTR2 at the margin. The next allu-vium recognized in the sequence (KTR6) wasdeposited at some time after the tenth-twelfthcentury AD and before 530±45 bp (1308-1453AD). An infilled palaeochannel, associatedwith this unit, indicates that the AgiosDimitrios was flowing closer to the valleywalls at this time. The youngest depositdescribed (KTR7) was laid down after 530±45bp and, judging from the minor degree ofpedogenic development, not long before thepresent. The Agios Dimitrios changed itscourse, migrating slightly to the south to occu-py its present bed, and incised, isolating theTl terrace. Deposits lying below the modernfloodplain (TO) have not been examined, but itseems likely that they are very recent.

Architecturally, the Agios Dimitrios sequenceis slightly different from the Gerakaris. Theoldest two alluvial units described (KTR1 andKTR2) are observed in vertical succession nearthe valley walls, while the vertically stacked

units KTR3 and KTR4 are inset into these.Valley-wide sedimentation resumed withdeposition of unit KTR5, which buried allolder alluvial fills. Finally, the youngest unitsKTR6 and KTR7 also accumulated verticallyabove the KTR5 alluvium. Two relativelydeep and narrow, infilled palaeochannels areassociated with units KTR1 and KTR6 andindicate that, unlike in the Gerakaris, therehas been little lateral migration of the AgiosDimitrios stream since the Late/Final Neo-lithic period.

All seven members of the Kitros alloformationare fine-textured (sandy loams, silt loams andsilts), reflecting the composition of the sourcematerial (the Upper Miocene-Lower Plioceneformations), and represent floodplain deposits.Colluvial deposits interfinger within theKTR4 and KTR7 fills near the valley walls,suggesting that aggradation of these fills wastriggered by slope destabilization and large-scale sediment input into the system.

The soils formed within the Kitros alloforma-tion are Inceptisols, with the exception of theyoungest, KTR7 soil, that shows minor pedo-genic development and can be classified as anEntisol (Birkeland 1999). Overall, there is adecrease in pedogenic maturity from older toyounger units. Occurrences of manganeseoxides in KTR1 and of redox mottles in KTR2,KTR3 and KTR5 suggest some saturation by afluctuating water table (Birkeland 1999).

Correlation of the Gerakaris and AgiosDimitrios Sequences

Overall, the Holocene sequences of both theGerakaris and Agios Dimitrios streams arecharacterized by long periods of stability andsoil formation, interrupted by infrequent, yetin some cases dramatic, intervals of valleyaggradation and occasional stream incisionand lateral migration. Moreover, in both

Holocene Alluvial History of Northern Plena, Macedonia, Greece 23

streams, the frequency of alluviation increaseddramatically over the last two millennia.Although the earliest Holocene deposits maybe buried, the relative rarity of alluviation inthe early to middle Holocene in both streamsis not an artefact of the exposure of relevantdeposits. In detail, however, the two streamsdo display some differences in the characterand timing of alluviation events and streamadjustments (Fig. 1.4). In part, but only in part,these differences can be attributed to the con-trasting level of chronological precision achie-ved for each sequence and to more restrictedfieldwork in the Agios Dimitrios stream.

The more complete and better dated Gera-karis sequence will be used as the basis for thefollowing synthesis. The earliest alluvial unit(KAG1) was deposited in the Gerakaris valleyfloor at some time before 8.5 ka bp. Depositsof similar age have not been identified in theAgios Dimitrios valley, owing perhaps to theincomplete exposure of its early Holocenealluvial sequence. On the basis of radiocarbondates for the Gerakaris sequence and of arch-aeological inclusions in the Agios Dimitriossequence (a Late/Final Neolithic site), KAG2in the Gerakaris seems to be contemporarywith KTR1 in the Agios Dimitrios and bothwere laid down before the Final Neolithic.KAG3 covered the entire Gerakaris valleybetween 5.3 ka and 5.1 ka bp, during the FinalNeolithic, and was followed by at least 1350years of landscape stability. No equivalentalluvial unit has yet been identified in theAgios Dimitrios valley. The next alluvium rec-ognized in the Gerakaris sequence (KAG4) wasdeposited between 3.7 ka bp and 2.7 ka bpand was followed by another prolongedphase of landscape stability and soil forma-tion, which lasted for a minimum of 1100years. In terms of pedogenic expression anddegree of pedogenic maturity, the soil formedon KTR2 and the KAG4 palaeosol are verysimilar (both exhibit Ab-Bwkb-Bkb soil pro-

files with stage II calcic horizons). Therefore,KAG4 and KTR2 could be contemporary,although at present any such correlation mustbe tentative and the stream downcutting,which accompanied deposition of KTR2, hasnot been observed in the Gerakaris sequence.KTR3 aggraded at some time before 2.1 ka bpand none of the units presently identified inthe Gerakaris valley is temporally equivalentto this fill. The Gerakaris aggraded again, theKAG5 fill being deposited some time after 1.6ka bp and most likely around the third-fourthcentury AD, and the stream migrated to thenorth-east. The KTR4 alluvium, laid downbetween 2.1 ka bp and the third-sixth centuryAD, but not accompanied by lateral migration,can be tentatively correlated with KAG5.Likewise, the three alluvial events, KAG6,KAG7 and KAG8, dated between the sixth-seventh century AD and the late eighteenthcentury AD, are tentatively correlated withKTR5, KTR6 and KTR7, which aggradedbetween the fifth-seventh century AD and thepresent. The Gerakaris changed its course andincised most likely between the fifteenth-sev-enteenth and eighteenth century AD, while theAgios Dimitrios also migrated laterally andvertically at some point after the fourteenth-fifteenth century AD.

The alluvial histories of the two streamsthus appear to correlate well over the last twomillennia. Both streams also appear to haveexperienced common phases of alluviation atsome time between the late ninth and earlysixth millennia bp and again in the fourth-third millennia bp. The two sequences cannotbe compared for the earliest part of theHolocene, because of the incompleteness ofthe record from the Agios Dimitrios stream.The two streams do appear to diverge at leastat two points: in the late sixth millennium bp,when the Gerakaris alluvium KAG3 is notmatched in the Agios Dimitrios valley, and atsome time between perhaps the fourth-third

Figure 1.4 Comparative chronology of alluvial activity in the Gerakaris and Agios Dimitrios streams. The length of each rectangleindicates not the duration of alluviation, but the uncertainty of dating; solid, dashed and missing borders of rectanglesindicate termini of increasing uncertainty.


millennium and 2100 bp, when the AgiosDimitrios experienced the KTR3 alluviation,apparently missing in the Gerakaris valley.

Landscape Change and the Formation of theArchaeological Record

The geomorphic processes that have alteredthe landscape of the study area have also seri-ously affected the survival and visibility of thearchaeological record. Most of the known set-tlements in the catchments of the Gerakarisand Agios Dimitrios streams have sufferederosion, which has removed much of the pre-existing soils and archaeological deposits(Pappa 1999), and total destruction of small,ephemeral or perhaps early sites cannot beprecluded. On the other hand, sediment trans-ported by both streams has buried the lowervalleys under thick alluvial deposits, obscur-ing 'ancient topographies' and archaeologicalsites from surface survey, however intensive.Chance finds in exposures demonstrate thatsites buried in this way range in date from atleast the Late/Final Neolithic to the Byzantineperiod (Krahtopoulou 2001).

Causality: Agents of Landscape Change

Landscape changes, such as have been docu-mented in the study area, tend to be attributedto 'natural' causes, when alluvial episodes indifferent catchments appear to be contempo-rary (Vita-Finzi 1969), and to anthropogeniccauses, when this is not the case (van Andel etal 1990). On this basis, it might be argued thatalluviation in the study area was driven by cli-mate, or some other natural factor(s), duringthe early Holocene and also during the last2000 years, with a more complex alternationbetween natural and cultural forcing in theintervening millennia. This interpretation

would be broadly consistent with the argu-ments advanced by van Andel et al. (1990) infavour of widespread erosion triggered byagricultural activity towards the end of theNeolithic and beginning of the Bronze Age.

In the case of this study area, however, anumber of considerations caution againstacceptance of this conclusion. First, the mid-Holocene parts of the Gerakaris and AgiosDimitrios sequences, which have been inter-preted as evidence for locally variable alluvialhistories, include some serious chronologicalambiguities. It is possible, therefore, that fur-ther dating evidence will undermine some ofthe apparent divergence between the twostreams. Secondly, the Gerakaris alluvial unitKAG3, for which no equivalent has beenfound in the Agios Dimitrios sequence, is veryclosely dated to the late sixth millennium bpor Final Neolithic. Available archaeologicalevidence suggests human occupation of bothcatchments at this time, whereas the contrast-ing alluvial records might lead one to expectoccupation of the Gerakaris catchment but notof the Agios Dimitrios. It must be stressed thatthe archaeological evidence is too coarselydated to establish unambiguously its contem-poraneity with KAG3. Moreover, an anthro-pogenic trigger for this or any other alluvialepisode presumably depends on the scale,longevity and severity of human impact, whichcannot be assessed from the crude evidence ofthe presence or absence of human settlement.On the other hand, the scant settlement evi-dence does nothing to strengthen the interpre-tation of KAG3, in anthropogenic terms, as theproduct of (intensive) human activity restrict-ed to only one of the two catchments studied.

The evidence from the study area can alsobe approached from a rather different direc-tion. A striking common feature of bothstreams is that, through most of the Holocene,the landscape has been characterized by longperiods of stability and pedogenesis, punctu-

26 Athanasia Krahtopoulou

ated by short-lived and relatively infrequentepisodes or phases of erosion and alluviation.This might suggest a landscape normallyunder limited threat of anthropogenic destabi-lization. Over the last two millennia, however,both catchments were subject to frequentdestabilization and alluvial deposition. In thecase of the KTR4 and KTR7 units in the AgiosDimitrios valley, interleaving with colluvialdeposits suggests that increased sedimentinput into the system was triggered by slopedestabilization. Moreover, an anthropogeniccontribution to these recent alluvial episodesreceives some support from both historicaland palynological evidence (Gerasimidis1984) for human activity in the Pierianuplands over the last two millennia. Thus,even though the late Holocene alluvial histo-ries of the Gerakaris and Agios Dimitrioscatchments apparently correlate well, a plau-sible argument could be mounted in favour ofa major causal role for anthropogenic distur-bance. It should also be noted that the alluvialhistories of the two streams developed in par-allel over a period of time far longer than theusual climatic suspects, such as the early mod-ern 'Little Ice Age7 (Lamb 1982).

It might be argued that the two catchmentsstudied here are too close for correlation oftheir alluvial histories to be able to discrimi-nate between climatic and anthropogenic cau-sation, but their proximity also factors outseveral other potential environmental con-trasts, such as in geology, altitude, topogra-phy or climax vegetation. On the other hand,it is also striking that the chronological resolu-tion of the present study, which is arguablyfiner and firmer than that of most precedingstudies in Greece, sets only proxima for thebeginning and end of each depositional event,and still leaves sufficient ambiguity to make itdifficult to argue for a particular form of cau-sation on the basis of the degree of synchro-nism between different alluvial sequences.

Conclusion

This study has revolutionized understandingof the degree to which the landscape of north-ern Pieria has changed during the Holoceneand also of the extent to which these changeshave shaped survival and visibility of thearchaeological record. An understanding ofthe nature of landscape change will also playa valuable role in future attempts to under-stand the changing pattern of human settle-ment in this region in terms of the distributionof resources. For example, the progressive cre-ation of an extensive coastal plain and relatedshift eastwards of the coastline are likely tohave exercised a significant influence over set-tlement location. Geoarchaeological studyalone is unlikely to resolve the critical issue ofthe cause(s) of landscape change. Well-datedsequences such as those described here, how-ever, in combination with independent arch-aeological and other paleoecological recordsof comparable quality, would surely underpinfruitful discussion of these issues.

Acknowledgments

This paper is based on research undertakenfor my PhD thesis at the University ofSheffield, partially funded by the Universityof Sheffield and the A.G. Leventis Foundation.Generous funding for the radiocarbon datespresented here was provided by INSTAPthrough the Makrigialos post-excavation pro-ject. I would like to thank my supervisors, DrCharles Frederick and Dr Paul Halstead, forfruitful and stimulating discussions duringthis research, and Manthos Besios of the 1STEphorate of Antiquities of Thessaloniki forsharing with me his deep knowledge of thelandscape and archaeology of NorthernPieria. Dr Paul Halstead has also helped infinalizing and editing the language of this


paper. Special thanks to my field companionsKostas, Themis and Christos who alwaysmanage to turn stressful days in the field intoentertaining expeditions.

Bibliography

Andel, T. van, E. Zangger and A. Demitrack1990 Land use and soil erosion in prehistoric and his-

toric Greece. Journal of Field Archaeology 17: 379-96.

Besios, M.1993 Martiries Pidnas. To psifisma tou Apollonos

Dekadriou. Ancient Macedonia 5 (2): 1111-221.Besios, M. and M. Pappa

1995 Pidna. Thessaloniki: Pieriki Anaptixiaki.Birkeland, P.W.

1999 Soils and Geomorphology. Oxford: Oxford Univ-ersity Press.

Endfield, G.H.1997 Myth, manipulation and myopia in the study of

Mediterranean soil erosion. In A. Sinclair, E.Slater and J. Gowlett (eds.), ArchaeologicalSciences 1995. Oxbow Monographs 64: 241-48.Oxford: Oxbow.

Gerasimidis, A.M.1984 Stathmologikes sinthikes kai metapagetodis exelixi

tis vlastisis sta dasi Lailia Serron kai KatafigiouPierion. Unpublished PhD thesis, University ofThessaloniki.

Gile, L.H., F.F. Peterson and R.B. Grossman1966 Morphological and genetic sequences of car-

bonate accumulation in desert soils. Soil Science101: 347-60.

Grammenos, D., M. Besios and S. Kotsos1997 Apo tous proistorikous oikismous tis Kentrikis

Makedonias. Makedoniki Vivliothiki 88. Thessa-loniki: Dimosievmata Etairias MakedonikonSpoudon.

IGME1986 Geological Map of Greece, 1:50,000: Katerini Sheet.

Athens: Institute of Geology and MineralExploration.

Krahtopoulou, A.2001 Late Quaternary Alluvial History of Northern

Pieria, Macedonia, Greece. PhD dissertation,University of Sheffield, in preparation.

Lamb, H.H.1982 Climate, History and the Modern World. London:

Methuen.Pappa, M.

1999 I organosi tou khorou stous Neolithikous oik-ismous tis Vorias Pierias. In Ancient Macedonia,6th International Symposium, Thessaloniki 1996: II,873-86. Thessaloniki: Institute for BalkanStudies.

Vita-Finzi, C.1969 The Mediterranean Valleys: Geological Changes in

Historical Times. Cambridge: Cambridge Univ-ersity Press.

Palynological Evidence for Human Influence onthe Vegetation of Mountain Regions in NorthernGreece: The Case of Lailias, Serres

Achilles Gerasimidis

Introduction

Man, directly or indirectly, has been the mostsignificant factor in the evolution of vegeta-tion over the last few thousand years. Theantiquity and intensity of human influence onthe natural vegetation differs from place toplace and is related to the local history ofhuman settlement. The magnitude of humanimpact on the natural environment, however,is also influenced by prevailing environmen-tal conditions.

In Greece, human impact on the vegetation,and on the natural environment in general, isof considerable antiquity. The establishmentand expansion of farming from the Neolithiconwards required the clearance of, initially,virgin forests, and agricultural activities havecontinued through time to influence the forestvegetation to varying degrees. An increasingneed for wood, the main source of fuel, alsoentailed intensive use of the forests, oftenthrough clear-cutting of large areas. Suchinfluences have resulted in fundamental shiftsin the evolution of the natural vegetation and,in areas where the natural environment wassensitive and unstable, have caused completedestruction of forest vegetation. Apart fromthese negative effects, however, human inter-ference has also long exercised a more positiveinfluence on the vegetation of Greece through

the introduction and establishment or the pro-tection of various forest species.

Palynology provides a means of investigat-ing the history of forest vegetation in Greece.For mountain regions in particular, whereother sources are almost non-existent, pollendiagrams constitute the best evidence for thehistory of forest vegetation and of humaninterference in its evolution. Palynologicalresearch in Greece in recent decades has pro-duced information on the history of vegeta-tion from more than 30 locations in variousparts of the country, particularly from areas oflow or intermediate altitude (Fig. 2.1).

Evidence of Human Influence on ForestVegetation in Greece

The effects of human activity on the evolutionof vegetation are apparent in all pollen dia-grams from Greece and especially those thatcover the last few millennia. These diagramsreveal fluctuations in vegetation coverbetween alternating periods of destructionand stability, sequential modifications of thecomposition and dominance of forest vegeta-tion, and changes in the frequency of pollenfrom plants associated with human activity.Directly or indirectly, and to varying degreesand on varying scales, humans have evidently

2

Palynological Evidence for Human Influence on Vegetation 29

1. Lerna2. Thermisia3. Kiladha4. Kleones5. Osmanaga6. Kaiafa7. Voulkaria8. Trikhonis

9. Kopais10. Halos11. Litochoro12. Aghia Galini13. Tristinika14. Xinias15. Viviis16. loannina

17. Gramousti18. Tenagi Philippon19. Gravouna20. Volvi21. Giannitsa22. Edessa23. Vegoritis24. Kastoria

25. Khimaditis26. Pertouli27. Rezina28. Pieria29. Voras30. Paiko31. Rhodopes32. Lailias

Figure 2.1 Location of palynological investigations in Greece.

30 Achilles Gemsimidis

influenced the evolution of vegetation overseveral millennia. This influence has affectedalmost every vegetation type in Greece(Gerasimidis 1995).

In the lower Eu-Mediterranean zone (Quer-cetalia ilicis),1 the natural vegetation includesextensive maquis (evergreen sclerophyllousshrubs), with phrygana (semi-shrub heath-land) dominant in degraded areas. In certainlocations, the forest vegetation comprisespines (Pinus halepensis, P. brutia and P. pinea).In this zone, the natural vegetation is particu-larly degraded, because of the antiquity ofhuman settlement and the intensity of humanactivity. Pollen diagrams from this zone indi-cate that the extent, and in some cases initialappearance, of the Quercetalia ilicis is anthro-pogenic. More specifically, the more degradedvegetation units of this zone, such as the mod-ern distribution of the Oleo-Ceratonietumassociation, are the result of long-term humanactivity and impact on the natural vegetation.Some pine forests in this zone also result,directly or indirectly, from human interven-tion.

The hilly/sub-montane zone of deciduousforest (Quercetalia pubescentis) replaces theQuercetalia ilicis at higher elevations andmore inland locations and covers much ofcontinental Greece. The vegetation of thiszone has also been affected by thousands ofyears of severe human interference. Apartfrom the destruction of forest cover by intensehuman use, the natural vegetation in placesappears to a greater or lesser extent degraded.Human influence is more marked in the vege-tation of the lower subzone, Ostryo-Carpinion, where hornbeams (Ostrya carpini-folia, Carpinus orientalis) and kermes oak(Quercus coccifera) are dominant, but is alsoapparent in the upper subzone of Quercionconfertae, where oak forests and other broad-leaved deciduous trees are present. Palyno-logical data show that this zone was also dom-

inant at lower elevations before its anthro-pogenic degradation to vegetation types ofQuercetalia ilicis. Anthropogenic impact alsoproduced qualitative degradation in the vege-tation types of this zone: dominant mixed oakforests were reduced either to pure oak forestof low growth or to the vegetation types of thesubzone Ostryo-Carpinion.

Above the Quercetalia pubescentis zone, ataltitudes higher than 900-1100 m, the climatechanges gradually from Mediterranean tocontinental and, naturally, a similar changeoccurs in the vegetation. Forest vegetation inthe higher zones (Fagetalia and Vaccinio-Picetalia) typically comprises beech (Fagus),fir (Abies) and other coniferous species ofgreater or lesser cold tolerance. In the forestvegetation Fagetalia zone, beeches and firs aredominant, with Austrian pine (Pinus nigra)locally more or less strongly represented. Thelatter species also expands into the Querce-talia pubescentis. According to pollen dia-grams from the Fagetalia zone, firs were morewidespread in the past and even dominant incertain areas where today this species is rareor completely absent. The reduction or disap-pearance of firs was mainly anthropogenic inorigin. Their removal often favoured beechtrees, the expansion of which was thus sub-stantially dependent on humans.

The highest forest vegetation zone is theVaccinio-Picetalia, in which coniferous forestswith Scots pine (Pinus sylvestris), Bosnian pine(Pinus leucodermis) and spruce (Picea) are pre-sent. Palynological data show that destructiveinterference at high elevations has at varioustimes favoured the expansion of pines, proba-bly Scots pine which is the main representa-tive of the genus in this zone. Spruce hasalways been restricted to the Rhodopi moun-tains. The disappearance of birch (Betula),which is a component of this zone, from moresoutherly locations than those in which itappears today may partially be attributed to


the effects of human activity on forest vegeta-tion (Gerasimidis 1996).

The destruction of forest vegetation at thehighest altitudes, starting several thousandyears ago, has resulted in the creation orexpansion of pseudo-alpine vegetation whichappears above the tree line in the Astragalo-Acantholimonetalia zone.

Palynological Records of AnthropogenicImpact from the Mountains of NorthernGreece

Palynological evidence for the vegetationalhistory of a mountain region can be derivedfrom nearby sites at low elevations. Thisassumes, however, that the pollen representsnot only the area adjacent to the coring sitebut a more or less broad region, depending onthe particular conditions of pollen production,dispersal and preservation. Reconstruction ofthe evolution of vegetation in mountain regionsis more reliable, therefore, when based on pol-len diagrams from within these regions. Suchdiagrams are of particular significance as cer-tain of the pollen types associated with hu-man activity have a short radius of dispersal.

Of the palynological investigations conduct-ed in Greece, only seven are from mountainregions. Of the latter, two (Fig. 2.1: sites 26 and27) are from central and north-west Greecerespectively (Athanasiadis 1975; Willis 1992),while the remaining five (Fig. 2.1: sites 28-32)are from northern Greece (Gerasimidis 1985;Athanasiadis and Gerasimidis 1986; 1987;Athanasiadis et al. 1993).

Past human activity in the mountains ofnorthern Greece is indicated in these dia-grams principally by plants regarded asanthropogenic indicators and by the ratio ofarboreal pollen to non-arboreal pollen (AP:NAP). Anthropogenic indicators includepollen types directly or indirectly associated

with human activity, such as those of cultivat-ed plants, weeds and plants associated withgrazing, overgrazing and also human creationof pasture (e.g. Cerealia, Olea, Vitis, Plantago).Other indicators are plants characteristic ofdestructive human impact on natural vegeta-tion by fires, clear-cutting and so on (e.g.Juniperus, Artemisia, Chenopodiaceae) andalso species introduced through human activ-ity (e.g. Juglans, Castanea). A reduction inAP.-NAP values reflects the reduction of forestcover and, when associated with anthro-pogenic indicators, may be attributed tohuman activity. Conversely, an increase in theAP:NAP ratio usually represents the expan-sion of forest cover, following a reduction orcessation of intense human interference in theforest vegetation.

The palynological data from the mountainsof northern Greece in most cases cover onlythe last few millennia (Gerasimidis andAthanasiadis 1995) and so provide limitedinformation on the more distant past. It seemsfrom the available palynological data, howev-er, that human influence during the prehis-toric period on the natural vegetation was farless marked in northern than in southernGreece. This can be attributed, in part, to themore intensive nature of past human activityin southern Greece and, in part, to the differ-ence in climatic conditions. In northernGreece, and particularly in the mountains, theclimate is more conducive to natural recoveryof forest vegetation from the effects of anyhuman impact.

Of the pollen diagrams from the mountainsof northern Greece, that from Voras covers thelongest period, approximately 7000 years(Athanasiadis and Gerasimidis 1986). In thisdiagram, evidence of human influence is min-imal before the last millennium BC and, in thepalynological record in general, intensehuman interference in the vegetation of themountains of northern Greece first becomes

32 Achilles Gerasimidis

apparent to varying degrees at about thistime. In the Rhodopes (Athanasiadis et al.1993), destructive interference, probably bypeople, had changed the appearance of theupland woodlands by the end of the secondmillennium BC. In particular, the dominantspruce and the fir, as well as other species thatwere part of the forest vegetation, almost dis-appeared. Only birch expanded, presumablyin areas where forest was destroyed. It seemsthat this destruction was temporary, however,as the forest vegetation of the mountainsrecovered, mainly in the form of conifers. Thelower section of the Paiko diagram indicatesthat, by the early centuries of the first millen-nium BC (Athanasiadis and Gerasimidis 1987),the forest cover of the area, and the pineforests in particular, had already contracted inthe face of human activity.

In the ensuing period and up to the begin-ning of the Roman era, the palynological datasuggest continuous human activity in themountains of northern Greece, but withoutintensive interference in the forest vegetation.The most significant impact on forest vegeta-tion during this period is the appearance ofwalnut and chestnut, trees which are thoughtto have been human introductions (Bottema1974; Athanasiadis 1975).

Intensive interference in the forest vegeta-tion took place during the last centuries BC, atabout the time of the Roman occupation,when the palynological data suggest thatintense interference at high elevations result-ed in the reduction of coniferous woodland.The firs were most affected, becoming lessabundant or almost extinct in many regionswhere they were previously dominant. Con-versely, this contraction favoured the expan-sion of beech forests, which occurred immedi-ately afterwards. Similar anthropogenic des-truction of fir and pine forests during thesame period is also reported for numerousareas in Bulgaria (Bozilova et al. 1995).

The pollen diagrams from the mountains ofnorthern Greece reflect the importance of themountains as a periodic refuge for those flee-ing insecurity or economic crisis in the plains.Palynological evidence of reductions in forestcover and increases in agricultural activitymay be correlated roughly with historicallydocumented periods of insecurity or crisis inthe plains. Conversely, in periods of stability,the return of population to the plains allowedthe expansion of forest in the mountains. Asyet, however, palynological evidence from themountains is not dated closely enough to dis-tinguish between the effects of short-terminsecurity, for example during periods of war-fare, and longer-term changes in land owner-ship, taxation, market prices or internationaltrade. Palynological evidence of intensehuman interference in forest vegetation wasmore or less localized prior to the fifteenthcentury AD.

Around the time of the Turkish conquest ofnorthern Greece, forest vegetation in themountains was subject to intense anthro-pogenic pressure, perhaps from refugee popu-lations, as reflected in the pollen diagrams bya reduction of forest cover and an increase inindicators of human presence and activity. Anexception is the Lailia diagram which is dis-cussed below.

Thereafter, during the course of the Turkishoccupation, a general decline is observed inhuman activity in the mountains of northernGreece, although periodic episodes of inten-sive human activity may be associated withhistorical events. For example, following thefailure of a revolutionary movement duringthe seventeenth century AD, refugees first fledto the western foot of the Pieria mountainsand later withdrew to near the summit wherethey established the village of Katafigi(Nastos 1971). This event may be reflected inthe palynological data from the Pieria moun-tains in a reduction of forest cover and an


increase in indicators of human settlement(Gerasimidis 1985).

The uppermost sections of pollen diagramsfrom the mountains of northern Greece indi-cate a gradual decline in anthropogenic influ-ence and an expansion of forest cover. Thiscan be attributed to the gradual reduction ofmountain population and to the introductionof systematic measures for the protection ofthe natural environment. The historical evolu-tion of mountain vegetation in northernGreece and its relationship with human influ-ence may be illustrated by analysis of theLailias pollen diagram in the light of historicalinformation for the same area (Gerasimidis1985).

Anthropogenic Influence on the Develop-ment of Vegetation in the Mountain Regionof Lailias

Environmental Setting

Geography, geology and climate. Lailias is situat-ed in the western part of the Vrontou moun-tains which extend north of Serres and southof the Greek-Bulgarian border. In the studyarea, acid igneous rocks prevail. Climatic con-ditions change with increasing altitude fromMediterranean to continental. In the moun-tain area, the climate may be characterized ashumid continental with warm summers andharsh winters.

Vegetation. The lower part of the mountainslies within the Quercetalia pubescentis zone.Scrub of varying density or low woodland,with hornbeams (Carpinus orientalis andOstrya carpinifolia), kermes oak (Quercus coc-cifera), field elm (Ulmus minor), flowering ash(Fraxinus ornus) and so on, prevail in thelower subzone, Ostryo-Carpinion, up to 600-800 m altitude. Between 600-800 m and 900-

1200 m, the majority of the area is in fields ormeadows or bare ground. Some relict oaktrees reveal that the area belongs to a greatlydegraded Quercion confertae subzone. Thehigher part of the mountain up to the summit(Ali Baba at 1849 m) is occupied by the Fagionmoesiacae subzone of the Fagetalia zone.Beech woods prevail from 1200 to 1800 m,although beech trees occur singly or in clus-ters in some places down to 800 m. Scots pine(Pinus sylvestris) is a very important elementof the Lailias forest, but is gradually givingway to the expanding beeches. In this area,where it is considered extrazonal, Scots pineoccurs both together with beech and, whereconditions are not appropriate for beech, inunmixed stands.

Material and Methods

The coring site. The profile for pollen analysiswas taken from a small (3 ha) peat bog at analtitude of 1420 m and at coordinates 41 °16' 4"N, 23° 35' 58" E Greenwich. In the surround-ing area, the vegetation belongs to theFagetum moesiacae association.

Radiocarbon dates. The following three radio-carbon dates were obtained:

Depth (cm)1. 58-612. 127-1323. 175-200

Lab. ref. no.VRI-746VRI-747VRI-748

Years bp250±80910±80

1870±140

The third date is considered too young for thissection of the profile, but it may be correct if itcorresponds to the upper part of this ratherlong (25 cm) sample.

The diagram. The pollen diagram (Fig. 2.2) wasdrawn by the TILIA program. The percent-ages are based on the total pollen sum exclud-ing the spores of fern and hydrophytes.

Figure 2.2 Pollen diagram of Lailias.


Analysis of Diagram: Results and Discussion

The diagram was divided into eight phases(A-H, from oldest to youngest), primarily onthe basis of human influence on the develop-ment of the vegetation. The dating of eachphase is approximate and based on the radio-carbon dates and on comparison with otherpollen diagrams from mountain areas. Localhistorical data are mainly derived fromSakellariou (1982).

Phase A: Tilia-Corylus-NAP, 260-255 cm,AP:NAP=56:44 (c. 4000 BC). In the oldest partof the core, forest vegetation appears to belimited. At the same time, certain anthro-pogenic indicators such as Plantago, Artemisia,Oxyria-type and Chenopodiaceae are presentat a low level. Interpretation of this initialphase is complicated, however, by the suddenchange in vegetation that follows.

Phase B: Tilia-Corylus, 255-215 cm, AP:NAP=87:13 (4000-2000 BC). Dense deciduous forestsdominate, with Tilia and Corylus the majorspecies. There is no clear evidence of humanactivity.

Phase C: Abies-Pinus, 215-195 cm, AP:NAP=93:7 (2000-300 BC)The vegetation of Quercetalia pubescentisretreats to lower elevations and the mountainis covered with dense coniferous forests, dom-inated by Abies. The change is probably attrib-utable to changing climatic conditions, asthere is no evidence of human influence.

Phase D: Abies-Pinus-Fagus-NAP, 195-145 cm,AP:NAP=42:58 (300 BC-800 AD). The phasestarts with a marked retreat of forest vegeta-tion. An anthropogenic cause for the changein vegetation is suggested by indicator taxa,such as Plantago, Oxyria-type and Chenopod-iaceae, and in particular by the sudden

increase of cereals and appearance of Juglans.Given the evidence of human activity, theabundant indicators of dry environmentshould probably be attributed to the clearanceof forest and the increase in Artemisia relatedto clear-cutting. This destructive interferenceis dated to the second century BC, when theregion was occupied by the Romans. The peri-od between the end of the second century BCand the middle of the first century AD also sawnumerous destructive raids by barbariantribes, using the valley of the Strimon as a cor-ridor from the North. As a result, inhabitantsof the lowlands may have moved to themountains for protection, causing largechanges in vegetation. Following the majordestruction of the forest vegetation, continuedhuman activity is indicated by the palynolog-ical evidence, and particularly by values forcereal pollen, but the forest expanded oncemore (with Fagus an important component)and, despite some fluctuations as a result ofhuman interference, maintained a large per-centage cover. The phase ended as it hadbegun, with an even more severe destructionof forest vegetation, which may be correlatedwith Slavic incursions into the area. Thesereached their peak at the end of the seventhcentury AD, perhaps forcing the local popula-tion once again to move to the mountains.This destruction caused a significant qualita-tive change in the composition of the Lailiasforest vegetation, leading to the disappear-ance of Abies, which was hitherto the domi-nant species in the forest vegetation of thearea. Following this there was a small recov-ery of Abies but then it became extinct.

Phase E: Fagus-Quercus-Pinus, 145-125 cm,AP:NAP=61:39 (800-1100 AD). During thisphase, a major increase is observed of forest,particularly of broad-leaved, deciduousspecies. Especially significant is the increasein Fagus, which becomes dominant. Increased

36 Achilles Gerasimidis

pollen values for Quercus and, especially,Ostrya/Carpinus orientalis-type indicate a gen-eral stability in the region which allow theexpansion of forests to lower elevations.Human presence is indicated by Juglans, Vitis,Plantago, Oxyria-type, and so on, but theimpact on forest vegetation is limited. Thedecrease of cereal pollen values also indicatesa reduced level of human activity.

Phase F: Pinus-NAP, 125-95 cm, AP:NAP=35:65(1100-1400 AD). During this phase, the forestcover is once again reduced. Fagus declinesand there is also a significant reduction ofQuercus. The decline of oak forest, togetherwith the reduction in vegetation of theOstryo-Carpinion subzone, indicates destruc-tion of the lower-elevation forests. This wasparticularly important for the oak forests ofthe Quercion confertae, as the forest vegeta-tion of this subzone did not subsequentlyrecover and its presence today is minimal.Indicators of dry environment, such asCaryophyllaceae, Asteraceae, Cichoriaceae,Poaceae, Juniperus and so on, confirm thatextensive clearings existed during this phase.Also present, but at low levels, are indicatorsof cultivation and grazing. During this period,the area was successively invaded and occu-pied (with some interludes of renewedByzantine rule) by Normans, Franks, Bulgar-ians and Serbs. These historical events seem tohave had a more or less destructive impact onthe forest vegetation.

Phase G: Pinus-NAP, 95-55 cm, AP:NAP=54:46(1400-1800 AD). The recovery of forest coverduring this phase is attributed to the increaseof Pinus, while Fagus is present in low num-bers. Indicators of cultivation and clearingsoccur throughout this phase, but with low val-ues. In general, no destruction of forest vege-tation is observed, although the extrapolateddate for the beginning of the phase is the time

when the Turks invaded the area. In pollendiagrams from other mountain areas in north-ern Greece, the Turkish occupation may havebeen accompanied by destruction of the forestvegetation, albeit to varying degrees in differ-ent areas. Serres was one of the first Macedo-nian cities to be taken by the Turks (in 1383AD) and showed no great resistance. To influ-ence the population of other areas, the Turkstreated Serres well, not converting churches intomosques and allowing the people of the cityto be self-governing. For this reason, the inhab-itants of the plains did not move to the moun-tains for protection, as in previous periods.

Phase H: Pinus-NAP, 55-0 cm, AP:NAP=54:46(after 1800 AD). During the last phase, theAP:NAP ratio suggests that the scale of forestcover remained as in the previous phase. Asignificant qualitative change took place,however, in the composition of the forest veg-etation, as Fagus became more dominant, par-ticularly at the beginning of the phase, whilePinus declined. The sudden but impressiveincrease of Fagus indicates an expansion overlarge areas and implies the replacement ofpines by beeches, perhaps partly because ofselective cutting of pine trees by humans.Thereafter Fagus remained dominant despiteincreased impact by humans and livestock(increased pollen values for cereals, Chenopo-diaceae, Plantago, Juniperus and so on). Theobserved recovery in Pinus during the phasemay be attributed, at least in part, to its re-colonization of areas previously devoid ofvegetation, perhaps because of fire at thebeginning of the phase (implied by the sud-den increase in Fabaceae and Artemisia).Perhaps in the first half of the twentieth cen-tury AD, Fagus declines and there is anincrease of indicators such as cereals, Juniperusand Oxyria-type. This suggests that humaninterference in the forest, direct or indirect,became particularly intense during this period


of major historical events, including theBalkan wars, liberation from the Turks, twoworld wars, German occupation and civilwar. A renewed but limited increase in Fagus,observed in the surface sample from the pro-file, reflects what is apparent in the Lailias for-est today—the expansion of beech as a resultof protective measures (particularly restric-tions on grazing) taken a few decades ago.

Conclusion

The analysis of the Lailias pollen diagramindicates that the evolution of the forest vege-tation of the region has a complex history. Theforest vegetation of the Lailias mountainregion, which is evidently still evolving, is theend-product of a long period of development,during which a dominant influence has beenexercised by humans. Comparison with otherpollen diagrams from the mountains of north-ern Greece reveals common trends in the evo-lution of vegetation under anthropogenicinfluence, but also local differences due to con-trasting environmental conditions and, espe-cially, to the divergent history of each area.

Note

1. The classification of Greek vegetation followsAthanasiadis (1986) and Dafis (1975).

Bibliography

Athanasiadis, N.1975 Zur postglazialen Vegetationsentwicklung von

Litochoro Katerinis und Pertouli Trikalon(Griechenland). Flora 164: 99-132.

1986 Dasiki Fitokoinoniologia. Thessaloniki: Giachoudi-Giapouli.

Athanasiadis, N. and A. Gerasimidis1986 Metapagetodis exelixi tis vlastisis sto Vora

Almopias (in Greek with German summary).Scientific Annals of the Department of Forestry andNatural Environment, Thessaloniki 29,4: 211-49.

1987 Metapagetodis exelixi tis vlastisis sto OrosPaikon (in Greek with German summary).Scientific Annals of the Department of Forestry andNatural Environment, Thessaloniki 30,11: 403-45.

Athanasiadis, N., A. Gerasimidis, E. Eleftheriadou and K.Theodoropoulos

1993 Zur postglazialen Vegetationsentwicklung desRhodopi-Gebirges (Elatia Dramas: Griechen-land). Dissertationes Botanicae 196: 427-37.

Bottema, S.1974 Late Quaternary Vegetation History of Northwest-

ern Greece. Doctoral dissertation, University ofGroningen.

Bozilova, E., S. Tomkov and T. Popova1995 Forest clearance, land use and human occupa-

tion during the Roman colonization in Bulgaria.Palaeoclimate Research 10: 37-44.

Dafis, S.1975 Vegetationsgliederung Griechenlands. Veroff-

entlichung Geobotanische Institut ETH, StiftungRiibel, Zurich 55: 23-36.

Gerasimidis, A.1985 Stathmologikes sinthikes kai metapagetodis

exelixi tis vlastisis sta dasi Lailia Serron kaiKatafigiou Pierion (in Greek with German sum-mary). Scientific Annals of the Department ofForestry and Natural Environment, Thessaloniki 26-27:1-133.

1995 Anthropogenis epidrasis stin exelixi tis dasikisvlastisis stin Ellada: stoikhia apo diagrammatagiris (in Greek with English summary). ScientificAnnals of the Department of Forestry and NaturalEnvironment, Thessaloniki 38,1:170-203.

1996 Parelthon kai paron ton emfaniseon tis simidas(Betula) stin Ellada (in Greek with English sum-mary). Proceedings of Sixth Botanical ScientificConference, Hellenic Botanical Society and Biolo-gical Society of Cyprus, Paralimni, Cyprus: 126-32.Hellenic Botanical Society and BiologicalSociety of Cyprus.

Gerasimidis, A. and N. Athanasiadis1995 Woodland history of northern Greece from the

mid Holocene to recent time based on evidencefrom peat pollen profiles. Vegetation History andArchaeobotany 4:109-16.

Nastos, K.1971 Katafigi Pierion Kozanis. Thessaloniki.

Sakellariou, M.B. (ed.)1982 Makedonia. Athens: Ekdotiki Athinon.

Willis, K.J.1992 The Late Quaternary vegetational history of

northwest Greece, 2: Rezina Marsh. NewPhytologist 121:119-38.

Local Vegetation and Charcoal Analysis: AnExample from Two Late Neolithic Sites inNorthern Greece

Maria Ntinou and Ernestina Badal

Introduction

The study of past vegetation records is a pri-mary concern of palaeoecology, as a means ofreconstructing prevailing biological and cli-matic conditions and of investigating changesin the environment caused by natural factorsor anthropogenic activities.

The pollen record of Greece includes longsequences and detailed information for theLate Quaternary. The Early Holocene naturalenvironment was the setting for the establish-ment of Neolithic farming communities, whenchanges in human activity initiated modifica-tion of the landscape and vegetation. Thissubject has been explored by various authorsand has triggered discussion of criteria for theidentification of human impact on vegetation,its timing and intensity. Pollen investigationin Greece offers a clear framework in thisrespect. Neolithic environments, vegetationand climate have been reconstructed in detailthrough pollen sequences in lakes and marsh-es, especially in the most humid areas of northand north-west Greece.

A complementary source of information onvegetation and natural environments is theanalysis of charcoal from archaeological sites.Charcoal is the result of daily, or periodic,domestic activities. Wood gathered for fueland burned during the occupation of a settle-

ment, then deposited in the archaeologicallayers, is the basis for palaeoecological inter-pretation (Stieber 1967; Vernet 1973). Charcoalretains the anatomical structure of wood.Charcoal structures are compared to thosedescribed in the atlases of wood anatomy(Jacquiot 1955; Jacquiot et al. 1973; Greguss1955, 1959; Schweingruber 1978, 1990), thuspermitting identification to family, genus orspecies level. In charcoal diagrams, compari-son of the quantitative and qualitative compo-sition of consecutive spectra reflects local veg-etation and human influence thereon and, inlong sequences, changes caused by naturalfactors or anthropogenic activity (Badal 1992;Heinz 1991; Chabal 1988,1997).

The results of charcoal analysis at two LateNeolithic sites in northern Greece, Makri inThrace and Dispilio in West Macedonia, arepresented here and compared to the pollenrecord of the same period. The way in whichthe complementary results of these two disci-plines can be combined in palaeoecologicalinterpretation is discussed in the context ofthese two sites. Charcoal analysis may shedlight on local conditions and on human per-ception of vegetation. It may reveal how nat-ural vegetation was managed, which (at leastfor the Neolithic) may have involved otherprocesses than just clearance, deforestation anddegradation. Besides their ecological value,

3

Local Vegetation and Charcoal Analysis 39

such results enrich our knowledge of multi-dimensional prehistoric subsistence strategies.

The Holocene Pollen Record in North andNorth-West Greece

The pollen record of the Holocene in Greecehas been investigated most intensively innorth and north-west Greece, in part becausewetter climatic conditions are more favour-

able than in the south of the country to thepreservation of sediments for palynologicalinvestigation. The following brief summary isbased on the pollen diagrams from TenaghiPhillipon (Wijimstra 1969), loannina, Edessa,Khimaditis, Kastoria, Giannitsa (Bottema1974), Gramousti, Rezina (Willis 1992a-c),Tseravinas and Ziros (Turner and Sanchez-Goni 1997) (Fig. 3.1) and on a synthesis of thevegetation history of the Balkans by Willis(1994a).

Figure 3.1 Map of Greece showing the location of pollen coring sites and Late Neolithic settlements mentionedin the text.

40 Maria Ntinou and Ernestina Badal

The expansion of oak and the simultaneousappearance of other species mark the LateGlacial-Holocene transition at the majority ofsites. There is no evidence of the time-trans-gressive appearance of taxa that would indi-cate differential migration rates. The pollenrecord points, therefore, to the existence ofmid-altitude vegetation refugia during thelast glacial in the Greek mountains, fromwhich tree taxa expanded rapidly followingthe amelioration of climatic conditions (Ben-nett et al 1991; Bottema 1974; Willis 1994a).The Holocene vegetation in all coring sitesfrom north, north-west and eastern Greecehas some common characteristics: the estab-lishment of open oak woodland and theincrease in the frequencies of other deciduousspecies. Differences in establishment ratesbetween sites are due to altitudinal and latitu-dinal variation (Willis 1994b).

After this first stage of more or less open oakformations, evidenced by the increase ofPistacia between 9000 and 8000 bp, this sun-loving species declines sharply, denoting theclosing up of the canopy (Bottema 1974;Turner and Sanchez-Goni 1997; Wijmstra1969; Willis 1994a). Up to 7500 bp, other tem-perate deciduous taxa proliferate in the pollendiagrams at the expense of oaks, reflectingchanges in the diversity and density of vege-tation. At higher altitudes, conifers (Abies,Pinus) expand, as evidenced by the Rezina,Edessa and Khimaditis sequences (Bottema1974; Willis 1992b; 1994b). Local climatic dif-ferences, maturation of soils and differingestablishment time for different taxa wereresponsible for these changes in the composi-tion of vegetation (Willis 1994a). From 7500 to5000 bp mixed deciduous oak forests pre-vailed in most sites, with increasing frequen-cies of Carpinus orientalis/ Ostrya; at higherelevations, conifers maintained their position(Bottema 1974).

Between 5000 and 4500 bp all sequences

show the first signs of anthropogenic distur-bance, manifested by reduction in the diversi-ty and density of the forest. After 4000 bp, thisis accompanied by an increase in open groundherbaceous types and the establishment oftree taxa directly or indirectly related tohuman intervention, such as Olea, Fagus,Juglans, Castanea and Platanus.

The Neolithic Site of Makri

Regional and Cultural Setting

The Neolithic site of Makri lies 10 km west ofmodern Alexandropoulis in Thrace, north-east Greece (Fig. 3.1). It is located on top of acliff at 40 m asl, overlooking the Aegean sea tothe south. To the east and west extends theThracian plain. To the north the landscaperises in altitude to the peaks of the Rhodopimountain range (1000 m), that forms the nat-ural border between Greece and Bulgaria.

The cliff on which the Neolithic site is situ-ated corresponds to the distal end of a broad,slightly concave platform of stepped waterfalltravertine deposits (Efstratiou et al. 1998);locally, the different growth phases of traver-tine usually favour swamp formation (Efstra-tiou et al 1998).

The climate and vegetation of the region areinfluenced by the sea to the south and theRhodopi mountains to the north. The climateand precipitation regime is Mediterranean,though continental influences become strongerwith increasing distance from the coast. Fromsea level to an altitude of 500 m, the biocli-matic conditions in the region are of themesomediterranean type—dry, with meanannual precipitation under 500 mm. Wintersare cold (average January temperature 0-5 °C)and summers relatively hot (average July tem-perature 20-25 °C) (Forest Service 1989;Polunin 1980). The fertile plain is intensively


used for agriculture, mainly cotton, tobacco,cereals and occasionally olive trees in thewarmest niches. Near the coast, the vegetationforms a tall maquis dominated by evergreens,Phillyrea media, Quercus ilex, Arbutus unedo andso on. Deciduous vegetation grows as well,represented by Pistacia terebinthus, Fraxinusornus, and Pyrus amygdaliformis. Less frequentare Arbutus andrachne and Carpinus orientalis.From 500 m upwards, the bioclimatic sequencefollows the supramediterranean vegetation ofsubhumid to humid type, characterized bydeciduous oak woodlands with hornbeam,ash, maple and so on. Oromediterraneanbeech forests cover the peaks of the Rhodopimountains (Forest Service 1989).

The excavation of Neolithic deposits at themound of Makri revealed a four-metre deepstratigraphic sequence that is characterized bytwo distinct depositional units. These corre-spond to two different cultural periods, ashort early Makri I and a longer late Makri II,separated by a 'destruction layer'. The potterystudy confirms this division into two periodsthrough changes in the shape and decorationrepertory Each of these two periods includesseveral habitation-architectural phases (Efstra-tiou et al 1998).

The site is ascribed to the Middle and LateNeolithic horizons of the southern Balkans.The three available radiocarbon dates confirmthis. One dates the deeper layers of thesequence (Makri I) to 5540 BC (GrN-20475:6640±50 bp), and the other two indicate a dateof 5500 BC (GrN-21266: 6580±40 bp; GrN-21267: 6560±30 bp) for the transitionaldestruction layer. The ceramic traits relate theMakri II cultural period with the phasesSitagroi I-II and Paradimi I-II in Greece,Karanovo III in Bulgaria, early Hoga-(^esme inEuropean Turkey and Ilipinar IV in Anatolia(Efstratiou et al 1998).

Charcoal Analysis at Makri

Charcoal analysis covers only part of thesequence at Neolithic Makri. For the earlierperiod Makri I, scattered remains of firewoodwere collected from all layers in one trench.For the later Makri II period, only two layersbeneath the second habitation phase weresampled.

The analysis of 1939 charcoal fragments ledto the identification of 20 plant taxa and onethat remained indeterminate. The plant listfrom Makri, in alphabetic order, consists of:Acer sp. (maple), conifer, Cornus sp. (Corne-lian cherry), Ficus carica (fig) (Fig. 3.2 [4]),Fmxinus sp. (ash) (Fig. 3.2 [5]), Juglans regia(walnut), Juniperus sp. (juniper), Legumino-sae, Maloideae, Paliurus spina-christi (Christ'sthorn), Phillyrea-Rhamnus (buckthorn), Pistaciaterebinthus (terebinth), Prunus sp., Rosaceae,Quercus deciduous (deciduous oak), Salix-Populus, Taxus baccata (yew), Tilia sp. (lime),Ulmus sp. (elm) and Vitis vinifera var. sylvestris(grape vine) (Fig. 3.2 [6]). In the case of Prunussp., we have included a few charcoal fragmentsattributable to Prunus amygdalus (almond).

The lignifying angiosperms are representedby 15 plant families including various genera,all deciduous except for Phillyrea-Rhamnus.Anatomical differentiation between the latteris not possible, so it must be noted thatPhillyrea genus, Rhamnus alaternus and R.lycioides are Mediterranean evergreens whileRhamnus alpinus and R. sibthorpianus aredeciduous mountain species. Conifers aregenerally scarce, including only Juniperus sp.and Taxus baccata.

Because of the scarcity of charcoal, the datahave necessarily been analysed independent-ly of the thickness of the archaeological layers.Some layers did not contain sufficient char-coal fragments for analysis and, for this rea-son, samples had to be amalgamated in orderto assess the representation of the taxa. Layer


1. Quercus deciduous, transverse section, x60 2. Pinus cf. nigra, transverse section, x50

3. Pinus cf. nigra, longitudinal radial section, x790 4. Ficus carica, transverse section, x!70

5. Fraxinus sp., transverse section, x40 6. Vitis vinifera var. sylvestris, transverse section, x50

Figure 3.2 SEM images of archaeological charcoal from Late Neolithic Dispilio (1-3) and Makri (4-6).


29 of Makri I and layers 1 and 2 of Makri IIwere relatively rich in charcoal. Charcoal fromlayer 29 has been dated to 5540 BC.

The charcoal diagram for Makri (Fig. 3.3)consists of six spectra, four from the earlierMakri I period and two from the later MakriII. Although the floristic composition of thesespectra is quite similar, there are variations inthe frequency of taxa. In all Makri I spectra,the dominant taxon is Quercus deciduous (c.30%), except in layer 29 where Ficus carica ismore abundant. In the Makri II spectra, thedominant taxon is Fraxinus sp., followedclosely by Quercus deciduous.

The charcoal diagram seems to indicate thepresence of deciduous oak woodland, inwhich oaks are accompanied by other decidu-ous taxa such as Fraxinus sp., Cornus, Acer,Maloideae genera and Tilia. Fraxinus couldbelong to the oak woodland or grow in areasof greater soil humidity, like water courses ortravertine swampy formations. Its increasingfrequency in the four lower spectra and pre-dominance in the upper two suggest two pos-sibilities, bearing in mind the limitationscaused by the segmented sequence:

1. Parts of the deciduous oakwoods weremanaged in a way that did not alter theoak forest, as implied by the stable Quercusdeciduous percentages, but opened up thecanopy, so favouring the growth and pro-liferation of the heliophilous manna ash.

2. Swamps near the site, as indicated bytravertine formations, were rich in ash andother resources and so were frequented byhumans as complementary sources to theoakwood.

Cornus and Acer have low but steady fre-quencies. Cornus sp. may have grown in thehedges of humid areas or in the oak wood-lands. Acer has not been identified to species,but this genus is very common in oak forests,as are Taxus baccata and Tilia, which are bothvery rare in the spectra. Maloideae genera aredifficult to differentiate and so the subfamilyname is retained, but Crataegus and Pyrus-Sorbus are probably represented in our sam-ples. Maloidae are more or less stable exceptfor the upper spectrum.

A distinctive feature of this deciduous oakwoodland is the presence of the heliophilous

Figure 3.3 Charcoal diagram for Late Neolithic Makri: percentage frequencies of taxa in the charcoal assemblagesfrom successive layers.


components, Pistacia terebinthus, Prunus sp.and Prunus cf. amygdalus, Juniperus sp.,Paliurus spina-christi, and Pyrus from theMaloideae. These genera demand plenty oflight and do not strictly belong to the bushyformations that prosper under tree cover.Such taxa indicate a plant cover rather open inplaces where sun-loving species could flour-ish. These formations could develop in a dryto subhumid pluviometric regime, a featurethat differentiates these particular deciduousoak woods from their European equivalentsneeding higher precipitation.

Pistacia terebinthus, although well represent-ed in Makri I, diminishes in the spectra ofMakri II. Ficus carica would likewise havebeen quite abundant in the environmentaccording to its frequencies in the Makri Ispectra, but is quite scarce in Makri II. Thewild form of the fig tree grows in theMediterranean region from sea level to 1000m. This deciduous tree, which tolerates drysummers and cold winters well, occupies rockcrevices, gorges and stream sides. In the terri-tory of Neolithic Makri, fig trees probablygrew on the steep slopes overlooking the sea.The anatomical characteristics do not permitdistinction between the wild and domesticat-ed forms. The fig tree was used for firewoodin prehistoric Makri, although it does not pro-vide good quality fuel unless dry. Ethno-graphic studies in Morocco report the occa-sional use of fig wood for firing the base ofpottery kilns (Pena-Chocarro et al. 2000). Inprehistoric sites, specific uses of wood are dif-ficult to define, but short-term or isolatedevents may cause unexpectedly high frequen-cies of taxa, the interpretation of which canonly be speculative. The sampling gapbetween Makri I and II makes it difficult tointerpret the low Ficus carica frequencies in theupper layers. Some protection of these treesfor their highly nutritious fruit is possible andfig seeds, as well as entire carbonized fruits,

are present in the archaeological levels ofMakri II (Efstratiou et al. 1998).

Riverside vegetation is represented at Makriby Ulmus sp., Vitis vinifera, Salix-Populus andJuglans regia.

In the charcoal diagram for Makri, a largepart of the sequence is missing. Nevertheless,the earliest occupation and part of the latestare represented, covering the period betweenthe middle of the sixth millennium BC and thebeginning of the fifth millennium BC. Duringthat time, the surrounding vegetation wasdominated by deciduous oak and in generalremained unaltered. This deciduous forma-tion was probably the climax vegetation of thearea and developed under mesomediter-ranean dry to subhumid climatic conditions.Although the climate was very similar to thatprevailing in the region today, the plant for-mations were drastically different. Nowadays,in the environs of the site, only evergreen for-mations thrive and among these the bioindi-cators of human activities are quite frequent.

Dispilio Lakeside Settlement

The Regional and Cultural Setting

The site of Dispilio, West Macedonia, is a lake-side settlement, located at an elevation of625 m, a few metres from the south shore ofLake Kastoria (Fig. 3.1). The region is moun-tainous, forming part of the North Pindosrange that reaches 2128 m on the Vitsi peak tothe north-east, 2111 m on the Siniatsiko moun-tains to the south-east, and 2520 m on theGrammos to the far south-west on the borderwith Albania. Rivers, some feeding the lakeand others receiving its outflow, dissect thismountainous landscape. The plain to thesouth of the lake is drained by the RiverAliakmon and its tributaries and is surround-ed by hills.


The climate and precipitation regime of theregion is transitional from Mediterranean tocontinental. Rainfall is more or less evenly dis-tributed through the year, the dry period isshort and mean annual precipitation rangesfrom 700 to 1000 mm. The annual temperatureamplitude is approximately 20 °C, while snowand frost are frequent in winter (Bottema1974; Ntafis et al 1997; Polunin 1980).

The natural vegetation at lower elevationsand close to the lake has almost completelydisappeared. The plain supports intensivearboriculture. Around the lake, Phragmitesaustmlis is dominant (Ntafis et al. 1997). Salix,Populus and Ulmus are the most frequent treesclose to the lake, while Pyrus amygdaliformis,Paliurus spina-christi and Prunus spinosa thriveon the limestone hills to the south-west. To thesouth-east, the hills present a stage of regener-ation where Juniper us oxycedrus is frequent,together with Carpinus orientalis, Fraxinusornus, Ostrya carpinifolia and some Quercus cer-ris. The potential vegetation under natural cir-cumstances would be a deciduous forest ofsupramediterranean bioclimatic conditions(Bottema 1974). The mountains are covered byPinus nigra forests with Abies borisii-regis andFagus woods (Polunin 1980).

The deposits excavated at the Dispilio lake-side settlement offered an approximately two-metre stratigraphic sequence and plentifulinformation on various aspects of the prehis-toric occupation of the site. The earlier occu-pation phases represent a settlement built overshallow water or swampy ground which driedup periodically. The later occupation phasewas built on dry ground, though probablyaffected periodically by lake waters (Karkanasin preparation). The Dispilio sequence canlargely be assigned to the Late Neolithic, butthe pottery study offers evidence for activitydating back to the Middle Neolithic andextending down to the Early Chalcolithic(Andreou et al. 1996; Hourmouziadis 1996).

Charcoal Analysis

At the lakeside settlement of Dispilio, theentire stratigraphic sequence was sampled invarious trenches. The charcoal diagram isbased, however, on samples from two trench-es which contained scattered charcoal, in con-trast to the other trenches where evidentremains of architectural timber complicatedpalaeoecological interpretation. Nonetheless,the nature of the waterlogged deposits, wheredifferentiation between charcoal from archi-tectural remains and fuel is difficult, makes itnecessary to consider the possibility that atleast some of the samples analysed includemixed material.

The analysis of 3743 charcoal fragments ledto the identification of 21 plant taxa. The plantlist of Dispilio, in alphabetical order, consistsof: cf. Abies sp. (fir), Acer sp. (maple), Alnusglutinosa (alder), Carpinus/Ostrya (horn-beam/hop-hornbeam), Cornus sp. (Corneliancherry), Corylus sp. (hazel), Fraxinus sp. (ash),Hedera helix (ivy), Juniperus sp. (juniper),Maloideae, Pinus cf. nigra (black pine) (Fig. 3.2[2-3]), Pistacia terebinthus (terebinth), Prunuscf. amygdalus (almond), Prunus sp., Quercusdeciduous (oak) (Figure 3.2 [1]), Rhus coriaria(sumach), Rosa sp. (rose), Rosaceae, Salix sp.(willow), Tilia sp. (lime) and Ulmus sp. (elm).This plant list almost certainly underestimatesthe diversity of species exploited, given thatidentifications to genus may include severalspecies, as for example in the case of Quercusdeciduous. In this plant list, the majority ofthe taxa identified (18) are deciduous angio-sperms. The remaining three taxa are conifers.

The charcoal diagram for Dispilio (Fig. 3.4)consists of 12 spectra, each corresponding toan approximately 10 cm excavation layer. Thelayers between 11 and 14 with timber remainshave been excluded. The dominant taxon inall spectra is Quercus deciduous, which fluctu-ates around 40-50%. Records of Quercus sp.


Figure 3.4 Charcoal diagram for Late Neolithic Dispilio: percentage frequencies of taxa in the charcoal assemblagesfrom successive layers.

also probably belong to deciduous oak, butare shown separately because the ring porouszone was not clearly observable.

Other deciduous components of oakwoods,like Fraxinus, Acer and Cornus, present low butsteady frequencies. Tilia, Carpinus/ Ostrya andHedera helix are rather infrequent. Frequenciesof Ulmus sp. are low but stable and this taxonmay have formed part of the riverside or lake-side woods. Corylus sp. would be a less fre-quent component of the forest undergrowth.

Maloideae is a big subfamily of the Ros-aceae; anatomical differentiation betweengenera is difficult, so the generic term hasbeen retained. Various members of the Mal-oideae, such as Sorbus, Mains and someCrataegus and Pyrus species, are componentsof deciduous woods. The frequency of Mal-oideae increases in the lower part of thesequence, reaching a maximum in spectrumlib, but thereafter clearly declines.

Pistacia terebinthus is present in decreasingfrequencies in the lower part of the sequenceand disappears from spectrum 9 onwards.Rhus coriaria is present in the two lower spec-tra. Both are sun-loving species growing inopen woods and dry places. Prunus sp. israther sparse. There are various Prunusspecies in northern Greece, the majority ofthem growing in open, sunny and rockyplaces. In our samples, Prunus amygdalus hasbeen differentiated, on the basis of its distinc-tive anatomy (Bazile-Robert 1980; Schwein-gruber 1990).

Lakeside vegetation is evidenced by Salixsp. and Alnus glutinosa. Salix reaches its maxi-mum (16.5%) at the bottom of the sequenceand from then on is rare or absent. Alnus israther infrequent.

Conifers are represented by Pinus cf. nigra,Juniperus sp. and sparsely by Abies sp. Pinuscf. nigra is the next most important taxon after


Quercus deciduous. The percentages in spec-tra 9, 7, and 4 may be rather exaggerated,probably as a result of the admixture of char-coal from construction timber, but its frequen-cies increase steadily from c. 10% in the lowerto c. 20% in the upper part of the diagram.Juniperus sp. occurs throughout the sequencewith variable frequency

The charcoal diagram points to the existenceof two dominant formations, a deciduous oakwoodland and a coniferous forest. Deciduousoak woodland is the prevalent formationthroughout the sequence and, judging fromthe frequencies of taxa, it was of stable com-position.

Black pine forests were the other dominantformation in the region. The increasing fre-quencies of Pinus nigra are interpreted interms of an intensification in human exploita-tion of the pine forest rather than a change invegetation cover, since there is no evidence foralterations to the predominant oak woodland.Juniperus thickets would border the oakwoods or grow in the less dense parts of thedeciduous formations.

Bushy, sun-loving, open formations repre-sented by Pistacia terebinthus, Rhus coriaria,Prunus sp. and probably some Maloideae,such as Pyrus amygdaliformis, are less well rep-resented in the upper part of the sequence.The same occurs with lakeside vegetation. Inour view, these formations would be the clos-est to the site and, during the initial stage ofoccupation, were probably the most affectedby clearance for housing and other subsis-tence activities. If the charcoal record showsthe vegetation used by prehistoric people dur-ing the time of occupation of the site (Badal etal. 1994), then these two formations were notextensive in the first place and, once cleared(and the wood used), either did not regeneratebecause their niche was occupied by humans(open places, fields?) or were ignored becauseother formations were more convenient

(wood gathering would certainly be more effi-cient in oak forests than in restricted gallery-forests along the lake).

Oak forests were probably the climax for-mation for the latitude (40°30'N, 21°18/E) andaltitude (625 m) of the site of Dispilio, whileblack pine forests would cover the surround-ing mountains not far from the site. The pre-vailing climatic conditions would be similarto the present, though the vegetation aroundthe site has changed considerably

Discussion

The Late Neolithic sequences of Dispilio andMakri cover part of the sixth millennium andat least the beginning of the fifth millenniumBC, corresponding to the period from 6500 to6000 bp (Demoule and Perles 1993: 366, fig. 2).The charcoal data from these sites are consis-tent with the pollen record for the same peri-od, when deciduous oakwoods are prevalent,though altitudinal variation is evident in thepresence, distribution and abundance ofmountain conifers and Mediterranean or tem-perate taxa at different locations.

The Late Neolithic vegetation sequence atDispilio presents similarities to the pollenrecord of this region in the Atlantic period(Bottema 1974). Zone Y in the Khimaditis andKastoria pollen diagrams points to the occur-rence, at medium elevations, of a dense decid-uous oak forest mixed with Ulmus, Fmxinus,Tilia, Carpinus and Corylus. Pine forests withsome fir would cover the mountain zone.

The pollen diagram from Tenaghi Philliponindicates the gradual expansion of forests inthe Holocene (zone Z). The oak is attributed toevergreen oak and there is reference to theexistence of a Fmxinus zone (Wijmstra 1969).The Tenaghi Phillipon coring site is the closestto Makri and there is agreement between thepollen and charcoal records as regards the


forested nature of the landscape and the rela-tive importance of ash and other heliophilousspecies.

Charcoal analysis at the two Neolithic sitesalso reflects the prevalent local vegetation for-mations. At both sites, oak woodlands consti-tute the main vegetation cover, but the sec-ondary formations reflect regional differencesin climate related to altitude and exposure tothe influence of the sea. In the case of Dispilio,the coniferous forest, mainly of black pineswith some fir, reflects the location of the site at625 m altitude in an inland basin, surroundedby high mountains. At Makri, by contrast,abundant sun-loving Mediterranean speciesreflect the low elevation (40 m) and proximityto the sea. The deciduous oak woodlandseems denser around Dispilio, where oakmakes up 40-50% throughout the sequence;other components are sparse, including ashthat does not exceed 6%. At Makri, on theother hand, the oakwood would have beenless dense, with ash a well-represented com-ponent and other Mediterranean species con-stantly present.

Despite these differences resulting fromlocation and altitude, both sequences are char-acterized by the unaltered quantitative andqualitative composition of the principal decid-uous oak formations. This implies minimalhuman impact on the local environment. Inthis respect, the pollen record too does notdetect anthropogenic disturbance and landdegradation until after 5000-4500 bp (Bottema1974; Willis 1995). Even though the firstNeolithic farming communities were estab-lished a few thousand years earlier, theirimpact on the landscape seems to have beennegligible. Oakwoods were predominant andchanges in the composition of the woodlandwere principally caused by competitionbetween species and changing climate ratherthan anthropogenic intervention (Willis1994a). In the pollen record, land degradation

and deforestation, accompanied by evidencefor agriculture and grazing, post-dates theNeolithic and the charcoal results from thetwo Neolithic sites provide corroborative evi-dence of stability in local vegetation cover.

In interpreting the charcoal data from theNeolithic sites, some issues deserve furtherconsideration. First, since the charcoal spectrareflect local conditions, these are likely to beinfluenced by the duration and intensity ofoccupation at the sites: occupation of shorteror longer duration, with or without interrup-tion, would probably have different effects onthe local environment. The charcoal sequencesare attributed to the Late Neolithic, but theexisting evidence is insufficient for precisedating. At Makri, a very short duration is sug-gested for the earliest Makri I period by theradiocarbon dates (5540 BC for the bottom and5500 BC for the top of this part of thesequence), while the rest of the sequencewould correspond to some time at the begin-ning of the fifth millennium BC. For Dispilio,no radiocarbon dates are as yet available andthe intensity of occupation is a subject stillunder study; more archaeological informationis needed. Assuming a 500-year occupationspan for these sites, however, which is consis-tent with existing information, the transfor-mation of the vegetation is minimal and thepollen record points in the same direction. Inthis respect, the climatic and geographicalcharacteristics of the two regions may playsome role. Higher precipitation at inlandDispilio would favour regeneration of thevegetation, while the gentle topography atMakri would be less vulnerable to erosion. Inthe pollen record, high-altitude sites in roughterrain show signs of degradation earlier thanlowland sites (Willis 1995).

Secondly, anthropogenic factors should beconsidered: the nature of subsistence strate-gies and techniques and the ways in whichnatural vegetal resources were exploited. The


stability of vegetation cover may indicate bal-anced exploitation of the natural vegetation.Some kind of forest management may havebeen practised, which did not result in clear-ance or deforestation but rather maintainedthe natural equilibrium; it seems that agricul-ture and animal husbandry caused little mod-ification. To what extent this is true, andwhether it reflects stable farming and grazingmethods or repeated use of the same pieces ofland and thus did not affect forest cover, is amatter for investigation in each individualcase and by the relevant disciplines. Tech-niques of land use are another important fac-tor and it is has been argued that, only duringthe Bronze Age, did innovations in this areabecome a serious agent of landscape change(Sherratt 1981).

Clearly, there is an extensive pollen recordfor Greece but charcoal data from archaeolog-ical sites are very restricted. As palaeoenvi-ronmental studies on prehistoric sites becomethe norm, it would be worthwhile to conductcharcoal analysis more widely, as it is comple-mentary to palynology and has the advantageof reflecting local environments directly relat-ed to human activities. Questions of vegeta-tion change, degradation and human impactwould be illuminated by systematic charcoalanalysis at sites over a broader temporal andgeographical range.

Conclusions

1. Charcoal analysis at two Neolithic sites innorthern Greece offers a picture of localvegetation at the time of occupation of thesites. Deciduous oak formations are pre-dominant at both, but secondary forma-tions reflect the altitudinal, geographicaland, consequently, climatic differencesbetween the two sites.

2. The results of on-site charcoal analysis arecomplementary to the broader off-sitepollen record for the Holocene in northernGreece.

3. The evidence of pollen and charcoal pointsto minimal human influence on vegetationduring the Late Neolithic.

4. The unchanging nature of the vegetationthroughout the sequence at both sites mayreflect:• sustainable methods of forest manage-

ment;• practices and techniques of farming and

grazing that would not cause large-scaledisturbance of vegetation cover;

• the duration and intensity of occupation(issues requiring further investigation).

5. It is necessary to broaden the charcoalrecord with results from other sites in thesame and other regions, and in earlier andlater periods, in order to achieve a moreprecise understanding of local vegetationand of the impact on this of human activi-ties.

Bibliography

Andreou, S., M. Fotiadis and K. Kotsakis1996 Review of Aegean prehistory 5: the Neolithic

and Bronze Age of northern Greece. A]A 100:537-97.

Badal, E.1992 L'anthracologie prehistorique: a propos de cer-

tains problemes methodologiques. bulletin de laSociete botanique de France 139: Actualitesbotanic/ties 2-4:167-89.

Badal, E., J. Bernabeu and L. Vernet1994 Vegetation changes and human action from the

Neolithic to the Bronze Age (7000-4000 bp) inAlicante, Spain, based on charcoal analysis.Vegetation History and Archaeobotany 3: 155-66.

Bazile-Robert, E.1980 Les groupements a Amygdalus et Prunus de la

fin du Tardiglaciaire et du debut du Post-glaciaire en mediterranee occidentale. Geobios13: 777-81.


Bennett, K.D., RC. Tzedakis and K.J. Willis1991 Quaternary refugia of north European trees.

Journal of Biogeography 18:103-105.Bottema, S.

1974 Late Quaternary Vegetational History of NorthwestGreece. Doctoral thesis, University of Gronin-gen.

Chabal, L.1988 Pourquoi et comment prelever les charbons de

bois pour la periode antique: les methodes util-isees sur le site de Lattes (Herault). Lattara 1:187-222.

Chabal, L.1997 Forets et societes en Languedoc (Neolithique final

Antiquite tardive). L'anthracologie, methode etpaleoecologie. Documents d'Archeologie Fran-c,aise 63. Paris: Editions de la Maison des sci-ences de Thomme.

Demoule, J-P. and C. Perles1993 The Greek Neolithic: a new review. Journal of

World Prehistory 7: 355-416.Efstratiou, N., M.P. Fumanal, C. Ferrer et al.

1998 Excavations at the Neolithic settlement ofMakri, Thrace, Greece (1988-1996): a prelimi-nary report. Saguntum 31:11-62.

Forest Service1989 Forest Maps of Greece, 1:200,000. Athens: Ministry

of Agriculture.Greguss, P.

1955 Xylotomische Bestimmung der heute lebendenGymnospermen. Budapest: Akademiai Kiado.

1959 Holzanatomie der Europai'schen Laubholzer undStraucher. Budapest: Akademiai Kiado.

Heinz, C.1991 Upper Pleistocene and Holocene vegetation in

the south of France and Andorra. Adaptationsand first ruptures: new charcoal analysis data.Review of Palaeobotany and Palynology 69: 299-324.

Hourmouziadis, G.1996 Dispilio (Kastoria), the Prehistoric Lakeside

Settlement. Thessaloniki: Codex.Jacquiot, C.

1955 Atlas d'anatomie des bois de coniferes. Paris: Centretechnique du bois.

Jacquiot, C., Y. Trenard and D. Dirol1973 Atlas d'anatomie des bois des angiospermes. Paris:

Centre techniques du bois.Karkanas, P.

Mikromorfologiki meleti anaskafis Dispilio.Unpublished manuscript (in Greek), in prepara-tion.

Ntafis, S., E. Papastergiadou, K. Georgiou et al.1997 Odigia 92/43/EOK. To Ergo Oikotopon stin Ellada:

Diktio FISI. EC contract no. B4-3200784/756.Athens: Mousio Goulandri Fisikis Istorias,Elliniko Kentro Viotopon Igrotopon.

Pena-Chocarro, L., L. Zapata Pena, J.E. Gonzalez Urquijoand J.J. Ibanez Estevez

2000 Agricultura, alimentacion y uso del com-bustible: aplicacion de modelos etnograficos enarqueobotanica. In Saguntum-Plau. III. Reuniosobre Sconomia en el Mon Iberic., 3:403-20.Valencia: Department of Prehistory and Arch-aeology, University of Valencia.

Polunin, O.1980 Flowers of Greece and the Balkans: A Field Guide.

Oxford: Oxford University Press.Schweingruber, F.H.

1978 Mikroskopische Holzanatomie. Teufen: F. Fliick-Wirth.

1990 Anatomic Europaischer Holzer. Bern and Stuttgart:Haupt.

Sherratt, A.1981 Plough and pastoralism. In I. Hodder, G. Isaac

and N. Hammond (eds.), Pattern of the Past:Studies in Honour of David Clarke: 261-305.Cambridge: Cambridge University Press.

Stieber J.1967 A Magyarorszagi Felsopleisztocen vegetatio-

tortenete az anthrakotomiai eredmenyek (1957IG) Tiikreben. Foldtani Kozlony 97: 308-17.

Turner, C. and M.-F. Sanchez-Goni1997 Late glacial landscape and vegetation in Epirus.

In G. Bailey (ed.), Klithi: Palaeolithic Settlementand Quaternary Landscapes in Northwest Greece. II.Klithi in its Local and Regional Setting: 559-85.Cambridge: McDonald Institute for Archaeolo-gical Research.

Vernet J.-L.1973 Etude sur 1'histoire de la vegetation du Sud-Est

de la France au Quaternaire d'apres Tetude descharbons de bois principalement. Paleobiologiecontinentale 4 (1): 1-90.

Wijmstra, T.A.1969 Palynology of the first 30 metres of a 120 m deep

section in northern Greece. Acta BotanicaNeerlandica 18: 511-27.

Willis, K.J.1992a The late Quaternary vegetational history of

northwest Greece. I. Lake Gramousti. NewPhytologist 121:101-17.

1992b The late Quaternary vegetational history ofnorthwest Greece. II. Rezina Marsh. NewPhytologist 121:119-38.


1992c The late Quaternary vegetational history ofnorthwest Greece. III. A comparative study oftwo contrasting sites. New Phytologist 121: 139-55.

1994a The vegetational history of the Balkans. Quater-nary Science Reviews 13: 769-88.

1994b Altitudinal variation in the late Quaternary veg-etational history of northwest Greece. HistoricalBiology 9: 103-16.

1995 Land degradation in the Balkans: variations intime and space. In L'homme et la degradation deI'environnement. XVe recontres Internationalesd'archeologie et histoire d'Antibes: 161-74. Juan-les-Pins: Editions APDCA.

Holocene Climate Change in Crete: AnArchaeologists View

Jennifer Moody

Major variations in climate such as the waxingand waning of glacial epochs have long beenrecognized as having had a profound influ-ence on the organization of human societies(Sherratt 1997), but lesser fluctuations, such asthe 'Medieval Little Ice Age7, have received lit-tle attention from Mediterranean archaeolo-gists until recently. During the last ten yearsincreasingly refined climate proxy data havebecome available (such as the GIPS2 core),demonstrating that the Holocene climate, aswe know it today, is not typical of the last10,000 years. In fact there have been at leastfive substantial 'blips' in this interstadial'sjourney to the present-day climate. One suchanomaly, occurring c. 2200 BC, has recentlyreceived much attention in the Near East andEgypt (Dalfes et al 1997).

In the Aegean, however, the views of vanAndel et al (1986) and Zangger (1994), whointerpret most geomorphological eventsoccurring since the beginning of farming (c.7000 BC) as anthropogenic, continue to prevail.Although not perhaps as marginal for agricul-ture as the Near East, parts of the Aegean,where near desert conditions occur naturally,can be considered precarious for the prehis-toric farmer. Such environments prevail in thesouth and east, especially on the lee side ofmountains where strong rain-shadow effectscome into play. Today, winter rains come to

Crete from the north-west, creating strongrain shadows to the south and east of theisland's 1500-2500 m mountain backbone.These areas are the most vulnerable todrought. Similarly, the north-west of theisland is more vulnerable to killing frosts andsnow damage if colder temperatures were tobecome a concern. In 1991-92, as part of theEC project 'Crete and the Aegean Islands:Effects of Changing Climate on the Environ-ment',1 Dick and Jean Grove, Oliver Rackhamand the author observed substantial frostdamage and limb breakage to olives in theWhite Mountains of western Crete as lowdown as 500 m. This weather event depressedolive production in the hard-hit parts of westCrete for about four years and prompted thereplacement of many ancient olive trees withyounger and more 'fashionable' stock. Simi-larly, in 1999 a withering series of sirrocosblasted Crete from the south-west in earlyMay, shrivelling olive flowers and effectivelywiping out the olive crop for the year in thehard-hit areas. Such scenarios are part of thewild annual variation that is characteristic ofCrete today, and modern farmers are welladapted to it. But what if the winter rains didnot come for several years in a row? What ifspring and autumn sirrocos increased inseverity and frequency? What if low-lyingfrosts and snow increased in frequency?

4

Holocene Climate Change in Crete 53

Would the farming system in place today beable to cope?

The Medieval Little Ice Age

Although these questions may seem far-fetched at the beginning of the twenty-firstcentury AD, such weather seems to have beenthe norm in Crete during the sixteenth, seven-teenth and eighteenth centuries AD, accordingto the archival studies of Grove and Conterio(1994; 1995). Numerous letters from the Dukesof Crete to the authorities in Venice describedevastating weather anomalies which des-troyed crops over and over again. For exam-ple, in January 1595 'the continuous rain andsnow, going on for 3 months...have beenexcessive, and so the low parts of the countryhave suffered from frost' (quoted in Groveand Conterio 1995: 239). Or, as Corner wroteon 2 February 1645:

The need for building material is now greatly

increased due to the conspicuous damage to and

collapse of buildings which took place in every

piazza, following the extraordinary violence of the

rain and wind...at Suda, Gravusse, and Spina-

longa... all the buildings are greatly damaged. At

Gerapetra and in other places many dwellings

were damaged and in the fields the sowing lost, the

trees uprooted, and the animals swept away (quot-

ed in Grove and Conterio 1995: 242).

Grove and Conterio (1994; 1995: 224) con-vincingly compile evidence for increased'winter and spring drought, exceptionallysevere winters, and summer rain' from 1548 to1715 (see Grove and Conterio 1995: figs. 8,10).These centuries correspond to the middle andlatter part of the Medieval Little Ice Age (LIA)in Europe, an increasingly well-documentedglobal climate event (Grove 1997). It seemslikely that the anomalous weather in Crete

was the result of perturbations in the prevail-ing westerlies—which normally bring winterrains to the Eastern Mediterranean—due tothe developing Little Ice Age climate inEurope. It is probable that this less predictableclimate regime began as early as the late thir-teenth century AD in Crete, with the globalonset of the Medieval Little Ice Age.

In addition to the archival data compiled byGrove and Conterio for significant impact ofthe Medieval Little Ice Age on Crete, there is alist of perennial rivers dating to 1625 AD(Rackham and Moody 1996: 44, n. 13). This listincludes 25 rivers, some 18 more than reachthe sea today (see Rackham and Moody 1996:40, fig. 4.3). Although it is possible that thereduction in year-round rivers is a result ofover irrigation, it seems more likely to berelated to increased evaporation due to theshift from Medieval Little Ice Age to the mod-ern climate.

Climate proxy data also help round out ourpicture of medieval climate in Crete. At themoment, such data consist of a pollen core, ahandful of geomorphologic studies, and per-sonal field observations.

The Asi Gonia B Core

This core (see Atherden and Hall 1999: 188,fig. 3) dates from c. 1100-1985 AD, and shouldreflect the transitions from the 'MedievalWarm Period' to the 'Medieval Little Ice Age'to the present-day climate. Atherden and Hallattribute most of the pollen changes visible inthe core to anthropogenic factors. Although Ido not disagree with their views for the mostpart, there are a few changes which seem tocorrespond to the climate regime described byGrove and Conterio.

For example the marked increase in Arbutus(probably Arbutus unedo) and Hedera (ivy),both considered moisture-loving plants in

54 Jennifer Moody

Crete today, suggests decreased evaporationor increased precipitation, which would beconsistent with the climate regime describedby Grove and Conterio. Most telling, however,is the olive pollen profile. Olive pollen is rela-tively abundant at the bottom and top of theAsi Gonia B core, but between 170 and 30 cmit is present in only trace amounts—except at55 cm where it resurges and then disappearsagain. Olive is a frost-intolerant tree. As notedabove, a single year with low-lying frost candevastate the olive crop. Imagine what aseries of frost years could do. Frosts alternat-ing with spring sirrocos would be equallydamaging—and possibly more likely—giventhe Grove and Conterio reconstruction. Theslight peak in olive pollen in the early eigh-teenth century may have been associated witha series of more favourable years, suggestingthat whatever weather was depressing oliveflowering, it was not severe enough to actual-ly kill the trees. This is an important point,since by 1730 Crete was exporting 1600 tons ofoil a year to Marseilles (Triandafyllidou-Baladie 1988).

Buried Chapels2

Over the years, it has become apparent thatnumerous Venetian and some Byzantinechapels on Crete are buried by sediments,some a metre or more in depth. In many cases200- to 400-year old trees (usually olives orcarobs) are growing on the surface of thedeposits burying the chapels. This allows usto determine the timing of the depositionalsequence within a century or two, squeezed inbetween the construction of the chapel andthe germination of the tree.

Figure 4.1 shows the chapel of AghiosNikitas in the Frangokastello Plain, south-west Crete, buried by about 50 cm of a greyishcoarse material containing Greco-Roman and

Byzantine sherds. The sediment is capped bya carob about 1 m in diameter. Exposed annu-al rings suggest an average growth rate ofabout 2 mm per year, resulting in a tree age ofc. 250 years. This indicates that the greydeposit has to have been laid down betweenthe construction of the chapel in the fifteenthcentury AD and the germination of the carobtree c. 1750 AD, that is, the middle of theMedieval Little Ice Age.

Figure 4.2 shows a similar, roughly contem-porary event, but in an entirely differenttopography. The chapel of Sotiris o Christos(1358 AD) is in the valley of the Aghia IriniRiver about 1.5 km west of the entrance to the

Figure 4.1 The chapel of Aghios Nikitas, Frangokast-ello plain, south-west Crete: the foundationsare buried by a deposit containing Greco-Roman and Byzantine sherds.

Figure 4.2 The chapel of Sotiris o Christos, south-westCrete: olive trees, c. 275-300 years old, sealthe deposit which has buried the founda-tions of the chapel.


Aghia Irini Gorge in Selinou, south-westCrete. It is buried by about 80 cm of sediment.Again, this sediment is capped by olive treesranging in diameter from 1.1 to 1.2 m. Exposedannual rings suggest a maximum growth rateof 2 mm per year, resulting in a tree age of275-300 years. This places the alluviationevent between the construction of the chapelin the fourteenth century AD and the ger-mination of the trees, around 1700 to 1775 AD.

These are not isolated events. BuriedVenetian chapels can be found in the moun-tains and plains, in the west, east, north andsouth of the island.

Flash-Floods and Debris Flows

A horizon of flood events and debris flowsalso seems to be contemporary with theMedieval Little Ice Age in Crete (Fig. 4.3). Forexample, in 1988 a horizon of medieval pot-tery—including a sgraffito ware sherd datingto the fifteenth-sixteenth century AD3—wasfound in a bulldozed channel in the Istronriver floodplain. It is buried by 1.7 m of very

coarse channel deposits and capped by anolive c. 400-500 years old (Fig. 4.4). Sand-wiched between the tree and the sherds, theflood deposits must date from 1400-1600 AD.

At Stomion in western Crete, the currentriver channel is filled with coarse gravel andcobble deposits about 3 m in depth, nowretrenched. Near the base of these deposits isa horizon of Roman to Byzantine sherds (Fig.4.5), indicating that at least 2 m of the deposithas accumulated since the Byzantine epoch (c.800-1200 AD). Upstream from the locale withthe sherds, this flood sequence is overgrownwith pine trees. The oldest trees were coredand found to be c. 100-150 years old, datingthis striking series of floods to between 1200and 1850 AD.

A recent study in the Omalos Plain (1150 m)indicates that the intensification of floodevents in Crete during the Medieval Little IceAge was not confined to coastal plains andriver systems. The south side of the OmalosPlain is dominated by a huge alluvial fanwhich probably originated in the MiddlePleistocene (Hempel 1991: 57-60). In recenttimes it has been retrenched and heavily

EM/MMI

Late Minoan

Greek-Roman

Venetian-Turkish

Figure 4.3 Map showing flash-flood and mudflow deposits. The majority of events seems to date to the Middle andLate Bronze Age or to the Venetian and Turkish periods, both periods of global glacial advance. The con-centration of flood and flow events in the west is probably the result of uneven fieldwork.

56 Jennifer Moody

Figure 4.4 Section in Istron river floodplain: coarsechannel deposits overlie a fifteenth-sixteenth century AD sgraffito ware sherd(indicated by the author).

Figure 4.5 Section at Stomion, western Crete: thecoarse deposits overlie a horizon (indicatedby arrows) with Roman to Byzantine sherds.

Figure 4.6 The Omalos fan: two pollarded pear trees(centre of photograph) overlie debris fromeighteenth-nineteenth century AD floodevents.

reworked (Maas et al 1998: 160, fig. 12.4).Radiocarbon dates the recent series of flashfloods to the seventeenth and eighteenth cen-tury AD. Two pollarded pear trees survive onan island in the middle of the retrenchedchannel. They have diameters of 80 cm and anaverage growth rate of 2 mm, indicating thatthey are about 200 years old (Fig. 4.6). Thisindicates that the Omalos flood events tookplace between the eighteenth and nineteenthcentury AD, or in the latter part of theMedieval Little Ice Age.

Summary

The evidence clearly indicates that theMedieval Little Ice Age impacted on theCretan landscape by burying earlier surfaceswith debris flow and flash-flood deposits. Theeconomic effects of these events is amply illus-trated in the archival work of Grove andConterio: loss of crops, soil, animals; damageto buildings; inability to pay taxes. The severedroughts which also plagued Crete during theMedieval Little Ice Age are harder to demon-strate geomorphologically and the pollen dataare not sensitive enough to pick them up, butperhaps with new, developing technologieswe will be able to document this aspect of theMedieval LIA climate in Crete in the future.

A Minoan Little Ice Age?

An intriguing result of the study of medievalflood deposits has been the discovery of anearlier flood and debris flow horizon. Thishorizon largely dates between the earlyMiddle Bronze Age and the Late Bronze IIIperiod, and can be found from one end of theisland to the other (see Fig. 4.3). For example,while working on the EC project ThreatenedMediterranean Landscapes: West Crete',4


channel deposits exposed in a road cut imme-diately south of the National Highway andthe village of Nerokourou in the Khania Plainwere noted but no datable pottery was found(Fig. 4.7). Subsequent examination of thescarp found Middle to Late Minoan I sherdsmixed in with the gravels and cobbles. Thetop of the cobble deposit is 60-80 cm belowthe surface. Similar deposits containingMinoan sherds can also be seen in recent roadcuts north of Mournies and north and south ofthe National Highway Roman to Turkish peri-od sherds litter the surface above both thesedeposits, indicating that the flood events haveto have occurred between the LMI and Romanperiods. The absence of any non-Minoan pot-tery in the channel deposits suggests, howev-er, that these were Bronze Age events.

In the Frangokastello Plain, on the southcoast of Crete, a debris flow deposit contain-ing MMIII/LMI pottery was discovered dur-ing the Sphakia Survey5 (Fig. 4.8). This depositis capped by the same grey, coarse sedimentsthat buried the chapel of Aghios Nikitas dis-cussed above (Fig. 4.9). The sedimentarydetails of this section are described by Nemecand Postma (1993: fig. 8).

South-west of the village of Angouseliana(in west-central southern Crete), an LMIIIpithos buried by a flood event was found bythe Aghios Vasilios Valley Survey6 (Fig. 4.10).The section is presently located about 15 mabove the modern river channel at the westend of a soccer field. The Late Minoan floodhorizon contains only Bronze Age sherds andis capped by another flood deposit containingLate Roman and Byzantine sherds. TheByzantine horizon is in turn capped by a six-teenth-seventeenth century AD chapel, whichis not buried. The absence of Greek or Romanpottery in the deposit burying the LMIIIpithos, in spite of nearby sites, suggests thatthis event took place near the end of the LateBronze Age.

Figure 4.7 Section at Nerokourou, showing coarsechannel deposits which contain MM-LMIsherds.

Figure 4.8 Debris flow with possible Minoan wall: thepen points to an LMI cooking pot base.

Figure 4.9 Section in Frangokastello plain, showingcoarse sediments which overlie a debrisflow with MMIII-LMI pottery (see Fig. 4.8)and are capped by the chapel of AghiosNikitas. The chapel is indicated in thebackground.

58 Jennifer Moody

Figure 4.10 Section near Angouseliana, west-centralsouthern Crete, showing two flood deposits.The older flood deposit contains Minoansherds and buries an LMIII pithos (indicat-ed by arrow). The younger flood depositcontains mostly Late Roman and Byzantinepottery and is capped by the sixteenth-sev-enteenth century AD chapel of AghiosGiorgos.

Discussion

Flashfloods are an integral part of theMediterranean environment and on their ownare not remarkable. The Pachyammos delugeof 1986 (Rackham and Moody 1996: 20-22;Moody 1997), where nearly a year's rainfallfell in 36 hours, is rare but not unusual in themodern Cretan climate. What is striking is thedearth of these kinds of deposits for theNeolithic to Early Minoan, and Early Iron Ageto Byzantine periods, especially when com-pared to their frequency in the Middle to LateMinoan and Venetian to Turkish periods. Sofar only two flood deposits dating to theArchaic-Hellenistic periods have been identi-fied: at Nopigia, in north-west Crete, andVathi, in south-west Crete.

Why do the Minoan and Venetian flooddeposits survive? I suggest that the magni-tude of the Minoan and Venetian eventswiped out the preceding flood deposits. Thefact that so many Minoan deposits survivedthe Venetian events suggests that they wereinitially even more extensive and bigger in

scale than those associated with the MedievalLittle Ice Age. The implication is that theMinoan climate was not only substantiallydifferent from today's, but also from that ofthe Medieval Little Ice Age.

Other Cretan climate proxy data, whichmight clarify the situation, are not forthcom-ing. The pollen data from Crete for the Middleto Late Minoan period are confined to theTersana core (Moody et al. 1996). Increases indeciduous trees and a decline in olive at thetop of the pollen preserving spectra areremarkably similar to the changes noted in theAsi Gonia core for the Medieval Little Ice Age,though on a larger scale.

Meagre as they are, when combined, theflood and pollen data from Crete suggest thatthe Middle and Late Minoan periods mayhave coincided with another Little Ice Ageevent. Is there any other independent evi-dence for such an event? In fact there is a nearglobal horizon of glacial advance between3600 and 3000 bp, which calibrates to c.2000-1250 BC, that is, MMI-LMIII (Table 4.1).Furthermore, there may have been five or sixsuch episodes during the last 10,000 years.

How would the proposed Little Ice Ageevent have affected Minoan weather and sub-sistence? Tempting as it is to transpose themanifestations of the Medieval LIA (severewinters, drought, summer rain and floods)onto the Minoan LIA, such correlations wouldbe simplistic, especially since the Minoan LIAwill have been imposed on a climate regime(that existing during the EMI-II period) quitedifferent from that on which the MedievalLittle Ice Age was imposed (the so-called'Medieval Warm Period'). Nevertheless, itseems clear that the rhythm of agricultural lifewhich formed the basis of Minoan civilizationwould have been disrupted by periodic debrisflows and floods that far exceeded those of theMedieval Little Ice Age in scale—let alonethose that we experience today. The construe-

Table 4.1. Episodes of significant Holocene glacial advances in Europe compared with cultural periods in Crete(after Grove 1998).

Date bp Date BC/AD* Cultural phase Comment

1250-1850 AD Venetian and Turkish occupation

800-900 AD

3000-3600 2000-1250 BC

4200

4600

6000-6600

7300-7700

8100-8400

9000-9400

2800 BC

3450 BC

5515^871 BC

6495-6097 BC

7450-6950 BC

8550-8029 BC

Early to Middle Byzantine

EMIII/MMI to LMIIIB

EMI/II?

FN/EMI?

Late Neolithic

Early Neolithic at Knossos

Pre-pottery Neolithic

Mesolithic

600 years. Medieval Little Ice Age.Advances in Swiss Alps, Caucasus, Rockies,Andes up to 2.5 km beyond modern front.Especially intense around 1350,1600-50,1770-80,1815-20, and 1850-60.

100 years, 3.3 generations. Advances in Alps.

750 years, 23 generations. Loebben Advancein Alps, advances greater than MedievalLittle Ice Age.

100 years?, 3.3 generations.

100 years?, 3.3 generations.

644 years, 21.5 generations. Less intense.

398 years, 13 generations.

500 years, 16.5 generations.

521 years, 17 generations. Schlaten Advancein Alps, advances on same scale asMedieval Little Ice Age

h calibrated by OxCal 14L.

tion of Minoan dams, such as the one atPseira, make more sense within this sort of cli-mate reconstruction.

Such environmental unpredictability is like-ly to have spawned a new approach to subsis-tence, as well as a serious revamping of beliefsystems and ritual in an effort to control theperceived chaos. The enormous food storesthat evolve at this time within the 'palaces'and then the Villas7 may witness the impor-tance of banking against crop failure due toerratic weather patterns. The well-document-ed proliferation of ritual sites—public and pri-vate; peak top, palace and town—may wit-ness the desire to control/appease/supplicatethe unpredictable powers that be. No doubtother variables came into play in the evolutionof Minoan civilization, but the importance of aclimate change at both the beginning and theend of the so-called palace period in Crete andthe southern Aegean and its ramificationsdeserves further consideration.

Conclusion

Environmental data for Middle and LateBronze Age Crete consistently argue for a dif-ferent climate from today's. Evidence that theMinoan palaces may have evolved, floweredand declined during a Little Ice Age that waseven more intense than the Medieval Little IceAge is beginning to accumulate. Exactly howthis climate shift affected Crete needs to bemore thoroughly investigated. In addition tocontinued fieldwork, a fruitful area ofresearch would be computer modelling ofMinoan LIA weather by combining ourunderstanding of the baseline Early Minoanenvironment with the parameters and magni-tude of change which accompany the devel-opment of a Little Ice Age, based on theMedieval LIA data now assembled.


60 Jennifer Moody

Acknowledgments

I am most grateful to the KE' Ephoria inKhania and Rethumno, the Greek Archae-ological Service in Athens, the AmericanSchool of Classical studies at Athens, theBritish School of Archaeology at Athens, theCanadian Archaeological Institute at Athens,and the Department of EnvironmentalResearch of the European Union, withoutwhose support these researches would neverhave taken place. Enormous thanks go to theInstitute for Aegean Prehistory (New York)and other financial sources, who have so gen-erously funded my work. Special thanks go toOliver Rackham for his encouragement andcollaboration over the years. Also, I greatlyappreciate the mentoring and companionshipof Dick and Jean Grove, which added newdimensions to the Cretan landscape. Finally, Iam endlessly indebted to the people of Cretefor their many kindnesses over the years.

Notes

5.

This project was funded by the EC and direct-ed by A.T. Grove (Cambridge) and N.Margaris (University of the Aegean).This phenomenon was first brought to myattention by Oliver Rackham, to whom I amgreatly indebted.This scarp was examined by the author,George Postma and Oliver Rackham whileworking on the Vrokastro Survey Projectdirected by Moody and Hayden in 1988(Hayden et al. 1992). Sherd date is courtesy ofMargrete Hahn.This project was funded by the EC and direct-ed by A.T. Grove (Cambridge) and V. Papa-nastasis (University of Thessaloniki).The Sphakia Survey is directed by Moodyand Nixon under the auspices of the Cana-dian Institute and the Greek ArchaeologicalService (Nixon et al 1990).

6. This discovery was made while running testtransects for the Aghios Vasilios ValleySurvey under the general direction of AlanPeatfield and Stavroula Markoulaki and fielddirection of Jennifer Moody (Moody et al.2001).

Bibliography

Atherden, M. and J. Hall1999 Human impact on vegetation in the White

mountains of Crete since AD 500. The Holocene 9:183-93.

Dalfes, H.N., G. Kukla and H. Weiss1997 Third Millennium EC Climate Change and Old

World Collapse. Berlin: Springer.Grove, J.

1997 The spatial and temporal variations of glaciersduring the Holocene in the Alps, Pyrenees,Tatra and Caucasus. In ESF Project 'EuropeanPaleoclimate and Man7 Special Issue: GlacierFluctuations during the Holocene. Paleoklima-forschungen: Palaeoclimatic Research 24: 95-103.

Grove, J. and A.L. Conterio1994 Climate in the eastern and central Mediter-

ranean 1675-1715. In ESF Project 'EuropeanPaleoclimate and Man' Special Issue: ClimateTrends and Anomalies in Europe 1675-1715;Paleoklimaforschungen: Palaeoclimatic Research 13:274-85.

1995 The climate of Crete in the sixteenth and seven-teenth centuries. Climatic Change 30: 223-47.

Hayden, B., J. Moody and O. Rackham1992 The Vrokastro Survey Project 1986-1989.

Hesperia 61: 293-353.Hempel, L.

1991 Forschungen zur Physischen Geographic der InselKreta im Quartar. Gottingen: Vandenhoek andRuprecht.

Maas, G.S., M.G. Macklin and M.J. Kirby1998 Late Pleistocene and Holocene River Develop-

ment in Mediterranean steepland environ-ments, southwest Crete, Greece. In G. Benito,V.R. Baker and K.J. Gregory (eds.), Paleohydrol-ogy and Environmental Change: 153-65. Chiches-ter: John Wiley.

Moody, J.1997 The Cretan environment: abused or just misun-

derstood? In P.N. Kardulias and M. Shutes(eds.), Aegean Strategies. Studies of Culture and

1.

2.

3.

4.


Environment on the European Fringe: 61-79.Lanham: Rowman and Littlefield.

Moody, J., A. Peatfield and S. Markoulaki2001 Report from the Aghios Vasilios Valley. Paper to

8th International Cretological Congress,Heraklion, September 1996, in press.

Moody, J., O. Rackham and G. Rapp1996 The environmental archaeology of prehistoric

NW Crete. Journal of Field Archaeology 23: 273-97.Nemec, V. and G. Postma

1993 Quaternary alluvial fans in southwestern Crete:sedimentation processes and geomorphic evo-lution. In Alluvial Sedimentation. InternationalAssociation of Sedimentologists Special Publi-cation 17: 235-76.

Nixon, L., J. Moody, S. Price, O. Rackham and V. Niniou-Kindeli

1990 Archaeological survey in Sphakia, Crete. Echosdu monde dassique / Classical Views 34: 213-20.

Rackham, O. and J. Moody1996 The Making of the Cretan Landscape. Manchester:

Manchester University Press.Sherratt, A.

1997 Climate cycles and behavioural revolutions: theemergence of modern humans and the begin-ning of farming. Antiquity 71: 271-87.

Triandafyllidou-Baladie, Y.1988 To Emporio kai i Oikonomia tis Kritis 1669- 1795.

Heraklion: Vikelia.Andel, T. van, C.N. Runnels and K. Pope

1986 Five thousand years of land use and abuse inthe Southern Argolid, Greece. Hesperia 55: 103-28.

Zangger, E.1994 Landscape changes around Tiryns during the

Bronze Age. AJA 98:189-212.

Human Impact on the Vegetation of SouthernGreece and Problems of PalynologicalInterpretation: A Case Study from Crete

Margaret Atherden

Introduction

The traveller in southern Greece and theGreek islands is presented today with a mosa-ic of vegetation types, ranging from woodlandto steppe and cultivated land. Although it issometimes assumed that the 'natural7 vegeta-tion of southern Greece would be evergreenwoodland, dominated by Quercus ilex, there islittle evidence in the present landscape to sup-port this theory. Another common assumptionis that the present vegetation cover representsthe result of centuries or millennia of degra-dation and soil erosion—a theory termed the'ruined landscape' theory by Rackham andMoody (1996). However, modern vegetationcommunities appear to be relatively resistantto change and well adapted to variations inmoisture availability and soil type. Theimpacts of grazing, burning and wood cuttinghave combined with the environmental fac-tors to produce a complex and fascinatingarray of plant communities.

At one end of the spectrum there are surviv-ing examples of mature woodland. Theseinclude woods dominated by Abies cephaloni-ca, Pinus brutia and (on Crete) Cupressus sem-pervirens in the mountains; Pinus halepensisaround many of the coasts; and a wide varietyof deciduous trees at low and middle alti-tudes. Rackham (1982) recognizes about 12

types of woodland in southern Greece todayand believes that there would have been moretypes in the past when deciduous trees weremore widespread. Scrub and maquis (on acidsoils) are dominated by shrub species whichare capable of growing up to form real wood-land in the absence of grazing/browsing orother pressures. They are dominated byspecies such as Quercus cocci/era, Pistacia lentis-cus, Arbutus unedo and Erica arborea and varygreatly in height and growth form. Othercommunities are described as garigue orphrygana, the former dominated by woodyundershrubs and the latter by aromatic herbs,although there is much overlap in speciesbetween the two. Grassland areas are found inthe mountains or as patches of steppebetween other vegetation types. Grasses arenot necessarily dominant and a wide range ofherbs and bulbs may be found growingamong them. There is also overlap in speciesbetween steppe/grassland and cultivatedland—a problem exacerbated for palynolo-gists by the fact that the pollen of many taxa isonly recognizable to family level (e.g.Apiaceae, Brassicaceae). Wetland communi-ties are generally confined to permanentwatercourses and coastal sites, although occa-sional surprises occur, including the peat bognear Asi Gonia which forms the case studybelow.

5

Human Impact on the Vegetation of Southern Greece 63

The impact of grazing/browsing in theMediterranean region goes back at least twomillion years (Naveh 1990) and includes theeffects of Pleistocene herbivores, for examplethe dwarf hippo and dwarf deer found as fos-sils on the island of Crete (Rackham andMoody 1996). In the Holocene period, muchof this grazing has been by domesticatedstock, although the continuing effects of wildanimals should not be discounted. The maindomesticated grazing animals are sheep andgoats, which are almost ubiquitous in uplandareas, but cattle, mules and donkeys also playa role. Most grazing by sheep and goats isorganized by a system of shepherding, whichenables the animals to take advantage ofextensive areas of grazing land without over-grazing specific areas. Although grazing ismost often associated with herbaceous ordwarf shrub communities today, browsingwithin woodlands is also practised in someareas and may have been more widespread inthe past, affecting both the lower branches ofthe trees and the undergrowth. In someupland areas of southern Greece, a form oftranshumance is practised, whereby men andboys spend the summer living in summer vil-lages (e.g. Asfendos on Crete), from whichtheir flocks and herds graze the better-watered pastureland at higher altitudes.Lewthwaite (1981) argues that this is an inher-ently unstable system of land management,introduced as a tactical adaptation in mediae-val and early modern times, but it seems to bean enduring feature of the landscape in manyareas. In Crete, archaeological and palynolog-ical evidence (see below) suggests that thesystem probably dates from about 1000 AD.

Grazing is often accompanied by burning,designed to remove unpalatable woody tissueand stimulate a fresh growth of herbaceousmaterial. Naveh (1990) suggests that deliber-ate burning has been practised since lateUpper Palaeolithic times in the Middle East.

Plant beneficiaries include grasses, legumes,bulbs and tuberous plants, but most of allburning gives a competitive advantage towoody shrubs, many of which sprout quicklyafter a fire from buds near the base of thestem. Pine trees also benefit through preferen-tial regeneration from seed after a fire.Mediterranean species, such as Pinus halepen-sz's, are fire adapted, in common with mostother members of the genus. Today, 'occupa-tional' and accidental burning are common insouthern Greece, affecting perhaps 2% of thelandscape each year (Rackham 1983). Docu-mentary sources record the practice from atleast the seventeenth century onwards. Forexample, Randolph recorded annual burningof grass and 'bryers' in Arcadia to promotenew pasture growth in the seventeenth centu-ry (Rackham 1986), and de Boblaye describedthe burning of forests by shepherds to providepasturage for their flocks in about 1825(Wright 1972).

Many of the deciduous tree species found insouthern Greece are highly palatable but rela-tively slow-growing, and this fact has led totheir retreat to inaccessible ledges and cliffs inareas where grazing pressure has beenintense. They are also slow to respond afterburning and take several years to reach theflowering stage. In Boeotia, for example, treessuch as Tilia, Ulmus and Ostrya are today con-fined to habitats such as cliff faces on MtHelicon, whereas pollen analysis suggeststhat they were formerly more widely distrib-uted. Deciduous oaks seem to be less palat-able than most other deciduous trees, whichoffers an explanation for their more frequentsurvival in woods today (Rackham 1982).Shrub species, on the other hand, often resp-ond more quickly after grazing or burningand can produce flowers within a few monthsor even weeks of serious fires or periods ofintense grazing. The typical maquis shrubs,such as Pistacia lentiscus, Genista spp. and the

64 Margaret Atherden

ubiquitous Quercus cocci/era, are distinguishedby their remarkable powers of regenerationafter such pressures and are often the mainbeneficiaries of increased grazing or burning.Rackham (1982) attributes the present wide-spread distribution of Quercus coccifem to theeffects of goat grazing, without which thespecies might occupy a more restricted rangeof both climate and soil. Garigue species, forexample Sarcopoterium spinosum, Lavandulaspp. and Cistus spp., also respond well tograzing but are less capable of rapid recoveryafter severe fires. Steppe species, especiallyPoaceae and Asphodelus spp., are extremelyresistant to both grazing and burning, andoften characterize areas of over-exploitation.

Wood cutting is another human impactwhich has affected southern Greek vegetationfor many centuries. Its impact varies accord-ing to the species' ability to regrow from cop-piced stumps; for instance, most of the decid-uous trees/shrubs coppice easily, whereaspine does not. Abies cephalonica and Cupressussempervirens both show some ability to regrowand are sometimes encountered today in acoppiced or pollarded form (Rackham 1983).On pollen diagrams it is often difficult to dis-tinguish between the effects of wood cuttingand those of grazing or burning, as all lead toa diminution in pollen production.

If these various forms of pressure arerelaxed or abandoned, changes occur as theplant communities adjust to the new situa-tion. Secondary woodland regenerates easilyfrom maquis or scrub and, as it does so,impacts on the other plant communities.Garigue and phrygana seem to be largelydependent on grazing for their continuanceand lose their competitive edge over steppewithout it. Rackham (1982) suggests that apossible explanation may lie in the greaterwater demand from maquis shrubs as theymature, depriving the garigue/phryganaplants of their moisture supply. He envisages

a landscape of woodland interspersed withsteppe in the prolonged absence of grazing,burning or wood cutting in southern Greece.

Using this knowledge of the modern ecolo-gy, it is possible to make some predictionsabout the probable effects on the vegetation asreflected in the pollen diagrams when humanimpact increases or decreases. A sustainedincrease in grazing pressure should result ini-tially in more vigorous growth of the under-storey plants and later in the reduction ofwoodland to maquis/scrub and the spread ofgarigue/phrygana and steppe communities.The early stages might be expected to show anincrease in arboreal pollen (AP), especiallylight-demanding taxa, but the later stagesshould show an increase in maquis shrubs(e.g. Ericaceae, Arbutus) and garigue/phry-gana taxa (e.g. Cistaceae). Poaceae would alsoincrease and are likely to be the most abun-dant pollen indicators of the steppe communi-ties, as species such as Asphodelus are under-represented on pollen diagrams (Bottema1980). A relaxation in grazing pressure shouldproduce a gradual decrease in diversity, asgarigue/phrygana give way to steppe andmaquis shrubs grow up into woodland, shad-ing out understorey species, such as ferns. Theeffects of burning as shown by pollen dia-grams might be seen in increases for pyro-phytic species such as Arbutus, Cistaceae,Quercus coccifera-lype, but these will be impos-sible to differentiate from the effects of graz-ing. An increase in Pinus might offer mostpromise, especially as its pollen is well dis-tributed. The effects of wood cutting areunlikely to be distinguishable from those ofgrazing/burning on the pollen diagrams.

The Palynological Record

Palynologists have long recognized certainnon-arboreal (NAP) pollen types as indicators


of human activity. Behre (1990) lists cultivatedspecies as 'primary' anthropogenic indicatorsand several types of NAP as 'secondary'anthropogenic indicators, although some ofthem (e.g. Artemisia, Chenopodiaceae) canalso indicate steppe conditions, not necessari-ly associated with human impact. It is usuallyeasier to recognize indicators of cultivationthan grazing, as the latter often involveschanges in the distributions or abundance ofnative species whereas the former usuallyinvolves the introduction of non-native ordomesticated species. However, the fact thatmany of the cultigens (e.g. the cereals) arenative to the Mediterranean area blurs the dis-tinction. The two groups are also closely asso-ciated in all but modern intensive horticul-ture, as cultivated fields in southern Greececommonly have fringing areas of grassland orare alternated with fallow. Indicators of graz-ing/browsing are particularly difficult to spoton pollen diagrams when the activity takesplace in woodlands, as this may result in verylittle change in species composition. As notedby Bottema and Woldring (1990), the initialeffects of 'clearance' in woodland environ-ments may be to stimulate production of AP,as the lower branches of the trees receive morelight and flower more profusely.

Despite these difficulties, Bottema andWoldring (1990) recognize several indicatorsof grazing/meadow management, includingPlantago lanceolata-type, Poaceae, Poterium/Sanguisorba minor-type and Rumex acetosa-type. They distinguish other taxa as beingassociated with both burning and grazing,namely, Quercus coccifera, Arbutus unedo,Buxus sempervirens, Pinus halepensis, P. brutiaand Cistus spp. A general assumption hasbeen made that increased grazing or burningwill lead to a decrease in arboreal or shrubpollen, and there has been a tendency to lookfor 'clearance phases' in the pollen record, asin northern Europe. Although it is acknowl-

edged that some tree and shrub species dorespond to increased light availability,decreases in arboreal and shrub pollen havenormally been interpreted as indicating anincrease in human impact on vegetation. Aparticular difficulty is encountered in theinterpretation of pollen of the taller maquisshrubs, such as Quercus coccifera and Arbutusunedo, many of which fit uncomfortably with-in northern European conventions of arborealpollen (AP) or shrub pollen (SP), owing totheir ability to exist in a variety of growthforms. Nevertheless, the APiNAP ratio hasbeen used as a crude indicator of humanimpact and has also been linked with the con-cept of botanical diversity (Bottema 1982;Jahns 1993). Some of these interpretationsmay be challenged in the light both of obser-vations made in the field in southern Greeceand also comparison of the palynological anddocumentary/archaeological evidence forgrazing activities.

Although several pollen diagrams havebeen published from southern Greece and theGreek islands, few of them consider theimpact of grazing/browsing in any detail.Some pollen diagrams—for example, Kopais(Greig and Turner 1974; Allen 1990), Tersana(Moody et al. 1996)—are limited in their use-fulness by the plotting of deciduous and ever-green oak pollen as a single curve. This makesit impossible to estimate how much of theQuercus pollen is from Q. coccifera and there-fore possibly from a maquis rather than awoodland situation. However, even whendivided into deciduous and evergreen types,the Quercus pollen curves always present aparticular challenge of interpretation, as willbe seen below. Several pollen diagrams showthe development of Pinus-Quercus woodlandin the early Holocene being replaced by awoodland dominated by Quercus species inthe middle Holocene—for example, MaloJezero (Jahns 1992), Xinias (Bottema 1979),


Aghia Galini (Bottema 1980). However, thereis evidence for non-woodland communitiesalongside the woodlands in nearly all the dia-grams, suggesting that although the wood-land cover was more extensive in the middleHolocene, it was not continuous. Certainly byNeolithic times, secondary woodland with ahigher proportion of evergreen oaks is record-ed at several sites, such as Thermisia (Sheehan1979), Lake Lerna (Jahns 1993), Kleonai(Atherden et al. 1993). These diagrams alsoshow an increase in the pollen of maquisshrubs following the first major agriculturalphase of the Bronze Age, as do those fromKiladha (Bottema 1990) and Osmanagalagoon (Wright 1972). Most writers showgreater interest in indications of cultivation onthe pollen diagrams than in signs of grazing,but the diagram from Lake Lerna (Jahns 1993)has good evidence for grazing and probablyfor burning as well from about 1000 ADonwards. In lowland areas, grazing activitiesare likely to have been closely associated withcultivation; for instance, livestock are oftengrazed under olive trees. This makes it diffi-cult to distinguish the effects of arable andpastoral agriculture on the pollen diagrams.In upland areas, on the other hand, the impactof grazing is likely to have been more signifi-cant than that of cultivation, so it is from sitesat higher altitude that the clearest evidencemight be expected of the effects of grazing/browsing on the local vegetation.

A Case Study from the White Mountains ofCrete

In order to explore some of these ideas further,a peat bog was studied from Asi Gonia in theWhite Mountains of central Crete (Atherdenand Hall 1994,1999; Grove et al. 1991). The sitelies at an altitude of 780 m; it is a spring-fedbog at the head of one of the feeder streams of

the River Koularas. It lies just above the pre-sent limit of cultivation for olives and othercrops and, although there are some old ter-races close to the site, it is likely that grazinghas been the dominant land use in most peri-ods. Today, the area is used for grazing flocksof sheep and goats, which are shepherdedfrom the summer village of Kallikrati, lessthan 2 km away. The site therefore presents anopportunity to study human impact on thevegetation in an area where grazing is likelyto have been the most significant influenceover the last few millennia. An added advan-tage is that a major archaeological survey hasbeen undertaken over the past few years inSphakia, a few kilometres to the west, whichprovides a source of archaeological and docu-mentary evidence to complement the palyno-logical record (Nixon et al 1988, 1989, 1990,1994).

The area of peat measures approximately190 m by 60 m and reaches a maximum depthof 4.7 m. The peat contains wood fragmentstowards the bottom and occasional mineralbands, but most of the sediment is organic,which has led to good preservation of pollenand spores. Such peat deposits are rare insouthern Greece, adding to the interest of thesite. However, tapping of the spring for irriga-tion purposes and sporadic burning of the bogsurface have led to considerable drying out ofthe peat over the past decade, so its days maybe numbered. Today the surface is mostly dryand dominated by bracken (Pteridium aquil-inum) and rushes (Juncus effusus), with a fewsedges (e.g. Scirpoides holoschoenus) and moss-es (e.g. Polytrichum commune) in the damperspots. High percentages of Cyperaceae pollenand significant levels of Sphagnum spores insome layers of the peat suggest that the bogwas considerably wetter in the past. A solitaryclump of Sphagnum auriculatum grows nearthe site today (Turland and Wilson 1995).

The vegetation surrounding the site is a


mosaic of Erica-Arbutus maquis, phryganaand steppe but there are also some maturetrees of Platanus orientalis on the edge of thebog and trees of Quercus ilex, Quercus pubes-cens and Castanea saliva within a few hundredmetres of the site. The changing balance ofthese plant communities (woodland, maquis,garigue/phrygana and steppe) as inferredfrom the pollen evidence should provide aproxy record of grazing pressure in the imme-diate area.

Two cores were taken from the site, one witha vibro-corer and one with a Hiller borer. Theformer (core AG-B) recovered only the toptwo metres of peat but the latter (AG-A) pro-vides a full record from the site. Seven radio-carbon dates have been obtained, one ofwhich (from near the base of the vibro-coreAG-B) is almost certainly unreliable (Table5.1). The other dates form a sequence rangingfrom 1510±100 bp to 85±50 bp, indicating thatthe peat accumulated since Early Byzantinetimes. Full details of the correlation of the twocores and the interpretation of the radiocar-bon dates have been published elsewhere(Atherden and Hall 1999). Correlationbetween the pollen diagrams from the vibro-core and the top two metres of the Hiller coreis close, so only the longer diagram (AG-A)will be presented here.

The use of incremental cluster analyses(CONISS) within the TILIA computer pro-gram enables an objective approach to be

taken to the zoning of the pollen diagram(Grimm 1991). The technique also revealswhich zone boundaries are most significant interms of vegetation change. In zoning the twopollen diagrams from the site, four differentmethods of cluster analysis were applied andzone boundaries drawn at points whichappeared as significant changes according toseveral methods. On all four methods, 305 cmmarked the most significant change on thediagram. Similarly sharp transitions may beseen at 15 cm, 125 cm and 275 cm. More vari-ation was found between the four methods inother parts of the cores, so gradual transitionshave been used for the zone boundaries at25-35 cm, 65-85 cm and 205-215 cm (Fig. 5.1).

Table 5.2 shows the pollen taxa recognizedon the diagram and gives examples of thespecies they are likely to represent, based onknowledge of the present flora of Crete(Turland et al 1993) and plants found growingin the area around the site today. Six broadvegetation groups are recognized, namelywoodland, maquis/scrub, garigue/phrygana,steppe, cultivated/planted land and wetland.The pollen taxa are arranged in Figure 5.1according to these groups. There is some over-lap between groups where taxa such as ever-green oaks occur in several different plantcommunities or where families such asLamiaceae or Fabaceae include species char-acteristic of different communities. Twopollen traps were set on the edge of the site to

Table 5.1. Radiocarbon dates from the two Asi Gonia cores.

Core

AG-BAG-BAG-BAG-B*AG-AAG-AAG-A

Depth of sample

13 cm72cm127cm191cm300cm459cm469cm

C-14 date bp

85±50320±60700±701220±751010±701430±751510±100

Lab. ref. no.

Q-2707Q-2706Q-2705Q-2704AA-16472Q-2709Q-2708

Position in core

Base of AG-8AG-5/6 boundaryTop of AG-4Near base of AG-4AG-1/2 boundaryNear base of AG-A coreBase of AG-A core

* This date is considered to be unreliable.

Dark grey-brown peat + mineral material Dark grey-brown monocot peat + mosses Fibrous monocot peat + wood fragments Compact wood peat

Figure 5.1 Pollen diagram AG-A from Asi Gonia. Figures are percentages of total dryland pollen. Charcoal fragments are also expressed as percentages of totaldryland pollen. Radiocarbon dates refer to this core only (see text).

Table 5.2. Ecological interpretation of pollen types

Pollen/spores

AbiesPinusCupressaceae

BetulaQuercus coccifera-type

Quercus undiff./indet.Ulmus

AlnusFagusCarpinus

PlatanusCastaneaJuglansOstryaAcerMorusOleaRutaceae

SalixCorylusSorbus-type

HederaRhamnaceaePhillyreaPistacia

Arbutus

Ericaceae

Cistaceae

SarcopoteriumEphedraCaprifoliaceae

Poaceae undiff .

Poaceae >40 micronsCyperaceaeRumexPolygonaceae undiff.Fabaceae

ApiaceaeArtemisia

Probable species

Abies cephalonicaPinus brutiaJuniperus oxycedrusCupressus sempervirensBetula pendulaQuercus cocciferaQuercus ilexQuercus pubescensUlmus minorZelkova abeliceaAlnus glutinosaFagus sylvaticaCarpinus betulusCarpinus orientalisPlatanus orientalisCastanea sativaJuglans regiaOstrya carpinifoliaAcer sempervirensMorus albaOlea europaeaRuta chalepensisCitrus fruitsSalix albaCorylus avellanaSorbus ariaSorbus umbellataCrataegus monogyna ssp. azarellaHedera helixRhamnus lycioidesPhillyrea latifoliaPistacia lentiscusPistacia terebinthusArbutus unedoArbutus andrachneErica manipulifloraErica arboreaCistus salvifoliusOther Cistus spp., Fumana spp.,Helianthemum spp.Sarcopoterium spinosumEphedra campylopodaSambucus ebulusLonicera etruscae.g. Anthoxanthum sp., Briza minor,Aira elegantissimaCereals, e.g. barleye.g. Scirpoides holoschoenuse.g. Rumex bucephalophoruse.g. Polygonum avicularee.g. Genista acanthoclada, Lotus spp.,Trifolium spp.e.g. Daucus carota, Eryngium campestreArtemisia arborescens

Habitat type

WoodlandWoodlandMaquis /ScrubWoodland/PlantedWoodlandWoodland / ScrubWoodlandWoodlandWoodlandWoodland /ScrubWoodland/WetlandWoodlandWoodlandWoodland/ScrubWoodland/PlantedWoodland/PlantedPlantedWoodlandWoodlandPlantedPlanted /MaquisSteppePlantedWetlandPlantedScrub /MaquisScrub /MaquisWoodland / ScrubWoodlandScrub / MaquisWoodland /MaquisMaquis / Phry ganaWoodland / MaquisMaquis / WoodlandMaquis / WoodlandMaquisMaquisMaquis / Garigue

Maquis / GarigueGarigue / Phry ganaMaquis /ScrubScrubWoodland / ScrubSteppeSteppeCultivated landWetlandSteppeSteppeMaquis / Phry gana /SteppeSteppe /Cultivated landPhrygana

Provenance

MainlandRegionalRegionalRegionalMainlandLocalLocalRegionalRegionalRegionalMainlandMainlandMainlandMainlandLocalLocalRegionalMainlandLocalRegionalRegionalRegionalRegionalRegionalRegionalRegionalRegionalLocalLocalLocalRegionalRegionalRegionalRegionalRegionalLocalRegionalLocal

RegionalLocalRegionalRegionalRegionalLocalLocalRegionalLocalRegionalRegionalLocal

RegionalRegional



Table 5.2 continued

Pollen/spores

Asteraceae undiff .

Asteraceae LactuceaePlantaginaceaeLamiaceae

Rosaceae undiff.BrassicaceaeCaryophyllaceae

ChenopodiaceaeRanunculaceae

BoraginaceaeScrophulariaceaeUrticaRubiaceaeDipsacaceaeEuphorbiaceaeMalvaceaeHypericumLiliaceaePotamogetonSparganium-typePolypodiumPteridiumOsmundaPteropsida

Sphagnum

Probable species

e.g. Carduus spp., Carlina spp.,Pulicaria dysentericae.g. Crepis vesicariae.g. Plantago afrae.g. Teucrium spp., Phlomis fruticosa,Coridothymus capitatuse.g. Pyrus amygdaliformise.g. Biscutella didymae.g. Arenaria cretica, Bufonia stricta,Silene spp.e.g. Chenopodium albumRanunculus ficariaAnemone coronariae.g. Echium italicume.g. Sibthorpia europaeaUrtica piluliferae.g. Asperula spp., Galium spp.e.g. Knautia integrifoliae.g. Euphorbia characiase.g. Lavatera bryoniifoliae.g. Hypericum empetrifoliumAsphodelus aestivuse.g. Potamogeton pectinatusSparganium erectumPolypodium cambricumPteridium aquilinumOsmunda regalise.g. Blechnum spicant, Athyriumfilix-feminaSphagnum auriculatum

Habitat type

Cultivated land/Steppe/WetlandSteppe /Cultivated landSteppe /Cultivated landPhrygana

Maquis /ScrubCultivated land

SteppeCultivated landWetlandSteppeCultivated landWetlandCultivated landCultivated land /SteppeCultivated land /SteppeSteppeCultivated landMaquis / PhryganaPhrygana / SteppeWetlandWetlandWoodland/ScrubWoodland / ScrubWoodland

WoodlandWetland

Provenance

Regional /Local

RegionalRegionalLocal/Regional

RegionalRegional

RegionalRegionalLocalRegionalRegionalLocalRegionalRegionalRegionalRegionalRegionalRegionalRegionalRegionalRegionalRegionalLocalRegional

LocalLocal

record the present day pollen rain and a mosspolster growing close to the coring site in themiddle of the bog was also analysed.Comparison of the present flora around thesite with the results from the pollen traps andmoss polster helps with interpretation of thepollen diagrams.

The woodland group includes trees such asPinus, Quercus (evergreen and deciduous),Alnus, Platanus and Ostrya as well as Hederaand various ferns. Pinus is not found close tothe site and never forms a substantial curve onthis pollen diagram. Most of the deciduoustrees are highly palatable to livestock andtherefore confined to woodlands or inaccessi-

ble cliff faces. Their pollen records may betaken as indicating woodland conditions inthis context. The only genera with significantpollen records are Alnus and Platanus, both ofwhich were probably associated with wetsites, possibly growing around the site itself.In the pollen traps, woodland trees excludingevergreen oaks constituted 26.4% and 14.5%of the total pollen; in the moss polster the fig-ure was slightly higher at 31.4%, suggesting agreater regional component to the pollen rainin the centre of the site. Records for Oleapollen could derive from wild olive growingin maquis vegetation or from cultivated olivegroves. The amounts of Olea pollen are never


very high on this pollen diagram and it isunlikely that olives were grown close to thesite. In the pollen traps, Olea pollen contri-buted 12.0% and 6.0% and it contributed 4.0%in the moss polster. These records confirm thatOlea pollen is well distributed today and thatlow amounts on the pollen diagram do notsignify olive cultivation close to the site.

Evergreen oak pollen presents a particularchallenge to interpretation, as it includespollen of Quercus ilex and Q. cocdfem. The for-mer is palatable to grazing animals and is nor-mally found in woodlands, whereas the latteris more resistant to grazing pressure andvaries in growth form from a shrub a few cen-timetres ,high to a mature tree, sometimesgrowing in scrub/maquis or garigue/phry-gana and sometimes forming part of a wood-land community. On the pollen diagram fromAsi Gonia (Fig. 5.1), the curve for evergreenoaks follows closely the curve for deciduousoaks. This close association between the twocurves suggests that in this case the evergreenoaks are behaving as woodland trees ratherthan shrubs, despite the fact that they can pro-duce pollen when less than one metre inheight. The records for evergreen oaks fromthe pollen traps and moss polster wereextremely variable, ranging from 5.0% to56.0%! The highest figure was from the mosspolster in the middle of the site, which couldsuggest that most of the pollen was regionalrather than from the local maquis. There is awell-developed Quercus coccifera wood a fewkilometres from the site, near Alones. Thisreinforces the idea that most of the evergreenoak pollen on this diagram has come fromwoodland trees rather than from the sur-rounding maquis/scrub. Spores from fernsindicate woodland conditions but may beexpected to increase in woodland edge situa-tions or when the canopy is opened up to cre-ate only a light shade. In the pollen traps, fernspores only accounted for 1.4% and 7.0% of

total pollen and in the moss polster, 4.9%.Most were Pteridium spores and brackengrows on the site today, so they are probablyof local origin. However, in the past when thebog surface was wetter, ferns probably grewfurther away.

The local maquis flora is currently dominat-ed by Erica arborea and Arbutus unedo, both ofwhich are useful indicators of maquis on thepollen diagram. Other typical maquis taxainclude Phillyrea, Pistacia and Fabaceae, all ofwhich tend to be under-represented on pollendiagrams from Greece/Crete. Interpretationof the extent of maquis will rely heavily on theEricaceae curve. Figures for maquis/scrubtaxa from the pollen traps were 8.6% and13.0%, compared to only 1.4% for the mosspolster. This suggests that high levels ofpollen from maquis shrubs are likely to indi-cate maquis/scrub growing close to the site.Taxa taken as representing garigue/phryganainclude Cistaceae, Sarcopoterium, Fabaceae,Lamiaceae and Euphorbiaceae. All are under-represented on pollen diagrams and none isstrictly confined to garigue/phrygana, forexample some members of the Fabaceae arecommon in maquis (such as Calicotome villosa).The two pollen traps set near the site recordedrelatively high levels of Sarcopoterium pollen(18% and 19%), but these were probably fromplants growing very close to the trap, as themoss polster from the middle of the site had amuch lower figure of 0.7%. Other garigue/phrygana taxa contributed negligible amountsto the traps and polster, confirming the poorrepresentation of this group in pollen records.

Taxa characteristic of steppe and cultivatedland have been grouped together in thispaper, as many types of NAP, such as Planta-ginaceae, are characteristic of both habitats.Figures from the pollen traps were 7.6% and31.0%, the latter figure being inflated by a fig-ure of 17.0% for Brassicaceae in one of thetraps. The moss polster had only 1.5% for


steppe/cultivated land taxa. Poaceae is thedominant taxon within this group, with well-distributed windblown pollen. At the presentday, most of the pollen probably derives fromsteppe, as the site is above the current level ofcultivation, but the existence of old terracesnear the site presents the possibility of someof the pollen coming from cultivated land inthe past. Finally, the wetland group of taxacomprised mainly Cyperaceae and Sphagnum,with very occasional records for Sparganium-type and Potamogeton. These taxa are likely tohave been derived from the peat bog itself, sothey have been excluded from the pollen sum.Modern figures are 8.0% and 10.0% from thepollen traps and 3.6% from the moss polsterbut, as noted above, the surface of the bog isdrying out fast today.

Changes in the impact of grazing should bereflected on the pollen diagrams in variationsin the balance between the vegetation com-munities, as outlined above. Clues to the sig-nificance of grazing/browsing may be soughtin the relative proportions of taller communi-ties, such as woodland, maquis or scrub, andthe low-growing communities, such as gari-gue/phrygana or steppe. Evidence for burn-ing may be sought in the status of garigue/phrygana communities, which are believed tobe particularly sensitive to this factor. A curvehas been plotted for microscopic charcoalfragments over 10 microns in diameter. Thisshould give an independent indication of thefrequency of burning (either accidental or'occupational') in the past, which may becompared with the pollen evidence.

The Pollen Diagram

The pollen diagram will now be interrogatedfor changes in vegetation which may havebeen produced by variations in the intensityof grazing/browsing and burning.

Zone AG1

This corresponds to the Early Byzantine andSaracen periods. The Sphakia Survey hasshown that a shift of settlement took placeaway from the uplands towards the coast,starting in the first century AD. By the timepeat accumulation started in the fifth or sixthcentury AD, the vegetation around the AsiGonia site would have had several centuriesof relatively light human impact. There is noarchaeological evidence for occupation ofsummer villages in the mountains at this time,so grazing/browsing is likely to have beenlight, although it is not likely to have beenabsent altogether. The prediction from thearchaeological evidence would be for a regen-eration of woodland and a reduction ingarigue/phrygana and steppe communities.This is in perfect accord with the situation inZone 1 on the pollen diagram. Oak pollen(both deciduous and evergreen) is dominant,with a scattering of records for other trees,including Pinus, Ostrya, Acer and Corylus.There are occasional records for Hedera butvalues for fern spores are low, suggesting fair-ly dense shade. The only pollen from cultivat-ed trees is from Olea, which could either havebeen growing in the local maquis or haveblown from cultivated land at lower altitude.Values for maquis/scrub taxa in general arelow, as are those for Poaceae and steppe taxa.There are low charcoal records, indicating lit-tle burning near the site.

Zone AG2

The lower boundary of this zone is the mostsignificant one using all four methods ofincremental cluster analysis. A radiocarbondate of 1010 ± 70 bp places it near the start ofthe Second Byzantine period. The SphakiaSurvey records the re-establishment of settle-


ments in the middle and upper zones of thelandscape around 1000 AD. It is probable thatthe upland pastures (madhares) were used forsummer shepherding from this time onwards,with men and boys spending the summer intemporary settlements at high altitude (Nixonet al. 1989). The exploitation of wood and tim-ber probably increased as new settlementswould have needed both firewood and build-ing timbers. The prediction would thereforebe for a major impact on the vegetation at thistime, with an increase in both wood cuttingand grazing/browsing pressure. The pollendiagram records a major decrease in AP at thestart of Zone 2, affecting mainly the oaks (bothdeciduous and evergreen). Pollen of the light-loving Platanus increases and there are alsomore records for Pteridium and other fernspores, as the woodland edge habitat expandsand light levels increase. Pollen of Ericaceaeand almost all maquis/scrub and garigue/phrygana elements increases. Poaceae andsteppe taxa also increase at this point. There isonly a slight increase in charcoal records, soburning may not have been important at thisstage. The surface of the bog itself may havebecome wetter, as there is a marked increasein the pollen of Cyperaceae. The palynologicalevidence thus supports the notion of a majorchange in the settlement and land use patternin the Second Byzantine period.

Zone AG3

This zone dates from the late Second Byzan-tine period. On the pollen diagram a recoveryof woodland taxa is seen, but not to the highlevels of Zone 1. The records for Platanuspollen are maintained and those for fernspores increase, suggesting plenty of wood-land edge habitats available, despite theregeneration of trees. Ericaceae and othermaquis/scrub taxa mirror the Quercus curves,

decreasing and later increasing again towardsthe top of the zone. Garigue undershrubsshow an increase, particularly Cistaceae.Poaceae levels are slightly lower but recordsfor other steppe/cultivated land taxa increasein frequency, for example Asteraceae andBrassicaceae. The charcoal curve rises steadilythroughout the zone. The picture from thepollen diagram is thus one of maquis growingup into woodland but with grazing pressuremaintained on garigue/phrygana and steppecommunities. An increase in burning wouldgive a competitive advantage to maquisshrubs, perhaps accounting for their recoverytowards the top of the zone. The archaeologi-cal evidence provides few clues to thesechanges, so in this case the palynologicalrecord may be used to suggest changes in thesettlement/land use history. The picture isconsistent with a slight decrease in humanimpact, perhaps as land at lower altitude wasexploited again, but with continuing graz-ing/browsing pressure on the uplands suffi-cient to prevent a major regeneration of wood-land to its former levels.

Zone AG4

In the Early Venetian period there are docu-mentary sources which suggest greater pros-perity and much use of timber and wood. Theprediction would be for renewed impact onthe upland environment, through both woodcutting and grazing activities. The pollen dia-gram shows a decrease in woodland taxa butan increase in Ericaceae and other scrub/maquis taxa and relatively high values forgarigue/phrygana taxa. Poaceae and steppeherbs also increase and charcoal levels arehigh. Cyperaceae levels are high, suggestingan increase in surface wetness on the bogitself. This fits in well with the prediction fromthe documentary evidence. Wood cutting


could account for some of the decrease inwoodland taxa but this clearly coincided withconsiderable grazing pressure and increaseduse of occupational burning to improve pas-tures. A mosaic of communities is indicated,with relatively little woodland but plenty ofscrub/maquis (probably with ferns growingin the understorey), garigue and steppe. Localwetland species on the site itself would haveadded to the vegetation diversity.

Zone AG5

In the Late Venetian and early Turkish peri-ods, documentary evidence and visibleremains in the landscape suggest a period ofrelative prosperity but one in which demo-graphic fluctuations were seen as a result ofoutbreaks of plague. The prediction would befor generally high levels of human impact,such as wood cutting, punctuated by periodicdeclines but probably sustained high levels ofgrazing throughout, as numbers of livestockdo not necessarily correspond closely to pop-ulation levels. On the pollen diagram, wood-land taxa remain very low but Ericaceae andother scrub/maquis levels are very high. Poa-ceae and other steppe taxa show a very slightdecrease but Cistaceae records are relativelyhigh, suggesting that garigue was holding itsown. This would fit in with the lower recordsfor charcoal, as garigue is susceptible to burn-ing. A period of intensive grazing pressure isindicated, as predicted, but with less burningand wood cutting. Maquis and garigue com-munities were favoured over woodland andsteppe. Records for ferns decreased as thedense maquis shaded out the undergrowth.

Zone AG6

This zone is also within the Turkish period.Various rebellions and outbreaks of plague in

the seventeenth century might be expected tohave led to relaxed pressure on upland vege-tation from time to time. On the pollen dia-gram, a small increase is seen in woodlandtaxa, especially the fast-growing evergreenoak. Ericaceae and other maquis/scrub taxareach high levels; garigue and steppe taxa staymuch the same as in the previous zone. Thereare peaks for charcoal at the beginning andend of the zone. Cyperaceae and Sphagnumrecords increase a little, indicating wetter con-ditions on the bog, possibly linked to the cli-mate changes in the Little Ice Age, when moreextreme conditions were recorded in the sev-enteenth century (Grove et al. 1992). The inter-pretation of the palynological evidence is forperiods of high grazing pressure being inter-rupted by years of lower impact, when somemaquis shrubs managed to grow up and formwoodland once again.

Zone AG7

During the late Turkish period, prosperity andpopulation increased in Crete. Olive growingexpanded from the eighteenth century on-wards. The prediction might be for sustainedhigh grazing pressure in the uplands but it isalso possible that more agricultural effort wasbeing concentrated on the lowlands and theolive groves. This seems to have been the case,as the pollen diagram records an increase inevergreen oaks and a gradual decrease inEricaceae. There are lower records for garigueand steppe taxa, too, and charcoal recordsdecrease. All this suggests a diminution ingrazing pressure, which allowed the fast-growing Quercus coccifera to grow up andflower more profusely. Garigue and steppetaxa were outcompeted by the tall shrubs anddeveloping woodland.


Zone ACS

The final zone on the pollen diagram datesfrom the Late Turkish period onwards, whendocumentary records are plentiful and aug-mented by oral history Political and demo-graphic changes led to rural instability untilthe mid-twentieth century, but the periodfrom 1950 onwards saw a revitalization ofagriculture in the lowlands and further expan-sion of olive groves. This was accompanied bysome rural depopulation in the uplands,which would be predicted to lead to anincrease in woodland at the expense ofmaquis, garigue and steppe. The pollen dia-gram confirms these predictions in all respectsbut one: there is relatively high Sarcopoteriumpollen. Perhaps this thorny undershrub dom-inated the remaining areas of garigue/steppe,as grazing pressure was concentrated onthem, as it is resistant to grazing by havingboth thorns and an unpleasant taste. Charcoallevels are relatively low, as would be expectedif the emphasis had shifted away from uplandgrazing lands and on to the cultivated lands inthe lowlands. The site at Asi Gonia is too faraway from these agricultural operations tohave recorded them in any detail but it is justpossible to pick out an increase in the Okacurve in Zone 8. The increase in the amount ofwoodland cover in the uplands is well illus-trated in photographic and sketchbook evi-dence, as detailed by Rackham and Moody(1996).

Conclusion

Close analysis of the pollen diagram from AsiGonia reveals a complex picture of changingvegetation cover over the past 1500 years.Most of the changes seen may be interpretedas the result of variations in the pressuresexerted by grazing/browsing and burning.

Where documentary or archaeological evi-dence is available, the correspondence withthe palynological evidence is striking. Thissuggests that the palynological interpretationsare probably correct and enables interpreta-tion of zones for which documentary sourcesare scarce to be made with some confidence. Agreat advantage of the Asi Gonia site is itsgood and consistent preservation of pollenand spores, as compared with some other sitesin southern Greece where pollen preservationis patchy, such as Kleonai (Atherden et al1993). There are few sites in Greece with sucha complete and detailed record of changesduring the recent past (compare, for instance,the diagram from Aghia Galini, which doesnot extend beyond the middle Holocene).

There is a need for further work on the his-toric period in southern Greece, particularlyfrom sites in the uplands. Such analyses havebeen carried out in northern Greece (e.g.Gerasimidis and Athanasiadis 1994), althoughthe impact of grazing/browsing was not high-lighted in the interpretation. Other researchquestions concern the role of transhumance inGreek mountain environments and the rela-tionship between upland and lowland sites.Despite the growing interest in Greek paly-nology over the past few decades, there is stillmuch to be learned about the human impacton the environment of southern Greece andthe Greek islands.

Acknowledgments

I should like to acknowledge in particular thework carried out by Jean Hall in analysing thelonger core from the Asi Gonia site. OliverRackham and Jennifer Moody must take thecredit for the initial discovery of the peat bogand for bringing it to my attention. They,along with Dick Grove, helped with the field-work. Lucia Nixon provided essential back-


ground information on the archaeological anddocumentary evidence from Sphakia. The ini-tial analysis of the cores was carried out underthe auspices of an EC project, directed by DickGrove and Oliver Rackham, contract no.EV4C-0073-UK. Radiocarbon dating wasundertaken at the Godwin Laboratory inCambridge and the University of ArizonaAMS Facility I am grateful to Paddie Mor-rison for assistance with the production of thepollen diagram. Finally, thanks to participantsin the Sheffield Round Table in January 1999for providing stimulating discussion on theideas presented in this paper.

Bibliography

Allen, H.1990 A postglacial record from the Kopais Basin,

Greece. In S. Bottema, G. Entjes-Nieborg and W.van Zeist (eds.), Man's Role in the Shaping of theEastern Mediterranean Landscape: 173-82. Rotter-dam: Balkema.

Atherden, M.A. and J.A. Hall1994 Holocene pollen diagrams from Greece. Histor-

ical Biology 9:117-30.1999 Human impact on vegetation in the White

Mountains of Crete since AD 500. The Holocene 9:183-93.

Atherden, M.A., J.A. Hall and J.C. Wright1993 A pollen diagram from the north-east Pelo-

ponnese, Greece. The Holocene 3: 351-56.Behre, K.-E.

1990 Some reflections on anthropogenic indicatorsand prehistoric occupation phases in the NearEast. In S. Bottema, G. Entjes-Nieborg and W.van Zeist (eds.), Man's Role in the Shaping of theEastern Mediterranean Landscape-. 219-30. Rotter-dam: Balkema.

Bottema, S.1979 Pollen analytical investigations in Thessaly

(Greece). Palaeohistoria 21: 19-40.1980 Palynological investigations on Crete. Review of

Palaeobotany and Palynology 31:193-217.1982 Palynological investigations in Greece with spe-

cial reference to pollen as an indicator of humanactivity. Palaeohistoria 24: 257-89.

1990 Holocene environment of the southern Argolid:

a pollen core from Kiladha Bay. In T.J. Wilkinsonand S.T. Duhon (eds.), Fmnchthi Paralia: TheSediments, Stratigraphy, and Offshore Investiga-tions. Excavations at Franchthi Cave, Fascicle 6:117-38. Bloomington: Indiana University Press.

Bottema, S. and H. Woldring1990 Anthropogenic indicators in the pollen record of

the eastern Mediterranean. In S. Bottema, G.Entjes-Nieborg and W. van Zeist (eds.), Man'sRole in the Shaping of the Eastern MediterraneanLandscape: 231-64. Rotterdam: Balkema.

Gerasimidis, A. and N. Athanasiadis1994 Woodland history of northern Greece from the

mid Holocene to recent time based on evidencefrom peat pollen profiles. Vegetation History andArchaeobotany 4: 109-16.

Greig, J.R.A. and J. Turner1974 Some pollen diagrams from Greece and their

archaeological significance. JAS 1: 177-94.Grimm, E.G.

1991 TILIA and TILIA-graph. Springfield: Illinois StateMuseum.

Grove, A.T., J. Moody and O. Rackham (eds.)1991 Crete and the South Aegean Islands: Effects of

Changing Climate on the Environment. Report onEC Contract no. EV4C-0073-UK.

Grove, J.M., A.T. Grove and A. Conterio1992 Little Ice Age climate in the eastern

Mediterranean. In T. Mikami (ed.), Proceedings ofthe International Symposium on the Little Ice AgeClimate: 221-26. Tokyo: Tokyo MetropolitanUniversity.

Jahns, S.1992 Untersuchungen iiber die holozane Vegetations-

geschichte von Suddalmatien und Sudgriechenland.Gottingen: Cuvillier Verlag.

1993 On the Holocene vegetation history of theArgive Plain (Peloponnese, southern Greece).Vegetation History and Archaeobotany 2: 187-203.

Lewthwaite, J.1981 Plain tails from the hills: transhumance in

Mediterranean archaeology. In A. Sheridan andG. Bailey (eds.), Economic Archaeology: towards anIntegration of Ecological and Social Approaches.BAR International Series 96: 57-66. Oxford:British Archaeological Reports.

Moody, J., O. Rackham and G. Rapp Jr1996 Environmental archaeology of prehistoric

north-west Crete. Journal of Field Archaeology 23:273-397.

Naveh, Z.1990 Ancient man's impact on the Mediterranean

landscape in Israel: ecological and evolutionary


perspectives. In S. Bottema, G. Entjes-Nieborgand W. van Zeist (eds.), Man's Role in the Shapingof the Eastern Mediterranean Landscape: 43-50.Rotterdam: Balkema.

Nixon, L., J. Moody, V. Niniou-Kindeli, S. Price and O.Rackham

1990 Archaeological survey in Sphakia, Crete. Echosdu Monde Classique/Classical Views 34: 213-20.

Nixon, L., J. Moody, S. Price and O. Rackham1989 Archaeological survey in Sphakia, Crete. Echos

du Monde Classicjue/Classical Views 33: 201-15.Nixon, L., J. Moody and O. Rackham

1988 Archaeological survey in Sphakia, Crete. Echosdu Monde Classicjue/Classical Views 32:159-73.

Nixon, L., S. Price, J. Moody and O. Rackham1994 Rural settlement in Sphakia, Crete. In P.N.

Doukellis and L.G. Mendoni (eds.), Structurerurales et societes antiques: 255-64. Paris: LesEditions Belles Lettres.

Rackham, O.1982 Land-use and the native vegetation of Greece.

In M. Bell and S. Limbrey (eds.), ArchaeologicalAspects of Woodland Ecology. BAR InternationalSeries 146: 177-98. Oxford: British Archaeolo-gical Reports.

1983 Observations on the historical ecology ofBoeotia. BSA 78: 291-351.

1986 Observations on the Historical Ecology of Laconia.Unpublished report.


Manchester University Press.Sheehan, M.C.

1979 The Postglacial Vegetational History of the ArgolidPeninsula, Greece. Unpublished PhD disserta-tion, Indiana University.

Turland, N.J., L. Chilton and J.R. Press1993 Flora of the Cretan Area. London: HMSO.

Turland, N.J. and C.C. Wilson1995 Sphagnum auriculatum Schimp.: a genus new to

the bryophyte flora of Crete. Journal of Bryology8: 27-28.

Wright, H.E.1972 Vegetation history. In W.A. McDonald and G.

Rapp (eds.), The Minnesota Messenia Expedition:188-99. Minneapolis: University of MinnesotaPress.

Deconstructing Agricultural Terraces: Examiningthe Influence of Construction Method onStratigraphy, Dating and Archaeological Visibility

Charles D. Frederick and Athanasia Krahtopoulou

Introduction

Geoarchaeological studies are a common com-ponent of many archaeological survey pro-jects in Greece. These studies often identifythe spatial and temporal framework of natur-al sedimentation that may impede archaeo-logical visibility None, however, have exam-ined the implications of anthropogenic sedi-mentation associated with terracing. Agricul-tural terraces are an ubiquitous component ofmany Aegean landscapes, and their construc-tion may hold significant implications forinterpreting archaeological survey data. Thispaper explores terraces from a geoarchaeolog-ical perspective, and examines how theirinternal stratigraphy and archaeological visi-bility may be influenced by the constructionmethod employed. The review demonstratesthat terrace stratigraphy is strongly influ-enced by the construction method and thatthey may impede or improve archaeologicalvisibility. Unfortunately determination of howa terrace was made and the degree to whicharchaeological visibility is affected is impossi-ble from surface examination alone. Attentionis also directed toward identifying the bestway to date terrace construction and therebyevaluate the antiquity of this agriculturalstrategy

Background

Terracing is one of the most common forms ofsoil conservation in the Aegean. Although ter-racing may be found across Greece, it is mostcommon in the southern mainland andislands, where it plays a significant role in thecreation of new agricultural land. Althoughubiquitous, the significance of terracing isonly fully appreciated when the magnitude ofthis agricultural strategy is examined indetail. For instance, Whitelaw (1991: 405) esti-mates that c. 70-80% of the entire island ofKeos had been terraced at some time in thepast. In landscapes such as this, the most sig-nificant factor affecting the visibility ofarchaeological sites to surface survey (exclud-ing surface visibility associated with vegeta-tion cover) is the transformation from a natur-al sloping landscape to an intensively manip-ulated and managed one. This is the oppositeof many landscapes where the most signifi-cant impediment to archaeological visibility isgeological processes such as erosion and sedi-mentation. Attempts to compensate for suchlimitations is one of the main thrusts of geoar-chaeological research, especially in the NewWorld, where the majority of such studiesseek to elucidate periods of landscape activityand delineate portions of the landscape wherearchaeological visibility may be impeded.

6

80 Charles D. Frederick and Athanasia Krahtopoulou

Armed with such geographic and geologicinformation it is then possible to tailor archae-ological survey strategies to compensate forcompromised visibility. In terraced land-scapes, it is clear that it is human activity asso-ciated with the physical restructuring ofslopes that poses the major limitation onarchaeological visibility.

Outside of the Aegean, the origins of ter-raced landscapes have been investigatedarchaeologically, resulting in significant infor-mation on the nature and timing of terraceconstruction (e.g. Denevan et al. 1987; Fish etal. 1984) as well as the long-term edaphiceffects of agriculture (Sandor 1997). Despitethe ubiquity of terraces in the Aegean, studiesconcerning the history of terracing are few,and most features are readily dismissed byfield workers who more often than notassume they are of recent origin (French andWhitelaw 1999; Krahtopoulou 1997) despiteevidence that indicates that the technologyfirst appears in the Middle and Late BronzeAge (cf. French and Whitelaw 1999: 173-75;Moody and Grove 1990:190). This may in partbe due to the fact that, although terracing hasbeen a viable agricultural strategy for severalthousand years, many are of relatively recentage (Krahtopoulou 1997; French and White-law 1999). Nevertheless, the absence of

detailed, systematic examination of agricul-tural terraces is a prominent gap in the region-al literature and one that holds significantimplications for studies of long-term settle-ment and land use.

This paper has two principal goals: (1) toexamine how terrace construction may influ-ence the visibility of cultural material andthereby affect the accuracy of pedestrianarchaeological surveys; and (2) to deconstructagricultural terraces in order to examine howthey may best be dated. The approach is large-ly theoretical, but grounded upon previousempirical results where applicable.

Basic Considerations

Nomenclature

Agricultural terraces are step-like, generallyslope or contour parallel platforms that areconstructed in order to permit cultivation ofslopes with minimal soil erosion. These man-made or anthropogenic landforms generallyconsist of a vertical or near vertical wall orbank, known as a riser, and a flat or nearly sosurface that lies up-slope and behind the riserwhich is known as the tread (see Fig. 6.1 for anillustration of the components of a basic slope

Figure 6.1 Terrace components discussed in the text.

Cultivation surface

Post-abandonment fill (sheetwash)

Tread fill

Riser fill

Deconstructing Agricultural Terraces 81

terrace). Risers may be of masonry or earthen,and the treads vary in slope from flat to gen-tly sloping. There is a tremendous literatureon methods of soil conservation, and the read-er is referred to a number of texts which dis-cuss the details of slope terracing (e.g.Hudson 1971; Morgan 1986).

Terraces are generally constructed in orderto facilitate the management of soil, water,crops or microclimate, although the construc-tion of a terrace clearly influences all of thesefactors. Owing to their multi-functionalnature there is an extensive range of descrip-tive terms for a variety of features that, asHudson (1992:153) observes, fall into the gen-eral category of 'cross-slope barriers7. Thesefeatures range from simple earthen (or conser-vation) banks to formal masonry walls andbenches.

Terraces may be classified by their geomor-phic occurrence, function, morphology, andconstruction method. For instance, Donkin(1979) recognized three kinds of terraces,according to their geomorphic occurrence: (1)cross-channel; (2) contour; and (3) valley floor.Morgan (1986: 226), on the other hand, alsoidentified three kinds of terraces, but primari-ly organized according to their function: (1)diversion; (2) retention; and (3) bench. Diver-sion terraces are intended to intercept over-land flow and divert it to an acceptable outlet,whereas retention terraces are designed toconserve slope water. Bench terraces are prin-cipally employed to permit cultivation ofsteeply sloping land. Moody and Grove (1990)classified terraces in Crete according to theirplan configuration and recognized three basickinds: (1) parallel (or stepped as they werelater referred to by Rackham and Moody1992); (2) braided; and (3) pocket. Althoughthis classification is based upon plan form,Moody and Grove (1990) infer a functionaluse for two of the three types. Braided terraceshave ends that slope together to form a con-

tinuous, albeit zigzag slope and are linkedwith cereal cultivation and use of a plough.Pocket terraces are small slope benches whichare generally semi-circular in plan form andare specifically linked to the cultivation oftrees. Parallel or stepped terraces follow con-tours and are not linked with a special func-tion other than arable agriculture.

Because these features may have been con-structed for a variety of reasons, no singleclassification scheme is widely accepted byresearchers. Modern studies tend to adopteither a nomenclature that stresses terracefunction or method of construction, whereasstudies of ancient fields tend to adopt descrip-tive schemes because it is often difficult toascertain with confidence the motivationbehind ancient terracing. This paper intendsto examine the manner in which terraces werebuilt and is not concerned with their function.Hence the term terrace is used in the mostgeneral sense to refer to cross-slope barriers ofvarious forms that lead to the formation of anear vertical riser and a near level tread,although both of these components may devi-ate from the vertical and horizontal, respec-tively. Although terrace morphology can beand often is related to the specific function(e.g. pocket terraces most commonly used forcultivation of trees; the distinct morphologyof irrigation terraces), this is not necessarilythe case and the following discussion shouldapply to all terraces regardless of their specif-ic function.

Reasons for Terrace Construction

As mentioned briefly before, terrace construc-tion can be instigated for a number of reasons,the most common of which are the manage-ment of soil, water, crops and microclimate(Table 6.1). The primary benefits of terraceconstruction are erosion control, water conser-


Table 6.1. Reasons for the implementation of terraces (modified from Hudson 1992: 152)

Soil management

Water management

Crop management

to modify slopeto contain erosion with low inputsto contain erosion with minimal earth-moving on steep slopesto increase depth of soil

to increase effective rainfallto catch and hold all run-offto absorb some run-off with emergency overflowto control reduced run-offto reduce the velocity of run-off and promote infiltration

to provide level areas on steep slopes or ease cultivationto ease harvestingto provide drainage for cropsto modify microclimate to facilitate plant growth

vation and land reclamation. Although anyone of these factors may stimulate terrace con-struction, once constructed, terraces will influ-ence all of these factors to varying degrees(see Treacy and Dene van 1994 for a good dis-cussion of the effects of terracing). Rackhamand Moody (1996:142) note several other pos-sible reasons for terrace construction, whichalthough they may fit into the general descrip-tions above, are not common suggestionsamong most modern soil reclamation books.Some of their alternative suggestions include:soil redistribution to mitigate patchy soils onbedrock surfaces, increased root penetration,and the removal of large stones from thefields. Although a common technique, terrac-ing is not necessarily an ideal solution to ero-sion, water or crop management goals as theircreation may expose less fertile subsoils, or bemore prone to slope failure on some geologicsubstrates (Morgan 1986: 230).

Modern Methods of Terrace Construction

There are numerous ways to construct a ter-race. The most common involve constructinga slope obstacle, and then forming a levelplanting surface by either filling behind it by

hand, borrowing earth from immediately up-slope, or encouraging natural erosion, a pro-cess known as 'controlled erosion'. However,it is possible to form a terrace by merely cut-ting a level platform, or employing a combi-nation of hand filling and controlled erosionto make up the tread fill.

Regardless of the specific method employ-ed, terrace construction always necessitatesthe shifting of earth, and decisions concerningthe morphology of the resulting fields strong-ly control how much earth is moved duringconstruction. The main factors are the widthof the bench, and the slope of the tread. Thevolume of earth moved is proportional to thesquare of the bench (or tread) width (Hudson1992:154-55) so it is beneficial to keep the treadas narrow as cultivation techniques permit;wide treads require higher risers and there-fore a greater volume of earth movement.Outward sloping treads require less earthmovement, but can spill earth over onto thenext lower tread and this can be problematicwith easily erodible soils (Hudson 1992:155).

How a terrace is filled has prominent impli-cations for the visibility of cultural materialsthat were on the slope prior to construction.There are four basic processes associated withterrace construction and use that may affect


the visibility of cultural material: (1) burial ofthe existing surface by the introduction of afill or controlled erosion; (2) controlled erosionor excavation of slope deposits during the fill-ing process; (3) the introduction of new (exo-genous) cultural material if earth is importedto make up the fill; and (4) tillage-related mix-ing.

It is most common for earth to be shiftedlocally on the slope because it is the mostlabour-efficient means. There are also numer-ous records of earth being imported fromnearby (often cited as adjacent alluvial low-lands, e.g. Donkin 1979: 17) to fill terraces.However, this is an incredibly labour-inten-sive method. A brief review of some of themodern methods of terrace construction pro-vides some clues as to how these processesmay influence terrace stratigraphy and thevisibility of pre-existing cultural materials.

Terraces which Bury the Topsail

One of the most common means of terraceconstruction is the formation of a level plant-ing surface by borrowing earth from immedi-ately upslope (cf. Morgan 1986: 232). Thismethod risks reducing soil fertility by usingand shifting less fertile subsoils or minimallyweathered bedrock on top of the pre-existingtopsoil. Generally this process will lead to anover-thickened buried A horizon at the base ofthe terrace fill that is overlain by sedimentderived from the subsoil or bedrock.

Terraces which Conserve the Topsoil

Modern methods often favour the reuse andconservation of the existing A horizon, but it isclear from the stratigraphy revealed by thestudy of relict terraces that ancient culturesdid not necessarily share this preference. One

example of a modern method favouring con-servation of the topsoil is illustrated by PeaceCorps (1986; see also Hudson 1992: 155). Inthis example, the lowest terrace tread is pre-pared with fill created by cutting into the hillup-slope, and then the topsoil is stripped fromthe area of the next higher terrace and used tofill the tread of the lower terrace. Terraces thusconstructed only bury the pre-existing topsoilon the lowest terrace, and shift all the rest ofthe topsoil down one step.

FanyaJuu

This method of progressive terracing is anincremental process that involves the excava-tion of a small ditch parallel to the slope con-tour and the deposition of the spoil from theditch immediately up-slope of the ditch. Theterm fanya juu means 'to throw upwards'(Hudson 1992: 161-63). The bank thus createdbegins to accumulate soil which is moveddownhill by natural erosion, controlled ero-sion, or merely cultivation tillage. Over a peri-od of years, the bank height is increased andthe terrace tread width gradually increases.This method creates a terrace which lacks amasonry wall, and the terrace fills generatedfrom this process will resemble those whichbury the topsoil, unless the A horizon is verythick, in which case it may create an over-thick-ened A horizon. The bank will show a similarstratigraphic progression, but may shift intoless weathered material more rapidly owing tothe near vertical excavation of the ditch.

Bulldozing

Although not mentioned in most soil conser-vation manuals, heavy machinery such asbulldozers are one of the most common mod-ern methods of terrace construction (personal


observation; Rackham and Moody 1992).Terraces constructed in this fashion generallycreate the treads by cutting or scraping awaythe soil on the up-slope side, and then usingthis earth at the end of the terrace to formramps between successive terrace levels,hence most of these fields are of the morpho-logic form Moody and Grove (1990) describeas braided. Some earth may be displaced lat-erally to form a fill in the traditional sense, butmost appears not to be, so these terraces exhi-bit soil profiles that are most complete nearthe riser and become progressively truncatedtoward the rear of the tread.

Cross-channel Terraces

Check dams or cross-channel terraces are gen-erally built across water gathering, concaveslopes such as channels and gullies in order totrap sediment and thereby reduce channeldepth and slope (Morgan 1986: 242). Thesefeatures are nearly always self-filling, and areoften constructed incrementally. Their depo-sits are typically stratified and reflect thenature of sediments transported down thechannel (see Smith and Price 1994, for a pro-file of one such field in Mexico). Unlike mostterraces, this method may permit the develop-ment of numerous buried soils if their use andconstruction occurred gradually over a verylong period of time.

Terrace Stratigraphy

The stratigraphic elements associated withterraces are a direct artefact of their construc-tion method and use, and scrutiny of thesefeatures may inform on these processes, aswell as the nature of the slope soil prior to ter-race construction. Six major components arerecognized (see Fig. 6.1): (1) the pre-existing

surface (generally a palaeosol); (2) the riser;(3) the riser fill; (4) the tread fill; (5) the culti-vation surface; and (6) the post-abandonmentfill.

The Pre-existing Surface

Previous investigations of prehistoric terraceshave found buried A horizons beneath the ter-race deposits and this pre-existing surfacemay be the entire soil or a fragment of it (e.g.Sandor 1997). The nature of the palaeosol mayindicate the completeness of the slope soil atthe time the terrace was constructed. Giventhat some construction methods may haveresulted in truncation of the existing soil,however, care must be taken in evaluating thesignificance of truncated profiles. The highestprobability of palaeosol preservation isbeneath the tread fill adjacent to the riserand/or beneath the riser.

The Riser

The riser may be a masonry wall, an earthenbank, or a cut vertical surface. Its characterdepends on the terrace construction method.In some instances more than one riser may bepresent if the field has been improved or usedover a long period of time (see the illustrationin Williams and Walter 1988: 184). If the riseris an earthen bank, then the stratigraphy ofthe riser may, if distinguishable, indicate howthe riser was constructed. If the terrace wasformed by cutting (as in some bulldozer con-structed fields), then no riser deposit or struc-ture may be present.

The Riser Fill

In some instances the space immediately adja-


cent to and behind the riser is filled withcoarse rubble (Krahtopoulou 1997: 256; Treacyand Denevan 1994: 100). These deposits forma V-shaped wedge immediately behind theriser, and are generally a feature of terraceswith masonry risers. The function of thesedeposits is not clear, but may be associatedwith drainage, disposal of waste created dur-ing the construction of riser masonry walls, ormerely a discrete fill added to fill the spaceformed between the earthen terrace and theriser wall.

The Tread Fill

The tread fill is generally the most significantdeposit in terms of both volume and internalstructure. The character of the tread fill will bea function of the construction method, tillage,and whether the terrace was constructed com-pletely before use, or was formed under culti-vation. Informative aspects of these depositsinclude: (1) their composition; (2) their struc-ture or stratigraphy; (3) their cultural materialcontent (and their temporal depth); and (4)their degree of pedogenic development.

Composition. The sediment that comprises thetread fill will inform on the source of materialused to construct the terrace, and how thatmaterial may have changed during the con-struction or use of the terrace. For instance,terraces that are formed by controlled erosionmay change from redeposited topsoil initiallyto redeposited subsoil, and eventually bed-rock in a sequence that is basically reversedsoil stratigraphy. Earth imported for treadfills, such as alluvium from nearby streams,should differ significantly from the naturalslope deposits and soils.

Structure or stratigraphy. The internal structureor stratigraphy of the tread fill may inform on

the construction process as well. The characterof this deposit is a function of the constructionprocess, tillage mixing, and inputs of materi-als to the tread (most often organic matter).Several kinds of deposits may occur withintread fills and these include: (1) anthropogenicfill earths; (2) natural slope deposits; and (3)soils and buried soils. Treacy and Denevan(1994) contrast 'self filling or accretional ter-races' with those filled by hand (an anthro-pogenic fill earth). Deposits generated bysheetwash or controlled erosion will be thin-bedded and in some cases laminated (Treacyand Denevan 1994: 105; see also Smith andPrice 1994 for an example). If these depositsare cultivated while sedimentation fills theterrace, tillage will most likely obliterate themand little evidence of this process will remainwhereas their preservation is a testament toeither minimal tillage cultivation or lack ofuse of the terrace prior to the completion offilling. Treacy and Denevan (1994) fail to pro-vide any criteria for recognizing hand filledterrace deposits, perhaps because it is difficultto distinguish sediments shifted by hand fromthose altered or mixed by tillage. Buried soilsare also common components within terracetread fills (see Fish 1994: 58 for a graphicexample). Sandor (1997: 238-40) noted thatburied A horizons were commonly encoun-tered within the tread fill of Peruvian terraceshe examined in the Colca Valley. These soilsand intervening fills were interpreted as evi-dence of multiple fill events associated withterrace reconstruction events, the originalsoils buried by terrace construction or addi-tions of organic material by farmers.

Perhaps the most common attribute of treadfills is the poorly sorted matrix-supportedgravelly deposit (see French and Whitelaw1999 for a good description of terraces exhibit-ing this attribute). This kind of deposit resultsfrom haploidization of the tread fill duringtillage, whereby fine matrix and coarse frag-


ments are mixed together destroying any ves-tiges of sedimentary structures or other formsof stratification. Extremely thick mixed zonesindicate either deeply dug soils or cultivationduring progressive formation of the tread fill.Figure 6.2 illustrates the nature of the twomost prominent processes that influence ter-race stratigraphy following the major con-struction event.

Cultural inclusions. Cultural material withinthe tread fill may preserve vestiges of stratig-raphy that inform on the period of construc-tion or the progressive denudation of pre-existing cultural deposits during the filling ofthe terrace. Although tillage may destroyinternal bedding associated with controllederosion, cultural material added to the terraceduring cultivation (such as is widely attrib-uted to manuring—cf. discussion in Millerand Gleason 1994) may provide informationon the period of construction or use of the ter-race. In order to detect if such relict stratigra-phy is present it is critical to examine thestratigraphic occurrence of cultural materialwithin the terrace. At least three basic scenar-ios can be envisioned (and many more combi-nations are possible): (1) incorporation of cul-tural material to the field once the terrace is

Figure 6.2 Processes affecting terrace stratigraphy fol-lowing initial construction.

completed; (2) cultivation and addition of cul-tural material to the terrace while controllederosion is employed to fill the terrace; (3) re-deposition of pre-existing cultural depositsfrom controlled erosion of slope sediments. Inthe first case, presuming that the slope depositdid not contain any artefacts to begin with,then the cultural material in the terrace treadfill will be concentrated in the plough zone atthe top of the tread fill. The depth of theiroccurrence will reflect the depth of tillage andtheir temporal duration will reflect the periodof use. In the second instance, again assumingan artefact sterile slope deposit, the artefactswithin the terrace will date the period of ini-tial use and construction, and crude strati-graphic order may be retained, despite themixing associated with tillage. In this scenariotillage will effectively act as a moving aver-age, mixing assemblages of adjacent periodsbut still providing a vestige of stratigraphicorder. Clearly, this will only work if the fillingprocess is prolonged. The last case (but by nomeans the last possible combination) pre-sumes that the slope deposit contains somepre-existing cultural material, in a crude stra-tigraphic order. In this situation, controllederosion of the slope deposits above the risermay lead to the formation of a reverse cultur-al stratigraphy which may mimic trends in thetread fill deposit composition.

Degree of pedogenic development. The last majoraspect of the tread fill is the degree of pedo-genic development. Prehistoric terraces con-structed several thousand years ago will havepersisted long enough to permit pedogenicalteration of the deposits. The initial attributewill be the formation of a new A horizon inproximity to the tread surface. This horizonwill form naturally, but its development canbe greatly accelerated by manuring. In a nat-ural setting such horizons may form in lessthan a hundred years (Birkeland 1984), but


with anthropogenic additions of organic mat-ter such horizons may form in less than adecade, as my own garden activities havedemonstrated. Subsoil horizons form moreslowly, but even significant B horizons mayform in as little as 1000 years (Krahtopoulou,this volume). Hence it is likely that ancientterraces will exhibit incipient to moderatelydeveloped soils.

Evaluating Terrace Construction Methods:Some Hypothetical Examples

In the field, excavation of trenches perpendic-ular to the long axis of the terrace (basicallyupslope from the terrace wall to the rear of theterrace) and down to bedrock may revealclues to the manner in which the terrace wasconstructed. Figure 6.3 is a hypothetical exam-ple of how eight different construction meth-ods may influence the internal stratigraphy ofmonogenetic terraces. For the sake of clarity,these illustrations do not depict tillage-relatedmixing nor soils that form during cultivation.For this reason they are perhaps a bit over-simplistic. Despite this obvious shortcoming,they are provided to illustrate how variousconstruction techniques will affect the terracestratigraphy. The basic variables at play arethe degree to which the terrace is cut into theslope, and whether the fill deposit is anthro-pogenic or formed by natural slope sedimen-tation behind the terrace wall, as in the case of'controlled erosion'. Table 6.2 summarizes theimplications of each method for archaeologi-cal visibility of cultural materials that werepresent in the soil A horizon prior to terraceconstruction, as well implications for terracestratigraphy which are illustrated in Figure6.3. It is clear from these basic examples thatterrace construction may either enhance orinhibit archaeological visibility and that it isimpossible to discern which has occurred

I. Cut

II. Fill

III. Cut & fill

IV. Deposition

V. Cut & deposition

VI. Fill & deposition

VII. Cut, fill &deposition

VIII. With topsoil conserved

Soil A horizon

Soil B horizon

Cultural fill

Sheetwash- slopewash

Figure 6.3 Hypothetical examples of how constructionmethod may affect terrace stratigraphy.

Key to Symbols


Table 6.2. Implications of terrace construction methods for archaeological visibility and terrace stratigraphy

Method Implications for archaeological visibility Implications for terrace stratigraphy

ICut

IIFill

IIICut and Fill

IVDeposition

VCut andDeposition

• Near total removal of cultural material.• Remaining material will be situated nearthe riser and have relatively high visibility.

• Complete or nearly complete burial ofpre-existing cultural material if the fill isintroduced.• If the fill is derived from the same slope,good visibility may be retained but theoriginal contextual significance lost. Hencethe distribution of cultural material will bean artefact of terrace construction, notprevious occupation.

• If the fill is derived from cutting/borrowingof earth from immediately up-slope, theterrace should retain good visibility of pre-existing material owing to exposure of theA horizon up-slope and redeposition ofmaterial towards the riser.• Visibility could be quite low if the fill islargely derived from the subsoil or bedrock.

• If the sediment comprising the tread fill islocally derived from the A horizon, visibilityshould be good.

• Cutting should truncate the pre-existing soil leavingthe most complete fragments adjacent to the riser.• Sediment created from this process will be usedlocally, most likely as fill to form graded slopes orbraids between adjacent treads.• Ancient terraces constructed by this method may bedifficult to identify owing to the complication oftillage and pedogenesis following construction, whichwould tend to obfuscate the remaining vestiges of theformer soil.

• The introduction of fill earth to form the terrace willmost likely result in the complete preservation of theoriginal slope soil (as a palaeosol) unless some formof A horizon retention program is employed. Thelatter may result in the removal of the A horizon priorto burial by fill.

• Similar to III until erosion cuts into thesubsoil or bedrock, at which point theleading edge of the tread may be buried insterile earth. The truncation of the A horizonin the middle of the tread, however, shouldassure some, albeit diminished, visibility.

• Filling should preserve the original soil beneath andimmediately up-slope of the riser, whereas the cuttingwill truncate the same soil immediately down-slopeof the next higher riser.

• Deposition may preserve complete or nearlycomplete portions of the original soil, especially inproximity to the riser. Whether this soil is preservedup-slope depends upon the thickness of the originalsoil, the width of the terrace, and the depth to whicherosion up-slope occurred during the filling process.• This process can yield a cumulic A horizon if the fillis derived solely from A horizons up-slope• Such fills should retain some semblance of beddingunless cultivation occurred concomitant with sedi-mentation, in which case the entire fill may be mixedby tillage.

• Deposition should preserve the original soil beneathand immediately up-slope of the riser, whereas thecutting will truncate the same soil immediately down-slope of the next higher riser.• The deposit behind the riser may preserve evidenceof bedding at a depth greater than the average tillageif cultivation did not occur concomitant withsedimentation.


Table 6.2 continued

Method Implications for archaeological visibility Implications for terrace stratigraphy

VIFill andDeposition

VIICut, FillandDeposition

VIIIFilling withTopsoilConservation

• Visibility in this scenario is entirelycontingent upon the source of the tread fill.If it is topsoil from up-slope, then visibilityshould be good. Sloping treads, however,together with cultivation, could promotesignificant down-slope movement of culturalmaterial if spill-over occurs.• If subsoil or bedrock is the source, visibilitymay be poor to non-existent.

• Similar to V; exposure of the A horizon inthe middle of the tread should provide goodexposure across part of the tread, but notnecessarily all of it.

• Pre-existing cultural materials will bevisible on the tread of every terrace andburied in the palaeosol beneath the treadfill of the lowest terrace.

• Complete preservation of the former soil may beachieved if fill is derived from elsewhere. Otherwisesome erosional truncation may occur in the up-slopeportions of the tread.• It may be difficult or impossible to identify fieldsconstructed in this fashion because tillage will mostlikely destroy diagnostic depositional attributes.

• Filling should preserve the original soil beneath andimmediately up-slope of the riser, whereas thecutting will truncate the same soil immediatelydown-slope of the next higher riser.• It may be difficult or impossible to identify thedepositional component of the fill because tillage willmost likely destroy diagnostic attributes of thesedeposits.

• This method forms a complete palaeosol beneaththe tread fill in the lowest terrace, and a truncatedpalaeosol beneath every other terrace.• Tread fill should be derived from the subsoil andbedrock.

Note: Discussion pertains to Figure 6.3. These inferences are predicated upon the following assumptions:(1) that the original cultural material was situated in the A horizon at the time the terrace was constructed;(2) that terraces have been maintained and have not collapsed;(3) that terraces are monogenetic and formed during one discrete constructional phase.

from surficial inspection alone. Furthermore,the addition of soil mixing associated withtillage may make it difficult to unravel theconstruction method with certainty.

Methods of Dating Terrace Construction

Dating terrace construction can be problemat-ic and a wide variety of evidence has beenused to estimate terrace age (see Table 6.3). Ingeneral terms, previous efforts to date terraceconstruction have based age estimates upon:(1) cultural material in the area around the ter-races; (2) various aspects of the riser or wallelements; (3) attributes or inclusions derivedfrom the fills and soils which comprise the ter-

race; and (4) the relationship of walls witharchaeological remains of known age. Esti-mates based upon surrounding scatters of cul-tural material or settlements provide no firmlinkage with the terraces in question and mustbe regarded carefully. It is possible to date theriser by architectural style but the fact thatconstruction methods may be copied leads tosome uncertainty in such ages. It is also possi-ble to directly date the rocks by means oflichenometry or cosmogenic isotopes, butboth of these methods have limitations. If therocks used to form the walls were collectedfrom the surface, then these methods mayyield anomalously old ages. Furthermore,lichenometry will only work for rocksfavoured by lichens, and therefore its suitabil-

Table 6.3. Methods of dating agricultural terraces.

Component Method Strength and implication of the derived date Reference

• Their relation or association withancient settlements.

• Age of cultural material found in thevicinity of the terraces.

• Ties to periods of known occupation(especially in regions where pastsettlement was temporally limited or ofshort duration).

• Masonry style and phasing of wallconstruction.

• Lichenometry.

> Cosmogenic isotopes from wall rocks.

Treacy and Denevan1994: 105

• A weak means of dating the fields.• Possibly erroneous unless stratigraphically linked.

• May not bear a direct relationship to the terraces.• Not necessarily associated with terrace construction and use.• Significance depends upon mode of arrival in the field.• Distribution in proximity to fields may have been affected byslope and tillage processes.

• Circumstantial argument.• Cannot be used to date terraces in any way other than toidentify periods when construction could have occurred.

• Wall construction methods may be copied or repeated by latergroups, leading to erroneous age estimates.

• May provide an accurate date for the point of lichen growth.• Difficult to rule out lichen growth that began prior toincorporation in the wall.• Lichens are not common on all rock types.• Studies require some information on rate of growth andtaxonomy.

• May provide an indication of how long rocks have beenexposed.• Can lead to anomalously old ages if rocks used in wall wereinitially gathered from the surface.

Whitelaw 1991;Given et al. 1999

Denevan 1987: 21;Whitelaw 1991

Riser or WallAttributes

SurroundingLandscape

Table 6.3 continued

Component Method Strength and implication of the derived date Reference

•Stratigraphic constraint (e.g. structuresbuilt on top of terrace, or terrace overlainby isochronous deposits such as tephra).

• Ceramic inclusions in the tread fill.

• Ceramics in the palaeosol beneath theterrace wall and tread fill.

• Radiocarbon dating of organicinclusions (scattered charcoal).

• Radiocarbon dating of buried soils.

Degree of soil development.

• Luminescence dating of controllederosion (sheet wash/rill wash) fill.

• Provides a minimum age for the construction event.• Potentially very accurate means of dating terrace constructionas it provides a firm terminus ante quem.

• Significance is conditional upon the manner in which they areincorporated.• Generally provide a maximum age for the field.• Can date from period of use if input as manure or incrementalwaste disposal.• Can be irrelevant to the period of construction or use of fieldwhen incorporated by construction process or controlled erosionof pre-existing deposits.

• Provide a maximum age for the construction of the terrace.• Perhaps one of the most accurate means of dating a terrace.

• Direct age on organic material within the field.• Does not necessarily date use or construction.

• Dates will yield a maximum age for soil burial.• Mean residence time effect will lead to chronometricuncertainty about time of burial.• Can provide support for other forms of data.

• Cannot date the field accurately.• Can be used to support inferences that the field has beenpresent for a significant period of time (one or more thousandyears).

• Dates period of filling.• Samples may not be adequately zeroed (or bleached) resultingin anomalously old ages.

Betancourt et al 1990;Betancourt and HopeSimpson 1992

Malpass 1987;Shea 1987

Malpass 1987

Sandor 1987

Deposit Attributes


ity will vary as a function of bedrock. Datesderived from terrace deposits are problematicowing to two principal factors: (1) that cultiva-tion mixes the deposits (Treacy and Denevan1994: 106); and (2) that inclusions within theterrace may be associated with pre-existingdeposits that comprise terrace fills, use of thefeature or post-depositional processes.

Given the complexity and problems associ-ated with these dating methods, it is clear thatthe best approach is to employ more than onetechnique in order to estimate terrace age. Sec-ondly, attention to stratification, soil develop-ment, stratigraphic distribution and sequenceof sherds, sedimentary character of thedeposits, and nature of and inclusions in thesoil beneath the terrace may provide clues tothe nature and timing of filling and use.Ultimately, such investigations can be bestexpected to provide brackets for terrace con-struction rather than firm dates.

Implications of Terrace Construction forArchaeological Visibility

The limitations imposed upon archaeologicalsurvey by geological processes are well appre-ciated in Greece, as is demonstrated by thefrequent inclusion of geoarchaeological stud-ies in survey projects. Most of these projects,however, have only considered natural andanthropogenically accelerated (indirect) sedi-mentation. As a result, such studies havefocused primarily upon alluvial environments,and less frequently on slopes. The thrust ofthis paper is that terraced slopes are potential-ly as problematic for archaeological surveys asare alluvial valleys.

The manner in which terrace constructionmay influence archaeological visibility is pri-marily a function of: (1) if and where in thepre-existing soil cultural material was present;(2) the construction method employed; and (3)

the amount of post-construction sedimenta-tion. Because these variables are wide rangingand almost impossible to estimate from surfi-cial examination, we should treat terraced landsas if they are dynamic landscapes with thepotential to obscure archaeological visibility.

Given the nature of sloping land, the proba-bility of large settlements being sited on andsubsequently obscured by slope terracing isprobably relatively small. The major problemlies with small farmsteads, special functionsites, and small, early sites which could easilybe buried or destroyed by extensive terracingprograms at any point in time. In a more reg-ulatory oriented environment such as contractarchaeology, this conclusion would necessi-tate pedestrian surveys augmented by someform of subsurface exploration to evaluate theseriousness of the problem. This responsereflects an attempt to tailor survey strategiesto the terrain in order to attempt to compen-sate for limited archaeological visibility.Where deposition is less than a metre, suchareas are generally surveyed in conjunctionwith shovel tests or some other form of key-hole excavation. Where deposition is in excessof a metre, there are presently no adequateand time efficient strategies that will permitexploration of such landscapes and a widerange of compensatory procedures have beenused. Most involve the use of mechanicalexcavation (i.e. a JCB or backhoe), but thiswould be problematic on many terracedslopes, especially steep lands such as are thesubject of this paper. Hence, knowledge of theproblem is key to interpreting the survey ofterraced lands.

Summary

If we continue to assume that all or most ter-races are recent, and infer that sherds havebeen incorporated via modern slope process-


es, then we are potentially missing a signifi-cant part of the archaeological record. Thesites thus missed may not be big ceremonialsites or villages, but rather small, rural familydwellings. Furthermore, given the rugged ter-rain of many Greek landscapes, the imple-mentation of terracing may have had signifi-cant implications for the subsistence base atone or more times in the past. Understandingwhen this occurred is critical to understand-ing settlement of these landscapes, and this isa significant research domain in the under-standing of rural Greek landscape history.Hence, the investigation of terrace history hasthe potential to yield information relevant tothe landscape archaeology of Greece, as hasbeen pointed out by numerous previous stud-ies. Unfortunately, to realize this goal requiressystematic investigation by means of excava-tion because this history generally cannot bereconstructed from surficial examinationalone. Hence, terraced landscapes should beconsidered what they are, dynamic culturallandscapes with a potential to hide or revealevidence of pre-existing occupations.

Acknowledgments

This paper arises from our first field season onthe Kythera Island project (1999), and wewould like to thank Cyprian Broodbank(University College London) for the opportu-nity to participate in the project and for hiscomments and ideas on this paper. We arealso indebted to Vangelio Kiratzi (ResearchFellow, Fitch Lab, British School at Athens) forthe time she spent with us in the field, andassisting us in complying with permit require-ments. We would also like to thank IGME forpermission to perform the field work fromwhich this paper evolved.

Bibliography

Betancourt, P.P. and R. Hope Simpson1992 The agricultural system of Bronze Age Pseira.

Cretan Studies 3: 47-54.Betancourt, P.P., P. Goldberg, R. Hope Simpson and CJ.

Vitaliano1990 Excavations at Pseira: The evidence for the

Theran eruption. In D. Hardy (ed.), Them and theAegean World III: 96-99. London: The TheraFoundation.

Birkeland, P.1984 Soils and Geomorphology. Oxford: Oxford Univ-

ersity Press.Denevan, W.M., K. Mathewson and G. Knapp

1987 Pre-Hispanic Agricultural Fields in the AndeanRegion. BAR International Series 359. Oxford:British Archaeological Reports.

Donkin, R.A.1979 Agricultural Terracing in the Aboriginal New

World. Tucson: University of Arizona Press.Fish, S.K.

1994 Archaeological palynology of gardens andfields. In N.F. Miller and K.L. Gleason (eds.), TheArchaeology of Garden and Field: 44-69. Phila-delphia: University of Pennsylvania Press.

Fish, S.K., PR. Fish and C. Downum1984 Terraces and Hohokam agricultural production

in the Tucson basin, Arizona. In S.K. Fish andPR. Fish (eds.), Prehistoric Strategies in theSouthwest. Arizona State University Anthro-pological Research Papers 33: 55-72. Tempe:Arizona State University.

French, C.A. and T.M. Whitelaw1999 Soil erosion, agricultural terracing and site for-

mation processes at Markiani, Amorgos, Greece:the micromorphological perspective. Gearch-aeology: An International Journal 14:151-89.

Given, M., A.B. Knapp, N. Meyer, T. Gregory, V.Kassianidou, J. Noller, L. Wells, N. Urwin andH. Wright

1999 The Sydney Cyprus Survey Project: an interdis-ciplinary investigation of long-term change inthe north central Troodos, Cyprus. Journal ofField Archaeology 26: 19-39.

Hudson, N.1971 Soil Conservation. London: Batsford.1992 Land Husbandry. London: Batsford.

Krahtopoulou, A.1997 Agricultural terraces in Livadi, Thessaly,

Greece. In A. Sinclair, E. Slater, and J. Gowlett


(eds.), Archaeological Sciences 1995. OxbowMonograph 64: 249-59. Oxford: Oxbow Books.

Malpass, M.A.1987 Prehistoric agricultural terracing at Chijra, in

the Colca Valley, Peru, preliminary report 2. InW.M. Denevan, K. Mathewson and G. Knapp(eds.), Pre-Hispanic Agricultural Fields in theAndean Region. BAR International Series 359: 45-66. Oxford: British Archaeological Reports.

Miller, N.F. and K.L. Gleason1994 Fertilizer in the identification and analysis of

cultivated soil. In N.F. Miller and K.L. Gleason(eds.), The Archaeology of Garden and Field: 25-43.Philadelphia: University of Pennsylvania Press.

Morgan, R.P.C.1986 Soil Erosion and Conservation. London: Longman

Scientific and Technical.Moody, J. and A.T. Grove

1990 Terraces and enclosure walls in the Cretan land-scape. In S. Bottema, G. Entjes-Nieborg and W.van Zeist (eds.), Man's Role in the Shaping of theEastern Mediterranean Landscape: 183-91.Rotterdam: Balkema.

Peace Corps1986 So/7 Conservation Techniques for Hillside Farms.

Washington DC: Peace Corps InformationCollection and Exchange.

Rackham, O. and J. Moody1992 Terraces. In B. Wells (ed.), Agriculture in Ancient

Greece. Skrifter Utgivna av Svenska Institutet iAthen, 4, 42: 123-30. Stockholm: SwedishInstitute at Athens.


Manchester University Press.Sandor, J.A.

1997 Long-term effects of prehistoric agriculture on

soils: examples from New, Mexico and Peru. InV.T. Holliday (ed.), Soz7s in Archaeology: Land-scape Evolution and Human Occupation: 217-45.Washington: Smithsonian Institution Press.

Shea, D.E.1987 Preliminary discussion of prehistoric settlement

and terracing at Achoma, Colca Valley, Peru. InW.M. Denevan, K. Mathewson and G. Knapp(eds.), Pre-Hispanic Agricultural Fields in theAndean Region. BAR International Series 359: 67-88. Oxford: British Archaeological Reports.

Smith, M.E., and T.J. Price1994 Aztec-Period agricultural terraces in Morelos,

Mexico: evidence for household-level agricul-tural intensification. Journal of Field Archaeology21: 169-79.

Treacy, J.M. and W.M. Denevan1994 The creation of cultivable land through terrac-

ing. In N.F. Miller and K.L. Gleason (eds.), TheArchaeology of Garden and Field: 91-110.Philadelphia: University of Pennsylvania Press.

Whitelaw, T.M.1991 The ethnoarchaeology of recent rural settlement

and land use in northwest Keos. In J.F. Cherry,J.L. Davis and E. Mantzourani (eds.), LandscapeArchaeology as Long-Term History: Northern Keosin the Cycladic Islands from Earliest Settlementuntil Modern Times. Monumenta Archaeologica5, 16: 403-54. Los Angeles: Institute ofArchaeology, University of California, LosAngeles.

Williams, L.S. and B.J. Walter1988 Controlled-erosion terraces in Venezuela. In

W.C. Moldenhauser and N.W. Hudson (eds.),Conservation Farming on Steep Lands: 177-87.Ankeny, Ohio: Soil and Water ConservationSociety.

Landscape Exploitation via Pastoralism: Examining the'Landscape Degradation' versus Sustainable EconomyDebate in the Post-Mediaeval Southern Argolid

Hamish Forbes

The most critical step in the degradation of the

Greek environment was the invention of the bull-

dozer7 (Rackham 1996: 42).

Introduction

In this chapter I examine some of the argu-ments surrounding the status of the soils andvegetation of Greece. In particular, I considerthe question of the role that grazing hasplayed in producing the apparently 'degrad-ed' nature of much of Greece's vegetationalcommunities and soils. Within this debate,however, is an underlying current, rarely visi-ble but nonetheless, highly treacherous: theissue of uniformitarianism in archaeology—the assumption that well-documented behav-iours in the present and immediate past canbe read back into the more distant past.

The application of uniformitarian assump-tions concerning human behaviour is a majormethodological issue in archaeology, relatingto fundamental questions about the applica-bility of analogies, whether conscious orunconscious (e.g. Johnson 1999: 54-63; Trigger1989: 362-67, 389-91). Whatever the rights andwrongs of uniformitarian assumptions, it isgenerally accepted that analogies based onthem are more convincing in contexts wherethere is a direct cultural continuity between

the present and the past (Johnson 1999: 61;Trigger 1989: 391)—a situation which pertainsin the context of the present discussion(though cf. Johnson 1999: 61, who suggeststhat it is inapplicable in European archaeolo-gy). Nevertheless, it is a matter of debatewhether the direct historical approach can bereadily applied to Greece in the Classical peri-od, for instance. Its applicability to prehistoricGreece is even more debatable (e.g. Johnson1999: 61-62; Trigger 1989: 362, 364).

Despite the uncertainties both about unifor-mitarian assumptions relating to humanbehaviour and about the direct historicalapproach, the idea that the apparently'degraded' vegetational communities andsoils of present-day Greece can be explainedas being the direct result of overgrazing,excessive wood cutting and other forms ofland abuse is dependent on uniformitarianarguments. The aim of the present contribu-tion is neither to question nor to uphold theuniformitarian approach in archaeology.Rather, I attempt partially to sidestep the issueby examining the actual historical and ethno-graphic evidence concerning the exploitationof a specific uncultivated Greek landscape. Inthis way I set the issue of the use/abuse of the'wild' landscapes of Greece within the moreintegrated cultural context of real, function-ing, rural economies over time.

7

96 Hamish Forbes

Two different strands of uniformitarianthinking are involved in more extreme ver-sions of the idea that Greece's present land-scape represents centuries, if not millennia, ofenvironmental abuse. The first is the attemptto impose ideas derived from temperate zoneecology onto Greece's ancient (and not soancient) past. This approach is perhaps betterdescribed as pseudo-uniformitarianism, andit is allied with what has been described aspseudo-ecology (Rackham 1996). Rackhamhas argued that the idea of ancient Greece as aland of noble forests and crystal fountains,now destroyed by centuries of mismanage-ment, has a history which dates back at leastto the later eighteenth century (1996: 27-28).These ancient forests and fountains owedmore to the France of Marie Antoinette, thenatural environment of their author, the AbbeBarthelemy (1788), and his fertile imagination,than to the reality of ancient Greece. This kindof uniformitarian thinking is a hardy perenni-al (Rackham 1996: 27-28). Less than 20 yearsago a forestry specialist declared that the lackof perennial grasses in Mediterranean plantcommunities was a result of pastoral misman-agement (Thirgood 1981: 68). In actuality,perennial grasses typify temperate zone graz-ing environments: in Greece 'natural' pasturesrecognizable in temperate European terms arelargely restricted to Alpine-type environ-ments for good climatic reasons, not becauseof mismanagement (Rackham 1983: 322;Forbes 1998: 21).

The other uniformitarian issue concerns theextent to which it is possible to apply obser-vations on present-day pastoralism to antiqui-ty and prehistory in the Mediterranean.Recent thinking has tended to suggest thatmodels based closely on present-day or evenmediaeval practices of transhumant pastoral-ism have little relevance to Classical antiquityor prehistory (Halstead 1987; Hodkinson1988; Garnsey 1988; though cf. Isager and

Skydsgaard 1992: 96-104). Nevertheless,claims that the degraded state of soils andvegetational communities in Greece stemfrom inappropriate grazing and wood-cuttingregimes over long periods are heavily depen-dent on uniformitarianist assumptions.

There are many variants of the basic themethat past mismanagement of landscapes inGreece explains the present state of soils andvegetation. Many are carefully documentedand cautiously worded (e.g. van Andel et al.1997). At their crudest, however, the argu-ments involved tend to use both the uniformi-tarian approaches identified here. A writerobserves that most vegetational communitiesin the uncultivated landscape are radicallydifferent from 'natural' plant communitiesfound 'back home' in the temperate zone,being usually both much lower-growing andoften more widely spaced than in temperatezone woodland. Signs of recent grazing activ-ity and woodcutting are then observed withinthese environments. The writer then jumps tothe conclusion that if humans and their herbi-vores had been excluded from these commu-nities, 'natural' forests would have grown upproviding fine stands of trees which couldhave been felled for good quality timber. Iexplore the Western cultural background tothis contradictory thinking elsewhere (Forbes1997: 187-91), but it is salutary to rememberthat English forests have been subject to regu-lar wood cutting and to grazing by bothdomesticated and managed 'wild' animals forhundreds of years (Rackham 1989: 1-19) yetare perceived by most who visit them to be'natural' environments, untrammelled by reg-ular interference from humans or their ani-mals. Also, those of us who live in the tem-perate zone and are tempted to decry thewanton mismanagement of the Greek land-scape should be aware of Rackham's (1983:346) observation on the vegetation of theBoeotian landscape, which has changed less

Landscape Exploitation via Pastoralism 97

in 2500 years than that of England in the last1000 years and that of New England in the last180 years.

Although anthropologists and others arebecoming increasingly aware of the utility ofsources, both oral and documentary, from pre-vious centuries (e.g. Bennet and Voutsaki1991; Cherry et al 1991; Davis 1991; Sutton1991), as yet forms of land use even two tothree centuries ago are not well understood(Forbes 2000). A clear and detailed under-standing of the ways in which exploitation ofthe Greek landscape has occurred in theimmediate past and how that exploitation hasor has not changed over recent centuries is anessential first step to understanding the histo-ry of postglacial soils and vegetation. It is alsoan essential prerequisite to uniformitarianistattempts to explain present-day Greek vegeta-tional communities and soils. Yet this strandin the investigation of the Greek past is still inits infancy. Thus we find ourselves in a situa-tion where there is little detailed understand-ing of the ways in which the Greek landscapewas exploited in the eighteenth, nineteenthand earlier twentieth centuries. At the sametime the complex details of research on thesubject relating to the later twentieth centuryare easily ignored by those who prefer to paintstark pictures on broad canvasses. This is evi-dently a fragile academic environment.

The Problem

The existence in present-day Mediterraneanlandscapes of 'degraded' vegetation typesand abundant examples of erosion is a leitmo-tif of Western scholars who concern them-selves with the region (e.g. Goudie 1990: 53-54; Vita-Finzi 1969). Of particular significancein this context has been the growing realiza-tion that substantial erosional events or phas-es have occurred in postglacial—including

historical—times. Although Vita-Finzi (1969)originally argued for a climatic cause for aphase or phases of postglacial erosion in theMediterranean, a number of observers haveblamed them on the destruction of the naturalvegetation by human communities and theiranimals (e.g. Thirgood 1981: 112, 153-55;Jameson et al. 1994: 325; van Andel et al 1997:54-55). The activities of pastoralists, andabove all the effects of the appetite of the goat,are still regarded by many as the main causeof degraded vegetational communities and oferosion, although others have questioned theextent to which human activities have beenthe cause of environmental changes in theAegean area (e.g. Moody 1997).

The idea that pastoralists, heedless of thelong-term consequences, over-exploit or oth-erwise damage their grazing if given thechance, is a long-standing one (e.g. Goudie1990: 52-53; Thirgood 1981: 66, 73, 75). Over-exploitation and resultant environmentaldegradation are considered particularly likelywhere, as in much of Greece, grazing lands arecommon lands (e.g. Thirgood 1981:123-49; fora much more balanced picture, see Koster1997). Moreover, the assumption that, allthings being equal, resources in commonownership will be over-exploited by all, hasbeen given something approaching scientificstatus (if not holy writ) in Garrett Hardin'sinfluential The tragedy of the commons'(1968).

It is often forgotten that the apparent causalrelationship between 'degraded' vegetationand eroded landscapes, on the one hand, anddestructive pastoralism, on the other, has hadvery little genuinely scientific study to sub-stantiate it. Hardin's original article, andmany of the publications which followedfrom it (e.g. Hardin and Baden 1977) are pri-marily political position-papers, not detailedecological or ethnographic studies. (Signifi-cantly, in England, Rackham (1989: 140) notes

98 Hamish Forbes

that the historical record demonstrates thatthe removal of commoners' rights generallyresulted in the rapid destruction of forests bytheir private owners, not their long-term pre-servation.) Other observers of Mediterraneanlandscapes, having noted the existence ofwhat they perceive as 'degraded' plant com-munities (though cf. Rackham 1983; 1996: esp.18-22) and the regular presence of flocks ofsheep and/or goats in them, have jumped tothe conclusion, without further investigation,that the one is directly caused by the other.

A small number of studies have examinedin some detail the relationship between capro-vines and the Mediterranean plant communi-ties on which they graze and browse,although they have usually been very limitedin both spatial and chronological terms. Someof the more detailed studies of the effects ofexploitation by humans and their animals onlocal vegetational communities have ques-tioned the received wisdom (e.g. Kolars 1966;Forbes and Koster 1976; Koster 1997). But theyseem to have been largely ignored by thosewho believe, with Thirgood (1981: 66) that'the proclivity of the Mediterranean shepherd... is too well established to require documen-tation'. Likewise, Rackham's (1983) detailedstudy of the effects of grazing and wood cut-ting on vegetational communities and erosionis still being discounted in favour of environ-mental mismanagement explanations for his-torical period erosion.

An example of the ways in which uniformi-tarian assumptions are employed even incarefully researched publications can be seenin van Andel et al (1997: 50). They presentobservations of erosion, made during the1970s and 1980s in one extremely small valleyin the southern Argolid; evidence for the pres-ence of grazing caprovines and of abandonedterraces is used to explain the observed ero-sion. These observations are then used toexplain widespread evidence of erosion in

classical antiquity, based on the unsupportedclaim that terraces 'have been used since atleast Classical times' (van Andel et al 1997:47). Foxhall (1996), however, has demonstrat-ed that clear historical or archaeological evi-dence for terracing in classical antiquity isvery hard to discover, and that trenchingrather than terracing was the land manage-ment system probably practised on the estatesof the elite—the group from whom we deriveour historical records. She concludes that it 'isunrealistic to expect that the present system ofland management can be projected back infi-nitely into the past on a substantial scale'(Foxhall 1996: 64).

The idea that pastoralists mismanage com-mon pastures to secure and indeed maximizetheir short-term advantage has been under-mined by a number of recent studies. First,although over two thirds of the pastures inmodern Greece are publicly owned (i.e. 'com-mons'), a large amount of grazing is privatelyowned (Koster 1997: 172). Not only are sub-stantial areas of uncultivated pasture underprivate ownership, but also a significant pro-portion of pastures consists of privatelyowned arable land which is exploited bycaprovines when not under cultivation. Theexistence of these extensive areas of privatelyowned grazing and the extent to which live-stock management strategies incorporate theuse of privately owned arable land are oftenforgotten by those who assume that pastoral-ism is associated almost exclusively with theuncontrolled exploitation of common land(see Forbes 1995: 330-31). In addition thereseems to be a correlation between high ruralpopulation densities and those areas in whichprivate pasture is found (Koster 1997: 167-70,173). In other words, the areas where popula-tion pressure on resources is highest, andtherefore the risk of over-exploitation by over-grazing is the greatest, tend to be those inwhich grazing is privately owned, and there-


fore not open to an unmanaged free-for-all.It is likewise a fact that claims of misman-

agement of pastures are frequently made bythose who have little understanding of theexigencies of making a living from the uncul-tivated landscape. These, therefore, are theviews of outsiders, some of whom are con-vinced that indigenous peoples are incapableof understanding their own long-term advan-tage or, indeed, of responsible management oftheir own environments. Those who adhere tothe 'Mediterranean-as-goat-desert' school ofthought too often seem to consider indige-nous populations as no different fromnaughty and ignorant children who do notknow how to behave; hence the complex waysin which local communities interact with theirown environments have rarely been consid-ered (Forbes 1997).

The present paper investigates two issues.(1) It views the exploitation of the uncultivat-ed landscape from the point of view of localpopulations: of the people who depend ontheir local environments for fuel and othernecessities, and the pastoralists whose liveli-hoods are dependent on the well-being of theiranimals. This viewpoint is in marked contrastto that of Western pundits, whose livelihoodsdepend on eye-catching headlines and doom-laden statements, but who have no such life-long relationship with these landscapes. (2) Itemphasizes the need to investigate genuinehistorical data, rather than making historicalassumptions based largely on synchronicdata—if anything at all. Both of these issueswill be considered within the context of a spe-cific area of Greece: the Southern Argolid.

The Indigenous Viewpoint

In this section I indicate that, although theuncultivated landscapes in the SouthernArgolid region are, technically speaking, com-

mon lands, there is no unregulated free-for-allin which pastoralists maximize their short-term gains at the expense of long-term sus-tainability Indigenous thinking and practiceswithin the community ensure that numbers ofanimals do not exceed the capacity of thelandscape to support them. In addition, Ishow that the majority of the rural population,and not merely pastoralists, has been depen-dent on the products of the region's commonlands. Nevertheless, the amounts of materialremoved in the last two or three decades havebeen well within sustainable levels and, fur-thermore, the tradition of grazing 'territories'in common land areas has ensured that pas-toralists have been able to control the levels ofexploitation of these landscapes both for graz-ing and for non-pastoralist purposes.

The Southern Argolid is part of the driestsector of Greece: mean rainfall is under 400mm annually (Kayser and Thompson 1964:1.03), and generally restricted to the wintermonths from November to March, even innormal years. Annual agricultural returns(NSSG unpublished) from Kranidhi, the mainurban centre of the region, between 1968 and1977 are especially instructive because duringthat time-span they included comments onrainfall and other aspects of weather condi-tions. In 50% of these years rainfall wasreported as less than adequate for arable agri-culture. Weather conditions for livestock werereported as favourable only once in thedecade, and growth of natural pasture wasreported as regular in only two years out ofthe ten.

The factors of low rainfall with a relativelyhigh interannual variability, allied with lowwinter temperatures in upland locations andthin soils on the hillsides, combine to restrictplant growth severely on non-cultivated hillslopes. Such conditions are, at least potential-ly, tailor-made for producing imbalancesbetween the rate of exploitation of the plant

100 Hamish Forbes

communities in these locations and plantregeneration rates. Were this to occur, wewould expect the result to be rapid environ-mental 'degradation' and potential erosion ofremaining slope soils. The extent to whichsuch imbalances do actually occur, or haveoccurred in the past, must depend on whetherthe local population has achieved some formof long-term regulation of the exploitation ofthese rather delicately balanced ecologicalsystems.

Greek agricultural service figures for theearly 1980s indicated the importance of pas-tor alism in the region's rural economy at thattime. Although only 1.6% of agrarian enter-prises were entirely dependent on pastoral-ism, 68% of the total agrarian enterprises keptsheep and/or goats. Furthermore, in the early1980s some 38,400 sheep and goats weregrazed in the Southern Argolid annually, giv-ing a density of over 90 caprovines for everysquare kilometre of the region's surface (0.9caprovines/ha). Only 27% of this figure com-prised the flocks of transhumant pastoralists;hence approximately three quarters of thesheep and goats in the region were herded byfarmer-shepherds employing a variety of stra-tegies to combine agriculture with the man-agement of their caprovines (see Koster 1977,especially chs. 4 and 7, for a detailed discus-sion of some of the varied strategiesemployed). This represents a figure of justunder five head of stock for every individualactively engaged in agriculture (sensu laid) inthe region or 2.3 caprovines for every humanin the region (based on unpublished agricul-tural service figures). Given the relatively lowlevels of plant growth and high numbers ofcaprovines supported by the local environ-ment, the region is an ideal location to evalu-ate ideas concerning overgrazing and envi-ronmental degradation.

In the Southern Argolid, as elsewhere inrural Greece, use of the uncultivated land-

scape as grazing for sheep and above all forgoats, has provided an income to those whomight otherwise not have been able to sup-port themselves on the produce of their ownagricultural land. Detailed studies by Koster(1977) indicated that various social, economicand biological mechanisms have ensured thatgrazing resources have not been over-exploit-ed. In particular, two features stand out in hiswork. One is the interdigitation of local-levelagro-pastoralists with seasonally a'ppearingtranshumants. In this coastal area transhu-mant pastoralists have traditionally broughttheir flocks to graze during the winter months.During the summer months, when plantgrowth is limited by high temperatures andlack of rainfall, the transhumant pastoralistshave removed their flocks to the high moun-tain pastures of the central Peloponnese. Thishas ensured a reduced level of grazing duringthe hot dry summer months when grazingresources are under most pressure (Koster1977: 72-82; Koster and Koster 1976).

The other feature which has controlled ani-mal numbers has been the explicit under-standing that herders' livelihoods havedepended on an adequate plane of nutritionfor their flocks, which in turn has dependedon controlling the numbers of animals grazingparticular areas of the landscape. The mostimportant product for these pastoralists' ani-mals has traditionally been, and still is, milk.Without adequate nutritional levels milk pro-duction will drop drastically. Because of this,as Koster (1997: 163) notes, shepherds do notneed to measure their pastures or the condi-tion of their animals visually: they feel thequality of the grazing with their hands—viamilking. Declining nutrition levels will rapid-ly be translated into declining milk yields. It isfor this reason that, despite the appearance ofcheap hired labour, via the employment ofillegal Albanian immigrants, they are stillreluctant to entrust the vital monitoring task


of milking to those who do not have a deeppersonal investment in their flock's success.

For these reasons overgrazing, and the con-sequent drop in animals' plane of nutrition,has not been an option: undernourished ani-mals produce little of the milk on which thepastoralist's economy depends. In addition,the period when natural grazing is frequentlyat its poorest occurs during the ewes' gesta-tion period. A poor level of nutrition in preg-nant ewes at this time leads to pregnancy tox-aemia ('twinning disease')—a fatal disease.Herders in this region have therefore ensuredthat they do not overstock their pastures. Thisis accomplished via the maintenance of exclu-sive grazing territories (Koster 1977: 181-85;Koster and Forbes 2000), a practice which hascontrolled other, non-pastoral, aspects of theexploitation of the uncultivated landscape inthe process. Each herder has aimed to main-tain a sustainable number of animals in hisown grazing territory and ensure that otheraspects of the exploitation of the uncultivatedlandscape would not materially affect thegrazing. Traditionally, when animal numbersexceeded the ability of the mountain grazingto support them adequately (e.g. in severedroughts), there was a significant increase instock mortality (Koster 1977: 167-68), whichkept stock numbers in a rough equilibriumwith grazing resources.

The ways in which local-level mechanismswork to regulate the numbers of caprovinesexploiting local natural grazing areas in theSouthern Argolid and elsewhere in Greecehave been documented by Koster (1997: 160-67). The system of maintaining de facto grazingterritories (the thesis system) in areas in which,de jure, an open-access free-for-all is allowed,is one way in which pastoralists ensure thelong-term viability of their grazing. Related tothe maintenance of territories is the existenceof density-dependent violence between com-peting pastoralists. Since pastoralists cannot

afford to increase stock numbers beyond thelevel at which their grazing can support them,any expansion must be at the expense ofherders with neighbouring territories. Thelosers in this system must of necessity reducestock numbers if they are forced to reduce thesizes of their territories (Koster 1997: 162).Obviously, stealing another pastoralist's ani-mals or grazing, or destroying his property,are not part of the state-sanctioned legal sys-tem, but these behaviours, like the de factoestablishment of grazing territories in contra-diction of the de jure rules on the exploitationof the commons, act as effective local-levelchecks on the overstocking of pasture.

The assumption, held by some, that theuncultivated landscapes of Greece are onlyused by a relatively small number of shep-herds and goatherds for pasturing their ani-mals is likewise erroneous. As I have des-cribed elsewhere (Forbes 1997) the exploita-tion of the uncultivated 'wastes' of Greece forall manner of products has been a vital ele-ment in rural economies until after the SecondWorld War. In particular, it has traditionallyprovided fuel both for domestic use and forsmall-scale commercial enterprises involvedin the production of charcoal and lime. Never-theless, despite the very substantial amountsremoved annually from the environment, esti-mates of the level of wood consumption forone southern Argolic community in the 1970s(Forbes and Koster 1976: 120-21) indicatedthat it was well below the one ton per hectareper year suggested by Rackham (1983: 326) asa sustainable level in Boeotia. Prior to theSecond World War the use of wood cut fromthe uncultivated landscape had been impor-tant in the region for the commercial produc-tion of charcoal and lime (see below).Nevertheless, the cutting of wood for theseactivities was only conducted with the per-mission of the shepherds and goatherds inwhose grazing territories it was carried out. In

102 Hamish Forbes

this way pastoralists were able to control thelevel of exploitation of their de facto territories,thus ensuring that their flocks' milk yieldswere not adversely affected (Forbes 1997: 200-201; Koster and Forbes 2000).

Historical Data

In this section I evaluate the 'tragedy of thecommons' argument via the examination ofhistorical documentary sources. So far I haveconsidered the situation concerning theexploitation of the uncultivated landscapes ofthe Southern Argolid largely from a synchron-ic viewpoint. The question arises: can theseessentially synchronic data be 'read back' fur-ther into the past? Part of the answer lies in adiachronic analysis of populations of capro-vines and humans in the region over a periodof 300 years, indicating that the region's indi-genous populations have been able to exploittheir environments in a sustainable way overthe long term.

While the evidence presented in the previ-ous section can tell us how uncultivated land-scapes have been exploited at sustainable lev-els over the last few decades, it does not tell uswhether it used to work in the same way inthe past. For example, in the 1960s, the Greekgovernment introduced regulations on theexploitation of the uncultivated landscape ofthe Southern Argolid for commercial activi-ties, such as the production of charcoal andlime. Thus by 1970 only 10 tonnes of brushwas cut for lime-burning—and none there-after—and production of charcoal fromnon-cultivated sources averaged 44 tonnes inthe years 1970 and 1971. These figures con-trast with recorded peaks of production in theinter-war years: over 2100 tonnes of brush cutfor lime burning in 1938 and almost 800tonnes of charcoal produced in 1937(Dhasonomio unpublished: the figures

include public and private land). The datatherefore suggest that government regulationswere instrumental in bringing about the com-plete cessation of the use of fuel cut from com-mon lands for commercial lime and charcoalmanufacture in the 1970s. In the past, centralgovernment in Greece was less able to exer-cise effective control over matters such aswood cutting and grazing. Does this meanthat in earlier periods there was no effectivemeans of controlling the exploitation of theuncultivated landscape?

In reality, the apparent picture of successfulgovernment intervention is largely illusory.Forest Service personnel emphasized that, inthe Southern Argolid at least, these regula-tions were largely superfluous. The use of fuelcut from the uncultivated environment forcharcoal and lime burning stemmed from theneeds of the local community for cash. Thework was highly unpleasant and sometimesdangerous; usually only the poorest in theregion participated in these activities. By thelater 1960s a surge in Greece's economicdevelopment was well under way. Othersources of paid employment, especially in themerchant marine and the tourist and con-struction industries, provided far better finan-cial reward for less effort. At the same time,the appearance of bottled gas cooking stovesand the ready availability of factory-producedcement and lime drastically undermineddemand for locally produced charcoal andlime. Forest Service personnel therefore attrib-uted the demise of these local industries lessto government regulations than to economicdevelopment.

The unpublished data provided by theForest Service indicate that the production ofcharcoal and lime burning was only impor-tant for a short period during the twentiethcentury. Its rise and demise closely parallelsthat of the commercial exploitation of anotherproduct of the 'wild' landscape, on the penin-


sula of Methana not far from the SouthernArgolid: wild oregano (Clark 1997). Like theMethanites' exploitation of wild oregano, limeand charcoal burning in the Southern Argolidrose to importance in the inter-war years ofthe twentieth century, as an opening for thosewith few other economic options for obtainingcash, at a time of increasing monetization ofthe Greek economy The Forest Service recordsindicate a rapid rise in the production both ofcharcoal and of lime during the 1930s. Thusthe amount of scrub cut for lime burning fromcommon lands rose from 4250 zighia in 1932 tomore than 12,800 zighia in 1938 (1 zighi = 100okadhes-, 1 oka = 1.27 kg). The annual produc-tion of charcoal from common lands rose from9500 okadhes in 1932 to 18,000 okadhes in 1937(the figure for charcoal was not recorded in1938).

The economic catastrophe of Greece's entryinto the Second World War, and the subse-quent occupation of the country by AxisAlliance powers, put a temporary halt to theseactivities. Nevertheless, it is important tostress that their cessation owes nothing to alack of fuel brought about by over-exploita-tion. Continuing economic problems associat-ed with the Greek Civil War following the endof the Second World War meant that charcoaland lime burning did not resume on a com-mercial scale until the beginning of the 1950s.The entry in the Forest Service log for 1958indicates no charcoal produced from commonlands but brush cutting for lime at mid-1930slevels. Nevertheless, just over a decade laterthese activities on the common lands had allbut ceased, because of economic changes, notbecause of the effectiveness of governmentregulations nor because of the exhaustion ofthe fuel resources.

It is evident therefore that the commercialexploitation of common lands in the SouthernArgolid was a short-term episode, respondingto particular circumstances of industrializa-

tion and monetization in the economy It can-not be directly applicable to previous cen-turies in uniformitarianist terms. It is likewiseevident that external, central government reg-ulation of these activities in this case was notprimarily responsible for their cessation.Rather, as noted in a previous section, effec-tive control over the exploitation of the fuelresources on the common lands was wieldedat the local level by the shepherds in whosegrazing areas the fuel-cutting occurred.

What of these shepherds controlling their defacto grazing territories? Is there any evidencethat they have resisted the urge to maximizetheir short-term advantage by over-grazingtheir territories? Koster's study of twentieth-century archival sources for a large village inthe Southern Argolid, where pastoralism wasa major economic activity, indicate that thetotal number of sheep and goats pasturedwithin the community remained remarkablystable from 1940, when the document began,to the mid-1970s (Koster 1977: 171-73).Stocking densities for this community havelargely stayed between 1.0 and 1.5 head ofstock per hectare of grazing land over a periodof some 35 years, despite the upheavalscaused by the Second World War, the ensuingcivil war and major changes in the Greekagrarian economy Only in one year did thestocking density rise above 2.0 head per ha(Koster and Forbes 2000).

Other historical documents provide us withlonger-term historical data on caprovine num-bers in the Southern Argolid. Although theylack the detail of Koster's archival sources,they nonetheless indicate that local-levelmechanisms for maintaining sustainable num-bers of animals were not merely a twentieth-century phenomenon. The following discus-sion is dependent on Table 7.1, which tabulatesthe data derived from historical sources.

A cadastral survey undertaken in the lastdecade of the seventeenth century is the start-

104 Hamish Forbes

Table 7.1. 300 years of pastoralism in the Southern Argolid

1980s1960s1930s1

1880s1860s2

1690s3

Total sheep+ goats

38,40041,075

50-55,000

23,900

Sheep + goatsper ha of total

area

0.900.94

1.15-1.26

(1.11)1.15-1.57

Goats per haof uncultivated

area

0.810.84

0.88

Sheep + goatsper head ofagriculturalpopulation

<5.0

(9.4)

Sheep + goatsper head of

totalpopulation

2.3

(4.4)14.5

Olivetrees

660,000

163,000

9,000

1. Informed estimate.2. Data for 1860s derived from nomos of Argolis and Korinthia, not eparkhia.3. Administrative area significantly smaller in 1690s than in the nineteenth and twentieth centuries.

ing-point of the sequence (Topping 1976;Forbes 2000). The catastico ordinario lists theregion's population and agrarian wealth fromthe period of the short-lived Second VenetianOccupation of the Peloponnese (1684-1715).Particularly noteworthy are the very lowhuman population—less than one quarter ofthe population in the region in 1928 (Pana-ghiotopoulos 1985: 247-48)—and the evidenceof ruined houses and uncultivated land. Theexistence of land lying uncultivated is sup-ported by a letter to Venice from the Venetianchief administrator: 'they love idleness andcultivate only enough for their precise needs,being inclined to maintain large flocks inorder to get a return without sweat' (Topping1976: 95).

Despite the administrator's statement,which may relate primarily to his frustrationat not being able to raise more taxes, analysisof the catastico ordinario suggests that theregion was not suffering from over-exploita-tion by pastoralists at that time. Thus the fig-ures for caprovines per ha of the total surfacearea indicate that stock numbers were beingcarefully controlled. Three groupings of com-munities identified as being independentgrazing nexuses (Forbes 2000) show a narrowrange of 1.15-1.57 head per ha, which is close-ly comparable to the estimated figure for the

inter-war years of this century of 1.15-1.26sheep and goats per ha of surface area.

If we examine the cadastral data in moredetail we see further similarities with thetwentieth century. Thus the overall figure of0.88 goats per ha of non-cultivable land areain the Southern Argolid in the 1690s is notmuch higher than the figure for goats per haof uncultivated land documented for the 1960s,which stands at 0.84 per ha. It is also virtuallyidentical to the 1982 figure for the village ofDhidhima, the main pastoralist community inthe region today: 0.87 goats per ha.

On the other hand, because of the lowhuman population, the figure of 14.5 capro-vines per person is extremely high. All theavailable evidence suggests that pastoralismwas the region's primary cash-generatingactivity, with agriculture being practisedlargely for subsistence (Forbes 2000). Thus atthat time there were less than 9000 olive treesin the region, compared to c. 660,000 in the1980s. On the other hand, some wealthy land-owners kept large numbers of beef cattle: twomonasteries, the region's main landowners,accounted for 330 between them. The exis-tence of beef cattle in large numbers at thattime is an important indirect indicator that,despite relatively large numbers of capro-vines, grazing was anything but under pres-


sure. Cattle need significantly better qualitygrazing than either sheep or goats.

Some 160 years after the period of the cata-stico ordinario we have a further documentwhich helps us to follow the course of pas-toralism in this part of Greece. In 1867Alexandras Mansolas published the results ofstatistical researches into the population andeconomy of Greece at that time. The data aremore general than we might wish, but arenevertheless suggestive. For the province ofArgolis and Corinthia, of which the SouthernArgolid was a part, we have an overall figureof 1.11 caprovines per ha. This generalized,province-wide figure is very close to that fromErmionis at the end of the seventeenth centu-ry. Significantly, however, stock numbers perhead of population had dropped since the ear-lier date, to c. 9.4 per head of populationengaged in agriculture. This is because humanpopulation levels had risen very sharply fol-lowing the end of the Greek War of Indepen-dence some 30 years previously. Caprovinenumbers, however, were apparently beingkept pegged at a relatively steady level whichwould not endanger the area's grazing poten-tial (Mansolas 1867: 89-90).

The fact that the human population inc-reased dramatically over approximately 150years, but that there was no correspondingincrease in caprovine numbers, emphasizesthe effectiveness of the local controls over ani-mal numbers. This is not at all what one wouldexpect from a population engaged in unregu-lated exploitation of the local environment.

Accurate figures of the caprovine popula-tion of the region in the twentieth centuryprior to the 1960s do not seem to exist.However, an informed estimate by a memberof staff at the region's agricultural service vet-erinary clinic suggests a figure of 50,000-55,000 during the inter-war years. This wouldgive figures of 1.15-1.26 sheep and goats perha of surface area. Unpublished Forest Service

statistics and accompanying explanatorynotes (Dhasonomio unpublished) for 1968support informant comments that stock num-bers dropped after the Second World Waralthough the decline was apparently not spec-tacular: 0.94 caprovines per ha of total surfacearea, and 0.84 goats per ha of grazing land.Statistics for the 1980s also confirm informantcomments on a continuing decline: 0.88 capro-vines per ha of the total area and 0.73 goatsper ha of grazing land (NSSG unpublished).

These bald statistics, while less than ideal,nevertheless give us an idea of the trend ofcaprovine numbers in the Ermionis region ofGreece over a time-span of some 300 years.Historically the time-span includes three inva-sions, two violent revolutions, phases of pro-found economic stagnation as well as majoreconomic development, and above all, mas-sive human population change. During muchof these three centuries, the amount of controlover the exploitation of the local environmentprovided by centralized government has beenminimal. These conditions could be consid-ered ideal for encouraging short-term think-ing and consequent over-exploitation of thenatural grazing resources of the area, especial-ly as the local human population expandedand more mouths had to be fed, bodiesclothed, and so on. Yet the documentary andoral history data indicate that this is preciselywhat the local population did not do.

Of course, the figures indicating that thenumbers of caprovines in the region haveremained remarkably constant over three cen-turies do not by themselves indicate a sustain-able level of grazing. However, as notedabove, detailed estimates of the amount ofvegetation removed by a rural community inthe Southern Argolid via grazing and woodcutting (Forbes and Koster 1976) indicatedthat the levels concerned were well withinsustainable limits. Since a substantial portionof the vegetation removed was for domestic

106 Hamish Forbes

fuel, it is likely that this element in the equa-tion would have been lower in previous cen-turies when population levels were lower.

Geological evidence supports these obser-vations on sustainability. Van Andel et al.(1997: 53-54—see also Jameson et al 1994: 412-14) have identified only thin and sparsely dis-tributed alluviation dating to the period fromthe mediaeval Prankish occupation to the pre-sent day (the 'Kranidi alluvium'). The age andhistory of the different deposits belonging tothis phase seem to vary in the differentdrainage basins, although there are problemswith dating the deposits (van Andel et al.1997: 53). It is suggested that these variableages and histories, even within the restrictedarea of the Southern Argolid, are 'related tovery local economic and demographicchanges, which bring about changes in landuse and, especially, the neglect of terracing'(Jameson et al. 1994: 414). Significantly, themost recent examples quoted are all wellaway from the sectors of the landscape inwhich grazing and fuel-cutting have tradi-tionally been concentrated (van Andel et al.1997: 53-54). Nevertheless, it is also admittedthat the period of the deposition of theKranidi alluvium includes the period of theclimatic changes brought about by the LittleIce Age. Although some of the small patchesof alluvium seem to postdate this climaticoscillation, van Andel et al. (1997: 54) acceptthat some examples might be attributable to it.Jameson et al. (1994: 194), however, place theoverall level of mediaeval and postmediaevalalluviation into a longer term perspective bynoting that 'Extrapolating generously, humanactivity has aggravated the erosion of themountains, but the landscape of Classicaltimes was not much richer in soil than it isnow'. Thus the varied lines of evidence pre-sent a strong case that the amounts of grazingand wood cutting conducted in the unculti-vated landscapes of the Southern Argolid over

the last 300 years have not been beyond sus-tainable levels.

How did the expanding local populationcontinue to support itself via the land ifexpansion of pastoralism was not an option?One answer can be seen in the right hand col-umn of Table 7.1. Over the course of 300 yearsthere has been a massive expansion of olivecultivation in the region. The reasons behindthe phenomenon seem to have been complex,but involved economic, demographic, socio-logical and technological factors. These arediscussed in some detail elsewhere (Forbes1993); here it is sufficient to note that,although some expansion of olive cultivationinto previously uncultivated areas wouldhave taken place, much would have occurredon land already under cultivation. Thus olivecultivation allowed new avenues for intensifi-cation of production without competingdirectly with pastoralism. In addition, duringthe nineteenth and earlier twentieth centuriesmost of those who owned substantial areas ofolives were from the wealthier end of thesocial spectrum. Oral testimony, backed up byForest Service records (Dhasonomio unpub-lished), suggests that households at the poor-est end of the social spectrum were the mostlikely to engage in the exploitation of theuncultivated landscape for forest products:lime-burning and charcoal burning in particu-lar, but also resin tapping. As noted above,these activities were only conducted with theexplicit permission of the pastoralists withinwhose grazing territories they occurred, andpastoralists would only give that permissionif their own livelihoods were not jeopardized.

Conclusions

Ethnographic studies indicate that, under theconditions prevalent in the Southern Argolidin the twentieth century, short-termism, over-


exploitation and degradation of grazingresources have simply not been an option. Inaddition, historical data indicate that local-level practices to maintain grazing animalsand grazing resources have been in existencein the region for at least the last 300 years. It isnot possible to move on from these observa-tions to say that no pastoralists over-exploittheir resources anywhere: what I have pre-sented here is merely a case study. Neither doI wish to claim that the very clear evidence ofHolocene period erosion in Greece cannothave an anthropogenic cause and thereforemust be explained climatically. Nor yet do Iwish to claim knowledge of levels of grazingand wood cutting in the Southern Argolidprior to the late seventeenth century.

This study does, however, indicate thatthose who maintain that pastoralists generally—or even Greek pastoralists generally—areconstitutionally incapable of exercising self-restraint in the exploitation of common graz-ing resources must provide more evidencethan the forestry specialist who stated:

In modern times, grazing is the most obvious, and

probably the most significant force in the destruc-

tion of the Mediterranean landscape. The effects of

over-grazing may be seen throughout the basin:

degraded soils, active erosion and plant denuda-

tion are apparent almost everywhere. Over many

hundreds of square miles, soils are trampled and

rock outcrops visible, perennial grasses are deplet-

ed and replaced by annuals and less palatable

herbs. Trees and bushes are browsed to cushions or

drastically lopped for fodder. Too often, genera-

tions of overgrazing by goats have stripped great

tracts of land of the plants which hold the soil

together, churning the topsoil to powder; wind and

rain complete the devastation, transforming great

expanses to desert wasteland (Thirgood 1981: 68).

Acknowledgments

Ethnographic fieldwork, from which some ofthe data in this article are derived, was con-ducted in the summers of 1980,1982 and 1983,with funding from Stanford University andthe University of Liverpool. I gratefully ack-nowledge the help and encouragement ofSusan Sutton, with whom I shared a fruitfulsummer of 'commando anthropology' in theSouthern Argolid, and the support of MichaelJameson, director of the Stanford UniversityArgolid Exploration Project. I also gratefullyacknowledge the help of numerous secretariesof the koinotites and dhimoi in the EparkhiaErmionidhos.

In addition, I am most grateful to MichaelJameson and Peter Topping for respectivelymaking available to me, and allowing me touse, a transcript of the catastico ordinario forthe Southern Argolid. Thanks also go toCharles Frederick and Paul Halstead for invit-ing me to present the paper on which thiswork is based to the Aegean studies roundtable in Sheffield, and for their consideredfeedback on a previous draft. It goes withoutsaying that none of the above are responsiblefor any of the views presented here.

Bibliography

Andel, T. van, C. Runnels and K. Pope1997 Five thousand years of land use and abuse in

the southern Argolid, Greece. In P.N. Karduliasand M.T. Shutes (eds.), Aegean Strategies: Studiesof Culture and Environment on the EuropeanFringe: 33-59. Lanham: Rowman and Littlefield.(Reprinted from Hesperia 55 (1986): 103-28.)

Barthelemy, J.J.1788 Voyage du jeune Ancharsis en Grece. Paris.

Bennet, J. and S. Voutsaki1991 A synopsis and analysis of travellers' accounts

of Keos (to 1821). In J.F. Cherry, J.L. Davis and E.Mantzourani (eds.), Landscape Archaeology asLong-Term History. Northern Keos in the CycladicIslands from Earliest Settlement until Modern

108 Hamish Forbes

Times: 365-82. Los Angeles: Institute of Arch-aeology, University of California, Los Angeles.

Cherry, J.F., J.L. Davis and E. Mantzourani1991 Introduction to the archaeology of post-Roman

Keos. In J.F. Cherry, J.L. Davis and E. Mantz-ourani (eds.), Landscape Archaeology as Long-TermHistory. Northern Keos in the Cycladic Islands fromEarliest Settlement until Modern Times; 351-64.Los Angeles: Institute of Archaeology, Univer-sity of California, Los Angeles.

Clark, M.1997 Wild herbs in the marketplace: gathering in

response to market demand. In P.N. Karduliasand M.T. Shutes (eds.), Aegean Strategies: Studiesof Culture and Environment on the European Fringe:215-36. Lanham: Rowman and Littlefield.

Davis, J.L.1991 Contributions to a Mediterranean rural archae-

ology: historical case studies in the OttomanCyclades. JMA 4: 131-216.

DhasonomioUnpublished. Annual log of forest managementin Ermionis eparkhia. Held in Greek NationalForest Service office, Kranidhi.

Forbes, H.A.1993 Ethnoarchaeology and the place of the olive in

the economy of the southern Argolid, Greece. InM.-C. Amouretti and J.-P. Brun (eds.), La produc-tion du vin et de I'huile en Mediterranee. BCHSupplement 26: 213-26.

1995 The identification of pastoralist sites within thecontext of estate-based agriculture in ancientGreece: beyond the 'transhumance versus agro-pastoralism' debate. BSA 90: 325-38.

1997 A 'waste' of resources: aspects of landscapeexploitation in lowland Greek agriculture. InP.N. Kardulias and M.T. Shutes (eds.), AegeanStrategies: Studies of Culture and Environment onthe European Fringe: 187-213. Lanham: Rowmanand Littlefield.

1998 European agriculture viewed bottom-side up-wards: fodder- and forage-provision in a tradi-tional Greek community. Environmental Arch-aeology I: 19-34.

2000 The agrarian economy of Ermionidha around1700: an ethnohistorical reconstruction. In S.B.Sutton (ed.), Contingent Countryside: Settlement,Economy and Land Use in the Southern Argolidsince 1700: 41-70. Stanford: Stanford UniversityPress.

Forbes, H.A. and H.A. Koster1976 Fire, axe and plough: human influence on plant

communities in the Southern Argolid, Greece.

In M. Dimen and E. Friedl (eds.), Regional Varia-tion in Modern Greece and Cyprus: toward a Pers-pective on the Ethnography of Greece. Annals of theNew York Academy of Sciences 268: 109-26.

Foxhall, L.1996 Feeling the earth move: cultivation techniques

on steep slopes in classical antiquity. In G.Shipley and J. Salmon (eds.), Human Landscapesin Classical Antiquity. Environment and Culture:44-67. London: Routledge.

Garnsey, P.1988 Mountain economies in southern Europe.

Thoughts on the early history, continuity andindividuality of Mediterranean upland pastor-alism. In C.R. Whittaker (ed.), Pastoral Economiesin Classical Antiquity. Cambridge PhilologicalSociety Supplement 14: 196-209.

Goudie, A.1990 The Human Impact on the Natural Environment.

3rd edn. Oxford: Basil Blackwell.Halstead, P.

1987 Traditional and ancient rural economy in Medi-terranean Europe: plus c.a change? JHS 107: 77-87.

Hardin, G.1968 The tragedy of the commons. Science 162: 1243-

48.Hardin, G. and J. Baden

1977 Managing the Commons. New York: W.H.Freeman.

Hodkinson, S.1988 Animal husbandry in the Greek polis. In C.R.

Whittaker (ed.), Pastoral Economies in ClassicalAntiquity. Cambridge Philological SocietySupplement 14: 35-74.

Isager, S. and J.E. Skydsgaard1992 Ancient Greek Agriculture: An Introduction.

London: Routledge.Jameson, M.H., C.N. Runnels and T.H. van Andel

1994 A Greek Countryside: The Southern Argolid fromPrehistory to the Present Day. Stanford: StanfordUniversity Press.

Johnson, M.1999 Archaeological Theory: An Introduction. Oxford:

Basil Blackwell.Kayser, B. and K. Thompson

1964 Economic and Social Atlas of Greece. Athens:National Statistical Service of Greece.

Kolars, J.1966 Locational aspects of cultural ecology: the case

of the goat in non-western agriculture. Geogra-phical Review 56: 577-84.


Koster, H.A.1977 The Ecology of Pastor alism in Relation to Changing

Patterns of Land Use in the Northeast Peloponnese.PhD dissertation, University of Pennsylvania.

1997 Yours, mine and ours: private and public pas-ture in Greece. In P.N. Kardulias and M.T.Shutes (eds.), Aegean Strategies: Studies of Cultureand Environment on the European Fringe: 141-86.Lanham: Rowman and Littlefield.

Koster, H.A. and H.A. Forbes2000 The 'commons' and the market: ecological

effects of communal land tenure and marketintegration on local resources in the Mediter-ranean. In S.B. Sutton (ed.), Contingent Country-side: Settlement, Economy and Land Use in theSouthern Argolid since 1700: 262-74. Stanford:Stanford University Press.

Koster, H.A. and J.B. Koster1976 Competition or symbiosis? Pastoral adaptive

strategies in the Southern Argolid. In M. Dimenand E. Friedl (eds.), Regional Variation in ModernGreece and Cyprus: toward a Perspective on theEthnography of Greece. Annals of the New YorkAcademy of Sciences 268: 275-85.

Mansolas, A.1867 Politiografikai Pliroforiai peri Elladhos. Athens:

Ethnikon Typografion. (Reprinted 1980. Athens:Vivliopolio Dhionisiou Noti Kara via.)


derstood? In P.N. Kardulias and M.T. Shutes(eds.), Aegean Strategies: Studies of Culture andEnvironment on the European Fringe: 61-78.Lanham: Rowman and Littlefield.

NSSGNational Statistical Service of Greece. Annualreturns of estimated agricultural production, bycommunity. Held in koinotis (community)offices.

Panaghiotopoulos, B.1985 Plithismos kai Oikismoi tis Peloponnisou. 13os-18os

Aionas. Athens: Historical Archive, CommercialBank of Greece.

Rackham, O.1983 Observations on the historical ecology of

Boeotia. BSA 78: 291-351.1989 The Last Forest: the Story of Hatfield Forest.

London: J.M. Dent.1996 Ecology and pseudo-ecology: the example of

ancient Greece. In G. Shipley and J. Salmon(eds.), Human Landscapes in Classical Antiquity.Environment and Culture: 16-43. London: Rout-ledge.

Sutton, S.1991 Population, economy, and settlement in post-

revolutionary Keos: a cultural anthropologicalstudy. In J.F. Cherry, J.L. Davis and E. Mantz-ourani (eds.), Landscape Archaeology as Long-TermHistory. Northern Keos in the Cycladic Islands fromEarliest Settlement until Modern Times: 383-402.Los Angeles: Institute of Archaeology, Univer-sity of California, Los Angeles.

Thirgood, J.V.1981 Man and the Mediterranean Forest: a History of

Resource Depletion. New York: Academic Press.Topping, P.

1976 Premodern Peloponnesus: the land and the peo-ple under Venetian Rule (1685-1715). In M.Dimen and E. Friedl (eds.), Regional Variation inModern Greece and Cyprus: toward a Perspective onthe Ethnography of Greece. Annals of the NewYork Academy of Sciences 268: 92-108.

Trigger, B.C.1989 A History of Archaeological Thought. Cambridge:

Cambridge University Press.Vita-Finzi, C.

1969 The Mediterranean Valleys. Cambridge: Cam-bridge University Press.

8

Land Use in Postglacial Greece: Cultural Causesand Environmental Effects

Paul Halstead

The relationship between landscape changeand land use has attracted the interest of bothpalaeoecologists and archaeologists. The paly-nological and geoarchaeological records docu-ment dramatic alterations to the landscape ofGreece during the postglacial period (e.g.Bottema 1994; van Andel et al 1990) and therecan be little doubt that these changes resultfrom the interaction of natural and anthro-pogenic processes. Human land use has thusbeen widely cited as a destructive agent bypalaeoecologists, while inferred changes invegetation or land forms have been treated asevidence for the nature and scale of agricul-tural or pastoral land use (e.g. Bottema 1982;Willis and Bennett 1994; van Andel et al 1986).Models of early land use have, in turn, beendeployed in attempts to understand suchdiverse aspects of human culture as Neolithicsettlement patterns (e.g. van Andel et al. 1995)and residential strategies (e.g. Kotsakis 1999),Bronze Age settlement nucleation (Gamble1982) and regional redistribution (e.g.Renfrew 1972; 1982), and Neolithic-Early IronAge upland colonization (e.g. Sampson 1992;Halstead 1991) and artefactual style zones(Cullen 1984; Kilian 1973).

Unfortunately, it is often hard to disentangleanthropogenic causes of landscape changefrom natural processes, such as climaticchange, rising sea-level, tectonic activity, eco-

logical succession, or expansion of plantspecies from glacial refuges. Neither anthro-pogenic nor natural processes have left whol-ly unambiguous signals in the palynologicaland alluviational records, so that mostattempts to infer causality have revolvedaround temporal and spatial patterning inlandscape change (Endfield 1997; alsoFrederick 2000). Climatic forcing tends to beinvoked if landscape changes appear to becontemporary over a wide area, while anthro-pogenic causation is favoured if landscapechanges coincide with cultural phenomena:for example, with alterations in the densityand distribution of settlements or, more enig-matically, with the chronological periods (LateBronze Age, Roman, and so on) into whichprehistorians and historians conventionallydivide the past (e.g. Bottema 1982).

Two major weaknesses of this approach areobvious. First, correlation is often assumedrather than demonstrated. In Greece, absolutedates for palaeoecological sequences tend tobe sparse and so practitioners may be reducedto circular reasoning based on similaritiesbetween palynological cores or geoarchaeo-logical sections. Dating evidence for archaeo-logical sites is usually more abundant, but thechronological resolution of surface surveydata may still be too coarse to detect signifi-cant short- or medium-term shifts in patterns

Land Use in Postglacial Greece I I I

of settlement or land use. Secondly, correla-tion is not the same as causation (Buckland etal. 1997), because of the complexity of naturaland cultural processes and of the relationshipbetween them (e.g. Frederick 2000). Becauseof threshold effects in environmental systems,there may be a time lag between natural oranthropogenic cause and its effect on the land-scape. Likewise, the sensitivity of the land-scape to such forces varies in space, in accor-dance with local and regional variation ingeology, relief, aspect, altitude, latitude and soon. Moreover, the severity of anthropogenicimpact on the landscape depends not only onthe amount of land use, which might beinferred very crudely from settlement patterndata, but also on the types of land use prac-tised. For example, the introduction of terrac-ing has been invoked to account for the incon-sistent relationship during the Bronze Agebetween alluvial deposition and settlementdensity in the hilly southern Argolid (vanAndel et al. 1986). The type of land use prac-tised is, in turn, related to human settlementand demography in complex ways which fur-ther obscure the relationship between culturalcauses and environmental effects (e.g Cherryet al 1991; Davis 1991; Whitelaw 1991;Acheson 1997).

This last issue, neglected in some previousliterature, is the initial focus of this paper. Tothis end, alternative models of land use areexplored, together with their possible rela-tionship with other aspects of human cultureand their likely impacts on the palaeoecologi-cal record of landscape change. Currentdebate on the nature of early agro-pastoralland use in Greece is then reviewed and, inconclusion, some implications for futureinvestigation of changing landscape and landuse are briefly discussed.

Land Use in Postglacial Greece: ResearchStrategies and Alternative Models

Analyses of charred plant remains and animalbones from excavated settlements present afairly consistent picture of the range of cropand livestock species raised by Neolithic andBronze Age farmers (e.g. Payne 1985; Hansen1988; Halstead 1994: 204-205, table 1; 1996: 28-29, table 1) and also sporadic insights into theways in which these species were managed,but fall far short of a clear picture of the natureand scale of early agro-pastoral land use. Forthe Late Bronze Age, Linear B texts providesome detail on patterns of land use, but onlyin so far as this was of administrative interestto the palatial elites (Halstead 1992a). For earlyhistoric Greece, classical writers may revealmore of contemporary agronomic theory andcultural values than of agricultural practice,while potential bioarchaeological evidence haslargely been neglected. In consequence, whe-ther implicitly or explicitly, rival reconstruc-tions of early land use have relied heavily onthe accommodation of the patchy and ambi-guous available data to competing models.

One influential model has been that of 'tra-ditional' (pre-mechanized) Mediterraneanland use, as described by historians and geog-raphers (e.g. Semple 1922; Braudel 1975;Grigg 1974). Two distinctive and inter-relatedfeatures of such traditional farming were: (1)extensive growing of cereals, in alternationwith bare (i.e. cultivated) fallow, in the low-land plains; and (2) 'transhumant' movementsof large flocks of sheep and goats betweenwinter pastures in the lowlands and summerpastures in the high mountains. These twincharacteristics have been widely regarded asnecessary or optimal responses to the long-term constraints and challenges of Mediterra-nean climate and topography and, on thesegrounds, have been extrapolated to later pre-history and early history (e.g. Jarman et al.

112 PaulHalstead

1982; Isager and Skydsgaard 1992). Recentexamples of sedentary herding (in both theuplands and lowlands) and of intensive con-tinuous cropping of cereals and pulses, how-ever, show that traditional farming is not anecessary response to Mediterranean environ-ment (Halstead 1987a). Closer examinationsuggests that extensive cereal agriculture andtranshumant pastoralism are only optimalresponses to the Mediterranean environmentunder specific historical conditions.

Extensive agriculture has been characteristicof large land holdings, on which the groundwas prepared for sowing by plough oxen,weeds were controlled by bare-fallowing(again using plough oxen rather than humanlabourers), and cereals were grown in prefer-ence to labour-intensive pulses. In this way,despite modest area yields, cereals weregrown on a large scale with minimal invest-ment of human labour, so that a large surpluswas available for sale to urban consumers.Large amounts of human labour were onlyneeded at harvest time, when gangs of reaperswere temporarily mobilized. Traditionalextensive agriculture is thus based on sharpinequalities of land ownership, whereby somecan maintain plough animals and use largetracts of land unproductively but profitably,because many others with little or no land pro-vide a market for staple grains and a source ofseasonal harvesting labour (Karavidas 1931;Vergopoulos 1975; Halstead 1995).

Mobile pastoralism has likewise been char-acteristic of large-scale, specialized herders,who produced cheese, wool or textiles and,latterly, lambs or kids for the urban marketand bought cereal staples from lowland farm-ers. The extent of dependence on the market ismade clear by the negative impact on thesepastoralists of competition from imports ofwool and cheese, in recent decades, and oftextiles, during the Industrial Revolution.Seasonal mobility between lowland winter

and upland summer pastures served twoends. On the one hand, it enhanced the quali-ty of nutrition and productivity of livestock.On the other hand, by securing access to largeexpanses of natural pasture rather than culti-vated or collected fodder, it made possible theextensive management of large flocks withmodest human labour. In winter, theseherders depended on the fallow fields of low-land arable estates to provide the consolidatedblocks of rich pasture essential for efficientlyrunning large flocks of milk-sheep. Followingthe nineteenth-twentieth century AD dissolu-tion of such estates in lowland land reform,numbers of transhumant animals declinedsharply, giving way to more intensive hus-bandry of smaller and more sedentary flocks(Karavidas 1931; Campbell 1964; Vergopoulos1975; Koster 1977; Psikhogios and Papapetrou1984; Nitsiakos 1985; Chang 1992).

Traditional extensive cereal agriculture andmobile pastoralism enjoyed a symbiotic rela-tionship (Karavidas 1931) and both were dep-endent on the existence of urban markets andon markedly inegalitarian patterns of landownership. In Braudel's terms, they are notexemplars of a timeless Mediterranean longueduree but are contingent on a medium-term,historical conjuncture and, as such, should notbe extrapolated uncritically to the distant past.

An alternative model of land use can be dis-tilled from the activities of recent small-scalefarmers, who cultivated modest areas of landand kept modest numbers of livestock, large-ly for domestic consumption (du Boulay 1974;Psikhogios 1987; Chang 1994; Forbes 1976;1982; 1998; Blitzer 1990; Halstead 1990;Halstead and Jones 1989; Jones et al 1999).Such farmers typically used more intensivemethods of husbandry (manual weeding, reg-ular manuring, cereal-pulse rotation, stall-feeding and so on) and raised a relatively bal-anced mixture of both crop and livestockspecies, partly to secure a broad range of

Land Use in Postglacial Greece 113

products, partly to make full use of householdlabour and land, and partly to minimize therisk of total subsistence failure. Such farmers,of course, were also products of a particularhistorical conjuncture. On the one hand, small-scale farming activity was often the result ofdeliberate circumscription by wealthier land-or herd-owners (e.g. Karakasidou 1997). Onthe other hand, the economic viability ofsmallholders was often dependent on period-ic employment, whether as seasonal labourersin extensive agricultural enterprises or as arti-sans within the urban market economy. None-theless, the intersection between small-scale,intensive and diversified farming may offer auseful model for a putative, prehistoric-earlyhistoric 'domestic mode of production', pre-ceding or peripheral to the appearance ofmarkedly unequal access to the means andfruits of production.

This heuristic opposition between exten-sive/specialized and intensive/diversifiedfarming masks a great deal of variability inhusbandry practices, with complex implica-tions for the relationship between land useand landscape change. Nonetheless, there areseveral reasons why, other things being equal(e.g. for a given size of human population anddegree of settlement dispersion/nucleation),extensive systems of land use might beexpected to leave a stronger palaeoecologicalsignal than intensive systems.

First, extensive crop husbandry is primarilygeared to minimizing labour costs rather thanmaximizing area yields and so, for a givenlevel of grain production, is likely to involveannual sowing of a larger area than would beneeded under intensive husbandry. Extensivehusbandry also tends to involve regular culti-vated fallowing, to control weeds, and so islikely to entail annual tillage (for sowing andfallowing combined) on a much larger scalethan would intensive husbandry, which tendsto be characterized by crop rotation with

infrequent fallowing. Extensive crop hus-bandry should thus be more amenable todetection in the palynological and geoarchae-ological records because it would entail agreater overall scale of clearance and tillage,respectively, than intensive husbandry. Secon-dly, the cultivated fallow commonplace inextensive cultivation leaves more of the land-scape bare of vegetation, and so vulnerable toerosion by wind or rain, for more of the year.

Thirdly, because cultivated land is oftendivided into units of a size which can be tilledand sown in a day, individual land parcelstend to be larger under extensive plough-agri-culture than under intensive hoe-cultivation.Extensive crop husbandry also tends to bemore specialized and so favours the devotionof large contiguous blocks of land to the samecrop sown at the same time, whereas small-scale husbandry typically involves a broadrange of crops and these are often scattered inspace to reduce the risk of wholesale failure.Likewise, with animal husbandry, contiguousblocks of land, sown and harvested at thesame time and subdivided by insubstantialboundaries, greatly facilitate grazing by largeherds and, in recent decades, both large arableestates and village communities have oftencoordinated crop-growing regimes for thisreason. Conversely, small-scale mixed farmersare more likely to favour plot boundarieswhich impede the movement of livestock andso allow them to be enclosed rather than herd-ed while grazing. Such walled and hedgedboundaries tend to harbour trees which areoften managed as a source of fresh leafy fod-der or dried leafy hay, but such provision isusually practicable only for modest numbersof livestock. Thus, extensive land use, botharable and pastoral, tends to create largeblocks of cleared land, relatively amenable tothe transport of 'anthropogenic indicator'pollen and relatively vulnerable to erosion.Conversely, intensive cultivation is character-

114 PaulHalstead

ized by more fine-grained cultural landscapesin which both pollen transport and erosionshould be impeded by surviving vegetationcover. The planting of trees and constructionof field terraces, which involve investments ofhuman labour more compatible with inten-sive than extensive farming, similarly createmore fine-grained and stable landscapes. Such'conservationist' practices have certainly beenundertaken widely by small-scale farmerswithin living memory and the hilly regions ofGreece, of which they are typical, were rela-tively free of large agricultural estates duringthe nneteenth and twentieth centuries AD(McGrew 1985; Psikhogios 1987; see alsoSilverman 1968).

Fourthly, although livestock are kept allyear round in all parts of Greece, the higheststocking densities have typically beenachieved by specialized pastoralists, whomove large flocks seasonally between comple-mentary summer and winter pastures. Inmountainous areas such as the Pindos, thetree-line is now advancing upwards with therecent contraction in transhumant pastoralism(e.g. Halstead 1991), making it clear that theextensive summer pastures of the high moun-tains have partly been created and maintainedby grazing pressure. This pressure has largelybeen exerted by seasonally transhumantflocks, because livestock resident year-roundin the high mountains must be stall-fed overwinter and so their numbers are limited by thelabour-costs of collecting leafy hay and otherforms of winter fodder (Halstead 1998a). Theshredding of trees for leafy hay plainly playeda major role in shaping the recent mid-altitudevegetation of the Pindos, but as part of amosaic of small fields interspersed with man-aged and unmanaged trees (Halstead andTierney 1998). Thus, specialized pastoralism,by suppressing tree cover on watersheds,might be expected to accelerate highland ero-sion, while a lowered tree-line might well lead

to reduced arboreal:non-arboreal pollen ratiosin a nearby, high-altitude coring site. Conver-sely, sedentary mixed farmers would again beexpected to create a less detectable, mid-alti-tude mosaic of open patches and woodland,while the impact of shredding for leaf fodderon pollen rain would depend on such detailsof management as whether or not the crownsof trees were cut (cf. Rasmussen 1990: 79).

In the lowlands, the contrasting influence oflarge- and small-scale animal husbandry onthe cultivated landscape has already beennoted. The extent to which grazing transformsuncultivated parts of the landscape depends, ofcourse, on the number and type of animals,and the duration, season and frequency oftheir presence. The effects of overgrazing arerapidly apparent to herders in declining milkyields or weight loss (e.g. Forbes 1997; Koster1997) and, contrary to the popular assumptionthat herding inevitably involves overgrazing,recent pastoralists in Greece have adopted anumber of formal or informal mechanisms forcontrolling access to pasture (e.g. Nitsiakos1985; Chang 1992; Koster 1997; Forbes thisvolume). Nonetheless, grazing and browsinghave plainly influenced the composition andstructure of vegetation, as has herders' peri-odic use of burning to create pasture which ismore palatable and nutritious or available at amore critical time of year (e.g. Turrill 1929;Rackham 1983). The extent to which grazingpressure or burning may be evident in thepalaeoecological record partly depends on thepalynological visibility of different plant com-munities, but the abundance of microscopiccharcoal in pollen cores may provide an inde-pendent measure of burning (see Atherdenthis volume).

Although herding activity is highly vari-able, and its impact on the landscape complex,a broad contrast can again be drawn betweenlarge- and small-scale animal husbandry. Inthe recent past, only large herds (sedentary or


transhumant, owned by pastoralists or richarable farmers) have tended to make muchuse of distant rough grazing lying beyond thelimits of cultivation, while small householdflocks have usually been run in the intersticesof the cultivated landscape. Thus, for a givenoverall number of livestock, a few large herdsare more likely than numerous small herds torange outside the cultivated sector and so totransform also the uncultivated part of thelandscape. Because the uncultivated parts ofthe landscape are often relatively steep, with arelatively thin soil covering, extensive herdingis also more likely to trigger severe erosionand, in consequence, radical degradation ofvegetation. Incendiary interventions aimed atimproving pasture quality are probably attrib-utable, for the most part, to those managinglarge herds, because herders of small groupsof animals can usually find adequate patchesof good grazing (e.g. along field edges, or byseasonal streams). Overall, therefore, exten-sive systems of land use, pastoral as well asarable, seem more likely than intensive sys-tems to have left their mark in the palaeoeco-logical record.

Early Agro-Pastoral Land Use in Greece:The Intensive Mixed Farming Model

A definitive diachronic account of the chang-ing nature of agro-pastoral land use in theAegean area would be premature and, any-way, beyond the scope of this paper. Instead abrief outline of previous arguments willattempt to highlight the key assumptionswhich underpin them and to identify areas ofuncertainty or contention.

From the Early Neolithic (seventh millenni-um BC) onwards, archaeobotanical and arch-aeozoological evidence from habitation sitesis strongly dominated by remains of domesticcrops and livestock, leaving little doubt that

some sort of agro-pastoral farming supportedmost of the archaeologically visible popula-tion (Halstead 1989). The relative importanceof crops and livestock to subsistence or landuse cannot be inferred directly from such evi-dence, but consideration of the size and distri-bution of human population allows this issueto be approached indirectly.

From a fairly early stage of the Neolithic, themost frequent form of settlement in the morefertile, and apparently most heavily populat-ed, lowland areas of eastern mainland Greeceseems to have been some form of Village',housing between several tens and a few hun-dreds of persons. In at least one region, east-ern Thessaly, long-lived settlements of thissort were spaced as closely as 2-3 km apartfrom the Early Neolithic (Perles 1999), whilethe patchy Neolithic evidence for season(s) ofoccupation is perhaps most economicallyinterpreted in terms of year-round habitation(Becker 1999; Halstead 1999). Local communi-ties of such size and permanence, and region-al populations of such density, could havebeen supported by crop production on a rela-tively modest scale, but a major contributionto subsistence from livestock would arguablyhave required intensive 'dairy' managementand/or the keeping of very large numbers ofanimals. Mortality evidence for the common-est domestic animals, sheep (or goats), sug-gests a 'meat' or mixed purpose strategy ofmanagement rather than intensive dairying(Halstead 1989; cf. Payne 1973; Halstead1998b), while the difficulty of detecting evi-dence for Neolithic clearance in the admitted-ly coarse palynological record from largebasins (Bottema 1982; Willis and Bennett 1994)does not favour belief in animal keeping on avery large scale. During the Bronze Age (thirdand second millennia BC) and early historicalperiod (first millennium BC), as a settlementpattern of villages interspersed with largercentres progressively characterized most of

116 PaulHalstead

lowland Greece, the nutritional dependenceof most of the population on crops is likely tohave become even more inevitable (e.g.Foxhall and Forbes 1982).

Nonetheless, livestock are likely to havebeen of considerable economic, cultural andsocial significance to such crop-growing com-munities: as a form of 'indirect storage'against the periodic failure of staple crops(Halstead 1993); as a vehicle for the accumula-tion and display of wealth and status (Changand Koster 1986); and as means of negotiatingand reaffirming social ties through theexchange of livestock and use of meat in hos-pitality (e.g. Halstead 1992b). The cultural sig-nificance of livestock in the Neolithic is per-haps mirrored by the frequent occurrence ofanimal figurines alongside representations ofhumans, houses and various domestic facili-ties and furniture (e.g. Toufexis 1994). For theLate Bronze Age and early historic period, thisis confirmed more explicitly by artistic andtextual evidence for the consumption ofdomestic animals in elite-sponsored feastsand sacrifices (e.g. Vanschoonwinkel 1996;Killen 1994; Jameson 1988). Late Bronze Agetexts also imply that woollen textiles, wovenin palatial workshops, played a role in themarking and perhaps negotiation of social sta-tus (Killen 1985) while, for the early historicperiod, the keeping of horses was an indicatorof high status (Hodkinson 1988; 1991).

As regards the scale and intensity of landuse, the diversity of crops (in particular, therelatively balanced representation of cerealsand pulses) and livestock in Neolithic andBronze Age assemblages from Greece is morereminiscent of recent small-scale, intensivehusbandry than of extensive farming(Halstead 1994; 1996; cf. Forbes 1976). The linkbetween the diversity of crops and livestockand the intensity and scale of husbandry ispartly based on assumptions of economicrationality (e.g. efficient use of land and

labour; risk avoidance) and so must be extrap-olated to the past with caution. The diversityof the bioarchaeological record contrasts strik-ingly, however, with the textual evidence fromLate Bronze Age southern Greece for highlyspecialized palatial involvement in farming.Palatial intervention in grain crops was appar-ently restricted to the large-scale productionof one or two types of cereal, using palatialoxen for tillage and dependent labourers forharvesting. Palatial interest in livestock wasconcentrated on the management of tens ofthousands of wool sheep, run in individualflocks of up to several hundred head. Thus,there is unambiguous evidence of large-scale,extensive agriculture and specialized animalmanagement in the Late Bronze Age, but per-haps restricted, as in the recent past, to a pow-erful minority, while the bulk of the popula-tion continued to practise small-scale, inten-sive husbandry (Halstead 1992a). More directevidence for the nature of arable farming,from the analysis of weed ecology, is so farrestricted to Late Bronze Age Assiros Toumba,a possible crop storage centre in northernGreece (Jones 1987), where growing condi-tions of cereals seem, in part, to have resem-bled those characteristic of intensive horticul-ture (Jones 1992; cf. Jones et al. 1999; Bogaardet al. this volume). Although a single site is aninadequate basis for generalization, this studylends support to the argument that extensiveagriculture was not practised universally inLate Bronze Age Greece. Bioarchaeologicalevidence is extremely sparse for the early his-toric period, but written sources again implythat extensive agriculture and large-scaleherding were restricted to the wealthier ele-ments in society and co-existed with intensivemixed farming by those less well-off(Hodkinson 1988; also Forbes 1995; Hordenand Purcell 2000).

Thus it is argued that early agro-pastoralland use was initially dominated by small-


scale mixed farming, with elements of exten-sive agriculture and large-scale herding beingadopted by part of the population during thelater Bronze Age and early historic period, butnot demonstrably from an earlier date. Thismodel of changing land use is consistent withincreasing evidence during the later BronzeAge for marked social inequality—a corollaryof recent extensive husbandry It is also con-sistent (if extensive farming is more amenablethan intensive farming to palynological detec-tion—see above) with the first unambiguoustraces during the second millennium BC ofanthropogenic impact, including both cropsand 'disturbance' indicators, in regional-scalepollen diagrams (Bottema 1982; Jahns 1993;Zangger et al 1997).

Van Andel and Runnells (1995) have sug-gested that Early Neolithic cultivation inGreece exploited naturally irrigated and fertil-ized floodplains. The predominant late winter(snow-melt) floods of major north Greekrivers, however, may have been incompatiblewith the probable autumn germination seasonof early cereal crops (Hillman 1981: 147-48)and, for simple topographic reasons, flood wa-ter farming can never have been practised bymany Neolithic communities (Wilkie andSavina 1997; Perles 1999). Anyway, if agro-nomically viable, the floodwater farmingmodel is not incompatible with small-scale,intensive husbandry. A more serious and rad-ical challenge to the intensive mixed farmingmodel is posed by an issue neglected in thepreceding discussion, that of the relationshipbetween land use and 'marginal colonization'.

Land Use in the Later Neolithic and InitialBronze Age: Pastoralism, Herding andMarginal Colonization

Whereas early farming villages are concen-trated in relatively fertile and well-watered

parts of lowland Greece, Late-Final Neolithic(late sixth to fourth millennia BC) and EarlyBronze Age (third millennium BC) sites arealso widespread in agriculturally more mar-ginal parts of the Greek landscape, includingthe southern mainland and islands of the cen-tral and southern Aegean. Over much of thisarea, growing-season rainfall is, on average,close to the minimum needed for successfulcereal and pulse crops and, in many years,falls below this level, while relatively thin andinfertile soils often exacerbate the risk ofdrought and pose other limitations on cropgrowth. Other things being equal, therefore,crop husbandry would have been a less reli-able subsistence base than in the core areas ofearly farming settlement, and an expansion ofherding or foraging may have been favoured.

If colonization of areas marginal for cropgrowing provided the motive for subsistencechange, associated changes in the pattern ofsettlement provided the opportunity. Earlysettlement in these marginal areas tended tobe relatively dispersed, often taking the formof 'hamlets', comprising just a handful of fam-ily households, or even perhaps of isolated'farmsteads' (e.g. Blackman and Branigan1977; Cherry 1982; Whitelaw 1983; Branigan1999). Such occupation sites also tend to berelatively short-lived, perhaps lasting just afew generations, making it harder to assessthe density of contemporaneously occupiedsettlements than in the core areas, where tellvillages were often occupied more or less con-tinuously for several millennia. As a result, itcannot be argued for these marginal areas thatcrop-based subsistence was enforced by thelarge size of individual communities or thedensity of regional population. Althoughearly settlers introduced various wild animalsto the Aegean islands (Jarman 1996; Halstead1987b), the sparse bioarchaeological recordfrom LN-EB dispersed sites, such as Phyla-kopi (Gamble 1982), Kalythies cave (Halstead

118 PaulHalstead

and Jones 1987), Skoteini cave (Mangafa 1993;Kotjabopoulou and Trantalidou 1993), andZas cave (Zachos 1999) is dominated byremains of crops and livestock. This suggeststhat, if subsistence change accompanied mar-ginal colonization, it would have taken theform of greater reliance on herding ratherthan foraging. Chronologically, this phase ofmarginal colonization partly overlaps withSherratt's proposed 'secondary products revo-lution' (Sherratt 1981), encouraging somescholars to assume that dairy products,woollen textiles and animal traction came intouse in the Aegean at this time (e.g. van Andeland Runnels 1988). Research elsewhere inEurope, however, particularly into the antiq-uity of dairying (e.g. Legge 1981; Halstead1989; Rowley-Conwy 1997), suggests that thisimportant model is better regarded as a sourceof stimulating questions than of ready-madeanswers. To what extent are motive andopportunity for increased dependence onlivestock during LN-EB marginal colonizationmatched by evidence of a less circumstantialnature?

Widespread human use of caves is firstattested in Greece in the later Neolithic, and isparticularly common in agriculturally mar-ginal areas of the southern mainland andislands. Many of these caves are ill-suited, interms of size, illumination or accessibility forlong-term residence, but some have been usedin the recent past as temporary shelters forherders and/or livestock. On the other hand,caves were obviously used for a variety ofpurposes in later prehistory, including burialand other forms of ritual (Demoule and Perles1993) and also storage (Sampson 1992), andneed not be linked with herding. Moreover,although caves tend to be located, for obviousreasons, in rocky terrain, many have nearbytraces of recent or even current cultivationand so cannot be regarded as herding sitessimply on the basis of location. For example,

the LN-FN cave of Skoteini, in the hills ofEvvia in central Greece, has been interpretedas a seasonal base for mobile herders partlybecause of its location in an area of broken ter-rain, more suited to herding than mechanizedcultivation (Sampson 1992; 1993). Non-mech-anized cultivation is still practised in thevicinity of the cave (Jones et al. 1999), howev-er, and abandoned terraces show that cropgrowing was very widespread a few decadesago. During the twentieth century, the cavehas regularly been used to shelter livestock—but by mixed farming households living year-round in the nearby village of Tharounia.Even if LN-EB cave use was largely related toherding, therefore, this might simply repre-sent the expansion of mixed farming commu-nities into areas well endowed with caves (i.e.,convenient, natural shelters) rather than a sig-nificant shift in the balance between arableand pastoral land use.

Some LN-EB open sites have also been inter-preted as herders' camps. In the hills ofEpirus, north-west Greece, in an area nowlargely given over to grazing rather than cul-tivation, excavation of the tiny FN site atDoliana revealed two successive floors whichcould have belonged to a small hut of the sortoccupied seasonally by recent transhumantpastoralists (Dousougli 1996), but flimsystructures are not a wholly reliable indicatorof seasonal habitation. In the southern main-land, several survey projects have encoun-tered a rash of LN and/or FN open sites, oftensmall in area and sometimes in similar loca-tions to the folds of modern shepherds orgoatherds (Johnson 1996; Cavanagh 1999).Several of these sites have yielded restrictedranges of artefacts which, if not the result ofpartial preservation, might indicate somefunction other than permanent habitation(Johnson 1996: 65-66; Cavanagh 1999: 36).There is no unambiguous link, however,between assemblages lacking ceramics or


quernstones and herding. There are also dan-gers in viewing the arable potential of suchsites from a modern, mechanized perspectiveand some of the steeper and barer slopes ofsouthern Greece may have been blessed witha significant covering of soil in the Neolithic(Wells et al 1990). On the other hand, Cavan-agh notes that, in some parts of southernGreece, LN-EB 'marginal colonization' tookplace against a background of apparentlysparse occupation of areas more favourable tocultivation. Thus the location of sites on bar-ren limestone outcrops may not have been theenforced outcome of population pressure andcould be seen instead in terms of the selectiveoccupation of parts of the landscape favour-able to grazing (Cavanagh 1999: esp. 36-37).Such areas may have been particularly sus-ceptible to destabilization by grazing, whichmight in turn be reflected in the palaeoecolog-ical record.

In a comparative analysis of geoarchaeolog-ical evidence from inland north-east Thessalyand from the Argive plain and adjacent hillsof the southern Argolid in the southern main-land, van Andel et al (1990) argued that thepostglacial period has been characterized forthe most part by landscape stability, periodi-cally interrupted by brief phases of erosionand deposition. The dating of these sequencesis rather coarse, but the phases of instabilityapparently occurred at different dates in eachstudy area, favouring anthropogenic ratherthan climatic causation, and in each case start-ed several centuries after local evidence forthe inception of farming, suggesting anthro-pogenic disturbance related to cultivation orgrazing. Broadly similar sequences have sub-sequently been reported for the coastal area ofsoutheast Thessaly (Zangger 1991) and forcoastal Pieria in central Macedonia (Krahto-poulou this volume). Substantial Neolithic-EB episodes of anthropogenic landscapedestabilization are not easily compatible with

the model of small-scale and stable early hor-ticulture (van Andel et al. 1990), but might beexplicable in terms of an expansion of herdingfrom the later Neolithic onwards. On the otherhand, phases of landscape instability appearto have been short-lived (van Andel et al.1986) and, in at least some cases, may havetaken the form of rare catastrophic events(Zangger 1991; 1994; Moody 1997). In the lat-ter case, human land use may well have madethe landscape more or less susceptible toextreme weather conditions (with extensivegrazing rather than intensive cultivation per-haps the more plausible agent of widespreaddisruption—see above), but alluvial depositsare unlikely to be a reliable guide to the timing,nature and scale of anthropogenic alterationof the landscape.

Palynological evidence is also ambiguous. Awidespread expansion of hornbeams duringthe sixth-third millennia BC has been inter-preted in terms of increases in both precipita-tion (van Zeist and Bottema 1982) and brows-ing or cutting (e.g. Jahns 1993; Halstead 1994;Willis 1994a; but cf. Willis 1994b). In the longQuaternary sequence from L. Giannina, how-ever, these taxa appear to have expanded at asimilar stage during earlier warm periods(Tzedakis 1994), thus casting doubt on ananthropogenic interpretation. Most of thepollen evidence from Greece comes from largelakes or bogs with extensive catchments andso would be insensitive to local anthropogenicinterference with vegetation. Small catchmentcores, however, spanning the mid-Holoceneexpansion of hornbeams, have been taken inEpirus, north-west Greece: from Rezina at1800 m (Willis 1994a) in the Pindos moun-tains; and from Gramousti at 400 m (Willis1994a) and Tseravinas at 450 m (Turner andSanchez-Goni 1997), both in the nearby inlandbasin of Doliana. At Rezina, some time afterthe initial expansion of hornbeams, a cleardecrease in arboreal pollen relative to that of

120 PaulHalstead

herbaceous taxa, possibly accompanied byinflux of sediment from surrounding slopes,occurs in zone RMS from c. 5000 BC (6000 bp).Willis argues for similar changes at a similardate in Gramousti zone GL4 and, on thisbasis, suggests anthropogenic impact in theform of transhumant herding (Willis 1994a;1997), but the claimed relative expansion ofherbaceous pollen is not unambiguous untilthe later zone GL5 and dating is complicatedby inconsistencies in the available C14 dates.Nonetheless, in the nearby Tseravinas core,the expansion of hornbeams after c. 5500 BC(6500 bp) in zone TS5 is accompanied by aclear reduction in arboreal pollen and anincrease in herbaceous pollen, and also byhigh magnetic susceptibility values and anabundance of charcoal fragments. Togetherthese complementary lines of evidence dosuggest anthropogenic disturbance in thevicinity of, and roughly contemporary with,the FN site of Doliana. This unusually earlyevidence for human impact in the palynologi-cal record from Epirus should probably beattributed to the sensitivity of small catch-ment cores, however, rather than to regionaldifferences in land use, because archaeologi-cal evidence for Neolithic settlement is muchmore abundant in most other parts of main-land Greece.

Neither the geoarchaeological nor the paly-nological record is easy to interpret, in termsof the causes of landscape change. In each case,however, there is evidence of mid-Holocenelandscape change, potentially compatible withanthropogenic disruption and perhaps, moreparticularly, with an expansion of herdingactivity. This evidence is sufficient to justify, atleast, a reconsideration of the archaeozoologi-cal evidence for the scale of LN-EB herding.

Heavy reliance on livestock for subsistencein agriculturally marginal areas would argu-ably have required one or more of the follow-ing strategies: the keeping of very large num-

bers of animals; highly productive, intensivedairying; and specialization in animal prod-ucts (e.g. cheese, meat or livestock, wool ortextiles) for exchange (Halstead 1996).Absolute numbers of livestock are rarelyretrievable from faunal evidence, but seasonalmovement and specialization in one specieshelp to relax the constraints on stock numbersposed by pasture and herding labour, respec-tively. As regards seasonality of occupation,the three cave sites of LN-FN Skoteini in cen-tral Greece (Kotjabopoulou and Trantalidou1993), LN-EB Zas on Naxos in the centralAegean (Halstead 1996), and LN-FNKalythies on Rhodes in the south-east Aegean(Halstead and Jones 1987), and the 'marginal'open sites of LN Kastri on Thasos in the north-ern Aegean (Halstead 1987b: 79 fig. 3b) andDoliana in Epirus (Halstead et al. in prep.)have yielded remains of very young livestock.This suggests a human presence at least in latewinter/early spring, the time of year when,according to some scholars, mobile herdersmight have abandoned the higher sites of Zas(630 m), Skoteini (450 m) and Doliana (400 m)for pasture at lower altitude. Thus the avail-able evidence does not support mobility overdistances long enough to entail seasonal occu-pation of sites, although it does not precludethe use of these sites as winter bases for sum-mer grazing at higher altitudes. In terms ofspecies composition, none of these sites ishighly specialized: at Doliana, sheep, cattleand pigs are all well represented; at Kastri andthe three cave sites, and also at the marginalopen site of FN Kefala on the central Aegeanisland of Kea, there are few cattle or pigs butfairly balanced proportions of sheep andgoats (Halstead 1996: 31 fig. 2).

At Zas, Skoteini, Kalythies and perhapsDoliana, the slaughter of sheep and goatscombined was consistent with an unspecial-ized 'meat' strategy and, at Zas, this was sep-arately demonstrable for each of these two


species. Thus at none of these sites does mor-tality evidence suggest intensive dairying andso strengthen the plausibility of direct depen-dence on livestock for subsistence, while nei-ther mortality evidence nor body part repre-sentation indicates specialization in animalsor animal products for exchange (Halstead1996: 32). Faunal evidence implies, therefore,albeit indirectly, that dependence on livestockfor subsistence in areas of LN-EB 'marginalcolonization' is unlikely. No support is offeredto Cavanagh's (1999) suggestion, that margin-al LN-EB sites in southern Greece may havebeen occupied by herders living in a symbiot-ic exchange relationship with other, crop-growing communities. On the contrary, thefaunal record is consistent with the argumentthat specialized pastoralism is dependent on amarket economy and so unlikely to havedeveloped in the Aegean in prehistory (Leesand Bates 1974; Cherry 1988; Halstead 1991;but cf. Chang and Koster 1986).

As Cavanagh (1999) has pointed out, how-ever, it is an unhelpful oversimplification ofthe problem to consider only the polarizedalternatives of small-scale, sedentary mixedfarming and large-scale, mobile pastoralism,ignoring the possibility of intermediate herd-ing strategies characterized by short-distancemobility. It was argued above that, at earlyfarming villages, livestock played an impor-tant role in indirect storage, in negotiatingwealth and status, and in lubricating socialrelationships. An LN-EB expansion in thescale of animal keeping might be expected,therefore, for two reasons. First, as alreadynoted, farmers in agriculturally marginalareas are likely to have been more frequentlyand heavily dependent on livestock as ameans of banking against, and responding to,crop failure. Secondly, even in core areas ofearly farming settlement, the progressiveLN-EB isolation of the family household andreduction in social pressures towards sharing

(Halstead 1999) would have increased theincentives to accumulate wealth in the form oflivestock. The faunal record may offer indirectsupport for such an expansion in numbers oflivestock.

Faunal assemblages from early village sitesare overwhelmingly dominated by sheep, thedomestic animal most suited to the grazingand manuring of cultivated land, as well asconsumption of crop residues and food waste.Early animal husbandry may, therefore, havebeen closely integrated with, and thus limitedby the scale of, arable farming. LN-EB assem-blages in the same areas tend towards a morebalanced mixture of sheep, pigs and cattle,while the marginal assemblages from Skoteini,Kalythies, Zas, Kastri and Kefala (but notDoliana) have low percentages of pigs andcattle and are dominated by sheep and goats.This divergence in the composition of faunalassemblages may be explicable in terms of thecontrasting environments of these two typesof site, with pigs and cattle exploiting wood-land pannage and browse in the fertile low-lands, and goats exploiting the more rugged,scrubby parts of the landscape in 'marginal'areas. This would imply that LN-EB stockrearing may have expanded, in terms both ofnumbers of animals and of the scale of herd-ing movements, beyond the limits set by closeintegration with the cultivated landscape.

The sorts of grazing movements hypothe-sized by Cavanagh and Johnson, therefore,may reflect such a growing scale of animalhusbandry within an essentially mixed farm-ing regime. The steep and barren areasdescribed by these authors would have sup-ported seasonal (early) flushes of grazing.Such patches of nutritious pasture are keenlysought after today, for their ability to fattenanimals, improve milk yields, or enhancereproductive success. For the Neolithic far-mer, the use of surplus grain or flushes of nat-ural pasture would have been alternative

122 PaulHalstead

ways of building up the household herdwhich served as both a subsistence bank anda store of wealth. For example, for a house-hold recovering from a bad harvest, with nograin to spare as fodder and with most of itslivestock either eaten or exchanged for grain,intensive use of good seasonal pasture mighthave been an effective way of fattening up, orimproving the fertility of, remaining animals.Likewise, a household with 'surplus' herdinglabour (e.g. with teenage children) could haveenhanced its wealth in this way. Under suchcircumstances, even in a lightly populatedlandscape, seasonal patches of distant butgood pasture might well have been soughtand efforts might well have been made toextend their availability in time or space bymeasures such as burning. Seasonal herdingat a distance from the 'home' village is alsolikely to have served other cultural roles (cf.Broodbank 1993; Barrett 1994: 132-53;Edmonds 1997: 105): as a rationale for explor-ing the wider landscape and acquiring arcaneknowledge or exotic resources; as an opportu-nity for meeting members of other communi-ties engaged in similar movements; and per-haps as a rite of passage for those guardingwandering livestock against predators.

This 'narrative' moves beyond the polaropposition between large-scale, mobile pas-toralism and the household herd tethered tothe vicinity of the settlement. The suggestionthat livestock, although subsidiary to cropgrowing, was increasingly important throughtime, is consistent with available faunal data,does not entail potentially anachronisticassumptions about pre-market symbiosisbetween herders and crop growers, and canaccommodate the possible hints in the sitesurvey and palaeoecological records that LN-EB livestock may have ranged some distanceaway from cultivated areas. Perhaps moreimportant than the capacity of the model toreconcile apparently conflicting lines of evi-

dence is that it carries implications for futureresearch. For a full understanding of thenature and scale of early animal husbandry,and of its relationship with crop husbandry, itis necessary to track the movement of live-stock through the prehistoric cultural land-scape. To this end, isotopic analysis may shedlight on the broad geographical scale of move-ments (across watersheds or between lowlandand highland—d'Angela 1992; Wiedemann etal. 1999), while dental microwear and copro-lite evidence of diet may provide finer-grained insights into the particular pasture orfodder niches occupied (e.g. Mainland 1998a;1998b; Anderson and Ertug-Yaras 1998;Charles 1998).

Conclusion

Previous debate on the nature of early farm-ing in the Aegean area has largely revolvedaround two related issues: the oppositionbetween extensive ('traditional') and intensivestrategies of crop and stock husbandry; andthe balance between arable farming and herd-ing. The available evidence, though sparse andoften circumstantial, is compatible with thedominance of intensive mixed farming duringthe Neolithic, with the patchy adoption ofextensive farming from the latter part of theBronze Age, and with the absence of special-ized pastoralism before the historical era.

Care must be taken, however, not to conflatepastoralism (a more or less specialized way oflife), herding (the management of livestock byeither mixed farmers or specialized pastoral-ists), and seasonal mobility (practised, onvarying spatial scales, by both mixed farmersand pastoralists). If, as seems almost inevit-able, prehistoric livestock served as a bufferagainst subsistence failure, as a medium ofsocial interaction, and as a form of wealth, it isalso highly likely that attempts were made to


fatten up livestock by taking advantage of sea-sonally available and spatially scattered flush-es of rich grazing. The scale of stock-rearingmay have expanded during and after the laterNeolithic, partly as a consequence of theincreasing isolation of the household as abasic unit of production and consumption andpartly, in 'marginal' areas, as a response toincreased risk of crop failure. These expecta-tions are consistent with the apparent diver-gence in composition of faunal assemblagesbetween 'core' village sites and 'marginal'caves and hamlets. The proliferation of cavesites and small, ephemeral open sites duringthe later Neolithic may also be related to suchherding activity, but may partly reflect theexpansion of settlement at this time into areaswell endowed with caves and widely subjectto intensive survey. In other words, the appar-ent changes in settlement evidence in the laterNeolithic may reflect increasing archaeologi-cal visibility and detection of herding activi-ties rather than an expansion of mobile herd-ing. Independent investigation of the move-ment of early livestock within the landscape,as reflected in bone chemistry and dentalmicrowear, might clarify this issue.

The relative contributions of natural andcultural agents to the postglacial developmentof the landscape of the Aegean will continueto be disputed until changing patterns of bothclimate and land use can be inferred reliablyand independently. The relationships betweenlandscape change and land use, and betweenland use and human settlement, are too com-plex and too variable for patterns of land useto be inferred solely from palaeoecologicalrecords or for the latter to be explained solelyin terms of settlement patterns. Rather thescale and nature of land use must be inferredthrough multi-disciplinary research, combin-ing both on-site and off-site approaches. Forexample, ecological analysis of arable weedsand demographic analysis of domestic ani-

mals may illuminate the intensity of farmingpractice; the study of plant inclusions in ani-mal dung may clarify the degree of integra-tion between arable and pastoral farming; andsurface survey data may serve to map boththe scale of human activity and its spatial dis-tribution in the natural and cultural land-scape.

While palaeoecological evidence should notbe regarded as a short-cut to the reconstruc-tion of land use, viable models of early farm-ing practice must obviously be compatiblewith the palaeoecological record. In the pre-ceding discussion, possible geoarchaeologicaland palynological traces of anthropogenicdestabilization of the landscape have playedan important heuristic role in encouraging are-evaluation of later Neolithic land use. Thisre-evaluation has, in turn, identified new pri-orities for future research into patterns of landuse. Similarly, it is hoped that the broad dis-tinction drawn here between extensive andintensive systems of land use, with contrast-ing implications for the severity and detect-ability of landscape change, may aid attemptsto interpret the postglacial palaeoecologicalrecord of Greece in terms of human impact.

Acknowledgments

I am grateful to Amy Bogaard, Bill Cavanagh,Charles Frederick, Valasia Isaakidou, GlynisJones, Nancy Krahtopoulou and IngridMainland for helpful discussions during thewriting of this paper.

Bibliography

Acheson, P.1997 Does the economic explanation work? Settle-

ment, agriculture and erosion in the territory ofHalieis in the Late Classical-Early Hellenisticperiod. JMA 10: 165-90.

124 PaulHalstead

Andel, T. van, K. Gallis and G. Toufexis1995 Early Neolithic farming in a Thessalian river

landscape, Greece. In J. Lewin, M.G. Macklinand J.C. Woodward (eds.), Mediterranean Quater-nary River Environments: 131-43. Rotterdam:Balkema.

Andel, T. van and C. Runnels1988 An essay on the 'emergence of civilization7 in

the Aegean world. Antiquity 62: 234-47.1995 The earliest farmers in Europe. Antiquity 69:

481-500.Andel, T. van, C. Runnels and K. Pope

1986 Five thousand years of land use and abuse in thesouthern Argolid, Greece. Hesperia 55: 103-28.


torical Greece. Journal of Field Archaeology 17:379-96.

Anderson, S. and F. Ertug-Yaras1998 Fuel fodder and faeces: an ethnographic and

botanical study of dung fuel use in centralAnatolia. Environmental Archaeology 1: 99-109.

Angela, D. d'1992 Palaeoclimatic conditions in the Po plain during

the Neolithic, Chalcolithic and Bronze Ages:first results. Rivista di Studi Liguri 57:119-26.

Barrett, J.C.1994 Fragments from Antiquity. Oxford: Basil Blackwell.

Becker, C.1999 The Middle Neolithic and the Platia Magoula

Zarkou: a review of current archaeozoologicalresearch in Thessaly (Greece). Anthropozoologica30: 3-22.

Blackman, D. and K. Branigan1977 An archaeological survey of the lower catch-

ment of the Ayiofarango valley. BSA 72: 13-84.Blitzer, H.

1990 Pastoral life in the mountains of Crete.Expedition 32: 34-41.

Bottema, S.1982 Palynological investigations in Greece with spe-


1994 The prehistoric environment of Greece: a reviewof the palynological record. In P.N. Kardulias(ed.), Beyond the Site: Regional Studies in theAegean Area: 45-68. Lanham: University Press ofAmerica.

Branigan, K.1999 Late neolithic colonization of the uplands of

eastern Crete. In P. Halstead (ed.), NeolithicSociety in Greece: 57-65. Sheffield: SheffieldAcademic Press.

Braudel, F.1975 The Mediterranean and the Mediterranean World in

the Age of Philip II, Volume 1. London: Fontana.Broodbank, C.

1993 Ulysses without sails: trade, distance, knowl-edge and power in the early Cyclades. WorldArchaeology 24: 315-31.

Buckland, PC., A.J. Dugmore and K.J. Edwards1997 Bronze Age myths? Volcanic activity and

human response in the Mediterranean andNorth Atlantic regions. Antiquity 71: 581-93.

Campbell, J.K.1964 Honour, Family and Patronage. Oxford: Oxford

University Press.Cavanagh, W.

1999 Revenons a nos moutons: surface survey andthe Peloponnese in the Late and Final Neolithic.In J. Renard (ed.), Le Peloponnese: archeologie ethistoire: 31-65. Rennes: Presses UniversitairesRennes.

Chang, C.1992 Archaeological landscapes: the ethnoarchaeolo-

gy of pastoral land use in the Grevena provinceof Greece. In J. Rossignol and L. Wandsnider(eds.), Space, Time, and Archaeological Landscapes:65-89. New York: Plenum.

1994 Sheep for the ancestors: ethnoarchaeology andthe study of ancient pastoralism. In P.N.Kardulias (ed.), Beyond the Site: Regional Studiesin the Aegean: 295-313. Lanham: University Pressof America.

Chang, C. and H. Koster1986 Beyond bones: toward an archaeology of pas-

toralism. In M. Schiffer (ed.), Advances inArchaeological Method and Theory 9: 97-148. NewYork: Academic Press.

Charles, M.1998 Fodder from dung: the recognition and inter-

pretation of dung-derived plant material fromarchaeological sites. Environmental Archaeology1: 111-22.

Cherry, J.F.1982 A preliminary definition of site distribution on

Melos. In C. Renfrew and M. Wagstaff (eds.), AnIsland Polity: the Archaeology of Exploitation inMelos: 11-23. Cambridge: Cambridge UniversityPress.

1988 Pastoralism and the role of animals in the pre-and proto-historic economies of the Aegean. InC.R. Whittaker (ed.), Pastoral Economies inClassical Antiquity. Cambridge PhilologicalSociety Supplementary Volume 14: 6-34.


Cherry, J.F, J.L. Davis and E. Mantzourani1991 Patterns in the landscape of Keos. In J.F. Cherry,

J.L. Davis and E. Mantzourani (eds.), LandscapeArchaeology as Long-Term History: Northern Keosin the Cycladic Islands from Earliest Settlementuntil Modern Times: 457-79. Los Angeles:Institute of Archaeology, UCLA.

Cullen, T.1984 Social implications of ceramic style in the

Neolithic Peloponnese. In W.D. Kingery (ed.),Ancient Technology to Modern Science, 1: 77-100.Columbus: American Ceramic Society.

Davis, J.L.1991 Contributions to a Mediterranean rural archae-

ology: historical case studies from the OttomanCyclades. JMA 4: 131-215.

Demoule, J-P. and C. Perles1993 The Greek Neolithic: a new review. Journal of

World Prehistory 7: 355-416.Dousougli, A.

1996 Epirus: the Ionian Islands. In G.A. Papathan-assopoulos (ed.), Neolithic Culture in Greece: 46-48. Athens: Goulandris Foundation.

Du Boulay, J.1974 Portrait of a Greek Mountain Village. Oxford:

Clarendon.Edmonds, M.

1997 Taskscape, technology and tradition. AnalectaPraehistorica Leidensia 29: 99-110.

Endfield, G.H.1997 Myth, manipulation and myopia in the study of

Mediterranean soil erosion. In A. Sinclair, E.Slater and J. Gowlett (eds.), ArchaeologicalSciences 1995. Oxbow Monographs 64: 241-48.Oxford: Oxbow.

Forbes, H.1976 'We have a little of everything7: the ecological

basis of some agricultural practices in Methana,Trizinia. Annals of the New York Academy ofSciences 268: 236-50.

1982 Strategies and Soils: Technology, Production andEnvironment in the Peninsula of Methana, Greece.PhD dissertation, University of Pennsylvania.

1995 The identification of pastoralist sites within thecontext of estate-based agriculture in ancientGreece: beyond the 'transhumance versus agro-pastoralism' debate. BSA 90: 325-38.

1997 A 'waste' of resources: aspects of landscapeexploitation in lowland Greek agriculture. InP.N. Kardulias and M.T. Shutes (eds.), AegeanStrategies: Studies of Culture and Environment onthe European Fringe: 187-213. Lanham: Rowmanand Littlefield.

1998 European agriculture viewed bottom-sideupwards: fodder- and forage-provision in a tra-ditional Greek community. EnvironmentalArchaeology 1: 19-34.

Foxhall, L. and H.A. Forbes1982 Sitometria: the role of grain as a staple food in

classical antiquity. Chiron 12: 41-90.Frederick, C.

2000 Evaluating causality of landscape change:examples from alluviation. In P. Goldberg, V.Holliday and R. Ferring (eds.), Earth Sciences andArchaeology. New York: Plenum.

Gamble, C.1982 Animal husbandry, population and urbanisation.

In C. Renfrew and M. Wagstaff (eds.), An IslandPolity: The Archaeology of Exploitation in Melos:161-71. Cambridge: Cambridge University Press.

Grigg, D.B.1974 The Agricultural Systems of the World: an

Evolutionary Approach. Cambridge: CambridgeUniversity Press.

Halstead, P.1987a Traditional and ancient rural economy in

Mediterranean Europe: plus c,a change? JHS107: 77-87.

1987b Man and other animals in later Greek prehisto-ry. BSA 82: 71-83.

1989 Like rising damp? An ecological approach to thespread of farming in southeast and centralEurope. In A. Milles, D. Williams and N.Gardner (eds.), The Beginnings of Agriculture.BAR International Series 496: 23-53. Oxford:British Archaeological Reports.

1990 Waste not, want not: traditional responses tocrop failure in Greece. Rural History 1: 147-64.

1991 Present to past in the Pindhos: specialisationand diversification in mountain economies.Rivista di Studi Liguri 56: 61-80.

1992a Agriculture in the bronze age Aegean: towardsa model of palatial economy. In B. Wells (ed.),Agriculture in Ancient Greece: 105-16. Stockholm:Swedish Institute at Athens.

1992b From reciprocity to redistribution: modellingthe exchange of livestock in Neolithic Greece.Anthropozoologica 16: 19-30.

1993 Banking on livestock: indirect storage in Greekagriculture. Bulletin on Sumerian Agriculture 7:63-75.

1994 The North-South divide: regional paths to com-plexity in prehistoric Greece. In C. Mathers andS. Stoddart (eds.), Development and Decline in theMediterranean Bronze Age: 195-219. Sheffield: J.R.Collis.

126 PaulHalstead

1995 Plough and power: the economic and social sig-nificance of cultivation with the ox-drawn ardin the Mediterranean. Bulletin on SumerianAgriculture 8:11-22.

1996 Pastoralism or household herding? Problems ofscale and specialization in early Greek animalhusbandry. World Archaeology 28: 20-42.

1998a Ask the fellows who lop the hay: leaf-fodder inthe mountains of northwest Greece. RuralHistory 9: 211-34.

1998b Mortality models and milking: problems of uni-formitarianism, optimality and equifinalityreconsidered. Anthropozoologica 27: 3-20.

1999 Neighbours from hell: the household inNeolithic Greece. In P. Halstead (ed.), NeolithicSociety in Greece: 77-95. Sheffield: SheffieldAcademic Press.

Halstead, P. and Jones, G.1987 Bioarchaeological remains from Kalythies cave,

Rhodes. In A. Sampson, I Neolithiki Periodos staDodekanisa: 135-52. Athens: Ministry of Culture.

1989 Agrarian ecology in the Greek islands: timestress, scale and risk. JHS 109: 41-55.

Halstead, P. and Tierney, J.1998 Leafy hay: an ethnoarchaeological study in NW

Greece. Environmental Archaeology 1: 71-80.Hansen, J.M.

1988 Agriculture in the prehistoric Aegean: data ver-sus speculation. AJA 92: 39-52.

Hillman, G.1981 Reconstructing crop husbandry practices from

charred remains of crops. In R. Mercer (ed.),Farming Practice in British Prehistory: 123-62.Edinburgh: Edinburgh University Press.


Whittaker (ed.), Pastoral Economies in ClassicalAntiquity. Cambridge Philological SocietySupplementary Volume 14: 35-74.

1991 Politics as a determinant of pastoralism: the caseof southern Greece, ca 800-300 BC. Rivista diStudi Liguri 56:139-63.

Horden, P. and N. Purcell2000 The Corrupting Sea: A Study of Mediterranean

History. Oxford: Basil Blackwell.Isager, S. and J.E. Skydsgaard

1992 Ancient Greek Agriculture: An Introduction.London: Routledge.

Jahns, S.1993 On the Holocene vegetation history of the

Argive plain (Peloponnese, southern Greece).Vegetation History and Archaeobotany 2:187-203.

Jameson, M.H.1988 Sacrifice and animal husbandry in classical

Greece. In C.R. Whittaker (ed.), PastoralEconomies in Classical Antiquity. CambridgePhilological Society Supplementary Volume 14:87-119.

Jarman, M.R.1996 Human influence in the development of the

Cretan fauna. In D. Reese (ed.), The Pleistoceneand Holocene Fauna of Crete and its First Settlers:211-29. Madison: Prehistory Press.

Jarman, M.R., G.N. Bailey and H.N. Jarman (eds.)1982 Early European Agriculture. Cambridge:

Cambridge University Press.Johnson, M.

1996 The Berbati-Limnes archaeological survey: theNeolithic period. In B. Wells (ed.), The Berbati-Limnes Archaeological Survey 1988-1990: 37-73.Stockholm: Swedish Institute at Athens.

Jones, G.1987 Agricultural practice in Greek prehistory. BSA

82: 115-23.1992 Weed phytosociology and crop husbandry:

identifying a contrast between ancient andmodern practice. Review of Palaeobotany andPalynology 73: 133-43.

Jones, G., A. Bogaard, P. Halstead, M. Charles and H.Smith

1999 Identifying the intensity of crop husbandrypractices on the basis of weed floras. BSA 94:167-89.

Karakasidou, A.N.1997 Fields of Wheat, Hills of Blood. Chicago. Chicago

University Press.Karavidas, K.D.

1931 Agrotika: Meleti Sigkritiki. Athens.Kilian, K.

1973 Zur eisenzeitlichen Transhumanz in Nord-griechenland. Archaologisches Korrespondenzblatt3: 431-5.

Killen, J.T.1985 The Linear B tablets and the Mycenaean econo-

my. In A. Morpurgo Davies and Y. Duhoux(eds.), Linear B: A1984 Survey: 241-305. Louvain:Lou vain University Press.

1994 Thebes sealings, Knossos tablets and Mycen-aean state banquets. BICS 1994: 67-84.

Koster, H.A.1977 The Ecology of Pastoralism in Relation to Changing

Patterns of Land Use in the Northeast Peloponnese.PhD dissertation, University of Pennsylvania.

1997 Yours, mine, and ours: private and public pas-ture in Greece. In PN. Kardulias and M.T.


Shutes (eds.), Aegean Strategies: Studies of Cultureand Environment on the European Fringe: 141-85.Lanham: Rowman and Littlefield.

Kotjabopoulou, E. and K. Trantalidou1993 Faunal analysis of the Skoteini cave. In A.

Sampson, Skotini Tharrounion: 392-434. Athens:A. Sampson.

Kotsakis, K.1999 What tells can tell: social space and settlement

in the Greek Neolithic. In P. Halstead (edL),Neolithic Society in Greece: 66-76. Sheffield:Sheffield Academic Press.

Lees, S. and D. Bates1974 The origins of specialised nomadic pastoralism.

American Antiquity 39:187-93.Legge, A.J.

1981 The agricultural economy. In RJ. Mercer (ed.),Grimes Graves Excavations 1971-72: 79-103.London: Her Majesty's Stationery Office.

McGrew, W.W.1985 Land and Revolution in Modern Greece, 1800-1881:

The Transition in the Tenure and Exploitation ofLand from Ottoman Rule to Independence. Kent:Kent State University Press.

Mainland, I.L.1998a Dental micro wear and diet in domestic sheep

(Ovis aries) and goats (Capra hircus): distin-guishing grazing and fodder-fed ovicapridsusing a quantitative analytical approach. JAS 25:1259-71.

1998b The lamb's last supper: the role of dental micro-wear analysis in reconstructing livestock diet inthe past. Environmental Archaeology 1: 55-62.

Mangafa, M.1993 Arkhaiovotaniki meleti tou spilaiou Skotinis sta

Tharrounia Evvoias. In A. Sampson, SkotiniTharrounion: 360-69. Athens: A. Sampson.


derstood? In P.N. Kardulias and M.T. Shutes(eds.), Aegean Strategies: Studies of Culture andEnvironment on the European Fringe: 61-77.Lanham: Rowman and Littlefield.

Nitsiakos, V.1985 A Vlach Pastoral Community in Greece: The Effects

of its Incorporation into the National Economy andSociety. PhD dissertation, Cambridge University.

Payne, S.1973 Kill-off patterns in sheep and goats: the mandi-

bles from Asvan Kale. Anatolian Studies 23: 281-303.

1985 Zoo-archaeology in Greece: a reader's guide. InN.C. Wilkie and W.D.E. Coulson (eds.), Contrib-

utions to Aegean Archaeology: Studies in Honor ofWilliam A. McDonald: 211-44. Minneapolis:University of Minnesota Press.

Perles, C.1999 The distribution of magoules in eastern Thes-

saly. In P. Halstead (ed.), Neolithic Society inGreece: 42-56. Sheffield: Sheffield AcademicPress.

Psikhogios, D.K.1987 Proikes, Foroi, Stafida kai Psomi: Oikonomia kai

Oikogenia stin Agrotiki Ellada tou 19 Aiona.Athens: Ethniko Kentro Koinonikon Erevnon.

Psikhogios, D. and G. Papapetrou1984 Oi metakinisis ton nomadon ktinotrofon.

Epitheorisi Koinonikon Erevnon 53: 3-23.Rackham, O.


Rasmussen, P.1990 Pollarding of trees in the Neolithic: often pre-

sumed—difficult to prove. In D.E. Robinson(ed.), Experimentation and Reconstruction inEnvironmental Archaeology: 77-99. Oxford:Oxbow Books.

Renfrew, C.1972 The Emergence of Civilisation: the Cyclades and the

Aegean in the Third Millennium BC. London:Methuen.

1982 Polity and power. In C. Renfrew and M.Wagstaff (eds.), An Island Polity. The Archaeologyof Exploitation in Melos: 264-90. Cambridge:Cambridge University Press.

Rowley-Conwy, P.1997 The animal bones from Arene Candide: final

report. In R. Maggi (ed.), Arene Candide: a Funct-ional and Environmental Assessment of the HoloceneSequence. Memorie dell' Istituto Italiano di Paleo-ntologia Umana 5: 153-277. Rome: II Calamo.

Sampson, A.1992 Late Neolithic remains at Tharounia, Euboea: a

model for the seasonal use of settlements andcaves. BSA 87: 61-101.

1993 Skotini Tharrounion. Athens: A. Sampson.Semple, B.C.

1922 The influence of geographic conditions uponancient Mediterranean stock-raising. Annals ofthe Association of American Geographers 12: 3-38.

Sherratt, A.1981 Plough and pastoralism: aspects of the sec-

ondary products revolution. In I. Hodder, G.Isaac and N. Hammond (eds.), Pattern of thePast: Studies in Honour of David Clarke: 261-305.Cambridge: Cambridge University Press.

128 PaulHalstead

Silverman, S.F.1968 Agricultural organization, social structure, and

values in Italy: amoral familism reconsidered.American Anthropologist 70: 1-20.

Toufexis, G.1994 Neolithic animal figurines from Thessaly. In La

Thessalie: Quinze Annees de Recherches Archeol-ogiques, 1975-1990: 163-68. Athens: Ministry ofCulture.

Turner, C. and M.-F. Sanchez-Goni1997 Late glacial landscape and vegetation in Epirus.

In G. Bailey (ed.), Klithi: Palaeolithic Settlementand Quaternary Landscapes in Northwest Greece,Volume 2: Klithi in its Local and Regional Setting:559-85. Cambridge: McDonald Institute forArchaeological Research.

Turrill, W.B.1929 The Plant-life of the Balkan Peninsula. Oxford:

Oxford University Press.Tzedakis, P.C.

1994 Vegetation change through glacial-interglacialcycles: a long pollen sequence perspective.Philosophical Transactions of the Royal Society ofLondon B 345: 403-32.

Vanschoonwinkel, J.1996 Les animaux dans Fart minoen. In D.S. Reese

(ed.), Pleistocene and Holocene Fauna of Crete andits First Settlers: 351-412. Madison: Prehistory.

Vergopoulos, K.1975 To Agrotiko Zitima stin Ellada, i Koinoniki

Ensomatosi tis Georgias. Athens: Exandas.Wells, B., C. Runnels and E. Zangger

1990 The Berbati-Limnes archaeological survey: the1988 season. Opuscula Atheniensia 18(15): 207-38.

Whitelaw, T.M.1983 The settlement at Fournou Korifi, Myrtos and

aspects of Early Minoan social organization. InO. Krzyszkowska and L. Nixon (eds.), MinoanSociety: 323-45. Bristol: Classical Press.

1991 The ethnoarchaeology of recent rural settlementand land use in northwest Keos. In J.F Cherry,J.L. Davis and E. Mantzourani (eds.), LandscapeArchaeology as Long-Term History: Northern Keosin the Cycladic Islands from Earliest Settlementuntil Modern Times: 403-54. Los Angeles:Institute of Archaeology, UCLA.

Wiedemann, F.B, H. Bocherens, A. Mariotti, A. von denDriesch and G. Grupe

1999 Methodological and archaeological implications

of intra-tooth isotopic variations (d!3C, d!8O)in herbivores from Ain Ghazal (Jordan,Neolithic). JAS 26: 697-704.

Wilkie, N.C. and M.E. Savina1997 The earliest farmers in Macedonia. Antiquity 71:

201-207.Willis, K.J.

1994a Altitudinal variation in the late Quaternary veg-etational history of northwest Greece. HistoricalBiology 9: 103-16.

1994b The vegetational history of the Balkans.Quaternary Science Reviews 13: 769-88.

1997 Vegetational history of the Klithi environment: apalaeoecological viewpoint. In G. Bailey (ed.),Klithi: Palaeolithic Settlement and QuaternaryLandscapes in Northwest Greece, Volume 2: Klithiin its Local and Regional Setting: 395-413.Cambridge: McDonald Institute for Archaeo-logical Research.

Willis, K.J. and K.D. Bennett1994 The Neolithic transition fact or fiction?

Palaeoecological evidence from the Balkans. TheHolocene 4: 326-30.

Zachos, K.L.1999 Zas Cave on Naxos and the role of caves in the

Aegean Late Neolithic. In P. Halstead (ed.),Neolithic Society in Greece: 153-63. Sheffield:Sheffield Academic Press.

Zangger, E.1991 Prehistoric coastal environments in Greece: the

vanished landscapes of Dimini bay and LakeLerna. Journal of Field Archaeology 18:1-16.

1994 Landscape changes around Tiryns during theBronze Age. AJA 98: 189-212.

Zangger, E., M.E. Timpson, S.B. Yazvenko, F. Kuhnke andJ. Knauss

1997 The Pylos Regional Archaeological Project, 2:landscape evolution and site preservation.Hesperia 66: 549-641.

Zeist, W. van and S. Bottema1982 Vegetational history of the eastern Mediter-

ranean and the Near East during the last 20,000years. In J.L. Bintliff and W. van Zeist (eds.),Palaeoclimates, Palaeoenvironments and HumanCommunities in the Eastern Mediterranean Regionin Later Prehistory. BAR International Series 133:277-321. Oxford: British Archaeological Reports.

The Scale and Intensity of Cultivation: Evidencefrom Weed Ecology

Amy Bogaard, Michael Charles, Paul Halstead and Glynis Jones

Fascination with the issue of agricultural ori-gins has perhaps diverted the attention ofarchaeologists from investigation of the veryvariable ways in which crops have been man-aged during the postglacial period. Methodsof tillage or sowing, practices of rotation orfallowing, the weeding or watering of grow-ing crops, and a range of other husbandrypractices are of interest for a host of reasons.Such practices play a major part in determin-ing the productivity, labour costs, reliabilityand long-term sustainability of crop growingand help to shape the potential for integrationwith animal husbandry (e.g. Fleming 1985;Halstead 1987; Hodkinson 1988; Palmer1998a). Different crop husbandry regimes mayalso entail very different degrees of impact onthe landscape (e.g. Acheson 1997) and,through reliance on scarce resources (such asland, work-animals or human labour), may bea trigger or symptom of social inequality (e.g.Goody 1976; Sherratt 1981; Halstead 1995).

Although details of agricultural technologyor agronomic know-how may be gleanedfrom artefactual or artistic evidence and, inhistoric periods, from written sources, theserarely provide a representative picture of crophusbandry practices. For Neolithic andBronze Age Greece, the range of staple graincrops commonly grown is now reasonablywell documented by archaeobotanical evi-

dence (Hansen 1988), but insights into likelyhusbandry practices from the ecological char-acteristics of crop species are rare. Most cropstolerate a wide range of growing conditionsand have also been subject to centuries or mil-lennia of selective breeding (e.g. Davies andHillman 1988). For this reason, the richestsource of archaeobotanical evidence for crophusbandry practices is the ecological charac-teristics of the weed species which onceaccompanied growing crops and whose seedscan be found with samples of ancient grain(Hillman 1981).

The use of archaeological weed assemblagesas evidence for past crop husbandry practicesmust be based on modern ecological studiesof relevant agricultural regimes. One of thefirst difficulties encountered in such studies,given the remarkable uniformity of modernagriculture, is that of locating present-dayexamples of the range of husbandry practiceslikely to have existed in the distant past. Forexample, whereas non-mechanized farmers insome parts of the Mediterranean basin stillpractise extensive cultivation of cereals andpulses (characterized by modest inputs ofhuman labour and organic fertilizers), earlycultivation may have been on a small scaleand characterized by intensive inputs ofhuman labour and/or organic fertilizers (e.g.Jones 1992). Today such a radically different

9

130 Amy Bogaard, Michael Charles, Paul Halstead and Glynis Jones

arable environment is normally associatedwith vegetable gardens rather than with sta-ple grain crops.

One area which satisfies at least some of therequirements for a comparative study ofintensive and extensive methods of cultiva-tion is found on the island of Evvia, centralGreece, where pulse crops, especially broadbeans, are grown on a range of scales and atvarying levels of intensity (Jones et al. 1999).Here plots of cultivated pulses range fromtiny gardens within the village to larger fieldsat some distance from habitation. Gardens aremore intensively tilled and more heavilymanured than fields and are more likely to beweeded and watered.

The weed species growing in these plotswere systematically recorded during a fieldstudy in April-May 1988. Analysis of these

data revealed that, in terms of species compo-sition, the weed floras of these plots wereclosely related to three proxy measures of cul-tivation intensity, that is plot size, distancefrom village and plot type (garden, field, andso on). For example, in Figure 9.1, correspon-dence analysis arranges garden and field plotson the basis of weed species compositionalone and reveals an ecological trend, from thesmallest and most intensively managed plotsin the top left of the diagram, through inter-mediate cases at bottom left, to the largest andleast intensively managed plots on the right.

The relationship here between weed floraand husbandry regime is sufficiently strongthat, if samples of harvested grain were col-lected from modern pulse farmers in thestudy area, it should be possible to gauge theintensity of cultivation on the basis of the

Figure 9.1 Correspondence analysis diagram showing Evvia pulse plots arranged along the first two correspon-dence axes according to their weed species composition (number of 1 m2 quadrats out of ten per plot inwhich each species occurred). Only species present in six or more plots (c. 10%) were included. Datapoints represent individual plots, with symbols indicating plot size. Axis 1 is plotted horizontally andaxis 2 vertically.

<20m2

20-50 m2

50-100m2

100-500 m2

>500 m2

The Scale and Intensity of Cultivation 131

associated weed seeds. Two important limita-tions should be noted, however, on the practi-cal application of this modern analogue toarchaeobotanical weed data. First, because levelof fertility and intensity of tillage are highlycorrelated in the Evvia pulse plots, it would bedifficult to determine which of these two fac-tors was responsible for the observed differ-ences in weed floras. Secondly, and even morecritically, it would be impossible to apply theresults of this study to other times and placeswith a different range of weed species.

To solve both of these problems, a furtherfield study was conducted in May-June 1997to explore the ecology of the species encoun-tered (Jones et al 2000). Plant attributes stud-ied included ones which measure species'ability to respond to a fertile environment (e.g.leaf area) and ones which indicates species'ability to recover from disturbance (e.g. lengthof flowering period). Both categories ofattribute are related to plot type in a mannerindicating that fertility and disturbance eachplayed a part in determining weed speciescomposition (Fig. 9.2). Because the sameattributes can be measured on any weedspecies, this provides a modern analoguewhich can be applied to archaeological weedfloras composed of species not encountered inthe Evvia plots.

Although the Evvia study has not yet beenapplied in an archaeological context, it isworth noting that Evvia garden plots tendedto be rich in weeds of the group Chenopo-dietea, while the field plots were rich in weedsof the Secalinetea. The Secalinetea are weedstypical of winter cereal fields while theChenopodietea are characteristic of more fer-tile and more disturbed habitats (e.g. summer'row-crops7 such as potatoes). Archaeologicalweed assemblages containing a mixture ofChenopodietea and Secalinetea species havebeen recovered from a range of sites in Europeof varying date and, in a Greek context, from

Late Bronze Age Assiros Toumba (Jones 1992).This encourages the view that the Evvia pulsegardens are a relevant and fruitful analogy forearly crop husbandry regimes.

Prospects for archaeobotanical applicationwould be greatly enhanced if the Evvia studywere extended to embrace intensive cultiva-tion of winter cereals. To this end, a fieldstudy is currently under way in Asturias,north-west Spain, where winter-sown speltand emmer wheat are grown in small andintensively farmed plots (see also Pena-Chocarro 1996), in rotation with summer row-crops (potatoes and maize). Here, garden-scale cultivation of winter cereals is character-ized by a mixture of the Chenopodietea andSecalinetea groups of weeds.

It should also be noted that neither theAsturias nor the Evvia study encompasses thefull range of variation in cultivation intensity.In both areas, cereal crops were grown untilrecently in plots larger, more distant and moreextensively farmed than those studied. A par-allel field study in Jordan has focused on theextensive end of the spectrum of cultivationregimes, investigating alternative rotation andfallowing practices (Palmer 1998b; Bogaard etal. 1999). The ability to distinguish betweencontinuous cropping and regular fallowingadds a useful additional dimension to theinvestigation of intensive versus extensivecrop husbandry.

Conclusion

As has been emphasized above, the potentialof weed ecology to unravel past husbandryregimes is dependent on the study of present-day weed floras and so is constrained by theavailability of relevant modern analogues.Weed floras growing under husbandry condi-tions of potential relevance to the distant pastare fast disappearing and so opportunities for


(a)

(b)

Figure 9.2. Correspondence analysis of the Evvia pulse plots—diagrams showing weed species arranged along thefirst two correspondence axes (extracted from the same analysis as shown in Fig. 9.1). Data points repre-sent species coded to show (a) ratio of leaf area per node to leaf thickness, an attribute measuring species'ability to respond to a fertile environment, and (b) length of the flowering period, an attribute measur-ing species' ability to recover from disturbance. Axis 1 is plotted horizontally and axis 2 vertically.

< 2 months

3 months

> 4 months

< 1000 mm

1000-20000 mm

> 20000 mm

The Scale and Intensity of Cultivation 133

actualistic research should be grasped withurgency.

Nonetheless, the preceding discussion hasdemonstrated the ability of weed palaeoecolo-gy to clarify the scale and intensity of ancientcultivation regimes and also to identify someof the husbandry practices which constitutedthese regimes. In addition to its intrinsicimportance for an understanding of past eco-nomies, the intensity of cultivation regimeshas great relevance to the degree to whichhuman land use destabilizes the landscape. Ingeneral, extensive regimes seem more likelyto be registered in the palynological andgeoarchaeological records (see Halstead, thisvolume). If the nature of husbandry regimescan be independently established from on-sitearchaeobotanical data, therefore, interpreta-tion of the off-site palaeoecological recordshould be less open-ended. Conversely, arch-aeobotanical evidence for the intensity of cul-tivation cannot reveal the aggregate scale ofagricultural land use at a local or regionallevel, which is more appropriately exploredthrough off-site palaeoecological studies.Closer integration of such archaeobotanicaland palaeoecological studies, coupled withoff-site evidence from intensive surface sur-veys, may in the future enable a serious inves-tigation of both land use and its impact on thelandscape.

Bibliography

Acheson, P.1997 Does the economic explanation work?

Settlement, agriculture and erosion in the terri-tory of Halieis in the Late Classical-EarlyHellenistic period. JMA 10: 165-90.

Bogaard, A., C. Palmer, G. Jones, M. Charles and J.G.Hodgson

1999 A FIBS approach to the use of weed ecology forthe archaeobotanical recognition of crop rota-tion regimes. JAS 26: 1211-24.

Davies, M.S. and G.C. Hillman1988 Effects of soil flooding on growth and grain

yield of populations of tetraploid and hexaploidspecies of wheat. Annals of Botany 62: 597-604.

Fleming, A.1985 Land tenure, productivity, and field systems. In

G. Barker and C. Gamble (eds.), BeyondDomestication in Prehistoric Europe: 129-46.London: Academic Press.

Goody, J.1976 Production and Reproduction. Cambridge:

Cambridge University Press.Halstead, P.

1987 Traditional and ancient rural economy inMediterranean Europe: plus c,a change? JHS107: 77-87.

1995 Plough and power: the economic and social sig-nificance of cultivation with the ox-drawn ardin the Mediterranean. Bulletin on SumerianAgriculture 8: 11-22.

Hansen, J.M.1988 Agriculture in the prehistoric Aegean: data ver-

sus speculation. AJA 92: 39-52.Hillman, G.

1981 Reconstructing crop husbandry practices fromcharred remains of crops. In R. Mercer (ed.),Farming Practice in British Prehistory: 123-62.Edinburgh: Edinburgh University Press.


Whittaker (ed.), Pastoral Economies in ClassicalAntiquity. Cambridge Philological SocietySupplementary Volume 14: 35-74.

Jones, G.1992 Weed phytosociology and crop husbandry:

identifying a contrast between ancient andmodern practice. Review of Palaeobotany andPalynology 73: 133-43.

Jones, G., A. Bogaard, P. Halstead, M. Charles and H.Smith

1999 Identifying the intensity of crop husbandrypractices on the basis of weed floras. BSA 94:167-89.

Jones, G., A. Bogaard, M. Charles and J.G. Hodgson2000 Distinguishing the effects of agricultural prac-

tices relating to fertility and disturbance: a func-tional ecological approach in archaeobotany.JAS 27: 1073-84.

Palmer, C.1998a The role of fodder in the farming system: a case

study from northern Jordan. EnvironmentalArchaeology 1: 1-10.

1998b An exploration of the effects of crop rotation


regime on modern weed floras. EnvironmentalArchaeology 2: 35-48.

Pena-Chocarro, L.1996 In situ conservation of hulled wheat species: the

case of Spain. In S. Padulosi, K. Hammer and J.Heller (eds.), Hulled Wheats: Promoting theConservation and Use of Underutilized andNeglected Crops 4: 129-46. Rome: InternationalPlant Genetic Resources Institute.

Sherratt, A.1981 Plough and pastoralism: aspects of the sec-

ondary products revolution. In I. Hodder, G.Isaac and N. Hammond (eds.), Pattern of thePast: Studies in Honour of David Clarke: 261-305.Cambridge: Cambridge University Press.

10

Settlement Instability and LandscapeDegradation in the Southern Aegean in theThird Millennium BC

Todd Whitelaw

Introduction

Intensive surface surveys are dramaticallytransforming our understanding of the prehis-toric Aegean (Cherry 1983; 1994; Rutter 1993;Shelmerdine 1997). To date, however, much ofthe discussion in the literature has been on themethodologies of field survey and site recov-ery, and rather less on methodologies of inter-pretation (though see Cherry et al. 1988;Alcock et al. 1994; James et al. 1994; Laxton andCavanagh 1995; Whitelaw 1998; French andWhitelaw 1999). Similarly, while there is anincreasing interest in palaeoenvironmentalinvestigation (van Andel et al. 1990; Bintliff1992; Bottema 1994), there are, as yet, relative-ly few studies which attempt to integrate fullythese two sets of data (e.g. Halstead 1984;1994; Jameson et al. 1994; van Andel andRunnels 1987; 1995). To a fair degree, this is aconsequence of the relatively coarse chrono-logical resolution of both datasets, such thatattempts at causal linkage generally rest oncircumstantial assumptions, or only broadsynchronizations of settlement and environ-mental sequences within a region. At the sametime, the models considered for human-envi-ronment interactions remain at a fairly simpleand generally deterministic level (e.g. vanAndel et al. 1986; Runnels and van Andel1987).

To move beyond current interpretativeassumptions will require the investment ofboth time and resources to establish greaterchronological resolution in both settlementand environmental sequences, and moredirect investigation of contexts where culturaland environmental sequences can be directlylinked.

Intensive Surface Survey and SettlementPattern Changes in the Early Bronze AgeSouthern Aegean

Nearly three decades ago, Colin Renfrew(1972a; 1972b) drew the attention of Aegeanprehistorians to issues of regional demogra-phy, and the importance of understanding thenumber, relative distribution, and patterns ofchange in population, to the construction ofAegean social histories. Subsequent research,primarily based on the results of intensivesurface survey, has demonstrated that thebasic patterns he identified owe more to bias-es in data collection than to patterns in pastbehaviour. In addition, however, his basicinterpretative assumptions about the size oftypical sites and the proportion of sitesremaining undiscovered can now be seen tohave been unrealistic. Equally, recognition ofthe significance of changing patterns of settle-

136 Todd Whitelaw

ment nucleation and dispersion (Bintliff 1977),calls into question all of Renfrew's originalinferences derived from comparing popula-tion trajectories based on site numbers fromdifferent areas of the southern Aegean. This ishardly surprising, since no surveys compara-ble to what would today be recognized asintensive had been conducted in the Aegeanprior to the publication of Renfrew's Emer-gence of Civilisation (Cherry 1983).

More substantively, Aegean prehistorianscontinue to be ambiguous about the signifi-cance of demography in the development ofpast societies. In Renfrew's work, this wasrepresented by placing 'population' outside ofthe core interacting elements of 'the system'(1972a: 225), in an intellectual climate inwhich population growth was often viewedas an 'exogenous' factor in cultural develop-ment (e.g. Hill 1977).

A more effective social and dynamic per-spective on regional settlement patterns andpopulation change was initiated in the prehis-toric Aegean context by John Cherry, in hisinterpretation of intensive survey data fromMelos (Cherry 1979; Wagstaff and Cherry1982a; 1982b). In particular, he considered thechanging character of sites, not simply changesin site numbers, in a reconstruction of chang-ing settlement systems, within a broaderinter-regional context. The shift from a dis-persed pattern of small sites in the EarlyBronze Age, to nucleation at a single majorsite—Phylakopi—in the Middle Bronze Age,has become widely extrapolated as a generalmodel for settlement pattern change in differ-ent regions throughout the southern Aegean(e.g. Dickinson 1994: 50-60).

While variants on this pattern are increas-ingly being documented, as other intensivesurveys reach final publication (e.g. Cherry etal 1991; Jameson et al 1994; Cavanagh et al1996; Wells and Runnels 1996; Mee and Forbes1997), the Cycladic data still carry consider-

able weight, since it is the only region of thesouthern Aegean where both settlement andcemetery sites of the Early and Middle BronzeAge are reasonably well known through exca-vation. For this reason, inferences about thecharacter of settlement sites of different sizesare somewhat better founded than in otherareas (Fig. 10.1).

The investigation of numerous Cycladiccemeteries of the Early Bronze Age indicatesthat the communities burying their dead inthem were small, in agreement with the settle-ment evidence. However, in addition, thesmall number of individuals buried in mostcemeteries demonstrates that the sites werealso relatively short-lived, few representingmore than the dead of a single family for morethan a couple of generations (Fig. 10.2). Suchtemporal definition is still not possible withthe coarse-grained chronologies monitoredthrough ceramic stylistic change, so that inother areas of the Aegean, where fewer ceme-teries have been investigated, documentedsettlement patterns are likely to be palimpsests.

General standardization of intensive surveyfield-walking methods, since the early 1980s,has documented the small size of most EarlyBronze Age sites in the southern Aegean withgreater precision (Fig. 10.3). A small numberof somewhat larger sites, usually situated inlow-lying valleys or plains, has generally beeninterpreted as the focal points of small-scalesettlement systems. However, if one considersnot just the distribution of sites of differentsizes, but also the boundaries of the areas sur-veyed intensively, any tendency toward theclustering of smaller sites around the largerexamples, may be more a function of the sizeof the survey 'windows', than a real behav-ioural pattern (Fig. 10.4). Similarly, the largegaps between areas intensively surveyed, ifnot dictated principally by topography, maygive an unrealistic idea of the spatial extent ofsuch small-scale settlement systems.

Settlement Instability and Landscape Degradation 137

Figure 10.1 Excavated Early Cycladic settlement size comparison: (a) Dhaskaleio, Keros; (b) Phylakopi I, Melos;(c) Markiani, Amorgos; (d) Ayia Irini, Keos; (e) Kastri, Syros; (f) Mount Kynthos, Delos; (g) Panormos,Naxos. Estimated occupation areas shaded; 5 m contours.

138 Todd Whitelaw

Figure 10.2 Excavated Early Cycladic cemetery sizes, byphase: Grotta-Pelos; Grotta-Pelos andKeros-Syros; Keros-Syros; unspecifiedphase; total.

Unfortunately, our understanding of thelarger sites is less clear than we might hope,since many of them have later occupationsuperimposed, making the definition of sitesize specifically during the Early Bronze Agedifficult. Excavation has preferentially focusedon such sites, giving some idea of their com-plexity (Konsola 1984; Hagg and Konsola1986). In several Peloponnesian surveys, thecharacter of the finds recovered from individ-ual sites, such as obsidian working debris,roof tiles and impressed hearth rims (Kardulias1992; Jameson et al 1994: 353-54, 356-66;Forsen 1996), serves to differentiate the largersites as probably playing some sort of centralorganizational role within their immediateregion. There is also a suggestion (as onemight expect) that larger sites are more likelyto be demographically viable in the long-term,with evidence for multiple periods of occupa-tion in the Early Bronze Age, and often conti-nuity into the Middle Bronze Age (Fig. 10.5).

While a small number of relatively largesites has been identified in the Cyclades in themiddle of the Early Bronze Age (Fig. 10.1:

Figure 10.3 Documented site sizes: Berbati-Limnes,Laconia, Methana, Nemea, Pylos andSouthern Argolid surveys.

Ayia Irini, Dhaskaleio), there is no evidencethat they formed focal points for surroundingsettlement systems, except in the case ofDhaskaleio, where the agricultural resourcesof the region immediately surrounding thesite are unlikely to have been able to supportits population, and resources were probablyimported from neighbouring islands such asthe Kouphonisia. Rather, it has been arguedthat these larger sites generally developed inthe context of inter-island exchange systems(Broodbank 1989; 1993).

In Crete, a relatively small number of largersites similarly stands out from the mass ofsmall hamlets and farmsteads (Whitelaw1983), though again, in the absence of inten-sive surveys in their immediate environs, itcannot yet be argued clearly that they stood atthe head of local integrated settlement sys-tems. Preliminary reports on several surveys,however, are suggestive (Watrous et al. 1993;Muller 1996), and the sequence of use oftombs in the Mesara suggests a pattern ofpopulation nucleation at a few focal sitesthrough time.


Figure 10.4 Surveyed areas and Early Helladic II site distributions: (a) Southern Argolid; (b) Berbati-Limnes;(c) Methana.

140 Todd Whitelaw

Figure 10.5 Relationship between site size and dura-tion of occupation, Berbati-Limnes, Laconiaand Southern Argolid surveys.

A considerable number of large EarlyBronze Age settlement sites is also knownfrom the past century of excavation in otherareas of the Aegean, usually situated in largelowland, often coastal plains (Fig. 10.6).Whether these stood at the head of local set-tlement systems, or represent a totally nucle-ated local settlement pattern, will only beestablished through intensive survey in theirhinterlands. The predominantly coastal loca-tion of such sites, while not entirely surpris-ing, since many lowland areas in the Aegeanopen onto the coast, raises the possibility thatthe development of such sites owes some-thing both to their links with smaller inlandsites, as well as extra-local connections, by sea.

There are also broad similarities in thechronological distribution of sites in differentregions, with large numbers of Early BronzeAge sites, and very restricted numbers of sitesin the later Early Bronze Age and MiddleBronze Age (Dickinson 1994: 50-60). Crete is aclear exception to this pattern, where, despitesome nucleation in the late prepalatial period,site numbers expand dramatically in mostregions surveyed in the Middle Bronze Age,undoubtedly part of the processes involved in

Figure 10.6 Distribution of excavated substantialnucleated sites of Early Bronze Age date.

the development of the Protopalatial states.While numerous surveys have been undertak-en in Crete, none in a major lowland locationhas yet been published in the detail which willbe necessary to understand this process.

In the Melian context, where these shiftswere first documented in detail, it was arguedthat the island population nucleated at the siteof Phylakopi, late in the Early Bronze Age. Onthe basis of information on site numbers andsite sizes, it was suggested that overall islandpopulation increased, at the same time thatsite numbers declined dramatically (Cherry1979; Wagstaff and Cherry 1982a). Attemptingto explain this process, it was argued that thiswas simply the first example of a recurrentpattern, in which population on Melos (andpotentially other small-scale regions) nucleat-ed when the island was integrally tied intooff-island economic and political systems(Wagstaff and Cherry 1982b; Renfrew 1982).In the prehistoric case, nucleation was consid-ered to be a response to the expansion of theinfluence of Minoan Crete into the Cyclades.


This remains the principal attempt toexplain, rather than simply to describe thesesettlement pattern transformations. Howeverit is problematic, in that the beginning of set-tlement nucleation on Melos can be firmlyplaced during the Phylakopi I period, c. 2200-2000 BC, whereas the first Minoan importsarrive in Phylakopi II, after 2000 BC. Eventhen, very few MM IA sherds have beenrecovered from the early phase of the MiddleBronze Age at the site, and intensive 'Minoan-ization' only begins late in the Middle BronzeAge, with large-scale importation and imita-tion of Minoan Neopalatial ceramics (Papa-giannopoulou 1991: 116-22). Settlement pat-tern changes in the Peloponnese, broadly con-temporary with those on Melos, similarly pre-date the earliest Minoan contacts, representedby imports of MM IA date at Ayios Stephanosand Lerna (Rutter and Zerner 1983). The situ-ation in the Cyclades may be further compli-cated, if the gap in the sequence proposed byRutter (1983; 1984) is sustained, since thetransformation from a dispersed EB II tonucleated EB III-MB pattern might not repre-sent a continuous process.

Even if one challenges the specific explana-tion proposed and widely accepted for Melos,the social perspective on settlement patternchange developed by Cherry remains crucial.The shift between dispersed and nucleatedsettlement systems must represent significantsocial transformations on Melos and else-where in the Aegean where a similar changehas been documented (Fig. 10.7), but whethersimilar social changes are documented,depends crucially on how similar the settle-ment changes were in the different areas.

While some differences have been identifiedamong Early Bronze Age settlement patternsin different parts of the southern Aegean (e.g.Cosmopoulos 1991), these have generallybeen noted at the descriptive level, and notpursued in terms of what they might mean for

Figure 10.7 Chronological distribution of sites identi-fied in Peloponnesian surveys: (a) by peri-od; (b) by phase, where data is available.

the social histories of the different areas. Infact, ostensibly similar patterns (Fig. 10.8),may have very different implications. OnMelos, no large centre is known through mostof the Early Bronze Age, until Phylakopiemerges late in the period, when its excavatedextent indicates that it was a substantial com-munity, and it would appear that the popula-tion of the earlier small, dispersed communi-ties, nucleated at Phylakopi, and that overallpopulation also expanded significantly. In theSouthern Argolid, in contrast, there was asmall number of substantial nucleated low-land communities occupied throughout theEarly Bronze Age, and a considerable numberof small hamlets and farmsteads which fluctu-ated through time, and disappeared late in theEarly Bronze Age. In this case, the large low-land communities formed a relatively stable

142 Todd Whitelaw

EARLYBRONZE

I

EARLYBRONZE

II

EARLYBRONZE

III

MIDDLEBRONZE

Figure 10.8 Settlement pattern comparison by phase, Melos and the Southern Argolid, large sites distinguished.Shading indicates areas not surveyed.


demographic pool throughout the period(and in some cases into the Middle BronzeAge), while the smaller, short-lived hamletsand farmsteads, despite their number, neverrepresented the majority of the regional popu-lation.

Therefore, superficially similar patternsappear to represent significantly differentdemographic and social histories (Fig. 10.9).Areas such as the Cyclades, generally lackingsuch larger, nucleated settlements duringmost of the Early Bronze Age, would havebeen much less densely occupied, and any set-tlement systems that existed would have beenless complex. In Melos, instability in smalldispersed sites was the norm during most ofthe Early Bronze Age, whereas in theSouthern Argolid, it was a minority variation,around the stable, large, lowland centres.Such a contrast would have engendered avery different pattern of interactions between

SOUTHERN ARGOLID: MIXED HIERARCHICALPATTERN (EH I-II) TO NUCLEATED (EH III-MH)

Figure 10.9 Site number and population estimate com-parisons, by phase: the Southern Argolidand Melos. For phase abbreviations, seeFigures 10.2 and 10.7 and text.

sites, and a different climate for cultural, aswell as for longer-term demographic develop-ment.

Critical to understanding such differences inregional development is the recognition andcomparative exploration of variation betweendifferent study areas. Broadly comparableintensive survey methodologies are beginningto allow such comparisons, and the distinc-tion between differences due to biases in sitepreservation and recovery, and those whichactually represent differences in behaviour inthe past. Despite concerns over biases, it ispossible to approach extensively or haphaz-ardly collected data in as critical a fashion,with results which also recognize and addressvariability in past behaviour (e.g. Forsen 1992;Broodbank 2000).

The Environmental Context for HumanSettlement in the Third Millennium BC

While geomorphological investigations wereconducted as part of the Melos project(Davidson et al. 1976; Davidson and Tasker1982), and proceeded in parallel with theintensive survey of the island, the implica-tions of the Melian environment for past set-tlement were approached in static, rather thandynamic terms (Wagstaff and Gamble 1982;Wagstaff et al 1982; Wagstaff and Cherry1982a; 1982b; Renfrew 1982; see also Bintliff1977). To some extent this has been the modelfor subsequent surveys, whether detailed geo-morphological studies have accompanied thesurvey fieldwork (e.g. Methana: James et al.1997; Berbati: Wells et al 1990; Zangger 1992)or not (Keos: Cherry et al 1991). A more inte-grated and dynamic perspective has charac-terized a number of studies reporting on thework of the Argolid Exploration Project (Popeand van Andel 1984; van Andel et al 1986; vanAndel and Runnels 1987; van Andel et al 1990;

MELOS: DISPERSED PATTERN(EC HI) TO NUCLEATED (MC)

144 Todd Whitelaw

van Andel and Zangger 1990; Jameson et al.1994; Runnels 1995). Regardless of how direct-ly such specific results have been integratedinto the interpretation of the archaeologicalsurvey data, each study, along with muchother, purely environmental research, has con-tributed to a move away from models of theAegean environment which argued thatchanges were climate driven (Vita-Finzi 1969;Bintliff 1977; Rackham 1982), toward therecognition of the importance of anthro-pogenic factors in landscape change (David-son 1980; Wagstaff 1981; Bruckner 1986; vanAndel et al 1990; Bintliff 1992).

There is an increasing number of soil andvegetation studies pointing to major changesin the southern Aegean environment duringthe fourth and third millennia, usually (butnot universally) argued to be the result ofhuman action (Fig. 10.10).1 By far the mostdetailed case has been made for the SouthernArgolid, by Tjeerd van Andel and his col-leagues (Pope and van Andel 1984; van Andelet al. 1986; van Andel and Runnels 1987;Jameson et al. 1994). There, a major phase ofsoil erosion has been linked to the expansionof settlement in the Final Neolithic and EarlyBronze Age.

In fact, the dating evidence for this episodein the Southern Argolid is extremely limited:an undated cist grave and a Late Helladic sitesit on top of the Pikrodaphni alluvium, a tho-rium/uranium disequilibrium determinationon carbonate yielded a date of 2700±1400 BC,and the most recent sherds recovered from thedeposits are said to have been Early HelladicII (Pope and van Andel 1984: table 3; Jamesonet al. 1994: 355). A similarly massive phase oferosion is better documented, dated, and pub-lished, in Zangger's work in the plain ofArgos, around Tiryns and Lerna (Zangger1991a; 1991b; 1993; 1994a) and in the Asineplain (Zangger 1994b). He has suggested asimilarly significant episode of slope instabili-

Figure 10.10 Locations with evidence for mid-Holocene environmental change in thesouthern Aegean (for sites and refer-ences, see note 1).

ty and large-scale erosion in his work in thenearby Berbati valley (Wells et al. 1990; Zangger1992), though in the latter case, no indepen-dent dating evidence has been reported.

Following the perspective developed in theSouthern Argolid, in each case catastrophicslope erosion is seen as a consequence of theexpansion of Final Neolithic and Early BronzeAge settlement, slope clearance, and thedegradation of vegetation through grazing bydomestic stock (Pope and van Andel 1984; vanAndel et al. 1986; 1990). This colonization ofslopes and upland areas is documented wide-ly throughout the Aegean (e.g. Warren andTzedakis 1974; Watrous 1982; Halstead 1984;1996), and parallels the widespread coloniza-tion of the smaller Aegean islands, well docu-mented as a fourth and third millennia phe-nomenon (Cherry 1981; 1990; Broodbank


1999). Both patterns appear to represent anexpansion of settlement into relatively mar-ginal landscapes, almost certainly facilitatedby the diversification of agricultural strategiesand an increased reliance on pastoral produc-tion (Halstead 1989).

Some have also linked increased soil erosionto possible evidence for the introduction ofthe plough in the middle of the Early BronzeAge (van Andel et al. 1986; Pullen 1992;Manning 1997:164), though the potential pro-ductive impact of the plough may be grosslyexaggerated, given the costs of traction ani-mals (Halstead 1987; Halstead and Jones 1989)and evidence that, in the Late Bronze Age,plough oxen may have been monopolized bythe palaces (Palaima 1989; 1992; Halstead1992).

Despite the enthusiasm with which thismodel for landscape change has been dissem-inated, it has been imposed on poorly datedsequences (Bintliff 1992; Endfield 1997), some-times actually against the cited evidence(Jameson et al 1994: 192 citing Demitrack1990). Extrapolating the date of this phenom-enon, from two well-dated contexts (theplains of Argos: Zangger 1993; Jahns 1993; andAsine: Zangger 1994b) to others without inde-pendent chronologies, also imposes a unifor-mity on the process which is likely to mask thelocal diversity which was initially so impor-tant in arguing for anthropogenic, rather thanclimatic, causes for landscape change in thesouthern Aegean (Davidson 1980; Wagstaff1981; van Andel et al 1990; Bintliff 1992).

While there is no precise synchronization ofgeomorphological change across the Aegeanregion, there are hints of a more dynamic rela-tionship with the pattern of development ofagriculture in different regions (van Andel etal 1990). In addition, relatively small scale cli-matic variations during the fourth and thirdmillennia (Wright 1972; Rackham 1982;Bottema 1985; 1994; Allen and Katsikis 1990;

Moody et al 1996) are likely to have affectedindividual areas in different ways, given dif-ferent geologies, vegetation patterns, historiesof human settlement, and micro-climatic con-ditions. In other words, we are likely to bedealing with much more subtle interactionsbetween climate, soils, vegetation, and humanexploitation strategies, which played throughin different ways, even in adjacent valley sys-tems. While local variation is acknowledgedto be significant (van Andel et al 1990;Zangger 1992), this is masked by attemptingto conflate different sequences to a commonmodel without independent chronologicalsupport (e.g. Zangger 1992; Jameson et al1994: 192). As developed to date, the model isalso over-deterministic (van Andel et al 1986;Runnels and van Andel 1987), allowing littlescope for the possibility of alternativeexploitation strategies, differentially pursuedby different local groups and individuals,depending on how they evaluated their socialas well as environmental opportunities andconstraints.

Overall, the coarse chronological resolutionof most environmental studies in the southernAegean allows considerable ambiguity as tothe date of environmental change in specificareas, and about the degree to which vegeta-tion and geomorphological changes can besynchronized with each other, and with settle-ment pattern changes in specific regions. Theevidence for anthropogenic causation is cir-cumstantial, synchronization of processesbetween regions is often assumed rather thandemonstrated, and even the correlationbetween the environmental and settlementsequences is ambiguous.

Dating of sediment or pollen sequences isproblematic and expensive, and it is undoubt-edly the case that environmental studies usu-ally focus on relatively long periods of time,such that a few dates are viewed as sufficientto anchor a sequence within the Holocene.

146 Todd Whitelaw

With so few absolute dates, individual pollencores are often dated by similarities in thepollen assemblages to other cores, thereby rul-ing out the possibility of detecting variationsin local sequences. However, as analysts areincreasingly considering anthropogenic caus-es for landscape and vegetation changes,more fine-grained dating will be crucial toestablish correlations with cultural sequences,and to document and explain variation inthese processes in time and space.

In the previous section, it was suggestedthat Aegean prehistorians have attempted tofit varying local settlement sequences to a sin-gle simple model, thereby obscuring the dif-ferences of detail which are likely to be crucialto understanding the variable processes actu-ally involved. In this section, it has been sug-gested that environmentalists have tended tofollow a similar path. At an early stage inanalysis, this is crucial for assembling dis-parate data and allowing the detection ofbroad trends, but at some point, forcing datato fit too rigid a model becomes counter-pro-ductive and inhibits the recognition andpotential for understanding complex process-es. If, in addition, both sequences are signifi-cantly over-simplified, there is likely to be lit-tle insight in bringing them into conjunction.We need to employ models which are specific,but also open-ended, such that we can recog-nize when they do not fit, yet learn from thedivergences from our expectations.

So to develop a more effective understand-ing of the ways in which humans interactedwith their environments, and in turn, howthose environments acted back on the humanpopulations, both in specific cases and moregenerally, we need far finer resolution in bothcultural and environmental datasets, and inparticular, more direct associations betweenboth sets of patterns—linking cause and effect—as well as more detailed modelling of thepotential dynamic interactions.

Linking Behavioural Cause andEnvironmental Effect: Occupation andErosion at Markiani, Amorgos

Some progress towards resolving these uncer-tainties can be made by more explicit integra-tion of environmental and archaeologicalinvestigations. A trial investigation alongthese lines has been undertaken at the site ofMarkiani on the island of Amorgos, in thesouth-east Cyclades (Fig. 10.11).

Markiani is a small site, occupied from thelate Grotta-Pelos, through to the Kastri phasesof the Cycladic Early Bronze Age (Davis 1992:752-53; Marangou 1994: 470-71). Later occupa-tion of the site was primarily in the Hellenisticperiod, when it may have been an isolatedfarmstead. Intensive surface collection hasdocumented a 14 ha distribution of EarlyCycladic sherds, which have eroded from anoriginal occupation area of c. 0.3 ha on thesummit of the hill. The site is badly eroded,and was investigated through limited excava-tion. Together with the surface material, thisdocuments the expansion of the occupied areafrom the summit of the hill, down onto theupper southern slope, during the course ofperhaps 700 to 800 years. While the initialassumption has been that the site graduallyexpanded through time, continuous occupa-tion from the late Grotta-Pelos through theKastri phase would make it exceptionalamong the usually short-lived Early Cycladicsettlement sites. However, the extremely bat-tered condition of the pottery from the firsttwo phases, and the very slight depth of pre-served deposit, suggest that occupation wasepisodic rather than continuous. Similarly,while differences in patterns and rates ofdeposition can be anticipated both betweensites, and at different times on the same site, agross comparison of depositional rates at pre-historic sites across the Aegean suggests thatthe depth of deposit at Markiani would be


INTENSIVESURFACECOLLECTIONSHERDDENSITY25 GM/M2

SHERDDENSITY100 GM/M2

Figure 10.11 Location of Markiani, Amorgos.

148 Todd Whitelaw

completely anomalous if occupation had beencontinuous (Fig. 10.12).

In trying to understand the processesinvolved in the post-depositional spread ofmaterial from the hilltop, sediments on thesite and the slope immediately below it werestudied through in situ examination of sec-tions at various points on the slope, andmicromorphological analysis (French andWhitelaw 1999). This allowed the identifica-tion of a redeposited red soil underlying theterraces at the base of the slope, which origi-nally formed during more wooded condi-tions, and subsequently eroded downslope.The structure of the sediments indicates atleast two subsequent phases of slope erosion,while the formation of an A horizon indicatesstability for a substantial period after slopeerosion ceased. The fill of the collapsed Kastriphase structures at the summit of the site indi-cates deflation and capping by a lag deposit,protecting the underlying sediments and

marking the onset of slope stability, relativelysoon after the abandonment of the site. A laterphase of erosion is documented by the sub-stantial fills which were subsequently re-worked in the construction of the recent ter-races still preserved on the slope. This erosionis likely to have resulted from the re-occupa-tion of the site during the Hellenistic period(Fig. 10.13).

Because of the physical linkage between thesite and the slope immediately below, it hasbeen possible to tie the erosion sequence onthe slope directly to the occupation sequenceon the site above. Sampling below the occupa-tion site at Markiani, we are probably not see-ing the erosion effects of clearance for fieldsper se, since the gentler-sloping and better-watered land to the north of the site (Fig. 10.11) would probably have been preferentiallycultivated. Rather, the erosion episodes aremore likely to represent the impact of fire-wood collection and over-grazing in the

Figure 10.12 Deposit build-up rates for stratified sites in the Bronze Age Aegean.

ESTIMATED LENGTH OF OCCUPATION (YEARS)


Ca. AD 1960-1990

MARKIANIRECONSTRUCTED SEQUENCE

OF SLOPE DEVELOPMENT

SLIGHT BREAKDOWN OF TERRACES IN RECENTDECADES, FOLLOWING ABANDONMENT FORAGRICULTURE, BUT CONTINUED ANIMALGRAZING.

CONSTRUCTION OF SLOPE TERRACES, DURINGRECENT PAST, MIXING MOST SLOPE DEPOSITS,EXCEPT FOR DEEPEST LEVELS, ON GENTLESLOPES.

RENEWED EROSION OF SLOPE SOILS, SEALINGEARLIER DEPOSITS, RESULTING FROM RENEWEDOCCUPATION, PRINCIPALLY DURING THEHELLENISTIC PERIOD.

SLIGHT DEVELOPMENT OF IN SITU SOIL,INDICATING RELATIVELY LONG PERIOD OFSLOPE STABILITY.

DEPOSITION OF CLAY LININGS OF VOID SPACES,AS THE RESULT OF FURTHER EROSION OF BARESLOPES DURING PERIOD OF EARLY BRONZE AGEOCCUPATION.

EROSION OF ORIGINAL RED SOIL, ANDDEPOSITION ON LOWER SLOPE, COMMENCINGWITH INITIAL OCCUPATION OF THE SITE.

Figure 10.13 Reconstructed sequence of slope development at Markiani, Amorgos.

Ca. AD 1850-1930

Ca. 800 BC-AD 400

Ca. 2200-800 BC

Ca. 2800-2200 BC

Ca. 2800 BC

150 Todd Whitelaw

immediate vicinity of the site, but even so,even a small-scale occupation can be seen tohave had a very significant impact on vegeta-tion and slope stability in its vicinity.

Indeed, given the amount of erosion docu-mented, it is worth exploring this linkagereflexively—just as the occupation sequencecan be held to be responsible for the pattern ofslope erosion around the site, erosion on sucha scale is, in turn, likely to have been a majorfactor in the periodic abandonment of the site.While soils would not have redeveloped fastenough to account for the reoccupation of thesite, it seems likely that, as other nearby areaswere also over-exploited, areas previouslysomewhat degraded were returned to andexploited further.

While this is only a single case, it does estab-lish a direct linkage between settlement histo-ry and landscape degradation which has beenhypothesized but not demonstrated, else-where in the southern Aegean. While method-ologically showing that such linkages can beestablished, substantively, it also indicates thesevere environmental impact that even small-scale and short-lived settlement could have onmarginal slopes in southern Aegean environ-ments.

Landscape Degradation, SettlementInstability and Inter-CommunityCompetition

It is now worth returning to our environmen-tal and cultural sequences, to see how thepotential linkages between them can be devel-oped more explicitly. It was argued above thatwe have a significant problem linking thesedifferent sequences in any causal fashion,which is not helped by simply assuming thatdifferent local sequences, whether cultural orenvironmental, fit a single process during onechronological horizon. We are likely to devel-

op a better understanding of the differentprocesses involved by paying more attentionto local differences, and by considering intheir own right the implications of differentvariables, such as overall population levels,the relative sizes of individual sites, the rela-tive distribution of population among sites ofdifferent sizes, and the location of differenttypes of sites within the landscape.

Any move toward explanation of thechanges in either settlement or environment,will also require a more subtle sense of thedynamics behind the settlement transforma-tions documented throughout the southernAegean near the end of the third millenniumBC. For this, we need to be clearer on what issimilar and different in the settlement trajecto-ries of different regions, and therefore, whatchanges and transformations need to beexplained at the local scale, and which requireexplanation in terms of more inclusive,regional and inter-regional processes. Withthe hints at diversity outlined above, no singlemodel is likely to be relevant to understand-ing settlement pattern changes throughout thesouthern Aegean. In this case, it seems worth-while to try to pull together observations onsome of the different phenomena involved,which may point toward productive areas forfuture research.

Given the short life span which can beinferred for most small Early Bronze Age sitesin the southern Aegean, all of our chronologi-cal distribution maps are likely to be extremepalimpsests, which do not allow us to appre-ciate the degree of instability of settlement.Even in south-central Crete, where many ofthe round tombs reveal evidence of use forover a millennium, we are not necessarilydealing with particularly stable communities.Some of the tombs may have been used con-tinuously, but that need not imply stability ofsettlement. In the intensively surveyedAyiopharango valley (Blackman and Branigan


1977; Vasilakis 1989-90), the existence of sev-eral occupation sites for each tomb has beenused to argue that each tomb served severalhamlets or farmsteads. However, the occupa-tion sites rarely demonstrate long-term occu-pation, and are more likely to represent asequence of occupation sites shifting locationthrough time, tethered to a tomb serving as aterritorial focus and legitimation of controlover productive resources in its vicinity. Simi-larly, while the ceramics published from anytomb other than Ayia Kyriaki (Blackman andBranigan 1982) represent tiny and unreliablesamples of the ceramics originally deposited,the chronological gaps in the documented useof many tombs may indicate episodic ratherthan continuous use. On this reading, the evi-dence for clearing out and burning insidemany tombs may represent, not part of thefunerary ritual (Branigan 1992: 119-27), butrather cleaning out for re-use by a new kin-group, after a period of abandonment.

Such small-scale, unstable occupations—predominantly farmsteads and small hamlets—characterize much of the Final Neolithicand Early Bronze Age throughout the south-ern Aegean, and it is their disappearancetoward the end of the period which can beseen as a broadly synchronous pattern acrossthe region. They appear to represent small-scale colonization of the most marginal envi-ronments, an extreme manifestation of theprogressive colonization of marginal environ-ments which characterizes much of theAegean during the fourth and third millenniaBC. Their development, and their eclipse, as anAegean-wide phenomenon, would appear torequire explanation at a broad, inter-regional,rather than local, level.

There is a much more variable distributionof larger, more demographically stable com-munities, whose viability will have beenlinked directly to the agricultural resourcesavailable in their vicinity. With generally

deeper soils and better water resources, agri-culture in their vicinity would generally havebeen more productive and more reliable.Where they appear, they can be seen as part ofa longer-term pattern of expansion of popula-tion through successful development ofNeolithic agricultural strategies, supplement-ed and made more resilient through diversifi-cation with new species, the development ofcomplementary pastoral resources, and theparallel development of new social strategiesbuffering local communities against risk anduncertainty (Halstead 1981; 1989; 1994).

While it was opportune and effective, in try-ing to develop a social perspective on theAegean Early Bronze Age, for Renfrew toemphasize similarities and draw upon differ-ent types of evidence from different areas todevelop a common outline of 'proto-urban'societies throughout the southern Aegean(1972a), subsequent research has only servedto further differentiate the cultures at thebroad regional level (e.g. Davis 1992; Rutter1993; Watrous 1994), and we are increasinglyable to pick up finer distinctions withinregions (e.g. Whitelaw 1983; 1999; Broodbank1989; Wiencke 1989). The presence and natureof nucleated settlements, and the degree towhich they may have acted as local centres,appears to vary significantly within andbetween different areas of the Aegean.Therefore, such sites are not necessarily a uni-tary phenomenon, and their pattern of devel-opment needs to be considered in terms of dif-ferent patterns of inter-play between broadinter-regional and local level variables.

The nucleation of population at large low-land, often coastal sites, at the end of the EarlyBronze Age and in the Middle Bronze Age, insome cases develops a pattern seen already inparticularly favoured locations in the EarlyBronze Age. In such regions, relatively largesites already existed in the Early Bronze Age,and 'nucleation' only appears to be a later

152 Todd Whitelaw

Early Bronze Age phenomenon because of theloss of the smaller, more ephemeral dispersedsites in the uplands. In other areas, such as theCyclades, such sites appear to represent acompletely new social formation, with a realaggregation of formerly dispersed popula-tions. That these communities did not havetheir roots in the few large sites of the midEarly Bronze Age, such as Dhaskaleio andAyia Irini, is suggested by the abandonmentof the latter. These early centres developed ina different social and economic context, owingless to local agricultural resources and more totheir location on communication routes, givenenhanced importance precisely because of theneed for inter-island interaction in such mar-ginal environments.2

The concentration of population in a smallnumber of large, lowland communities dur-ing the Middle Bronze Age, was likely to havebeen a variable phenomenon, in some areaspart of a longer-term process of developmentfrom the Neolithic, but in others, responsiveto short-term, local ecological dynamics.Perhaps not completely fortuitous in the moremarginal areas, the erosion of upland soils islikely to have benefited agricultural exploita-tion of the lowlands, where such soils wereredeposited. The widespread, near completenucleation of population into such communi-ties, while undoubtedly contributing to thenew social contexts out of which developedmore politically-centralized societies in theMiddle Bronze Age, is unlikely to be explainedas the result of a single process throughout thesouthern Aegean.

Closer consideration of settlement patternsshould involve not just the individual com-munities and their relations with each other,but also their specific locations within theenvironment. Site locations in prehistorythroughout the southern Aegean, as can bedocumented in detail for Melos and north-west Keos (Fig. 10.14), but assessed more

superficially for other published surveys,clearly favoured low-slope land, which couldbe cultivated without requiring slope terrac-ing for soil conservation. While agriculturalterraces, or at least gully check-dams, of laterBronze Age date have been investigated onPseira (Betancourt and Hope Simpson 1992)and in the western Mesara (Watrous et al.1993; Gifford and Reese 1995), and morerecently suggested for Kythera (Broodbank inBlackman 2000: 27), no survey anywhere inthe Aegean has yet revealed a distribution ofprehistoric sites relative to slope which wouldsuggest that terraces were used in soil conser-vation on any scale. If farmers were avoidingthe most unstable slopes, this would seem, onthe face of it, to limit the potential impact ofslope erosion on cultivated areas. However,there is no absolute cut-off such that fieldsabove a certain angle of slope are at risk andthose below are not—the situation is one ofdegrees of risk, which will also be affected bylocal conditions such as the nature of the soil,the cultivation techniques, the vegetation, theaspect and exposure. Even fields on gentleslopes and in valley bottoms would be affect-ed by increased water run-off from higherslopes, deforested or denuded through clear-ance and grazing. Similarly, the re-depositionof the quantity of soil moved downslope, doc-umented in areas such as the Argolid, willhave had short-term catastrophic impacts onfields and crops.

Regardless of the local details, the instabili-ty of small sites and their widespread disap-pearance late in the Early Bronze Age, appearsto represent more than demographic process-es alone; without effective soil conservationmeasures, Early Bronze Age populations inmarginal areas were literally eating their waythrough the landscape, denuding slopes ascatastrophic soil erosion eventually followedclearance for cultivation and, particularly,local grazing. Ironically, it may have been an


Figure 10.14 Slope angle and distribution of prehistoric sites: (a) Northwest Keos survey; (b) Melos survey. Slopesgreater than 15° shaded.

154 Todd Whitelaw

increasing reliance on animals as part of adiversified agro-pastoral strategy—whichmade colonization of such environmentsviable in the first place—which caused ero-sion and large-scale abandonment of uplandareas throughout the southern Aegean by theend of the Early Bronze Age.

Given the impact on the local landscape ofeven small-scale occupation, as documentedat Markiani, settlement instability would havebeen particularly acute in areas of shallowsoils, and steep and unstable slopes. Ignoringlocal details of soil quality and access to water,these areas are also those most marginal foragriculture, and were the landscapes colo-nized latest—both regionally and locally. Thedenudation of the uplands would have been acumulative process, since slope soils were lostfar faster than they could develop. We shouldtherefore anticipate increasing pressure on landas a resource during the course of the EarlyBronze Age, and an increasing frequency oflocal resource crises, as traditional risk-buffer-ing mechanisms came under increasing stress.

The distinctive horizon of fortified sites,variably dated in different areas, but broadlydeveloping in the second half of the third mil-lennium BC, is puzzling when considered inthe context of distinct local settlement trajec-tories. Traditionally, such defences wereassumed to be a response to threats from out-side each local region, but no intrusive popu-lations have been convincingly identified(Forsen 1992; Sotirakopoulou 1993). Defensivemeasures, documented at small and remotesites as well as large lowland centres, wouldfit in a context of increasingly acute competi-tion for land, far better than competition formetalwork and other elite goods, as has usu-ally been suggested to date, since the evidencefor the latter is far more variable inter-region-ally (Cosmopoulos 1991). Just as the processof colonization of marginal landscapes is rele-vant to the entire region, though the timing

varies as a result of local settlement history, soare the consequences, as landscapes becamesubject to similar pressures and were affectedby similar processes.

Finally, social inequality, defined as unequalaccess to basic productive resources, needs tobe considered, not simply as a trend throughthe period, but as an element in the dynamicswhich worked in different ways in differentareas of the Aegean. This was not obviouslyan issue in the earlier Neolithic, when occupa-tion was focused on lowland basins. How-ever, expansion of at least some of the region-al population onto more marginal areas of thelandscape, with lower and less reliable agri-cultural productivity, will have engenderedsignificant inequalities in access to the basicmeans of production (Halstead 1995). Increasinginstability and unreliability in productionthrough the third millennium, differentiallyconcentrated in such marginal contexts,would only have exacerbated such inequali-ties, within and particularly between commu-nities, creating new forms of stress and inter-community competition.

This contribution has not intended to putforward population expansion, cultivation ofmarginal slopes and environmental degrada-tion as a universal process, or the prime-mover in the settlement pattern and socialchanges which marked the end of the EarlyBronze Age throughout the southern Aegean.However, it has tried to bring another elementinto the arguments, which mediates betweenthe purely social, and largely deterministicenvironmental perspectives developed todate. That the expansion of small-scale settle-ment into upland areas does appear to be ageneral phenomenon within a long-termprocess, its long-term consequence—the over-exploitation and degradation of marginal andupland areas—may provide a common ingre-dient which links otherwise more divergentlocal settlement trajectories.


Prospects: Integrating Settlement andEnvironmental Sequences in LandscapeArchaeology

The development of intensive survey has dra-matically affected the way Aegean prehistoryis approached, emphasizing long-term pro-cesses, regional scale patterns, and compara-tive perspectives. While this has led to anincreasing convergence both in questions anddata resolution, to date, archaeological andenvironmental investigations in the Aegeanregion have progressed largely in parallel,even where the practitioners have workedtogether on specific site-based or regional pro-jects. Fully integrated research, which wouldallow ecologically sophisticated explorationof social and demographic development with-in dynamic environmental contexts, is diffi-cult to achieve.

Methodologically, we have tended to com-pare parallel sequences, only broadly correlat-ed chronologically, and have simply assertedcausal connections which cannot be demon-strated because of the coarseness or uncertain-ty of our respective chronologies. More directlinkages, however, can be developed, if rele-vant investigation contexts are sought, andappropriate investigative strategies are pur-sued. Similarly, more integrated approachesto research design are required to ensure thatenvironmental research is fully integratedfrom the start, rather than running in parallelto the archaeological project (or even being anafterthought), and being reported as a back-ground chapter or relegated to an appendix.The research that has been undertaken to datedemonstrates that the potential is there, but 25years after the initiation of multi-disciplinaryresearch in the Aegean (McDonald and Rapp1972; Renfrew and Wagstaff 1982; Jameson etal. 1994), full and effective integration is still inthe future.

Acknowledgments

I would like to thank the editors for the invi-tation to contribute to the original workshop,and to the other participants for comments onthe presented paper. Fieldwork at Markianiwas encouraged and facilitated by LilaMarangou and Colin Renfrew, and assisted bySimos Giannakos, Georgios Gavalas, NatashaAngelopoulou, Kasia Gdaneic and Julie Boast.Permission to take soil samples was grantedby Lila Marangou and the KB' Ephorate forPrehistoric and Classical Antiquities (Dode-canese), and all soil analyses were undertakenby Charly French. The field and laboratorywork was funded by the British Academy, theMcDonald Fund for Archaeological Research,and St John's College Cambridge. I am grate-ful to all of these colleagues and institutionsfor their assistance and support. My views onEarly Bronze Age settlement strategies havebeen honed by two decades of debate withPaul Halstead, John Cherry, Tjeerd van Andeland Cyprian Broodbank, though it wouldn'tbe fair to blame them for the views expressedhere. Helpful comments on a draft of this con-tribution were received from the editors, LisaNevett, Cyprian Broodbank and SteveShennan, to whom I am also indebted.

Notes

1. Locations where environmental investigationshave documented later prehistoric vegetationand landscape change in the southern Aegean,numbered as in Figure 10.10.

Vegetation changes in the fourth and third millennia

BC

Copais (1): Turner and Greig 1975; Greig andTurner 1974; Rackham 1983; Allen 1990Kitsos (3): Renault-Miskovsky 1981Lerna (7): Jahns 1990; Jahns 1993

156 Todd Whitelaw

Naxos (14): Renault-Miskovsky 1983; Dalonge-ville and Renault-Miskovsky 1993Tersana (18): Moody et al 1996

Vegetation changes with broader chronological limitsKoiladha (11): Bottema 1990Thermisia (12): Whitehead and Sheehan 1981Amorgos (16): French and Whitelaw 1999Aghia Galini (19): Bottema 1980; 1996

Vegetation changes in the second millennium BCMessenia: Wright 1972; Zangger et al 1997

Major erosion events in the fourth and thirdmillennia BCCopais (1): Allen 1990Southern Argolid (10): Pope and van Andel1984; van Andel et al 1990; van Andel andZangger 1990; van Andel et al 1986Tiryns (8): Zangger 1991a; 1993; 1994aLerna (7): Zangger 1991b; 1993;Asine (9): Zangger 1994bNaxos (14): Dalongeville and Renault-Miskovsky 1993Amorgos (16): French and Whitelaw 1999Kythera (13): Broodbank, in Blackman 2000: 27West Mesara (20): Watrous et al 1993

Major erosion events with broader chronologicallimitsAyiofarango (22): Doe and Holmes 1977Berbati (6) (not directly dated): Wells et al 1990;Zangger 1992

Major erosion events in the second millennium BCThorikos (2): Paepe and Deraymaeker 1973Melos (15): Davidson et al 1976; Davidson andTasker 1982Thera (17): Davidson 1978; Limbrey 1990Kommos (21): Gifford and Reese 1995; Parsonsand Gifford 1995

Possible minor erosion events within broad limitsMethana (5): James et al 1994; 1997

Soil and vegetation changes at Nemea (4) are datedearlier in the NeolithicDemitrack 1990; Atherden et al 1993; Atherdenand Hall 1994

2. Relatively large Early Cycladic sites such asSkarkos on los (Marthari 1997) and Chalan-driani on Syros (Marthari 1998) may have clos-er parallels with the large lowland EarlyHelladic sites, than with contemporary sitessuch as Dhaskaleio and Ayia Irini in the islands.

Bibliography

Alcock, S., J. Cherry, and J. Davis1994 Intensive survey, agricultural practice and the

classical landscape of Greece. In I. Morris (ed.),Classical Greece. Ancient Histories and ModernArchaeologies: 164-65. Cambridge: CambridgeUniversity Press.

Allen, H.1990 A postglacial record from the Kopais basin,

Greece. In S. Bottema, G. Entjes-Nieborg and W.van Zeist (eds.), Man's Role in the Shaping of theEastern Mediterranean Landscape: 173-82. Rotter-dam: Balkema.

Allen, H. and A. Katsikis1990 Environmental and climatic change in the

Aegean during the late Quaternary. In D. Grove,J. Moody and O. Rackham (eds.), Stability andChange in the Cretan Landscape: 9-15. Petro-marula 1. Perivolia, Khania: Boutsounaria.

Andel, T. van and C. Runnels1987 Beyond the Acropolis: a Rural Greek Past. Stanford:

Stanford University Press.1995 The earliest farmers in Europe. Antiquity 69:

481-500.Andel, T. van, C. Runnels and K. Pope

1986 Five thousand years of land use and abuse inthe southern Argolid. Hesperia 55:103-28.

Andel, T. van and E. Zangger1990 Landscape stability and destabilization in the

prehistory of Greece. In S. Bottema, G. Entjes-Nieborg and W. van Zeist (eds.), Man's Role inthe Shaping of the Eastern Mediterranean Land-scape: 139-58. Rotterdam: Balkema.


torical Greece. Journal of Field Archaeology 17:379-96.


Atherden, M. and J. Hall1994 Holocene pollen diagrams from Greece.

Historical Biology 9: 117-30.Atherden, M., J. Hall and J. Wright

1993 A pollen diagram from the northeast Pelopon-nese, Greece: implications for vegetation historyand archaeology. The Holocene 3: 351-56.

Betancourt, P. and R. Hope Simpson1992 The agricultural system of Bronze Age Pseira.

Cretan Studies 3: 47-54.Bintliff, J.L.

1977 Natural Environment and Human Settlement inPrehistoric Greece. BAR Supplementary Series 28.Oxford: British Archaeological Reports.

1992 Erosion in Mediterranean lands: a reconsidera-tion of pattern, process and methodology. In M.Bell and J. Boardman (eds.), Past and Present SoilErosion. Archaeological and Geographical Perspec-tives. Oxbow Monograph 22: 125-32. Oxford:Oxbow Books.

Blackman, D.2000 Annual Report of the Council, British School at

Athens, 1998-99. London: British School atAthens.

Blackman, D. and K. Branigan1977 An Archaeological Survey of the Lower Catch-

ment of the Ayiofarango Valley. BSA 72: 13-84.1982 The excavation of an Early Minoan tholos tomb

at Ayia Kyriaki, Ayiofarango, southern Crete.BSA 77: 1-57.

Bottema, S.1980 Palynological investigations on Crete. Review of

Palaeobotany and Palynology 31:193-217.1985 Palynological investigations in Greece with spe-


1990 Holocene environment of the Southern Argolid:a pollen core from Kiladha Bay. In T.J. Wilkinsonand S. Duhon. Franchthi Paralia. The Sediments,Stratigraphy, and Offshore Investigations. Excava-tions at Franchthi Cave, Greece, Fascicle 6: 117-38. Bloomington: Indiana University Press.

1994 The prehistoric environment of Greece: a reviewof the palynological record. In N. Kardulias(ed.), Beyond the Site: Regional Studies in theAegean Area: 45-68. Lanham: University Press ofAmerica.

1996 Notes on the Holocene vegetation history ofCrete. In D.S. Reese (ed.), Pleistocene and Holo-cene Fauna of Crete and Its First Settlers.Monographs in World Archaeology 28: 53-59.Madison: Prehistory Press.

Branigan, K.1992 Dancing with Death. Life and Death in Southern

Crete c. 3000-2000 BC. Amsterdam: Adolf M.Hakkert.

Broodbank, C.1989 The longboat and society in the Cyclades in the

Keros-Syros culture. A]A 93: 319-37.1993 Ulysses without sails: trade, distance, knowl-

edge and power in the early Cyclades. WorldArchaeology 24: 315-31.

1999 Colonization and configuration in the insularNeolithic of the Aegean. In P. Halstead (ed.),Neolithic Society in Greece: 15-41. Sheffield:Sheffield Academic Press.

2000 An Island Archaeology of the Early Cyclades.Cambridge: Cambridge University Press.

Bruckner, H.1986 Man's impact on the evolution of the physical

environment in the Mediterranean region in his-torical times. Geojournal 13: 7-17.

Cavanagh, W., J. Crouwel, R. Catling and G. Shipley1996 Continuity and Change in a Greek Rural Landscape:

the Laconia Survey, 2. BSA SupplementaryVolume 27. London: British School at Athens.

Cherry, J.F.1979 Four problems in Cycladic prehistory. In J.

Davis and J. F. Cherry (eds.), Papers in CycladicPrehistory. UCLA Monograph 14: 22-47. LosAngeles: UCLA Institute of Archaeology.

1981 Pattern and process in the earliest colonisationof the Mediterranean islands. Proceedings of thePrehistoric Society 47: 41-68.

1983 Frogs round the pond: perspectives on currentarchaeological survey projects in the Mediter-ranean region. In D. Keller and D. Rupp (eds.),Archaeological Survey in the Mediterranean Area.BAR International Series 155: 375-416. Oxford:British Archaeological Reports.

1990 The first colonization of the Mediterraneanislands: a review of recent research. JMA 3: 145-221.

1994 Regional survey in the Aegean: the 'New Wave'(and after). In N. Kardulias (ed.), Beyond the Site:Regional Studies in the Aegean Area: 91-112.Lanham: University Press of America.

Cherry, J.F., J. Davis, A. Demitrack, E. Mantzourani, T.Strasser and L. Talalay

1988 Archaeological survey in an artifact-rich land-scape: a Middle Neolithic example from Nemea,Greece. A]A 92: 159-76.

Cherry, J., J. Davis and E. Mantzourani1991 Landscape Archaeology as Long-term History:

Northern Keos in the Cycladic Islands. Monumenta

158 Todd Whitelaw

Archaeologica 16. Los Angeles: University ofCalifornia Press.

Cosmopoulos, M.1991 The Early Bronze 2 in the Aegean. Studies in

Mediterranean Archaeology 98. Jonsered: PaulAstrom.

Dalongeville, R. and J. Renault-Miskovsky1993 Paysages passes et actuels de Tile de Naxos. In

R. Dalongeville and G. Rougement (eds.),Recherches dans les Cyclades: 9-57. Lyon: Maisonde TOrient Mediterranean.

Davidson, D.1978 Aegean soils during the second millennium BC

with reference to Thera. In C. Doumas (ed.),Thera and the Aegean World, 1: 725-39. London:The Thera Foundation.

1980 Erosion in Greece during the first and secondmillennia B.C. In R.A. Cullingford, D. Davidsonand J. Lewin (eds.), Timescales in Geoarchaeology:143-58. Chichester: Wiley.

Davidson, D., A.C. Renfrew and C. Tasker1976 Erosion and prehistory in Melos: a preliminary

note. JAS 3: 219-27.Davidson, D. and C. Tasker

1982 Geomorphological evolution during the lateHolocene. In C. Renfrew and M. Wagstaff (eds.),An Island Polity: the Archaeology of Exploitation inMelos: 82-94. Cambridge: Cambridge UniversityPress.

Davis, J.1992 Review of Aegean prehistory 1: the islands of

the Aegean. A]A 96: 699-756.Demitrack, A.

1990 Geological and geomorphological studies. In J.Wright, J.F. Cherry, J.L. Davis, E. Mantzourani,S.B. Sutton and R.F. Sutton Jr., The NemeaValley Archaeological Project: a preliminaryreport. Hesperia 59: 587-91.

Dickinson, O.1994 The Aegean Bronze Age. Cambridge: Cambridge

University Press.Doe, A. and D. Holmes

1977 The physical environment. In D. Blackman andK. Branigan, An archaeological survey of thelower catchment of the Ayiofarango valley. BSA72:14-24.

Endfield, G.1997 Myth, manipulation and myopia in the study of

Mediterranean soil alluviation. In A. Sinclair, E.Slater, and J. Gowlett (eds.), ArchaeologicalSciences 1995. Oxbow Monographs 64: 241-48.Oxford: Oxbow Books.

Forsen, J.1992 The Twilight of the Early Helladics. Studies in

Mediterranean Archaeology and LiteraturePocket-book 116. Jonsered: Paul AstromsForlag.

1996 The Early Helladic period. In B. Wells and C.Runnels (eds.), The Berbati-Limnes ArchaeologicalSurvey 1988-1990. Skrifter utgivna av SvenskaInstitutet i Athen 44: 75-120. Stockholm: PaulAstrom.

French, C. and T. Whitelaw1999 Soil erosion, agricultural terracing and site for-

mation processes at Markiani, Amorgos, Greece:the micromorphological perspective. Geoarch-aeology 14:151-89.

Gifford, J. and D. Reese1995 The physical geology of the western Mesara and

Kommos. In J.W. Shaw and M.C. Shaw (eds.),Kommos 1: the Kommos Region and Houses of theMinoan Town: 30-90. Princeton: PrincetonUniversity Press.

Greig, J. and J. Turner1974 Some pollen diagrams from Greece and their

archaeological significance. JAS 1:177-94.Hagg, R. and D. Konsola (eds.)

1986 Early Helladic Architecture and Urbanization.Studies in Mediterranean Archaeology 76.Goteborg: Paul Astrom.

Halstead, P.1981 From determinism to uncertainty: social storage

and the rise of the Minoan palace. In A.Sheridan and G. Bailey (eds.), Economic Archa-eology. BAR International Series 96: 187-213.Oxford: British Archaeological Reports.

1984 Strategies for Survival: an Ecological Approach toSocial and Economic Change in the Early FarmingCommunities of Thessaly, N. Greece. UnpublishedPhD dissertation, Cambridge University.

1987 Traditional and ancient rural economy in Medi-terranean Europe: plus c,a change? JHS 107: 77-87.

1989 The economy has a normal surplus: economicstability and social change among early farmingcommunities of Thessaly, Greece. In P. Halsteadand J. O'Shea (eds.), Bad Year Economics: CulturalResponses to Risk and Uncertainty: 68-80.Cambridge: Cambridge University Press.

1992 Agriculture in the Bronze Age: towards a modelof palatial economy. In B. Wells (ed.), Agriculturein Ancient Greece: 105-17. Stockholm: SwedishInstitute at Athens.

1994 The north-south divide: regional paths to com-plexity in prehistoric Greece. In C. Mathers and


S. Stoddart (eds.), Development and Decline in theMediterranean Bronze Age. Sheffield Archaeol-ogical Monograph 8: 195-219. Sheffield: J.R.Collis Publications.

1995 From sharing to hoarding: the Neolithic foun-dations of Aegean Bronze Age society. In R.Laffineur and W.-D. Niemeier (eds.), Politeia:Society and State in the Aegean Bronze Age, 1.Aegaeum 12: 11-21. Liege: University of Liege.

1996 The development of agriculture and pastoral-ism in Greece: when, how, who and what? InD.R. Harris (ed.), The Origins and Spread ofAgriculture and Pastoralism in Eurasia: 296-309.London: University College London Press.

Halstead, P. and G. Jones1989 Agrarian ecology in the Greek islands: time

stress, scale and risk. JHS 109: 41-55.Hill, J. (ed.)

1977 Explanation of Prehistoric Change. Albuquerque:University of New Mexico Press.

Jahns, S.1990 Preliminary note on human influence and the

history of vegetation in southern Dalmatia andsouthern Greece. In S. Bottema, G. Entjes-Nieborg and W. van Zeist (eds.), Man s Role inthe Shaping of the Eastern MediterraneanLandscape: 333-40. Rotterdam: Balkema.

1993 On the Holocene vegetation history of theArgive plain (Peloponnese, southern Greece).Vegetation History and Archaeobotany 2:187-203.

James, P., M. Atherton, A. Harvey, A. Firmin and A.Morrow

1997 The physical environment of Methana. In C.Mee and H. Forbes (eds.), A Rough and RockyPlace: the Landscape and Settlement History of theMethana Peninsula, Greece: 5-32. Liverpool:Liverpool University Press.

James, P., C. Mee and G. Taylor1994 Soil erosion and the archaeological landscape of

Methana, Greece. Journal of Field Archaeology 21:395-416.

Jameson, M., C. Runnels and T. van Andel1994 A Greek Countryside: the Southern Argolid from

Prehistory to the Present Day. Stanford: StanfordUniversity Press.

Kardulias, PN.1992 The ecology of Bronze Age flaked stone tool

production in southern Greece: evidence fromAgios Stephanos and the southern Argolid. A]A96: 421-42.

Konsola, D.1984 / Proimi Astikopoiisi stous Protoelladikous

Oikismous. Athens.

Laxton, R. and W. Cavanagh1995 The rank-size dimension and the history of site

structure from survey data. Journal of Quanti-tative Anthropology 5: 327-58.

Limbrey, S.1990 Soil studies at Akrotiri. In D. Hardy (ed.), Them

and the Aegean World 3, Volume 2: 377-82.London: The Thera Foundation.

McDonald, W.A. and G. Rapp Jr. (eds.)1972 The Minnesota Messenia Expedition. Reconstruct-

ing a Bronze Age Regional Environment. Minnea-polis: University of Minnesota Press.

Manning, S.1997 Cultural change in the Aegean c. 2200 BC. In H.

Nuzhet Dalfes, G. Kukla and H. Weiss (eds.),Third Millennium BC Climate Change and OldWorld Collapse-. 149-71. Berlin: Springer-Verlag.

Marangou, L.1994 Nees martiries gia ton Kikladiko politismo stin

Amorgo. In Phigos: Timitikos Tomos gia ton SotiriDakari: 467-88. loannina: University of loannina.

Marthari, M.1997 Apo ton Skarko stin Poliokhni: paratirisis gia tin

koinoniko-oikonomiki anaptixi ton oikismon tisProimis Epokhis tou Khalkou stis Kiklades kaita nisia tou Vorioanatolikou Aigaiou. In C.Doumas and V. La Rosa (eds.), / Poliokhni kai iProimi Epokhi tou Khalkou sto Vorio Aigaio: 362-82.Athens: University of Athens and Italian Schoolof Archaeology.

1998 Syros, Chalandriani, Kastri. From the Investigationand Protection to the Presentation of an Archaeol-ogical Site. Athens: Ministry of the Aegean.

Mee, C. and H. Forbes (eds.)1997 A Rough and Rocky Place: the Landscape and

Settlement History of the Methana Peninsula,Greece. Liverpool: Liverpool University Press.

Moody, J., O. Rackham and G. Rapp, Jr1996 Environmental archaeology of prehistoric NW

Crete. Journal of Field Archaeology 23: 273-97.Muller, S.

1996 Prospection archeologique de la plaine deMalia. BCH120: 921-28.

Paepe, R. and D. Deraymaeker1973 Geomorphological and Quaternary mapping of the

Adami-Potami area (S.-E. Attica). Thorikos 1969:79-99. Brussels: Comite de Fouilles Beiges enGrece.

Palaima, T.1989 Perspectives on the Pylos oxen tablets: textual

(and archaeological) evidence for the use andmanagement of oxen in Late Bronze AgeMessenia (and Crete). In T. Palaima, C. Shel-

160 Todd Whitelaw

merdine and P. Ilievski (eds.), Studia Mycenaea(1988). Ziva Antika Monograph 7. Skopje: 85-124.

1992 The Knossos oxen dossier: the use of oxen inMycenaean Crete, part 1: general backgroundand scribe 107. In J.-P. Olivier (ed.), Mykena'ika.Actes du 9e Colloque international sur les textesmyceniens et egeens. BCH Supplement 25: 463-74.Paris: Boccard.

Papagiannopoulou, A.1991 The Influence of Middle Minoan Pottery on the

Cyeludes. Studies in Mediterranean Archaeologyand Literature, Pocketbook 96. Jonsered: PaulAstrom.

Parsons, M. and J. Gifford1995 Soil and land use studies at Kommos. In J.W.

Shaw and M.C. Shaw (eds.), Kommos I: theKommos Region and Houses of the Minoan Town:292-324. Princeton: Princeton University Press.

Pope, K. and T. van Andel1984 Late Quaternary alluviation and soil formation

in the Southern Argolid: its history, causes, andarchaeological implications. JAS 11: 281-306.

Pullen, D.1992 Ox and plow in the Early Bronze Age Aegean.

A]A 96: 45-54.Rackham, O.

1982 Land-use and the native vegetation of Greece.In M. Bell and S. Limbrey (eds.), ArchaeologicalAspects of Woodland Ecology. BAR InternationalSeries 146: 177-98. Oxford: British Archaeo-logical Reports.


Renault-Miskovsky, J.1981 Analyse pollinique des sediments de la grotte

de Kitsos (Lavrion, Grece). In N. Lambert (ed.),La grotte prehistorique de Kitsos (Attique). Missions1968-1978, Volume 1: 633-55. Paris: EcoleFranchise d'Athenes.

1983 Les connaissances actuelles sur les vegetationset les climats quaternaires en Grece, d'apres lesdonnees de 1'analyse pollinique. In G.Rougemont (ed.), Les Cyclades: 99-109. Paris:CNRS.

Renfrew, A.C.1972a The Emergence of Civilisation. The Cyclades and the

Aegean in the Third Millennium BC London:Methuen.

1972b Patterns of population growth in the prehistoricAegean. In P. Ucko, R. Tringham, and G.Dimbleby (eds.), Man, Settlement and Urbanism:383-400. London: Duckworth.

1982 Polity and power: interaction, intensificationand exploitation. In A.C. Renfrew and J.M.Wagstaff (eds.), An Island Polity. The Archaeologyof Exploitation in Melos: 264-90. Cambridge:Cambridge University Press.

Renfrew, A.C. and J.M. Wagstaff (eds.)1982 An Island Polity. The Archaeology of Exploitation in

Melos. Cambridge: Cambridge University Press.Runnels, C.

1995 Environmental degradation in ancient Greece.Scientific American 272(3): 72-75.

Runnels, C. and T. van Andel1987 The evolution of settlement in the Southern

Argolid, Greece: an economic interpretation.Hesperia 56: 303-34.

Rutter, J.1983 Some observations on the Cyclades in the later

third and early second millennium. AJA 87: 69-76.

1984 The 'Early Cycladic III Gap': what it is and howto go about filling it without making it go away.In J.A. MacGillivray and R. Barber (eds.), ThePrehistoric Cyclades: Contributions to a Workshopon Cycladic Chronology: 95-107. Edinburgh:Department of Classical Archaeology, Univer-sity of Edinburgh.

1993 Review of Aegean prehistory 2: the prepalatialBronze Age of the southern and central Greekmainland. AJA 97: 745-97.

Rutter, J. and C. Zerner1983 Early Hellado-Minoan Contacts. In R. Hagg and

N. Marinates (eds.), The Minoan Thalassocracy:Myth and Reality: 75-83. Goteborg: Paul Astrom.

Shelmerdine, C.1997 Review of Aegean prehistory 6: the palatial

Bronze Age of the southern and central Greekmainland. AJA 101: 537-85.

Sotirakopoulou, P.1993 The chronology of the 'Kastri group' reconsid-

ered. BSA 88: 5-20.Turner, J. and J. Greig

1975 Some pollen diagrams from Greece. Review ofPalaeobotany and Palynology 20:171-204.

Vasilakis, A.1989-90 Proistorikes thesis sti Moni Odigitrias: Kalous

Limenes. Kritiki Estia 3:11-79.Vita-Finzi, C.

1969 The Mediterranean Valleys. Cambridge: Cam-bridge University Press.

Wagstaff, J.M.1981 Buried assumptions: some problems in the

interpretation of the 'Younger Fill' raised byrecent data from Greece. JAS 8: 247-64.


Wagstaff, J.M., S. Augustson and C. Gamble1982 Alternative subsistence strategies. In A.C.

Renfrew and J.M. Wagstaff (eds.), An IslandPolity: The Archaeology of Exploitation in Melos:172-80. Cambridge: Cambridge UniversityPress.

Wagstaff, J.M. and J.F. Cherry1982a Settlement and population change. In A.C.

Renfrew and J.M. Wagstaff (eds.), An IslandPolity. The Archaeology of Exploitation in Melos:136-55. Cambridge: Cambridge UniversityPress.

1982b Settlement and resources. In A.C. Renfrew andJ.M. Wagstaff (eds.), An Island Polity. The Arch-aeology of Exploitation in Melos: 246-63.Cambridge: Cambridge University Press.

Wagstaff, J.M. and C. Gamble1982 Island resources and their limitations. In A.C.

Renfrew and J.M. Wagstaff (eds.), An IslandPolity: The Archaeology of Exploitation in Melos:95-105. Cambridge: Cambridge UniversityPress.

Warren, P. and I. Tzedhakis1974 Debla, an Early Minoan settlement in western

Crete. BSA 69: 299-342.Watrous, L.V.

1982 Lasithi: a History of a Settlement on a HighlandPlain in Crete. Hesperia Supplement 18.Princeton: American School of Classical Studiesat Athens.

1994 Review of Aegean prehistory 3: Crete from ear-liest prehistory through the Protopalatial peri-od. AJA 98: 695-755.

Watrous, L.V, D. Hatzi-Vallianou, K. Pope, et al.1993 A survey of the western Mesara plain in Crete:

preliminary report on the 1984, 1986 and 1987field seasons. Hesperia 62: 191-248.

Wells, B. and C. Runnels (eds.)1996 The Berbati-Limnes Archaeological Survey 1988-

1990. Skrifter utgivna av Svenska Institutet iAthen 44. Stockholm: Paul Astrom.

Wells, B., C. Runnels and E. Zangger1990 The Berbati-Limnes archaeological survey: the

1988 season. Opuscula Atheniensia 18: 207-38.Whitehead, D. and M. Sheehan

1981 The late-postglacial vegetational history of theArgolid peninsula, Greece. National GeographicSociety Research Reports 13: 693-708.

Whitelaw, T.1983 The settlement at Fournou Korifi, Myrtos, and

aspects of Early Minoan social organisation. InO. Kryzszkowska and L. Nixon (eds.), MinoanSociety: 323-45. Bristol: Bristol Classical Press.

1998 Colonisation and competition in the polis ofKoressos: the development of settlement innorth-west Keos from the Archaic to the LateRoman periods. In L. Mendoni and A.Mazarakis Ainian (eds.), Kea - Kythnos: Historyand Archaeology. Meletimata 27: 227-57. Athens:National Hellenic Research Foundation.

1999 Alternative pathways to complexity in theprepalatial southern Aegean. Seminar onArchaeology and Anthropology, Athens.

Wiencke, M.H.1989 Change in Early Helladic II. AJA 93: 495-509.

Wright, H.E.1972 Vegetation history. In W McDonald and G.

Rapp (eds.), The Minnesota Messenia Expedition.Reconstructing a Bronze Age Regional Environ-ment: 188-99. Minneapolis: University ofMinnesota Press.

Zangger, E.1991a Tiryns Unterstadt. In E. Pernicka and G.A

Wagner (eds.), Archaeometry '90. Basel: Birk-haiiser Verlag.

1991b Prehistoric coastal environments in Greece: thevanished landscapes of Dimini bay and LakeLerna. Journal of Field Archaeology 18:1-15.

1992 Neolithic to present soil erosion in Greece. In M.Bell and J. Boardman (eds.), Past and Present SoilErosion: Archaeological and Geographical Perspec-tives: 133-48. Oxbow Monograph 22. Oxford:Oxbow Books.

1993 The Geoarchaeology of the Argolid. Argolis 2.Berlin: Mann.

1994a Landscape changes around Tiryns during theBronze Age. AJA 98:189-212.

1994b The island of Asine: a palaeogeographic recon-struction. Opuscula Atheniensia 20: 221-39.

Zangger, E., M. Timpson, S. Yazvenko, F. Kuhnke and J.Knauss

1997 The Pylos Regional Archaeological Project, part2: landscape evolution and site preservation.Hesperia 66: 549-641.

11

Soils and Site Function: The Laconia Rural SitesProject

Christopher Mee and Peter James

Introduction

The focus of this paper is palaeoenvironmen-tal research at the level of the individual sitebut in the context of a project which has aregional perspective. We hope to show howsoil studies can help us to understand thefunction of the most common type of siteencountered by recent intensive surveys,namely small rural sites. The Laconia RuralSites Project (Cavanagh and Mee 1999) hasbeen set up specifically to investigate suchsites. It has evolved from the Laconia Surveywhich identified over 400 sites east of the RiverEurotas (Cavanagh et al. 1996a). Phosphateanalysis of soil samples from a number ofthese sites was undertaken (Buck et al. 1988;Cavanagh et al. 1996b).

Rural Sites

Intensive surveys have identified hundreds ofrural sites. Although more common in someperiods than others, they have been a featureof the Greek landscape at least since the end ofthe Neolithic. Characteristically they are5000 m2 or less in size. Often their period ofuse was relatively brief. Most consist of adense scatter of sherds and tile. There may bestone implements, more substantial items

such as press beds, or traces of walls. Theassumption is that these sites were isolatedfarmsteads (Cherry et al. 1991: 336-37; Jame-son et al. 1994: 249-50; Snodgrass 1990:125-26)but can we be certain that they were in factoccupied and were not simply storehouses orshelters? Were they occupied continuously oronly when the agricultural cycle made resi-dence in the countryside more convenient?Were they occupied by their owners, by ten-ants or by slaves (see inter alia Alcock 1993: 60-62; Alcock et al. 1994:163; Cavanagh 1991:113-14; Osborne 1992: 25; Snodgrass 1987:117-18)?

Texts and inscriptions provide some infor-mation about the function of rural sites in theClassical, Hellenistic and Roman periods,although most Greek writers were more inter-ested in the city than the countryside and theircomments can be ambiguous or even contra-dictory (Osborne 1985:119). Ethnographic evi-dence has also been utilized, in particular byWhitelaw (1991; 1994) who has undertaken adetailed analysis of recent rural settlementand land use on Keos, based on documentaryand statistical data, interviews and study ofthe field systems and structures. Nevertheless,we felt that it would be useful to investigatethe surface characteristics of rural sites as rig-orously as possible and thereby resolve, or atleast redefine, some of the questions abouttheir function.

Soils and Site Function: The Laconia Rural Sites Project 163

The Laconia Rural Sites Project

In the first phase of the project, we revisitedapproximately 60 of the sites identified by theLaconia Survey. These were single periodsites, in most cases less than 2500 m2. We thenselected 20 (Fig. 11.1) which seemed suitablefor intensive survey and analysis. The rangeof periods represented is: Early Helladic (2),Middle/Late Helladic (2), Archaic (1), Clas-sical (4), Hellenistic (3), Roman (6) and Byzan-tine (2). The sites were examined in 1993-94and there was a study season in 1995.

The procedure adopted on each of the siteswas as follows. A 5 m x 5 m grid of squareswas laid out across the whole of the site. Themembers of the survey team counted thenumber of artefacts and recorded surface visi-bility along transects which crossed the siteand its periphery. This enabled us to identifypeaks in artefact density and to decide howwe would sample the site. Then every artefactwas collected from each of the squares. Tilewas sorted by type of decoration and thick-ness, the different categories were weighedand recorded, and the tile was then discardedon site. Sherds, stone artefacts and other findswere taken away to be cleaned, marked,drawn and recorded on our database. Thisenables us to generate density maps for thedifferent types of artefact on each site, so thatwe can see whether there is a correlation in thedistribution of sherd and tile for instance.

In order to identify possible sub-surface fea-tures, Neil Brodie undertook a geophysicalsurvey of every site. A gradiometer was usedto detect local variations in the magnetic field.Readings were taken at 1 m intervals in aseries of 20 m x 20 m grids. Soil resistivity wasmeasured on a resistance meter with a twinprobe array. The high temperatures did affectthe instruments and the fact that the soil wasso dry impeded the use of the resistivity meterin our second season. However, we took

advantage of the much wetter conditions inthe spring of 1995 to complete the programmeof geophysical prospection. The results arestill being processed but features have beenidentified on over half of the sites.

The third element in the analysis of each sitehas involved detailed soil studies. The aimhas been to assess the likely impact of past soilerosion and deposition, and soils have beensampled for analysis of selected chemical andmineral magnetic properties. The purpose ofthe analyses is to determine any spatial rela-tionships between soil properties and arte-facts, and to consider whether the present soilproperties may reflect past human activitiesassociated with each site, such as the discardof organic waste, the use of a hearth/kiln, ormetallurgy Residues can accumulate in thesoil from a great range of organic and inor-ganic materials in food, textiles, householdutensils, building material such as mudbrick,agricultural and industrial implements, pro-cesses such as tanning, as well as the faeces,urine and bodies of humans and animals.Manure, wood ash and other waste productswere used to improve soils for intensive cropgrowth. Many soils, particularly base-richsoils in non-humid environments, are capableof retaining a portion of the chemical elementsderived from these materials in a relativelyimmobile form. The smelting and working ofmetals are also likely to leave marked chemi-cal signatures. Soil mineral magnetic proper-ties are altered when oxidation/reductionprocesses caused by intense burning trans-form iron oxides, for example around hearthswhere magnetic susceptibility tends to beenhanced.

The elements chosen for analysis were phos-phorus (P), lead (Pb), copper (Cu) and zinc(Zn). These, and a number of other elements,have been shown to occur in archaeologicalsoils in both the Old and New Worlds at con-centrations above those off-site (Griffith 1980;

Figure 11.1 The location of the 20 sites selected for the Laconia Rural Sites Project.


1981; Konrad et al 1983; Buck et al 1988;Moore and Denton 1988; Bintliff et al 1990;Lewis et al 1993; Aston et al 1998). Total nick-el (Ni) and manganese (Mn) were determinedfor more than half of the sites. Nickel wasselected as an element which is not associatedwith pre-industrial activity, manganese as anelement with a distribution expected to bedominated by its abundance in soil minerals,though both elements are excreted by humans(Snyder et al 1975). In the case of site LP1,which is discussed in more detail below, totaliron (Fe), calcium (Ca), magnesium (Mg),potassium (K), organic carbon (C), soil parti-cle size distribution, percentage CaCO3 andsecondary iron oxides were also analysed. Inaddition to analysis of mineral magnetic prop-erties in the laboratory (following Thompsonand Oldfield 1986), volume susceptibility wasmeasured along selected transects in the fieldby F-probe at 20 cm depth and on the groundsurface.

Soil samples were taken at a depth of 20 cmfrom the centre of each grid square or from theboundaries between grid lines. A note wasmade of the soil texture and colour, consisten-cy, stone and root content, and the number ofartefacts in the topsoil was counted. In each ofthe grid squares sampled, the gradient andaspect of slope, surface soil characteristics andvegetation were also recorded. In addition,off-site samples were taken, although in somecases the soil conditions were markedly dif-ferent from those on the site.

In the laboratory the samples were thor-oughly air-dried, once pH had been deter-mined. The analyses were undertaken on<2 mm material, subsamples for the differentanalyses being sieved through brass, stainlesssteel or nylon sieves. The total amount of eachelement was determined by nitric and per-chloric acid digestion and atomic absorptionspectrophotometry. In addition to total iron(Fet), dithionite extractable Fe (Fed)(Mehra and

Jackson 1960)—chiefly that in fine crystallineiron oxides formed within the soil—was mea-sured for LP1 because, together with Fe in pri-mary minerals inherited from the soil parentmaterial, it is a major determinant of mineralmagnetic properties. Iron oxyhydroxides alsoadsorb heavy metals. Secondary amorphousor paracrystalline Fe, extracted with ammoni-um oxalate (Fe0)(McKeague and Day 1966),was also determined for LP1 as, in relation toother fractions of Fe, it is used to interpret thepathway and extent of soil development.Organic matter in the soil, measured as organ-ic carbon (Walkley and Black 1934), may carrysignificant amounts of all the elements deter-mined and may itself be a residue from mate-rial deposited when the site was in use.Secondary Fe and magnetic susceptibility (Xlf)tend to be associated mainly with the clayfraction. The three size fractions, sand(63-2000 mm), silt (2-63 mm) and clay(<2 mm), and the full particle-size curve forthe <2 mm soil were determined.

Spatial relationships between the soil attrib-utes and between these and the ground-sur-face artefact weights have been examined firstby subjective comparison of plots which showthe distribution of each variable (simplifiedinto five classes of values) across the grid, andsecondly by examining (again subjectively)data for selected grid cells. In addition datafrom some of the sites have been examined bycorrelation, principal components, and clusteranalysis. Particular attention has been paid tovariables which appear to be related at somespatial scale, as it is found that such relation-ships are of interest, whether or not there isevidence of clear on-site/off-site differences insoil attributes.

It is through the integration of these differ-ent approaches that we hope to understandhow these rural sites functioned. As the analy-sis of the data is still in progress this is obvi-ously not the moment for definitive state-

166 Christopher Mee and Peter James

ments but we can offer some comments basedon the results from LP1, a Late Roman site,and more briefly from four further sites.

LP1

The site is in an olive grove on a terrace abovethe river Eurotas, approximately 1 km east ofSparta (Figs. 11.2-11.3). The west margin ofthe site is cut by the edge of the terrace, an 85°bluff which falls 6 m to the modern Eurotasflood plain. The south-east margin of the sitehas been eroded by a 3.5 m deep gully. Nowstabilized, the gully forms part of one of thelarge gully systems which drain the Neogenegravels on the east side of the Eurotas valleyErosion of the east side of the site is attestedby the exposure of a Roman tile grave in thegully wall to the south of grid square Ell. It islikely, however, that the gully has followedapproximately its present course at least sinceLate Roman times. The surface of the river ter-race in the area of the grid has also remainedstable. Because of the gentle slope (<2° to the

south) only the finer fractions of the soilwould be transported by overland flow andthe terrace is protected by the gully fromdeposition of sediment eroded off the hill-slope to the east. The southern limit of the site,which has not been located, lies in the densevegetation of an irrigated Citrus grove. To thenorth, site LP2 lies at a distance of 150 m.

We sampled 99 squares in toto and so thiswas one of the largest sites surveyed (Fig. 11.4).Artefact density peaks at three points (Fig.11.7). Square A10 stands out in a cluster whichalso includes A9, B9-10 and C9- 10. B2^t andC3 represent a second cluster, while the thirdoccupies squares E8-11. It is in these clustersthat the highest numbers of sherds were re-corded, and there is a close, although not exact,correlation between the pottery and tile distri-butions. It is possible that sherds, potentiallythe most mobile of the artefacts on the groundsurface, have been displaced from where theywere originally discarded. The roof tile wasunpainted and weighed 181.85 kg. There werealso 47.85 kg of brick and floor tile, concen-trated in squares B2, BIO, C3, CIO and E8.

Figure 11.2 Location of LP1.


Figure 11.3 LP1: view of the site from the north-east.

Figure 11.4 LP1: site plan.


A high proportion of the pottery from LP1,possibly as much as 90%, is closed. There werepithos sherds, especially in the squares definedas clusters 1 and 3, a range of amphorae, alsojars, jugs and a lid. Open shapes includebowls, basins and dishes. A number of sherdspreserve traces of black glaze decoration anda Hellenistic date is likely but not certain.Most of the pottery should be Roman or LateRoman and there were also nine mediaevalsherds. A bronze coin, which was in squareD9, has been identified as a follis of theemperor Maurice Tiberius which was struckin AD 590/591.

Geophysical prospection on LP1 indicated alarge magnetic anomaly in squares A8 and A9,which continues into square A10 and coversan area of about 50 m2 (Fig. 11.5). The maxi-mum intensity recorded was in excess of100 nT and suggests the presence of a firedstructure, such as an oven or kiln. There is alinear magnetic anomaly traversing the sitefrom square H2 to square D10, which is alsovisible as a diffuse, high resistance feature(Fig. 11.6). This may mark the path of an oldwall or track.

The soil studies revealed that the artefacts inthe topsoil were concentrated in the sameareas of the site, the south-west and south-eastcorners of the grid, as those on the surface(Fig. 11.7). This suggestion of the 'reliability7

of the ground surface artefacts as an archaeo-logical indicator is supported by the geomor-phological stability of the river terrace uponwhich LP1 is situated. Associated with thehigh artefact concentrations in the southernpart of the grid are relatively high magneticsusceptibility (Ms) values, total phosphorus(P), lead (Pb), copper (Cu), zinc (Zn), potassi-um (K), organic carbon and clay content, lowpH and darker soil colour (Figs. 11.7-11.8).The spatial association is particularly markedfor phosphorus, copper and magnetic suscep-tibility which highlight the two centres

defined by the artefacts. The values for leadare high in relation to other sites. The maxi-mum for copper, 259.2 mg kg'1 in D9, is one ofthe highest recorded in the survey andexceeds the concentration recommended assafe for food production (Baker 1990). Cal-cium concentration is high in the area of hightile and sherd weights, but an extreme valueof 14,555 mg kg1 occurs at I/J16. The organiccarbon and the dark colour of the soil maywell have survived from the use of manure inintensive food production when the site wasinhabited. Organic carbon can reside in thesoil for many thousands of years, particularlyin relatively dry climates and where organicmatter is stabilized by interaction with clayminerals (Huang and Schnitzer 1986; Wangand Amundsen 1996). Dark soil colours, suchas those in the south of the grid, may formunder intensive cultivation and persist forlong periods after cultivation ceases (Smith1980). Illuviation of humus, clay and siltresults in darkening of subsurface horizonsduring long-term cultivation in Mediter-ranean environments (Soil Survey Staff 1997:10) and may have occurred in the soil at LP1.Relatively high values for clay content occurthroughout the grid, but the highest are inrows B, C and D. The clay might have beenbrought in when floors were being laid orcould derive from mudbricks. The possibilitythat the soil in square E6 has been modified byintense heat is suggested by a 'hard7 mineralmagnetic remanence (frequency dependentand low frequency susceptibility are low),characteristic of haematite (Oldfield andRichardson 1990). One of the processes bywhich haematite forms within soil is byintense heating. The 'hard7 remanence insquare E6 is appreciably higher than else-where in the grid.

It would seem that there were two, or possi-bly three, structures on LP1, which were cer-tainly in use in the Roman/Late Roman period.


Figure 11.5 LP1: gradiometer plot.

H -

Figure 11.6 LP1: resistivity plot.

Figure 11.7 LPI: tile weight, sherd weight, topsoil artefacts, magnetic susceptibility, phosphorus, lead, copper and zinc.


Figure 11.8 LPI: nickel, manganese, calcium, magnesium, potassium, pH, organic carbon, secondary crystallineiron, clay and soil colour. For each soil colour the Munsell value and chroma are given.


The lack of domestic pottery and the highproportion of storage vessels suggest that thiswas not primarily a residential site but a com-plex of storehouses in which agricultural pro-duce was kept and processed. The relativelyhigh organic carbon, phosphorus and otherelement concentrations indicate that the landaround these storehouses was intensively cul-tivated and there may have been middens.The magnetic anomaly could reflect associat-ed 'industrial' activity.

Other Sites

A brief account of the results from four moresites will illustrate the way in which the soilanalyses both complement and extend theinferences drawn from the artefact data. LP3is a Hellenistic site of the third or second cen-tury BC. Most of the artefacts—pottery and tile—were concentrated in four squares in thecentre of the site. Our impression is that therewas a single structure, most likely a farm-stead, which was occupied at least on a sea-sonal if not a permanent basis. There werehigh values for magnetic susceptibility, cop-per, lead, zinc and phosphorus in the squaresin which the artefact density was high, andadditionally for nickel, manganese and cobaltin squares adjacent to these. The peaks inthese variables are consistent with the use offire and of a range of materials at the site cen-tre. The maximum for phosphorus (50.1 mgkg'1) is high for the LRSP soils and could resultfrom a midden located near the site centre. Atthe southern end of the site, high phosphorus,lead, manganese, copper and zinc suggest asecond focus of more restricted activity. Thesituation on LP13, a Roman site on a singleagricultural terrace, is rather similar. An areaof five squares in which the pottery and tilewere clustered also yielded high values formagnetic susceptibility, organic carbon, phos-

phorus and lead. However, the squares in thenorth of the grid were the most distinctive interms of their soil chemistry, with high valuesfor trace metals, magnetic susceptibility andorganic carbon. The soil data for LP13 indicatea range of activities and support the conclu-sion, based on the presence of domestic pot-tery, that this was a residential site.

Multivariate analyses of the data from LP1(see also James 1999) and LP13 usefully high-light differences between the two sites.Selected results from principal componentsanalysis are shown in Table 11.1. For both sitesthere is a 'mineral' component with naturallyinter-related soil variables showing relativelyhigh loadings (>4.0), and an 'archaeological'component in which artefacts, organic carbon,phosphorus and lead are among those vari-ables with higher loadings. The 'archaeologi-cal' component is the first component forLP13 and the second for LP1. There is a sig-nificant difference in the role of copperbetween the sites: in LP1 it has the highestloading on the 'archaeological' component,but the lowest on this component in LP13. Inthe latter, manganese has a loading of 0.57 andzinc plays a greater part than in LP1, with aloading of 0.36. The results of the principalcomponents analyses suggest a stronger—orsimpler—archaeological signal in the soil ofLP13 and differences in materials usedbetween the two sites, with copper beingimportant in LP1.

Two Mycenaean sites provide an insightinto the variable effects of post-depositionalprocesses. LP10 occupies the summit of aridge that slopes up to the north and south.Most of the artefacts were at the north andsouth ends of the site rather than in the centre.The mineral magnetic, lead, copper and zincvalues have a similar distribution. Phos-phorus has particularly high values in twosquares in the south (73.3 mg kg: and 98.3 mgkg1). It seems likely that soil has been eroded


Table 11.1 Principal components analysis (unrotated) of data from LPI and LPI 3. Selected variable loadings on thefirst two components. Soil art = number of artefacts recorded in topsoil. Percentage figures are for vari-ance. For LPI, extreme values for Mg and Ca at location I/J16 are excluded. For LP13, Xlf, pH, particle-sizedata and topsoil artefacts are not included.

downslope from the squares in the south ofthe grid. The removal of the brown topsoil bya combination of cultivation and overlandflow has left the sherds exposed there andburied the artefacts in the centre of the site.Where topsoil has been removed the soil isredder but still carries the magnetic andchemical traces of activities associated withthe site. On LP20, the settlement associatedwith the Mycenaean chamber tomb cemeteryat Melathria (Cavanagh and Crouwel 1992),there is a derelict house. The sherd and tile onthe south side of the site evidently derive fromthis. Most of the Mycenaean pottery was con-centrated on the east side of the grid wherethe mineral magnetic and phosphorus levelswere also high.

Conclusions

We thought at first that analysis of totalamounts of chemical elements in the soilwould be too blunt an instrument for definingpedochemical signals of former activity at thesites, but it has proved not to be. The chemicalpatterns derived correspond quite closelywith those both of artefact density and of arange of magnetic, mechanical and physicalproperties on 18 of the 20 sites examined.There occur clear associations between arte-fact distribution and organic carbon, certainmineral magnetic properties, soil colour andtotal phosphorus, copper, zinc, lead, calciumand potassium. It is encouraging that thesetraces of activity have survived in situ, togeth-er with the artefacts, for a period of up to 5000years, commonly in sloping terrain at consid-

ZnMgSandClayxlfKNiCaFed

MnPbPPHSoil artCuOrgC

0.900.83-0.820.820.810.780.770.760.730.720.660.590.490.310.190.16

CuOrgCPSoil artPbpHTile wtSherd wtxlf

0.740.650.560.510.44-0.430.410.330.28

OrgCPbTile wtPMnSherd wtZnCaC03

Cu

0.770.750.700.670.570.540.36-0.610.16

Zn 0.79Ni 0.78Cu 0.70CaC03 0.46Mn 0.42

Component 1(Mineral)

34.4%

LP1

Component 2(Archaeological)

15.6%

LP13

Component 1(Archaeological)

31.8%

Component 2(Mineral)

23.9%


arable potential erosion risk. One inevitableconclusion is that actual erosion has been lim-ited throughout this period.

Acknowledgments

The Laconia Rural Sites Project is directed byWilliam Cavanagh and Christopher Mee. Thepermits for the project were issued by theGreek Ministry of Culture and we are mostgrateful to the former Director of Prehistoricand Classical Antiquities, Dr Y. Tzedakis, toDr T. Spyropoulos, the Ephor of Prehistoricand Classical Antiquities for Arcadia andLaconia, and to his colleagues in Sparta, inparticular S. Raftopoulou. We wish to thankthe director and staff of the British School atAthens for their advice and assistance. IreneCooper, Alan Henderson, Hilda Hull and BobJude of the Geography Department at theUniversity of Liverpool assisted with the lab-oratory analyses. Generous grants in supportof the project have been made by the BritishSchool at Athens, the British Academy, theUniversity of Liverpool and the University ofNottingham.

Bibliography

Alcock, S.E.1993 Graecia Capta: The Landscapes of Roman Greece.

Cambridge: Cambridge University Press.Alcock, S.E., J.F. Cherry and J.L. Davis

1994 Intensive survey, agricultural practice and theclassical landscape of Greece. In I. Morris (ed.),Classical Greece: Ancient Histories and ModernArchaeologies: 137-70. Cambridge: CambridgeUniversity Press.

Aston, M.A., M.H. Martin and A.W. Jackson1998 The use of heavy metal soil analysis for archae-

ological surveying. Chemosphere 37: 465-77.Baker, D.E.

1990 Copper. In BJ. Alloway (ed.), Heavy Metals inSoils: 151-96. London: Blackie and Son.

Bintliff, J.L., B.E. Davies, C.F. Gaffney and A. Waters1990 Trace accumulation in soils on and around

ancient settlements in Greece. In S. Bottema, G.Entjes-Nieborg and W. van Zeist (eds.), Man'sRole in the Shaping of the Eastern MediterraneanLandscape: 159-72. Rotterdam: Balkema.

Buck, C.E., W.G. Cavanagh and C.D. Litton1988 The spatial analysis of site phosphate data. In

S.P.Q. Rahtz (ed.), Computer and QuantitativeMethods in Archaeology 1988. BAR InternationalSeries 446, ii: 151-60. Oxford: BritishArchaeological Reports.

Cavanagh, W.G.1991 Surveys, cities and synoecism. In J. Rich and A.

Wallace-Hadrill (eds.), City and Country in theAncient World: 97-118. London: Routledge.

Cavanagh, W.G. and J.H. Crouwel1992 Melathria: a small rural settlement in Laconia.

In J.M. Sanders (ed.), FILOLAKON: LakonianStudies in Honour of Hector Catling: 77-86.London: British School at Athens.

Cavanagh, W.G., J. Crouwel, R.W.V. Catling and G.Shipley

1996a Continuity and Change in a Greek Rural Landscape:The Laconia Survey ii. British School at AthensSupplementary Volume 27. London: BritishSchool at Athens.

Cavanagh, W.G., R. Jones and A. Sarris1996b The phosphate and geophysical surveys. In

W.G. Cavanagh, J. Crouwel, R.W.V. Catling andG. Shipley, Continuity and Change in a GreekRural Landscape: The Laconia Survey ii. BritishSchool at Athens Supplementary Volume 27:261-325. London: British School at Athens.

Cavanagh, W.G. and C. Mee1999 Diversity in a Greek landscape: the Laconia

Survey and Rural Sites Project. In W.G.Cavanagh and S. Walker (eds.), Sparta in Laconia:the Archaeology of a City and its Countryside.British School at Athens Studies 4: 141-48.London: British School at Athens.

Cherry, J.E, J.L. Davis and E. Mantzourani1991 Landscape Archaeology as Long Term History:

Northern Keos in the Cycladic Islands from EarliestSettlement until Modern Times. MonumentaArchaeologica 16. Los Angeles: Institute ofArchaeology, University of California.

Griffith, M.A.1980 A pedological investigation of an archaeological

site in Ontario, Canada, 1: an examination of thesoils in and adjacent to a former village.Geoderma 24: 327-36.


1981 A pedological investigation of an archaeologicalsite in Ontario, Canada 2: use of chemical datato discriminate features of the Benson Site.Geoderma 25: 27-34.

Huang, P.M. and M. Schnitzer1986 Interactions of Soil Minerals with Natural Organics

and Microbes. Soil Science Society of AmericaSpecial Publication 17. Madison: Soil ScienceSociety of America.

James, PA.1999 Soil variability in the area of an archaeological

site near Sparta, Greece. JAS 26: 1273-88.Jameson, M.H., C.N. Runnels and T.H. van Andel

1994 A Greek Countryside: the Southern Argolid fromPrehistory to the Present Day. Stanford: StanfordUniversity Press.

Konrad, V.A., R. Bonnischen and V. Clay1983 Soil chemical identification of ten thousand

years of prehistoric human activity areas at theMunsungun lake thoroughfare, Maine. JAS 10:13-28.

Lewis, R.J., J.E. Foss, M.W. Morris, M.E. Timpson andC.A. Stiles

1993 Trace element analysis in pedo-archaeologystudies. In J.E. Foss, M.E. Timpson and M.W.Morris (eds.), Proceedings of the First InternationalConference in Pedo-Archaeology. Special Publica-tion 93-03, University of Tennessee AgriculturalExperiment Station: 81-88. Knoxville: Univer-sity of Tennessee.

McKeague, J.A. and J.H. Day1966 Dithionite and oxalate-extractable Fe and Al as

aids in differentiating various classes of soils.Canadian Journal of Soil Science 46:13-22.

Mehra, O.P. and M.L. Jackson1960 Iron oxide removal from soils and clays by a

dithionite-citrate-bicarbonate system bufferedwith sodium bicarbonate. Clay and Clay Minerals7: 317-27.

Moore, T.R. and D. Denton1988 The role of soils in the interpretation of archae-

ological sites in northern Quebec. In J. Bintliff,D.A. Davidson and B.C. Grant (eds.), ConceptualIssues in Environmental Archaeology: 25-37. Edin-burgh: Edinburgh University Press.

Oldfield, F. and N. Richardson1990 Lake sediments and atmospheric deposition.

Philosophical Transactions of the Royal Society ofLondon: Biological Series 327: 325-30.

Osborne, R.1985 Buildings and residence on the land in Classical

and Hellenistic Greece: the contribution of epig-raphy. BSA 80:119-28.

1992 Ts it a farm?7: the definition of agricultural sitesand settlements in ancient Greece. In B. Wells(ed.), Agriculture in Ancient Greece: 21-25.Stockholm: Paul Astroms Forlag.

Smith, N.1980 Anthrosols and human carrying capacity in

Amazonia. Annals of the Association of AmericanGeographers 70: 552-66.

Snodgrass, A.M.1987 An Archaeology of Greece: The Present State and

Future Scope of a Discipline. Berkeley: Universityof California Press.

1990 Survey archaeology and the rural landscape ofthe Greek city. In O. Murray and S. Price (eds.),The Greek City from Homer to Alexander: 113-36.Oxford: Clarendon Press.

Snyder, W.S., M.J. Cook, E.S. Nasset, L.R. Karhausen, G.Parry Howells and I.H. Tipton

1975 Report of the Task Group on Reference Man.International Commission on RadiologicalProtection 23. Oxford: Pergamon Press.

Soil Survey Staff1997 Keys to Soil Taxonomy (7th edition). US

Department of Agriculture, Soil ConservationService. Blacksburg: Pocahontas Press.

Thompson, R. and F. Oldfield1986 Environmental Magnetism. London: George Allen

and Unwin.Walkley, W.A. and LA. Black

1934 An examination of the Degtjareff method fordetermining soil organic matter and a proposedmodification of the chromic acid titrationmethod. Soil Science 37: 29-38.

Wang, Y. and R. Amundsen1996 Radiocarbon dating of soil organic matter.

Quaternary Research 45: 282-88.Whitelaw, T.M.

1991 The ethnoarchaeology of recent rural settlementand land use in northwest Keos. In J.F. Cherry,J.L. Davis and E. Mantzourani (eds.), LandscapeArchaeology as Long Term History: Northern Keosin the Cycladic Islands from Earliest Settlementuntil Modern Times. Monumenta Archaeologica16: 403-54. Los Angeles: Institute of Archaeol-ogy, University of California.

1994 An ethnoarchaeological study of rural land-usein north-west Keos: insights and implicationsfor the study of past Aegean landscapes. In PN.Doukelis and L.G. Mendoni (eds.), StructuresRurales et Societes Antiques: 163-86. Paris: BellesLettres.

landscape and land use in postglacial greece

Documents

departamento de prehistoria

nea trapezous alloformation

twelfth century ad

seventh century ad

hellenic botanical society

istron river floodplain

depicting stratigraphic relationships

dr paul halstead