The Origins of Transhumant Pastoralism in Temperate Southeastern Europe: A zooarchaeological perspective from the Central Balkans 9781841719702, 9781407329949

Table of contents :
Front Cover
Title Page
Copyright
BOOK JUSTIFICATION
ABSTRACT
CHAPTER SUMMARY
FOREWORD
ACKNOWLEDGEMENTS
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
LIST OF APPENDICES
CHAPTER 1: INTRODUCTION
CHAPTER 2: TRANSHUMANT PASTORALISM: SOME ETHNOGRAPHIC CONSIDERATIONS
CHAPTER 3: PREVIOUS RESEARCH ON ORIGINS OF TRANSHUMANT PASTORALISM IN PREHISTORIC SOUTHERN EUROPE
CHAPTER 4: REGIONAL ENVIRONMENT
CHAPTER 5: EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY
CHAPTER 6: METHODOLOGY
CHAPTER 7: DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES
CHAPTER 8: THE IDENTIFICATION OF TRANSHUMANT PASTORALISM THROUGH HARVEST PROFILE AND CEMENTUM ANALYSES
CHAPTER 9: CONCLUSIONS
REFERENCES CITED
APPENDIX A
APPENDIX B
APPENDIX C
APPENDIX D
APPENDIX E

Citation preview

BAR S1538 2006 ARNOLD & GREENFIELD

The Origins of Transhumant Pastoralism in Temperate Southeastern Europe A zooarchaeological perspective from the Central Balkans

THE ORIGINS OF TRANSHUMANT PASTORALISM

Elizabeth R. Arnold Haskel J. Greenfield

BAR International Series 1538 B A R

2006

The Origins of Transhumant Pastoralism in Temperate Southeastern Europe A zooarchaeological perspective from the Central Balkans

Elizabeth R. Arnold Haskel J. Greenfield

BAR International Series 1538 2006

ISBN 9781841719702 paperback ISBN 9781407329949 e-format DOI https://doi.org/10.30861/9781841719702 A catalogue record for this book is available from the British Library

BAR

PUBLISHING

BOOK JUSTIFICATION This book is designed as a test case for the identification of transhumant pastoralism using zooarchaeological techniques that have been proposed in the literature. Most studies that try to identify transhumance do not use zooarchaeological methods and hence are open to a variety of interpretations. By moving the study out of an arid environment and into a temperate environment, we remove the important variable of environment as a constraint. Most studies have tried to identify transhumance as deriving from the Early Neolithic (and beginning of animal domestication), where it is difficult to see differences because of similarities between hunter-gatherer and early pastoral behaviour. In our study, we try to get beyond these issues by working in an area without any expectation of transhumance associated with the earliest farming cultures and testing for its later appearance as part of a package of larger changes associated with the secondary products revolution. Little is still known about the evolution of this important and ancient form of land use and domestic animal management, especially in temperate environmental zones. Yet, zooarchaeological data can be used to answer basic questions concerning the origin and nature of early transhumant pastoralism. Such research has yielded data suggesting that transhumant pastoralism may have initially appeared in the central Balkans early in the Post Neolithic (Eneolithic – ca. 3300 B.C., calibrated radiocarbon dating). This is the earliest dates for the appearance of transhumant pastoralism in the temperate zones of Europe.

i

ABSTRACT This book addresses the issue of the temporal origins of transhumant pastoralism in temperate southeastern Europe (northern half of the Balkan Peninsula). In this region, there is little of the environmental imperative frequently cited to account for the origins of transhumance, in contrast to the Mediterranean littoral. The climate does not force the migration of animals from the lowlands in the summer into the highlands, and back into the lowlands because of insufficient graze and harsh temperatures. Yet, this form of land use and animal exploitation pattern has a very long history in southern Europe, extending at least to the Roman period. However, little is known about its origins and development. In recent years, several hypotheses have been suggested to explain when and why transhumant pastoralism with domestic animals appeared across the southern Mediterranean. Each hypothesis proposes a different point in time when transhumance would appear, ranging from the appearance of the earliest domestic animals (advent of the Early Neolithic), to the appearance of secondary product exploitation (advent of the Post Neolithic), and to the appearance of complex societies (advent of the Iron Age). Previous attempts to test these hypotheses has indicated that transhumance with domestic stock did not appear in the temperate zone until the advent of the Eneolithic (c. 3300 BC), long after the beginning of animal domestication in the Early Neolithic (c. 6100 BC). The hypotheses are tested by examining the tooth remains from three domestic animal taxa (Ovis/Capra, Bos taurus and Sus scrofa) from archaeological sites in the central part of the northern Balkans (also known as the Central Balkans). Data from eleven sites in the region, with statistically sufficient samples and spanning the period from the Early Neolithic through to the Early Iron Age, were tabulated to test the hypotheses. The primary technique involved the creation of harvest profiles from mandibular tooth wear and eruption data of domestic animal to examine age of death, the associated season of death and exploitation strategies. Season of death and the seasonality of culling practices were also examined for each taxon through additional graphical evaluation of the data and were supplemented by the secondary technique of cementum analysis of modern and archaeological mandibular Ovis aries and Capra hircus teeth. The specific hypothesis used in this investigation was that transhumant pastoralism would appear at the temporal point where complementary culling patterns between highland and lowland sites in the region appear. Based on other sources of data, such a pattern was expected to appear at the advent of the Post Neolithic. Several overriding taphonomic issues affecting sample size greatly hampered the creation of the traditional harvest profiles and limited the capability to evaluate the original hypotheses by these means. However, the harvest profile data does lend itself to provide further support for the secondary products revolution model, which is hypothesized to occur at the advent of the Post Neolithic (Eneolithic-Iron Age) in the region. The seasonality data, both graphically and from the cementum analysis, show complementary profiles between highland and lowland areas in the Post Neolithic, lending support to the original hypotheses of transhumant movement. At the same time, problematic divisions of age groups of Ovis/Capra are revealed as potentially masking herd movements that might otherwise be revealed in the traditional harvest profiles. It is clear from the analysis that strong changes in not only culling patterns, but also land use take place during the Eneolithic and Bronze Age of the region. These are interpreted in light of two regional developments – the advent of transhumant pastoralism and the beginning of the Secondary Products Revolution.

iii

CHAPTER SUMMARY Chapter 1 presents the theoretical background to the research problem and outlines the methodology, techniques and data that are utilized in the research. Chapter 2 provides a definition and discussion of transhumant pastoralism and relevant environmental and ecological parameters. Chapter 3 examines previous research on the origins of transhumant pastoralism in Europe focusing on both the Mediterranean and the northern temperate region of the Balkans. The research of Geddes, Halstead and Greenfield are the focus. Chapter 4 introduces the characteristics of the regional environment including topography, climate and vegetation in order to consider the viability of transhumant pastoralism within the northern Balkans. Chapter 5 summarizes the culture history of southeastern Europe from the Neolithic through the Early Iron Age. Aspects of settlement, fauna, and evidence for sedentism and mobility are examined. Chapter 6 describes the methodology to be utilized and details the two chosen techniques, tooth wear and eruption and cementum analysis. Chapter 7 describes the data examined in this investigation. In the chapter, each site is described, including site location, environment and nature of deposits. The mandibular and loose teeth remains are quantified by species and by major period. It also describes the thin sectioning data, both the modern comparative collection and the archaeological sample. Chapter 8 presents the results of the data analysis. The first part focuses on the tooth wear and eruption data, examining both production strategies and implications for the transhumant movement of herds. The second part focuses on the thin sectioning results. Chapter 9 presents the final conclusions regarding the data and discusses their implication in terms of the origins of transhumant pastoralism in the northern Balkans.

v

FOREWORD This work is an example of an idea that had a long gestation period and that has gone through almost 20 years of data collection. It was one of those tasks that required the collection of data from a long range of time periods and over a vast geographic scale. As a result, it has its origins in the PhD thesis research of Haskel Greenfield and it became complete with the MA thesis research of Elizabeth Arnold. In 1977, Greenfield began to work in Serbia under the supervision of H. Arthur Bankoff (Brooklyn College). This initial field season of survey and test excavations in the central Balkans led to their return and the beginning of systematic excavation at the Post Neolithic site of Novačka Ćuprija. The faunal data from this and other excavations by Bankoff were offered to Greenfield for analysis as part of his developing PhD thesis on Post Neolithic subsistence practices. In preparation for the identification of zooarchaeological specimens from the central Balkans, Greenfield received training in Budapest by the late Sandor Bökönyi. The experiences at Novačka Ćuprija eventually led to the analyses of many other collections in the region from both the Neolithic and Post Neolithic, many of which were reported in Greenfield (1986) and subsequent publications. Greenfield’s thesis and early publications were primarily concerned with testing Andrew Sherratt’s hypothesis on the advent of secondary products exploitation in Europe. Early on in the analysis of the specimens for the testing of the secondary products hypothesis, Greenfield noticed an apparent discrepancy between the patterns for highland and lowland Post Neolithic harvest profiles of ovicaprines and cattle, which he interpreted as evidence for the advent of transhumant pastoralism (1988, 1991, 1999a, 2001a). However, he felt that he only had sufficient data to point out the possible pattern rather than to comprehensively test this hypothesis. One gap in his early research was the paucity of highland (alpine or subalpine assemblages). In order to rectify this gap, he joined Blagoje Govedarica’s research project which was in the process of excavating Post Neolithic settlements on the Glasinac Plateau (east Bosnia). The project eventually focused on the site at Kadica Brdo which was excavated from 1986-1990. The excavations, as many others, were terminated by the Yugoslavian civil wars and have never resumed. Only preliminary reports of the excavations were ever published. However, sufficient fauna were recovered and analyzed to finally provide a test for transhumance from the highland zone for the EIA. One of the major problems that Greenfield encountered in his initial analyses was that he included all fragments in his ageing of specimens in order to maintain high sample sizes. The detailed tooth eruption and wear data were never separately analysed because of problems with the early computer recording of the information. In the 1990’s, with the appearance of easily accessible computer spreadsheets, Greenfield was able to recover all of his old data from the region and make them comparable for reanalysis. This included the tooth wear and eruption data. During the later years of Greenfield’s research in the region, new and exciting techniques in zooarchaeology were beginning to appear and spread throughout the discipline – i.e. tooth cementum analysis. Influenced by its potential for the study of transhumance, Greenfield began to collect what were then considered to be suitable specimens for such analysis. These were brought back to the University of Manitoba for eventual study, and there they lay until Elizabeth Arnold entered the picture. Arnold worked with Greenfield on a variety of data sets from the region, upgrading them and putting them all into compatible computer formats (Excel spreadsheets). From this database, the tooth wear and eruption data were extracted by Arnold for a more complete analysis. Some of the archaeological faunal assemblages described herein were stored at the University of Manitoba. These were sorted and all the teeth were separated for analysis by Arnold. Under the guidance of Dr. Ariane Burke, Arnold gained the knowledge and experience of thin sectioning and cementum analysis. A sample of the domestic Ovis/Capra material was selected for testing of the transhumance hypotheses using these techniques. Comparative modern Ovis/Capra material was collected on monthly trips to local abattoirs and small-scale farms over the course of a year and prepared in a similar manner. This work is a result of these combined efforts and author order does not imply relative effort in completing this monograph.

vii

ACKNOWLEDGEMENTS There are too many people to properly thank for access to data used in this work. They include all of Greenfield’s various collaborators through the years in ex-Yugoslavia. In particular, we would like to thank the directors (or codirectors) of the various projects or people who arranged access to the data from each of the sites used in the analyses presented within this book, including Florin Draşovean (Foeni-Salaş), Željko Jež (Petnica), Mirjana Vukmanović and Petar Popović (Livade), H. Arthur Bankoff (Novačka Ćuprija), Blagoje Govedarica (Kadica Brdo), Milenko Bogdanović (Ljuljaci), Mihalis Fotiadis (Megalo Nisi Galanis), Ljubomir Bukvić (Opovo) and the late Svetozar Stanković (Blagotin, Stragari-Šljivik), and the late Professors Milutin Garašanin and Dragoslav Srejović (Vinča-Belo Brdo). Other individuals also played vital roles in the data collection from the region, including Vesna Jeremenko, Tina Jongsma, Dimitrije Madas, Guilmine Eygun, Igor and Andrej Starović and the staff of the Istraživačka Stanica Petnica, Alexandr Radoman, Bojana Vojković, Zev Greenfield, and the late Vladimir Leković. Without the help of each and every one of the above, it would never have been possible to have accumulated such a wealth of comparative information from the region. Thanks to Clayton Robins (Manitoba Sheep Association) and Sharon Peddler (Manitoba Goat Association) for providing contacts and direction; to Monica Griffiths, for supplying all the modern goat specimens as well as valued information, and for continued interest; to Lee Perreault and the staff at Prairie Abattoir and Jim and Doris Holmes and staff from Carmen Meats; and Randy and Solange Eros for having the interest and taking the time to provide sheep specimens. Thanks must go to our colleagues and graduate students at the University of Manitoba who helped at various stages in the preparation of specimens, organization of data, and presentation of results. In particular, we would like to thank Val McKinley and Ariane Burke for Elizabeth Arnold’s training and their assistance in the University of Manitoba’s thin sectioning laboratory. Val’s support went far beyond simple technical advice. Special thanks must also be accorded to Tina Jongsma who aided in the field collection and analysis of many of the specimens described here and has allowed them to be incorporated into our analysis. In addition, we would like to Chris Meiklejohn and Karin Wittenberg for their involvement, patience and advice during the various revisions of this work and Dennis Murphy (Statistical Assistance Center, University of Manitoba) for his help with statistical issues. Figures 6.3, 8.4, 8.24, and 8.25 are reproduced with the permission of Sebastian Payne. Most of all, special thanks must be extended to our families for their constant support. None of this could have been accomplished without them. They continuously sacrificed in order to ensure that this work would see the light of day. This work is dedicated to the memory of Rita Fecher, Haskel’s mother, who passed away on June 13, 2003. This research could not have been undertaken without her inspiration for Haskel to persevere in the face of insurmountable odds during his studies and especially during the long periods of field work that were required to collect the body of data. She would have been proud to have seen its completion before her untimely passing.

ix

TABLE OF CONTENTS CHAPTER AND SECTION TITLES Book Justification Abstract Chapter Summary Foreword Acknowledgments Table of Contents I. Chapter and section titles II. List of Figures III. List of Tables IV. List of Appendices

i iii v vii xi xi xi xv xvii xix

CHAPTER 1: INTRODUCTION

1

I. Theoretical background to research problem II. Research hypotheses III. Method and Technique IV. Data A. The region B. Temporal span C. Data V. Conclusions

1 1 2 3 3 3 3 5

CHAPTER 2: TRANSHUMANT PASTORALISM: SOME ETHNOGRAPHIC CONSIDERATIONS I. Introduction to pastoralism II. Types of pastoralism A. Nomadic pastoralism B. Semi-nomadic pastoralism C. Transhumant pastoralism III. Aspects of modern transhumant pastoralism in the northern Balkans IV. Conclusion CHAPTER 3: PREVIOUS RESEARCH ON ORIGINS OF TRANSHUMANCE PASTORALISM IN PREHISTORIC SOUTHERN EUROPE I. Introduction II. Theories on the origins of transhumant pastoralism A. Political instability B. Regional symbiosis C. A by-product of initial animal domestication in an ecologically varied habitat D. Secondary Products Revolution III. Previous research on the origins of transhumant pastoralism in Mediterranean Europe A. Mesolithic/Neolithic continuity B. Complex societies IV. Previous research in temperate southeastern Europe VI. Differences between models – comparing apples and oranges VII. Conclusions

xi

7 7 7 7 7 8 8 11 13 13 13 13 13 14 14 14 15 15 16 17 18

CHAPTER 4: REGIONAL ENVIRONMENT

19

I. Introduction II. Landforms A. Topography B. Rivers III. Climate A. Temperature B. Precipitation IV. Environmental Regions A. Highland areas B. Lowland areas VI. Regional environment and climate in prehistory A. Environment and climate in the Balkans during the Neolithic B. Anthropogenic environmental changes during the Neolithic C. Environment and climate in the Balkans during the Post Neolithic D. Anthropogenic environmental changes during the Post Neolithic VII. Site locations in relation to major environmental regions VIII. Conclusion CHAPTER 5: EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY

19 19 19 20 20 20 21 22 22 22 23 23 23 23 24 24 24 25

I. Introduction II. Early Neolithic A. Chronology B. Cultures C. Settlement D. Fauna E. Evidence for sedentism/mobility III. Middle Neolithic A. Chronology B. Cultures C. Settlement D. Fauna E. Evidence for sedentism/mobility IV. Late Neolithic A. Chronology B. Cultures C. Settlement D. Fauna E. Evidence for sedentism/mobility V. Eneolithic (Chalcolithic) A. Chronology B. Cultures C. Settlement D. Fauna E. Evidence for sedentism/mobility VI. Early Bronze Age A. Chronology B. Cultures C. Settlement D. Fauna E. Evidence for sedentism/mobility VII. Middle and Late Bronze Age A. Chronology B. Cultures C. Settlement D. Fauna

25 25 25 25 25 26 26 27 27 27 27 27 27 28 28 28 28 28 28 29 29 29 29 29 29 30 30 30 30 30 31 31 31 31 31 31 xii

E. Evidence for sedentism/mobility VIII. Early Iron Age A. Chronology B. Cultures C. Settlement D. Fauna E. Evidence for sedentism/mobility IX. Conclusions

31 32 32 32 32 32 32 32

CHAPTER 6: METHODOLOGY

33

I. Introduction II. Tooth wear and eruption III. Establishing absolute age using tooth eruption and wear IV. Production strategies and construction of harvest profiles V. Cementum analysis VI. The control sample VII. The fossil sample VIII. Age determination with cementum analysis IX. Season of death determination with cementum analysis X. Conclusion CHAPTER 7: DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES I. Introduction II. Blagotin A. Site description B. Mandibular and loose tooth remains III. Foeni-Salaş A. Site description B. Mandibular and loose tooth remains IV. Kadica Brdo A. Site description B. Mandibular and loose tooth remains V. Livade A. Site description B. Mandibular and loose tooth remains VI. Ljuljaci A. Site description B. Mandibular and loose tooth remains VII. Megalo Nisi Galanis A. Site description B. Mandibular and loose tooth remains VIII. Novačka Ćuprija A. Site description B. Mandibular and loose tooth remains IX. Opovo A. Site description B. Mandibular and loose tooth remains X. Petnica A. Site description B. Mandibular and loose tooth remains XI. Selevac A.Site description B. Mandibular and loose tooth remains XII. Stragari-Šljivik A. Site description B. Mandibular and loose tooth remains

33 34 35 35 38 38 39 39 39 39 41 41 41 41 41 43 43 44 44 44 46 46 46 48 48 48 50 50 50 50 53 53 53 56 56 56 56 56 56 59 59 62 62 62 62

xiii

XII. Vinča A. Site description B. Mandibular and loose tooth remains XIII. The modern tooth wear and eruption control sample XIV. Cementum analysis data: archaeological and modern A. The modern control sample B. Archaeological cementum analysis sample XV. Conclusions CHAPTER 8: THE IDENTIFICATION OF TRASHUMANT PASTORALISM THROUGH HARVEST PROFILE AND CEMENTUM ANALYSES I. Introduction II. Construction of harvest profiles – first stage of analysis A. Sus scrofa dom. B. Ovis/Capra C. Bos taurus III. Regional scale of analysis – second stage of analysis A. Transhumant movement of pigs B. Transhumant movement of ovicaprines C. Transhumant movement of cattle IV. Comparison of tooth wear and eruption ageing methods – third stage of analysis V. Cementum analysis – fourth stage of analysis A. The modern comparative sample B. The archaeological sample VI. Conclusions CHAPTER 9 CONCLUSIONS I. II. III. IV. V.

62 62 64 64 64 65 65 65 71 71 71 71 80 92 102 103 103 104 104 116 116 116 117 119

Introduction Methodological concerns The Secondary Products Revolution Evidence for transhumant pastoralism Conclusions

119 119 119 120 122

REFERENCES CITED

123

xiv

LIST OF FIGURES 1.1

The Central Balkans and site locations

4

4.2

Map showing climatic divide between Mediterranean and temperate central Europe

21

6.3

Proportional allocation example

36

8.4 8.5 8.6 8.7 8.8 8.9 8.10 8.11 8.12 8.13 8.14 8.15 8.16 8.17 8.18 8.19 8.20 8.21 8.22 8.23 8.24 8.25 8.26 8.27 8.28 8.29 8.30 8.31 8.32 8.33 8.34 8.35 8.36 8.37 8.38 8.39 8.40 8.41 8.42 8.43 8.44 8.45 8.46 8.47 8.48 8.49 8.50 8.51 8.52 8.53

Payne’s production models Harvest profile (Sus scrofa) Middle Neolithic Petnica Harvest profile (Sus scrofa) Late Neolithic Petnica Harvest profile (Sus scrofa) Late Neolithic Vinča Harvest profile (Sus scrofa) Late Neolithic Opovo Harvest profile (Sus scrofa) Middle/Late Neolithic Selevac Harvest profile (Sus scrofa) Eneolithic Petnica Harvest profile (Sus scrofa) Early Bronze Age Novačka Ćuprija Harvest profile (Sus scrofa) Early/Middle Bronze Age Ljuljaci Harvest profile (Sus scrofa) Middle Bronze Age Vinča Harvest profile (Sus scrofa) Late Bronze Age Livade Harvest profile (Sus scrofa) Early Iron Age Kadica Brdo Harvest profile (Ovis/Capra) Early Neolithic Foeni-Salaş Harvest profile (Ovis/Capra) Early Neolithic Blagotin Harvest profile (Ovis/Capra) Early Neolithic Harvest profile (Ovis/Capra) Middle Neolithic Stragari Harvest profile (Ovis/Capra) Late Neolithic Vinča Harvest profile (Ovis/Capra) Middle/Late Neolithic Selevac Harvest profile (Ovis/Capra) Late Neolithic Petnica Harvest profile (Ovis/Capra) Late Neolithic Paynes Milk Production Paynes Wool Production Harvest profile (Ovis/Capra) Eneolithic Novačka Ćuprija Harvest profile (Ovis/Capra) Eneolithic Petnica Harvest profile (Ovis/Capra) Early Bronze Age Novačka Ćuprija Harvest profile (Ovis/Capra) Middle Bronze Age Vinča Harvest profile (Ovis/Capra) Late Bronze Age Novačka Ćuprija Harvest profile (Ovis/Capra) Late Bronze Age Livade Harvest profile (Ovis/Capra) Late Bronze Age Petnica Harvest profile (Ovis/Capra) Late Bronze Age Harvest profile (Ovis/Capra) Early Iron Age Kadica Brdo Harvest profile (Ovis/Capra) Final Neolithic Megalo Nisi Galanis Harvest profile (Ovis/Capra) Final Neolithic/Early Bronze Age Megalo Nisi Galanis Harvest profile (Bos taurus) Early Neolithic Foeni-Salaş Harvest profile (Bos taurus) Early Neolithic Blagotin Harvest profile Early Neolithic Bos vs. Ovis/Capra Harvest profile (Bos taurus) Middle Neolithic Stragari Harvest profile (Bos taurus) Middle Neolithic Petnica Harvest profile (Bos taurus) Middle Neolithic Harvest profile (Bos taurus) Late Neolithic Opovo Harvest profile (Bos taurus) Late Neolithic Vinča Harvest profile (Bos taurus) Late Neolithic Petnica Harvest profile (Bos taurus) Middle/Late Neolithic Selevac Harvest profile (Bos taurus) Late Neolithic Harvest profile (Bos taurus) Eneolithic Blagotin Harvest profile (Bos taurus) Early/Middle Bronze Age Ljuljaci Harvest profile (Bos taurus) Middle Bronze Age Vinča Harvest profile (Bos taurus) Late Bronze Age Livade Harvest profile (Bos taurus) Early Iron Age Kadica Brdo Percentage of age groups (Sus scrofa) Neolithic

73 75 75 76 76 77 77 78 78 79 79 80 82 82 83 83 84 84 85 85 86 87 88 89 89 94 94 95 95 96 96 97 97 98 98 99 99 100 100 101 105 105 106 106 107 107 108 108 109 109

xv

8.54 8.55 8.56 8.57 8.58 8.59 8.60 8.61 8.62 8.63 8.64 8.65 8.66 8.67

Percentage of age groups (Sus scrofa) Post Neolithic Percentage of age groups (Ovis/Capra) Neolithic Percentage of age groups (Ovis/Capra) Post Neolithic Percentage of age groups (Bos taurus) Neolithic Percentage of age groups (Bos taurus) Post Neolithic Modern comparative thin sectioning sample (Sheep #1) Modern comparative thin sectioning sample (Goat 14 b) Modern comparative thin sectioning sample (Goat #13 b) Archaeological thin sectioning sample (Kadica Brdo sample #2) Archaeological thin sectioning sample (Kadica Brdo sample #4) Archaeological thin sectioning sample (Kadica Brdo sample #6) Archaeological thin sectioning sample (Kadica Brdo sample #10) Archaeological thin sectioning sample (Vinča sample #1) Archaeological thin sectioning sample (Vinča sample #10)

xvi

110 110 111 111 112 113 113 113 114 114 114 115 115 115

LIST OF TABLES 6.1 6.2 6.3

Sheep/Goat mandibular wear stages (MWS) and suggested ages Cattle mandibular wear stages (MWS) and suggested ages Pig mandibular wear stages (MWS) and suggested ages

36 36 37

7.4 7.5 7.6 7.7 7.8 7.9 7.10 7.11 7.12 7.13 7.14 7.15 7.16 7.17 7.18 7.19 7.20 7.21 7.22

Stage distribution of Ovis aries mandibles and loose teeth from Early Neolithic Blagotin Stage distribution of Ovis/Capra mandibles and loose teeth from Early Neolithic Blagotin Stage distribution of Ovis/Capra mandibles and loose teeth from Early Iron Age Blagotin Stage distribution of Bos taurus mandibles and loose teeth from Early Neolithic Blagotin Stage distribution of Bos taurus mandibles and loose teeth from Eneolithic Blagotin Stage distribution of Ovis aries mandibles and loose teeth from Early Neolithic Foeni-Salaş Stage distribution of Ovis/Capra mandibles and loose teeth from Early Neolithic Foeni-Salaş Stage distribution of Ovis/Capra mandibles and loose teeth from Early Iron Age Foeni-Salaş Stage distribution of Bos taurus mandibles and loose teeth from Early Neolithic Foeni-Salaş Stage distribution of Ovis aries mandibles and loose teeth from Early Iron Age Kadica Brdo Stage distribution of Ovis/Capra mandibles and loose teeth from Early Iron Age Kadica Brdo Stage distribution of Bos taurus mandibles and loose teeth from Early Iron Age Kadica Brdo Stage distribution of Sus scrofa mandibles and loose teeth from Early Iron Age Kadica Brdo Stage distribution of Ovis/Capra mandibles from Late Bronze Age Livade Stage distribution of Bos taurus mandibles and loose teeth from Late Bronze Age Livade Stage distribution of Sus scrofa mandibles and loose teeth from Late Bronze Age Livade Stage distribution of Bos taurus mandibles and loose teeth from Early/Middle Bronze Age Ljuljaci Stage distribution of Sus scrofa mandibles and loose teeth from Early/Middle Bronze Age Ljuljaci Stage distribution of Ovis aries mandibles and loose teeth from Late Neolithic/Final Neolithic Megalo Nisi Galanis Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic/ Final Neolithic Megalo Nisi Galanis Stage distribution of Ovis aries mandibles and loose teeth from Final Neolithic Megalo Nisi Galanis Stage distribution of Ovis/Capra mandibles and loose teeth from Final Neolithic Megalo Nisi Galanis Stage distribution of Ovis aries mandibles and loose teeth from Final Neolithic/ Early Bronze Age Megalo Nisi Galanis Stage distribution of Ovis/Capra mandibles and loose teeth from Final Neolithic/ Early Bronze Age Megalo Nisi Galanis Stage distribution of Ovis/Capra mandibles and loose teeth from Eneolithic Novačka Ćuprija Stage distribution of Ovis/Capra mandibles and loose teeth from Early Bronze Age Novačka Ćuprija Stage distribution of Ovis/Capra mandibles and loose teeth from Late Bronze Age Novačka Ćuprija Stage distribution of Sus scrofa mandibles and loose teeth from Early Bronze Age Novačka Ćuprija Stage distribution of Bos taurus mandibles and loose teeth from Late Neolithic Opovo Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic Opovo Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic Petnica Stage distribution of Ovis/Capra mandibles and loose teeth from Eneolithic Petnica Stage distribution of Ovis/Capra mandibles and loose teeth from Late Bronze Age Petnica Stage distribution of Bos taurus mandibles and loose teeth from Middle Neolithic Petnica Stage distribution of Bos taurus mandibles and loose teeth from Late Neolithic Petnica Stage distribution of Sus scrofa mandibles and loose teeth from Middle Neolithic Petnica Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic Petnica Stage distribution of Sus scrofa mandibles and loose teeth from Eneolithic Petnica Stage distribution of Ovis/Capra mandibles and loose teeth from Middle Neolithic Stragari Stage distribution of Bos taurus mandibles and loose teeth from Middle Neolithic Stragari Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic Vinča Stage distribution of Ovis/Capra mandibles and loose teeth from Middle Bronze Age Vinča Stage distribution of Bos taurus mandibles and loose teeth from Late Neolithic Vinča Stage distribution of Bos taurus mandibles and loose teeth from Middle Bronze Age Vinča Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic Vinča Stage distribution of Sus scrofa mandibles and loose teeth from Middle Bronze Age Vinča Modern Ovis/Capra comparative cementum analysis – Summary of readings Archaeological Ovis/Capra cementum analysis – Summary of readings

42 42 42 43 43 45 45 45 46 47 47 47 48 49 49 49 51 51 51

7.23 7.24 7.25 7.26 7.27 7.28 7.29 7.30 7.31 7.32 7.33 7.34 7.35 7.36 7.37 7.38 7.39 7.40 7.41 7.42 7.43 7.44 7.45 7.46 7.47 7.48 7.49 7.50 7.51

xvii

52 52 52 54 54 54 55 55 57 57 57 58 58 60 60 60 61 61 61 63 63 63 65 66 66 66 67 68 69

8.52 8.53 8.54

Summary of sieving and weathering Expectations for movement in transhumant pattern Summary of strata and weathering

xviii

72 73 91

LIST OF APPENDICES A. SUMMARY OF AGEABLE TOOTH WEAR AND ERUPTION DATA BY SITE, TAXON AND PERIOD.

133

B. STAGE DISTRIBUTION DATA OF SMALL SAMPLES:

135

Table B1. Stage distribution of Ovis/Capra mandibles and loose teeth from Eneolithic Blagotin. Table B2. Stage distribution of Bos taurus mandibles and loose teeth from Early Iron Age Blagotin. Table B3. Stage distribution of Sus scrofa mandibles and loose teeth from Early Neolithic Blagotin. Table B4. Stage distribution of Sus scrofa mandibles and loose teeth from Early Iron Age Blagotin. Table B5. Stage distribution of Ovis aries mandibles and loose teeth from Early Iron Age Foeni-Salaş. Table B6. Stage distribution of Bos taurus mandibles from Early Iron Age Foeni-Salaş. Table B7. Stage distribution of Sus scrofa mandibles and loose teeth from Early Neolithic Foeni-Salaş. Table B8. Stage distribution of Capra hircus mandibles and loose teeth from Early Iron Age Kadica Brdo. Table B9. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Bronze Age Ljuljaci. Table B10. Stage distribution of Ovis/Capra mandibles and loose teeth from Early/ Middle Bronze Age Ljuljaci. Table B11. Stage distribution of Ovis/Capra mandibles and loose teeth from Middle Bronze Age Ljuljaci. Table B12. Stage distribution of Bos taurus mandibles and loose teeth from Early Bronze Age Ljuljaci. Table B13. Stage distribution of Sus scrofa mandibles and loose teeth from Early Bronze Age Ljuljaci. Table B14. Stage distribution of Sus scrofa mandibles and loose teeth from Middle Bronze Age Ljuljaci. Table B15. Stage distribution of Bos taurus mandibles and loose teeth from Final Neolithic Megalo Nisi Galanis. Table B16. Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic/ Final Neolithic Megalo Nisi Galanis. Table B17. Stage distribution of Sus scrofa mandibles and loose teeth from Final Neolithic Megalo Nisi Galanis. Table B18. Stage distribution of Bos taurus mandibles and loose teeth from Eneolithic Novačka Ćuprija. Table B19. Stage distribution of Bos taurus mandibles and loose teeth from Early Bronze Age Novačka Ćuprija. Table B20. Stage distribution of Bos taurus mandibles and loose teeth from Late Bronze Age Novačka Ćuprija. Table B21. Stage distribution of Sus scrofa mandibles and loose teeth from Eneolithic Novačka Ćuprija. Table B22. Stage distribution of Sus scrofa mandibles and loose teeth from Late Bronze Age Novačka Ćuprija. Table B23. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic Opovo. Table B24. Stage distribution of Ovis/Capra mandibles and loose teeth from Middle Neolithic Petnica. Table B25. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic/Eneolithic Petnica. Table B26. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Bronze Age/Early Iron Age Petnica. Table B27. Stage distribution of Bos taurus mandibles and loose teeth from Eneolithic Petnica. Table B28. Stage distribution of Bos taurus mandibles and loose teeth from Late Bronze Age Petnica. Table B29. Stage distribution of Bos taurus mandibles and loose teeth from Late Bronze Age/Early Iron Age Petnica. Table B30. Stage distribution of Sus scrofa mandibles and loose teeth from Late Bronze Age Petnica. Table B31. Stage distribution of Sus scrofa mandibles and loose teeth from Late Bronze Age/Early Iron Age Petnica. Table B32. Stage distribution of Ovis aries mandibles and loose teeth from Middle Neolithic Stragari. Table B33. Stage distribution of Sus scrofa mandibles and loose teeth from Middle Neolithic Stragari. Table B34. Stage distribution of Ovis/Capra mandibles and loose teeth from Eneolithic Vinča. Table B35. Stage distribution of Sus scrofa mandibles and loose teeth from Eneolithic Vinča.

xix

C. COMPARISON OF THE KNOWN AGE OF DEATH FROM MODERN SPECIMENS WITH GRANT (1975) AND PAYNE’S (1973) TOOTH ERUPTION AND WEAR SEQUENCES RECORDING METHODS.

147

D. DETAILS OF THE DENTAL CEMENTUM ANALYSIS FROM ARCHAEOLOGICAL AND MODERN SPECIMENS.

155

E. STATISTICAL ANALYSIS DATA

157

Table E1. Statistical analysis data - Late Neolithic Sus scrofa. Table E2. Statistical analysis data - Bronze Age Sus scrofa. Table E3. Statistical analysis data - comparison of major periods (Sus scrofa). Table E4. Statistical analysis data - Early Neolithic Ovis/Capra. Table E5. Statistical analysis data - Late Neolithic Ovis/Capra. Table E6. Statistical analysis data - Eneolithic Ovis/Capra. Table E7. Statistical analysis data - Early/Middle Bronze Age Ovis/Capra. Table E8. Statistical analysis data - Late Bronze Age Ovis/Capra. Table E9. Statistical analysis data - Early Neolithic Bos taurus. Table E10. Statistical analysis data - Early Neolithic Bos vs. Ovis/Capra. Table E11. Statistical analysis data - Middle Neolithic Bos taurus. Table E12. Statistical analysis data - Late Neolithic Bos taurus.

xx

CHAPTER 1 INTRODUCTION I. Theoretical background to research problem

Europe. Greenfield (1986, 1988, 1991, 1999a, 2001b) hypothesized that transhumant pastoralism became a major element of the subsistence strategies in temperate southeastern Europe (northern Balkans) only at the beginning of the Post Neolithic, that is, the Eneolithic and Bronze Age (ca. 3300 BC). At this time, there were significant changes in land use, such as the colonization of agriculturally marginal highlands. He proposed that the appearance of transhumant pastoralism was part of the shifts in settlement patterns, mortuary practices, architectural and artifactual remains, and economy (including the appearance of secondary animal products—following Sherratt 1980, 1982) that occurred at this time.

Transhumant pastoralism is an economic activity involving the seasonal movement of domestic herds between altitudinally differentiated and complementary pastures (Geddes 1983; Khazanov 1984). It is still an important form of land use in many parts of the world, in particular in mountainous regions in Africa (e.g. EvansPritchard 1940; Dahl and Hjort 1975), Europe (southern Europe—Bartosiewicz and Greenfield 1999; Greenfield 2001b; Norway—Paine 1994), South America (Argentina —Glatzer 1982; the Andes—Lynch 1971, 1980, 1983), Near East (Barth 1961; Bates 1974; Irons 1975; Sweet 1965), and Asia (Pakistan—Ehlers and Kreutzmann 2000; and Tibet—Ekvall 1968). With a stronger understanding of its origins, we can better understand its development and effects on the shaping of a culture’s social organization (Bartosiewicz and Greenfield 1999).

Although it has been argued that transhumant pastoralism, as a strategy of domestic animal management and land use, has had a long history of development in the arid regions of the Mediterranean littoral (e.g. Barker 1973; 1975; Chapman 1981; 1982; Chapman and Müller 1990; Geddes 1982; Higgs et al. 1967; Hesse 1982; Hole 1978; Hole, Flannery and Neely 1969; Wheeler Pires-Ferreira 1975), little is known about its evolutionary development in the temperate environmental zones found in the mountainous interior of the European sub-continent. Transhumant pastoralism can be historically documented in the Balkans at least as far back as the Roman era (e.g. in Dalmatia— Antonijević 1982; Dedijer 1916; Cvijić 1918; Sterud 1978). Few studies dealing with earlier periods in the temperate zone have convincingly established its presence or absence in the temperate zone. This investigation seeks to delimit the origins of transhumant pastoralism in Europe, specifically within the temperate regions of southeastern Europe (northern Balkans). The results of this study will provide an important comparative example and determine if the ancestry of transhumant pastoralism lies deeper in the pasts than suggested by historical reconstructions.

Historically, transhumant pastoralism has been a significant part of the economy in southern Europe, including the Mediterranean littoral and the Balkan Peninsula (Bartosiewicz and Greenfield 1999; Chang and Koster 1986; Cherry 1988; Nisbet et al. 1991; Whickham 1985). Archaeologically, many researchers assume and some explicitly argue (Geddes 1983; Greenfield 1986a, 1988, 1999a, 2001a; Halstead 1981, 1996; Harding 2000; Sherratt 1981, 1983a; Sterud 1978) that it was also an important element of the economy in prehistoric times. However, the exact temporal origins of transhumant pastoralism in southern Europe are still debated within the archaeological literature (Bartosiewicz and Greenfield 1999), and generate intense discussions. Over the past twenty years, several research projects have been directed at elucidating these temporal origins. Most scholars have focused on the Mediterranean littoral environment. Geddes (1983) and others (e.g. Miracle and Forenbaher 2005) proposed that transhumant pastoralism appeared at the Mesolithic-Neolithic junction because they argued that prehistoric hunters and gatherers and early subsistence farmers would have easily incorporated domestic migratory stock into their seasonal exploitation of land and resources. Halstead (1981, 1991, 1996) suggested that transhumance only appears with the emergence of Late Bronze Age and Classical complex societies in the Mediterranean. Lewthwaite (1981, 1984) and Walker (1983) argued that transhumance appeared even later, during the Classical or Medieval periods, with the need for the movement of specialized pastoralism for urban markets.

II. Research hypotheses Temporally, three major temporal “moments” have been hypothesized to be the points in time when transhumant pastoralism might have appeared. They are the following: 1.

In contrast, few projects have focused on the temperate European environment, immediately to the north of the mountainous divide between Mediterranean and central

1

Early Neolithic—Transhumant pastoralism has long been proposed to have existed from the beginning of animal domestication in Europe. It is thought to be either a continuation of the migratory patterns of Mesolithic indigenous hunter-gatherers who adopted domestic animals into their repertoire (Geddes 1982; Sterud 1978) or appeared as a natural consequence of pastoralism during the Neolithic periods (Barker 1973;

INTRODUCTION 1975; Chapman 1981; 1982; Chapman and Müller 1990; Hesse 1982; Hole 1978; Nisbet et al. 1991; Miracle and Forenbaher 2005; Wheeler Pires-Ferreira 1975).

It is with the shift to longer distance transhumance, where the animals are away from the village for a lengthy period of the year, that transhumance may become archaeologically identifiable.

2.

Early Post Neolithic—The Secondary Products Revolution model posits that the diffusion of new production technologies and domestic breeds from the East at the onset of the Post Neolithic (Eneolithic or Early Bronze Age) enabled domestic animal exploitation patterns to shift from primary (meat, hide and bone) to secondary products (wool, milk and traction). Herds increased in size to more effectively produce secondary products. To avoid placing strains upon local economies (to produce and store winter fodder), herds were moved to unoccupied highland pastures for the summer and returned to the lowlands with the onset of inclement weather (Sherratt 1981, 1983a).

3.

Iron Age (or later)—Transhumant pastoralism is historically known from vast areas of the Mediterranean littoral by classical times (e.g. Dalmatia and Greece Classical era (Antonijević 1982; Halstead 1987, 1991; Nisbet et al. 1991; Sterud 1978). The large scale long distance specialized transhumant adaptations characteristic of the area (Antonijević 1982; Sterud 1978) and elsewhere in southern Europe (Chang and Tourtelotte 1993; Halstead 1981, 1987; Lewthwaite 1981, 1984; Walker 1983) appear to be very late (early historical periods) phenomenon. Transhumant pastoralism has been hypothesized to be a function of the appearance of large urban markets and productive specialization that appear in Classical or Medieval times (seen also in the Near East at an earlier time—Lees and Bates 1974). As a result, it can be hypothesized to be a result of the formation of early complex societies in the Iron Age or with the introduction of the Roman era.

Long versus short distance transhumant migration would have different archaeological signatures because of the differences in the scale of movement. Long distance movements by specialists require the construction of pens and huts in both highland and lowland pastures (e.g. Chang and Tourtellotte 1991; Creighton and Segui 1998). This additional architectural investment results in identifiable archaeological signatures. In contrast, short distance village based herders rarely require the construction of such temporary structures, except where they are very distant from their home bases (e.g. among lowland herders who have moved into summer highland pastures). These differences are borne out by the ethnoarchaeological studies (e.g. Barker 1991; Baker 1999; Chang and Koster 1986; Nandris 1985, 1991) and we maintain should be apparent in the zooarchaeological data, as well. On the basis of ethnographic analogy, several scholars have argued over the past twenty years that transhumance pastoralism should be identifiable on the basis of the presence or absence of different age classes of domestic animals (e.g. Arnold and Greenfield 2003; Geddes 1983; Grant 1991; Greenfield 1999a; Nisbet et al. 1991). Transhumant pastoralism involves the movement of herds as part of a regional subsistence strategy. As a result, there should be a complementary pattern in season of death of these animals between highland and lowland sites. The transhumant movement of domestic stock is predictable in a mountainous temperate environmental zone. The herd will move into highland pastures in the early spring, soon after lambing/calving occurs and returns to the lowlands during the autumn. In a subsistence economy, the age groups that are slaughtered in the highlands and the lowlands will be different. Therefore, animal remains should be particularly relevant to answering the question.

The hypotheses can be aptly summarized to propose that transhumant pastoralism appears at the beginning of the Neolithic, the beginning of the Post Neolithic, or in later periods with the emergence of complex urban societies.

Therefore, the appearance of transhumant pastoralism in a region may be identified through the initiation of complementary culling practices between highland and lowland settlements. For example, if transhumant pastoralism appears in temperate southeastern Europe at the advent of the Post Neolithic, then a complementary seasonal culling pattern will begin to be apparent between from this period onwards between highland and lowland sites. The previous periods should have little or no evidence of complementarity. This hypothesis presupposes that herds were largely absent from parts of the region for part of the year. The null hypothesis is, therefore, that transhumant pastoralism was not practiced. The null hypothesis can be accepted if herds were resident year-round in each region. This would be demonstrated by a random culling pattern between highland and lowland sites.

III. Method and Technique Can transhumant pastoralism be identified from faunal remains? Several decades ago, Fleming (1971) argued against the identification of pastoralism using archaeological material. Most scholars in the intervening years, in contrast, agree that zooarchaeological (animal remains from archaeological contexts) material can be a suitable base for testing hypotheses about animal exploitation strategies (e.g. Barker 1985; Davis 1987; Hesse and Wapnish 1986; O’Connor 2000; Reitz and Wing 1999). This economic adaptation probably arose as a result of short term movement of animals between highland and lowland pastures by shepherds embedded within villages. 2

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Two techniques within zooarchaeology will be utilized in this investigation—tooth wear and eruption and dental cementum analysis. These two sources of data are used here because they are appropriate for testing the above hypothesis concerning herd movements.

often juxtaposed in close proximity. This is the kind of environment in which one would expect transhumant pastoralism to exist. Historically, transhumant pastoralism is part of the regional subsistence system (e.g. Cvijić 1918; Dedijer 1916). Hence, it is logical to expect that this practice extended back in time. Also, it is one of the few regions which have experienced transhumant pastoralism in Europe that have easily available archived zooarchaeological data from a variety of environments (including highland and lowland).

First, harvest profiles are created from the analysis of the tooth wear and eruption data from the mandibles of domestic animals, specifically, sheep (Ovis aries), goat (Capra hircus), cow (Bos taurus) and pig (Sus scrofa dom.). This technique provides information on both age and season of death information for each species. If transhumant pastoralism was present, then the tooth eruption and wear patterns from archaeological teeth of the youngest age classes will show complementary slaughtered age groups between highland and lowland regions. Those ages that are slaughtered in highland sites will be missing from the lowland sites and vice versa. The null hypothesis is therefore, if transhumant pastoralism was not practiced, all age groups will be present in both areas.

B. Temporal span Data from the beginning of the Early Neolithic through the Early Iron Age of the central Balkans will be evaluated. This time range was selected because all previous research indicates that the origins of transhumant pastoralism in the region occurred within this time span. Therefore, the time range has to extend beyond the point where transhumant pastoralism occurs in order to be identified. It is necessary to include periods where it probably did not yet exist in order to pin point the appearance, rather than simply the existence, of transhumant pastoralism.

Second, the results of the tooth wear and eruption analysis are supplemented by cementum analysis of mandibular Ovis/Capra teeth to provide additional seasonality estimates. If transhumant pastoralism was present in temperate southeastern Europe during the Post Neolithic, then the season of death information will show complementary season of occupation of highland and lowland sites. The null hypothesis is that if transhumant pastoralism was not practiced, then the data will not show a complementary pattern of seasonal culling between highland and lowland sites.

C. Data There are two sources of data in this investigation—ancient and modern. The ancient data from the region are used to directly test the above hypotheses. The modern data were collected and used as a control for the application of both techniques (above) used to test the hypotheses. The ancient data can be divided into two types: computerized database and limited archaeological tooth material. Greenfield collected a large database on the zooarachaeology of the region between 1977 and 1994. These data are computerized and archived at the University of Manitoba.

While not new, techniques such as tooth wear and eruption and cementum analyses have not previously been applied to this problem at the same time. The mandibular remains of four species of domestic animals (Ovis aries, Capra hircus, Bos taurus and Sus scrofa) are examined for evidence of transhumant movement. The first three species are believed to have engaged in such movements, while the latter species is included as a control, as pigs are not commonly herded in a transhumant manner.

The identification stage of analysis, including species determinations from all of the sites to be presented here, have already been completed and presented elsewhere (e.g. Greenfield 1986a, 1994, 1997, 2005b, in press a, b; Greenfield and Fowler 2002, 2005). These data were the basis for his previous analyses (Greenfield 1986a, 1988a, 1999a, 2001a) to identify the advent of transhumant pastoralism. Greenfield’s research lumped together the cranial and post-cranial data. However, the cranial material, specifically the mandibles and teeth, were not given the separate attention they deserve. Many researchers (e.g. Payne 1973; Grant 1975, 1982; Crabtree 1982) would argue that mandibular data provide the most accurate age information for archaeological material since it is less affected by differential preservation and recovery. Mandibular tooth eruption and wear data will provide better control over both age and season of death information for each species.

IV. Data A. The region In order to fill the lacunae in our knowledge concerning the emergence of transhumant pastoralism, the above hypotheses will be tested by re-examining the evidence from one region—the central Balkans (central region of the northern Balkans in southeastern Europe—Figure 1.1). This region was chosen because it was hypothesized by Greenfield to be the first region in temperate Europe (north of the Mediterranean littoral) to experience the advent of transhumant pastoralism. The region is appropriate for this type of analysis because of the nature of its topography. Both highlands and lowlands exist, 3

HUNGARY

Da nu be

N

Mures

ROMANIA

Vojvodina

Foeni-Salas

Ti sa

Drava Danube

BOSNIA AND HERZEGOVINA

Sava

*

Transylvania Alps

Opovo v Vinca

Danube

Belgrade

v

Novacka Cuprija

Petnica Ljuljaci Kadica Brdo

Stragari

Blagotin BULGARIA a av or

M

YUGOSLAVIA

Montenegro

Livade

Balkan Mts.

Kosovo

Adriatic Sea

THE FORMER YUGOSLAV REPUBLIC OF MACEDONIA Republic boundary Autonomous province boundary

ALBANIA

River

*

Modern city

Megalo Nisi Galanis

Archaeological site Scale 1: 3,550,000

0

50km

Pindos Mts.

GREECE

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE This study presents the analysis of the mandibular tooth wear and eruption data of domestic animals, specifically, sheep (Ovis aries), goat (Capra hircus), cow (Bos taurus) and pig (Sus scrofa dom.), from the central Balkan assemblages previously analyzed by Greenfield. It also includes new and unpublished data from previously unreported sites. It is expected that the specific focus on this material will provide more concise data than has been provided by previous analysis. It will also expand the regional coverage of the study of transhumant pastoralism by including a larger sample of sites. Greenfield also brought back from his fieldwork selected samples of animal bone remains from many of the sites, including mandibular tooth remains. A portion of these remains were sectioned for cementum analysis in order to determine seasonality of slaughter, and by implication, season of occupation of sites. It is hoped that this would allow for determination of the presence of transhumant herds at sites. The University of Manitoba thin sectioning laboratory provided access to the equipment and training necessary for this investigation.

The proposed results of this research should reveal the temporal origins of transhumant pastoralism in a temperate environment, specifically in the northern Balkans. This will have a significant impact on the models for cultural development in this region. It is not until the origins of transhumance are established that its role in the evolution of complex systems of land use in temperate climate zones of southeastern Europe can be fully understood. In this book, both tooth wear and eruption and cementum analysis will be used to test for the advent of transhumant pastoralism in the central Balkans. Both tooth wear and eruption and cementum analysis have been critically evaluated for their utility in age and season of death determination within the archaeological literature. Each has problems which can affect the overall analysis. For example, tooth wear and eruption can be subject to error due to variations in diet, nutrition and health of the animals (Monks 1981: 189). These same problems may be relevant to cementum analysis, as the same factors may affect the deposition of cementum (Lieberman 1993a, b). Cementum analysis has also been criticized as being too subjective (A. Stutz, pers. comm., 2000). It is hoped that through the utilization of these techniques in conjunction with each other, some of the difficulties that limit their use may be overcome. If the application of this methodology proves successful, it should end the debate as to whether transhumant pastoralism can be revealed through archaeological remains. The application of these zooarchaeological techniques to the question of pastoral movements in prehistory promises to yield valuable information, given appropriate samples. If the application of this methodology is successful, it may be extrapolated to increase understanding of other cultures with a transhumant pastoral basis. As a result, its methodology can be applied to other areas of the world, to cultures that have a similar economic basis. Understanding the nature of pastoral patterns in an area will have significant implications for the understanding of behavioral and social patterns of the culture. This discussion continues in the next chapter with a summary of previous approaches to the archaeological investigation of transhumant pastoralism in Europe, including both the Mediterranean and the northern temperate region of the Balkans.

The modern data consists of comparative data from modern livestock breeders in Manitoba, including Ovis aries and Capra hircus mandibles with teeth. These data were collected by Arnold. Tooth wear and eruption data were recorded on all mandibles and a sample of teeth was selected for cementum analysis. A modern control sample is necessary to establish the time of formation of incremental growth structures in the cementum in order to apply this information to the study of an archaeological sample (Burke and Castenet 1995). It is hoped that the cementum analysis will provide additional seasonal information to the question of the origins of transhumant pastoralism. V. Conclusions Transhumant pastoralism is hypothesized to become a major element of the subsistence strategies of cultures in southeastern Europe during the Post Neolithic. Its adoption is proposed as the essential ingredient for the significant shifts in economic organization, settlement patterns, and social and political organization that occur during this time (Greenfield 1988, 1999a, 2001b).

5

CHAPTER 2 TRANSHUMANT PASTORALISM: SOME ETHNOGRAPHIC CONSIDERATIONS outbreaks, localized drought, the number of people involved in the herd movement and relations with nonpastoral groups. The mobility of livestock does not always correlate with human mobility. The proportion of the human population traveling with the transhumant herds can vary culturally, but also in response to a variety of political, economic and ecological circumstances (Niamir-Fuller and Turner 1999: 22). The species composition, the age and sex structure of herds are determined by several factors. Major domestic species used by pastoralists include sheep and goats, cattle, camel, horses, reindeer, llamas, and alpacas. The biology of the species and geographic conditions will be the primary determinate. These same factors will affect whether animals are utilized for primary products (such as meat), or for secondary products (such as wool, milk, or traction). Additionally, it is important to realize that economic, social, political and cultural factors are also influential (Khazanov 1984: 27).

I. Introduction to pastoralism One of the problems in investigating the origins of any type of pastoralism is the lack of any clear or agreed upon definition of pastoralism, generally, and transhumant pastoralism specifically (Bartosiewicz and Greenfield 1999; Khazanov 1984; Nisbet et al. 1991). Pastoralism is both a land use strategy and a system of animal production (Krader 1959: 499). It is a distinctive form of human subsistence economy in which domestic animals play a predominant, but not an exclusive role in the shaping of the economic and cultural lives of the people who depend on them (Galaty and Johnson 1990). There is a wide spectrum in the forms of pastoralism (Ehlers and Kreutzmann 2000: 13). This is due to a range of factors and can include the quantitative and qualitative characteristics of the herds, the extent and range of mobility, degree of inclusion of agricultural products, environment and ecological aspects of the region and the extent of ties with an external market (Logashova 1982: 53). In this chapter, we will discuss the various forms of pastoralism that are suspected to have relevance to the archaeological record in the region.

A. Nomadic pastoralism Nomadic pastoralism (also known as pure pastoralism) has been characterized by the absence of agriculture, “even in a supplementary capacity” (Khazanov 1984: 19). However, some researchers (e.g. Whittaker 1988) maintain that nomadic pastoralism never existed in a pure form, in other words with the total absence of agriculture. There is “always a spectrum in the relative importance of one towards the other” (Whittaker 1988: 1).

II. Types of pastoralism Ethnographic research on pastoral societies provides some general elements of a pastoral economy and society. In general, pastoralism is grouped into two basic types, nomadic pastoralism and semi-nomadic pastoralism. On a very basic level, both types of pastoralism can be defined as the movement of domestic herds between seasonally complementary pastures. Two basic variables are used to distinguish the types of pastoralism: the degree of mobility of the human community accompanying their herds (Ehlers and Kreutzman 2000; Rafiullah 1966), and degree of reliance upon animal and agricultural products (Khazanov 1984: 19). Further distinctions arise from consideration of the spatial (geographical and altitudinal) relationship between herd pastures (Chang 1993: 709). The choice of the variable to be emphasized often leads to confusion and contradictions in the type of pastoralism being discussed.

Nomadic pastoralism involves the movement of people and animals within a large and defined geographic area according to a set schedule. This is an economic adaptation where the major economic orientation of the culture relies upon domestic stock (Barth 1961). In nomadic pastoralism, all or the majority of the human population migrates with their domestic herds year-round within a system of pastures and do not have a fixed (i.e. sedentary) settlement (Chang 1993: 709). These pastoralists are highly mobile, move over vast areas, and are characterized by the absence of, or minimal investment in agriculture. This is the form of pastoralism has been long practiced by the Basseri tribe of South Persia (Barth 1961). Their migratory pattern involves movement between arid steppes and mountainous environments in order to utilize extensive but localized pasturelands for their herds.

The mobility of pastoral groups is not limitless. Constraints, such as geographic knowledge, social contacts, and access to resources (such as pasture), often result in pastoralists revisiting encampment points on an annual or semi-annual basis. The mobility of livestock has both spatial and temporal elements. A variety of factors will affect the distance, timing and location of the movement of livestock. These can include the location and seasonal changes of water and pasture, disease

B. Semi-nomadic pastoralism The difference between nomadic and semi-nomadic pastoralism is the degree of mobility of the human population (and settlement). In semi-nomadic 7

TRANSHUMANT PASTORALISM pastoralism, the herd moves between fixed locations (winter vs. summer pastures), but the human population has less residential flexibility and a lower degree of mobility over space and through time in comparison to nomadic pastoralism (Chang 1993: 709). Semi-nomadic pastoralism is characterized by an economy where pastoralism is the predominant activity but includes varying emphasis on agriculture as a supplementary activity. All elements of a semi-nomadic pastoral existence, including species composition, routes and timing of migration, will be affected by even limited investment in agriculture. Within semi-nomadic pastoral groups, two main alternatives are observed. It may be that the entire population in a given society is involved in both agriculture and pastoralism. Alternatively, there are specialized groups within the society that devote themselves primarily, or even exclusively, to pastoralism, alongside groups that are primarily occupied with agriculture (Khazanov 1984: 19-20).

from movement of the entire community, to no movement of the community but rather solely movement of animals under the supervision of professional shepherds (Rafiullah 1966: 6). It is possible that the same groups in a given society (or sub-society) are occupied with both agriculture and pastoralism, or conversely, there are specialized groups that devote themselves primarily, or even exclusively, to pastoralism, in conjunction with groups which are primarily occupied with agriculture (Khazanov 1984: 20). Transhumant pastoralism may also involve specialized groups within a community. The majority of the population may be involved in agriculture or other economic activities, such as hunting or gathering, while a specific group within the community is focused on the maintenance of herds as a specialized occupation. The existence of variation is also supported by recent ethnographic investigations, which have shown that in modern transhumant economies, shepherds are wage labourers hired by livestock owners. This is in contrast to modern nomadic pastoral groups, where relatives manage their personal resources. This further distinction of transhumance from other forms of pastoralism in recent ethnographic literature stems from a focus on the relationship of the shepherds to the animals (Ehlers and Kreutzmann 2000: 16).

C. Transhumant pastoralism There are two main types of movement in all types of pastoralism: vertical and horizontal. Length of transhumant migrations vary highly depending upon topography, climate, political conditions, and other variables. These periodical movements may involve journeys of several hundred kilometers or only a few kilometers (Rafiullah 1966: 5; Walker 1983: 37). Horizontal (or lateral) pastoralism occurs in large flat basins and steppes and involves the movement between spatially differentiated pastures within the same altitudinal zone. In this situation, herds are moved laterally within the same altitude zone in order to exploit seasonally available resources. Vertical (or transhumant) pastoralism occurs between altitudinally differentiated pastures, typically located in mountainous and highland zones. In this instance, herds are moved up and down mountains in order to exploit seasonally available resources at different altitudes. In transhumance, the seasonal migration of domestic herds occurs between summer pastures in the mountains and winter pastures in the lowlands (Ehlers and Kreutzmann 2000: 16). Both types of movement are apparent for Mediterranean and temperate southern Europe. Vertical pastoralism is found under conditions of topographic variability (e.g. in the mountainous areas of the Alps and Dinarics). Lateral pastoralism is found under conditions of little topographic variability (e.g. on the steppes of Eastern Europe and in the Hungarian/Pannonian Plain—Matley 1970; Szabadfalvi 1968; Vincze 1980).

Historically, transhumant pastoralism has been, and is, a significant part of the economy in the Mediterranean and the Balkans. Even today, seasonal migrations of herds and herdsmen takes place on a fairly large scale in different parts of the Balkans and neighboring countries (i.e. Albania, Bulgaria, Greece, Romania, and the various countries of the former Yugoslavia) (Bartosiewicz and Greenfield 1999; Rafiullah 1966). Many researchers (Halstead 1981; Geddes 1983; Greenfield 1986a, 1991, 1999a, 2001a) agree that it was also an important element of the economy in prehistoric times (Harding 2000: 138). As a result, this investigation will focus on a specific type of semi-nomadic pastoralism, that is, transhumant pastoralism, rather than nomadic pastoralism. III. Aspects of modern transhumant pastoralism in the northern Balkans Descriptions of the traditional way of life of transhumant shepherds are far less abundant in the ethnographic and historical literature of the northern than southern (along the Mediterranean littoral) Balkans. The dividing line is the Dinaric Mountain range. In the more arid Mediterranean littoral (including the Adriatic coast) or alpine environments, the advantages of pursuing transhumant strategies are more obvious than in temperate climatic zones. In the arid zones, stock is moved from lowland to highland pastures to escape the devastating summer heat and to secure adequate water and pasture. This would be the case in the southern Balkans or along the Mediterranean littoral. Similarly, alpine communities would move stock back up to the mountains so that their products may be more efficiently exploited. The herds of both would be moved down the mountains during the autumn to avoid the freezing temperatures

Transhumant pastoralism is part of a more broadly based economic system, which incorporates crop cultivation and transhumance in a single economic scheme (Geddes 1983: 51). The practitioners of this type of pastoralism, oftentermed mixed or specialized pastoralists (Cherry 1988: 7), are often semi-sedentary (Geddes 1983: 51). In this situation, they would be semi-nomadic pastoralists. The degree of mobility for the human populations can vary 8

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE and precipitation in the mountains, while taking advantage of the abundant winter grass growth in the lowlands (Bates 1973; Hole 1978; Rhoades and Thompson 1975; Sterud 1978).

villages. Highland shepherds practice vertical transhumance as a means to avoid variations in climate and in order to economically exploit two environmental zones that would otherwise be unable to support livestock (Nandris 1975; Sterud 1978; Zdanovski 1954). This is the environment that presents the greatest difficulties in terms of herd survival. Highland winters are severe and over wintering creates great challenges for the herd and the herder. It is also the environment most subject to overgrazing (and consequent erosion), and the need for shorter stays (Matley 1968; Zdanovski 1954).

In the temperate environment of the northern Balkans, the benefit of doing so is not as apparent. Pastoralists in nonalpine temperate environmental zones may herd their animals up into the mountains during the summer to avoid the greater heat and humidity in the lowlands and to take advantage of delayed plant growth, and to access the potentially a higher quality summer graze in the highlands. However, the need to do so is not as great. Temperature extremes in the lowland are not sufficiently extreme during the summer to drive lowland-based livestock into the mountains in search of grazing and water. Sufficient water and graze are available year-round in most low and mid-altitude pastures in temperate climatic zones. In the northern Balkans, sufficient microenvironments exist in and close to the lowlands, such as marshes, streams, hills and plains, to allow for stock to be safely herded throughout the year in the lowlands. As a result, ecologically, there are fewer incentives for pastoralists from low and mid-altitude settlements in temperate regions to practice transhumance.

Under conditions of vertical transhumance, two different strategies of settlement are found: normal and inverse. In normal transhumance, a permanent village base is maintained in the lowlands. In inverse transhumance, the permanent base is maintained in the mountains. The former is more common in the Mediterranean zone, while the latter is more common in the temperate zone (Alpine—Sterud 1978: 389). In this way, herders are able to spend a substantial amount of time close to home. A portion of the village population tends to be absent seasonally from the village since they (often men and boys) accompany the herds in their seasonal migrations. Studies of traditional pastoralism in temperate southeast Europe may be generalized into five basic types of herd movements and village settlement (cf. Matley 1970; Nandris 1985, 1991; Vincze 1980):

Lowland based herds do not have to be moved between altitudinally differentiated and seasonally complementary pastures in the search for graze and water. Instead, they are moved laterally. Even today (and in the ethnohistoric past), most livestock in the temperate lowlands are (were) grazed locally, rather than moved between distant highland and lowland pastures (Gaál and Gunst 1972; Halpern 1967; Szabadfalvi 1968; Vincze 1980). Forests (and clearings), marshes and swamplands served as pastures in the winter and early spring. In winter, the snow was thawed by the high temperatures of the “decaying, steaming boggy soil” enabling animals to find forage. This was such a favorable environment for winter that highland herders often sought them out (Szabadfalvi 1968: 145). This system was used for cattle, sheep, goats and horses, whereas pigs were often left to roam freely through nearby forests (Halpern 1967; Halpern and Kerewsky-Halpern 1986; Szabadfalvi 1968). In the summer, lowland based herds are often brought to nearby hills or to non-cultivated steppe environments (if hill country is lacking). In drier environments, the geographical range of annual migration is wider in order to access graze and water resources since they are often spread out farther (Gaál and Gunst 1972; Szabadfalvi 1968; Vincze 1980).

1.

2.

3.

4.

5.

The highland and lowland zones present different opportunities (and trials) for lowland and highland based temperate zone pastoralists (Sterud 1978; as well as Mediterranean pastoralists—Koster and Koster 1976). While the lowland based herder is not necessarily driven to a transhumant lifestyle in the temperate zone, the same situation does not exist for herders based in highland

Village is located proximate to both lowland and highlands. There will be localized herding from a village base, without overnight stays. There are no pens or huts away from the village. Village is in or surrounded by mountains. There will be localized herding on summer pasture near the village, and wintering in the village. There are pens and huts in the summer pastures. Village is in the lowlands. Herds will be moved to summer pastures in the mountains, but wintered in the village. There are pens and huts in the summer pastures. Village is in the lowlands. Spring pasturing will take place near the village, during which arable land is manured. Summer pasturing will occur in the mountains. Herds will be wintered outside of the village, often near the edge of forests or in swamps. There may be pens and huts in both the highlands and lowlands. Herds are not village based. In this situation, herds will be moved by professional transhumant shepherds from lowland winter to highland summer pastures.

In terms of seasonality of movement, the following model can be postulated. Winter settlement is in the lowlands and the herd moves into the mountains in late April or early May. The herds return to the lowlands in late September or early October (Popović 1971; Ryder 1999: 191; Sterud 1978: 391-392; Szabadfalvi 1968; 9

TRANSHUMANT PASTORALISM Vinšćak 1983, 1988; Zdanovski 1954: 121). The exact time of the movements depends largely on environmental conditions such as the melting of mountain snows in spring and the onset of bad weather in the autumn. Originally, entire families made the seasonal migration, while today it is largely the work of men (Dedijer 1916; Gürsan-Salzman 1984; Ryder 1999: 191).

and Koster 1994; Gürsan-Salzman 1984; Vinšćak 1988). Lowland based herders often retain seasonal huts or lodgings in the highlands (Lockwood 1975; Vincze 1975: 397-8; Vinšćak 1999). Traditionally, transhumant pastoralists were integrated into the communities through which they passed, by networks of exchange and lineage (marital, filial relationships, etc.—Antonijević 1982; Chalkea 1999; Lockwood 1975; Vincze 1983). These ensured access to pastures and supplies, and entailed various social obligations. But, it also meant that herders were never isolated from the mainstream of village life in the surrounding countryside.

In terms of specialization of herds, the production of milk (stored as cheese) is the major focus of modern transhumant herds (Ryder 1999: 190). This is followed by meat and wool. In the 17-18th century, 65% of the revenues in Bosnia came from milk, whereas only 1013% came from wool (Šmalcelj 1954: 203). In modern times with the widespread appearance of synthetic materials and importation of cotton, the exploitation of wool has further declined throughout the region. Shifts in meat consumption have also affected the nature of milk exploitation. For example, only goats are milked in Dalmatia and most of their milk is used for making cheese and for shepherd’s food. In contrast, sheep (ewes) are not milked. Instead, their milk is given to the lambs in order that the lambs can gain weight faster. The weight of the lamb affects its sale price—the heavier the lamb, the more it will yield in the market place and the more the herder will reap (Vinšćak 1988). This is a reflection of the shift in sheep exploitation away from wool toward meat. The same is not true for Romania, where sheep were milked as well. However, sheep are milked less in order to allow lambs to rapidly gain weight (GürsanSalzman 1984). Among herds in Bosnia, ewes lactate from 5.5-7.0 months. They tend to finish lactating and the lambs are weaned by the end of September (Dracun 1954: 257). Historically, however, sheep and goats were exploited for their full combination of potential products—milk, meat, hide and wool (Gaál and Gunst 1972: 10, 38; Serić 1956).

During 1985 (and subsequent visits in 1987-88), Greenfield was able to visit and interview shepherds in the area of the Glasinac Plateau (E. Bosnia) near the site of Kadica Brdo about contemporary patterns of animal husbandry in the area. In combination with ethnographic and ethnohistoric studies of pastoralism in the region and from nearby areas (e.g. Antonijević 1982; Halpern 1967; Lockwood 1975), it is apparent that modern practices are as complex as are the hypotheses (and implications) of the prehistoric economy. There are two patterns that are apparent: long distance and local movements. These are similar to those found even among lowland villages (Vincze 1974, 1980). Most domestic animal production in the highlands involves sheep and goats, with some cattle (see Plate 1). Locally, few farms raise pigs, but this may reflect the strong Moslem influence exerted over the region until the latter half of the last century. Cattle are grazed locally and brought home every night in order to milk them. Cattle are stalled and fodder fed through the winter. Almost all transhumance is conducted with ovicaprines (sheep and goats).

Cattle exploitation depended upon whether the animals were part of migratory herds or were stabled locally. It was only in locally stabled herds that cattle were intensively milked and used for draught (Gaál and Gunst 1972: 10). With the shifts in transport and opening of new markets, cattle production undergoes profound changes. While previously, cattle were bred with an eye to their exploitation for draught potential, meat and milk production become the principal aims (Gaál and Gunst 1972: 30). With the appearance of the tractor as a normal part of local agricultural industries, the draught potential of cattle further declined (Mileusnić 1987). The same is not true in Hungary with respect to sheep meat. There was an aversion to sheep meat, while wool remained the major product for herders (Gaál and Gunst 1972: 38). It is apparent that market conditions and local tastes affect which products are exploited. Transhumant pastoralists are summer visitors to the mountains of temperate SE Europe. In the past, they walked (or rode horses or mules) to and from pastures, moving with their flocks. Presently, many use vehicles to truck their animals and supplies across the region (e.g. Baker 1999; Chalkea 1999; Chang

Local farmers tend to move their ovicaprine flocks around the region in order to take advantage of seasonally available pasture and/or fodder. Locally moved herds are small in size and there is enough natural pasture and fodder to support the animal population. Some fodder crops are also cultivated to supplement winter supplies, mostly for the stalled cattle. Of the few sheep kept in the highlands during the winter, many of these animals are eaten through the winter as fodder supplies dwindle. Transhumant pastoralists bring flocks into the mountain basins during the summer and return to the lowlands (via the Drina) for the winter. Local herders will often buy transhumant stock in the lowlands at the end of winter (when they are cheapest), then bring them to highlands (e.g. Glasinac) during the spring. They are fattened over the summer and slaughtered during the fall/early winter. This avoids the costs involved in supporting large resident herds throughout the winter within the basin and, at the same time, satisfies local demand for meat without causing undue pressure on local graze. An interesting 10

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE subjective observation offered by several informants is that the quality of meat (as measured by taste) from animals who annually migrate is considered to be poorer than that of locally bred stock. This may be the result of the increased toughness in the meat of animals who are more active (as would occur in transhumant populations), thereby lowering taste value (cf. Jochim 1976).

IV. Conclusion From the above discussion, it is obvious that a common element of all pastoral societies is the mobility of domestic herd animals. This movement of livestock has been noted historically as a method to reduce uncertainty and risk in the environment. The mobility of pastoral herds functions to: 1) take advantage of spatially and temporally variable ecosystems, 2) take advantage of unpredictable resources, 3) utilize pastures distant from settlements, 4) make use of seasonal pastures, and 5) provide fodder for livestock at minimal labour and economic cost (Niamir-Fuller and Turner 1999: 21).

The evolution of the large scale long distance transhumant adaptations characteristic of the area (Antonijević 1982; Serić 1956; Sterud 1978) seem to be a function of the appearance of large urban markets and productive specialization (Halpern 1999). Toward the end of the 18th century, the traditional pattern of agriculture is disrupted with the development of new modes of transportation. With the creation of extensive canals in the Hungarian lowlands, large quantities of grain could be moved inexpensively from producers to consumers. This trend is accelerated with the appearance of railroads in the 19th century (Gaál and Gunst 1972: 8, 18). Prior to this, cattle could be moved to distant markets on the hoof, but it was not economic to move animals such as pigs. Pigs, sheep, and horses were raised and mostly consumed locally. Sheep and pigs were raised as food animals and in order to provide clothing, while horses were raised for riding and draught (Gaál and Gunst 1972: 8). From this point in time, we begin to see large scale movement of pigs to distant markets (Halpern 1999) and a shift in emphasis from cattle to sheep production, especially on the large estates. The large estates turn to sheep production as a consequence of increases in wool prices and the need to raise cash to pay land taxes. There are fewer changes in the small-scale peasant land holdings (Gaál and Gunst 1972: 8).

Additionally, the movement of domestic herds also protects against problems such as disease outbreaks and droughts (Niamir-Fuller and Turner 1999: 22). These functions have been noted for arid environments, where the majority of pastoral economies exist. But, they are rarely taken into consideration in more temperate environments, such as will be investigated in this work. In sum, transhumant pastoralism is a risk reducing economic strategy for increasing one’s survival in an inherently unstable environment. Environments are never stable, changing between different political, social, economic, and/or climatic conditions. Transhumance reduces risk by allowing people to shift their focus from immobile (agricultural) to mobile (animals) resources and utilize the same landscape in different ways (Sterud 1978: 383). It is for this reason that transhumance has survived as a widespread practice throughout the Mediterranean and temperate zone of southern Europe (Bartosiewicz and Greenfield 1999; Nisbet et al. 1991).

11

TRANSHUMANT PASTORALISM

Plate 1: Photograph of traditional pastoralists on the Glasinac Plateau, eastern Bosnia (1988). Photograph by Haskel J. Greenfield.

12

CHAPTER 3 PREVIOUS RESEARCH ON ORIGINS OF TRANSHUMANT PASTORALISM IN PREHISTORIC SOUTHERN EUROPE patterns and material culture then become evidence of similar shifts toward a more pastoral economy. For example, the shift to a more ephemeral settlement pattern in the Eneolithic is frequently interpreted by prehistorians as evidence of the migrations of the steppe peoples (also known as Kurgans—e.g. Bukvić 1979; Brukner, Jovanović, and Tasić 1974). It is assumed that the local economies became more pastorally oriented and transhumance arrived with nomads from the eastern steppes. However, there are frequent confusions between nomadic and transhumant pastoralism (cf. Milisauskas 1978) and between specialized and mixed agro-pastoral economies (cf. Lees and Bates 1974). Additionally, the evidence for political instability or increased pastoralism in the economy is not assured. In the prehistoric archaeological record of temperate southeast Europe, there is no unambiguous evidence for early migrations, warfare and the kinds of state-sponsored politicaleconomic instability so characteristic of the past millennium (Bankoff and Greenfield 1984; Greenfield 1986a; Sterud 1978).

I. Introduction Transhumant pastoralism has long been a significant part of the economy in the Mediterranean and the Balkans. It has been well documented from Classical through Medieval times and into the modern era (e.g. Bartosiewicz and Greenfield 1999; Cherry 1988; Hodkinson 1988; Nisbet et al. 1991; Whickham 1985). However, there are strong disagreements and much confusion between researchers when the data are pushed farther back in time. In this chapter, we will review and discuss the various archaeological theories that have been proposed for the origins of transhumant pastoralism, review some of the studies that have attempted to identify it prehistorically, and present a series of hypotheses that can be used for testing its presence with zooarchaeological data. II. Theories on the origins of transhumant pastoralism There are a plethora of theories that have been proposed to account for the origins of transhumant pastoralism. It is possible to group them into four major categories. Most of them are focused on the exploitation of agriculturally marginal environments. Some have applicability to temperate climates, such as southeast Europe. Each of the major theories is summarized and their applicability to the archaeological record of the central Balkans evaluated next.

B. Regional symbiosis The development of symbiotic relationships between groups and/or regions (cf. Barth 1956) has been another widely cited hypothesis. More complex adaptations to an environment are made possible through the specialization of productive relations. Goods and services, lacking in one group or region, are provided by another. Transhumance provides the modus operandi by which goods are moved between highland and lowland areas (Lees and Bates 1974; Bates and Lees 1977; Chapman 1981; Moreno García 1997, 1999).

A. Political instability Political instability has been the most popular perspective among prehistorians working in the Balkans (and Eastern Europe, in general). In this view, the agriculturally marginal highland zones were colonized by transhumant pastoralists as an adaptation to political instability. Herders use their migratory patterns to flee from invading armies, tax collectors, etc. The unstable political conditions existing in the Balkans (during the past millennium) and consequent detrimental effects upon agro-pastoral systems of food production are frequently cited to explain why transhumant pastoralism has been an important adaptation in the region (e.g. Brice 1974; Navy 1945).

This concept has been applied to two different periods in the Balkans. In earlier periods, such as the beginning of the Post Neolithic, the development of larger scale metallurgical extraction and the need to control both mineral/metal sources and routes of movement have been hypothesized to create an impetus for the colonization of agriculturally marginal highland marginal zones (Greenfield 2001b; Sherratt 1976). There is no evidence for the sponsorship of settlement in the highlands by lowland groups to warrant support for this model. In the Balkans and other parts of the Mediterranean, this theorem is frequently linked with the emergence of the productive specialization of large-scale complex societies. These begin to emerge in the Late Bronze Age of the Aegean and become paramount in Classical antiquity (Cherry 1988; Fotiadis 1980; Halstead 1981, 1996; Lewthwaite 1981, 1984; Walker 1983).

Balkan prehistorians have frequently seized upon the historical association often existing in the Balkans between the decline in agricultural, increased investment in pastoralism, appearance of long-distance transhumance and political instability to extrapolate similar conditions back into the past (e.g. Garašanin 1972, 1973, 1983; Gimbutas 1965, 1973). Shifts in prehistoric settlement 13

PREVIOUS RESEARCH ON ORIGINS OF TRANSHUMANT PASTORALISM IN PREHISTORIC SOUTHERN EUROPE and production strategies (including new domestic breeds) from the Near East at the onset of the Post Neolithic (Eneolithic or Early Bronze Age). Domestic animal exploitation patterns shift from a focus upon primary products (meat, hide and bone) to secondary products (wool, milk and traction). It is assumed that herd sizes increase to more effectively produce secondary products. At the same time, human populations were also increasing across the landscape. This would have resulted in the twin dilemma of increased human and domestic animal population pressure upon resources and consequent habitat degradation. The Secondary Products Revolution model posits that the agriculturally marginal highlands were colonized at this time in order to avoid placing strains upon local economies (to produce and store winter fodder). Herds were moved to highland pastures for the summer and returned to the lowlands with the onset of inclement weather.

C. A by-product of initial animal domestication in an ecologically varied habitat In this model, hunter-gatherers (i.e. Mesolithic) exploited both high- and lowlands in the search for seasonally available resources. As agriculture was introduced into the region, hunter-gatherers increasingly include domestic livestock into their subsistence system becoming herder-hunter-gatherers. Movement patterns continued as in earlier pre-agricultural periods in the seasonal round of hunting and gathering, with the addition of domestic stock. The need to find adequate seasonal pasture for the livestock begins to dominate and alter the annual migratory route, rather than the available wild resources for hunting and gathering. As wild game dwindled, transhumance was transformed from an initially advantageous agro-pastoral practice into an ecologically determined necessity in the maintenance of protein availability (Chapman 1982; Geddes 1982; Higgs et al. 1967; Jacobsen 1984; Miracle and Forenbaher 2005). Along the Mediterranean littoral, domestic stock must be moved out of the lowlands during the summer to avoid the searing heat and out of the highlands in the winter to avoid freezing temperatures and precipitation (cf. Hole, Flannery and Neely 1969). The continued spread of agricultural fields in the lowlands would eventually limit the available pasture. As time passed, herders would be forced to exploit seasonally available upland pastures to avoid over-exploiting the limited lowland pastures (cf. Geddes 1982). If protein availability was to be maintained, this model assumes that colonization and transhumance are synonymous and byproducts of initial animal domestication. But it does not deal with the possibility that a long period of time may have existed during which transhumance was not a factor in local pastoral strategies (which would also be the refuting hypothesis in this case).

Several aspects of this model are difficult to test archaeologically, such as the assumption of population pressure (Bankoff and Greenfield 1984; Greenfield 1984, 1986a) and of increased herd size (Hesse 1982). There is no logical reason that the posited shift in production strategies required growth in the size of individual herds since modern agropastoralists in the region typically operate on a small scale. Additionally, the timing of arrival of some of these new technologies is in dispute (Chapman 1982). Further, analysis of zooarchaeological data from the Balkans and other regions do not show the appearance of specialized secondary products economy at this juncture, but rather more diversified production strategies including components of both secondary and primary products (Arnold and Greenfield 2003; Greenfield 1986a, 1988, 1989, 1991, 1999a, 2001a). III. Previous research on the origins of transhumant pastoralism in Mediterranean Europe

Archaeologically, transhumant pastoralism is often assumed to be a by-product of initial animal domestication and does not deal with the possibility that a long period of time may have existed during which transhumance was not a factor in local pastoral strategies (which would also be the refuting hypothesis in this case). In most parts of southern Europe, there is very poor evidence for indigenous development or adoption of agricultural or pastoral lifestyles. This argues against the notion that transhumant pastoralism evolved out of a Mesolithic subsistence base. Additionally, in cases where there is evidence for the entry of Early Neolithic cultures into a region, there is a dearth of evidence for transhumance (Halstead 1981, 1996; Greenfield 1993).

There is extensive documentary evidence that transhumant pastoralism was well established throughout much of the Classical and Medieval Mediterranean (e.g. Grassl 1985; Gyóni 1951; Hodkinson 1988; Skydsgarrd 1974; Wickham 1983). How far back in time can it be pushed? Most early studies of the origins and nature of pre-modern transhumant pastoralism in Europe are relatively facile attempts. Usually they are simply visual associations, wherein modern transhumant routes are compared to archaeological distributions. If they are coincident, then transhumance is used as an explanation for the archaeological patterns (e.g. Bosch-Gimpera 1944; Brice 1974; Higgs 1976; Higgs et al. 1967; Sterud 1978). As Chapman (1979) long ago noted for Spain, the distributions of archaeological material do not closely coincide with transhumant routes and particularly termini of the migration routes where transhumant pastoralists would be expected to spend much of their time. Relatively little time is spent along the migration routes, in contrast to the amount of time

D. Secondary Products Revolution In this model, colonization and transhumance appear as part of the package deal associated with the Secondary Products Revolution (Sherratt 1981, 1983a). This model posits the diffusion of new domestic animal exploitation

14

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE spent in the pastures at the termini. Another pattern of research is to associate types of tombs (e.g. tumuli) or cave burials with transhumant pastoralists. This, however, is an unwarranted assumption since burial type may reflect shared heritage or class, more than an economic type such as transhumant pastoralism (Walker 1987; Govedarica 2005).

In the Early Neolithic, there is increased agricultural activity, including the appearance of cultivated cereals and legumes. Further, there is an increased ovicaprine frequency in the zooarchaeological samples. Wild resources continue to be an important element of the subsistence economy with wild boar as frequent as the sheep/goat at all sites. Other wild animals, such as red deer, roe deer, aurochs and ibex, accompany the domestic animals in upland sites. Geddes (1983) highlights the evidence for Early Neolithic seasonal occupations concomitant with transhumant movements between upland and lowland sites. Seasonal evidence from lowland sites specifically indicates an occupation during the winter and the spring lambing season. Additionally, the proximity of arable soil suitable for the cultivation of cereals and legumes suggests occupation for a greater part of the year. Year-round occupation of the Upper Aude valley (higher altitudes) did occur in historic periods. However, it required the cultivation of supplementary fodder crops and the sheltering of domestic stock from the rigorous winter cold. As a result, so it is believed this type of perennial occupation of upland areas was not occurring in the Early Neolithic periods (Geddes 1983).

Recently, a few studies have taken a much more sophisticated approach. A summary of the previous research conducted on transhumance in Europe both within southern Balkans, with its Mediterranean environment, and the temperate environment of the Northern Balkans is necessary. Some of the more intensive ones are described below. A. Mesolithic/Neolithic continuity Several studies propose that transhumance appears as a by-product of initial animal domestication in a varied ecological habitat (Geddes, 1983, 1985; Voytek 1991). The most widely cited is Geddes (1983, 1985) research in the Aude Valley of southern France, wherein he proposes that transhumant pastoralism appears at the MesolithicNeolithic junction. This region falls into the Mediterranean climatic regime and provides classic geographic conditions, such as varied relief, and a diversity of microenvironments favourable to the adoption of transhumance as a dominant economic strategy. He argues that prehistoric hunters and gatherers and early subsistence farmers would have easily incorporated domestic migratory stock into their seasonal exploitation of land and resources. Transhumance has played an important role in modern and historic times in this area.

Together, the sites of Grotte Gazel (lowland), Abri Jean Cros (middle altitude) and Abri de Dourgne (upland) are argued to demonstrate the appearance of transhumance of domestic herds during the Early Neolithic. Herds were moved seasonally into middle and high altitude environments for the exploitation of pastures. In addition, these areas continued to be exploited for boar, red and roe deer, aurochs, ibex and, undoubtedly, wild plant resources. Increasing dependence on domestic stock including ovicaprines, cattle and pigs is evident throughout the period (Geddes 1983).

In the Mesolithic of the Aude valley, the subsistence economy is largely dominated by wild species including wild boar (Sus scrofa scrofa), red deer (Cervus elaphus), and the Pyrenean ibex (Capra pyrenaica). The latter becomes more common as altitude increases. Other archaeological material and macrobotanical evidence suggest that these sites were occupied on a seasonal basis. Geddes (1983, 1985) proposes that the faunal, geographical, and archaeological evidence indicates that ovicaprine transhumance began at this time with the diverse exploitation of the lowland, middle altitude and mountain environments as small domestic herds of sheep were incorporated into the seasonal round. These domestic animals make up a small percentage of the faunal remains of these sites in both upland and lowland regions (Geddes 1983, 1985). Recently, doubt has been cast upon Geddes’ claim for domesticated sheep in late Mesolithic depositional contexts. The paucity of supposedly domesticated remains, uncertain stratigraphic association of the bones and the difficulty of distinguishing from wild and domestic forms have undermined his claim that domesticated sheep were adopted by late Mesolithic hunters and gatherers (Binder 2000).

B. Complex societies Halstead (1981, 1985, 1991, 1996) proposes that transhumant pastoralism only appears with the Late Bronze Age and Classical complex societies in Greece (and the Mediterranean littoral) and is associated with large-scale economic specialization of transhumance. He argues that the extensive use of upland pastures in a transhumant manner would not have occurred on a large scale until the development of extensive agriculture. It is only when flocks become too large to be fed year-round in either the lowlands or uplands and large scale clearance is occurring in both areas that transhumant pastoralism becomes an element of the subsistence and economic system. This requires that extensive blocks of fallow land suitable for grazing be maintained in lowland regions, as well as an urban market or comparable outlet for pastoral produce. Halstead argues that these preconditions begin to exist only from the latter half of the second millennium BC and in subsequent periods. Even at this late date, the evidence for extensive agriculture and sheep herding suggests that these practices may only have been characteristic of land use 15

PREVIOUS RESEARCH ON ORIGINS OF TRANSHUMANT PASTORALISM IN PREHISTORIC SOUTHERN EUROPE under the direct control of the late Bronze Age palaces (Halstead 1991).

Sterud (1978). In this work, he argued for local transhumance by the Early Neolithic occupants of sites in Bosnia (e.g. Obre). While he cites the faunal data collected by Bökönyi (1974b) in support of the presence of transhumance, he does not bring to bear any evidence for seasonality or complementarity between highland and lowland assemblages. This is largely because the data are absent. It is unlikely that localized and smaller-scale forms of transhumance (especially where there would only be localized movement of herds around the site) did not exist at this time in the mountain valleys of Bosnia. This kind of transhumance may have always existed, but it is impossible to demonstrate its presence or absence since it would leave behind minimal evidence (Chang and Koster 1986: 104).

A fundamental difference between the data and models offered by Geddes and Halstead relates to the nature of transhumance. Geddes envisions that transhumance is conducted as part of a group’s annual subsistence cycle. In other words, early agricultural groups would employ transhumance as part of their diversified approach to subsistence. In contrast, Halstead views the origins of subsistence from the perspective of the appearance of productive specialization. Implicit in his analyses is the view that transhumant pastoralists are specialists in a larger economic system, and that this economic adaptation would not appear until the evolution of largescale complex societies. This is a view echoed by Cherry (1988). Hence, the nature and scale of transhumance would be different. In essence, they are discussing vastly different situations and they are both probably correct with respect to the types of transhumance. Presently in the Mediterranean, transhumance is practiced largely by specialists (Chalkea 1999; Moreno García 1999; Perez and Saez 1990).

Greenfield’s research attempts to determine the origins of transhumant pastoralism in the neighboring region— Serbia (central Balkans) (Greenfield 1984, 1986, 1988, 1989, 1991, 1999a, 2001b). He evaluates the issue through the analysis of zooarchaeological remains from highland and lowland sites, ranging from the Neolithic to the Early Iron Age. He also considers other archaeological and ecological evidence for the adoption of transhumance in the Post Neolithic of the region.

Other recent research on the origins and nature of transhumance in Mediterranean Europe fall into these two categories. Most have argued for the appearance of transhumance as part of the productive specialization characteristic of later complex societies (Classical and Medieval). For example, Lewthwaite (1981) proposes that transhumance in Corsica and Sardinia appeared during the Medieval period as part of a tactical adaptation to difficult economic and political circumstance. Wickham (1985) demonstrated that, transhumance is not an ecologically determined activity, but is embedded into the social and economic context of the medieval societies of Spain.

Ecological considerations highlight the advantages of the adoption of transhumance in the Balkans during the Post Neolithic (c. 3300 BC). This is reflected in the archaeological material from eastern and central Europe which indicates minimal exploitation of agriculturally marginal areas of the highlands during the Neolithic (6500-3300 BC) (Greenfield 1999a, 2001b). These areas are not extensively utilized or colonized until the Eneolithic and Bronze Age (3300-1000 BC). Colonization of these areas includes the widespread adoption of transhumant pastoralism as part of the overall domestic animal management strategy (Greenfield 1989: 37).

IV. Previous research in temperate southeastern Europe

Archaeological evidence from the Post Neolithic also shows extensive reorganization of settlement patterns that includes more closely spaced and smaller sites with horizontally displaced occupations, lower artifact densities and insubstantial structures. These sites tend to not occupy the same loci as Late Neolithic sites, and there is little evidence for large and internally well-organized communities. Greenfield and others interpret this evidence for population redistribution. There is a move from substantial nucleation of settlements, with the regional population concentrated in a few large settlements to a pattern of an increased number of smaller and less intensively occupied residential localities. The introduction of the wagon and the plough into this region at this time may well have contributed to this redistribution, offering increased efficiency in transportation and cultivation respectively (Bankoff and Greenfield 1984; Greenfield 1984, 1986, 1999a; Sherratt 1981, 1983).

Studies such as Geddes (1983), Flannery (1965) and others emphasize the element of ecological necessity to move stock from the hot and dry lowlands in the summer to the cool and moist highlands in the winter. This concept is frequently used in order to explain the development of transhumant pastoralism in arid environments, such as the Mediterranean. However, such an ecological necessity is less obvious beyond the Mediterranean littoral. The climate north of the mountains surrounding the Mediterranean is more temperate, with a reverse seasonal relationship—wet summers in both highland and lowland. This allows stock to graze year-round in the lowlands. Hence, the discussion shifts from a focus on ecological necessity to the inclusion of cultural choices. The earliest attempt at explicitly incorporating transhumant pastoralism into discussions of prehistoric subsistence in temperate southeastern Europe was by

16

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Transhumant pastoralism would have been an efficient response to the new challenges occurring at this time. Human populations were increasing and bringing larger areas under cultivation in the low and mid-altitudes resulting in less available and predictable pasture. Additionally, the paleo-environmental data indicate that significant change in the physical, vegetative and micromammalian environments took place at this time. The net result of these shifts was a decrease in the upper altitudinal limit for cultivation, mostly affecting highland communities. The adoption of transhumance at this time encourages exploitation of minimally utilized highland zones, less suited for agriculture. It is likely that members of the lowland and mid-altitude communities would have been responsible for the colonization of these areas in order to ensure continued access to these new and vital grazing resources for their stock. The highland settlements would also remain dependent upon lower altitude settlements and pastures to buffer the seasonal variability in resource in the highlands (Greenfield 1999a, 2001b).

appears as a byproduct of the exploitation of appearance of productive specialization. In most cases, this assertion is made without bringing to bear the necessary zooarchaeological data (e.g. Clark 1991; Lewthwaite 1984). Greenfield’s analysis of the zooarchaeological remains from Late Neolithic and early Post Neolithic (Eneolithic and Bronze Age) sites lumped together the cranial and post-cranial data. An abundance of recent research has indicated that mandibular tooth remains can be a more precise means for reconstructing age of death and culling patterns (Crabtree 1982; Grant 1975; Payne 1973; and many others cited in Lyman 1994 and Reitz and Wing 1999). This study will test the validity of Greenfield’s results by extending the temporal range to include sites from the Early Neolithic through to the Early Iron Age and take advantage of the increased precision from a focus on the teeth. VI. Differences between models – comparing apples and oranges

Greenfield’s analyses (1984, 1986, 1988, 1989, 1991, 1999a, 2001a) utilized ternary (three-pole) graphs to illustrate the respective percentages of the three age cohort variables (very immature, sub-adult and adult). It was clear from the analysis that there was a distinct shift in the percentage of slaughtered age cohorts at the advent of the Post Neolithic—highland sites are beginning to slaughter individuals at younger ages than lowland sites. As a result, Greenfield interprets these data to indicate that transhumant strategies have appeared at the onset of the Post Neolithic.

While the proposals of the various scholars appear to be greatly at odds, this is not really the case. They are essentially arguing differing explanations for different times, places and adaptations. Geographically, Geddes (1982) and Halstead (1991) argue for the appearance of transhumant pastoralism in an essentially Mediterranean European environment (southern France and Greece, respectively). Within a Mediterranean environment, the benefits of pursuing such practices, escaping summer heat and the provision of adequate water and pasture are obvious (Greenfield 1999a: 15). In contrast, Greenfield (1986, 1988, 1991, 1999a, 2001b) discusses its origins in a temperate European environment (northern Balkans). Within the temperate environment, the benefits of transhumant pastoralism are less obvious as there are few ecological incentives to the movement of herds to highland areas. The temperature extremes found in the Mediterranean environment do not exist in temperate lowland areas and a variety of microenvironments exist in the northern Balkans which provide sufficient water and graze in the lowland areas year round (Greenfield 1999a, 2001b). All the researchers agree that transhumant pastoralism was an important part of the pre-modern economy of these regions.

As the advent of transhumance is hypothesized to occur at roughly the same time as this shift to secondary products, this diversified economic focus complicated the results in Greenfield’s test for the origins of transhumant pastoralism in southeastern Europe. Consideration of the use of secondary products of these herds is a factor which cannot be ignored in this region. Once it was recognized that the type of exploitation strategy will complicate identification of transhumance in a region, a new series of hypotheses was generated (see Greenfield 1999a). Greenfield (1999a) offers a series of sub-hypotheses to come up with a means for identifying transhumance amidst the complications of a diversified animal exploitation. His analysis supports Sub-Hypothesis 2C, which states that if herds (cattle and ovicaprines) were being exploited for both primary and secondary products during the Post Neolithic, then mortality profiles would show less clear cut differences between highland and lowland sites (Greenfield 1999a: 24).

Both Geddes and Greenfield propose that the adoption of transhumant pastoralism is simply a change in local adaptation. In contrast, Halstead (1981, 1991), Lewthwaite (1981, 1984) and others suggest that transhumant pastoralism only appears with the Late Bronze Age and/or Classical complex societies in the Mediterranean and is associated with large-scale economic specialization of transhumance. Halstead (1981: 334) and Walker (1983) further maintain that extensive use of upland pastures in a transhumant manner would not have occurred on a large scale until

Similar to Greenfield, several researchers have attempted to examine the relationship between the advent of transhumance and secondary products exploitation in the Mediterranean region. These studies invariably rely upon and note the changes in settlement pattern and subsistence accompanying the advent of the Post Neolithic. They argue that transhumant pastoralism 17

PREVIOUS RESEARCH ON ORIGINS OF TRANSHUMANT PASTORALISM IN PREHISTORIC SOUTHERN EUROPE the development of extensive agriculture. They explicitly link this type of economic strategy with the use of secondary animal products. Several researchers, notably Lees and Bates (1974) and Cribb (1990) also argue in support of a later date for specialized pastoralism in the Near East, suggesting that large scale mobility of people and their domestic herds had to occur at a time when specialization in either a pastoral or agricultural production had became a possibility.

VII. Conclusions Many researchers (Halstead 1981; Geddes 1983; Greenfield 1986a, 1999a, 2001b) agree that transhumant pastoralism was an important element of the economy in prehistoric times. This realization has become widely accepted in the literature (e.g. Harding 2000: 138; Milisauskas 2000). However, the exact temporal origins of transhumant pastoralism in Europe are still debated. Geddes (1983), focusing on the Mediterranean environment, proposes that transhumant pastoralism appears at the Mesolithic-Neolithic junction. Halstead (1981, 1991) suggests that transhumant pastoralism appears with the Late Bronze Age and Classical complex societies in the Mediterranean (Harding 2000: 138). In contrast, Greenfield (1986, 1988, 1991, 1999a, 2001a) focuses on the temperate European environment, and tested the hypothesis that transhumant pastoralism becomes a major element of the subsistence strategies in southeastern Europe only at the beginning of the Post Neolithic, that is, the Eneolithic and Bronze Age, along with the advent of secondary products exploitation. Most importantly, each of the recent approaches incorporates a regional approach to the question. This is necessary since the comparison of upland and lowland assemblages is the basic ingredient necessary for the identification of productive complementarity, as implied by transhumant pastoralism under mixed farming conditions.

Once again, it appears that the various models are not really discussing the same situation. In fact, it may seem as if they are investigating “apples and oranges”. Greenfield and Geddes are focusing upon the evolution of transhumant pastoralism as part of a diversified subsistence strategy, while Halstead, Lewthwaite, Lees and Bates, Walker, and Cribb are concerned with the evolution of specialized transhumant pastoralism as part of complex societies. It appears that Halstead is arguing that transhumance would not be visible in Greece until it became a specialized economic adaptation. From the above summary, it should be clear that the presence or absence of transhumant pastoralism in prehistoric periods has not been firmly established (Greenfield 1999a).

18

CHAPTER 4 REGIONAL ENVIRONMENT I. Introduction

valleys often exhibit very different combinations of regional environmental variables, yet retain the general pattern of environmental diversity within the area as a whole (Greenfield 1991).

An important factor in the consideration of the origins of transhumant pastoralism in any geographic area is the regional environment. Several questions need to be addressed including: Is transhumance a viable and productive economic strategy for that area? Would aspects of the regional environment affect or encourage its adoption as a dominant economic strategy? To provide answers, the area of interest must be defined and described. In this chapter the regional environment, including topography, climate and vegetation, is described in order to consider the viability of transhumant pastoralism within the study area.

A transhumance strategy within this area would allow for the productive exploitation of the variety of resource areas. Prehistoric peoples had a wealth of environmental and ecological knowledge of a variety of plants and animals, which enabled successful adaptation to a variety of conditions (Flannery 1965). The adoption of transhumant pastoralism would have offered a viable strategy in order to take full advantage of the variety of micro-environments created by the topographic and climatic variations within the northern Balkan area. Additionally, cultural factors may have played a significant role in the adoption of transhumant pastoralism. As agricultural practices required more arable land, the movement of domestic herds may have been a practical response to changes in settlement patterns and restriction of pasture for domestic stock in lowland areas.

The area of interest for this investigation is commonly referred to as the central Balkans (Figure 1.1) (and sometimes, northern Balkans). It is somewhat of an arbitrary geographic region in that it has few defining topographic, geographic or political borders. In terms of modern political units, it includes Serbia and the Vojvodina, eastern Bosnia, northwestern Bulgaria and southwestern Romania (the Banat and Oltenia). Topographically, its boundaries to the east are defined by the mountain ranges of eastern and central Serbia. The northeast division follows along the topographic break between the lowlands of the Banat and southern edge of the Transylvanian plateau. The boundary to the south is defined by the mountains dividing Serbia (including Kosovo) from Macedonia. The Dinaric Alps divide the interior from the Mediterranean littoral along the Croatian coast in the southwest. To the west, the Bosna and Sava rivers form the western boundary of the region while the Danube marks the northwest limit. The Fruška Gora defines the northeastern limit. There are no obvious geographical features to define the northern boundary (Ehrich 1965; Greenfield 1986a).

II. Landforms A. Topography The central Balkans is characterized by extreme topographic complexity and there are no rigid topographic boundaries that clearly define the extent of the region. A diversity of environmental conditions exists in both the highland and the lowland areas. Some highland areas are heavily forested, while others lack heavy forest cover. Lowland areas may be marsh-like, while others are well-drained (Greenfield 1986a).

The topography of the central Balkans displays great variation within a small area. Generally, the area can be divided into highland and lowland areas. Highland regions include the rolling hills and low mountains of central Serbia (Šumadija), the high mountain areas of eastern Bosnia, and the Iron Gates. The plains of Pannonia, the Dacian Plain and the Morava valley are lowland areas. Local climate is greatly influenced by this variation in landforms, and can be described as transitional between the arid Mediterranean climate and the more temperate climate of central Europe (Figure 4.2.) (Greenfield 1986a, 1991; Pounds 1969).

“The word “Balkan” is of Turkish origin and denotes “…a chain of mountains” (Navy 1944: 1), and is an accurate descriptive word for the region. Several major mountain chains dominate and divide the Balkan Peninsula into the major vegetation and climatic divisions that characterize the region. Perhaps the most important, and the one from which the area derives its name, is the Balkan range (or the Stara Planina in Slavonic). This mountain range is an extension of the Carpathian system which forms an inverted ‘S” as it curves through southeastern Europe. Most of the Carpathian mountain range curve is in Romania, and encircles the Transylvanian basin (Gottmann 1969). The Carpathian range is interrupted by the Danube at the Iron Gates.

Climate and plant and animal communities take on different characteristics not only in a general N-S gradient but also with increasing altitude. Neighbouring

South of this division is the Balkan mountain range which curves to the east and south of the Lower Danube. The northern slope of the Balkan range descends steeply to the 19

REGIONAL ENVIRONMENT low and flat Danube Plain, while the southern slope is a complicated system of highlands and ranges that occupies almost the entire Balkan Peninsula (Gottmann 1969).

the Morava which flows northwards from Kosovo and southern Serbia, and the Sava and the Drava which flow northwards and eastwards, respectively, from the Dinaric and Julian Alps. The Sava also has several important tributaries including the Drina, Bosna and Una. Other important rivers in the region include the Neretva, Vardar and Martisa. Only the Danube and the Vardar are capable of supporting major water transport, as the others are too shallow, rapid or rocky to allow for safe passage of even small boats. While the river system in the Balkans is extensive, there are few large lakes in the region (Danta and Hill 1996; MacDonald 1973).

To the northwest is a system of mountains ranges and plateaus that are an extension of the Italian Alps. These ranges, often called the Dinaric Alps, extend from Slovenia to Macedonia (Marković 1968; Navy 1944; Greenfield 1986a), with their highest point overlooking the Adriatic Sea. Steep cliffs overlook small plains in the lowland areas along the coast. Inland, there are an alternating series of highlands and lowlands, with valleys and basins where altitudes range from roughly 300 to 650 meters asl, which descend towards the Pannonian Plain to the north. Extensive plateaus are found above 1300 and even 2000 meters asl (Gottmann 1969).

III. Climate There are two major and two minor climatic zones found within the Balkan Peninsula (Figure 4.1). They are:

A third mountain range, which adds to the balkanization or division of the region into sectors, is the Rhodopian Massif. It extends into southeastern Serbia, Macedonia and Bulgaria. This range runs roughly from east to west to the Black Sea (Danta and Hall 1996; Marković 1968; Navy 1944). The Balkan Peninsula is a region of complex tectonic activity. There are numerous fault, fold and joint systems that influence the relief of the region. Steep rugged slopes are the dominant features of the landscape, especially along the waterways. Additionally, the predominance of limestone, especially in the western half of the peninsula, has a great effect on landform development in the area. Limestone is porous and easily eroded. This causes abundant subsurface solution weathering which creates the karst topography characteristic of this region. Features of this type of landscape include underground caves, travertine terraces, sinkholes, dolines and poljes. Dolines are closed hollows, larger than sinkholes. These can combine to form poljes, large flat areas that can extend for several kilometers and are often the only viable agricultural land in mountainous areas (Danta and Hall 1996).

1.

The Mediterranean (and Mediterranean Transitional) Zone, which includes the regions of the southern Balkans (Dalmatia, Macedonia, Greece, southern Bulgaria and the southwestern Black Sea Coast);

2.

The Continental (temperate) Zone – this includes all countries north of the regions detailed above. This is a temperate climatic zone. This zone can be divided into two sub-zones: a. The Humid Continental Warm Summer Zone, which includes the northern Balkans (Serbia, northwestern Bulgaria and southern Romania); b. The Humid Continental Cool Summer Zone, which is found to the north, closer to central Europe (Pound 1969).

The climate varies throughout the region as a function of altitude and proximity to the two neighbouring climatic zones. In general, the climatic pattern is continental with very cold winters and very hot summers (Figure 2.2) (Halpern 1967; MacDonald 1973; Navy 1944). The area of investigation lies within climatic zone 2a. It can be described as a southern temperate climate (similar to that of central Europe) that retains features of the more arid Mediterranean climate zone to the south.

B. Rivers Various rivers cross cut the major mountain ranges of the region and play an important role in placing boundaries between areas. The Sava River in the west and the Danube in the northeast have been the traditional geographic limit between the Balkans and central Europe (by those who exclude Romania). At the same time, the major river systems and their tributaries are essential in communication and movement through the mountainous, forested and marshy areas of the region from prehistory to the present (Ehrich 1965; Greenfield 1986a).

A. Temperature Throughout the region, July is the warmest month of the year, while January is the coldest. Temperatures vary in the different regions due to variability in altitude and relief. In the Vojvodina, July temperatures can reach as high as +35°C, in combination with high humidity. North of the Sava River January temperatures can fall to as low as –15°C. In the interior of the Balkan Peninsula winter temperatures are slightly warmer, ranging from –15 to –10°C. The difference between winter minima and summer maxima over the area is generally between +50 to +55°C. The most extreme temperature range is found in the Banat (eastern Vojvodina) where July maxima may reach 42°C and February minima may reach –12°C (Furlan 1977; Halpern 1967; MacDonald 1973; Navy 1944).

North of the Dinaric Mountain divide, all of the rivers and streams flow into the Danube, which empties into the Black Sea. Along the way, it runs parallel to sections of the borders between Serbia, Romania and Bulgaria. Major tributaries of the Danube systems include the Tisza and Maros which flow southwards from the Carpathians, 20

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

Figure 4.1 Map showing climatic divide between Mediterranean and temperate central Europe, (cf. Pounds 1969) B. Precipitation

Again, as with temperature, the various microenvironments vary in the amount of precipitation that they receive. The annual mean in the Šumadija is between 600 and 1200 mm, the Banat receives somewhat less (600-700 mm) and the Iron Gates region receives the highest levels, between 800 and 1200 mm (Furlan 1977).

The region is characterized by a continental rainfall pattern, where the highest levels of precipitation fall in the summer (June) and the lowest levels fall in the winter (February). There is little variation in precipitation levels throughout the region over the course of the year and seasonal variations are not as extreme as in the Mediterranean areas further south. The majority of precipitation (nearly one-half) falls between April and July and the number of days with rain is high, often more than 150 days. This pattern generally ensures adequate water throughout the agricultural growing season, although in some areas, a high evapo-transpiration rate exceeds precipitation (Barker 1975; Furlan 1977; Halpern 1967; Navy 1944).

As temperatures begin to drop in the late fall and early winter, precipitation begins to fall in the form of snow. While the earliest frost can occur as early as the end of September, the average date is October 21st. Snow accumulations occur from January to March. The last frost generally occurs around April 13th, although it can occur during the first week of May. The amount and timing of frost and snow is largely a function of altitude. In the uplands, snow accumulations can often remain for long 21

REGIONAL ENVIRONMENT periods and would have had significant effect on settlement and occupation, as movement would have been inhibited (Furlan 1977; Halpern 1967; MacDonald 1973; Navy 1944).

Dacian Plain and the Morava River valley. Grasslands and mixed forests predominate in the plains. In contrast to highland areas, the soils in the plains and river valleys are generally deep and fertile. Consequently, agriculture is limited to these lowland areas in much of the region (Danta and Hill 1996; Halpern 1967).

IV. Environmental Regions The extensive mountain ranges have created several micro-regions within the Balkan Peninsula. These can generally be divided into highland and lowland regions, each with distinguishing environmental conditions.

Pannonia, as the plains north of the Danube-Sava line are called, is also known as the Great Hungarian Plain. It is located north of the Danube and Sava Rivers and is surrounded by the Carpathian, Dinaric and Alpine Mountains. As part of the Middle Danube Basin, the Pannonian plain is characterized by the extensive river system including the Sava, Drava, Tisza, Körös, Mureş and Timiş Rivers. Pannonia is an area of low, flat relief and lies between 100 and 200 m above sea level extending into Croatia, Serbia, Romania and Hungary (Greenfield 1986a). The floodplain of the rivers is wide and is bordered by low terraces and plains. The lowland areas experience frequent and devastating flooding. There are extensive permanent and seasonal marshes that are impassable and unusable for agriculture (Barker 1975; MacDonald 1973; Nandris 1970). The annual precipitation is 600-700 mm and increases slightly as one moves westward (Furlan 1977). Soils are predominantly loess deposits, although small areas of alluvial and brown forest soils are found. Today, some lowland areas are used for agriculture. Additionally, some mixed oak forests are found in the higher elevations. On the upper terraces, pine and spruce dominate the vegetation (Navy 1944).

A. Highland areas Highland regions include the rolling hills and low mountains of Central Serbia, also known as Šumadija, and the high mountain areas of eastern Bosnia, eastern Serbia, and the Iron Gates. The vegetation of highland areas is characterized by coniferous forests at the higher elevations, while scrub forests cover the coastal areas and the more southerly portions of the peninsula. Throughout the more mountainous regions of the area, vegetation is sparser than in the rolling hills and lowlands due to tree-line limits, the limestone environment and its associated karst topographic features. Additionally, the soils that cover the limestone are often acidic (Danta and Hill 1996). Šumadija is the northern-most extension of the Dinaric Alps. This is an area of high altitude (1000 m asl) that decreases as it descends into the foothills and terraces of the Pannonian Plain (200-300 m asl - Grubić 1980). In general, the average summer temperature in this region ranges from +20-25°C, while the average winter temperature ranges from 0 to +2.5°C.

The Morava valley is an extensive and important river system that is part of the Middle Danube Drainage Basin. It is split into a southern (the Južna) and a western branch (the Zapadna) and drains a large area south of the Danube. North of Niš, the western and southern branches converge to form the Lower Morava that empties into the Danube near Golubac. While the river meanders within a zone only three or four kilometers wide, the river plain is extremely wide (over 16 km) and flat. Fertile areas are found near the river, as flooding is common. The rest of the river plain is relatively dry (Barker 1975; Chapman 1981; Navy 1944).

The Balkan Mountains of eastern Serbia and their extension across the Danube in Romania are an area of jagged, broken relief. They are highest in the south, generally more than 2000 m asl, decreasing to roughly 1000 m asl in the north. In the north, while still mountainous, the relief is gentler and broken up by various rivers and streams. Settlement in the northern part of this region appears to be restricted to a series of small isolated basins that provide the only areas suitable for agriculture. The southern basins are larger and more contiguous (Navy 1944; MacDonald 1973; Stoianović 1966).

Abundant forests covered the hills and mountains bordering the Morava valley in prehistoric times. These were the focus of extensive clearing in prehistoric and historic periods, and so little remains today. However, the soils are very fertile and today are under cultivation. The main crops include corn, sunflowers, wheat, alfalfa, vines, vegetable, and fruit orchards. A continental climate characterizes the area. Rainfall (roughly 640 mm annually) is evenly distributed throughout the year, and in combination with the water retentive soil, farming is possible without the use of irrigation (Halpern 1967; Marković 1968).

The Iron Gates is a series of gorges that connect the Pannonian (or Hungarian) Plain and the Dacian basin. It is a corridor formed by the Danube cutting through the Carpathian and Balkan mountain range. While the mountains on either side of the river are steeply sloped, small basins where the river widens offers areas appropriate for settlement (MacDonald 1973; Navy 1944). B. Lowland areas

The Dacian basin is found where the Danube emerges from the Iron Gates, east of the arc of the Carpathians.

The lowland areas include the plains of Pannonia, the 22

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Here, the Danube changes to a north-south orientation and the relief of the area is much more subdued. It falls into the eastern Balkan region (Greenfield 1986a; Grubić 1980; Navy 1944).

6000 and 2500 BP. As agriculture and animal husbandry practices developed throughout this period, the increased proportion of cattle and sheep/goat would have caused progressive disturbances of the vegetation by humans. While this was a gradual process there would be long-term environmental effects resulting from continuous animal and human exploitation of vegetation with ruthless forest clearance for cultivation and the collection of fodder for domestic animals. These interpretations are supported by the faunal data that show the increase in the proportion of domestic species within assemblages (Halstead 1981). Additionally, pollen diagrams indicate extensive manipulation of the ecosystem throughout the Neolithic period (Nandris 1976). Even if one takes a highly conservative view of the magnitude of the climatic changes at the Atlantic to Sub-Boreal transition, when these are combined with the human actions of deforestation and cultivation over the course of the Neolithic, they undoubtedly contributed to ecological changes during the Bronze Age (Greenfield 1986a).

VI. Regional environment and climate in prehistory A. Environment and climate in the Balkans during the Neolithic The Neolithic period is characterized by the spread of agriculture throughout Europe and corresponds with the warm, moist Atlantic phase (ca. 6200-3300 BC), dominated by a warm, maritime climate. This period was distinguished by warm, wet winters and cool, wet summers (Butzer 1971). The climatic optimum is reached at the very beginning of the Atlantic period followed by a decrease in temperatures thereafter. During the climatic maximum, the average July temperature for Hungary ranged from a high of about +19°C to lows of +16°C toward the end of the period. Annual precipitation levels reached a peak at about 5700 BC, then decrease towards the end of the Atlantic period (Kordos 1978a).

C. Environment and climate in the Balkans during the Post Neolithic

The abundance of moisture during the Atlantic enabled maximum deciduous forest growth and spread, roughly 200-300 m higher in altitude than today. This is evident from the dramatic increases in arboreal pollen throughout the Lower Danube drainage. Pine and spruce species dominated the upper altitudes, while oak, hazel, birch and spruce were common in other areas, dependent on soil type. Loess plateaus were dominated by steppe-grasses. Dry grassland conditions appear and spread during this period as well (Kordos 1978a).

The transition from Atlantic to Sub-Boreal times (33003000 BC) is characterized by a climatic trend of gradual cooling). There is a shift to a more continental climate and is characterized by increased dryness as compared with that of the preceding Atlantic period (Butzer 1971; Frenzel 1966). Climate was significantly cooler with an increase in frequency of temperature oscillations. Winter temperatures decease while summer temperatures increase slightly. While average July temperatures for Hungary at the beginning of the sub-Boreal demonstrated highs of roughly +16°C, these decreased to lows of roughly +15°C towards the end of the period. Annual precipitation levels are at their lowest levels of the postGlacial period at the beginning of the Sub-Boreal. Both the vertebrate remains (Kordos 1978a, 1978b) and palynological evidence from the northern Balkans support these interpretations (Frenzel 1966; Willis and Bennett 1994; Zolyomi 1980).

While minor cold fluctuations occurred throughout the Atlantic period, changes in temperature were slight. However, the last five hundred years of the period is characterized by a much sharper oscillation toward a colder climate. Changes towards the end of the period include a reduction of the altitudinal extent of the forest as spring and summer temperatures are cooler and frosts arrive earlier in the fall (Lamb 1977). Winters also become cooler. Elm and ivy species decline in the pollen spectra and spruce forests are replaced alpine meadows. The height of this cold period was reached at c. 3200 BC. This was followed by rapid improvement (stabilization) towards a more pronounced continental climate (Frenzel 1966).

Fossil molluscan evidence from eastern Moravia dated to the end of the Atlantic period show that species characteristic of forest-steppes or open forests increased steadily while those mollusc species adapted to lakes, ponds and other moist environments decreased at this time. This data presents evidence of a worsening of the water budget, with reductions in the availability of water (Frenzel 1966).

B. Anthropogenic environmental changes during the Neolithic

Changes in forest composition occurred as palynological evidence indicates a decline in elm and ivy pollen spectra (Frenzel 1966) and in increased dominance of oak and beech (Butzer 1971). Several small climatic fluctuations take place, with marked rainfall fluctuations (Butzer 1971; Frenzel 1966; Lamb 1977).

It is also important to consider human action on the environment. While agriculture moved into temperate southeastern Europe at roughly 5900 BC, the human effects on the environment were not immediately apparent. Recent palaeoecological research (Willis and Bennett 1994) indicates that the first indication of anthropogenic change on the landscape occurs between 23

REGIONAL ENVIRONMENT D. Anthropogenic environmental changes during the Post Neolithic

environmental units. The sites are described in greater detail in Chapter 7.

Sedimentary analysis has been conducted in the Mediterranean in order to provide data on processes of deposition and erosion in the past. Geomorphological and pedological changes, particularly in the formation of soils at the end of the Neolithic and in the Early Bronze Age, are evident. These are seen as mainly human-induced due to reduced vegetation cover and impeded drainage. Through a combination of deforestation and intensive agriculture caused erosion from the hills and deposition in the plain. This resulted in a reduction of areas suitable for cereals in the lowlands and resulted in the abandonment of tell sites, which had been occupied for significant periods of time. These settlement shifts were the result of the reduction of environmental possibilities. Erosional processes in these areas caused shifts in settlement with an increase in settlements in the valley bottoms (Nandris 1977). Unfortunately, such research has not had parallels in the central Balkans. But, settlement does not shift to the valley bottoms. In fact, it spreads to include the highlands during the Post Neolithic.

Kadica Brdo is located on the Glasinac Plateau amidst the mountains of eastern Bosnia. The sites of Blagotin, Ljuljaci, Novačka Ćuprija, Petnica, and Stragari-Šljivik are found nestled among the rolling hills of the Šumadija region of central Serbia. Vinča-Belo Brdo is found on the bank of the Danube at the southern edge of Pannonia, with the hills of Šumadija to its back. Foeni-Salaş and Opovo are located within the lowlands of Pannonia. Livade is found on the bank of the Danube at the edge of the Dacian basin with its back to the mountains of eastern Serbia (Figure 1.1) (Greenfield 1986a).

It is not suggested that the climatic shifts and effect upon the environment of the region occurring at the Atlantic/sub-Boreal boundary were severe enough to have been the sole cause of the significant cultural changes seen at the Late Neolithic/Post Neolithic boundary. There were undoubtedly other factors, both social and economic, at work. However, the climatic shifts may have added some impetus to changes already taking place. It has been suggested that the adoption of transhumant pastoralism as an economic and land use strategy was an adaptation to these new ecological constraints (Greenfield 1986a).

VIII. Conclusion

When the sites are classified according to elevation and environment, they can be placed into three groups. Lowland sites include Foeni-Salaş, Livade, Novačka Ćuprija, Opovo, Stragari-Šljivik and Vinča-Belo Brdo. Mid-altitude sites are Blagotin, Ljuljaci and Petnica. The Greek Macedonian site of Megalo Nisi Galanis (included in the analysis) would fall into this category. The sample of highland sites is limited to Kadica Brdo.

The nature of the environment makes pastoralism a possible economic strategy in the region. At the same time, it does not negate the pursuit of an agricultural lifestyle. There often exists a correlation between pastoral forms of production and a range of ecological factors. While these factors may serve to effectively limit agricultural production, this is not true in all areas and does not lead directly to the adoption of pastoralism (Bonte 1981; Dhavalikar 1989). Particularly in temperate environments, ecological conditions may not directly limit agriculture. Rather it is the combination of climatic, topographic and vegetative factors that present pastoralism as an equally viable economic strategy that may be pursed. Whether it is pursued or not lies in the realm of the decision-makers rather than the environment.

VII. Site locations in relation to major environmental regions Using the above information, it is possible to allocate the archaeological sites used in this analysis to different

24

CHAPTER 5 EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY I. Introduction

II. Early Neolithic

The hypotheses to be investigated here propose that the origin of transhumance took place at one of three time periods, the advent of the Early Neolithic, the Late Neolithic/Post Neolithic junction, or with the advent of productive specialization (in this area, during the EIA or later periods). To properly investigate the other periods of interest, it is necessary to examine data from the periods preceding and following each of these temporal divisions. In order to ensure that the reader has sufficient background to evaluate the information from the various periods under discussions, the culture history of the central Balkans from the Neolithic through the Early Iron Age is summarized below.

A. Chronology The Early Neolithic began in the northern half of the Balkans about 5900 BC, as agriculture spreads north from the southern Balkans. The period ends roughly at 4900 BC, based on radiocarbon dates (Ehrich and Bankoff 1990; Manson 1990), although more recent citations indicate slightly older dates 6500-5200 BC (Tringham 2000). For the purposes of this research, we will use the older system of dates since the later periods have not yet been reassessed in this respect. B. Cultures The cultures of the northern Balkans are referred to collectively as the Karanovo I-Kremikovci-StarčevoKörös-Anza-Criş culture group. Each represents a geographical variation on the same larger material culture theme — Karanovo I in southern Bulgaria, Kremikovci in western Bulgaria and southern ex-Yugoslavia (Macedonia), Starčevo in eastern ex-Yugoslavia (Serbia and Bosnia) and Körös in southeastern Hungary and northernmost Serbia (Vojvodina), and Criş in Romania (Tringham 1971, 2000).

The manufacture and decoration of pottery has traditionally been the focus of culture historical research in the Balkans and has been used to distinguish the different cultural groups, as well as to mark distinctions between periods. A great deal of research has been done detailing the characteristics and changes through time of pottery (e.g. Dumitrescu 1983; Garašanin 1983; Bailey 2000). However, as the focus of this investigation is the economic activities and subsistence practices, the discussion will not dwell on the typological history of pottery. This chapter provides a brief summary of the major periods of interest, and the general economic and culture historical characteristics of each.

The abundance of different cultural names can generate a misleading picture of the cultural landscape. The different names correspond to variations in settlement type, location, and economy. There is little variation in material culture (Tringham 1971: 78). Others suggest that the distinct names reflect adaptations to the microenvironments of the Balkan Peninsula (Kaiser 1984: 46), or are the basis of distinct regional pottery styles (Barker 1985: 90). While these are all valid interpretations, archaeological nationalism can certainly not be ignored as a major factor (Jongsma 1997: 55; Tringham 1971).

It is widely recognized, particularly within the Neolithic periods, that there are considerable differences between the traditional relative chronology and the radiocarbon dates. These differences are often as great as one thousand years for the periods of concern (Garašanin 1983: 84). Attempting to reconcile the two systems is an exercise in frustration. It is beyond the scope of this investigation to attempt to resolve these issues. As such, the relative chronological ages of the period will be focused on, as the data have been analyzed within this context. It would be preferable to provide calibrated (BC) radiocarbon dates, as these have been corrected for the fluctuations in the formation of radioactive carbon (Cauver 2000). Unfortunately, uncalibrated dates are the most common form of reporting from the region, often without the necessary information to allow calibration. As such, dates will be given in both uncalibrated (bc) and calibrated (BC) radiocarbon years, where possible.

C. Settlement Early Neolithic settlements are generally located on the terraces surrounding river valleys or on the edge of plateaus. These areas offered dry dwelling places on abandoned levee/channel systems and much arable land, which provided a buffer against drought years, enabling populations to survive and flourish. These environments were rich in natural resources, such as wild deer and boar, fish and shellfish, water birds and a wide range of plants,

25

EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY that could be exploited (Barker 1975; Sherratt 1983b; van Andel and Runnels 1995).

these sites, wild resources may dominate the assemblage (Bökönyi 1974a; Greenfield 1993, 1994, in press a; Whittle 1996).

These early farming communities were located on the brown forest soils and alluvial deposits of the rivers. It is clear that a range of soils for easy cultivation and proximal grazing were chosen. Settlements in the flat plains to the north of the Serbian hill country were commonly found on the well-drained loess deposits of the Danube basin (Barker 1975; Chapman 1981; Sherratt 1983b; Tringham 1971; van Andel and Runnels 1995).

E. Evidence for sedentism/mobility The Early Neolithic period has been assumed to be associated with greater sedentism (than in the preceding Mesolithic), with some researchers going so far as to say that the production of food demanded permanent settlements (Kalicz 1970). Researchers have utilized many aspects of settlements such as depth of deposits, architectural remains, site size or organization and reliance on domesticates in an effort to substantiate claims of sedentism for the Early Neolithic. Although increased permanence and sedentism is listed as a defining difference between the Neolithic and the earlier periods, it is a characteristic that researchers admit is difficult to prove. In the past it has been argued that tell sites are evidence of increased permanent settlement during the Neolithic (Kotsakis 1999; Whittle 1985). Evidence from tell sites, such as Achilleion, indicate periods of abandonment and relocations within the mound (Gimbutas et al. 1989). In central Bulgaria, sites, such as Karanovo, demonstrate depth of deposits is over 12 meters, although this frequently cited figure is misleading given that it includes later occupations at the site. In the more central part of the Balkans (from northeastern Bulgaria through SW Romania and the southern part of the Hungarian plain), most settlements were relatively briefly occupied, with only thin occupation layers. Recent research (Greenfield 2000; Greenfield and Draşovean 1994; Greenfield and Jongsma in press a, b, c; Whittle 1996) has indicated that Starčevo sites have largely been misinterpreted. It is clear that settlements were not permanent sedentary affairs and probably experienced seasonal or annual abandonment (Greenfield and Jongsma in press a, b, c; Jongsma and Greenfield 2001). There is no evidence for permanent sedentism in the Early Neolithic of the central Balkans and it should not be considered a defining characteristic of the earliest phase.

There is little evidence of multi-level occupations. Most sites have thin layers, often from a single occupation (Whittle 1996). Early Neolithic sites are small, usually less than two hectares in size. Settlements are not differentiated into functionally distinct areas and there is no evidence of settlement hierarchy, social ranking or occupational specialization (Greenfield 1993; Tringham 2000). Evidence of spatial arrangement from large-scale horizontal excavations of Early Neolithic sites has revealed that settlements were organized around a large central pit house structure surrounded by a ring of smaller pit houses with an open plaza between them (Greenfield 2000; Greenfield and Jongsma 2003, in press a and b: Jongsma 1997; Jongsma and Greenfield 2001). D. Fauna Domestication was introduced into the southern Balkans from the Near East (ca. 6500 BC). In this region, the similar climatic conditions resulted in no significant changes in domesticated species or exploitation systems. However, as agriculture spreads northwards across the climatic divide into the northern Balkans during the Early Neolithic, the existing domesticated plants (wheat and barley) and animals (sheep and goat) required adaptation to the temperate environment. Additionally, there was the need for the domestication of new species already indigenous to the area, such as cattle and pigs (Bökönyi 1974a; Greenfield 1993; Tringham 1971), although recent genetic research indicates that European cattle were probably descended from Near Eastern ancestors (Edwards et al. 2004).

Increased investment in architecture within a settlement can also be used as an indicator of increased sedentism (Whittle 1996). Semi-subterranean houses, or pit houses, are a common type of architecture during the Early Neolithic of the central Balkans. These structures are characterized by very little modification of the living area. They were often large and irregular in shape, contained no internal structural features of postholes, and roofs were simple wooden affairs (Jongsma 1997; Jongsma and Greenfield 2001). These structural elements suggest structural impermanence and would have been appropriate for a lifestyle of considerable mobility (Whittle 1996: 52).

In the overall subsistence economy, hunting and animal husbandry were of roughly equal importance. When sites are examined individually, hunting dominated over husbandry in some sites, and the opposite occurred in others. There is a tendency for the proportion of domestic animals on any given site to decrease as one goes from south to north. The common domestic animals included sheep and goat, cattle and pig. Sheep dominated the fauna assemblages over cattle, goats and pigs. Only small numbers of goat and pig remains were found. Wild species included red deer, aurochs and wild pig. In settlements near appropriate water sources, fish, waterfowl and other marine resources were exploited. In

The evidence for permanence rests on the assumption that increased investment in architecture and the constraining needs of subsistence activities will result in increased 26

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE sedentism. However, the current reinterpretation of tell sites as discussed above and the extensive investigation into architectural styles in this area during the Early Neolithic does not support these assumptions (Jongsma 1997; Jongsma and Greenfield 2001). Additionally, it should be realized that the adoption of animal husbandry and plant cultivation does not inevitably lead to sedentism (Whittle 1996). Greenfield and Jongsma (in press a, b, and c) believe that most flat Early Neolithic sites in the northern Balkans were occupied for short time periods, for at most two or three years. These cannot be considered permanent settlements, since a degree of mobility was still an important element of the cultures.

that dominates the central Balkans is the Vinča A and B culture (c. 4500-4100 BC) (Chapman 1981; Greenfield 1986a; Tringham 1971), also known as the VinčaTordoš culture (Garašanin 1983). C. Settlement In the Middle Neolithic, more sites appear and these are more widely spread across the landscape and encompass a wider range of sizes. Sites are located in an increasing number of environments, but are still restricted to the lowlands or upland riverine basins (e.g. Obre I). Sites began to expand in size, with some reaching over 50 hectares. However, it is likely that this large area of occupation was the result of a series of shifting smaller occupations (Tringham and Krstić 1990).

III. Middle Neolithic A. Chronology

D. Fauna

The Middle Neolithic period of the northern Balkans covers the period from 4900-4100 BC (Chapman 1981; Greenfield 1986a). The chronology for this period may appear confusing, as there are two major chronological systems in use. Milojčić (1949) divides the culture into four major periods, A through D, each with additional sub-phases. Also widely used is the system by Garašanin (1983). It divides the culture into only two major periods (with subdivisions), the earlier Vinča-Tordoš (I and II) and later Vinča-Pločnik (I, IIA and IIB). These two systems are easily reconcilable as they are nearly coterminus with “Vinča-Tordoš I = Vinča A, Vinča-Tordoš II = Vinča B, Vinča-Pločnik I = Vinča C, Vinča-Pločnik II = Vinča D” (Chapman 1981; Greenfield 1991: 163).

The Early Vinča sites were agricultural communities, but had less dependence on wild resources than in previous periods (Leković 1991; Tringham 1971). Most characteristic about this period was the increase in the importance of domestic cattle. In the more heavily forested areas, the primary domestic species was cattle, followed in decreasing order of frequency by pigs and then by sheep/goat. In the drier more open ecological zones, sheep/goat predominated, with cattle in second place, and pigs a distant third (Greenfield 1986a). There is evidence for increased local domestication as remains of aurochs increased as well as the transitional forms between domestic and wild cattle. The same pattern of local domestication is seen in pig. Wild pig remains increased as well as the occurrence of transitional forms. In contrast, species such as red deer decreased in importance (Bökönyi 1974a).

However, in Chapman’s (1981: 31) extensive exploration of the Vinča culture, a new division of the culture into early and late phases is suggested. This new interpretation is based on radiocarbon dates and artifactual evidence that indicate only a part of the total length of the Vinča culture is represented by the type-site of Vinča Belo Brdo. Therefore, Chapman suggests an early phase, contemporary with the Vinča A, B, and C phases at the type site, (c. 4500-3950 bc), and a late phase, (c. 39503300 bc), represented by the Vinča D phase at the type site and later occupations at other Vinča sites, such as Divostin. All this being said, the system to be used in this investigation is the most well known and most widely cited in other publications, the Milojčić and Garašanin subdivisions of the Vinča culture. In this analysis, the Middle Neolithic period is represented by the Vinča A and B cultures (c. 4500-4100 BC) (Ehrich and Bankoff 1990; Greenfield 1986a).

E. Evidence for sedentism/mobility Sedentism was a gradual process during the Neolithic period. It is clear that the degree of sedentism of settlements increases throughout the Neolithic period. Researchers do not make solid claims for sedentism until later phases of the Neolithic (Kaiser and Voytek 1983). As in the Early Neolithic, the evidence indicates that the Early Vinča sites were not permanently (yearround) occupied, but, in contrast, were more often reoccupied. This caused an increase in the formation of tell sites (Leković 1991; Tringham 1971). However, flat sites without significant occupational development continue to exist, highlighting the lack of permanent sedentism and further demonstrating the range of variation throughout the region (Whittle 1996).

B. Cultures In northern Bulgaria and southern Romania, the Middle Neolithic is represented by the Vădastra I culture in the west (Oltenia) and the Dudeştri culture in the east (Muntenia). In southern Bulgaria, the Middle Neolithic is characterized by the Veselinovo culture. The culture

There is also evidence for increases in population based upon regional survey, in particular the Selevac valley (Chapman 1990). Further evidence comes from the growth of architectural units over the course of time at various sites. Houses increase in number and become 27

EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY more solidly built and longer lasting (Tringham and Krstić 1990; Whittle 1996).

analyzed) separate into lowland and mid-altitude clusters. This is due to differences in the pig and cattle ratio between these areas. Cattle dominate in all sites but pigs are more common in the lowlands than in the uplands. The major domestic species were all exploited primarily for meat, although goat gives some indication of also being exploited for secondary products such as milk (Greenfield 1991).

IV. Late Neolithic A. Chronology The Late Neolithic period of the northern Balkans covers the period from 4100-3300 BC (Chapman 1982; Ehrich and Bankoff 1990).

Additionally, there is evidence for a significant increase in the hunting of wild species in the Late Neolithic settlements. Faunal remains from red deer, wild pig, aurochs, and fish in appropriate areas, constitute a high proportion of the faunal assemblage (Bökönyi 1974a, b; Greenfield 1986a, 1991; Russell 1998; Tringham 1971). The same type of lowland/highland division seen in the domestic species is seen in the wild species. There is a lower domestic:wild ratio in the lowland sites than in the upland sites. It has been suggested that because the mid-altitude areas were better suited for agriculture with well-drained soils and easily cleared land wild animals were pushed out of this environment. In contrast, the lowlands, surrounded by wetlands in the plains, specialized in the exploitation of microenvironments, and incorporated both wild and domestic species into the subsistence system (Greenfield 1991). Fishing continued to play an important role in the subsistence economy (Greenfield 1986a: 27).

B. Cultures The Late Neolithic is represented by the Vinča-Pločnik (I and II) culture. As discussed above, this is equivalent to the Vinča C (c. 3700-3400 bc; 4100-3900 BC) and Vinča D (c. 3400-3000 bc; 3900-3300 BC) phases (Chapman 1981; Ehrich and Bankoff 1990; Garašanin 1983). C. Settlement Vinča-Pločnik settlements are often located in the same areas as the Middle Neolithic Vinča-Tordoš settlements, on the lower terraces of the rivers. However, shifts in settlement patterns are evident as Vinča-Pločnik settlements become organized into a two level site size hierarchy. Numerous small sites surround larger sites, and they more dispersed across the landscape. It is during this period that permanent and highly organized agricultural villages are established (Chapman 1981; Kaiser and Voytek 1983; Tringham 1971; Tringham and Krstić 1990).

E. Evidence for sedentism/mobility The archaeological evidence, specifically the architectural evidence, points to increases in settlement permanency. The Middle Neolithic Vinča-Tordoš settlements were overlain in later periods by thick habitation layers of the Late Neolithic cultures (Chapman 1981; Tringham 1971). There were increases in building durability, and more rigid demarcation of space including a focus on specialized areas for economic and production activities (Bailey 2000). Sites are larger and there is greater longevity of settlement. This may be taken as an indication of increases in population size and a high degree of agricultural productivity (Chapman 1981). Many of the Late Neolithic sites show relatively deeper (commonly over four metres in depth), and stratigraphically superimposed deposits and structures. This resulted in better feature preservation and higher artifact and feature density over large areas (Greenfield 1986).

A significant change in houses is also apparent as buildings became more substantial. At this time, houses were built of thick clay on a framework of wooden posts. A central row of posts in the interior of the house supported a gabled roof. Increased investment in houses is evident with evidence for the external decoration including painting, incised decoration and even clay representations of cattle heads. Many houses had thick clay floors over a foundation of horizontal logs (Chapman 1981; McPherron and Srejović 1988; Tringham 1971). D. Fauna Trends in animal exploitation patterns established in earlier periods continue through the Late Neolithic. Domestic animals continue to dominate faunal assemblages and the significant species (cattle, sheep, goat and pig) remained unchanged. The movement away from ovicaprine exploitation to a focus on cattle continues (Bailey 2000; Bökönyi 1974a, b; Greenfield 1986a, 1991; Tringham 1971). At this time, cattle typically constitute 53-76% of the major domestic fauna. What is most interesting to note is that in this period the faunal assemblages (when quantitatively

In addition to this architectural evidence, the faunal remains from Late Neolithic sites in the central Balkans can be used to argue for sedentary lifestyles without marked occupational seasonality. The age distributions of several of the domestic species include all age groups implying year-round occupation, as do the antler data. Furthermore, comparisons of lowland and mid-altitude sites did not find any shifts in seasonality between them (Greenfield 1991).

28

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE In addition, settlement size appears to increase, but this is a very problematic assertion given the fact that sites became characterized by laterally-displaced deposits. Probably, the size of actual occupied area and population within settlements declined in this period since sites were characterized by occupation by a few family homesteads instead of a large population agglomeration. This pattern becomes even more prevalent in the next period (Bankoff and Greenfield 1984; Greenfield 1986a, 2001a).

V. Eneolithic (Chalcolithic) A. Chronology The Eneolithic period, also known as the Chalcolithic or Copper Age, was the period during which metallurgy begins to spread through the region. Stone tools, however, continue to be the dominant technology, since copper is a relatively soft metal and was mostly used for decoration. The Eneolithic period of the northern Balkans covers the period from 3300-2500 BC (Ehrich and Bankoff 1991; Garašanin 1983; Greenfield 1986a, 1999b). During this period there were significant changes in settlement patterns and subsistence that show a marked break from the Neolithic. These new patterns show continuity into the following Bronze Age (Greenfield 1986a, 2001a; Champion et al. 1984; Harding 2000; Milisauskas 2000).

While settlement patterns in the Late Neolithic were characterized by the continuous localized use of a small area, there was a change to a geographically more extensive approach in the Eneolithic. Linked with this new system were the increasing importance of domestic animals and the use of plough agriculture. As settlements expanded onto soils that were inappropriate for sustained crop agriculture, more land was cleared, which in turn left increasing amounts of land fallow to feed more domestic animals (Champion et al. 1984; Sherratt 1981, 1983a).

B. Cultures

D. Fauna

The Balkan Eneolithic is characterized by geographically wide-spread cultural complexes that are composed of a series of regional variants. The main Eneolithic cultures of interest in this investigation include the Baden group to the north, which consists of a series of regional and temporal (Baden, Kostolac, and Vučedol) variants, and the Bubanj-Hum group to the south (I-II – local southern Serbian variant). The Baden culture has been subdivided with an initial Baden subculture, followed by the Kostalac group. In central Serbia, the temporal variants (Baden and Kostolac) often cannot be separated and are described as the Baden-Kostalac variant. The Vučedol group appears after the Kostolac group in the Vojvodina (Garašanin 1983).

The changes apparent with the Secondary Products Revolution (Sherratt 1981, 1983a) begin during the Eneolithic. This included the dramatic increase in the use of domestic animals for their secondary products, such as cattle for traction with the plough and the cart, the exploitation of sheep for wool, and the milking of domestic cattle, sheep and goats. The data signal an increase in the scale of domestic animal keeping over the previous periods (Greenfield 1986a, 1988, 1989, 2004; Sherratt 1981, 1983a). As in earlier periods, the animal segment of the economy was dominated by the raising of domestic stock, primarily sheep, goats and cattle (Bökönyi 1974a; Greenfield 1986a, 1988, 1989, 2004). There is also the sense that there are more animals out on the landscape than previously (Sherratt 1981). The variation in fauna across the region is an indication that agricultural systems were closely adapted to their specific regions and environments (Champion et al. 1984).

C. Settlement During the Eneolithic, tell sites are common in the plains to the north of the Danube-Sava line, but flat sites with laterally displaced deposits are common in the hill country to the south. However many tells were abandoned during the Eneolithic. In some areas, a small number were fortified. Fortified settlements are often located on “naturally dominating, fortified positions or in places suitable for defense” (Garašanin 1983: 149). The development of metallurgy and the new importance of metal sources created a new sense of territoriality and the need to secure access to these resources. As a result, clashes often occurred and introduced the need for fortification of sites (Garašanin 1983; Greenfield 2001a).

E. Evidence for sedentism/mobility It is obvious that the Late Neolithic/Post Neolithic boundary represents an important transition period in the central Balkans. Prehistoric societies underwent dramatic reorganization as well as a demographic redistribution. The Late Neolithic pattern is one of high intra-settlement population density and size, with regional population heavily concentrated in large settlements. In contrast, the Post Neolithic pattern is one of low intra-settlement population density and size accompanied by population dispersion into a larger number of more closely spaced residential localities. This shift appears not be associated with a population decline, but rather is simply redistribution over the landscape. The largest settlements have broken up and their population dispersed while the smaller settlements continued to exist, but usually in

In the Eneolithic, there was an expansion of the geographic range of site locations. Settlement expands to include the agriculturally-marginal highlands and the spaces between previously settled areas (Champion et al. 1984: 160; Greenfield and Bankoff 1984; Greenfield 1986a, 2001a).

29

EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY different locations (Bankoff and Greenfield 1984; Greenfield 1986a).

defensible positions such as on hilltops. While these sites are small, they are considered to be regional centres and were permanently occupied. Each river valley area is dominated by one or two such sites. Lower level sites were small and undifferentiated villages or hamlets. River valleys may contain several of these sites. However, this pattern of settlement hierarchies did not occur uniformly throughout the region. Some areas have a clearer dominance hierarchy than others (Greenfield 2001a).

The evidence for sedentism and mobility within the Eneolithic is difficult to interpret. Two explanations (not necessarily incompatible) have been offered. Sherratt (1981: 292) suggested that settlements were occupied for shorter amounts of time and there was a more rapid turnover of cultivated areas, as compared with earlier periods. Greenfield (1986a: 32) suggested that part of the population began to move around the landscape rather than seasonal abandonment of settlements. Domestic stock was moved to between lowland and highland areas to provide greater protection from winter conditions and better quantity and quality of graze.

Another feature of Bronze Age settlements is that they are essentially shallow, usually with deposits of only 50 cm or less (excluding pits or disturbed plow zones). These Bronze Age settlements show insubstantial wattle and daub architecture, rarely more than one vertically definable occupation level, features that are easily disturbed by ancient and modern plowing and horizontally displaced stratigraphy. Additionally, these sites are spatially extensive and have low surface and subsurface artifact densities (Bankoff and Greenfield 1984; Greenfield 1986a).

VI. Early Bronze Age A. Chronology The Bronze Age period begins in 2500 BC, ends at 1000 BC in the central Balkans and is commonly divided into early, middle and late phases. The Early Bronze Age extends from 2700-2000 BC; 2000-1700/1600 bc (Garašanin 1983; Greenfield 1986a, 2001a; Harding 2000).

While statements on community organization in the Early Bronze Age are limited by the lack of extensive excavation, two dominant patterns emerge. Most Early Bronze Age sites demonstrate a haphazard layout of structures (Greenfield 2001a).

B. Cultures D. Fauna Cultural groups in the area, while mutually related, had their own regional limits and differences, based mainly on pottery types. These groups include Vinkovci in Srem and Slavonia, Slatina in central Serbia, and Bubanj-Hum III in the valley of the Southern (Južna) Morava River of southern Serbia (Garašanin 1983).

The changes described above for the Eneolithic, in terms of the Secondary Products Revolution, come to dominate the domestic animal assemblages (Greenfield 1986a, 1988, 1989, 2004). In the Bronze Age, farmers practiced a mixed subsistence strategy with cattle, sheep, goat and pig as the main domestic animals, exploiting them for both their primary and secondary products. But there is no evidence for specialization. Some variation in species importance occurs in response to local environmental conditions. Several changes were evident in domestic livestock frequency. Cattle decrease in frequency while still remaining the dominant species, with a corresponding increase in sheep, goat and to a lesser extent pig (Bökönyi 1974a; Greenfield 1986a).

C. Settlement At this time, the major changes in settlement patterns that begin in the Eneolithic become dominant throughout the region. The data from the Eneolithic are suggestive, but too sparse at present to arrive at more definitive conclusions. In contrast to earlier periods, the Early Bronze Age pattern is one of low intra-settlement population density and size accompanied by population dispersion into a larger number of more closely spaced residential localities (or settlement sites). Population increases and is accompanied by a redistribution of the population over the landscape. The settlements of the Bronze Age are located in a wider range of environments, utilizing most of the major environmental zones of the Balkan Peninsula. This shift may be the result of the greater diversification of subsistence that necessitated a change in land use. Settlements were no longer restricted to the rich alluvial soils along water sources, but now were found in all areas, highland and lowland (Greenfield 1986a, 2001a).

The domestic horse also appears in the region during this period. The appearance of this species is one of the features that most sharply distinguish Bronze Age animal husbandry from preceding periods. It is the first new species to appear in the region since the Early Neolithic (Bökönyi 1974a: 33). Horse remains appear to be associated with “elite” settlements during this period (Greenfield 1986a, 2006). Wild species are still present in faunal assemblages and evidently continued to be part of the subsistence economy. But their importance was greatly reduced. Fish continue to be an important component of the wild fauna, particularly in lowland sites (Bökönyi 1974a; Greenfield 1986a, 2001a).

In certain areas, such as the hills surrounding the Pannonian plain, settlement hierarchies emerge. Upper level sites are often fortified and located in naturally 30

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Age continue into the later phases. There is low intrasettlement population density and size accompanied by population dispersion into a larger number of more closely spaced residential localities. The redistribution of the population over the landscape results in few large settlements and many smaller settlements. A distribution pattern of several sites per valley, dominated by a single regional center, is common (Greenfield 1986a; Harding 2000).

E. Evidence for sedentism/mobility Significant shifts in subsistence and settlement patterns begin during the Eneolithic and continue during the Early Bronze Age (Greenfield 1986a). Most importantly is the increase in geographic range of settlement to include highland environments. This pattern began in the Eneolithic. It has been suggested that these important changes were the result of developments in agriculture and husbandry. Sherratt (1981, 1983a) argued that increased specialization in the secondary products of the main domestic species, specifically the introduction of the cart, would have resulted in increased ease of movement for people at this time. A complementary suggestion has been the adoption of transhumant pastoralism as the dominant economic strategy (Greenfield 1986a, 1988, 1999a). Transhumance has historically been a significant part of the economy in the Mediterranean and the Balkans and probably was also an important element of the economy in prehistoric times (Geddes 1983, Greenfield 1986a, Halstead 1981; Harding 2000). Each of these hypotheses suggests greater mobility of populations across the landscape. The settlement data in particular supports this idea.

Bronze Age settlements show shallow occupation horizons and have insubstantial wattle and daub architecture, rarely more than one vertically definable occupation level. These sites are spatially extensive and have low surface and subsurface artifact densities (Bankoff and Greenfield 1984; Greenfield 1986a, 2001a). D. Fauna It is important to stress that there is little variation in animal husbandry across the region. The relative importance of each species in an assemblage seems to vary in response to local environmental conditions (Bökönyi 1974a; Greenfield 1986a; Harding 2000: 134). However, it is possible to highlight some trends within the period. Cattle are the dominant species, although it has declined somewhat in importance in comparison to the previous period. Sheep/goat increases in importance at the expense of cattle and pigs, although pigs increase in frequency somewhat as well (Bökönyi 1974a: 34; Greenfield 1986a).

VII. Middle and Late Bronze Age A. Chronology The data from the Middle and Late Bronze Ages of the region are extremely similar. As a result, they will be discussed together. Within the central Balkans, the Middle Bronze Age extends from 1900-1400 BC (15001300 bc), while the Late Bronze Age extends from 14001000 BC (1300-1200 bc) (Garašanin 1983).

In contrast to the variety seen in the exploitation of domestic animals, the hunting of wild species was barely incidental and much more homogeneous throughout the Bronze Age. Aurochs were less frequent than red deer and generally less than other wild species. This indicates two features of the later Bronze Age animal exploitation patterns. First, aurochs were becoming rarer throughout the landscape. Additionally, this indicates the decline of the domestication process. Animal husbandry had advanced to the point where it can “support itself not only by increasing the stock of domestic animals without further domestication and providing the population with meat but also supplying other foodstuffs (milk), giving draught power and material for clothing” (Bökönyi 1974a: 34).

B. Cultures As the Bronze Age progressed, closely linked cultural groups evolved that shared many common features. This makes the delimitation of their exact territories rather difficult as they often overlapped. As a result, the Middle and Late Bronze Age are often grouped together in publications (e.g. Garašanin 1983; Harding 2000). Major cultural groups often extend from the first into the second sub-period and include the Vatin and Dubovac-Žuto Brdo, found in southern Pannonia (the Vojvodina) and Danubian Serbia, and the Verbicioara ceramic groups found in Romania (Dumitrescu 1983; Garašanin 1983). These cultural groups all appear to be contemporary and fall within the Middle-Late Bronze Ages of the central Balkans (Reinecke – Br A2/B1 to D – ca. 1600-1200 bc). The end of the Bronze Age in the region is usually associated with the Halstatt A (c. 1200-1000 bc) (following Reinecke – Garašanin 1983; Greenfield 1986a).

E. Evidence for sedentism/mobility Whether it was due to the increased use of the cart (Piggott 1982), or the adoption of transhumant pastoralism (Greenfield 1999a), the evidence seems to suggest that there was an increase in mobility of people across the landscape during the Bronze Age periods. Smaller site size and lack of significant architectural structures suggests a degree of impermanence in the settlements.

C. Settlement The settlement patterns established in the Early Bronze

31

EVIDENCE FOR SETTLEMENT, SEDENTISM AND MOBILITY IN CENTRAL BALKAN CULTURE HISTORY VIII. Early Iron Age

exploited for both primary and secondary products. Horse remains became a small, but common part of faunal assemblages. The most noted difference between faunal assemblages of this period and earlier periods is the almost complete lack of any bones of wild species, except in the highlands. Fishing may have continued in certain areas, but was relatively unimportant away from the riverine environments (Bökönyi 1974a; Garašanin 1983; Greenfield 2005b).

A. Chronology The chronology of the Early Iron Age in the central Balkans is once again complex. Before World War II, Reinecke’s system for central Europe was extended to the Balkans. The Late Bronze and Early Iron Age of the region was divided along the same lines as in Central Europe. The application of this system to the Balkan area was problematic since it assumed that the Iron Age began with Halstatt A.

E. Evidence for sedentism/mobility The dominant settlement pattern of the Early Iron Age is a continuation of the patterns established in the Early Bronze Age. The occurrence of smaller more dispersed settlements seems to suggest an increased mobility of the population across the landscape. These observations have been interpreted as “facilitating the growing importance of pastoralism in the economy” (Raish 1992: 15).

Subsequently, Halstatt A has been recognized to be part of the Late Bronze Age for the central Balkans (14001000 BC). It is subdivided into two sub-periods (Halstatt A1-A2). Halstatt B is the transitional period between the Late Bronze Age and Early Iron Age (1000-800 BC) and subdivided into through sub-periods (Halstatt B1-B3). Finally, the true or classic Early Iron Age of the central Balkans is characterized by Halstatt C and D (Garašanin 1983; Greenfield 1986a).

IX. Conclusions The extensive temporal range of this study, from the Early Neolithic to the Early Iron Age, necessitated a brief summary of relevant aspects of central Balkan culture history. Over the course of this vast time frame, local cultures experienced the initial colonization by early farmers in the Early Neolithic through to the evolution and dominance of the region by chiefdoms in the Early Iron Age. Other significant cultural change occurs throughout this time, specifically developments in population distribution and stability, intra- and inter-regional interaction, agriculture, transport, the development of pottery manufacture and the use of metals.

B. Cultures The Early Iron Age in southeastern Europe is represented by the Halstatt C and D (800 - 500 BC) cultures (Garašanin 1983; Greenfield 1986a). While traditionally, Halstatt material culture covers a vast area of central Europe, from Budapest into Germany (Wells 1984), it clearly extends based on similarities in material culture into the central Balkans (Garašanin 1983). C. Settlement During this period, there is a continuation of the trend of the abandonment of tell sites and settlement dispersion into smaller communities. These sites were commonly located on hilltops or other places well placed for defense (Garašanin 1983: 591) and were often fortified (Raish 1992: 15). The pattern of highland settlement continues with the location of sites along inter-valley routes of movement, probably to control the movement of goods, animals, and people (Greenfield 1999a).

The first major animal domesticates (sheep/goat, cattle and pig) appear during the Early Neolithic and continue to be the dominant fauna through all periods, including the Early Iron Age. During the Post Neolithic, the exploitation of cattle and ovicaprines shift to include their secondary products. Most importantly for this study are the indications for the appearance of transhumant pastoralism in the region at the advent of the Post Neolithic. These include dramatic shifts in human population distribution over the landscape to include the agriculturally marginal highlands and the exploitation of animals for their secondary products. It is this linkage between the shifting importance of the various domestic species over time and the potentially interrelated changes in settlement patterns that hints at the appearance of transhumance. This hypothesis will be tested in subsequent chapters.

D. Fauna It appears that the patterns of animal husbandry and exploitation that were common in the Bronze Age continue into the Early Iron Age. There is the continued dominance of cattle over the other common domestic species of sheep/goat and pig and these species were

32

CHAPTER 6 METHODOLOGY I. Introduction

uses of the domestic species. Finally, epiphyseal data are often biased because of differential attrition between early and late fusing bone ends. As a result, early fusing bone elements tend to be more resistant to attrition. This leads to a bias against certain age classes especially against the very young age classes. Their bones are not fused and are very subject to attrition (Greenfield 1986a, 1991; Lyman 1994; Munson 2000). The disadvantages of epiphyseal fusion for ageing can be overcome by the use of tooth eruption and wear data.

The general approach or methodology chosen for this investigation is zooarchaeology (also known as faunal analysis). Zooarchaeology is the study of animal bones from archaeological sites, in other words, from cultural as opposed to natural deposits (which would be palaeontology—Olsen 1971; Olsen and Olsen 1981; Reitz and Wing 1999). The primary purpose of zooarchaeology is to investigate the interactions between humans and animals. Among hunter-gathers, these data form the basis of subsistence systems. In food producing, it forms the basis for the regional economy.

Two techniques of analysis within zooarchaeology were selected for this study: tooth eruption and wear and cementum analysis. Tooth eruption and wear is the second most commonly used technique in the ageing of individual animals from zooarchaeological collections (Payne 1973; Reitz and Wing 1999). Cementum analysis is a technique that is often separated from primary faunal identification because it requires specialized equipment (Reitz and Wing 1999), is significantly more time consuming, and is destructive of specimens. Nevertheless, its application has proven useful for the determination of both age and season of death in mammals (Burke and Castenet 1995; Lieberman and Meadow 1992).

Zooarchaeology is particularly suited to the investigation of transhumant pastoralism because it utilizes the remains of the animals of interest (Geddes 1983; Greenfield 1999a). While the proportion of the human population that was involved in the movement of the livestock is still debated, the majority of animals were moved around the landscape in a mixed pastoral economy (Khazanov 1984). As a result, the remains of the animals should yield a clear pattern of complementary movement between highland and lowland sites (Greenfield 1986a, 1991, 1999a, 2001b).

II. Tooth wear and eruption

In order to test the hypotheses presented in earlier chapters, the techniques chosen for this investigation must establish both age of death and season of death of the archaeological faunal remains of the major domestic species (cattle, goats, pigs, and sheep). There are a variety of techniques for the determination of age of death in zooarchaeological samples. These include epiphyseal fusion and closure of cranial sutures, tooth growth and replacement sequences, tooth wear, incremental structures, and antler and horn development (Davis 1987; Reitz and Wing 1999; Wilson et al. 1982).

Tooth wear is one of the oldest techniques for age determination, and is particularly applicable to large herbivores (Davis 1987). It is both easy to apply and provides accurate age determination when tied into a known age sequence (Lowe 1967). The sequence and timing of the eruption of the teeth in the mandible for domestic animals has been established for a number of species (Silver 1969; Habermehl 1961; Getty 1975). It is widely utilized today (Hambleton 1999: Munson 2000).

The major technique of age determination in zooarchaeology has traditionally been epiphyseal fusion of bones (Silver 1969). There are several problems associated with this method. First, the chronological age of fusion is not known for most wild species. Second, the epiphyseal age classes are sometimes too coarse to establish meaningful mortality profiles (Klein and CruzUribe 1984: 43). It is this second characteristic which is of great concern for this investigation. Additionally, the fusion of epiphyses ceases relatively early in an animal’s life (Payne 1973: 283) when an animal reaches maturity. As a result, it is difficult to distinguish a prime adult (young but full-grown) from an aged one, based purely on the epiphyseal data (Crabtree 1982: 242; Klein and Cruz-Uribe 1984: 43). This is a necessary analytical step if we are to reconstruct the economic

The eruption of deciduous and permanent teeth in animals allows for ageing in much the same manner as the state of epiphyseal fusion of the post-cranial skeleton. The dentition of animals offers several advantages over epiphyseal fusion. Teeth can monitor age more or less continuously throughout the life of an individual, and allows for ageing beyond the range of epiphyseal fusion (Reitz and Wing 1999). Analysis of tooth wear and eruption enables a clear distinction between prime adults and senile adults (Klein and Cruz-Uribe 1984). Further, the mandibles and mandibular teeth are less affected by preservation bias (taphonomy) (Payne 1973; Munson 2000). This greater durability of teeth makes them more abundant in samples.

33

METHODOLOGY There are three main disadvantages associated with tooth wear and eruption. The primary disadvantage is that the nature of the animals’ environment, specifically, the foods eaten, and the amount of grit consumed when eating can affect the degree of wear. As well, the general nutritional health of the animal can affect wear stages of teeth (Reitz and Wing 1999). While tooth eruption and wear stages can provide a good estimation of physiological age of an individual, there are problems in attempts to assign an absolute age to wear stages (Hambleton 1999). Finally, while cow and pig can easily be identified and distinguished morphologically from each other, it is more difficult to distinguish sheep and goat mandibles, particularly with fragmentary material (Boessneck et al. 1963; Ewbanks 1964). Recent techniques have been proposed for the separation of the taxa among young animals (Payne 1985) and have recently been extended to older animals as well (Halstead et al. 2002).

Absolute age for mandibles in the database was established with the following procedure. Each mandibular tooth row is assigned a score based on the tooth eruption and wear pattern of the three molar teeth (Grant 1975; Payne 1973). The score for the mandible is then converted to an absolute age (Tables 6.1-6.3). For mandibles that are missing one of the three molar teeth, it is still possible to assign a mandibular score. The wear stage of the missing tooth can be predicted with reference to complete molar rows in which the wear stages of the teeth are the same as those that are present in the incomplete mandibles (Grant 1982; Payne 1973). It is possible to establish a range of mandibular stages, and so, a range of absolute ages, utilizing the charts discussed above (Tables 6.1-6.3). Loose teeth are useful especially in those sites where there are few or no mandibles. Normally, one does not like to use them because of the possibility of inflating the count of Minimum Number of Individual’s. But in many cases in this investigation, it is necessary to use them due to the problem of small sample size. The only loose teeth included in the examination were those molar teeth that were specifically identified to number, i.e. first, second or third molar teeth. Lower loose teeth will be dealt with as if they were mandibles with two missing teeth.

There are two main systems for the recording of tooth eruption and wear data. Both systems utilize a series of diagrams showing successive tooth wear stages, where individual wear or eruption stage of each tooth is recorded by reference to the diagrams (Grant 1982; Payne 1973). The first system to be established is by Payne (1973) and focuses on sheep and goat remains. The second system is from Grant (1975) and is applicable to sheep, goats, cattle and pigs. Halstead (1985) adapted Payne’s technique for use with cattle as well. Both systems identify tooth wear stages for permanent mandibular molars and the deciduous and permanent fourth premolar. The wear stages of each molar tooth are combined to provide a mandibular wear stage. The systems differ in terms of number of wear stages, how the tooth wear stages are defined and the relation to absolute age. As both systems have been used to record tooth wear and eruption for the sites considered here, data were converted to the same format to make them comparable using the methodology detailed by Hambleton (1999). The system ultimately utilized to describe the tooth eruption and wear from the various assemblages is Payne’s because his results allow for easier recognition of herd age structure and kill-off patterns. Additionally, Grant’s method provides neither indication of absolute age nor the relative duration of different wear stages (Hambleton 1999).

Sometimes, the mandibular score does not correspond with a single age stage. This normally happens when some teeth are missing in the tooth row. Under these conditions, the tooth wear for the missing tooth must be estimated (based on the other teeth) and the numerical score calculated. In these cases, the numerical score will cover a range of stages, instead falling into a single stage (e.g. stages B-C in the Payne system). The data will be recorded in the following manner. The raw count represents mandibles and loose teeth that have been identified as belonging to each stage on the basis of the defined criteria as defined above (Payne 1973; Grant 1975; Hambleton 1999). Following the method proposed by Payne (1973) the corrected count represents specimens that cannot be placed more closely than within a group of stages. These specimens are proportionally allocated to the age stages based on the raw count data. This is explained by Payne’s (1973) example (Figure 6.3). The modern goat comparative collection provides a control of the assigned ages of sheep/goat given by Payne (1973). The suggested ages listed by Payne for sheep/goat will be compared to the known ages of the collected specimens. While it would appear more logical to assume that the earlier breeds would be more applicable to archaeological populations, the research does not bear out that assumption.

III. Establishing absolute age using tooth eruption and wear In some investigations, it is adequate to assign individual bones to age classes, such as juvenile, immature and adult as was done by Greenfield (1986a) previously. But where assessment of the economic significance of domestic species is required, establishing the absolute age of the species is highly desirable (Bullock and Rackham 1982). For the purposes of this investigation, the assignment of absolute ages is a necessary step and will be assigned to each species as above.

While it has been argued that the 19th century data for the relationship between tooth eruption and wear and age would be most applicable to prehistoric zooarchaeological data, Payne (1985) suggests that modern 20th century eruption timetables are more 34

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE applicable to archaeological populations. Additionally, research on the dental eruption and wear of Soay sheep, a feral breed believed to be the closest living analogy to prehistoric breeds, also agrees with Silver’s modern estimates and Payne’s suggested absolute ages (Hambleton 1999). Therefore, the use of a modern collected sample for this type of testing seems both appropriate and applicable. The suggested ages for cattle and pig cannot be tested in this way, as there is no adequate comparative modern mandible collection of known age available to the researchers. The suggested absolute ages from Hambleton (1999) for these breeds will be accepted as valid.

pastoralism on a regional basis, the sample of sites needed to be extended to provide a baseline from which to test the hypotheses. As this is the first zooarchaeological approach to this question in this region, we were seeking to lay the essential foundation and to highlight areas for necessary future work. Where we see the effects of small sample size on the analysis, this immediately highlights where larger faunal assemblages are required and where future fieldwork needs to be directed. The chi square statistic was used in all cases where there are two or more profiles per period in order to compare the degree of similarity between harvest profiles from different sites. The youngest age group was eliminated from consideration due to the taphonomic issues and the age groups E through to I were combined. The analysis was performed using StatXact 4 software. The statistical approach and software was specifically chosen to deal with the issue of small sample size. A significance level (p value) of less than 0.05 was considered to be statistically significant. The software provided both a usual p value that is the original chi square test and is best for large samples. Additionally, an exact p value was calculated, which corrects for small sample size (i.e. cells less than 5) and provides a more accurate calculation (Dennis Murphy, personal communication 2001).

IV. Production strategies and construction of harvest profiles Harvest profiles of faunal remains are widely utilized by archaeologists to elucidate the exploitation strategies of a society. Herders will differentially slaughter age and sex groups according to the type of products they wish to produce. The harvest profile is the distribution of age at death for a species (Hesse 1982; Greenfield 1988, 1991; Payne 1973). Production strategies, whether the domestic animals were utilized for primary products, such as meat, or whether they were utilized for secondary products, such as milk, traction, wool or hair, will be reflected in the harvest profile derived from the data. The use of harvest profiles for the examination of exploitation strategies has several advantages over other methods. First, the species frequencies in faunal samples are subject to taphonomic biases. Second, it is nearly impossible to reconstruct the original herd structure. Finally, the goal for archaeologists is to determine the ways the animals were managed.

Through the course of analysis, it became apparent that the traditional harvest profiles constructed in the initial stages of analysis provided less than concrete evidence of transhumant movement. As such, a reanalysis of the data was undertaken. They are graphed differently. A second set of mortality profiles, more sensitive to seasonality issues, was constructed. Bar graphs were found to be more useful to illustrate the pattern. New profiles were established for all three domestic species (sheep/goat, cattle and pig) for each site and for each major period. All staged mandibles are counted and the percentage of the total sample calculated. However, rather than being sequentially subtracted from 100 and the resulting values graphed the percentage that is graphed as percentage of the total assemblage in a bar graph format. This form of presentation of the data is more sensitive to seasonality issues in the transhumant movement of herds. One can evaluate where (highland or lowland) particular age groups were predominantly slaughtered based on the original hypothesized timing of transhumant movements in the Post Neolithic.

The initial harvest profiles for this investigation are established for all three domestic taxa (sheep/goat, cattle and pig) for each site and for each major period using the methodology proposed by Hambleton (1999), following Payne (1973). Cumulative frequency graphs were used to visually depict the patterns. All staged mandibles are counted and the percentage of the total sample calculated. These percentages are then sequentially subtracted from 100 and the resulting values graphed. Loose teeth are included where they provide a sufficiently narrow age stage or are necessary to provide a sufficient sample size. Minimal sample size for inclusion was ten elements per period per site. Shennan (1988) maintains that samples of mandibles less than 15-20 should be considered too small to provide accurate harvest profiles. For more accurate statistical analyses, the minimum sample size should be forty mandibles. If we had been restricted to this number, a regional approach would have been impossible and some of the major periods of interest would not have any samples. This would have resulted in the inclusion of less than 20% of the samples. While larger samples would be more desirable, they were not available from many sites. Since our goal was to test for the origins of transhumant

V. Cementum analysis Incremental growth structures have been observed in a variety of organisms. These may be present in mineralized tissues, such as bone, molluscs, teeth, otoliths, fish spines and antler pedicles. Correlations between growth lines in teeth have been recognized for more than 30 years as the most reliable means of establishing age and season of death of individual specimens (Lieberman 1993a, b; Lieberman and Meadow 1994; Pike-Tay 1991). 35

METHODOLOGY

If there are 3As, 6 Bs, And 1C, one of the 3Abs is allocated to A and 2 to B according to the ratio 3A:6B; similarly the 14BCs are allocated 12 to B and 2 to C according to the ratio 6B:1C, giving 4As, 20Bs and 3Cs. ABCs are now allocated not according to the original 3A:6B:1C ratio, but according to the new ratio 4A:20B:3C (Payne 1973: 296). Figure 6.3. Proportional allocation example (from Payne 1973). Table 6.1. Sheep/Goat mandibular wear stages (MWS) and suggested ages Payne MWS A B C D E F G H I

Grant MWS 1-2 3-7 8-18 19-28 29-33 34-37 38-41 42-44 45+

Suggested Age (Payne 1973) 0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

Table 6.2. Cattle mandibular wear stages (MWS) and suggested ages Payne MWS A B C D E F G H I

Grant MWS 1-3 4-6 7-16 17-30 31-36 37-40 41-43 44-45 46+

Suggested Age (Halstead 1985) 0-1 months 1-8 months 8-18 months 18-30 months 30-36 months young adult adult old adult senile

36

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 6.3. Pig mandibular wear stages (MWS) and suggested ages Payne MWS

Grant MWS

A B C D E F G-I

0-1 2-8 9-17 18-32 33-42 43-46 46+

Suggested Age (Higham 1967; Bull and Payne 1982) 0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months adult

Teeth are composed of three distinct tissues, enamel, dentine and cementum. Enamel covers the crown of the tooth, while cementum covers the roots of a tooth. Dentine forms the main structure of the tooth and underlies both the enamel and the cementum (Hillson 1986). While incremental structures have been identified in all three of the dental tissues and previous study has focused on both the dentine and the cementum, it is the analysis of cementum that has proven more effective in determining both age and season of death information (Lieberman and Meadow 1994; Pike-Tay 1991). Therefore, it is the incremental structures in cementum of domestic animal teeth that are of interest in this investigation.

(1979) provide a comprehensive summary. It is generally recognized that the formation of incremental structures in cementum are governed to a considerable degree by an internal rhythm related to the animal’s growth and metabolism. However, external factors can still exert some effects on cementum formation. These can include environment and climate (Burke 1995; Grue and Jensen 1979), dietary restrictions, hormonal changes, reduced food intake (Saxon and Higham 1969) and reduced nutrition and biomechanical stress (Lieberman 1993a, b). It is likely that it is the result of a complex of factors and the need for further investigation as to their influence is required. Several techniques are available for the observation of growth structures in cementum. These include polarized light, ordinary light, stained, histological thin sections, transmitted or scanning electron microscopy and microradiography (Burke 1995). The chosen technique for this investigation is the examination of undecalcified, ground thin sections of teeth to be viewed using a polarizing light microscope with a rotating stage under both transmitted and polarized light. This is one of the few techniques that are applicable to both modern and archaeological teeth.

Acellular cementum is the most reliable for indications of age at and season of death. It is deposited towards the cervical end of the tooth root and is composed of seasonally deposited bands that appear either opaque or translucent in polarized light (Lieberman and Meadow 1994). Dental cementum has been considered extremely reliable, since it is deposited through the entirety of an animal's life and is rarely reabsorbed (Gordon 1993). Seasonal differences in the deposition of cementum are generally manifest and described as growth zones (which represent the period of active deposition of cementum), annuli (which occur as a result of growth slowdown) and Line of Arrested Growth (LAG) that occur as a result of growth cessation. Together, these structures may be referred to as a Growth Line Group and is the equivalent of a year in time (Burke 1995; Grue and Jensen 1979; Klevezal 1996).

The wider, growth zones of cementum appear light in colour and opaque under reflected light microscopy and transmitted polarized light. In contrast, these same structures will appear dark under transmitted white (or natural) light. The narrower annuli in cementum appear dark in color and translucent under reflected light microscopy as well as with transmitted polarized light. In contrast, these increments will appear bright under transmitted white (natural) light (Burke 1995; Pike-Tay 1995).

Cementum is deposited around the roots of teeth by cementoblasts. The cementum is deposited on the surface of the dentine, within and around the Sharpey’s fibers that protrude from the periodontal ligament (Lieberman and Meadow 1992). Therefore, the earliest layers of cementum are closest to the dentin, while those formed later are closer to the outside part of the root (Burke 1995; Grue and Jensen 1979; Klevezal 1996). For a full discussion of the histology and structure of the various types of cementum see Grue and Jensen (1979).

Acellular cementum is deposited at the cementum-enamel junction, or the cervical region of the tooth. This makes this area of the tooth ideal for the counting of the incremental structures in the cementum. In this region, the lines are most distinct and even. Additionally, in ungulates it has been shown that layers in cementum are often disturbed in the apical area of the root, and layers here are less distinct. As a result, the coronal half of the root is preferred for interpretation (Grue and Jensen 1979).

The causes of the annual growth cycle of cementum are still largely debated within the literature. Grue and Jensen

37

METHODOLOGY system. Then, the lower molar 1 (M1) of the left side of the mandible was extracted for thin sectioning. Only mandibular M1 was selected for sectioning for a variety of reasons.

VI. The control sample The purpose of a modern control sample is to establish the time of formation of incremental growth structures in the cementum in order to apply this information to the study of an archaeological sample (Burke and Castenet 1995). A sample of sheep and goat skulls were collected from August 2000 to March 2001 from local Manitoba abattoirs. Approximately six sheep per month were collected from Carmen Meats in Carmen, Manitoba (Canada). The goats were collected from Prairie Abattoir in Portage la Prairie, Manitoba. The abattoir slaughtered two animals per week, which were collected each month. The utilization of these local animals is considered applicable to archaeological material in the Balkans because the animals are obtained from an environment with roughly similar patterns of seasonality. The animals of the comparative collection are goats raised on smallscale farms in Manitoba. They are domestic farm animals raised with seasonal differences both in food availability and seasonal environment.

1. 2. 3. 4.

The establishment of this control sample is for the purpose of interpretation of archaeological teeth. Mandibular teeth are easier to section due to their simpler root forms. Incisors and canines are less frequent than cheek teeth in archaeological samples. Incisors and canines are more difficult to identify to the species level in archaeological samples than are cheek teeth (Pike-Tay 1991).

A total of fourteen goats were included in the sample collected monthly from August to March of the same year. As these were animals slaughtered for meat consumption, they were generally young animals (less that a year and half). Due to the young age of the goats, the comparative sample was supplemented by five sheep specimens (over the age of 18 months) in order to bulk out the older age groups of the sample.

All the goats were born in the month of April, with a mean birth date estimated by the producer to be April 15th. The goats range in age from six months to a year and a half. As the pattern of slaughter was two animals per week and the specimens were collected at roughly the middle of each month throughout the collection period, date of death was estimated over a period of two months. This changed through the time of collection when collection was at the end of each month.

The modern thin section sample of goats will provide an interesting methodological contribution by testing whether the incremental growth visible in the teeth of the goats agrees or does not agree with the eruption/wear stages and/or with the actual ages. It will act as a threeway comparison between tooth eruption and wear, incremental structures and true ages (Ariane Burke, personal communication, 2001).

It was possible to obtain several older sheep, from a private producer who raises sheep for wool. This small sample included one sheep specimen that was four and a half years old (54 months) and two animals that were a year and a half-year-old at death. These were also included in the final sample in order to expand the age range covered by the sample. Additionally, two older sheep with no age information that were collected from the abattoir were included. As a result, the final modern comparative collection consists of both sheep and goat. There are forty-three animals, twenty-nine goats and fourteen sheep, in the sample that compares tooth eruption and wear. The thin sectioning sample consists of nineteen animals, fourteen goats and five sheep.

VII. The fossil sample As cementum is the least hard of the dental tissues, it can easily be stripped from the tooth as the result of postmortem damage. As a result, only lower molar 1 (M1) still encased in the mandible are selected for the archaeological sample (Lieberman and Meadow 1994). A sample of twenty left mandibular M1 teeth in total - ten teeth from a highland site and ten teeth from a lowland site. The sites chosen were Vinča (Belo-Brdo, Serbia), a lowland site that includes material from the Late Neolithic, Eneolithic and Middle Bronze Age, and Kadica Brdo (Knežina, Bosnia) a highland site that includes material from the Early Iron Age.

It was not detrimental to the analysis to include both sheep and goats in the sample. They are often lumped together in zooarchaeological investigations due to difficulties in reliably distinguishing the two species (Boessneck et al. 1963; Payne 1973). The teeth of sheep and goat are very similar in structure and size (Hillson 1986). Other researchers have utilized the application of goat thin sections to Gazella gazella (Lieberman 1993b) based on the close phylogenetic relationship (family Bovidae).

The teeth selected for sectioning must first be removed from the mandible. This was accomplished through the use of a small handsaw and a high-speed rotary saw. Two cuts were made on either side of M1 and the mandible was broken along these lines using a pair of cutters and/or a screwdriver. The modern teeth were then soaked in a degreasing agent (in a 50:50 solution with water) in order to remove as much of the fat and grease from the teeth. Teeth were soaked for five days, and then allowed two full days to air dry. This degreasing step was omitted for the archaeological teeth. All teeth were measured

Mandibles were extracted from the skull and defleshed. The tooth wear and eruption was recorded for each mandible, both with Payne’s system and with Grant’s 38

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE according to guidelines in Von den Driesch (1976; see Appendix A and B below for these data).

scratches and polish the surface. Polishing was finished by applying a small amount of 0.05 micron powder polish to the specimen and rubbing with a chamois cloth. The slide was then accurately labeled with permanent ink.

The tooth must be encased in resin (Buehler Epoxide) in order to fill any voids in the sample and protect fragile materials. Samples were placed into vessels, or boats, suitable to the size of the tooth. The appropriate amount of both hardener and resin was calculated, weighted and then mixed together (100 parts of resin to 36 parts of hardener by weight). The mixture was poured into the boat to fully cover the sample. The vessel was placed into the vacuum chamber for 15 to 20 minutes, removed and allowed to harden for 24 hours.

VIII. Age determination with cementum analysis In most species, cementum deposition begins shortly before or just at the time of eruption (Grue and Jensen 1979). As a result, by adding the number of seasons of cementum growth to the age at which the tooth erupted the age of death is established. However, to be most accurate, one must determine age, one must establish the time of the formation of the first increment line, from which counting should begin (Klevezal 1996).

In order for the sample to be mounted on a slide for thin sectioning, an initial cut through the sample was necessary. The cut was made longitudinally, parallel to the cementum surface. An Isomet saw with a wafering blade was used. The block was then polished on a polishing/grinding machine as evenly as possible using first a 320 micron abrasive paper and, then with a 400 micron abrasive paper prior to mounting on glass slides.

The eruption time for M1 in sheep and goats will be taken to be three months, following Silver’s (1969) eruption timetable of primitive modern breeds. Because the timing of the eruption is before the beginning of the first winter, some amount of cementum is present on the root before the first winter. As a result, the first cementum layer visible corresponds to the first year of an animal’s life. In these circumstances, there is a rule to be accepted in age determination by cementum layers. It is stated as follows:

Beuhler Epoxide was then used to glue specimens to slides. A small amount of resin was spread on the slide in a “T” shape, roughly the same length as the specimen to be glued. Beginning at the top of the “T”, one edge of the specimen was placed on the slide. The block was then quickly lowered. The sample was moved around slightly to allow the resin/hardener to spread under the entire sample and release any air bubbles that may be trapped underneath the specimen. The sample was then positioned in the centre of the slide. Slides were left to dry, approximately 24 hours or more.

When we see the first (innermost) incremental line close to the dentin, it should be considered to be formed in the first winter of a given tooth function. When the first incremental line is separated form the dentin by a wide cementum band, then this line should be considered to be a line of the second winter of the tooth function (Klevezal 1996: 122).

Next, the slide was mounted to the vacuum chuck of the thin sectioning machine that cuts away the majority of the specimen block. Using 320 and 400 micron abrasive papers, the freshly cut side was then hand thinned as evenly as possible on the grinding/polishing wheel. This was necessary to allow light to pass through the sample and provide a clear representation of growth increments. During this process the slide was checked periodically under the microscope to estimate how much more thinning was required to allow for optimal visibility of the growth increments.

IX. Season of death determination with cementum analysis Researchers have correlated growth layers in cementum with seasonal growth of many mammalian species. As a result, it is possible to deduce the season of death for archaeological samples. Under transmitted polarized light, the bands that appear opaque are usually deposited in most animals during the season of reduced growth; in contrast the bands that appear translucent are most often deposited during the growth season. The establishment of a control sample will determine in which season these different bands of acellular cementum are formed, from which the season of death of an individual can be determined through an examination of the nature of the outermost band (Grue and Jensen 1979; Lieberman 1993a, b).

Once the structures were clearly visible and readable under the microscope, a thin layer of resin/hardener was applied. This prevents the specimen from drying out and lifting from the slide. The appropriate amount the Epoxide resin/hardener is mixed and a thin layer spread on the slide and left to harden for at least twenty four hours.

X. Conclusion The final stage of the thin sectioning process was to remove as much of this final layer of resin without exposing and further abrading the specimen. The 320 micron abrasive paper was used to carefully remove some of the top coat of resin. The 400 micron and finally the 600 micron abrasive papers were used in turn to remove

The tooth wear and eruption data has only been analyzed in a coarse manner (combined with the post-cranial material) in previous research. As a potentially informative source of information for this question, it deserves more significant and precise attention. By 39

METHODOLOGY analyzing it separately, a new dimension of the age-ofdeath of domestic animals can be estimated, and information on production strategies and seasonality inferred. While providing additional age estimation, the cementum analysis will also supply season of death information, an important element in the consideration of the transhumant movement of herds.

The two zooarchaeological techniques chosen here utilize the most relevant source of information on the transhumant movement of domestic herds, that is, the remains of the animals of interest. It is hoped that the combination of both techniques will provide a complete picture of the economic movement of these animals and may minimize some of the problems of interpretation associated with each technique.

40

CHAPTER 7 DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES I. Introduction

leaving forty-five mandibles and three loose teeth in the final analysis (Appendix A). Table 7.4 summarizes the age stage distributions of Ovis aries remains from the Early Neolithic. Both the Eneolithic and Early Iron Age samples of Ovis aries consisted of only one mandible. Each was identifiable to an age stage.

Primary zooarchaeological analysis including species identification from all sites has been completed by Greenfield and have been described in detail elsewhere (1984, 1986a, b, 1994, 1996, 2005b, in press b, n.d. a, b, c, and d; Greenfield and Fowler 2003, 2005). In this chapter, the samples will be described in terms of the site, surrounding environment, period(s) of occupation, and the size and nature of the sample selected for this investigation. Only identified domestic taxa mandibles and loose teeth will be considered in the present analysis. The sites are discussed in alphabetical order, with the remains in each site discussed by time period. Several samples were too small for the types of analyses for consideration here and these are detailed in Appendix B. A description of the remains and the final sample size for each species and period within each site is summarized below.

A total of 299 domestic Ovis/Capra mandibular remains (mandibles – n=233 and loose teeth – n=66) were recovered from Early Neolithic deposits at Blagotin. Most mandibles (n=191) and many loose teeth (n=19) were unidentifiable to age, leaving forty-two mandibles and forty-seven loose teeth in the final analysis (Appendix A). Table 7.5 summarizes the age stage distributions of Ovis/Capra remains from the Early Neolithic. As the Eneolithic sample of Ovis aries was too small for effective analysis, it was lumped into the Ovis/Capra category to increase sample size. The Ovis/Capra sample probably is composed of only Ovis aries since no goat remains were found. As a result, nine Ovis/Capra mandibles and loose teeth were recovered from Eneolithic levels of Blagotin (three mandibles, one of Ovis aries and six loose teeth). Three mandibles were unidentifiable to age, leaving six loose teeth in the final analysis (Appendix A). This sample is too small to be included in the analysis. The age stage distributions of Ovis/Capra remains from the Eneolithic are summarized in Appendix B (Table B1).

II. Blagotin A. Site description Blagotin, a site in central Serbia (Figure 1.1), is located at the base of the Blagotin Mountain (250 m asl) at the headwaters of the Blagotin stream. It is a mid-altitude site set amidst rolling hills and low mountains, and surrounded by mixed oak forests and rich agricultural areas. The site is on a gentle slope above the stream and extends onto the flatter terrace above the slope. The slope has contributed to severe erosion in some areas of the site (Greenfield 2000; Greenfield and Jongsma in press b, c).

The Early Iron Age sample of Ovis/Capra included a single Ovis aries mandible from this period as well. Eighteen domestic Ovis/Capra mandibular elements were recovered from Early Iron Age levels at Blagotin (9 mandibles and 9 loose teeth). Five of the mandibles and two of the loose teeth were unidentifiable to age, leaving four mandibles and seven loose teeth in the final analysis (Appendix A). Table 7.6 summarizes the age stage distributions of Ovis/Capra remains from the Early Iron Age.

The site has three periods of occupation, Early Neolithic, Eneolithic and Early Iron Age. It is relatively small, c. 100x80 m during the Early Neolithic, 100x50 m in the Eneolithic, and 200x200 m during the Early Iron Age. The deposits range up to a depth of 3 m at the center of the site, with vertically superimposed remains. There was little laterally displaced stratigraphy at the site. Approximately 50% of the site was wet-sieved and floated. Over 35,000 animal bone fragments were recovered and analyzed from the 1989-1994 excavations at the site (Greenfield and Jongsma n.d.).

A total of 281 Bos taurus mandibles and loose teeth were recovered from Early Neolithic deposits at Blagotin (232 mandibles and 49 loose teeth). There were 194 mandibles and nine loose teeth unidentifiable to age, leaving thirtyeight mandibles and forty loose teeth in the final analysis (n=78 – Appendix A). Table 7.7 summarizes the age stage distributions of Bos taurus remains for the Early Neolithic.

B. Mandibular and loose tooth remains Eighty-one mandibles and four loose teeth from domestic Ovis aries were recovered from Early Neolithic levels at Blagotin. Thirty-six mandibles and one loose tooth were unidentifiable to an absolute age stage (due to absence of or damage to appropriate teeth),

Nineteen Bos taurus mandibular elements were recovered from Eneolithic levels at Blagotin (8 mandibles and 11 loose teeth). Five mandibles and two loose teeth were

41

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.4. Stage distribution of Ovis aries mandibles and loose teeth from Early Neolithic Blagotin Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 6 3 19 4 1 2 2 0 0 37

Corrected Count % 16 9 51 11 3 5 5 0 0 100

No. 6 3 21.8 6.4 1.9 2.9 6 0 0 48

% 12 7 45 13 5 6 12 0 0 100

Table 7.5. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Neolithic Blagotin Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 4 1 10 7 1 1 1 3 0 28

Corrected Count % 14 4 35 25 4 4 4 10 0 100

No. 4 1.1 13.9 26 6.6 11.6 9 16.8 0 89

% 4 1 16 30 7 13 10 19 0 100

Table 7.6. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Iron Age Blagotin Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 1 2 0 0 1 0 0 4 42

Corrected Count % 0 0 25 50 0 0 25 0 0 100

No. 0 0 1.6 4.4 0 0 5 0 0 11

% 0 0 15 40 0 0 45 0 0 100

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

Table 7.7. Stage distribution of Bos taurus mandibles and loose teeth from Early Neolithic Blagotin Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult old adult Senile

No. 9 3 12 15 2 4 0 1 1 47

Corrected Count % 20 6 26 32 4 8 0 2 2 100

No. 9 3 12.4 22.9 3.9 12.8 0 7.2 6.3 77.5

% 12 4 16 29 5 17 0 9 8 100

Table 7.8. Stage distribution of Bos taurus mandibles and loose teeth from Eneolithic Blagotin Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult old adult Senile

No. 0 0 2 3 2 1 0 0 0 8

Corrected Count % 0 0 25 38 25 12 0 0 0 100

unidentifiable to age, leaving three mandibles and nine loose teeth in the final sample (n=12 – Appendix A). Table 7.8 summarizes the age stage distributions of Bos taurus remains from the Eneolithic.

No. 0 0 2 4.8 3.2 2 0 0 0 12

% 0 0 17 40 26 17 0 0 0 100

unidentifiable to age, leaving no Eneolithic Sus sample. In the Early Iron Age levels only one Sus scrofa mandible was found. It is identifiable to age and the final sample size is one. This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Tables B3 and B4).

Twenty-two Bos taurus mandibular elements were recovered from Early Iron Age deposits at Blagotin (15 mandibles and 7 loose teeth). Thirteen mandibles and two loose teeth were unidentifiable to age, leaving two mandibles and five loose teeth in the final analysis (n=7 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B2).

III. Foeni-Salaş A. Site description

Foeni-Salaş is a lowland site located in the western edge of the Romanian Banat (Figure 1.1) near the village of Foeni (approximately 3 km away from the Serbian border). The site is in the midst of the flat alluvial plain of the Banat. It is located on the bank of the Timişat, a tributary of the Timiş River. This is an area distinguished by low altitude (c. 80 m asl) and low relief (Greenfield n.d. d; Greenfield and Draşovean 1994; Greenfield and Jongsma in press a).

All of the domestic pig remains were too few to be included in the analysis. Nine mandibles were recovered from Early Neolithic levels at Blagotin. Seven were unidentifiable to age, leaving two mandibles in the final analysis (Appendix A). One Sus scrofa mandible was recovered from the Eneolithic levels at Blagotin. It was 43

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Two features of this area, a high water table and the gentle slope of the land, combine to contribute to frequent flooding of the rivers and the formation of swamps. The area was drained in the 19th century. Forests existed along the river edge, but these are no longer in existence. Most of the area is presently used for cultivation, but the soil tends to be very heavy. It was likely not extensively cultivated before the advent of the plough (Greenfield and Draşovean 1994).

Thirteen Ovis/Capra mandibles and loose teeth were recovered from Early Iron Age levels at Foeni-Salaş (11 mandibles and 2 loose teeth). Four mandibles were unidentifiable to age, leaving seven mandibles and two loose teeth in the final analysis (n=9 – Appendix A). Table 7.11 summarizes the age stage distributions of Ovis/Capra remains from the Early Iron Age. Forty-two Bos taurus mandibular remains (mandibles – n=34 and loose teeth – n=8) were recovered from Early Neolithic deposits at Foeni-Salaş. Twenty-four mandibles and one loose tooth were unidentifiable to age, leaving ten mandibles and seven loose teeth in the final analysis (n = 17 – Appendix A). Table 7.12 summarizes the age stage distributions of Bos taurus remains from the Early Neolithic.

There are at least six periods of occupation at the site, the Early Neolithic, Eneolithic, Middle Bronze Age, Early Iron Age, Dacian (later Roman), and Medieval (12th century) periods. For the most part, only the Early Neolithic and Early Iron Age periods will be considered here. Few or no relevant bone (mandibles and teeth) remains could be assigned to the Eneolithic or Middle Bronze Age deposits. The site is found on a low natural mound, part of an ancient alluvial terrace. It is small, roughly half a hectare and consists of a single, thin Early Neolithic occupation horizon and discrete features from the Early Neolithic and other periods (Greenfield and Draşovean 1994; Greenfield and Jongsma in press a). The sample was almost entirely dry and wet sieved (c. 85%). Over 10,000 fragments were recovered and analyzed from the 1992-94 excavations at the site (Greenfield n.d. d).

Eleven Bos taurus mandibular remains (mandibles – n=6 and loose teeth – n=5) were recovered from Early Iron Age levels at Foeni-Salaş. Four mandibles and two loose teeth were unidentifiable to age, leaving two mandibles and three loose teeth in the final analysis (n=5 – Appendix A). These data were too small to be included in the analysis. Appendix B (Table B6) summarizes the age stage distributions of Bos taurus remains from the Early Iron Age.

B. Mandibular and loose tooth remains Eight Sus scrofa mandibles and loose teeth were recovered from Early Neolithic levels at Foeni-Salaş. Two were identified as belonging to wild species (Sus scrofa fer.) and were eliminated from the sample. The total domestic sample size was six (4 mandibles and 2 loose teeth). Three mandibles and one loose tooth were unidentifiable to age, leaving one mandible and one loose tooth in the final analysis (n=2 – Appendix A). The Sus scrofa sample from Middle Bronze Age Foeni-Salaş includes a single mandible from a domestic pig. It was unidentifiable to age, leaving no Middle Bronze Age sample. There are no Early Iron Age Sus scrofa remains recovered from Foeni-Salaş. The age stage distribution data for the Early Neolithic are described in Appendix B (Table B7).

Twelve Ovis aries mandibles were recovered from Early Neolithic levels at Foeni-Salaş. Two mandibles were indeterminate to a specific age, leaving ten mandibles in the final analysis (Appendix A). Table 7.9 summarizes the age stage distributions of Ovis aries remains from the Early Neolithic. Four Capra hircus mandibles were recovered from Early Neolithic deposits of Foeni-Salaş. Two mandibles were unidentifiable to age, making two the final sample size. In the Early Iron Age levels, there was only one Capra hircus mandible. As both the Ovis aries and the Capra hircus samples were small, they were lumped into the category Ovis/Capra for analysis (below).

IV. Kadica Brdo

Ninety-three Ovis/Capra mandibular remains (mandibles – n=76 and loose teeth – n=17) were recovered from Early Neolithic levels at Foeni-Salaş. Forty mandibles and three loose teeth were unidentifiable to age, leaving thirty-six mandibles and fourteen loose teeth in the final sample (n=50 – Appendix A). Table 7.10 summarizes the age stage distributions of Ovis/Capra remains from the Early Neolithic.

A. Site description Kadica Brdo is located on a hilltop overlooking the western end of the Glasinac plateau in eastern Bosnia (Figure 1.1). It overlooks and is close (5 km) to the town of Knežina. Glasinac is the largest highland plateau in southeast Europe at this elevation. It is found at elevations mostly above 1000 m asl, with surrounding mountain heights extending up to 2000 m asl. Extensive sub-alpine grasslands and coniferous forests cover the plateau. Today, the area is known mainly for grazing and forestry as it is not an environment suitable for extensive agriculture. Pockets of small-scale agriculture are located in depressions within the basin (Greenfield 2005b).

Five Ovis aries mandibles were recovered from Early Iron Age levels at Foeni-Salaş. All were identifiable to age using tooth wear and eruption, making the final sample size five (Appendix A). The age stage distribution data are summarized in Appendix B (Table B5). Since these data were too small to be used on their own in the analysis, this sample was combined with the Ovis/Capra remains for the analysis (below). 44

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.9. Stage distribution of Ovis aries mandibles and loose teeth from Early Neolithic Foeni-Salaş Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 1 7 1 0 0 0 0 0 9

Corrected Count % % 0 11 78 11 0 0 0 0 0 100

0 1 7.9 1.3 0 0 0 0 0 10.2

0 10 80 10 0 0 0 0 0 100

Table 7.10. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Neolithic Foeni-Salaş Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 1 2 12 3 0 2 3 0 0 23

Corrected Count % 4 9 52 13 0 9 13 0 0 100

No. 1.5 2.1 16.5 12.6 0 6 11.7 0 0 50.4

% 3 4 33 25 0 12 23 0 0 100

Table 7.11. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Iron Age Foeni-Salaş Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 4 1 0 0 1 1 0 7

Corrected Count % 0 0 58 14 0 0 14 14 0 100

45

No. 0 0 4 1 0 0 2.5 1.5 0 9

% 0 0 44 11 0 0 28 17 0 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.12. Stage distribution of Bos taurus mandibles and loose teeth from Early Neolithic Foeni-Salaş Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months young adult adult old adult senile

No. 3 1 3 2 0 2 0 0 1 12

The site at Kadica Brdo was occupied during the Eneolithic, Early Iron Age, and Late Iron Age. Only the faunal remains from the Early Iron Age levels derive from secure temporal contexts. As a result, only these will be considered in this investigation.

Corrected Count % 25 8 25 17 0 17 0 0 8 100

No. 3 1 3 2 0 5 0 0 3 17

% 18 6 18 12 0 28 0 0 18 100

Ovis/Capra remains from the Early Iron Age. Sixty-six Bos taurus mandibular remains (mandible – n=36 and loose teeth – n=30) were recovered from Early Iron Age levels at Kadica Brdo. Eighteen mandibles and one loose tooth were undeterminable to age, leaving eighteen mandibles and twenty-nine loose teeth in the final analysis (n=47 – Appendix A). Table 7.15 summarizes the age stage distributions of Bos taurus remains from the Early Iron Age.

Kadica Brdo is a small site (100x50 m), surrounded by a thick fortification wall, which preserved the internal stratigraphy at the site. Only one trench was dry sieved, but selected samples across the site were also floated. Approximately 10% of the spatial contexts were systematically recovered. Over 20,000 fragments were recovered and analyzed from the 1980, and 1987-89 excavations at the site (Greenfield 2005b).

Seventy-two Sus scrofa mandibles and loose teeth were recovered from Early Iron Age levels at Kadica Brdo (65 mandibles and 7 loose teeth). Twenty-three mandibles were unidentifiable to age, leaving forty-two mandibles and seven loose teeth in the final analysis (n=49 – Appendix A). Table 7.16 summarizes the age stage distributions of Sus scrofa remains from the Early Iron Age.

B. Mandibular and loose tooth remains Fifty-seven Ovis aries mandibular remains (mandibles n=49 and loose teeth – n=8) were recovered from Early Iron Age deposits at Kadica Brdo. Ten mandibles were unidentifiable to age, leaving thirty-nine mandibles and eight loose teeth in the final analysis (n=47 – Appendix A). Table 7.13 summarizes the age stage distributions of Ovis aries remains from the Early Iron Age.

V. Livade A. Site description Livade is a lowland site located on the right bank of the Danube in northeastern Serbia at the western edge of the Dacian Plain (Figure 1.1). On the left bank of the river, the broad Dacian plain begins. The site is located in the wide alluvial plain that borders the Danube, just beyond the eastern edge of the Iron Gates, on the eastern side of the mountains of eastern Serbia. The mountains of East Serbia are less than 5 km distant. It is near the modern village of Mala Vrbica, on the eastern edge of the area of investigation.

Eleven Capra hircus mandibular remains (mandibles n=10 and loose teeth – n=1) were recovered from Early Iron Age deposits at Kadica Brdo. Three mandibles were unidentifiable to age, leaving seven mandibles and one loose tooth in the final analysis (n=8). The age stage distribution data are summarized in Appendix B (Table B8). Since these data were too small to be used on their own in the analysis, this sample was combined with the Ovis/Capra remains for the analysis (below). There were a total of 254 mandibular remains (mandible – n=153 and loose teeth – n=101) of Ovis/Capra recovered from Early Iron Age levels at Kadica Brdo. Forty-two mandibles and four loose teeth were undeterminable to age, leaving 111 mandibles and 97 loose teeth in the final analysis (n=208 – Appendix A). Table 7.14 summarizes the age stage distributions of

The site contains ceramic and other artifactual remains with strong cultural affinities to the other sites in the study region. The ceramic assemblage and figurines place Livade in the Dubovac-Žuto Brdo ceramic group. Some ceramic elements also reveal affiliation with the Verbicioara ceramic group across the river in Romanian Oltenia. 46

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.13. Stage distribution of Ovis aries mandibles and loose teeth from Early Iron Age Kadica Brdo Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 4 5 20 5 1 0 1 0 0 36

Corrected Count % 11 14 55 14 3 0 3 0 0 100

No. 4 5 25.6 9 1.4 0 2 0 0 47

% 8 10 54 20 4 0 4 0 0 100

Table 7.14. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Iron Age Kadica Brdo Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 1 1 15 10 5 16 17 3 3 71

Corrected Count % 1 1 22 14 7 23 24 4 4 100

No. 1 1 25.5 40.5 19.2 69.8 39.8 5.6 5.6 208

% 1 1 12 19 9 34 18 3 3 100

Table 7.15. Stage distribution of Bos taurus mandibles and loose teeth from Early Iron Age Kadica Brdo Stage

Suggested Age

Raw Count

Corrected Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult old adult

No. 4 0 3 8 0 2 1 1 6

% 16 0 12 32 0 8 4 4 24

No. 4 0 3 14.5 0 7.1 3.8 3.3 11.2

% 8 0 6 32 0 15 8 7 24

25

100

46.9

100

Senile

47

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.16. Stage distribution of Sus scrofa mandibles and loose teeth from Early Iron Age Kadica Brdo Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months adult old adult senile

No. 2 11 4 7 6 0 0 0 1 31

Corrected Count % 6 36 13 23 19 0 0 0 3 100

As a result, Livade is considered to be a Late Bronze Age site. Chronologically, the Dubovac-Žuto Brdo horizon falls within the Middle-Late Bronze ages of the central Balkans (Reinecke – Br A2/B1 to D – ca. 1600-1200 bc) (Greenfield 1986a).

No. 2 11.7 10 17.7 6.5 0 0 0 1 48.9

% 4 25 20 36 13 0 0 0 2 100

and one loose tooth in the final analysis (n=27 – Appendix A). Table 7.19 summarizes the age stage distributions of Sus scrofa remains from the Late Bronze Age. VI. Ljuljaci

The size of Livade is difficult to determine because much of the site was buried beneath a thick sandy horizon (from river flooding). It extends for a distance of up to 100 m from the riverbank, and was investigated over a length of 200 m along the river. The sandy horizon preserved the site so that there was little mixing of remains in the cultural horizon (contrary to contemporary sites that are usually disturbed by plowing). None of the sediments from the site was sieved or floated. Over 5,000 fragments were recovered and analyzed from the 1980-81 excavations at the site (Greenfield 1986a).

A. Site description Ljuljaci is found on the top of the hill, Milica Brdo, in the village of Ljuljaci, about 20 km west of the city of Kragujevac, in Serbia (Figure 1.1). It is located on the western half of a natural plateau overlooking the surrounding valleys at approximately 300 m asl and is considered to be a mid-altitude site. Three sides of the plateau are steep, leaving the eastern side as the only approach to the site. It is considered to be a fortified position due to its less accessible location and the possible palisade and ditch along one side. A small stream, the Glavica, winds itself around the base of the plateau on the western, northern and eastern sides. The plateau area was subject to deforestation and cultivation during its period of occupation. The surrounding environment is one of rolling hills and low mountains covered by mixed oak forests and some agricultural cultivation (Greenfield 1986a, 2006).

B. Mandibular and loose tooth remains Sixteen domestic Ovis/Capra mandibles (n=15) and loose teeth (n=1) were recovered from Late Bronze Age levels at Livade. This included a single mandible of Ovis aries. Four mandibles and one loose tooth were unidentifiable to age, leaving eleven mandibles in the final analysis (Appendix A). Table 7.17 summarizes the age stage distributions of Ovis/Capra remains from the Late Bronze Age.

Ljuljaci includes material from the late Early Bronze Age and the Middle Bronze Age. The stratigraphy of the site is relatively simple consisting of three layers. Dates are based on ceramic groups, and cross-dating with other sites.

Twenty-one Bos taurus mandibular remains (mandibles – n=14 and loose teeth – n=7) were recovered from Late Bronze Age levels at Livade. Four mandibles and one loose tooth were undeterminable to age, leaving ten mandibles and six loose teeth in the final analysis (n=16 – Appendix A). Table 7.18 summarizes the age stage distributions of Bos taurus remains from the Late Bronze Age.

1. 2. 3.

Forty-two Sus scrofa mandibular elements (mandibles – n=41 and loose teeth – n=1) were recovered from Late Bronze Age levels at Livade. Fifteen mandibles were undeterminable to age, leaving twenty-six mandibles

Ljuljaci I – Early Bronze Age (ca. 1950 bc, uncalibrated) Ljuljaci II – late Early Bronze Age- early Middle Bronze Age (1730-1690 bc) Ljuljaci III – Middle Bronze Age (ca. 1600-1550 bc) (Greenfield 1986a: 125).

The site is relatively small, 100x50 m. In the early periods (Ljuljaci I and II), the settlement remained at about the same size. It expanded in size in the final period

48

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.17. Stage distribution of Ovis/Capra mandibles from Late Bronze Age Livade Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 0 1 0 2 3 2 0 8

Corrected Count % 0 0 0 12 0 25 38 25 0 100

No. 0 0 0 1 0 2.8 5.01 2.19 0 11

% 0 0 0 9 0 25 46 20 0 100

Table 7.18. Stage distribution of Bos taurus mandibles from Late Bronze Age levels of Livade Stage

Suggested Age

Raw Count

Corrected Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult Old adult

No. 0 0 1 1 0 0 0 1 2

% 0 0 20 20 0 0 0 20 40

No. 0 0 1 3.86 0 1.3 1.58 3.19 5.01

% 0 0 6 24 0 8 10 21 31

5

100

15.94

100

Senile

Table 7.19. Stage distribution of Sus scrofa mandibles and loose teeth from Late Bronze Age Livade Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months Adult Old adult Senile

No. 0 1 2 7 7 1 0 0 0 18

49

Corrected Count % 0 6 10 39 39 6 0 0 0 100

No. 0 2 4 9.5 10.38 1.12 0 0 0 27

% 0 8 15 35 38 4 0 0 0 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES (Ljuljaci III). The stratigraphy was relatively thin, with a maximum depth of c. 1 m beneath the surface (except for the occasional pit). None of the site was sieved or floated. Over 5,000 fragments were recovered and analyzed from the 1975-77 excavations at the site (Greenfield 1986a).

VII. Megalo Nisi Galanis A. Site description Megalo Nisi Galanis is in the southern Ptolemais basin, located in Greek Macedonia near the city of Kozani. The Ptolemais basin is the southernmost extension of the Pelagonian plain stretching from Greek Macedonia into the Former Yugoslav Republic of Macedonia (Figure 1.1). The plateau is surrounded by mountainous height. The site is considered a mid-altitude site (c. 500 m asl) in terms of regional long distance transhumant pastoralists. It is a small site, with thick deposits, ranging from a depth of 23m, covering an area of c. 100 x 100 m. The entire excavated deposit was dry and wet sieved. The site is located on a paleo-lake drained in the 19th century, and falls within the Mediterranean climatic zone. Surrounding heights are currently used for grazing, the flats of the plateau for agriculture. The excavation uncovered material from the Late Neolithic, Final Neolithic (=Eneolithic of central Balkans), and Early Bronze Age. Underlying Middle and Final Neolithic deposits exist but are not yet analyzed (Greenfield and Fowler 2003, 2005).

B. Mandibular and loose tooth remains Five Ovis/Capra mandibles and loose teeth were recovered from Early Bronze Age levels at Ljuljaci (4 mandibles and 1 loose tooth). Two mandibles were unidentifiable to age, leaving two mandibles and one loose tooth in the final analysis (n=3 – Appendix A). Since these data were too few to be used on their own, they were combined with the rest of the Bronze Age material and designated Early/Middle Bronze Age. There were a total of ten Ovis/Capra mandibles and loose teeth from the site (9 mandibles and 1 loose tooth); including one mandible identified as Ovis aries. Only one mandible was unidentifiable to age, that of the Ovis aries specimen. Therefore, only eight mandibles and one loose tooth were included in the final analysis (n=9). Lumping of the data still results in too small a sample size for inclusion in the analysis. Age stage distribution data are summarized in Appendix B (Tables B9-B11).

Though out of the geographical area of study, Megalo Nisi Galanis is included for two main reasons; 1) it is a large and systematically collected and analyzed faunal sample from western Macedonia spanning the critical Late Neolithic-Post Neolithic temporal divide that is the focus of this investigation; 2) its inclusion in the sample provides a connection with the southern Balkans. Over 20,000 fragments were recovered and analyzed from the 1980-1986 excavations at the site (Greenfield and Fowler 2003, 2005).

Seven Bos taurus mandibles were recovered from the Early Bronze Age levels at Ljuljaci. Five mandibles were unidentifiable to age, leaving two mandibles in the final analysis (Appendix A). Since these data were too few to be used by themselves, all of the Bronze Age material from the site was combined and designated Early/Middle Bronze Age. Age stage distribution data for the Early Bronze Age are summarized in Appendix B (Table B12). To increase sample size, the Middle Bronze Age assemblage was combined with Early Bronze Age material. There were a total of twenty-four Bos taurus mandibles (n=16) and teeth (n=8). Eleven mandibles and one loose tooth were unidentifiable to age, leaving five mandibles and seven loose teeth in the final analysis (n=12 – Appendix A). Table 7.20 summarizes the age stage distributions of Bos taurus remains from the Early/Middle Bronze Age.

B. Mandibular and loose tooth remains Three Ovis aries mandibles were recovered from Late Neolithic/Final Neolithic levels at Megalo Nisi Galanis. All were identifiable to age, and all were included in the final analysis (n=3). Table 7.22 summarizes the age stage distributions of Ovis aries remains from the Late Neolithic/Final Neolithic. Fourteen domestic Ovis/Capra mandibles were recovered from Late Neolithic/Final Neolithic levels at Megalo Nisi Galanis. Five were unidentifiable to age, leaving nine mandibles in the final analysis (n=9 – Appendix A). Table 7.23 summarizes the age stage distributions of Ovis/Capra remains from the Late Neolithic/Final Neolithic.

There were five ageable remains recovered from Early Bronze Age deposits and six ageable remains recovered from Middle Bronze Age deposits and age stage distributions are detailed in Appendix B (Tables B13 and B14). To increase sample size, all Bronze Age material was combined for Sus scrofa. In total, there were forty-eight domestic Sus scrofa mandibles (n=46) and loose teeth (n=2) from the Bronze Age deposits. Twenty-eight of the mandibles were unidentifiable to age, leaving eighteen mandibles and two loose teeth in the final sample (n=20 – Appendix A). Table 7.21 summarizes the age stage distributions of Sus scrofa remains from the Early/Middle Bronze Age.

Nine Ovis aries mandibles and loose teeth were recovered from Final Neolithic levels at Megalo Nisi Galanis (7 mandibles and 2 loose teeth). Four mandibles could not be assigned an age, leaving three mandibles and two loose teeth in the final analysis (n=5). Table 7.24 summarizes the age stage distributions of Ovis aries remains from the Final Neolithic.

50

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.20. Stage distribution of Bos taurus mandibles from Early/ Middle Bronze Age Ljuljaci Stage

A B C D E F G H I

Suggested Age

Raw Count

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult Old adult Senile

No. 0 2 2 2 0 0 0 0 1 7

Corrected Count % 0 28 28 28 0 0 0 0 15 99

No. 0 2 2 2 0 0.9 2.6 0 2.9 12.4

% 0 16 16 16 0 7 22 0 23 100

Table 7.21. Stage distribution of Sus scrofa mandibles and loose teeth from Early/Middle Bronze Age Ljuljaci Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months Adult

No. 0 1 1 5 5 1 0 0

% 0 8 8 38 38 8 0 0

No. 0 1 1.2 7.3 9 1.5 0 0

% 0 5 6 37 45 7 0 0

0 13

0 100

0 20

0 100

Old adult Senile

Corrected Count

Table 7.22. Stage distribution of Ovis aries mandibles and loose teeth from Late Neolithic/Final Neolithic Megalo Nisi Galanis Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 1 0 0 0 0 0 0 1

Corrected Count % 0 0 100 0 0 0 0 0 0 100

51

No. 0 0 2 0.5 0.5 0 0 0 0 3

% 0 0 66 17 17 0 0 0 0 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.23. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic/Final Neolithic Megalo Nisi Galanis Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 2 0 0 0 0 0 0 2

Corrected Count % 0 0 100 0 0 0 0 0 0 100

No. 0 0 3 0.5 1 1.9 1.3 0.6 0.6 8.9

% 0 0 33 6 11 21 15 7 7 100

Table 7.24. Stage distribution of Ovis aries mandibles and loose teeth from Final Neolithic Megalo Nisi Galanis Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 1 0 2 0 0 0 0 0 0 3

Corrected Count % 33 0 67 0 0 0 0 0 0 100

No. 1 0 4 0 0 0 0 0 0 5

% 20 0 80 0 0 0 0 0 0 100

Table 7.25. Stage distribution of Ovis/Capra mandibles and loose teeth from Final Neolithic Megalo Nisi Galanis Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 1 0 5 3 2 1 1 0 0 13

Corrected Count % 8 0 38 23 15 8 8 0 0 100

52

No. 1 0 8.8 11.7 8.6 5.8 8.1 0 0 44

% 2 0 21 26 20 13 18 0 0 100

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B17). There were no Sus species remains for the Final Neolithic/Early Bronze Age period.

Fifty-seven Ovis/Capra mandibles and loose teeth were recovered from Final Neolithic levels at Megalo Nisi Galanis (24 mandibles and 33 loose teeth). Twelve mandibles and one loose tooth were unidentifiable to age, leaving twelve mandibles and thirty-two loose teeth in the final analysis (n=44 – Appendix A). Table 7.25 summarizes the age stage distributions of Ovis/Capra remains from the Final Neolithic.

VIII. Novačka Ćuprija A. Site description Novačka Ćuprija is a lowland site that is located roughly 5 km north of the city of Smederevska Palanka in central Serbia (Figure 1.1). It is located on the western edge of a low plateau east of a small tributary of the Jasenica River. The site is located in a non-obtrusive position, offering no significant defensive advantage. In close proximity are suitable agricultural land and access to water and wood supplies (Greenfield 1986a).

Six Ovis aries mandibles and loose teeth were recovered from Final Neolithic/Early Bronze Age levels at Megalo Nisi Galanis (5 mandibles and 1 loose tooth). One mandible was unidentifiable to age, leaving four mandibles and one loose tooth in the final analysis (n=5 – Appendix A). Table 7.26 summarizes the age stage distributions of Ovis aries remains from the Final Neolithic/Early Bronze Age. As these samples were too small for effective analysis, they were combined with the Ovis/Capra remains.

The site is divided into four areas, of which only the data from the central area is relevant and utilized. This area is approximately four hectares in area. The plateau on which the site is situated is generally flat. It increases in width and rises slightly to the east (Greenfield 1986a).

Eighteen Ovis/Capra mandibles and loose teeth were recovered from Final Neolithic/Early Bronze Age levels at Megalo Nisi Galanis (8 mandibles and 10 loose teeth). Four mandibles and one loose tooth were unidentifiable to age, leaving four mandibles and nine loose teeth in the final analysis (n=13 – Appendix A). Table 7.27 summarizes the age stage distributions of Ovis/Capra remains from the Final Neolithic/Early Bronze Age.

Novačka Ćuprija is a lowland site that includes material from the Eneolithic through to the Roman period, although there is no evidence for occupational continuity over this entire time span. The most intensive occupation took place during the Early Bronze Age (Greenfield 1986a).

There were insufficient Late Neolithic/Final Neolithic remains of Bos taurus from Megalo Nisi Galanis to include in the analysis. Only two mandibles were recovered, both unidentifiable to age (Appendix A).

The prehistoric settlements at Novačka Ćuprija represents the remains of a series of small villages that laterally moved up and down the length of the terrace overlooking the stream. Each settlement probably represented no more than 100x100 m in size. Because the site has moved laterally over time, the total length of the site is almost 1 km. The deposits were relatively thin since there was little vertical deposition. Most of the material above the pit levels were disturbed and mixed by modern and ancient plowing. The soils from the entire excavation were sieved and/or floated. Over 5,000 bone fragments were recovered and analyzed from the 1977 and 1980 excavations at the site (Greenfield 1986a, 1986b).

Ten Bos taurus mandibular remains (mandibles – n=4 and loose teeth – n=6) were recovered from Final Neolithic levels at Megalo Nisi Galanis. Two mandibles were unidentifiable to age, leaving two mandibles and six loose teeth in the final analysis (n=8 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B15) Two Bos taurus mandibles and loose teeth were recovered from Final Neolithic/Early Bronze Age levels at Megalo Nisi Galanis (1 mandible and 1 loose tooth), both are unidentifiable to age. As a result, there was no sample included in the analysis (Appendix A).

B. Mandibular and loose tooth remains Ten domestic Ovis/Capra loose teeth were recovered from Eneolithic levels at Novačka Ćuprija. All the teeth were assigned an absolute age and were included in the final analysis (Appendix A). Table 7.28 summarizes the age stage distributions of Ovis/Capra remains from the Eneolithic.

Two Sus scrofa mandibles were recovered from Late Neolithic/Final Neolithic levels at Megalo Nisi Galanis. Only one was identifiable to age. As a result, the sample size is one. This sample is too small for inclusion in the analysis. Stage distribution data is included in Appendix B (Table B16).

Thirty-six Ovis/Capra mandibular remains (mandibles n=12 and loose teeth – n=24) were recovered from Early Bronze Age levels at Novačka Ćuprija. Four mandibles were unidentifiable to age, leaving eight mandibles and twenty-four loose teeth in the final analysis (n=32 – Appendix A). Table 7.29 summarizes the age stage

Nine Sus scrofa mandibular remains (mandibles - n=6 and loose teeth – n=2) were recovered from Final Neolithic levels at Megalo Nisi Galanis. Two mandibles were unidentifiable to age, leaving four mandibles and two loose teeth in the final analysis (n=6 – Appendix A). This sample 53

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.26. Stage distribution of Ovis aries mandibles and loose teeth from Final Neolithic/Early Bronze Age Megalo Nisi Galanis Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 1 2 2 0 0 0 0 0 5

% 0 20 40 40 0 0 0 0 0 100

Table 27. Stage distribution of Ovis/Capra mandibles and loose teeth from Final Neolithic/Early Bronze Age Megalo Nisi Galanis Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 1 2 3 0 0 0 0 0 6

Corrected Count % 0 17 33 50 0 0 0 0 0 100

No. 0 1 2 4.7 0.6 1.2 1 1.5 1 13

% 0 8 15 35 5 9 8 12 8 100

Table 7.28. Stage distribution of Ovis/Capra mandibles and loose teeth from Eneolithic Novačka Ćuprija Stage A B C D E F G H I

Suggested Age 0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

Raw Count No. 0 0 0 1 0 3 0 0 1 5

54

% 0 0 0 20 0 60 0 0 20 100

Corrected Count No. 0 0 0 1.5 0 6.5 0 0 2 10

% 0 0 0 15 0 65 0 0 20 100

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.29. Stage distribution of Ovis/Capra mandibles and loose teeth from Early Bronze Age Novačka Ćuprija Stage A B C D E F G H I

Suggested Age 0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

Raw Count No. 0 1 1 8 1 3 1 0 2 17

% 0 6 6 46 6 18 6 0 12 100

Corrected Count No. 0 1.5 1.5 10.9 1.7 5 7.8 0 3.6 32

% 0 5 5 34 5 16 24 0 11 100

Table 7.30. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Bronze Age Novačka Ćuprija Stage A B C D E F G H I

Suggested Age 0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

Raw Count No. 0 0 1 1 1 1 0 0 0 4

distributions of Ovis/Capra remains from the Early Bronze Age.

% 0 0 25 25 25 25 0 0 0 100

Corrected Count No. % 0 0 0 0 1 6 1.6 10 3.6 23 3.6 23 4.5 28 1.6 10 0 0 15.9 100

Four Bos taurus remains (mandibles – n=1 and loose teeth – n=3) were recovered from Late Bronze Age levels at Novačka Ćuprija. All were assigned an absolute age and the final sample size was four. These samples were too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Tables B18B20).

Seventeen Ovis/Capra mandibular remains (mandibles— n=4 and loose teeth – n=13) were recovered from Late Bronze Age deposits at Novačka Ćuprija. One mandible was unidentifiable to age, leaving three mandibles and thirteen loose teeth in the final analysis (n=16 – Appendix A). Table 7.30 summarizes the age stage distributions of Ovis/Capra remains from the Late Bronze Age.

Nine Sus scrofa remains (mandibles – n=5 and loose teeth – n=4) were recovered from Eneolithic levels at Novačka Ćuprija. One mandible was unidentifiable to age, leaving four mandibles and four loose teeth in the final analysis (n=8 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B21).

An insufficient sample size of Bos taurus remains exists for Novačka Ćuprija for any period. Three Bos taurus mandibles were recovered from Eneolithic levels at Novačka Ćuprija. One mandible was unidentifiable to age, leaving two mandibles in the final analysis (Appendix A).

Sixteen Sus scrofa remains (mandibles – n=11 and loose teeth – n=5) were recovered from Early Bronze Age levels of Novačka Ćuprija. Five mandibles and one loose tooth were unidentifiable to age, leaving six mandibles and four loose teeth in the final analysis (n=10 – Appendix A). Table 7.31 summarizes the age stage distributions of Sus scrofa remains from the Early Bronze Age.

Nine Bos taurus mandibular remains (mandibles – n=5 and loose teeth – n=4) were recovered from Early Bronze Age levels at Novačka Ćuprija. Three mandibles were unidentifiable to age, leaving two mandibles and four loose teeth in the final analysis (n=6 – Appendix A).

55

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Four Sus scrofa remains (mandibles – n=3 and loose teeth – n=1) were recovered from Late Bronze Age deposits at Novačka Ćuprija. One mandible was unidentifiable to age, leaving two mandibles and one loose tooth in the final analysis (n=3 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B22).

n=19 and loose teeth – n=1) were recovered from Late Neolithic Opovo. Nine mandibles were undeterminable to age, leaving ten mandibles and one loose tooth in the final sample (n=11 – Appendix A). Table 7.33 summarizes the age stage distributions of Sus scrofa remains from the Late Neolithic deposits at Opovo.

IX. Opovo

A. Site description

A. Site description

Petnica is located in western Serbia, c. 5 km from the city of Valjevo. It is at the edge of the village of Petnica (Figure 1.1). The site is located on a slope overlooking a depression at the end of a valley. The site is situated at the base of a steep escarpment. Above (to the south) the site is a series of plateaus and ridges, beyond which the mountains of western Serbia rise. The highlands above and the slopes around the site are still heavily forested and present a dissected landscape of flat and sloping terrain interspersed with steeper inclines. Below the site, a small stream flows down to the Kolubara River (which is a tributary of the Sava). The valley is a rich agricultural area, but forests cover the ridges and mountains to the south of the site (Greenfield 1986a).

X. Petnica

Opovo is a Late Neolithic settlement in the Banat, a region of northern Serbia located between the Danube and Romania (Figure 1.1). Opovo is situated on a slightly elevated location and was surrounded on three sides by an extensive series of annually flooded lowland marshes. The area around the site is not well suited to either crop cultivation or the herding of domestic animals. Drainage is poor due to the high water table. However, a wide range of wild resources including fish, birds, shellfish, amphibians, reptiles, wild pig, roe and red deer was available in the aquatic and semi-aquatic environments of the surrounding habitat (Greenfield 1986a).

There were five major occupation horizons at the site. From the earliest to the latest, Middle Neolithic Vinča B, Late Neolithic Vinča C, Late Neolithic Vinča D, Eneolithic Baden-Kostalac, and late Bronze-Early Iron Age Halstatt A, B and C (Greenfield 1986a). Most of the Halstatt occupation is Halstatt A and therefore, most is considered to be Late Bronze Age.

Ceramic analysis indicated the presence of two phases of the Late Neolithic Vinča-Pločnik period (Tringham et al. 1985). The data analyzed here derive from the earlier Yugoslav excavations at the site and the phases were not separable stratigraphically at the time. As a result, the faunal remains were analyzed as a single temporal unit. None of the site was sieved or floated. Over 2,000 fragments were recovered from excavation of three trenches during the 1979-1980 excavations at the site. These were analyzed in 1982 (Greenfield 1986a). Subsequent excavations produced a large sample of remains that were unavailable for comparative analysis due to the revision of phasing (Russell 1998; Tringham et al. 1985, 1992).

Petnica is a small site, occupying roughly one hectare. The stratigraphy extends to depth of 2-3 m across much of the site. Approximately 20% of the site was sieved or floated (Greenfield 1986a). Over 20,000 fragments were recovered and analyzed from the 1980-1986 excavations at the site (Greenfield 1991, 1999b, n.d. c). B. Mandibular and loose tooth remains

B. Mandibular and loose tooth remains Nine Ovis/Capra mandibular remains (mandibles – n=6 and loose teeth – n=3) were recovered from Middle Neolithic levels at Petnica. Two mandibles were unidentifiable to age, leaving four mandibles and three loose teeth in the final analysis (n=7 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B24).

Six Ovis/Capra remains (mandibles – n=5 and loose teeth – n=1) were recovered from Late Neolithic levels at Opovo. All were identifiable to age (Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B23). Nineteen Bos taurus mandibular remains (mandibles – n=12 and loose teeth – n=7) were recovered from Late Neolithic levels at Opovo. Five mandibles were unidentifiable to age, leaving seven mandibles and seven loose teeth in the final analysis (n=14 – Appendix A). Table 7.32 summarizes the age stage distributions of Bos taurus remains from the Late Neolithic.

Twenty-one Ovis/Capra remains (mandibles – n=7 and loose teeth – n=14) were recovered from Late Neolithic levels at Petnica. Two mandibles were unidentifiable to age, leaving five mandibles and fourteen loose teeth in the final analysis (n=19 – Appendix A). Table 7.34 summarizes the age stage distributions of Ovis/Capra remains from the Late Neolithic.

Twenty Sus scrofa mandibular remains (mandibles – 56

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.31. Stage distribution of Sus scrofa mandibles and loose teeth from Early Bronze Age Novačka Ćuprija Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months Adult Old adult Senile

No. 0 0 0 3 4 0 0 0 0 7

Corrected Count % 0 0 0 43 57 0 0 0 0 100

No. 0 1 1 4 4 0 0 0 0 10

% 0 10 10 40 40 0 0 0 0 100

Table 7.32. Stage distribution of Bos taurus mandibles and loose teeth from Late Neolithic Opovo Stage

A B C D E F G H I

Suggested Age

Raw Count

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult Old adult Senile

No. 0 0 0 6 0 1 0 1 0 8

Corrected Count % 0 0 0 75 0 12 0 12 0 99

No. 0 0.5 0.5 7 0 4.01 0 1.99 0 14

% 0 4 4 50 0 28 0 14 0 100

Table 7.33. Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic Opovo Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months Adult Old adult Senile

No. 0 2 2 3 2 0 0 0 0 9

Corrected Count % 0 22 22 33 22 0 0 0 0 99

57

No. 0 2 2.4 3.6 3 0 0 0 0 11

% 0 18 22 33 27 0 0 0 0 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.34. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic Petnica Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 2 2 1 2 1 1 0 9

Corrected Count % 0 0 22 22 11 22 11 11 0 99

No. 0 0 3.5 4.2 1.9 4.8 3.2 1.5 0 19.1

% 0 0 18 22 10 25 17 8 0 100

Table 7.35. Stage distribution of Ovis/Capra mandibles and loose teeth from Eneolithic Petnica Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 4 2 1 1 1 0 0 9

Corrected Count % 0 0 45 22 11 11 11 0 0 100

Seven domestic Ovis/Capra mandibles and loose teeth were recovered from Late Neolithic/Eneolithic levels at Petnica (3 mandibles and 4 loose teeth); including one loose tooth of domestic Ovis aries. Only the Ovis aries loose tooth was undeterminable to age, leaving three mandibles and three loose teeth in the final analysis (n=6 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B25).

No. 0 0 6.5 3.3 1.1 1.1 1 0 0 13

% 0 0 50 26 8 8 8 0 0 100

As such, the final analysis included four mandibles and nine loose teeth (n=13 – Appendix A). Table 7.36 summarizes the age stage distributions of Ovis/Capra remains from the Late Bronze Age. Eleven Ovis/Capra mandibular remains (mandibles - n=4 and loose teeth – n=7) were recovered from Late Bronze Age/Early Iron Age deposits at Petnica. One mandible and one loose tooth were unidentifiable to age, leaving three mandibles and six loose teeth in the final analysis (n=9 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B26).

Thirteen Ovis/Capra mandibles and loose teeth were recovered from Eneolithic levels at Petnica (9 mandibles and 4 loose teeth). All elements were assigned an age. The final sample size is thirteen (Appendix A). Table 7.35 summarizes the age stage distributions of Ovis/Capra remains from the Eneolithic.

Thirty-four Bos taurus mandibular remains (mandibles – n=27 and loose teeth – n=7) were recovered from Middle Neolithic deposits at Petnica. Twelve mandibles were unidentifiable to age, leaving fifteen mandibles and seven loose teeth in the final sample (n=22 – Appendix A). Table 7.37 summarizes the age stage distributions of Bos

Fourteen Ovis/Capra remains (mandibles – n=5 and loose teeth – n=9) were recovered from Late Bronze Age levels at Petnica. Only one mandible was indeterminable to age.

58

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE taurus remains from the Middle Neolithic. Twelve Sus scrofa mandibular remains (mandibles – n=10 and loose teeth – n=2) were recovered from Eneolithic levels at Petnica. Two mandibles were unidentifiable to age, leaving eight mandibles and two loose teeth in the final analysis (n=10 – Appendix A). Table 7.41 summarizes the age stage distributions of Sus scrofa remains from the Eneolithic.

Thirty-six Bos taurus mandibular remains (mandibles – n=19 and loose teeth – n=17) were recovered from Late Neolithic levels at Petnica. Fifteen mandibles and two loose teeth were unidentifiable to age, leaving four mandibles and fifteen loose teeth in the final sample (n=19 – Appendix A). Table 7.38 summarizes the age stage distributions of Bos taurus remains from the Late Neolithic.

Seven Sus scrofa mandibular remains (mandibles – n=5 and loose teeth – n=2) were recovered from Late Bronze Age levels at Petnica. Two mandibles were unidentifiable to age, leaving three mandibles and two loose teeth in the final analysis (n=5 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B30).

All subsequent periods (Late Neolithic through the Early Iron Age) from Petnica have insufficient sample size for inclusion in the analysis. Four Bos taurus mandibular remains (mandibles – n=2 and loose teeth – n=2) were recovered from Late Neolithic/Eneolithic levels. Both mandibles were unidentifiable to age, leaving two loose teeth in the final analysis (Appendix A). It can be noted that both the loose teeth were aged to cover mandibular stages G and I (adult-senile).

Seven Sus scrofa mandibular remains (mandibles – n=6 and loose teeth – n=1) were recovered from Late Bronze Age/Early Iron Age levels at Petnica. Two mandibles were unidentifiable to age, leaving four mandibles and one loose tooth in the final analysis (n=5 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B31).

Twelve Bos taurus mandibular remains (mandibles – n=6 and loose teeth – n=6) were recovered from Eneolithic deposits. Three mandibles were unidentifiable to age, leaving three mandibles and six loose teeth in the final analysis (n=9 – Appendix A). Twelve Bos taurus mandibular remains (mandibles – n=2 and loose teeth – n=10) were recovered from Late Bronze Age deposits. Two mandibles and one loose tooth were unidentifiable to age, leaving nine loose teeth in the final analysis (n=9).

XI. Selevac Site description Selevac (also known as Selevac-Staro Selo) is a settlement dated to the Middle to Late Neolithic in central Serbia, near the modern village of Selevac. It is nestled in the hills to the west of the Morava River, close to its confluence with the Danube. It is on the south-facing slope of the hill above a large stream (the Vrbica) which eventually drains into the Morava River. The site is close to well-drained and fertile soils, a variety of forest and riverine resources, and high quality clay sources. The site is located on the nearest high ground above the Morava River’s floodplain (150 m asl). It should be considered a lowland occupation in the context of this study.

Eight Bos taurus mandibles and loose teeth were recovered from the Late Bronze Age/Early Iron Age levels at Petnica (4 mandibles and 4 loose teeth). Two mandibles were unidentifiable to age, leaving two mandibles and four loose teeth in the final analysis (n=6). All these samples were too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Tables B27-B29)). Thirty-two Sus scrofa mandibular remains (mandibles – n=31 and loose teeth – n=1) were recovered from Middle Neolithic deposits at Petnica. Nineteen mandibles were unidentifiable to age, leaving twelve mandibles and one loose tooth in the final analysis (n=13 – Appendix A). Table 7.39 summarizes the age stage distributions of Sus scrofa remains from the Middle Neolithic.

The site covers a relatively large area (53 ha), with a maximum dimension of 780x1080 m (Tringham and Krstić 1990). Excavations took place over three field seasons (1976-1978). The deposits ranged in thickness from c. 1.5 to 3 m in thickness. The site has two main periods of occupation, an earlier phase known as VinčaTordoš and a later phase referred to as Vinča-Pločnik. Four stratigraphic-architectural phases were identified during excavation (I-IV). These corresponded with the latter half of the Vinča-Tordoš (I-II) and to the earlier half of the Vinča-Pločnik (III-IV) culture phasing. Between these two cultures lies the intermediate Gradac phase of the Vinča culture, which is present at the site. The Gradac or transitional phase was originally considered to be part of the Late Neolithic, but was absent from the type-site and identified at the Gradac site

Twenty-four Sus scrofa mandibular remains (mandibles – n=22 and loose teeth – n=2) were recovered from Late Neolithic deposits at Petnica. Eleven mandibles were unidentifiable to age, leaving eleven mandibles and two loose teeth in the final analysis (n=13 – Appendix A). Table 7.40 summarizes the age stage distributions of Sus scrofa remains from the Late Neolithic. Two Sus scrofa mandibles were recovered from Late Neolithic/Eneolithic deposits at Petnica. Both were unidentifiable to age. There was no Late Neolithic/Eneolithic Sus scrofa sample for Petnica.

59

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.36. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Bronze Age Petnica Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 1 1 0 0 3 2 0 0 7

Corrected Count % 0 14 14 0 0 43 29 0 0 100

No. 0 1 2 0.5 0.9 5.2 3.4 0 0 13

% 0 8 15 4 7 40 26 0 0 100

Table 7.37. Stage distribution of Bos taurus mandibles and loose teeth from Middle Neolithic Petnica Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months young adult adult old adult senile

No. 2 0 1 4 0 1 0 0 2 10

Corrected Count % 20 0 10 40 0 10 0 0 20 100

No. 2 0 1 8.2 0 3.8 0 0 6.9 21.9

% 9 0 4 38 0 17 0 0 32 100

Table 7.38. Stage distribution of Bos taurus mandibles and loose teeth from Late Neolithic Petnica Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months young adult adult old adult senile

No. 1 0 0 3 1 0 0 0 1 6

Corrected Count % 17 0 0 50 17 0 0 0 17 101

60

No. 1 1 1 5.8 5.5 0.8 1.6 0 2.3 19

% 5 5 5 31 30 4 8 0 12 100

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.39. Stage distribution of Sus scrofa mandibles and loose teeth from Middle Neolithic Petnica Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months adult old adult senile

No. 1 5 2 2 0 0 0 0 2 12

Corrected Count % 8 41 17 17 0 0 0 0 17 100

No. 1 5 2 3 0 0 0 0 2 13

% 8 39 15 23 0 0 0 0 15 100

Table 7.40. Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic Petnica Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months adult old adult senile

No. 0 2 3 2 0 0 0 0 1 8

Corrected Count % 0 25 38 25 0 0 0 0 12 100

No. 0 2.4 3.6 6 0 0 0 0 1 13

% 0 18 28 46 0 0 0 0 8 100

Table 7.41. Stage distribution of Sus scrofa mandibles and loose teeth from Eneolithic Petnica Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months adult old adult senile

No. 0 2 0 1 0 0 0 0 1 4

61

Corrected Count % 0 50 0 25 0 0 0 0 25 100

No. 0 4 0 4 0.5 0.5 0 0 1 10

% 0 40 0 40 5 5 0 0 10 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES near Niš in southern Serbia. Some later material was found (Roman and Medieval), but is present in very minor deposits. The Neolithic occupation horizons have been dated to 4390-3650 bc (5020-4400 BC, calibrated) (Tringham and Krstić 1990). Through an examination of the frequency of data by architectural-stratigraphic phase from the site, it would seem that most of the relevant data are from the later Neolithic Vinča-Pločnik phase (Legge 1990; Tringham and D. Krstić 1990).

Mandibular and loose tooth remains Four domestic Ovis aries mandibles were recovered from Middle Neolithic levels at Stragari. All were identifiable to an age and included in the final analysis. Appendix B (Table B32) summarizes the age stage distributions of Ovis aries remains. Only one domestic Capra hircus mandible was recovered from Stragari. It was unidentifiable to age. Due to these limited samples sizes, the Ovis aries and Capra hircus remains were lumped into the category Ovis/Capra to provide an adequate sample size.

B. Mandibular and loose tooth remains Selevac is the only site in the sample that was not collected by Greenfield. But the nature of the analysis and publication makes it useful for comparison to the other samples and for inclusion in this study. Recovery of remains from the site was systematic, and included both dry (0.7 mm) and wet sieving, over the entire area of excavation. Preservation of bones was excellent. The zooarchaeological analysis is based on the 7,035 bones identified to genus or species. Harvest profiles for the major domestics (sheep/goat, cattle and pigs) were generated by Legge (1990). In our reanalysis, all mandibles from all Middle and Late phases at Selevac are combined and considered as a single Middle/Late Neolithic assemblage.

Twenty-four Ovis/Capra remains (mandibles – n=9 and loose teeth – n=15) were recovered from Middle Neolithic levels at Stragari. Two mandibles were unidentifiable to age, leaving seven mandibles and fifteen loose teeth in the final analysis (n=22 – Appendix A). Table 7.42 summarizes the age stage distributions of Ovis/Capra remains from the Middle Neolithic. Forty-eight Bos taurus mandibular remains (mandibles n=31 and loose teeth – n=17) were recovered from Middle Neolithic levels at Stragari. Eighteen mandibles and one loose tooth were unidentifiable to age, leaving thirteen mandibles and sixteen loose teeth in the final analysis (n=29 – Appendix A). Table 7.43 summarizes the age stage distributions of Bos taurus remains from the Middle Neolithic.

Legge (1990) provides some information on quantities of remains that are relevant to this study. The majority of Ovis/Capra remains from Selevac are identified as sheep. There are few goats in the sample. As a result, there was no attempt to separate the mandibles into species and they are analysed as Ovis/Capra. There are 110 ageable Ovis/Capra mandibles, of which 71 were included in Legge’s analysis. There are 78 cattle mandibles, partial mandibles and loose permanent third molars, of which only 41 mandibles were used in the final analysis. No specific information on sample size was provided for pigs (Legge 1990).

Nine Sus scrofa mandibles were recovered from the Middle Neolithic levels at Stragari. Two were unidentifiable to age, leaving seven mandibles in the final analysis (n=7 – Appendix A). This sample was too small to be included in the analysis. The age stage distributions are detailed in Appendix B (Table B33). XII. Vinča

XII. Stragari-Šljivik

A. Site description

A. Site description

The site of Vinča-Belo Brdo is a lowland site located on the right bank of the Danube, 15 km southeast of the city of Belgrade, in Serbia (Figure 1.1). It is positioned in an optimal position to monitor and control movement along the major waterways of the region. It is near the confluence of the Danube with the Morava, the Tisza, the Timiš and the Sava rivers. On the opposite bank of the Danube are found the flat plains of Pannonia.

Stragari-Šljivik is a lowland site located on a tributary of the Western Morava at the edge of the village of Stragari, near the city of Kruševac, in central Serbia (Figure 1.1). The site is located in the floodplain of a deeply incised stream, surrounded by low rolling hills. Presently, it is a rich agricultural area. Stragari is a Middle Neolithic site, with two levels of occupation from the Vinča A and B cultures. It is a small site, extending c. 200x200 m, with deposits extending for about 1 m beneath the plough zone. None of the sediments from the site were sieved or floated. Over 10,000 fragments were recovered and analyzed from the 1987-1989 excavations at the site (Greenfield n.d. b).

The soils surrounding the site are high quality fertile soils including river alluvium and chernozems. These were highly cultivable with the available Neolithic technology and continue to support modern grasslands and cultivated crops. Sporadic areas of deciduous forests are also found within the area (Barker 1975; Chapman 1981).

62

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

Table 7.42. Stage distribution of Ovis/Capra mandibles and loose teeth from Middle Neolithic Stragari Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 1 3 5 1 7 0 0 0 17

Corrected Count % 0 6 18 29 6 41 0 0 0 100

No. 0 1 3.9 8.3 1.4 7.3 0 0 0 21.9

% 0 5 18 38 6 33 0 0 0 100

Table 7.43. Stage distribution of Bos taurus mandibles and loose teeth from Middle Neolithic Stragari Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months Young adult Adult old adult Senile

No. 2 0 6 2 0 0 1 0 1 12

Corrected Count % 17 0 50 17 0 0 8 0 8 100

No. 2 0 7.5 4.3 0 0 9.2 0 6 29

% 7 0 26 15 0 0 32 0 20 100

Table 7.44. Stage distribution of Ovis/Capra mandibles and loose teeth from Late Neolithic Vinča Stage

A B C D E F G H I

Suggested Age

Raw Count

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 1 6 5 1 3 5 3 2 26

Corrected Count % 0 4 23 19 4 12 19 12 7 100

63

No. 0 1.1 8.0 5.8 1.5 5.6 6.6 3.2 2.1 34

% 0 3 25 17 4 16 20 9 6 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Vinča is a lowland site that extends throughout the entire Neolithic period, as well as having Eneolithic, Middle Bronze Age, and Medieval levels. This is the type-site for the Middle and Late Neolithic Vinča culture and is a ‘yardstick’ for the Balkan Neolithic (Chapman 1981). The faunal data described here derive from the Late Neolithic, Eneolithic and Middle Bronze Age levels.

Table 7.47 summarizes the age stage distributions of Bos taurus remains from the Middle Bronze Age. Forty-four Sus scrofa mandibular remains were recovered from Late Neolithic deposits at Vinča. Eleven mandibles were undeterminable to age, leaving thirty-four mandibles in the final analysis (Appendix A). Table 7.48 summarizes the age stage distributions of Sus scrofa remains from the Late Neolithic.

The site of Vinča is a large artificial mound that extends for several hundred meters along the riverbank. It is estimated that over one-third of the site has been eroded by the river. The site is believed to have also extended several hundred metres back from the river as well. It is similar in construction to a tell site with the stratigraphy extending greater than nine metres (Chapman 1981).

In the Eneolithic layers, there were only two domestic Sus scrofa mandibles. Only one was determinable with an age. Age stage distribution is detailed in Appendix B (Table B35). Thirty Sus scrofa mandibular remains (mandibles – n=29 and loose teeth – n=1) were recovered from Middle Bronze Age deposits at Vinča. Ten mandibles were unidentifiable to age, leaving nineteen mandibles and one loose tooth in the final analysis (n=20 – Appendix A). Table 7.49 summarizes the age stage distributions of Sus scrofa remains from the Middle Bronze Age.

The fauna reported here derive from the 1982 excavations at the site. They were hand-collected, but over 10,000 fragments were recovered and analyzed (Greenfield n.d. a). B. Mandibular and loose tooth remains

XIII. The modern tooth wear and eruption control sample

Fifty-one domestic Ovis/Capra mandibles were recovered from Late Neolithic levels at Vinča. Seventeen were unidentifiable to age, leaving thirty-two mandibles in the final analysis (Appendix A). Table 7.44 summarizes the age stage distributions of Ovis/Capra remains from the Late Neolithic.

The modern control sample included forty-three animals: twenty-nine goats and fourteen sheep. All animals were produced locally in Manitoba by small-scale producers. They are domestic farm animals raised with seasonal differences both in food availability and seasonal environment. General age data was able to be estimated for all animals based on information provided by the producer and extrapolation backwards from the date of death. All the goats were born in the month of April, with a mean birth date estimated by the producer to be April 15th. The goats range in age from 6 months to a year and a half. As the pattern of slaughter was two animals per week and the specimens were collected at roughly the middle of each month throughout the collection period, date of death was estimated over a period of two months. Further specific age data was provided by ear tag information where available. Unfortunately, all animals did not have ear tags.

The total sample of Ovis/Capra from Eneolithic deposits was a single mandible. The final sample size is one and age stage distribution data is summarized in Appendix B (Table B34). One domestic Ovis aries mandible and one domestic Capra hircus mandible from Middle Bronze Age levels were lumped into the Ovis/Capra heading. As a result, there was a total of twenty-three Ovis/Capra mandibular remains (mandibles – n=13 and loose teeth – n=10) from the Middle Bronze Age deposits at Vinča. All were included in the final analysis (Appendix A). Table 7.45 summarizes the age stage distributions of Ovis/Capra remains from the Middle Bronze Age. Nineteen Bos taurus mandibular remains (mandibles – n=11 and loose teeth – n=8) were recovered from Late Neolithic deposits at Vinča (eleven mandibles and eight loose teeth). Five mandibles were undeterminable to age, leaving six mandibles and eight loose teeth in the final analysis (n=14 – Appendix A). Table 7.46 summarizes the age stage distributions of Bos taurus remains from the Late Neolithic. There were no Bos taurus mandibular remains from the Eneolithic levels.

Tooth eruption and wear was recorded for all mandibles using the Payne (1973) and Grant (1975) methods. This data will be utilized to test the agreement between tooth wear and eruption and the assignment of absolute age. The modern tooth wear and eruption data are presented in Appendix C.

Twenty-four Bos taurus mandibular remains (mandibles n=16 and loose teeth – n=8) were recovered from Middle Bronze Age deposits at Vinča. Eight mandibles were unidentifiable to age, leaving eight mandibles and eight loose teeth in the final analysis (n=16 – Appendix A).

Data for the cementum analysis included both a modern control sample of sheep and goats and an archaeological sample. A control sample of the same species of known age and known season of death was required in order to accurately determine the timing of the deposition of the

XIV. Cementum analysis data: archaeological and modern

64

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE different cementum growth layers in order to accurately interpret the archaeological teeth. At the same time, the comparative sample was utilized to test the agreement between the two most common methods of recording tooth wear and eruption (Payne 1973; Grant 1975) and the assignment of absolute age to the wear stages. Additionally, the modern thin section sample of goats provided an interesting methodological contribution by testing whether the incremental growth visible in the teeth of the goats agrees or did not agree with the eruption/wear stages and/or with the actual ages. It acted as a three-way comparison between tooth eruption and wear, incremental structures and true ages (Ariane Burke, personal communication, 2001). Details on the modern and archaeological samples for cementum analysis are provided in Appendix D.

transhumance is hypothesized to be present. Finally, there was the spatial issue. The sample needed to cover both highland and lowland areas. As such, two sites were selected, Kadica Brdo, an Early Iron Age highland site, and Vinča, a lowland site. All the teeth from Vinča were from Late Neolithic deposits. Ten teeth were sectioned from Kadica Brdo and nine teeth were sectioned from Vinča. Details of the data from Kadica Brdo and Vinča appear in Table 7.51. XV. Conclusions Zooarchaeological remains were available for analysis from seventeen sites. Ten sites from the central Balkan region had sufficient data to be utilized in this research. One site, Megalo Nisi Galanis (in Greek Macedonia) in the southern Balkan region, is also included for comparison. The sites range in time from the Early Neolithic to the Early Iron Age. Only if the mandibular and loose tooth samples from sites included ten or more elements per species and per period, were they considered in the final analysis. Samples with a sample size lower than ten were considered too small to provide accurate harvest profiles. These were eliminated from the analysis, but are described in Appendix B in the event that more data becomes available.

A. The modern control sample The modern control cementum (thin sectioning) sample consisted of fourteen goats and five sheep. All animals had their left mandibular M1 sectioned using the methodology discussed in Chapter 6. Where possible, two slides per tooth were produced. A description of the cementum observations appears in Table 7.50. B. Archaeological cementum analysis sample

The central Balkan region can be divided into lowland, mid-altitude and highland areas. Lowland sites to be examined include Foeni-Salaş, Livade, Novačka Ćuprija, Opovo, Stragari and Vinča. Mid-altitude sites are Blagotin, Ljuljaci and Petnica. The only highland site is Kadica Brdo.

The ancient sample for cementum analysis had to meet several criteria. First, there was the simple issue of availability. The faunal material had to be available at the University of Manitoba. Second, there was consideration of the nature of the sample. As cementum is the least hard of the dental tissues, it can easily be stripped from the tooth as the result of post-mortem damage. Only teeth still encased in the mandible should be selected (Lieberman and Meadow 1994). Third, there was a consideration of the temporal issue. The archaeological sample would ideally cover the periods within which

A large modern sample of sheep and goats from the province of Manitoba were selected as a control on the relationship between absolute age and tooth eruption and wear. Samples of the modern goat and the ancient sheep and goat remains were selected for cementum analysis.

Table 7.45. Stage distribution of Ovis/Capra mandibles and loose teeth from Middle Bronze Age Vinča Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-6 months 6-12 months 1-2 years 2-3 years 3-4 years 4-6 years 6-8 years 8-10 years

No. 0 0 3 5 1 1 1 1 1 13

Corrected Count % 0 0 27 46 9 0 9 9 0 100

65

No. 0 0 3 6.7 2 3.6 3.6 2.9 1 22.8

% 0 0 13 29 9 16 16 13 4 100

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES Table 7.46. Stage distribution of Bos taurus mandibles and loose teeth from Late Neolithic Vinča Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months young adult adult old adult senile

No. 0 1 1 1 0 0 0 0 2 5

Corrected Count % 0 20 20 20 0 0 0 0 40 100

No. 0 1.5 1.2 6 0.25 0.25 0.25 0.25 4 13.7

% 0 10 9 44 2 2 2 2 29 100

Table 7.47. Stage distribution of Bos taurus mandibles and loose teeth from Middle Bronze Age Vinča Stage

Suggested Age

Raw Count

A B C D E F G H I

0-1 month 1-8 months 8-18 months 18-30 months 30-36 months young adult adult old adult senile

No. 1 1 4 1 0 0 0 1 1 9

Corrected Count % 11 11 45 11 0 0 0 11 11 100

No. 1 1 4 1 0 0 0 5.5 3.5 16

% 6 6 25 6 0 0 0 34 23 100

Table 7.48. Stage distribution of Sus scrofa mandibles and loose teeth from Late Neolithic Vinča Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months Adult old adult Senile

No. 0 5 8 9 1 0 0 0 0 23

Corrected Count % 0 23 34 39 4 0 0 0 0 100

66

No. 0.0 6.1 11.3 13.3 2.3 0.0 0.0 0.0 0.0 33.0

% 0 18 33 42 7 0 0 0 0 100

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.49. Stage distribution of Sus scrofa mandibles and loose teeth from Middle Bronze Age Vinča Stage

Suggested Age

Raw Count

A B C D E F G H I

0-2 months 2-7 months 7-14 months 14-21 months 21-27 months 27-36 months Adult old adult Senile

No. 1 1 5 6 2 1 0 0 0 16

Corrected Count % 6 6 31 38 13 6 0 0 0 100

67

No. 1.0 1.0 5.5 8.8 2.8 1.0 0.0 0.0 0.0 20.0

% 5 5 27 44 14 5 0 0 0 100

68

a

b

a

b

a

b

a

b

Goat #11

Goat #11

Goat #12

Goat #12

Goat #13

Goat #13

Goat #14

Goat #14

Goat #15

b

a

Goat #8

Goat #10

b

Goat #7

a

a

Goat #7

Goat #10

b

Goat #6

b

1 year, 5-6 months

a

Goat #6

Goat #8

1 year, 4-5 months

Goat #4

9-10 months

9-10 months

9-10 months

8 months

8 months

8 months

8 months

6-7 months

6-7 months

6-7 months

6-7 months

1 year, 5-6 months

1 year, 5-6 months

1 year, 5-6 months

1 year, 5-6 months

1 year, 5-6 months

1 year, 4-5 months

Goat #2

Age

1 year, 4-5 months

Slide

Goat #1

Sample

Jan/Feb 2001

Jan/Feb 2001

Jan/Feb 2001

December 2000

December 2000

December 2000

December 2000

Oct/Nov 2000

Oct/Nov 2000

Oct/Nov 2000

Oct/Nov 2000

Sept/Oct 2000

Sept/Oct 2000

Sept/Oct 2000

Sept/Oct 2000

Sept/Oct 2000

Sept/Oct 2000

Aug/Sept 2000

Aug/Sept 2000

Aug/Sept 2000

Date of Death

growth zone

indeterminable

growth zone

growth zone

growth zone

undeterminable

undeterminable

undeterminable

undeterminable

undeterminable

no annulus/zone final

growth zone?

Nature of final increment

0 GLG

0 GLG

reading taken at end of root

reading taken at end of root

bright outer increment = annulus forming? unreadable

unreadable

reading taken at end of root

reading taken at end of root

root area too thick for accurate reading

reading taken at end of root

reading taken at end of root

reading taken at end of root

slide too thick

Comments

growth reading taken at end of root zone/annulus final tumor tooth growth zone

2 annuli are undeterminable formed low on the root 0 GLG undeterminable

undeterminable

no GLG

indeterminable

no GLG

no GLG

no GLG

undeterminable

1GLG

1GLG

undeterminable

1GLG

1GLG

1GLG

undeterminable

1GLG (partial)

indeterminable

# of increments (GLGs)

Table 7.50. Modern Ovis/Capra comparative cementum analysis – Summary of readings

DESCRIPTION OF SITES, TIME PERIODS, AND NATURE OF SAMPLES

Vinca Vinca Vinca Vinca Vinca Vinca Vinca Vinca

V #3 V #4 V #5 V #6 V #7 V #8 V #9 V #10

b

Goat #17

Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Kadica Brdo Vinca Vinca

a

Goat #17

KB #1 KB #2 KB #3 KB #4 KB #5 KB #6 KB #7 KB #8 KB #9 KB #10 V #1 V #2

b

Goat #16

Site

a

Goat #16

Sectioning Code

Slide

Sample

Jan/Feb 2001

Jan/Feb 2001

Jan/Feb 2001

Jan/Feb 2001

Date of Death

undeterminable

undeterminable

0 GLG

0 GLG

# of increments (GLGs)

growth zone

growth zone

Nature of final increment

unreadable

unreadable

69 Lowland Lowland Lowland Lowland Lowland Lowland Lowland Lowland

Highland Highland Highland Highland Highland Highland Highland Highland Highland Highland Lowland Lowland

Location

5 6 6 5 5 5 5 5

4 24 23 12 1 16 24 3 25 13 6 6

Level

Ovis aries Ovis aries Capra hircus Ovis aries Ovis aries Capra hircus Ovis aries Ovis/Capra

Ovis aries Ovis/Capra Ovis aries Ovis aries Ovis aries Ovis aries Ovis aries Capra hircus Possible goat Ovis aries Capra hircus Ovis aries

Taxon

8-10 years 4-6 years 6 months - 4 years 4-6 years 2-3 years 6-8 years 1-2 years 3-10 years

3-4 years 2-3 years 1-2 years 4-6 years 6-8 years 4-6 years 2-3 years 6-12 months 3-4 years 6-12 months 6-8 years 4-6 years

unreadable 2 GLG 1 GLG 6 GLG unreadable 5 GLG unreadable 1 GLG unreadable 0 GLG 9 GLG Tooth completely broken during extraction unreadable unreadable unreadable 4 GLG unreadable unreadable 1 GLG 8 GLG

Absolute Age (from tooth # of increments wear and eruption) counted

indeterminate possible annulus final

possible annulus final

growth zone possible annulus final

indeterminate

growth zone

growth zone indeterminate growth zone

Nature of Final Increment

right side – bright outer increment = annulus forming? number of secondary GLGs

Comments

Table 7.51. Archaeological Ovis/Capra cementum analysis – Summary of readings

9-10 months

9-10 months

9-10 months

9-10 months

Age

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

CHAPTER 8 THE IDENTIFICATION OF TRANSHUMANT PASTORALISM THROUGH HARVEST PROFILE AND CEMENTUM ANALYSES I. Introduction

taphonomic issues of differential preservation and excavation practices, specifically the extent of sieving at sites (Table 8.52). Differential preservation is necessary to consider because the remains of the very young age classes are very fragile. It can be assumed that “a very high proportion of those originally present in archaeological contexts would be lost through pre- and post depositional processes and excavation procedures” (Munson 2000: 391; Cribb 1985). The presence/absence of this age group is too affected by the taphonomic issues mentioned above to be considered reliable in making conclusions on the existence of transhumance. Munson (2000) goes as far as to exclude the consideration of neonates and notes only presence/absence. As such, any indications of transhumance must be established by looking at the other age groups involved in the movement of herds, specifically the 2-6 months and the 6-12 month group in sheep/goat and the 1-8 month and 8-18 month group in cattle. The expected patterns for the lowland, mid-altitude and high altitude sites for each species are summarized in Table 8.53. Excavation practices have a similar attritional effect. Hand collection (as opposed to sieving) favours the gathering of the remains of larger (older) animals over small (younger) animals. These issues will be considered during the analyses below.

This chapter will undertake several types of analyses. The first section deals with the analysis of the tooth wear and eruption data discussed in Chapter 7. This first stage of the analysis will present the interpretation of the data by taxon within each period and site. Harvest profiles of each taxon/site/period are constructed and analyzed utilizing the methods discussed in Chapter 6. The analysis will consider the harvest patterns exhibited by each species over the time period covered in this investigation, that is, from the Early Neolithic to the Early Iron Age. Interpretations on animal exploitation strategies and the possibility of transhumance are examined for each taxon/site/period individually. The second stage of analysis in this chapter will synthesize the harvest profile data on a regional scale to test for the appearance of transhumance at a moment in time. The focus will be on changes over time. The third stage of analysis will evaluate the tooth wear and eruption recording techniques against the modern comparative sample to determine their validity. Finally, the results of the cementum (thin-sectioning) analysis are discussed. The analysis was limited by the paucity of high altitude sites gathered for this investigation, of which there is only one (Kadica Brdo). The reason is that the high altitude environments are not colonized until the Eneolithic (beginning of the Post Neolithic). Hence, it is difficult to find directly comparable sites from both the Neolithic and Post Neolithic. The mid-altitude sites are hypothesized to be a relevant substitute in that they should show harvest profiles most similar to the high altitude sites in the transhumant movement of herds. This is because the over-wintering of herds in these mid-altitude regions would be difficult without sufficient shelter for the herds and collected fodder. As such, the environment of the mid-altitude sites in the region, as would that of the high altitude sites, would force the movement of domestic herds out of this altitude during the colder months of the year (i.e. transhumance). This is in contrast to the lowland sites, where the variety of microenvironments in these areas can provide sufficient graze and water for herds year round. As discussed earlier, there is no environmental factor that forces the transhumant movement of lowland herds.

II. Construction of harvest profiles – first stage of analysis A. Sus scrofa dom. Domestic pig can be considered as a control for any analysis of the transhumant movement of herds (Greenfield 1988, 1999a). It is expected that the exploitation patterns of pig should not change over time as these animals were unlikely to be subjected to the transhumant movement characteristic of either cattle or sheep/goat that is hypothesized to occur. This is due to the fact that pigs are less suited to the types of movement that is required of transhumant herds. Some researchers have noted that pigs are capable of long distance movements, such as the driving of herds from Serbia to Vienna in the 19th century AD, (Halpern 1999). This tends to be a one time movement under highly developed market conditions and should not be equated with the regular movements expected in transhumance under the pre-market conditions of prehistory.

The absence of the youngest age classes for all the domestic species is a problem throughout this investigation. The hypotheses are stated in such a way as to place significant emphasis on the presence or absence of this age class. However, one must consider the

Given previous analyses, it is expected that domestic pigs were exploited predominantly for their primary products, as there are no secondary products that are readily available from pigs (Greenfield 1988, 1989). As a result, 71

72

Vinča

Stragari

Petnica

Novačka Ćuprija Opovo

Ljuljaci

Megalo Nisi Galanis Livade

Kadica Brdo

Foeni-Salaş

Blagotin

Site

0

0

0

0

0

10

20

50

100

100

85

0

0

0

0

0

0

0

2

1

1

10

0-2 month

11-25 26-50 51-75 76-100

0

1-10

Age class (Ovis/Capra)

% Sieved

1

2

2

0

1

0

0

2

1

3

4

2-6 month

11

7

23

2

2

2

0

22

25

32

39

6-12 month

1

2

3

0

0

0

0

0

4

3

9

0-1 month

Age class (Bos taurus)

Table 8.52. Summary of sieving and presence of youngest age

3

0

3

1

0

3

0

0

0

1

3

1-8 month

5

8

4

1

0

2

1

0

3

3

14

8-18 month

1

0

1

0

0

0

0

0

2

0

0

0-2 month

Age class (Sus scrofa)

7

1

11

2

2

1

2

2

11

0

1

2-7 month

17

3

12

2

1

3

4

0

10

0

0

7-14 month

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

Table 8.53. Expectations for movement in transhumant pattern Ovis/Capra Lowland Mid-altitude Highland Bos taurus Lowland Mid-altitude Highland

Age class 0-2 Present Absent Absent

2-6 month Absent Present Present

6-12 month Present Absent Absent

Age Class 0-1 Present Absent Absent

1-8 months Absent Present Present

8-18 months Present Absent Absent

Figure 8.4. Payne’s meat production – mortality profile (from Payne 1973). 73

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM the harvest profiles constructed for pigs are expected to show patterns similar to Payne’s (1973) meat production kill off pattern (Figure 8.4).

may be due to taphonomic issues discussed above. The presence of the earliest age groups (2-7 through to 7-14 months) indicates that herds were present in some proximity to the site during all seasons.

As predicted above, the harvest profiles for all sites and periods demonstrate a pattern most similar to Payne’s meat production.

The harvest profile from the site of Selevac (Legge 1990) was included for consideration and shows a significant difference from other Late Neolithic sites in the central Balkans (Figure 8.9). This difference is the result of the fact that the report (Legge 1990) combined both wild and domestic forms of pigs when the harvest profiles were constructed. As only domestic forms were considered from the other sites, the data from Selevac are not comparable and cannot be considered valid for this analysis.

Additionally, and most importantly for this investigation, the harvest patterns do not change over time. All the profiles can be characterized as showing a high mortality of animals between the ages of 2-7 months and 27-36 months. Virtually all animals are slaughtered before reaching adulthood. In the following sections, the remains from each period are separately considered.

4. Eneolithic Only one site contained sufficient Eneolithic material to construct a harvest profile — Petnica (Figure 8.10). The profile indicates the absence of the youngest age class (02 months). There is high mortality of animals between the ages of 2-7 months, and again between 14-21 months. The intervening age group is missing. Subsequently, there is a low rate of mortality until 27-36 months. Then, there is no mortality until the senile age group, which is a single senile individual. The Eneolithic profile indicates continuity of exploitation from the previous Late Neolithic. While the youngest age group (0-2 months) is absent, this may be due to the taphonomic issues discussed above. The presence of the earliest age groups (2-7 through to 7-14 months) indicates that herds were present in some proximity to the site during all seasons.

1. Early Neolithic None of the Early Neolithic assemblages had a large enough sample of pig remains to construct a harvest profile. This is due to the dearth of pig remains in such assemblages. This is typical of most Early Neolithic sites in the region (e.g. Bökönyi 1974a; Greenfield 1993, in press a). 2. Middle Neolithic In the Middle Neolithic, only the deposits from Petnica had a sufficient sample size to reconstruct a harvest profile. The harvest profile (Figure 8.5) shows a rapid mortality of the youngest age classes (0-21 months). The adult end of the profile shows the absence of subadults and adults, except for the presence of two senile individuals. The profile basically conforms to the expected exploitation pattern for meat. The presence of the earliest age groups (0-2 months through to 7-14 months) indicates that herds were present in some proximity to the site during all seasons.

5. Early/Middle Bronze Age The Early and Middle Bronze Age material will be discussed together since one of the sites (Ljuljaci) overlaps both periods. There are three sites with Early and/or Middle Bronze Age materials (Vinča, Ljuljaci and Novačka Ćuprija). The profile from Novačka Ćuprija (Figure 8.11) indicates a very high mortality of animals between the ages of 2-7 and 21-27 months. There is a total dearth of remains from the 0-2 and >27 month classes.

3. Late Neolithic Three Late Neolithic assemblages had sufficient pig remains to construct a harvest profile (Petnica, Vinča and Opovo). Each of the sites shows essentially the same pattern, regardless of whether the sites are in lowland or mid-altitude locations. Chi square analysis indicates that the harvest profiles for each of the three sites show no statistically significant difference (Appendix E, Table E1). All are missing the youngest age class (0-2 months). There is a very high mortality of animals between the ages of 2-7 and 27-36 months. The Late Neolithic Petnica deposits (Figure 8.6) continue to show a presence of small quantities of senile individuals (n=1). The sites of Vinča (Figure 8.7) and Opovo (Figure 8.8) show an absence of senile individuals. In these two sites, all animals are slaughtered before the age of 21-27 months.

In the mid-altitude site of Ljuljaci (Figure 8.12), the harvest pattern remains the essentially the same. The only difference is that a few animals are being slaughtered at a slightly older age (up to 36 as opposed to 27 months at Novačka Ćuprija). The same general pattern exists in the Middle Bronze Age lowland site of Vinča (Figure 8.13). Just as in Ljuljaci and Novačka Ćuprija, there is a very high mortality in the 2-7 to the 21-27 month age classes. Vinča, however, is more similar to Ljuljaci in that the age classes extend to the 27-36 month range. It differs from both of them by the presence of the youngest age classes (0-2 months).

The Late Neolithic pattern is a continuation of the Middle Neolithic pattern. The profiles indicate exploitation of pigs for meat. The presence of very young individuals in the sample confirms the hypothesis that no domestic animals are moving between highland and lowland sites. While the youngest age group (0-2 months) is absent, this

Statistical analysis indicates that the harvest profiles for each of the three sites show no statistically significant 74

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months B

7-14 months 14-21 months 21-27 months 27-36 months C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.5. Harvest Profile (Sus scrofa) Middle Neolithic Petnica

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months B

7-14 months 14-21 months 21-27 months 27-36 months C D E F

adult G

Mandibular Stage and Absolute Age

Figure 8.6. Harvest Profile (Sus scrofa) Late Neolithic Petnica

75

old adult H

senile I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

60

40

20

0 0 months

0-2 months A

2-7 months B

7-14 months 14-21 months 21-27 months 27-36 months C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.7. Harvest Profile (Sus scrofa) Late Neolithic Vinča

100

80

% Age Survival

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months 7-14 months B C

14-21 months D

21-27 months E

27-36 months F

adult G

Mandibular Stage and Absolute Age

Figure 8.8. Harvest profile (Sus scrofa) Late Neolithic Opovo

76

old adult H

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% of age survival

80

60

40

20

senile I

old adult H

adult G

27-36 months F

21-27 months E

14-21 months D

7-14 months C

2-7 months B

0-2 months A

0

0

Mandibular Stage and Absolute Age Selevac (Middle/Late Neolithic)

Petnica (Middle Neolithic)

Opovo (Late Neolithic)

Vinča (Late Neolithic)

Figure 8.9. Harvest profile (Sus scrofa) Middle/Late Neolithic Selevac

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months B

7-14 months 14-21 months 21-27 months 27-36 months C D E F

adult G

Mandibular Stage and Absolute Age

Figure 8.10. Harvest Profile (Sus scrofa) Eneolithic Petnica

77

old adult H

senile I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months B

7-14 months 14-21 months 21-27 months 27-36 months C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.11. Harvest Profile (Sus scrofa) Early Bronze Age Novačka Ćuprija

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months 7-14 months B C

14-21 months D

21-27 months E

27-36 months F

adult G

old adult H

Mandibular Stage and Absolute Age

Figure 8.12. Harvest profile (Sus scrofa) Early/Middle Bronze Age Ljuljaci

78

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months 7-14 months B C

14-21 months D

21-27 months E

27-36 months F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.13. Harvest profile (Sus scrofa) Middle Bronze Age Vinča

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months 7-14 months B C

14-21 months D

21-27 months E

27-36 months F

adult G

Mandibular Stage and Absolute Age

Figure 8.14. Harvest profile (Sus scrofa) Late Bronze Age Livade

79

old adult H

senile I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-7 months B

7-14 months 14-21 months 21-27 months 27-36 months C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.15. Harvest profile (Sus scrofa) Early Iron Age Kadica Brdo difference (Appendix E, Table E2). This result indicates continuity of exploitation in pigs over time. The presence of young individuals in lowland and mid-altitude sites also supports the hypothesis that pigs are not moving in a transhumant fashion. While the youngest age group (0-2 months) is absent, this may be due to taphonomic issues discussed above. The presence of the earliest age groups (2-7 through to 7-14 months) indicates that herds were present in some proximity to the site during all seasons.

proximity to the site during all seasons. 8. Implications of pig exploitation patterns The exploitation pattern of pigs does not change significantly over the entire time period considered in this investigation. Statistical analysis of sites by major periods indicates no statistically significant changes (Appendix E, Table E3). The exploitation of pigs for their primary products continues to be the dominant feature of all the harvest profiles.

6. Late Bronze Age During the Late Bronze Age, there is only a single sample with sufficient pig remains, the lowland site of Livade (Figure 8.14). This profile follows the trend of the preceding periods. The absence of the 0-2 month class in a lowland site is likely a result of differential destruction of younger age classes at the site (Cribb 1985; Greenfield 1986a; Munson 2000). The presence of the earliest age groups (2-7 through to 7-14 months) indicates that herds were present in some proximity to the site during all seasons.

Most importantly for the investigation into the origins of transhumant pastoralism, there is no difference in the exploitation patterns of domestic pig between highland and lowland sites either before and after the Late Neolithic/Post Neolithic juncture or at any other time. B. Ovis/Capra Separate harvest profiles for Ovis aries and Capra hircus were not possible to produce. Only three sites had sufficient Ovis aries remains to produce separate harvest profiles and no sites had sufficient remains of Capra hircus. Therefore, the remains of both taxa were combined in order to achieve a number of sites with sufficient remains for analysis. The drawback of this approach is that it is difficult to separate out the pattern of sheep versus goat. This is somewhat negated by examining the frequency distribution of sheep and goat in each site. In every case, sheep remains far outnumber those of goats (often at a 10:1 ratio). Goats tend to be an insignificant part of total identified assemblages. In addition, most previously identified goat remains were from adult

7. Early Iron Age The Early Iron Age highland site of Kadica Brdo (Figure 8.15) yields a pattern broadly similar to those of the other sites. The highest mortality rate is between 2-7 months and 21-27 months. In addition, there are both very old and very young individuals at the site. The presence of the 0-2 month age class would indicate that animals are born at or near the site. This would indicate the lack of seasonal transhumant movement of pigs at the site. The presence of the earliest age groups (0-2 through to 7-14 months) indicates that herds were present in some

80

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE individuals (Greenfield 1986a, 1986b, 1988, 1991, 1994, 1996). Therefore, it is proposed that the perceived pattern probably represents the remains of sheep, with a slight influence by goats on the older age classes.

production model and from the Early Neolithic assemblages is in the absence of the oldest age groups (4-6 to 8-10 years). Otherwise, this Middle Neolithic site follows the same general pattern established in the Early Neolithic.

1. Early Neolithic During the Early Neolithic period, both the lowland site of Foeni-Salaş and the mid-altitude site of Blagotin have sufficient sample sizes to produce harvest profiles for Ovis/Capra. At Foeni-Salaş (Figure 8.16), there is some, but very little mortality of the youngest age classes (0-2 to 2-6 months). This is followed by very high mortality of the age groups 2-6 months to 1-2 years. Then there is a plateau in the mortality rate between 1-2 and 2-3 year age classes. This is followed by a rapid mortality of the age groups from 1-2 years to 4-6 years, but at a decreased rate than during the previous rapid mortality phase. This pattern is essentially repeated at Blagotin (Figure 8.17), with two major differences: the absence of the plateau of the mortality rate between 1-2 and 2-3 year age classes and the oldest age group (8-10 years). The general Early Neolithic mortality pattern for Ovis/Capra is shown in Figure 18. Statistical analysis indicates that there are no statistically significant differences between the profiles of Foeni-Salaş and Blagotin (Appendix E, Table E4).

There is a lack of evidence for transhumance during the Middle Neolithic. Stragari is a lowland site. The 0-2 month age class would be expected to be present. The absence of the 0-2 month age class would normally imply that the animals are not being born in and around the site. The absence of this age class may be because this is an unsieved sample, but such remains are present in other unsieved or partially sieved samples. The presence of the 2-6 and 6-12 month age classes would indicate that the herds were around the site for the majority of the year. The continuity of occupation and exploitation from these age classes would imply that there is no evidence for transhumance during this period. 3. Late Neolithic Two Late Neolithic assemblages have sufficient Ovis/Capra remains to construct a harvest profile (midaltitude Petnica and lowland Vinča). The lowland site of Selevac (Legge 1990) is also included in the analysis. The harvest profile from Vinča (Figure 8.20) indicates an absence of the youngest age group (0-2 months) followed by a very low mortality of the 2-6 month age groups. Selevac (Figure 8.21) shows a higher rate of mortality in the 2-6 month group (than at Vinča) followed by a relatively rapid mortality through the 6-12 months and 1-2 years age groups. There is a rapid, but continued high mortality rate, in comparison to the earlier groups, of the older age groups (2-3 through 8-10 years). At the midaltitude site of Petnica (Figure 8.22), a slightly different pattern emerges. Both the 0-2 months and the 2-6 months are absent (whereas they were present in all earlier and contemporary assemblages). Following this, there is a rather steady rate of mortality from the 6-12 month group through to the 6-8 year stage. The oldest individuals are also absent. The harvest profiles for Ovis/Capra (Figure 8.23) from the Late Neolithic sites show no statistically significant differences between the profiles (Appendix E, Table E5).

The harvest profiles from both sites indicate the exploitation of the herd for primary products, that is, meat production. It most closely resembles Payne’s Model A kill-off pattern (see Figure 8.4). The Early Neolithic sites do not show any evidence for a transhumant pattern. The presence of the 0-2, 2-6, and 612 month age classes in both sites implies a year round availability of the herds. This suggests a non-transhumant movement and continuous culling of the animals and residential stability throughout the year. There is no complementarity between the sites and transhumance is not occurring between mid-altitude Blagotin and lowland Foeni-Salaş. 2. Middle Neolithic There is only one site that provides information on the Middle Neolithic harvest profile patterns of Ovis/Capra the lowland site of Stragari (Figure 8.19). The harvest profile of Stragari shows the same pattern as those from Early Neolithic. The 0-2 month class is missing, and there is very low mortality of the youngest age groups (2-6 months), followed by a rapid mortality from 6-12 months to 1-2 years. This is followed by a rapid mortality, but at a decreased rate in the older age groups (1-4 years). What is unusual about this profile in relation to the Early Neolithic is the complete lack of the oldest age groups (3-4 through to 8-10 years). However, these age stages are not sensitive to the hypotheses being tested.

The harvest profiles of all the Late Neolithic sites most closely resemble the herd exploitation model for primary products, that is, meat production, during this period (Payne’s 1973 Model A kill-off pattern—see Figure 8.4). None of the harvest profiles appear to fully fit the expected pattern for Neolithic ovicaprine exploitation with its absence of the earliest age groups. These sites stand in contrast to the profiles created for the other sites in the sample for the Neolithic period. It cannot be argued that the absence of these youngest age groups was the result of differential recovery practices as this absence is most pronounced at Petnica, where approximately 20% of the sample was sieved. Also, there is no evidence for additional taphonomic issues that might plague Petnica in contrast to other sites from this and later time periods (Greenfield 1986a, 1991).

The harvest profile most closely resembles the herd exploitation model for primary products, that is, meat production, during this period (Payne’s 1973 Model A killoff pattern—see Figure 8.4). Where it differs from the 81

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

60

40

20

0 0 months

0-2 months A

2-6 months 6-12 months B C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Ages

Figure 8.16. Harvest Profile (Ovis/Capra) Early Neolithic Foeni-Salaş

100

80

% Age Survival

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months 6-12 months B C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

Mandibular Stages and Ages

Figure 8.17. Harvest Profile (Ovis/Capra) Early Neolithic Blagotin

82

8-10 years I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

Blagotin

Foeni

100

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.18. Harvest Profile (Ovis/Capra) Early Neolithic

100

80

% Age Survival

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

Mandibular Stage and Absolute Age

Figure 8.19. Harvest Profile (Ovis/Capra) Middle Neolithic Stragari

83

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months 6-12 months B C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.20. Harvest Profile (Ovis/Capra) Late Neolithic Vinča

100

% of age survival

80

60

40

20

0 0

0-1 months A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

old adult H

Mandibular Stage and Absolute Age Selevac

Opovo

Petnica

Vinča

Figure 8.21. Harvest Profile (Bos taurus) Middle/Late Neolithic Selevac 84

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.22. Harvest profile (Ovis/Capra) Late Neolithic Petnica

Vinca

Petnica

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

Mandibular Stage and Absolute Age

Figure 8.23. Harvest profile (Ovis/Capra) Late Neolithic

85

6-8 years H

8-10 years I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

8.24 Payne’s (1973) Milk Production

86

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

8.25 Payne’s (1973) Wool Production

87

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.26. Harvest profile (Ovis/Capra) Eneolithic Novačka Ćuprija While the absence of the youngest age groups (especially in the lowland site) may be seen as an indication of transhumant movement in the Late Neolithic, this is not believed to be the case here. If transhumance was present, the second youngest age group (2-6 months) would be expected to be present in mid-altitude sites. The absence of both the 0-2 and 2-6 month class in the mid-altitude sites (e.g. Petnica) would argue that these age classes were not being exploited. The absence of the youngest age groups from Late Neolithic Petnica, a mid-altitude site, would argue against the appearance of transhumance during the Late Neolithic.

value is given as 0.0110, the exact p value is 0.048. If one were to consider only the usual p value, one might assume that the results are extremely significant. However, as it is the exact p value that corrects for small sample sizes (as is the case here), its values indicate that the difference is not as strong as it is implied by the other value. If transhumance was occurring, these differences would be expected between mid altitude Petnica and lowland Novačka Ćuprija at this time. However, the small sample size mitigates our ability to confidently identify its presence. Neither of these patterns fit any of Payne’s proposed specialized herd production models (Figures 8.4, 8.25 and 8.26). Both, however, conform to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with a mixed subsistence economy (both primary and secondary products) and an emphasis on secondary products.

4. Eneolithic There are two sites from the Eneolithic with sufficient samples to construct harvest profiles - lowland Novačka Ćuprija and mid-altitude Petnica. At Novačka Ćuprija (Figure 8.26), there is a complete absence of the early age groups (0-12 months). There is low mortality of the 1-2 year group, none in the 2-3 year group and extremely high mortality of 3-4 year animals. Then, the mortality rate plateaus until 8-10 years. At Petnica (Figure 8.27), there is an absence of the youngest age groups (0-2 and 26 months) followed by a very high mortality of the 6-12 month age group. Subsequently, there is a decreased rate of mortality at 1-2 years, which slightly levels off at 2-3 years. It continues at this rate until the 4-6 year class. There is a complete absence of the two oldest age groups (6-8 and 8-10 years).

During the Eneolithic, clear indications for transhumance are absent. If transhumance was to appear during the Eneolithic, the youngest age group, then the 0-2 month class should be present at lowland sites. Can their absence be the result of differential attrition? Sieving is an unlikely cause since Novačka Ćuprija was completely sieved. However, the assemblage was not deeply buried and was subject to higher rates of weathering than many of the other contemporary assemblages (Table 8.54). The absence of the expected 0-2 month class at Novačka Ćuprija cannot be used to monitor the presence or absence of the transhumant movement of herds.

Chi square analysis indicates that there is a marginally statistically significant difference between these two sites in this period (Appendix A, Table E6). While the usual p

88

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.27. Harvest profile (Ovis/Capra) Eneolithic Petnica

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

Mandibular Stage and Absolute Age

Figure 8.28. Harvest profile (Ovis/Capra) Early Bronze Age Novačka Ćuprija

89

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM As a mid-altitude site, Petnica is expected to show mortality profiles similar to that of a highland site. This would require the absence of the youngest age class (0-2 months) and the presence of the second youngest age class (2-6 months), which would indicate a seasonal presence in the highlands between May and August. However, both of these age classes are missing from the profile. As a result, it can be concluded that there is no evidence for transhumance, since Eneolithic Petnica does not conform to the expectation of a highland site involved in the transhumant movement of herds.

of assemblage attrition, given the lack of sieving at Vinča (Table 8.49). The slightly older age classes should be less affected (cf. Munson 2000). The absence of the 2-6 month class and presence of the 6-12 month class at Vinča are in accordance with the expectations for a lowland site involved in the transhumant movement of herds. The problem is that the presence of the 2-6 month classes at Novačka Ćuprija is not accordance with the expectations for a lowland site involved in the transhumant movement of herds. One site may be interpreted to be part of a transhumant system and the other may not. A major problem with this period is that there is an absence of mid- and high altitude site locations with sufficient data with which to compare them.

The mid-altitude site of Petnica is missing the expected 2-6 month class for a transhumant pattern and the lowland site of Novačka Ćuprija is missing its 6-12 month age group.

6. Late Bronze Age The situation is different for the Late Bronze Age. Three sites yielded sufficient data for this period—mid-altitude Petnica and lowland Novačka Ćuprija and Livade. Novačka Ćuprija (Figure 8.30) shows an absence of very young age groups (0-6 months). There is a medium mortality rate of age classes 6-12 months and 1-2 years followed by a higher mortality rate through the 2-6 years age classes. After this, the rate declines (6-8 years). There are no remains from oldest age group (8-10 years) sample. At Livade (Figure 8.31), there is complete absence of age groups 0-12 months and a steep decline in the profile indicating a rapid mortality of subadult/adults (2-3 through to 6-8 years). There is an absence of the oldest age group (8-10 years). Petnica (Figure 8.32) shows a different pattern. Even though the youngest age groups (0-2 months) are absent, the next age group (2-6 months) is present at the site. There is a pattern of gradual mortality from 2-6 months through to 2-3 years followed by a more rapid mortality of the age groups 34 years and 4-6 years. There is an absence of the oldest age groups (6-8 and 8-10 years).

In addition, the 6-12 month class is present in Petnica. The animals in this age class would be expected to have returned to their lowland pastures for the autumn and winter months. As a result, it is less likely that transhumant pastoralism is taking place at this time. 5. Early/Middle Bronze Age As with pigs, the Early and Middle Bronze Ages are considered together. There is a great deal of cultural continuity between these periods. Only two sites yielded samples, both of which are from the lowlands - Early Bronze Age Novačka Ćuprija and Middle Bronze Age Vinča. The Novačka Ćuprija (Figure 8.28) mortality profile shows the absence of the youngest age group (0-2 months) followed by little mortality of the 2-6 and 6-12 months age groups. This is followed by rapid mortality of age groups 1-2 years and finally a rapid mortality, but at a decreased rate, of the older age groups (2-10 years) in comparison to the earlier groups. The harvest profile from Vinča (Figure 8.29) shows an absence of the youngest age groups (0-2 and 2-6 months). There is continued and rapid mortality of the age groups beginning with 6-12 months and continuing through to 8-10 years, with slight changes along the way. The profiles of Early Bronze Age Novačka Ćuprija and Middle Bronze Age Vinča are very similar. There are no statistically significant differences between the profiles of these sites (Appendix E, Table E7).

It would be expected that the lowland sites of Novačka Ćuprija and Livade would be similar to each other and significantly different from the mid-altitude site of Petnica (Figure 8.33). However, chi square analysis of the profiles from all three sites from this period indicates that there is no statistically significant difference between them (Appendix E, Table E8). It is probable that the sample sizes are not sufficiently large to reflect the expected differences using statistics.

Neither of these patterns fit any of Payne’s proposed specialized herd production models (Figures 8.4, 8.24 and 8.25). Both, however, conform to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with a mixed subsistence economy (both primary and secondary products) and an emphasis on secondary products.

None of the sites fit any of Payne’s proposed specialized herd production models (Figures 8.4, 8.24 and 8.25). Both conform to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with a mixed subsistence economy (both primary and secondary products) and an emphasis on secondary products.

The evidence for transhumance from this period is somewhat mixed. Since both samples come from lowland sites, they should have evidence of the 0-2 and 6-12 month, and absence of 2-6 month age classes if transhumance is occurring. The 0-2 age class is missing in both cases. The absence of the expected 0-2 month class is not surprising in these sites. It may be a function

The evidence for transhumance in this period is ambiguous. At Petnica, the evidence is mixed. On the one hand, the presence of the 2-6 month class and the absence 90

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE Table 7.54. Summary of strata and weathering Site

Superimposed strata

Weathering

Blagotin Foeni-Salaş Kadica Brdo Kozani Livade Ljuljaci Novačka Ćuprija Opovo Petnica Stragari Vinča

Deep Shallow Deep Deep Shallow Shallow Shallow Deep Deep Deep Deep

Low Medium Low Low Very high, except in deep pits Medium Medium Low Low Low Low

of the younger class would indicate the presence of herds during the warmer half of the year. This is in accordance with expectations since Petnica is a mid-altitude site (and not a pleasant area to over winter with ovicaprine herds). On the other hand, the presence of the 6-12 month age class at Petnica, which would be during the colder half of the year, is not in accordance with the stated hypotheses – that herds would be moved out of the hill country during the winter. At the same time, this pattern would indicate that Petnica is a sedentary year-round occupation.

gaps in the youngest age groups (0 to 1-2 years). The harvest pattern at Kadica Brdo does not fit any of Payne’s (1973) proposed models for exploitation strategies (Figures 8.4, 8.24 and 8.25), but conforms to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with an emphasis on secondary products. The remains from Kadica Brdo are not indicative of transhumance. The harvest profile shows the presence of all age classes, including the youngest, which implies a year round availability of the herds. This, in turn, suggests a non-transhumant movement and continuous culling of the animals and residential stability throughout the year. It may be that at this later period, the economy of highland areas was developed enough to support year round residence of domestic herds through the production and storage of winter fodder and appropriate shelter available for the animals. This does not preclude the possibility that the bulk of the herd may have been moved in a transhumant fashion, while a part may have been kept year-round in proximity to the settlement.

The data from the two lowland sites (Livade and Novačka Ćuprija) implies that the occupants are participating in a transhumant economy. The 2-6 month age class is absent at both lowland sites, which is in accordance with expectations for a transhumant economy—herds are absent from the lowlands during the summer months. Additionally, the presence of the 6-12 month class at Livade, which is contrary to expectations for a lowland site, and can be traced to the taphonomic history of the assemblage, with its high rate of weathering (Table 8.54). Other taphonomic issues arise when considering the youngest age class (0-2 months), complicating any interpretations of transhumance. At all of these sites, the Late Bronze Age levels were subjected to a great deal of weathering. The youngest age classes are most likely to disappear under such conditions. The degree of sieving may also have an effect upon assemblage diversity, having a greater effect upon some assemblages than other, but clearly influencing the presence of the youngest age classes (Table 8.52).

8. The Southern Balkans Megalo Nisi Galanis was included in the analysis in an effort to make a comparison with the southern Balkans. The data from each period are considered separately. Only ovicaprine remains were abundant enough to be included in the analysis. a. Final Neolithic The harvest profile from Megalo Nisi Galanis shows low mortality of the 0-2 month and absence of the 2-6 month age groups. There is a rapid mortality of the 6-12 month through to the 4-6 year age groups. There are no indications of the presence of the 6-8 years or 8-10 year age groups.

7. Early Iron Age The harvest profile from the high altitude Early Iron Age site of Kadica Brdo (Figure 8.34) shows a very low mortality in the youngest age groups (0-6 months), followed by rapid mortality of age groups 6-12 months and 1-2 year, followed by a slowing in the 2-3 year, and a final very rapid mortality rate between 3-4 and 4-6 years. There is a very low rate of older age groups (6-8 and 8-10 years) in comparison to the earlier groups. There are no

The mortality profile of the Final Neolithic (Figure 8.35) of Megalo Nisi Galanis most closely resembles the herd exploitation model for primary products, that is meat production, during this period (Payne’s 1973 Model A 91

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM kill-off pattern—see Figure 8.4).

1-8 month age class followed by rapid mortality of the 830 months age groups. The 30-36 month age group is missing from the profile. There is again a rapid mortality rate of the young adult age group. Adult and old adult age classes are also missing. This site was totally sieved, so the sample is believed to be well representative of the youngest age groups that may have been omitted in unsieved sites. This pattern is essentially repeated at Blagotin, with two major differences: both the 30-36 months age group and the old adult age group are present and there is an absence of senile individuals.

The Final Neolithic period at Megalo Nisi Galanis is temporally contemporaneous to the Eneolithic period of the central Balkans. Therefore, one would expect to see a harvest profile similar to those seen in the Eneolithic from further north. The harvest profile shows the presence of the youngest age group (0-2 months), the absence of the next age group (2-6 months), and a high mortality rate of the subsequent age groups. This appears to fit exactly with the expectations of a lowland site involved in a transhumant movement of herds. However, Megalo Nisi Galanis is a site located on a mid-altitude plateau. As a result, one would have expected the opposite pattern.

The harvest profiles from both sites indicate the exploitation of the herd for primary products, that is, meat production. It most closely resembles Payne’s Model A kill-off pattern (see Figure 8.4).

b. Final Neolithic/Early Bronze Age The harvest profile of the Final Neolithic/Early Bronze Age of Megalo Nisi Galanis (Figure 8.36) shows the absence of the youngest age group (0-2 months), followed by a somewhat rapid mortality of the 2-6 month groups and the 6-12 months group. A much more rapid mortality rate is seen for the 1-2 year group, followed by a gradual mortality of the remaining groups.

The Early Neolithic harvest profiles imply a year round availability of the herds which suggests non-transhumant movement and continuous culling of the animals and residential stability throughout the year. The profile from both Foeni-Salaş and Blagotin shows the presence of the youngest age classes (0-1, 1-8 and 8-18 months). The harvest profiles from the two sites have no statistically significant differences (Appendix E, Table E9). Additionally, there are no statistically significant differences between an average Bos taurus pattern and an average Ovis/Capra pattern (Appendix E, Table E10) for the same period (Figure 8.39). The similarity between both profiles is remarkable, given that one is lowland and the other is mid-altitude. It would be correct to state that there is no complementarity between them and that transhumance is not occurring between highland and lowland areas.

The mortality profile of the Final Neolithic/Early Bronze Age remains from Megalo Nisi Galanis most closely resemble the herd exploitation model for primary products, that is meat production, during this period (Payne’s 1973 Model A kill-off pattern—see Figure 4). Culling of the youngest age groups is later than in the Final Neolithic within Megalo Nisi Galanis. The harvest profile from both the Final Neolithic and Final Neolithic/Early Bronze Age deposits at Megalo Nisi Galanis do not fit the pattern hypothesized for a midaltitude site involved in transhumance. This is a region in which transhumance is expected as an ecological necessity (cf. Geddes 1983). The environmental differences between the two regions (Mediterranean and temperate Europe) leads to different expectations about the nature of transhumance—it is expected due to climatic forces in the Mediterranean, while it is not in the lowlands of the central Balkans. The limited evidence that we present here supports the expectation that the two regions would have different patterns of transhumance. In order to have an effective comparison between the patterns of the central and southern Balkans, a larger sample of sites, including those from both high and low altitude locations, is necessary.

2. Middle Neolithic Two sites provide information on the Middle Neolithic harvest profile patterns of Bos taurus. These are the lowland site of Stragari (Figure 8.40) and mid-altitude Petnica (Figure 8.39). The harvest profile of Stragari shows the same general pattern as those from Early Neolithic. There is low mortality of the youngest age group (0-1 month) followed by an absence of the 1-8 month age group. A rapid mortality rate of the 8-30 months groups is seen. The 30-36 months and young adult groups are both missing from the profile. A rapid mortality rate of the adult age group is seen and an absence of the old adult group. The harvest profile of Petnica shows similar patterns. But there are several differences: there is a low mortality from 8-18 months and rapid mortality of the young adult group. The adult group is also missing from the profile. While both sites have adequate sample sizes, Stragari (n=29) and Petnica (n=22), both are missing important age classes for this investigation (Figure 8.40). Significantly, while both sites include the remains of the 0-1 month age class, both are missing the 1-8 months age class.

C. Bos taurus 1. Early Neolithic In the Early Neolithic, there are two sites with sufficient sample sizes to produce harvest profiles for Bos taurus the lowland site of Foeni-Salaş (Figure 8.37) and the mid-altitude site of Blagotin (Figure 8.38). The harvest profile shows a rapid mortality of the youngest age groups (0-1 month), a slight reduction in mortality of the

The harvest profiles from both sites indicate the exploitation of the herd for primary products, which is for 92

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE meat, hide and bone production. It most closely resembles Payne’s Model A kill-off pattern (see Figure 8.4).

Petnica, Vinča, Opovo and Selevac show the presence of the youngest age classes (0-1, 1-8 and 8-18 months). This implies a year round availability of the herds, which in turn, suggests a non-transhumant movement and continuous culling of the animals and residential stability throughout the year. It would be correct to state that there is no complementarity between them and that transhumance is not occurring between highland and lowland areas. This pattern conforms to the pattern established in the Early Neolithic. There are no statistically significant differences between the harvest profiles of the four Late Neolithic sites (Appendix E, Table E12).

The harvest profiles would seem to imply that the occupants of both sites are moving their herds in a transhumant fashion. This would be the pattern for lowland sites. While Stragari can be considered a lowland site, Petnica is a clear mid-altitude site. This contradiction is not a function of small sample size, since both have adequate samples (above). The statistical analysis indicates that there are some statistically significant differences between the profiles for this period (Appendix E, Table E11). However, this difference should only be considered to be a moderate (as the usual p value is greater than 0.05 and the exact value is only slightly less than 0.05). Even though the sample size is high for this investigation, the age groups are not appropriately representative. However, since a significant number of age groups are missing from both profiles, it may be suggested that there is a problem with how representative is the sample, rather than indication of transhumance.

4. Eneolithic In the Eneolithic period, there is only one site with sufficient data—Blagotin. The harvest profile from Blagotin (Figure 8.48) shows a complete absence of the youngest age groups (0-8 months) followed by a steep mortality rate for the 8-18 month group through to the young adult group. There is the complete absence of the oldest age groups, adult, old adult and senile.

3. Late Neolithic There are four sites with data from the Late Neolithic. The lowland sites of Opovo (Figure 8.43) and Vinča (Figure 8.44), the mid-altitude site of Petnica (Figure 8.45) and Selevac (Figure 8.46) analyzed by Legge (1990) all have sufficient sample sizes of Bos taurus for the Late Neolithic.

The pattern does not fit any of Payne’s proposed specialized herd production models (Figures 8.4, 8.24 and 8.25). It does, however, conform to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with a mixed subsistence economy (both primary and secondary products) and an emphasis on secondary products.

The site of Opovo shows a low rate of mortality for the youngest age groups (0-1 through to 8-18 months). This is followed by a steep mortality of the 18-30 months group. The 30-36 month age class is missing from the sample. Finally, there is a rapid rate of mortality for the young adult age class, although more gradual than the previous classes. The adult and senile age classes are missing from the sample. At Vinča, there is a more rapid mortality rate of the youngest age classes and a very low mortality rate of the 30-36 month group through to the adult age group. The senile age group is missing at Vinča. The mid-altitude site of Petnica indicates the same harvest profile pattern. There is a gradual rate of mortality for the youngest age groups (01 months through to 8-18 months). The steep mortality of the 18-30 months seen Opovo and Vinča, extends to the 30-36 months group. This is followed by the low rate of mortality for the oldest age groups. The harvest profile from Selevac shows slightly greater mortality of the youngest age groups, and then follows a similar pattern through the older age groups.

Blagotin is a mid-altitude site and would be expected to conform to the highland pattern within a transhumant strategy. There would be an expected absence of the youngest age classes (0-1 month). This is the pattern that is seen. However, as discussed above, this age class is unreliable for indications of transhumance based on taphonomic issues. One should then look at the expected presence of the next age group (1-8 months) that would be expected to dominate the harvest profile. The complete absence of this age class would indicate that the harvest profile of Blagotin does not show results that conform to the expectations of a highland site involved in the transhumant movement of herds. A problem with this period is that there is an absence of lowland site locations with sufficient data with which to compare them. 5. Early/Middle Bronze Age Two sites yielded sufficient data to construct harvest profiles for the Early/Middle Bronze Age - Ljuljaci and Vinča. The mid-altitude site of Ljuljaci (Figure 8.49) combines the data from both the Early/Middle Bronze Age period. The profile shows an absence of the youngest age class (0-1 months). This is followed by a rapid mortality rate of the 1-8 month group through to the 1830 month group. The age class 30-36 months is missing from the profile. This is followed by a low mortality rate of young adults and a more rapid mortality of adult age groups. The old adult group is missing from the profile.

The harvest profiles of all four Late Neolithic sites (Figure 8.47) most closely resemble the herd exploitation model for primary products, that is, meat production, during this period (Payne’s 1973 Model A kill-off pattern—see Figure 8.4). In terms of transhumance, the harvest profiles from the 93

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

80

% Age Survival

60

40

20

0 0 months

0-2 months A

2-6 months 6-12 months B C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.29. Harvest profile (Ovis/Capra) Middle Bronze Age Vinča

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

Mandibular Stage and Absolute Age

Figure 8.30. Harvest profile (Ovis/Capra) Late Bronze Age Novačka Ćuprija

94

8-10 years I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.31. Harvest profile (Ovis/Capra) Late Bronze Age Livade

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

Mandibular Stage and Absolute Age

Figure 8.32. Harvest profile (Ovis/Capra) Late Bronze Age Petnica

95

6-8 years H

8-10 years I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

Petnica (mid-altitude)

Novacka Cuprija (lowland)

Livade (lowland)

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.33. Harvest profile (Ovis/Capra) Late Bronze Age

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

Mandibular Stage and Absolute Age

Figure 8.34. Harvest profile (Ovis/Capra) Early Iron Age Kadica Brdo

96

8-10 years I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.35. Harvest profile (Ovis/Capra) Final Neolithic Megilo Nisi Galanis

100

% Age Survival

80

60

40

20

0 0 months

0-2 months A

2-6 months B

6-12 months C

1-2 years D

2-3 years E

3-4 years F

4-6 years G

6-8 years H

8-10 years I

Mandibular Stage and Absolute Age

Figure 8.36. Harvest profile (Ovis/Capra) Final Neolithic/Early Bronze Age Megilo Nisi Galanis

97

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

old adult H

senile I

Mandibular Stages and Absolute Ages

Figure 8.37. Harvest profile (Bos taurus) Early Neolithic Foeni-Salaş

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

Mandibular Stages and Absolute Ages

Figure 8.38. Harvest profile (Bos taurus) Early Neolithic Blagotin 98

old adult H

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

Bos taurus Early Neolithic Pattern

Ovis/Capra Early Neolithic Pattern

100

% Age Survival

80

60

40

20

0 A

B

C

D

E

F

G

H

I

old adult H

senile I

Mandibular Stage

Figure 8.39. Harvest profile Early Neolithic Bos vs. Ovis/Capra

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

Mandibular Stage and Absolute Age

Figure 8.40. Harvest profile (Bos taurus) Middle Neolithic Stragari

99

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.41. Harvest profile (Bos taurus) Middle Neolithic Petnica

Stragari

Petnica

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

Mandibular Stage and Absolute Age

Figure 8.42. Harvest profile (Bos taurus) Middle Neolithic

100

old adult H

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.43. Harvest profile (Bos taurus) Late Neolithic Opovo The Middle Bronze Age mortality profile of Bos taurus remains from Vinča (Figure 8.50) indicates some interesting patterns. There is a gradual mortality of the youngest age groups (0-1 month and 1-8 months). This is followed by a rapid mortality of the 8-18 month group and then a low mortality of the 18-30 month group. There is a complete absence of age stages 18-30 months through to adult.

pattern of a highland site, with the absence of the 0-1 month group and the occurrence of rapid mortality of the 1-8 months age group. The lowland site of Vinča does not fit the transhumant pattern for lowland sites. As expected the youngest age group (0-1 month) is present, however the next age group (1-8 months) is also present contrary to expectations for transhumance.

The Ljuljaci harvest profile does not fit any of Payne’s proposed specialized herd production models (Figures 8.4, 8.24 and 8.25). It does, however, conform to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with a mixed subsistence economy (both primary and secondary products) and an emphasis on secondary products. The unusual pattern at Middle Bronze Age Vinča indicates that there is no culling of the subadult/adult age groups. This pattern strongly indicates a traction profile. It seems likely that the subadult/adult age classes were kept in the lowland sites as agents of traction for agricultural cultivation.

6. Late Bronze Age Only a single site, the lowland site of Livade (Figure 8.51) has sufficient sample size of Bos taurus to construct a harvest profile for the Late Bronze Age period. The harvest profile indicates an absence of the youngest age classes (0-1 and 1-8 months). There is a rapid mortality of the 18-30 month age group and the 30-36 month group is absent from the profile. This is followed by a low mortality rate of the young adult and adult age classes and rapid mortality of the old adult and senile age classes.

The data from Middle Bronze Age Vinča indicates that there is no culling of the subadult/adult age groups. This pattern strongly indicates a traction profile, which is very similar to the wool profile for sheep (Figures 8.25; Greenfield 1988). It seems likely that the subadult/adult age classes were kept in the lowland sites as agents of traction for agricultural cultivation. As Ljuljaci is a mid-altitude site, the harvest profile is expected to show a highland pattern in a regional transhumant economy. Ljuljaci shows the expected

101

The harvest profile from Livade does not fit any of Payne’s proposed models for kill of patterns indicating a single exploitation strategy (Figures 8.4, 8.24 and 8.25). Instead, it appears to follow Greenfield’s (1988) pattern for mixed exploitation strategy of both primary and secondary products, with a greater emphasis upon the latter, and herd stability. Assuming a transhumant movement of the domestic herds, the absence of the 1-8 month group is expected in the lowland site of Livade. The presence of the next age group 8-18 months also fits with the expectations of a transhumant movement. As such, it seems that the data implies that

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM animals are being moved in a transhumant fashion.

analysis can take place. We suggest that transhumance can only be identified during the next step of the analysis, when sites are compared on a regional scale. It is the shifts in age group representations in the harvest scale of all sites across time that transhumance becomes apparent for the first time. In this section, the data to support this assertion are presented.

7. Early Iron Age The high altitude site of Kadica Brdo (Figure 8.50) provides the only harvest profile for the Early Iron Age. There is a very low mortality of the youngest age group (01 months), and complete absence of the 1-8 month age group. This is followed by a low mortality rate of the 8-18 and rapid mortality of the 18-30 month groups. The 30-36 month group is missing from the profile. There is a rapid mortality rate of the young adult group through to senile, although at a reduced rate from the previous age groups. The harvest pattern at Kadica Brdo does not fit any of Payne’s (1973) proposed models for exploitation strategies (Figures 8.4, 8.24 and 8.25), but conforms to Greenfield’s (1988) expectations for a herd based on long-term stability and animal exploitation with an emphasis on secondary products. The Bos taurus remains from the Early Iron Age highland site of Kadica Brdo yield a problematic harvest profile. It is one of the larger sample sizes of remains included in this study, yet there is a significant gap in the youngest age group (1-8 months). This is exactly the age group that is expected to be present if there were indications of transhumance since young animals would likely be present in the warm weather highland pastures. Therefore, the absence of this age group is startling, to the say least, and difficult to explain as a function of simply transhumance. An alternative explanation comes from the fact that the youngest age categories are not exploited when a strategy of herd stability is pursued. Such a pattern is found in very unpredictable climates and marginal environments, such as existed in the high mountains of eastern Bosnia (where Kadica Brdo is located). In such environments, herds often tend to pursue a herd management style that minimizes risk. Part of this strategy is to keep as many animals alive as possible. The consequence would be minimal harvesting of veal (as represented by the immature age groups – Arnold and Greenfield 2003; Redding 1984). This is supported by the absence or near absence of individuals in most of the immature age groups and the high rate of mortality among adults. This is probably the best way to interpret the harvest profile.

Several problems exist with the use of harvest profiles for the investigation of transhumance. One of the major problems is that the age classes of the standard techniques for the recording of tooth eruption and wear do not match those necessary for identifying the transhumant movement of herds. This lack of precision limits the elucidation of transhumance from these traditional zooarchaeological approaches. Additionally, within this investigation, hypotheses for expectations of transhumance were constructed with the assumption of a complete absence of specific age groups, an extreme expectation. This problem can be seen in the timing of movement and birth in the northern Balkans. The time of lambing/calving for all the major transhumant domestic species (cattle and sheep/goat) is February. As the herds would not be moved immediately after lambing/calving, the herd would be found in the lowland areas from February to April. Animals found in lowland areas (or archaeological sites) would be expected to be between birth and 3 months old, plus the >1 year old animals from previous seasons. Herds are moved into the highland areas from roughly May to September (based on ethnographic observations). Animals expected in the highland areas from that seasons’ birthing are between four and eight months old. The domestic herds return to the lowland areas in the autumn (roughly October) when “the youngest cohort is at least 8 months old” (Greenfield 1999a: 19). In the transhumant movement of both domestic cattle and ovicaprine herds, it is expected that the youngest age group (class A, which is 0-2 months in sheep/goat, 0-1 months in cattle) would be found only in the lowlands and would indicate a late winter/spring occupation. Unfortunately, the paucity of this age group, due to the taphonomic issues discussed above, limits the identification and interpretations for a transhumant movement of herds. Utilization of the next age group for sheep, goat and cattle for the elucidation of transhumance is also problematic. As the sheep and goat herds move into the highlands from roughly May to September, the next age stage (class B, 2-6 months in sheep/goat, 1-8 months in cattle) overlaps the hypothesized timing of movement into the highland areas. A two months old individual would be found only in lowland areas, while a six months old individual would be expected to be found in the highlands. The same overlap is present with the return movement to the lowlands in the autumn. The division of age grades is simply not subtle enough. It would be ideal to be able to split the ovicaprine 2-6 month class and the cattle 1-8 month age class into finer categories to enable finer distinctions. However, there is as yet no valid analytical justification for this.

III. Regional scale of analysis – second stage of analysis In the above section, where each site or group of sites is examined period by period, we subjected the reader to a litany of problems in the identification of transhumance with zooarchaeological data. The essential flaw in such an approach is that transhumance is a regional economic strategy and therefore must be searched for on a regional scale. Trying to identify whether one or another site is part of a transhumant pattern is an exercise in frustration, only occasionally yielding data in accordance with expectations. It is a less productive, but necessary first step in any analytical approach to transhumance. The data must be examined in such a manner before the next step in the 102

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE The lack of precision in the age class structure in combination with the assumption of a complete absence of specific age groups (Table 8.55) in the original statements of the hypotheses limits the search for transhumant movement using traditionally structure harvest profiles (Payne 1973). As a result, the original hypotheses were reformulated to expect an increased percentage of the relevant age groups in each region. For example, one should expect to see a greater percentage of the 0-2 month age group for sheep and goat, and 0-1 month age group for cattle in the lowlands, in comparison to mid-altitude and highland areas. A greater percentage of the 2-6 months sheep and goat age group and the 1-8 months cattle group were expected in mid-altitude and highland sites in comparison to the lowland sites. Finally, the proportion of the 6-12 months sheep and goat age group and the 8-18 month cattle age group is expected to be higher in lowland sites in comparison to mid-altitude and highland sites. This reformulation of hypotheses is a means of dealing with the problem of the overlap between the age classes and the transhumant movement of domestic herds.

transhumance was occurring, one expects the youngest age group (0-2 months) to show the highest percentage of the assemblage in the lowland sites. This hypothesis is difficult to evaluate with the Sus scrofa data, as no lowland sites have produced any remains from this age group. The 2-7 months age group would be expected to account for a higher percentage of the assemblage in the mid-altitude and highland sites. In the Neolithic period, a random pattern is evident (Figure 8.53). Early Neolithic Blagotin, a lowland site, shows the highest percentage, Late Neolithic Vinča and Opovo, both lowland sites, show an equal percentage as Late Neolithic Petnica, a mid-altitude site. Finally, the 7-14 months age group should dominate the lowland sites. While Late Neolithic Vinča does show the highest percentage, it is only minimally larger than Petnica, a mid-altitude site. There are no consistent patterns of presence/absence or increased percentage in lowland, midaltitude or highland sites that might indicate a transhumant movement of domestic pig in the Neolithic. The pattern in the Post Neolithic is somewhat different. Again the youngest age group cannot be considered due to its absence. Interestingly, the first expectation of the hypothesis is shown in the Post Neolithic (Figure 8.54). The 2-7 months age group accounts for a higher percentage of the assemblage in the mid-altitude and highland sites. However, this age group is missing from the Early and Middle Bronze Age levels of Ljuljaci, limiting a full regional expression of the pattern. In considering the final expectation of the hypothesis, one expects the 7-14 months age group should dominate the lowland sites. This is clearly not the pattern with the Post Neolithic pig data, where there is no consistent pattern of increased percentage of this age group in lowland or mid-altitude sites that might indicate a transhumant movement of domestic pig in this period (Figure 8.53). As a result, the hypothesis for transhumant movement of pigs cannot be supported. There are no indications of the transhumant movement of domestic pig in the Post Neolithic.

For all domestic animals (Sus scrofa, Ovis/Capra and Bos taurus), the paucity of the youngest age group (0-2 months for pig, sheep and goat, and 0-1 month in cattle) is problematic. Taphonomic issues, specifically the extent of sieving and weathering at the sites, have greatly affected the recovery of the 0-2 month age group throughout this investigation. This makes it difficult to evaluate the reformulated hypotheses with this age group. Therefore, the focus is on the other age groups. The high-altitude site of Kadica Brdo is not considered with the reformulated hypotheses due to several observations discussed above. It may be that by this later period, highland areas have developed enough to support year round residence of domestic herds through the production and storage of winter fodder and appropriate shelter available for the animals. Alternatively, the unpredictable climate and marginal environment of the high mountains of eastern Bosnia (where Kadica Brdo is located) may have contributed to a herd management style that minimizes risk.

B. Transhumant movement of ovicaprines The Ovis/Capra data present some interesting observations. No remains from the two youngest age groups are present in mid-altitude or highland sites from Neolithic sites. One can only evaluate the 2-6 months and 6-12 months age groups for their agreement with the hypothesis. The 2-6 months age group would be expected to account for a higher percentage of the assemblage in the mid-altitude and highland sites. In the Neolithic, it is the lowland site of Foeni-Salaş that shows the highest percentage (Figure 8.55). Mid-altitude Blagotin and lowland Vinča show an equal percentage. The 6-12 months age group should dominate the lowland sites. A random patterning of age groups between lowland and mid-altitude sites is evident and implies no transhumant movement of domestic sheep and goats in the Neolithic.

A. Transhumant movement of pigs Seasonal statements about pigs are based upon the assumption of a single birthing period during the year. This is because pigs were not kept ethnographically in stalls or fed throughout the year in this region. Instead, they were left to forage in the forests, with a pig herder. Whenever pigs were needed for food, the herder brought back one or more (Halpern 1999). As a result, it is unlikely that the modern conditions for multiple births throughout the year would have occurred in the past. Domestic pig is considered a control for the transhumant movement of herds as these animals were not subjected to the transhumant movement characteristic of either cattle or sheep/goat. With the restructured hypotheses above, if

In the Post Neolithic (Figure 8.56), again the 0-2 month age group is eliminated from consideration due to lack of 103

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM availability of data. The 2-6 month age group depictions fit with the reformulated hypotheses, with the mid-altitude site showing a greater percentage of 2-6 months animals than the lowland site. Again, taphonomic issues play a role here with only one mid-altitude site and a single lowland site with sufficient remains.

in a transhumant pattern. Blagotin again dominates, with Middle Neolithic Stragari also high. However, the other sites display a random patterning of age groups between lowland and mid-altitude sites and imply no transhumant movement of domestic cattle in the Neolithic. The bar graph regional analysis of the Post Neolithic cattle data conforms to the expectations of transhumant movement (Figure 8.58). Paucity of remains from the 0-1 month age group eliminates this group from evaluation. In the 1-8 months age group, one observes a greater percentage of individuals are slaughtered in the midaltitude site of Ljuljaci in comparison to the lowland site of Vinča. The reverse pattern is present in the 8-18 months age group between the sites of Livade and Vinča. A greater percentage of 8-18 month individuals are slaughtered in the lowlands than the mid-altitude sites. The original hypotheses state that the youngest individuals would be culled from the herds in the highland and mid-altitude pastures. This would be the infants and the juveniles (0-8 months old).

Interestingly, the 6-12 month age group shows exactly the opposite of the reformulated hypotheses. The mid-altitude sites show more 6-12 months individuals than are present at the lowland sites. Petnica displays significantly more individuals (50% of its total assemblage). An explanation for these observations is the divergence between the division of the age stages established from the tooth eruption and wear and the hypothesized timing of the transhumant movement. The original hypothesis states that the transhumant movement of domestic stock is predictable and identifiable. The herd will move into highland pastures in the early spring, soon after lambing/calving occurs and returns to the lowlands during the autumn. In the northern Balkan area, the time of lambing/calving for sheep/goat and cattle is February. As the herds would not be moved immediately after lambing/calving, the herd would be found in the lowland areas from February to April. It is hypothesized that the herds would then be moved into the highland areas from roughly May to September. The domestic herds return to the lowland areas in the autumn (roughly October). The specific hypotheses state that the youngest individuals would be culled from the herds in the highland and mid-altitude pastures. This age category would be the infants and the juveniles 0-8 months old (Greenfield 1999a). If these animals were being slaughtered in the mid and high-altitude sites towards the end of that period, i.e. at 7-8 months, just before the herds move back into the lowlands, they would appear in the harvest profiles in the 6-12 month age group. This is what is indicated by the Post Neolithic data. Consequently, while the individual site tooth wear and eruption age grades are somewhat problematic for seasonality and elucidation of the temporal aspects of a transhumant movement, a reconsideration of the data does argue for support for the transhumant movement of sheep, goat and cattle herds occurring in the Post Neolithic.

The data from the lowland site of Livade are problematic, in that they do not fit the pattern. While it would be easy to dismiss this anomaly as a limitation imposed by small sample size (n=16), the mid-altitude site of Ljuljaci has a similar sample size (n=15). It is likely that the problem arises again from the division of age groups at 8 months. One must question that when an 8 month old individual is coded for tooth wear and eruption, are they coded into the end of class B, 1-8 months, or the beginning of class C, 818 months? However, the pattern emerging in the data suggests a transhumant movement. In comparison, the Neolithic bar graphs display only a random patterning of age groups between lowland and mid-altitude sites, implying no transhumant movements. IV. Comparison of tooth wear and eruption ageing methods – third stage of analysis The modern ovicaprine data, while collected as the control sample to establish the timing of the formation of the cementum increments to accurately investigate the nature of the increments visible in the archaeological sample, also provides a methodological contribution. It is possible to test the agreement of the incremental cementum layers in the teeth with the eruption/wear stages and with the actual ages. It will act as a three-way comparison between tooth eruption and wear, incremental structures and true ages (Ariane Burke, personal communication, 2001). Additionally, the agreement of the two recording systems used for tooth eruption and wear can be evaluated (Grant 1975; Payne 1973).

C. Transhumant movement of cattle Better representation of the youngest age group (0-1 month) of Bos taurus from the Neolithic allows a full evaluation of the hypotheses (Figure 8.57). One expects a greater percentage of the 0-1 month age group in the lowland sites, in comparison to mid-altitude and highland sites if transhumance was occurring. This expectation is not met, with mid-altitude Blagotin showing the greatest percentage of the assemblage. Other sites show a random pattern. The 1-8 months age group would be expected to account for a higher percentage of the assemblage in the mid-altitude and highland sites. Again, Blagotin shows the highest percentage of the assemblage, but the differences between the sites are not large. Finally, the 818 months age group should dominate the lowland sites

The two tooth wear and eruption recording methods utilize similar diagrams for their representation of the different stages. Grant (1975) has many more stages and the diagrams are more representational. In contrast, Payne (1973) utilizes less stages and more stylistic representations.

104

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.44. Harvest profile (Bos taurus) Late Neolithic Vinča

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

Mandibular Stage and Absolute Age

Figure 8.45. Harvest profile (Bos taurus) Late Neolithic Petnica

105

old adult H

senile I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% of age survival

80

60

40

20

0 0

0-1 months A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age Selevac

Opovo

Petnica

Vinča

Figure 8.46. Harvest Profile (Bos taurus) Middle/Late Neolithic Selevac

Petnica

Vinca

Opovo

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

Mandibular Stage and Absolute Age

Figure 8.47. Harvest profile (Bos taurus) Late Neolithic

106

old adult H

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.48. Harvest profile (Bos taurus) Eneolithic Blagotin

100

% Age Survival

80

60

40

20

0 Suggested Age Stage

0 months

0-1 month A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

old adult H

Mandibular Stage and Absolute Age

Figure 8.49. Harvest profile (Bos taurus) Early/Middle Bronze Age Ljuljaci

107

senile I

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.50. Harvest profile (Bos taurus) Middle Bronze Age Vinča

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months 8-18 months B C

18-30 months D

30-36 months E

young adult F

adult G

Mandibular Stage and Age

Figure 8.51. Harvest profile (Bos taurus) Late Bronze Age Livade

108

old adult H

senile I

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

100

% Age Survival

80

60

40

20

0 0 months

0-1 month A

1-8 months B

8-18 months 18-30 months 30-36 months young adult C D E F

adult G

old adult H

senile I

Mandibular Stage and Absolute Age

Figure 8.52. Harvest profile (Bos taurus) Early Iron Age Kadica Brdo

60

50

% of assemblage

40 Blagotin Early Neolithic Lowland Petnica Middle Neolithic Mid-altitude Petnica Late Neolithic Mid-altitude Vinča Late Neolithic Lowland Opovo Late Neolithic Lowland

30

20

10

0 0-2 months

2-7 months

7-14 months

14-21 months

21-27 months

27-36 months

adult

old adult

senile

Age groups

Figure 8.53. Percentage of Age Groups (Sus scrofa) Neolithic

109

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

45

40

35

% of assemblage

30 Petnica Eneolithic Mid-altitude Novačka Ćuprija Early Bronze Age Lowland Ljuljaci Early Bronze Age Mid-altitude Ljuljaci Middle Bronze Age Mid-altitude Livade Late Bronze Age Lowland Kadica Brdo Early Iron Age Highland

25

20

15

10

5

0 0-2 months

2-7 months

7-14 months

14-21 months

21-27 months

27-36 months

adult

old adult

senile

Age groups

Figure 8.54. Percentage of age groups (Sus scrofa) Post-Neolithic

30 Foeni-Salaş Early Neolithic Lowland 25

Blagotin Early Neolithic Mid-altitude

Vinča Late Neolithic Lowland

% of assemblage

20

Petnica Late Neolithic Mid-altitude 15

10

5

0 0-2 months

2-6 months

6-12 months

1-2 years 2-3 years 3-4 years 4-6 years 6-8 years

8-10 years

Age groups

Figure 8.55. Percentage of age groups (Ovis/Capra) Neolithic

110

THE ORIGINS OF TRANSHUMANT PASTORALISM IN TEMPERATE SOUTHEASTERN EUROPE

70

Novačka Ćuprija Eneolithic Lowland Novačka Ćuprija EBA Lowland

60

Vinča MBA Lowland Novačka Ćuprija LBA Lowland

50

% of assemblage

Livade LBA Lowland Petnica Eneolithic Mid-altitude

40

Petnica LBA Mid-altitude Kadica Brdo EIA Highland

30

20

10

0 0-2 months

2-6 months

6-12 months

1-2 years 2-3 years 3-4 years 4-6 years 6-8 years

8-10 years

Age groups

Figure 8.56. Percentage of age groups (Ovis/Capra) Post Neolithic 25

% of assemblage

20

Foeni-Salaş Early Neolithic Lowland Blagotin Early Neolithic Mid-altitude Stragari Middle Neolithic Lowland Petnica Middle Neolithic Mid-altitude Opovo Late Neolithic Lowland Vinča Late Neolithic Lowland Petnica Late Neolithic Mid-altitude

15

10

5

0 0-1 month

1-8 months

8-18 months

18-30 months

30-36 months

young adult

adult

old adult

senile

Age groups

Figure 8.57. Percentage of age groups (Bos taurus) Neolithic

111

THE IDENTIFICATION OF TRANSHUMANT PASTORALISM

40 Vinča MBA Lowland 35 Livade LBA Lowland 30

% of assemblage

Ljuljaci E/MBA Mid-altitude 25 Kadica Brdo EIA Highland 20

15

10

5

0 0-1 months 1-8 months

8-18 months

18-30 months

30-36 months

young adult

adult

old adult

senile

Age groups

Figure 8.58. Percentage of Age Groups (Bos taurus) The comparative collection demonstrates excellent agreement between the two methods (see Appendix C). Within this sample, • • • •

Payne’s mandibular wear stage C is Grant mandibular wear stages 9-13; Payne’s mandibular wear stage D is Grant’s mandibular wear stage 21-26; Payne’s mandibular wear stage E is Grant’s mandibular wear stage 33; and Payne’s mandibular wear stage G is Grant’s stage 41.

increments that correspond well with absolute age. However, this observation must be cautiously applied as only one sheep in the collection demonstrates this pattern (Sheep #1 – Figure 8.59). This may be due to the fact that the modern comparative collection consists predominantly of younger (