Geophysical Monograph Series

Including IUGG Volumes Maurice Ewing Volumes Mineral Physics Volumes Geophysical Monograph Series

163 Remote Sensing of Northern Hydrology: Measuring 181 Midlatitude Ionospheric Dynamics and Disturbances Environmental Change Claude R. Duguay and Alain Paul M. Kintner, Jr., Anthea J. Coster, Tim Fuller-Rowell, Pietroniro (Eds.) Anthony J. Mannucci, Michael Mendillo, and 164 Archean Geodynamics and Environments Keith Benn, Roderick Heelis (Eds.) Jean-Claude Mareschal, and Kent C. Condie (Eds.) 182 The Stromboli Volcano: An Integrated Study of 165 Solar Eruptions and Energetic Particles the 2002–2003 Eruption Sonia Calvari, Salvatore Natchimuthukonar Gopalswamy, Richard Mewaldt, Inguaggiato, Giuseppe Puglisi, Maurizio Ripepe, and Jarmo Torsti (Eds.) and Mauro Rosi (Eds.) 166 Back-Arc Spreading Systems: Geological, Biological, 183 Carbon Sequestration and Its Role in the Global Chemical, and Physical Interactions David M. Christie, Carbon Cycle Brian J. McPherson and Charles Fisher, Sang-Mook Lee, and Sharon Givens (Eds.) Eric T. Sundquist (Eds.) 167 Recurrent Magnetic Storms: Corotating Solar 184 Carbon Cycling in Northern Peatlands Andrew J. Baird, Wind Streams Bruce Tsurutani, Robert McPherron, Lisa R. Belyea, Xavier Comas, A. S. Reeve, and Walter Gonzalez, Gang Lu, José H. A. Sobral, and Lee D. Slater (Eds.) Natchimuthukonar Gopalswamy (Eds.) 185 Indian Ocean Biogeochemical Processes and 168 Earth’s Deep Water Cycle Steven D. Jacobsen and Ecological Variability Jerry D. Wiggert, Suzan van der Lee (Eds.) Raleigh R. Hood, S. Wajih A. Naqvi, Kenneth H. Brink, 169 Magnetospheric ULF Waves: Synthesis and and Sharon L. Smith (Eds.) New Directions Kazue Takahashi, Peter J. Chi, 186 Amazonia and Global Change Michael Keller, Richard E. Denton, and Robert L. Lysal (Eds.) Mercedes Bustamante, John Gash, and Pedro Silva Dias (Eds.) 170 Earthquakes: Radiated Energy and the Physics of Faulting Rachel Abercrombie, Art McGarr, 187 Surface Ocean–Lower Atmosphere Processes Hiroo Kanamori, and Giulio Di Toro (Eds.) Corinne Le Quèrè and Eric S. Saltzman (Eds.) 188 Diversity of Hydrothermal Systems on Slow 171 Subsurface Hydrology: Data Integration for Properties Spreading Ocean Ridges Peter A. Rona, and Processes David W. Hyndman, Colin W. Devey, Jérôme Dyment, Frederick D. Day-Lewis, and Kamini Singha (Eds.) and Bramley J. Murton (Eds.) 172 Volcanism and Subduction: The Kamchatka Region 189 Climate Dynamics: Why Does Climate Vary? John Eichelberger, Evgenii Gordeev, Minoru Kasahara, De-Zheng Sun and Frank Bryan (Eds.) Pavel Izbekov, and Johnathan Lees (Eds.) 190 The Stratosphere: Dynamics, Transport, 173 Ocean Circulation: Mechanisms and Impacts—Past and Chemistry L. M. Polvani, A. H. Sobel, and Future Changes of Meridional Overturning and D. W. Waugh (Eds.) Andreas Schmittner, John C. H. Chiang, and 191 Rainfall: State of the Science Firat Y. Testik Sidney R. Hemming (Eds.) and Mekonnen Gebremichael (Eds.) 174 Post-Perovskite: The Last Mantle Phase Transition 192 Antarctic Subglacial Aquatic Environments Kei Hirose, John Brodholt, Thorne Lay, and David Yuen Martin J. Siegert, Mahlon C. Kennicut II, (Eds.) and Robert A. Bindschadler 175 A Continental Plate Boundary: Tectonics at 193 Abrupt Climate Change: Mechanisms, Patterns, South Island, New Zealand and Impacts Harunur Rashid, Leonid Polyak, David Okaya, Tim Stem, and Fred Davey (Eds.) and Ellen Mosley-Thompson (Eds.) 176 Exploring Venus as a Terrestrial Planet Larry W. Esposito, 194 Stream Restoration in Dynamic Fluvial Systems: Ellen R. Stofan, and Thomas E. Cravens (Eds.) Scientific Approaches, Analyses, and Tools 177 Ocean Modeling in an Eddying Regime Andrew Simon, Sean J. Bennett, and Janine M. Castro Matthew Hecht and Hiroyasu Hasumi (Eds.) (Eds.) 178 Magma to Microbe: Modeling Hydrothermal Processes 195 Monitoring and Modeling the Deepwater Horizon Oil at Oceanic Spreading Centers Robert P. Lowell, Spill: A Record-Breaking Enterprise Yonggang Liu, Jeffrey S. Seewald, Anna Metaxas, and Amy MacFadyen, Zhen-Gang Ji, and Michael R. Perfit (Eds.) Robert H. Weisberg (Eds.) 179 Active Tectonics and Seismic Potential of Alaska 196 Extreme Events and Natural Hazards: The Complexity Jeffrey T. Freymueller, Peter J. Haeussler, Perspective A. Surjalal Sharma, Armin Bunde, Robert L. Wesson, and Göran Ekström (Eds.) Vijay P. Dimri, and Daniel N. Baker (Eds.) 180 Arctic Sea Ice Decline: Observations, Projections, 197 Auroral Phenomenology and Magnetospheric Processes: Mechanisms, and Implications Eric T. DeWeaver, Earth and Other Planets Andreas Keiling, Eric Donovan, Cecilia M. Bitz, and L.-Bruno Tremblay (Eds.) Fran Bagenal, and Tomas Karlsson (Eds.) Geophysical Monograph 198

Climates, Landscapes, and Civilizations

Liviu Giosan Dorian Q. Fuller Kathleen Nicoll Rowan K. Flad Peter D. Clift Editors

American Geophysical Union Washington, DC Published under the aegis of the AGU Books Board Kenneth R. Minschwaner, Chair; Gray E. Bebout, Kenneth H. Brink, Jiasong Fang, Ralf R. Haese, Yonggang Liu, W. Berry Lyons, Laurent Montési, Nancy N. Rabalais, Todd C. Rasmussen, A. Surjalal Sharma, David E. Siskind, Rigobert Tibi, and Peter E. van Keken, members.

Library of Congress Cataloging-in-Publication Data

Climates, landscapes, and civilizations / Liviu Giosan...[et. al.], editors. p. cm. – (Geophysical monograph ; v. 198) Includes bibliographical references and index. ISBN 978-0-87590-488-7 (alk. paper) 1. Cultural landscapes. 2. Landscape changes. 3. Climatic changes. 4. Human beings–Effect of climate on. 5. Civilization, Ancient. I. Giosan, Liviu. GF41.C54 2012 304.2–dc23 2012040662

ISBN: 978-0-87590-488-7 ISSN: 0065-8448

Cover Image: (top left) The Sun is the principal natural driver of Earthʼs climate (http://commons.wikimedia.org). The water cycle translates climate signals into diverse landscapes ranging from (top right) highly productive floodplains (http://commons.wikimedia.org) to (bottom left) completely inhospitable deserts (http://commons.wikimedia.org). In the process of becoming a geological-scale force, humans have conquered even the most hazardous landscapes and devised strategies for survival and rapid recovery. (bottom left) Reconstructed pre-Columbian house structures built in El Baga National Park, Cayo Coco, Cuba. Photograph credit: Jago Cooper.

Figures, tables and short excerpts may be reprinted in scientific books and journals if the source is properly cited.

Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by the American Geophysical Union for libraries and other users registered with the Copyright Clearance Center (CCC). This consent does not extend to other kinds of copying, such as copying for creating new collective works or for resale.The reproduction of multiple copies and the use of full articles or the use of extracts, including figures and tables, for commercial purposes requires permission from the American Geophysical Union. Geopress is an imprint of the American Geophysical Union.

Printed in the United States of America. CONTENTS

Preface Liviu Giosan, Dorian Q. Fuller, Kathleen Nicoll, Rowan K. Flad, and Peter D. Clift...... vii

Bridging a Disciplinary Gap William F. Ruddiman...... 1

Section I: Americas Arguments and Evidence Against a Younger Dryas Impact Event M. Boslough, K. Nicoll, V. Holliday, T. L. Daulton, D. Meltzer, N. Pinter, A. C. Scott, T. Surovell, P. Claeys, J. Gill, F. Paquay, J. Marlon, P. Bartlein, C. Whitlock, D. Grayson, and A. J. T. Jull...... 13 Environment and Agency in the Ancient Maya Collapse James J. Aimers...... 27 Rainfall Variability and the Rise and Collapse of the Mississippian Chiefdoms: Evidence From a DeSoto Caverns Stalagmite Paul Aharon, David Aldridge, and John Hellstrom...... 35 Building Resilience in Island Communities: A Paleotempestological Perspective Jago Cooper...... 43 Recent Seasonal Variations in Arid Landscape Cover and Aeolian Sand Mobility, Navajo Nation, Southwestern United States Amy E. Draut, Margaret Hiza Redsteer, and Lee Amoroso...... 51

Section II: Asia Utilization of Distal Tephra Records for Understanding Climatic and Environmental Consequences of the Youngest Toba Tuff Emma Gatti and Clive Oppenheimer...... 63 An Abrupt Shift in the Indian Monsoon 4000 Years Ago M. Berkelhammer, A. Sinha, L. Stott, H. Cheng, F. S. R. Pausata, and K. Yoshimura...... 75 Late Holocene Drought Pattern Over West Asia Michael Staubwasser...... 89 Geomorphological Constraints on the Ghaggar River Regime During the Mature Harappan Period Hideaki Maemoku, Yorinao Shitaoka, Tsuneto Nagatomo, and Hiroshi Yagi...... 97 A Simulation of the Neolithic Transition in the Indus Valley Carsten Lemmen and Aurangzeb Khan...... 107 Mid-to-Late Holocene Agricultural System Transformations in the Northern Fertile Crescent: A Review of the Archaeobotanical, Geoarchaeological, and Philological Evidence Simone Riehl, Konstantin Pustovoytov, Aron Dornauer, and Walther Sallaberger...... 115 Late Holocene Evolution of the Fuzhou Basin (Fujian, China) and the Spread of Rice Farming Barry V. Rolett...... 137 Climate-Induced Changes in Population Dynamics of Siberian Scythians (700–250 B.C.) Irina P. Panyushkina...... 145

Section III: Africa Geoarchaeological Perspectives on Holocene Climate Change as a Civilizing Factor in the Egyptian Sahara Kathleen Nicoll...... 157 Heavy Rainfalls in a Desert(ed) City: A Climate-Archaeological Case Study From Sudan Jonas Berking, Janina Körper, Sebastian Wagner, Ulrich Cubasch, and Brigitta Schütt...... 163

Section IV: Europe The Influence of Transgressive Paleogeography on the Development and Decline of Cardium Pottery Culture (Mediterranean Neolithic) A. Amorosi and A. Morelli...... 171 A Dynamic Human Socioecology of Prehistoric and Protohistoric Ulster Tina Thurston and Gill Plunkett...... 177

Section V: New Approaches and Discussions Influences of Active Tectonism on Human Development: A Review and Neolithic Example Eric R. Force and Bruce G. McFadgen...... 195 Irragric Anthrosols—Artifacts of Human Adaptation to Arid Conditions: Examples From the High Himalaya and the Coastal Desert of Peru J. Baade...... 203 Hunter-Gatherers, Agriculturalists, and Climate: Insights From a Cross-Disciplinary Review Daniel M. Cadzow...... 209 AGU Category Index...... 217 Index...... 219 PREFACE

The current volume brings together papers presented at each of these disciplines is used at a level below its potential, the AGU Chapman Conference “Climates, Past Landscapes, and in the process, the complexity of phenomena is diluted. and Civilizations,” held in March 2011 in Santa Fe, New Solutions to bridging this divide are not simple, but they are Mexico. We wish to thank all participants and organizers of not beyond reach and may include a wider presentation of the conference and are grateful to all contributing authors, existing interdisciplinary research to multiple stakeholders reviewers, and editorial staff who have helped to produce this and funding agencies and cross-disciplinary working groups book. The meeting was attended by more than 100 scientists, within professional societies and dedicated meetings. Further scholars, and journalists across the fields of Earth sciences, progress may require integration and standardization of anthropology, and archaeology, with the overarching goal of databases and establishing new organizations dedicated to enhancing the cross-disciplinary dialogue on the history of interdisciplinary research on coupled socioenvironmental complex interrelationships between humans and their envir- systematics. Both the differences and convergence of opportu- onment. Discussions, thematic group sessions, and answers nities discussed during the Chapman meeting are reflected in the to individual questionnaires revealed differences among current volume and in other conference papers published outside disciplines on the design, methodology, and interpretation of this volume. research but also pointed out a strong collective interest to The early anthropogenic hypothesis provided a larger develop collaborative pathways toward bridging any perceived context for presentations at the meeting. The centerpiece of disciplinary divides. this hypothesis, developed by paleoclimatologist Bill Ruddi- Research on the history of interactions between humans and man in a series of papers since 2003, proposes that humans the environment is intrinsically interesting to diverse began to exert an influence over the global climate thousands audiences and engaging for the wider public. The fate of past of years ago through greenhouse gas emissions linked to cultures also presents us with completed intricate experiments the expansion of agriculture (see the introduction to this that provide a wealth of data for exploring models of the volume by Ruddiman and references therein). While archae- resilience and sustainability of coupled socioenvironmental ologists have always been concerned with the interactions systems. At a time when climate change, overpopulation, and between past cultures and their environment, the global scope scarcity of resources are increasingly affecting our ways of and fingerprint of these interactions suggested by the early life, the lessons of the past provide multiple reference frames anthropogenic hypothesis introduces a new level of complex- that are valuable for informing our future decisions and action ity in Earth sciences and provides a path for future plans. Despite this wide interest and investigational potential, interdisciplinary research. In the introduction to the present collaboration across disciplines is uncommon, and adequate volume, Ruddiman argues for an increased role of archaeology funding to explicitly support this style of interdisciplinary and anthropology in validating competing models of land use. research remains scarce. The two broad fields of inquiry, Earth Novel use of historical information on social organization and sciences and archaeology, have distinct customs and rhythms resources and regional and global assessments of the scale and of publishing and discussion of new ideas, and hypotheses are spatial distribution of past societies, as well as better criteria for mostly generated within nonoverlapping professional socie- discriminating between anthropogenic and natural landscapes, ties. Consequently, Earth scientists and archaeologists, experts are just a few elements that are critical for advancing this goal. in these fields analyzing the same phenomena at various Along similar lines, Cadzow argues in his chapter for research temporal and spatial scales, rarely overlap effectively in on the primary drivers of long-term environmental impacts planning and performing their research. Information outside such as hunter-gatherers and agriculturists rather than focusing only on major sites or civilizations. The impact of these “unsung” societies, although more diffuse, may have left a more profound (and not necessarily harmful) fingerprint on Climates, Landscapes, and Civilizations landscapes and ecosystems. Geophysical Monograph Series 198 The overlap of human agency and natural climate change © 2012. American Geophysical Union. All Rights Reserved. on landscapes and ecosystems is often confounding and, in 10.1029/2012GM001426 many cases, difficult to disentangle. The extinction of

vii viii PREFACE

Pleistocene megafauna and the termination of the Clovis A modeling approach is taken by Berking and colleagues, lithic technology/culture during the Younger Dryas is one who use downscaling of atmospheric general circulation example of such complex phenomena that has elicited an model results to investigate the rise and fall of the city of active debate recently. In this volume, Boslough and Naga along the middle Nile during the first millennium colleagues present new data and argue against a recent before the Common Era. hypothesis that proposed that a large impact or airburst The cause-effect relationships between climate and human caused simultaneous climate cooling, extinction events, and history are often nonintuitive, and multidisciplinary ap- cultural changes at the Younger Dryas around 12.9 ka. proaches are required to reconstruct them. An unexpected When climate changes can be detected and isolated from relationship between climate and the Siberian Scythians is anthropogenic overprints, high-resolution records can reveal revealed by Panyushkina, who posits that a decrease in new relevant aspects for the socioenvironmental systems at habitation in the Altai Mountains during warmer climate fine scales. Along this line, Berkelhammer et al. present a intervals may have increased mobility and possibly resulted high-resolution speleothem record of the monsoon regime in the development of transhumant pastoralism. In contrast, from northeast India and document for the first time the the civilizing value of prehistoric climatic stress leading to 4.2 ka climate event on the Indian subcontinent against which acculturation, social complexity, and relocation is discussed the reorganization of the Indus Valley civilization can be by Nicoll, who analyzes the effect of droughts on the assessed. Aharon and colleagues provide a high-resolution Neolithic culture at Nabta Playa, west of the Nile Valley, and climate reconstruction from a speleothem from DeSoto the rise of the Pharaonic culture in Egypt. Force and Caverns in Alabama and address the role of instability rather McFadgen discuss the role of active tectonics in landscape than singular climate events on the fate of Mississippian development over long time scales by providing diverse chiefdoms in the southeast United States. environments and also through creative destruction events Sea level changes represent a cumulative and more gradual that accelerate the development of cultural complexity. aspect of global climate variability, often with profound local A wide array of archaeobotanical, geoarchaeological, and effects on culture. Landscape formation in coastal settings is philological data is synthesized by Riehl et al. to analyze the tightly constrained by sea level variability, leading to the multiple drivers controlling the transformation of agricultural progressive development of habitability niches. This is systems in northern Mesopotamia. A combination of discussed by Amorosi and Morelli in their chapter on the environmental reconstructions, archaeological methods, and fate of Neolithic Cardium Pottery Culture in the Mediterra- evidence from historical documents is also employed by nean and by Rollet in his analysis of the advent of large-scale Thurston and Plunkett, who reconstruct the “invisible” history irrigated rice agriculture in the Fuzhou Basin of China. of human activity under continuous pasture cover in Northern Regional complexity requires synoptic reconstructions Ireland. In a review of resilience to storms in Caribbean island of climate changes and associated landscape responses for communities, Cooper highlights the importance of examining in-depth examination of their links to cultural events. High- cultural life cycles from a long-term perspective that brings resolution geospatial imaging, areal expansion of paleoen- forward the capacity for rapid recovery rather than strategies vironmental databases, their integration with archeological for robust resistance to disasters. Development of novel metainformation, and scenario-based modeling of coupled proxies, monitoring, and dynamical reconstructions are socioenvironmental systems are increasingly applied to advocated in several papers in this volume. Adding to the advance these research directions. Maemoku et al. couples increasing body of literature that recognizes sedimentary high-resolution terrain models with estimates of river flow materials as important components of the archaeological and chronologies of eolian landforms to provide constraints record, Baade discusses irrigation-linked anthropogenic soils on the interpretation of the Ghaggar-Hakra Valley as the in two contexts where their identification contributes to our lost, legendary Sarasvati River of the Indian Vedas. A understanding of regional cultural developments: the high common mechanism for regional climate change in the west Himalaya and coastal Peru. Draut et al. address the role of Asia and Indian monsoon domain is proposed by Staub- detailed monitoring of sand transport and vegetation in wasser, after analyzing the common pacing of reconstructed analyzing the effects of climate variability on landscape Indus River outflow and water column stratification in the dynamics and cultures in arid regions with eolian elements. Red Sea over the last 5000 years. Focusing also on South Gatti and Oppenheimer address extreme events such as the Asia, Lemmen and Khan model the transition to agriculture Youngest Toba eruption, using a modeling method to better in the Indus Valley region, taking into account the constrain the distribution of ash from the eruption while biophysical forcing factors as well as sociotechnological noting that the environmental effects remain largely un- innovation, migration, population, and subsistence changes. quantified due to the low resolution of the proxy data. GIOSAN ET AL. ix

Covering the range from collapse to transformation, both the “site scale” of the archaeology as well as at the cultural responses to climate and landscape change may regional scale. Furthermore, it is essential that archaeologists, include abandonment, redistributions, and reorganizations geologists, and environmental scientists working within the of settlement types and patterns, recalibration of food same region collaborate in data collection, assessment, and procurement strategies, trade network development, migra- synthesis. This will help temper disciplinary biases and tion and colonization, and technological transitions. Con- provide opportunities for new observations and theoretical tinuing discussions on this topic started at the conference, developments based upon a better understanding of the and in accord with many points raised in other papers in this complexities of human-environment interactions. volume, Aimers argues for developing and using dynamic models of cultural development in his analysis of the Liviu Giosan cultural transformations of the lowland Maya in the ninth Woods Hole Oceanographic Institution century A.D. The author underscores the need for close collaboration between Earth scientists and archaeologists to Dorian Q. Fuller understand the broad spectrum of cultural responses. University College London Research design should approach this complexity with an open mind that moves beyond deterministic assumptions Kathleen Nicoll about civilization collapse and instead focuses on under- University of Utah standing the resilience strategies involved in past cultural transformation. Rowan K. Flad One key issue that emerged at the conference as a partic- Harvard University ularly important focus going forward is the need for developing increasingly precise chronologies for both paleoen- Peter D. Clift vironmental and archaeological data. This is important at Louisiana State University Bridging a Disciplinary Gap

William F. Ruddiman

Department of Environmental Sciences, University of Virginia, Charlottesville, Virginia, USA

This paper examines a perceived dividebetweentwogroupsreferredtoas “archeologists” and “physical climate scientists.” The former group encompasses field scientists in related disciplines such as geoarcheology, archeobotany, and those aspects of sedimentology and paleoecology that focus on the Holocene. Also included are human geographers who study written historical records of civilizations spanning the last 2 millennia. The latter group covers those working in fields such as atmospheric sciences, paleoclimatology, paleoceanography, hydrology, ice core and marine geochemistry, numerical (general circulation) modeling, and carbon-cycle modeling. Few scientists on the two sides of this barrier have successfully bridged this gap, even though the potential benefits of doing so are considerable. This paper makes the case that climate scientists, trying to understand the middle and late Holocene, need to consider how the spread of agriculture transformed past landscapes and potentially altered regional and larger-scale climate. It also points to ways in which archeologists can benefitfromplacing regional studies in the kind of “big picture” view common in climate studies.

1. INTRODUCTION 2. FORGOTTEN MILL PONDS

Several AGU members organized and held a March 2011 In the 1950s, two highly respected hydrologists, Gordon Chapman conference with the title “Climate, Past Land- Wolman and Luna Leopold, investigated a region of second- scapes, and Civilizations.” Their intention was to bring to- and third-order streams in the mid-Atlantic Piedmont of the gether people in different disciplines to bridge disciplinary United States. Acting under the assumption that these barriers related to these three topics. From my perspective, streams were natural in origin, they formulated the under- the conference was a successful step toward an important lying physical laws that appeared to govern their behavior goal that deserves much more attention in the future. [Wolman and Leopold, 1957; Leopold and Wolman, 1960]. The ﬁrst two parts of this chapter focus on examples in This research played a prominent role in the development of which physical climate scientists have at times shown a the science of ﬂuvial geomorphology. surprising lack of awareness of ways in which our own Half a century later, Walter and Merritts [2008] reexa- species has altered the face of the planet, even in relatively mined the same region and found a strong human overprint recent centuries. The last section suggests ways that arche- on the landscape (Figure 1). Logging of the piedmont had ologists could help to bridge these disciplinary barriers and begun with the earliest European settlement in the late 1600s, provides examples of progress in that direction during the accelerated in the 1700s, and continued into the middle or last few years. late 1800s. Small water-powered sawmills were built along these streams at average intervals of a few kilometers to process cut timber. Mills also provided waterpower for grain Climates, Landscapes, and Civilizations grinding, textile production, paper production, and iron Geophysical Monograph Series 198 forges. Each mill had an upstream millpond dammed by © 2012. American Geophysical Union. All Rights Reserved. earthen structures to divert water for power. The piedmont 10.1029/2012GM001222 region was well suited to mills because of its gentle stream 1 2 BRIDGING A DISCIPLINARY GAP

Figure 1. Location of mid-1800s mill dams in eight mid-Atlantic counties based on historic atlases: in Pennsylvania, 1, Centre County; 2, Huntingdon County; 3, Cumberland County; 4, York County; 5, Lancaster County; 6, Chester County; and in Maryland, 7, Baltimore County; and 8, Montgomery County. From Walter and Merritts [2008] and Merritts et al. [2011]. gradients and location near navigable tidal water not far ble lowland forests were cut, later clearance removed forests downstream. from steeper and more remote hillsides, where erosion desta- Based on LIDAR (Light Detection And Ranging) surveys bilized the soil and sent fine sediment into the streams and that showed flat-lying areas (“valley flats”), historical re- millponds. But as steam power gradually replaced waterpower, cords, and field examinations, Walter and Merrits found most of the mills were no longer needed. And with no more evidence for at least 8000 mills and millponds along prom- forest to cut, the wooden and steel structures in the sawmills inent creeks and streams in the Pennsylvania piedmont, and were recycled to new locations farther west or south or to Merritts et al. [2011] also found numerous others across the higher terrain where timbering was still underway. With the Mid-Atlantic (Figure 1). Peak mill use occurred between sawmills gone, the millponds filled in with eroded sediment, 1780 and 1860, with more than 65,000 water-powered mills and with new forest taking over, the Piedmont began to in the eastern United States by that time. revert to what looked like a natural state. By the middle 1800s, more than a century of timbering A century later, when Wolman and Leopold studied the (and agriculture) had transformed the mid-Atlantic piedmont region, they interpreted the landscape as natural, analyzing into a largely humanized landscape. After the more accessi- the meandering patterns of piedmont streams flowing mostly RUDDIMAN 3 in single channels across the “valley flats” they thought were Several attempts have been made to quantify the history of floodplains. But Walter and Merritts showed that these land clearance [Houghton, 1999; DeFries et al., 1999; Ra- “floodplains” are actually the remnants of millponds that had mankutty and Foley, 1999; Goldewijk,2001;Joos et al., filled with sediment, in some cases breaching the mill dams. 2004; Pongratz et al., 2008; Strassmann et al., 2008]. Most The stream channels were meandering across these millpond of these studies were based on estimates that per capita deposits. cultivation has amounted to just a few tenths of a hectare to Based on historical accounts and on sediment deposits in a one hectare [Seiler and Crutzen,1980;Goldewijk,2001; few undammed regions that survive today, Walter and Mer- Ramankutty et al., 2002]. Several of these reconstructions ritts concluded that the natural streams that had previously made the key simplifying assumption that the same small per existed in the piedmont were small branching channels flow- capita clearance numbers applied back into the pre-industrial ing through forested wetland (sedge) meadows separated by Holocene. islands stabilized by alder trees and shrubs. Woody debris This critical assumption tied those land use reconstructions that jammed the stream channels helped produce these com- more or less linearly to estimates of the past population of the plex branching flows. Little fine sediment had accumulated planet (Figure 2). In the year 1500, global population was in those natural pre-clearance channels, although they were 450–500 million, or less than 7% of the modern value, but by rich in organic matter. 1850, it had risen to 900 million (~13% of modern), and now These remarkable papers serve as a stark warning to anyone we number 7 billion people. As a result, these land use studying the more distant past, including the effects of early reconstructions inevitably placed most global forest clear- Americans on the landscape prior to European arrival, as well ance within the population explosion of the industrial era, as the long history of land use in Eurasia. If hydrologists in the with very little clearance prior to about 1700. 1950’s had entirely “forgotten” the history from just one or This assumption of roughly constant per capita clearance two centuries earlier, what might have been “forgotten” about and land use sounds reasonable at first. More than once I things that happened several millennia ago? have heard it justified by a simple challenge: “Why would (early) farmers have farmed any more land than necessary to 3. FORGOTTEN (OR OVERLOOKED) feed their families?” The clear implication is that per capita PREINDUSTRIAL RECORDS OF LAND USE land use has had no reason to vary for thousands of years. But this constant land use assumption is demonstrably From direct personal experience, I offer here a second wrong. It is refuted by field-based studies in archeology, example of how scientists trained in physically based areas anthropology, and related disciplines, as well as by historical have been unaware of fundamental knowledge in disciplines evidence summarized below [Buck, 1937; Rackham, 1980; that have traced the imprint of humans on Earth’s surface Chao, 1986; Ellis and Wang, 1997; Mather and Needle, prior to the industrial era. Assessing the magnitude of this 2000; Williams, 2003; Bradshaw, 2004]. This evidence past human imprint has implications that extend into other shows that early farmers used far more land per capita than disciplines, including early emissions of greenhouse gases those in recent centuries. (CO2 and CH4) and their potential role in keeping late Ruddiman [2003, 2007] noted a striking historical example Holocene climate warmer than it would otherwise have been of the failure of the constant per capita assumption: a survey [Ruddiman, 2003]. of England ordered in 1086 by William the Conqueror. The

Figure 2. Estimated pre-industrial global population for the middle and late Holocene. Based on McEvedy and Jones [1978] and Denevan [1992]. Trend is truncated at 800 million people living at the end of the pre-industrial era (the early 1800s). 4 BRIDGING A DISCIPLINARY GAP results, recorded in the 1089 Domesday Book, counted a to 1800, but a few went back to the start of the European population of 1.5 million people and found that 85% of the historical era near 2000 years ago. These historical data arable land in Britain and parts of Wales and Scotland had again revealed surprisingly extensive early clearance long been cleared (Figure 3). Because this survey was done al- before the last few centuries. They found that forest clear- most 1000 years ago, its accuracy might be thought ques- ance does not track population in a one-for-one way but tionable. But Oliver Rackham, an exacting and meticulous instead follows a sigmoidal trend [Mather and Needle, botanist/paleobotanist, tested it using a range of innovative 2000; Kaplan et al., 2009]. Clearance accelerates quickly methods [Rackham, 1980]. Among other things, he consid- at low population densities and is nearly complete by the ered well-dated archeological remains of homes and villages, time population densities reach an intermediate level of 100 place names that constrain the founding of villages to partic- people per km2 (Figure 4). Additional population increases ular eras, and the degree of match between wooded areas have little or no effect, because most of the forests have recorded in the Domesday survey and modern woodlands already been cut. (generally a very close match). Rackham concluded that Historical data from China also conﬁrm that per capita forest clearance in 1089 was at least as high as the 85% value land use was much larger one or two millennia ago and had recorded in the Domesday Book and quite possibly a little decreased by the centuries just before the industrial era. higher. Working in pre-WWII Nanjing, Buck [1937] compiled land This Domesday example completely contradicts the pro- use trends during the last 2000 years across the entire agri- posed one-for-one link between population and clearance cultural area of east central China based on central dynastic (Figure 3). Today, roughly 60 million people live in Britain, administrative records (later updated and reﬁned by Chao, Wales, and Scotland, with ~80–85% of the arable land 1986). The survey covered both dry land crops like millet, cleared. Based on an assumed one-to-one relationship be- soybeans and wheat in the north, and irrigated rice in the tween population and clearance, the 1.5 million people alive south. The trend shows a decrease in per capita cultivated in 1089 should have cleared about 2% of the arable land. Yet area from 0.6–0.7 hectares per person nearly 2000 years ago the Domesday survey showed ~85% clearance, at least as to 0.15–0.2 hectares by the early 1800s (Figure 5). Ellis and much as today. Wang [1997] found a similar decrease in per capita cultiva- Other records of early forest clearance in other parts of tion during the last 1000 years in a county-sized region of Europe (France, Denmark, Sweden, Greece and Ireland) rice irrigation near the Yangtze River. were summarized by Mather and Needle [2000], Williams This range of evidence from Europe and China reveals a [2003], Bradshaw [2004] and previous studies cited in pervasive trend toward smaller per capita land use through those papers. Most of the records are from the years 1000 time, but why did this happen? The most likely answer

Figure 3. The assumption of constant per capita land use projects 2–3% forest clearance in Britain and Wales for the year 1089, but the Domesday survey shows 85% clearance at that time. RUDDIMAN 5

Table 1. The Boserup Sequence of Changing Population and Land Usea Changes Through Time (Long Fallow to Short Fallow to Annual Cropping to Double Cropping) Earlier Later Population density low high Labor required per acre low high Productivity per acre low high Per capita area farmed high low aBased on Boserup [1965, 1981].

and farmers moved to new plots and repeated the sequence. Plots were reoccupied (if at all) after lying fallow for several decades, long enough to allow significant recovery of soil fertility. Figure 4. Sigmoidal trend of forest clearance versus population This long-fallow phase of early agriculture used large density based on historical data in Europe [Mather, 1992; Kaplan amounts of land because of the cleared plots that were et al., 2009]. abandoned and left lying fallow. Each year’s actively farmed plot may have occupied no more than one hectare per person, comes from economist Ester Boserup. Decades ago, Boserup but the constant shifting from old to new areas left behind a [1965, 1981] proposed that a gradual shift in the style of large footprint of cleared forest long after each plot was agriculture occurred over many millennia because of popu- abandoned (Figure 6). Because abandoned plots take a few lation growth, innovation and the adaptation of new farming decades to develop even semi-mature trees, only a small skills (Table 1). fraction of total reforestation (and carbon sequestration) During the earliest “long-fallow” phase, farmers shifted would have occurred several decades after abandonment. constantly from plot to plot. They either cut down trees or These changes in above-ground carbon in vegetation were girdled them to stop the flow of sap in the outer bark layers. accompanied by similar changes in shallow soil carbon, with Then they set fire to the dead debris during dry seasons and initial loss after clearance, and slow recovery later. This used dibble sticks to plant seeds in the ash-enriched soil. sequence of footprints left behind in effect boosts the cumu- After a few years of growing crops, the soil fertility dropped lative area cleared per family to many hectares per person.

Figure 5. Per capita cultivation in China based on historical data [Buck, 1937; Chao, 1986]. 6 BRIDGING A DISCIPLINARY GAP

Figure 6. Cleared forest regrows slowly, leaving a “footprint” of abandoned land.

Over many millennia, the more reliable nutrition provided ~4 hectares per person 7000 years ago to less than 0.4 hectares by agriculture led to population growth. As this happened, just before the industrial era. In a parallel effort, Kaplan et al. claims on the locally available land began to increase, as [2009] used the historical evidence from Europe noted earlier farmers began to be hemmed in by neighbors. As farming and found that the population density needed to cause a families became constrained to smaller holdings, they were given fraction of forest clearance had increased through time forced to produce more food from their land. (in effect, the sigmoidal curve plotted in Figure 4 moved This shift to smaller plots was possible because new farm- from left to right through time). This trend indicates that ing methods were being adopted that gradual shortened the farmers gradually cleared smaller amounts of forested land fallow period from several decades to just a few years and per capita as time passed, again consistent with the Boserup also lengthened the interval during which each plot could be synthesis. actively cultivated before its nutrients were depleted. Even- Kaplan et al. [2011] used the historical evidence from tually, plots could be cultivated every year or even twice a Europe to estimate global clearance during the last several year. Key to these innovations was a new attention to enrich- millennia, with adjustments for higher primary productivity ing soil quality, at ﬁrst by mixing in grass and other available levels and longer growing seasons in the tropics. They vegetable debris, and later by spreading animal and human estimated gradually decreasing per capita land use in all manure. With these new techniques, farmers used less land regions, with a marked acceleration during the last 3000 but still kept food productivity high. years. Their simulated clearance shows much of China, This “Boserupian” view of the effects of population India, Mesopotamia and Europe largely cleared of forest growth on land use is widely known to scientists in the ﬁelds and other vegetation already cleared by 2000 years ago of archeology, human geography and anthropology, and it is (Figure 7a). The clearance simulated by Kaplan’s method often used as a standard of reference against which to assess matches the regions where Lewthwaite and Sharratt [1980] supporting or contradictory evidence [e.g. Grigg, 1979; Net- mapped organized cultures and well-developed agriculture ting, 1993; Turner and Shajaat, 1996]. Yet Boserup’s in- at that time (Figure 7b). Reconstructions of anthropogenic sights, and the published evidence for extensive early alteration of natural vegetation biomes by Ellis [2011] also clearance, was either unknown to, or ignored by, most nu- show major human overprints in these regions long before merical land use modelers. the industrial era. In contrast, a simulation based on one of Boserup’s view of past land use has recently gained re- the models that assumes small and nearly constant per capita newed attention. Ruddiman and Ellis [2009] used various clearance (in this case the HYDE model of Goldewijk sources to estimate the average decrease in per capita land [2001]) shows very little deforestation by 2000 years ago, use during the middle and late Holocene. They inferred that even in heavily populated areas that hosted highly advanced per capita land use fell by a factor of about ten from cultures (Figure 7c). RUDDIMAN 7

tion levels would imply. Fuller et al. [2011] also estimated an exponential growth of CH4 emissions that closely resembles and accounts for most of the CH4 trend measured in ice cores for the interval 5000 to 1000 years ago. In summary, a wide range of historical and archeological data shows that early land use was much higher on a per capita basis than during recent centuries. These results inval- idate previous land use simulations that were based on assuming a small, roughly constant amount of clearance. Future simulations will need to incorporate this evidence for changing land use values. Earlier, I mentioned the simple but deceptive challenge: “Why would (early) farmers have farmed any more land than was necessary to feed their families?” One response to this reasonable question is an equally simple counter-challenge: "Could farmers really have been so dim-witted that they learned nothing new about their basic livelihood during the last 7000 yrs, not a single new skill that enabled them to get more food from each acre of land they farmed?" Ester Boserup’s synthesis provides a more thoughtful response to this question by placing land use in the perspective of what people had to do to survive in a world where population was growing rapidly. With neighbors gradually crowding in, most farmers could no longer rotate freely from wooded plot to wooded plot when they chose. Slowly, they were forced to produce more food from the shrinking plots of land available, and eventually from small and constant land holdings. As populations grew, farmers no longer had any choice. Historical evidence further suggests a surprising reversal in deforestation trends during the industrial era. Mather [1992] found evidence that Britain, France, Sweden, and Switzerland had begun reforesting by 1850 and continued to do so through the entire industrial era (Table 2). Mather Figure 7. (a and c) Two simulations of vegetation clearance 2000 called this reversal from the ongoing deforestation of previ- years ago from Kaplan et al. [2011] compared to (b) map of ous millennia the “forest transition”. Reforestation has also agriculture based on archeological information (from Roberts occurred in western Russia since 1920, because of popula- [1998] adapted from Lewthwaite and Sherratt [1980]). tion losses during and since the two world wars, mortality caused by forced relocations of ethnic populations, and a In another recent effort, Fuller et al. [2011] synthesized archeological evidence on the spread of irrigated rice agriculture beginning near 5000 years ago. They found that by Table 2. Time of the Start of Industrial Era Reforestation (the “ ” 1000 years ago irrigated rice was being grown in every Forest Transition ) region in Asia where it is grown today. By assuming that the Country Start of Reforestation subsequent infilling of irrigation farming in each area fol- England/Wales 1750 lowed the log of population growth (based on the modern Sweden 1800–1830 spatial relationship), they reconstructed an exponential in- France 1825 crease in the area of irrigated rice after 5000 years ago. They Switzerland 1850 estimated that the area of irrigated rice had reached 35% of Hungary 1925 the modern level by 1000 years ago, even though the popu- Eastern United States, Canada 1900 lation was only 5–6% of modern. Once again, this evidence Russia 1920 suggests that early land use was far higher than the popula- China 1980s 8 BRIDGING A DISCIPLINARY GAP recent population decline. Net reforestation has also been this kind of labor are inevitably specific to that one site in that occurring in China since the 1980’s, in large part because of one region, although often the published results are usually directives from the central government.. Even eastern North compared to nearby sites to see how well they fit into larger America, which had been rapidly deforested as recently as regional-scale interpretations. the 1700s and 1800s, was already reforesting by the late What I have not found in my reading is much evidence of 1800s or early 1900s, as Walter and Merritts [2008] noted. efforts to integrate this hard-won site-specific information In many regions, reforestation occurred after 1850 because into a larger story at subcontinental to continental scales. mechanization of farm equipment made larger-scale, more Books like “First Farmers” by Bellwood [2005] very capa- cost-effective agriculture possible on fertile prairie and bly summarize general trends in words, but few such sum- steppe soils by enabling farmers to plow out deep-rooted maries have tried to take advantage of the larger-scale trends vegetation typical of semi-arid regions. that can now be revealed by mapping the enormous number In summary, early attempts by land use modelers to recon- of well-dated archeological sites available in most regions. struct the pre-industrial history of forest clearance and land One admirable exception (and a good example of what I use for agriculture greatly underestimated the extent of early mean) was the effort by Zohary and Hopf [1993] who used clearance. These modelers had “forgotten” (or never knew archeobotanical information from hundreds of well-dated about) key work decades ago that told a very different story. archeological sites in southwest Asia and Europe to map the first arrival of the fertile-crescent package of crops between 4. ARCHEOLOGICAL CONTRIBUTIONS 10,000 and 5,500 years ago. Turney and Brown [2007] later TO INTERDISCIPLINARY EFFORTS displayed this information in an informative way by showing sites color-keyed to different ages. The book The Holocene Because I am a marine geologist/paleoclimatologist, it by Roberts [1998] also added several regional-scale maps may be thought presumptuous for me to comment on whether from primary sources. or not similar barriers to interdisciplinary communication But otherwise, at least to my knowledge, these efforts exist within archeology and related fields. Nevertheless, I stood alone in attempting large-scale map integration of have enough recent personal experience, confirmed by infor- archeological and archeobotanical data. Much of what I mal discussions with a few archeological colleagues, to point learned about this larger picture came from reading two trade out at least one area where room for improvement seems to books, Guns, Germs, and Steel by Diamond [1999] and exist (and has now begun to occur). 1491,byMann [2006]. Although both were very informa- A decade ago, I became interested in the “big picture” tive, they had the relatively limited graphic treatments typical aspects of archeology: the large-scale story of how agricul- of trade books. ture developed in different regions and gradually spread Recently, the situation has improved considerably. For across the continents. Not long afterward, I also became example, Li et al. [2009] mapped archeologically dated sites interested in how the agricultural practices used to obtain in east central China between the interval from 8000 to 7000 food from the land changed through time. Taken together, I years ago and the one from 5000 to 4000 years ago. In those see these advances as one of the most fascinating “stories” 3000 years, the number of sites shows a 30-fold increase. our species has to tell, and arguably the most important. Because these people were mainly agriculturalists, their pres- Without these advances, the modern world we know would ence in naturally forested regions implies the start of major not have come into existence. forest clearance in China during this interval, along with My early investigations of past land use were unsystematic, increased CO2 emissions. but gradually I reached the point where I felt I was seeing the Two independent but convergent efforts [Gong et al., field more or less whole. So I offer the following comments on 2007; Ruddiman et al., 2008] mapped the spread of rice archeology from this decidedly non-expert outside perspective. paddy agriculture in China after irrigated rice was domesti- Despite all the admirable work done in this field, my cated near 6500 years ago. As noted earlier, Fuller et al. impression is that archeology and its closely related disci- [2011] has recently used archeobotanical criteria to refine plines are to some extent “balkanized.” By this, I mean the and extend these efforts to all of southern Asia for the field often lacks a “big picture” view, especially the kind that interval between 6500 and 1000 years ago. Fuller et al. comes from integrating discrete point-source information [2011] also used archeological data to map the spread of into larger-scale regional maps. I realize that most of the livestock across Asia and Africa during the last 7000 years, actual work of archeology is the product of long “dirty although they did not attempt to estimate methane emissions. boots” field seasons spent intensively examining a particular Previously, Chen et al. [2010] had mapped the spread of site of a certain age in a specific region. The first results of Zebu cattle across India. RUDDIMAN 9

As some of these examples show, synthesizing archeolo- Diamond, J. (1999), Guns, Germs, and Steel, W. W. Norton, New gical data is important for an interdisciplinary reason. These York. data sets can be used to test the hypothesis that the spread of Denevan, W. M. (Ed.) (1992), The Native Population of the Amer- agriculture long before the industrial era was already having icas in 1492, Univ. of Wis. Press, Madison. a major effect on greenhouse-gas emissions [Ruddiman, Ellis, E. C. (2011), Anthropogenic transformations of the terrestrial – 2003, 2007]. Because several regions still lack archeological biosphere, Philos. Trans. R. Soc. A, 369, 1010 1035, doi:10. syntheses of crop data, future efforts are needed to fill in the 1098/rsta.2010.0331. largest time/space gaps in current coverage. Another impor- Ellis, E. C., and S. M. Wang (1997), Sustainable traditional agriculture in the Tai Lake region of China, Agric. Ecosyst. Environ., tant interdisciplinary effort in the future will be to develop 61, 177–193. models that can transform discrete point-source archeologi- Fuller, D. Q., J. Van Etten, K. Manning, C. Castillo, E. Kingwell- cal data into integrated estimates of regional greenhouse-gas Banham, A. Weisskopf, Q. Ling, Y.-I. Sato, and R. J. Hijmans emissions. (2011), The contribution of rice agriculture and livestock to prehistoric methane levels: An archeological assessment, The 5. CONCLUSIONS Holocene, 21(5), 743–759, doi:10.1177/0959683611398052. Goldewijk, K. K. (2001), Estimating global land use change over As of 2011, scientific disciplines related to archeology and the past 300 years: The HYDE Database, Global Biogeochem. those covering various physically based sciences seem Cycles, 15(2), 417–433, doi:10.1029/1999GB001232. poised for a mutually beneficial and explosive growth in Gong, Z. T., H. Z. Chen, D. G. Yuan, Y. G. Zhao, Y. J. Wu, and G. L. areas of overlapping interest. Artificial barriers that have Zhang (2007), The temporal and spatial distribution and constrained cross-disciplinary exchanges have begun to the significance of ancient rice in China, Chin. Sci. Bull., 52, break down. 562–567. Grigg, D. (1979), Ester Boserup’s theory of agrarian change: A Acknowledgments. My horizons have been greatly expanded critical review, Prog. Human Geogr., 3,64–84, doi:10.1177/ by discussions with scientists knowledgeable about a wide range 030913257900300103. of aspects of past land use, particularly Erle Ellis, Dorian Fuller, Houghton, R. A. (1999), The annual net flux of carbon to the and Jed Kaplan. I thank Dorothy Merritts, Robert Walter, and atmosphere from changes in land use 1850-1990, Tellus, Ser. B, Mike Rahnis for providing Figure 1. I thank two anonymous 51, 298–313. reviewers, and I acknowledge support from the National Science Joos, F., S. Gerber, I. C. Prentice, B. L. Otto-Bliesner, and P. J. Foundation. Valdes (2004), Transient simulations of Holocene atmospheric carbon dioxide and terrestrial carbon since the Last Glacial Max- REFERENCES imum, Global Biogeochem. Cycles, 18, GB2002, doi:10.1029/ 2003GB002156. Bellwood, P. (2005), First Farmers: The Origins of Agricultural Kaplan, J. E., K. M. Krumhardt, and N. Zimmerman (2009), The Societies, Blackwell, Oxford, U. K. prehistorical and preindustrial deforestation of Europe, Quat. Sci. Boserup, E. (1965), The Conditions of Agricultural Growth, Allen Rev., 28, 3016–3034. and Unwin, London, U. K. Kaplan, J. E., K. M. Krumhardt, E. Ellis, W. F. Ruddiman, Boserup, E. (1981), Population and Technological Change: A Study C. Lemmen, and K. K. Goldewijk (2011), Holocene carbon of Long Term Trends, Univ. Chicago Press, Chicago, Ill. emissions as a result of anthropogenic land cover change, The Bradshaw, R. H. W. (2004), Past anthropogenic influence on Euro- Holocene, 21(5), 775–791, doi:10.1177/0959683610386983. pean forests and some possible genetic consequences, For. Ecol. Leopold, L. B., and M. G. Wolman (1960), River meanders, Geol. Manage., 197, 203–221. Soc. Am. Bull., 71(6), 769–793, doi:10.1130/0016-7606(1960)71 Buck, J. L. (1937), Land Utilization in China, Commercial Press, [769:RM]2.0.CO;2. Shanghai, China. Lewthwaite, J. W., and A. Sherratt (1980). Chronological Atlas, in Chao, K. (1986), Man and Land in Chinese History: An Economic Cambridge Encyclopedia of Archaeology, edited by A. Sherratt, Analysis, Stanford Univ. Press, Stanford, Calif. pp. 437–452, Cambridge Univ. Press, Cambridge, U. K. Chen, S., et al. (2010), Zebu cattle are an exclusive legacy of the Li, X., J. Dodson, J. Zhou, and X. Zhou (2009), Increases of South Asia Neolithic, Mol. Biol. Evol., 27,1–6, doi:10.1093/ population and expansion of rice agriculture in Asia, and molbev/msp213. anthropogenic emissions since 5000 YBP, Quat. Int., 202, DeFries, R. S., C. B. Field, I. Fung, G. J. Collatz, and L. Bounoua 41–50. (1999), Combining satellite data and biogeochemical models to Mann, C. C. (2006), 1491, Random House, New York. estimate global effects of human-induced land cover change on Mather, A. S. (1992), The forest transition, Area, 24, 367–379. carbon emissions and primary productivity, Global Biogeochem. Mather, A. S., and C. L. Needle (2000), The relationships of Cycles, 13(3), 803–815. population and forest trends, Geogr. J., 166,2–13. 10 BRIDGING A DISCIPLINARY GAP

McEvedy, C., and R. Jones (1978), Atlas of World Population Ruddiman, W. F., Z. Guo, X. Zhou, H. Wu, and Y. Yu (2008), Early History, Penguin, New York. rice farming and anomalous methane trends, Quat. Sci. Rev., 27, Merritts, D., et al. (2011), Anthropocene streams and base-level 1291–1295, doi:10.1016/j.quascirev.2008.03.007. controls from historic dams in the unglaciated mid-Atlantic re- Seiler, W., and P. J. Crutzen (1980), Estimates of gross and net gion, USA, Philos. Trans. R. Soc. A, 369, 976–1009, doi:10. fluxes of carbon between the biosphere and the atmosphere from 1098/rsta.2010.0335. biomass burning, Clim. Change, 2, 207–247. Netting, R. M. (1993), Smallholders, Householders: Farm Families Strassmann, K. M., F. Joos, and G. Fischer (2008), Simulating and the Ecology of Intensive Sustainable Agriculture, Stanford effects of land use changes on carbon fluxes: Past contributions Univ. Press, Stanford, Calif. to atmospheric CO2 increases and future commitments due to Pongratz, J., C. Reick, T. Raddatz, and M. Claussen (2008), A losses of terrestrial sink capacity, Tellus, Ser. B, 60, 583–603, reconstruction of global agricultural areas and land cover for the doi:10.1111/j.1600-0889.2008.00340.x. last millennium, Global Biogeochem. Cycles, 22, GB3018, Turner, B. L., II, and A. Shajaat (1996), Induced intensification: doi:10.1029/2007GB003153. Agricultural change in Bangladesh with implications for Malthus Rackham, O. (1980), Ancient Woodland: Its history, Vegetation, and and Boserup, Proc. Nat. Acad. Sci. U. S. A., 93, 1484–1491. Uses in England, Edward Arnold, London, U. K. Turney, C. S. M., and H. Brown (2007), Catastrophic early Holo- Ramankutty, N., and J. A. Foley (1999), Estimating historical cene sea level rise, human migration, and the Neolithic transition changes in global land cover: Croplands from 1700 to 1992, in Europe, Quat. Sci. Rev., 26, 2036–2041. Global Biogeochem. Cycles, 13(4), 997–1027, doi:10.1029/ Walter, R., and D. Merritts (2008), Natural streams and the legacy of 1999GB900046. water-powered milling, Science, 319,299–304, doi:10.1126/ Ramankutty, N., J. A. Foley, and N. J. Olejniczack (2002), People science.1151716. on the land: Changes in global population and croplands during Williams, M. (2003), Deforesting the Earth, Univ. of Chicago Press, the 20th century, Ambio, 31, 251–257. Chicago, Ill. Roberts, N. (1998), The Holocene, Blackwell, Oxford, U. K. Wolman, M. G., and L. B. Leopold (1957), River flood plains: Ruddiman, W. F. (2003), The atmospheric greenhouse era began Some observations on their formation, Geol. Surv. Prof. Pap. thousands of years ago, Clim. Change, 61, 261–293. 282-C, U.S. Geol. Surv., Washington, D. C. [Available at http:// Ruddiman, W. F. (2007), The early anthropogenic hypothesis: pubs.usgs.gov/pp/0282c/report.pdf] Challenges and responses, Rev. Geophys., 45, RG4001, doi:10. Zohary, D., and M. Hopf (1993), Domestication of Plants in the Old 1029/2006RG000207. World, Oxford Univ. Press, Oxford, U. K. Ruddiman, W. F., and E. C. Ellis (2009), Effect of per-capita land- use changes on Holocene forest clearance and CO2 emissions, W. F. Ruddiman, Department of Environmental Sciences, Uni- Quat. Sci. Rev., 28, 3011–3015, doi:10.1016/j.quascirev.2009. versity of Virginia, P. O. Box 400123, Charlottesville, VA 22904, 05.022. USA. ([email protected]) Arguments and Evidence Against a Younger Dryas Impact Event

M. Boslough,1 K. Nicoll,2 V. Holliday,3 T. L. Daulton,4 D. Meltzer,5 N. Pinter,6 A. C. Scott,7 T. Surovell,8 P. Claeys,9 J. Gill,10 F. Paquay,11 J. Marlon,10 P. Bartlein,12 C. Whitlock,13 D. Grayson,14 and A. J. T. Jull15

We present arguments and evidence against the hypothesis that a large impact or airburst caused a significant abrupt climate change, extinction event, and termination of the Clovis culture at 12.9 ka. It should be noted that there is not one single Younger Dryas (YD) impact hypothesis but several that conflict with one another regarding many significant details. Fragmentation and explosion mechanisms proposed for some of the versions do not conserve energy or momentum, no physics-based model has been presented to support the various concepts, and existing physical models contradict them. In addition, the a priori odds of the impact of a >4 km comet in the prescribed configuration on the Laurentide Ice Sheet during the specified time period are infinitesimal, about one in 1015. There are three broad classes of counterarguments. First, evidence for an impact is lacking. No impact craters of the appropriate size and age are known, and no unambiguously shocked material or other features diagnostic of impact have been found in YD sediments. Second, the climatological, paleontological, and archeological events that the YD impact proponents are attempting to explain are not unique, are arguably misinterpreted by the proponents, have large chronological uncertainties, are not necessarily coupled, and do not require an impact. Third, we believe that proponents have misinterpreted some of the evidence used to argue for an impact, and several independent researchers have been unable to reproduce reported results. This is compounded by the observation of contamination in a purported YD sample with modern carbon.

1Sandia National Laboratories, Albuquerque, New Mexico, USA. 7Department of Earth Sciences, Royal Holloway University of 2Department of Geography, University of Utah, Salt Lake City, London, Egham, UK. Utah, USA. 8Department of Anthropology, University of Wyoming, Laramie, 3School of Anthropology and Department of Geosciences, Uni- Wyoming, USA. versity of Arizona, Tucson, Arizona, USA. 9Earth System Science, Vrije Universiteit Brussel, Brussels, 4Department of Physics and Center for Materials Innovation, Belgium. Washington University, St. Louis, Missouri, USA. 10Department of Geography, University of Wisconsin-Madison, 5Department of Anthropology, Southern Methodist University, Madison, Wisconsin, USA. Dallas, Texas, USA. 11Department of Geology and Geophysics, University of Hawai‘i 6Department of Geology, Southern Illinois University, Carbon- at Mānoa, Honolulu, Hawaii, USA. dale, Illinois, USA. 12Department of Geography, University of Oregon, Eugene, Ore- gon, USA. 13Department of Earth Sciences, Montana State University, Boze- man, Montana, USA. Climates, Landscapes, and Civilizations 14Department of Anthropology, University of Washington, Seat- Geophysical Monograph Series 198 tle, Washington, USA. © 2012. American Geophysical Union. All Rights Reserved. 15AMS Radiocarbon Facility, University of Arizona, Tucson, 10.1029/2012GM001209 Arizona, USA. 13 14 YOUNGER DRYAS IMPACT EVENT

1. INTRODUCTION nature of the impactor and the impact mechanisms. Support for the hypothesis is drawn entirely from indirect evidence The earliest form of the “Younger Dryas (YD) impact and, paradoxically, from the lack of observed physical evi- hypothesis” was published by Firestone and Topping dence. For instance, the lack of any recognized impact struc- [2001] and was substantially extended in a book by Firestone ture of YD age in North America prompted speculation that et al. [2006]. Adopting some (but not all) of these earlier the impactor struck the Laurentide Ice Sheet or detonated as ideas, the YD impact hypothesis was formally presented in multiple fragments in the atmosphere. However, there is no 2007 [Firestone et al., 2007]. Firestone et al. [2007] pro- direct physical evidence that either of these events occurred posed their impact hypothesis as a unifying causal mecha- (see discussion in the following sections). Regarding pur- nism to explain the abrupt cooling that marks the onset of the ported markers of the impactor or products of the impact, YD at 12.9 ka, extinctions of Pleistocene megafauna, and a independent studies conclude that this physical evidence is presumed termination of the Clovis lithic technology/culture. irreproducible, misidentified, and/or misinterpreted as According to the impact hypothesis, one or more large, low- unique to only impact mechanisms. For a detailed review, density extraterrestrial objects exploded over the Laurentide see the work of Pinter et al. [2011a]. Ice Sheet around 12,900 years ago. The resulting shock wave The proposed nature of the hypothetical impact event was destabilized the ice sheet, causing the sudden cooling. Heat not uniquely defined by Firestone et al. [2007], but the from the explosion ignited wildfires across the North Amer- proponents seem to rule out an asteroid in favor of a low- ican continent. The combined effects resulted in profound density comet that may or may not have broken up, may or environmental change that forced a major ecological reorga- may not have exploded before it reached the surface, and if it nization and food deprivation that adversely affected mega- impacted, it did so at an oblique angle on thick ice. However, fauna as well as human populations. Firestone et al. [2007] estimate the impactor size by assum- As stated by Firestone et al. [2007], the impact hypothesis ing that it had effects over the entire continent. They argue is based on the interpretation of a purported carbon-rich that it needed to be more than 4 km wide and explode at the “black mat” stratigraphic layer and its constituent minerals, optimum height for blast damage at the surface. Moreover, which are presumed to represent a single horizon of the YD the authors suggest that it may have fragmented to produce a boundary. The term “black mat” refers to a wide array of late barrage of airbursts that generated continental-scale wildfires Quaternary wetland deposits and soils found in stream, lake, and destabilized the ice sheet, but produced no craters. and pond settings, primarily across the Southwest and Great A more recent YD impact paper [Israde-Alcántara et al., Plains of North America ranging in age from 6 to more than 2012] was coauthored by several of the original proponents 40 ka [Pigati et al., 2012]. The impact hypothesis is based of the impact hypothesis and proposes that the impactor upon seven indicators present in the black mat that are could have been either a comet or asteroid with an initial described as evidence for an extraterrestrial impact and as- diameter greater than several hundred meters. It may or may sociated biomass burning: magnetic grains associated with not have been fragmented, and its entry angle could have elevated iridium (Ir) concentrations, magnetic microspher- been as great as 30° from the horizontal. This hypothesis is a ules, charcoal, soot, carbon spherules, glass-like carbon, major departure from the Firestone et al. [2007] concept. The nanodiamonds, and fullerenes with extraterrestrial helium proposed object is now 3 orders of magnitude smaller in (3He). Other professed evidence includes Pt-group element terms of mass and energy, and unusual and low-probability peaks, geochemical evidence for biomass burning during the characteristics (low-density comet, fragmented state, low YD in Greenland ice cores, and radioactive animal bones at impact angle, ice sheet target) are not prescribed. the YD boundary. We begin our critique with a summary of the Firestone et Coauthors of the original work presented by Firestone et al. [2007] YD impact hypothesis. Although it is not well al. [2007] have subsequently published papers in support of constrained, the impact mechanism has variously been de- YD impact hypothesis variants [e.g., Kennett et al., 2008, scribed as an airburst, a cluster of airbursts, an ice sheet 2009b; Firestone, 2009; Firestone et al., 2010; Kurbatov et impact, and multiple continent-spanning impacts, there are al., 2010; Israde-Alcántara et al., 2012]. A number of key theoretical arguments that can be applied to the proposed YD details in previous versions conflict with later versions of the impact scenarios. hypothesis, but have not been abandoned. The YD impact The scenarios are inconsistent with our understanding of hypothesis is therefore not a single concise or coherent impact and airburst physics. There is no process capable of evolving explanation, but rather an overlapping set of multi- causing a comet or asteroid to break up or explode at suffi- ple hypotheses that are sometimes conflated. The YD impact ciently high altitude to lead to the claimed effects on the hypothesis proponents make a variety of inferences about the Earth’s surface, whereas an impact into the Laurentide Ice BOSLOUGH ET AL. 15

Sheet by a 4 km diameter comet would shock the underlying and Daulton et al. [2010] concluded that the same research rock strata, leaving an impact structure. Moreover, the prob- group misidentified graphene/graphane aggregates as lons- ability of the fragmented comet impact event is infinitesimal. daleite in YD boundary sediments. In addition, arthropod The combination of proposed size, configuration, and trajec- coprolites and fungal sclerotia in the sediments were mistaken tory of the putative impactor is exceedingly unlikely to have for fire-generated carbon spherules [Scott et al., 2010]. An occurred together as a single event in the entire history of the aliquot of carbon spherules that was provided by this group for Earth. the detection and verification of nanodiamonds was contami- We discuss three broad classes of evidence-based counter nated with a modern-aged carbon spherule. arguments. Taken together, these observations provide arguments 1. Evidence for impact is lacking. There are no impact against the idea that an impact was responsible for an envi- craters of the appropriate size and age. None of the classic ronmental catastrophe, extinction, and culture change at the markers of impact shock processing, unambiguously onset of the YD. shocked material (coesite, stishovite, shatter cones, etc.), and no meteoritic material from an impactor, have been found in 2. FUNDAMENTAL FLAWS any YD dated material. Proposed impact markers such as magnetic grains, soot, charcoal, and carbon spherules are not Even in the absence of counterevidence, we consider the diagnostic of impact and occur in other environments. The Firestone et al. [2007] impact hypothesis to be fatally flawed black mats in which these materials are found do not repre- because it violates physical conservation laws and is incon- sent a single stratigraphic horizon and, in many cases, have sistent with conventional understanding of impact physics. not been age dated. Suggested YD impact craters, the Car- Several subsequent elaborations of the hypothesis incorpo- olina Bays (unusual elliptical depressions across the Atlantic rate increasingly low probability events, leading to a pro- Coastal Plain from Georgia to Virginia) do not exhibit impact posed impact configuration that has only an infinitesimal crater morphology nor are they of YD age. probability. 2. Effects that the impact proponents are attempting to explain, namely, a cooling event, megafaunal extinctions, 2.1. Inconsistencies and Contradictions and termination of the Clovis culture, may not be instanta- neous or synchronous and may not have any common causal The scope of analysis in the present paper is limited to the link. Moreover, none of these transitions require impact as a hypothesis stated by Firestone et al. [2007] and subsequent cause. The abrupt cooling at 12.9 ka is of no greater magni- publications by coauthors. For context, it should be noted tude or climatological consequence than many such climate that the 2007 hypothesis is an outgrowth of the book “Cycle variations that occurred throughout the Quaternary [Broecker of Cosmic Catastrophes” by Firestone et al. [2006], in which et al., 2010]. Any special status of the YD derives from the the hypothesized comet was just one manifestation of a much fact that it was the last such event before the Holocene and is greater cycle of catastrophes initiated by a nearby supernova the most recent and therefore best characterized. The extinc- at 41 ka that subjected the Earth to cosmic and thermal tion event was not abrupt, and at least some members of the neutron radiation so strong that terrestrial materials still Pleistocene megafauna were already in decline before 12.9 ka remain radioactive. According to the 2006 book, the super- [Grayson, 2007; Faith and Surovell, 2009]. The YD time nova also generated an interstellar shock wave that reached frame does not mark the termination of the Clovis culture or the Earth at 34 ka and then a wave of cosmic debris at the a population collapse, but coincides broadly with a change in YD. The 2006 version of the hypothesis proposed a large lithic technology. impact that created the present-day Hudson Bay, generating 3. In our view, there has been a pattern of irreproducibility ejecta that formed the Carolina bays, which the authors and misinterpretation of evidence used to support a YD described as secondary impact craters. This event also desta- impact hypothesis. Independent investigators have failed to bilized the Earth’s magnetic field and exposed it to another confirm the reports of enhanced concentrations of spherules wave of particle irradiation at its surface. and platinum-group elements in YD boundary sediments. The title of Firestone et al. [2007, p. 16,016] asserts, Specimens collected during an expedition for the 2008 film- “Evidence for an extraterrestrial impact. . .” and according to ing of the PBS NOVA documentary “The Last Extinction,” the abstract, the authors “. . . provide evidence for an extraterres- that were purported to contain shock-formed (lonsdaleite) trial (ET) impact event at ~12.9 ka. . .” The list includes seven nanodiamonds [Kurbatov et al., 2010] have not yet been markers, all of which they say “are evidence for an ET impact made available for independent analysis. The data published and associated biomass burning at ~12.9 ka.” The Carolina by Kurbatov et al. [2010] are inconsistent with lonsdaleite, Bays are highlighted as “unique, elliptical depressions” in the 16 YOUNGER DRYAS IMPACT EVENT

2007 paper, but their relevance is unclear; they propose but do strength and density, entry angle, and velocity. Toon et al. not document evidence for 15 bays with impact-related markers [1997] plotted the burst height for various impactor classes and provide no stratigraphic data or numerical dating. The as a family of curves (Figure 1). For a given class of object previous conjecture that these features are secondary impact (every other variable held constant), there is only one size craters appears to have been abandoned in current versions of that will naturally explode at its optimum altitude and dam- the hypothesis. age a greater area on the ground than if it had exploded at a different altitude. It can be determined from the curve inter- 2.2. Physical Implausibility section points in Figure 1. The highest possible optimal airburst altitude for any object is defined as the point where There are three aspects to the hypothesis that are inconsis- the Glasstone and Dolan [1977] curve intersects with the tent with conventional physics: (1) a 4 km diameter comet long-period comet curve, corresponding to a 120 m comet at cannot possibly explode at its optimum height; (2) an object 15 km altitude. More recent work by Boslough and Crawford traveling at escape velocity or greater cannot sufficiently [2008] shows that the effective airburst altitude of a given- disperse fragments, even if it is disrupted at the Roche limit sized impactor is significantly lower than the Toon et al. (the distance from the Earth at which tidal forces exceed self- [1997] curves indicate, suggesting that the intersection point gravity of an orbiting or approaching object); and (3) for any in Figure 1 actually overestimates the highest possible air- plausible entry angle, even a 2 km impactor would not fail to burst altitude. penetrate a 2 km thick ice, leaving a crater in the underlying When the optimum altitude for creating blast waves strata. (which neglects the Earth’s curvature) is extrapolated, a Firestone et al. [2007] place constraints on the impactor 107 Mt explosion must be detonated at 500 km to generate (>4 km diameter comet generating 107 megaton (Mt) explo- continental-wide effects [Glasstone and Dolan, 1977]. sive event) based on the surface damage model of Toon et al. Whereas nuclear weapons can be set off at any altitude, there [1997] and cite the model to suggest that such an impact is is no physical mechanism that can cause a comet to explode capable of continental-scale damage. However, the Toon et in outer space (e.g., 500 km). Such an explosion would al. [1997] model invokes a crater-forming impact that ejects require the conversion of significant kinetic energy to inter- solid debris at high velocity that reenters as shooting stars nal energy for heating and vaporizing the comet. This would whose thermal radiation ignites wildfires. Simply stated, necessitate momentum loss through drag on the cometary such ballistic ejecta will not be produced in the absence of mass. In the absence of air, there is no mass to which crater formation. Crater-forming impact is a necessary com- momentum can be transferred, and such an explosion would ponent because that is how the ballistic ejecta are produced. violate the laws of physics. Therefore, one cannot use the However, there is no physical evidence of a crater in North America dating to the YD. To explain this lack, Firestone et al. [2007] argue that prior fragmentation prevented a crater from forming, but this argument contradicts prior assumptions invoked in the Toon et al. [1997] model. As such, the hypothesis is internally inconsistent. Firestone et al. [2007] also cite Toon et al. [1997] as noting that “. . .if airbursts explode with energy of 107 megatons at optimum height, they will cause blast damage over an area the size of North America. . .” Optimum height of burst is a concept from the nuclear weapons effects literature. It is the prescribed altitude for a point-source explosion to maximize surface damage. According to Glasstone and Dolan [1977], it is “that at which it is estimated a weapon of a specified energy yield will produce a certain desired effect over the maximum possible area.” For the YD impact hypothesis, the “desired effect” is damage due to blast waves. Optimum height of burst is a function of explosive yield, so it can be expressed as a locus of altitudes. The altitude at Figure 1. Airburst altitude for comets and asteroids [from Toon et which an object actually explodes is determined by physics al., 1997] with optimum height of burst [from Glasstone and Dolan, and also depends on other parameters such as impactor 1977]. BOSLOUGH ET AL. 17 optimum height concept to constrain the mass (and explosive experiments for which the crater formation is dominated by equivalence) of large comets. strength effects such as spall. It is inappropriate to extrapolate Firestone et al. [2007] also suggest that a debris shower from scales of millimeters to kilometers without accounting from a heavily fragmented comet “would have produced an for the change from strength-dominated to gravity-dominated airburst barrage that was similar to, although exponentially crater growth; for a detailed discussion of scaling laws and larger than Tunguska, while causing continent-wide biomass impact cratering, see the work of Melosh [1989]. burning and ice-sheet disruption.” The Tunguska event refers For high–strain rate phenomena at kilometer scales, the to the only unambiguous example of an observed naturally strength of the ice is negligible, and cratering is dominated occurring low-altitude airburst. It occurred in 1908 over by its hydrodynamic equation of state that is very similar to central Siberia and is estimated to have exploded with an that of liquid water. It is therefore instructive to consider an effective yield of 3 to 5 Mt [Boslough and Crawford, 1997]. oceanimpactasananalogtoanimpactintothickice. This description is consistent with the hypothesis presented Numerical models of such an impact were performed by at the 2007 AGU Joint Congress, where an animation from Shuvalov [2003] to determine the size of an impactor for the 2006 National Geographic documentary “Ancient Aster- which a bottom crater does not form in 4 km deep water. The oid” was shown in which a tumbling asteroid breaks apart as vertical impact of a 1.5 km stony asteroid excavated a clearly it approaches the Earth, resulting in an array of explosions pronounced 2 km deep crater in the ocean floor. According to and plumes across Southeast Asia. (The animation was pro- Shuvalov and Trubetskaya [2007], the cratering process deduced by TV6 Limited, a British production company, based pends critically on impactor diameter-to-water depth ratio on simulations by M. Boslough. It is available on Youtube: d/H.For0.11, the water column has press conference in Acapulco, May 2007, stating “. . .we little effect on the cratering process. The Laurentide Ice Sheet think a similar thing happened for this event.” (http://www. is estimated to be ~2–3 km thick during the YD period youtube.com/watch?v=I2ld-lohrPw#t=4m47s).) [Paterson, 1972]. Therefore, a 4 km diameter comet would For continent-wide effects, the fragmented comet would need to fragment into at least several thousand (<0.2–0.3 km need to significantly disperse over hundreds of kilometers diameter), well-separated, impactors during atmospheric entry along its trajectory to create widely space airbursts. However, in order not to shock the underlying rock upon impact on the there is no lateral aerodynamic force that can separate frag- ice sheet. This particular scenario is implausible. ments by a large distance between the upper and lower atmosphere. Likewise, no lateral force exists to accelerate 2.3. Infinitesimal Likelihood pieces apart between the Roche limit and atmospheric entry. Fragments of a broken comet would drift apart at a speed of The burden of evidence for a hypothesis increases with its tens of centimeters per second if spun apart by tidal disrup- a priori improbability. Firestone et al. [2007] invoke an tion, even if boosted by release of volatiles. In the 10 min or extraordinary sequence of events that compounds improba- so between fragmentation and impact, fragments would be bility to the point of virtual impossibility. This puts an separated by much less than the initial diameter of the object. extraordinary onus on the proponents to show that there are The impact of such a tight cluster would be indistinguishable no other explanations compatible with the evidence. Impacts from a single impact of a lower-density object because the on the Earth became rare events after its initial planetary total mass and kinetic yield are the same, regardless of the formation and the late heavy bombardment period. The best fragmentation state of the impactor. Greater fragment sepa- method to estimate the current impact flux on the Earth is by ration would require radial velocities that are a significant using the observed population of near-Earth objects (NEOs) fraction of escape velocity. There is no source of energy that [Stuart, 2001; Harris, 2002], which are in orbits that can can provide sufficient radial acceleration. bring them close to the Earth. Observed bolide frequencies Additionally, Firestone et al. [2007] cite unpublished data constrain the population of small objects [Brown et al., suggesting that a low-impedance layer such as an ice sheet 2002]. The full population can be used to determine the would minimize cratering in the underlying target rock. They probability of impact per year as a function of size (Figure 2). argue on this basis that multiple 2 km objects could strike the This can be converted to mean impact interval. A 4 km 2 km thick central zone of the Laurentide Ice Sheet at a asteroid collides with the Earth, on average, about once every shallow angle, leaving little evidence of craters other than 14 million years [NASA NEO Science Definition Team, 2003, depressions in the Great Lakes or Hudson Bay. Firestone et Table 3.1], which is a number that is consistent with lunar al. [2007] based their argument on laboratory-scale impact crater counts [Werner et al., 2002]. The flux can also be 18 YOUNGER DRYAS IMPACT EVENT

Figure 2. Observed population of near-Earth objects [NASA NEO Science Definition Team, 2003] with updates by A. Harris (personal communication, 2012). converted to a probability density function that indicates the Definition Team, 2003, Table 3.6]. The timescale for signif- probability that an impact of an asteroid of a given size will icant reduction in the population of NEOs by orbital interac- be the largest in a specified time. The best estimate for the tions is millions of years [Hut et al., 1987], so current fluxes largest impact of the past 20 ka is about 250 m (Figure 3). are an upper-bound limit for determining the probability of The population of comets in the kilometer size range in the impact at the YD. inner solar system is about 1% that of asteroids (D. Yeomans, Each attribute of the hypothetical impact scenario can be Jet Propulsion Laboratory, personal communication, 2011), assigned an approximate a priori probability to arrive at an and the mean impact interval for impact by a 4 km comet is order-of-magnitude estimate. The following factors are about once every 2.5 billion years [see NASA NEO Science rounded to the nearest order of magnitude and are not

Figure 3. The largest asteroid impact expected over a 20,000 year interval is 250 m. Impact by an object larger than 2 km is exceptionally unlikely (probability less than 1%). BOSLOUGH ET AL. 19 intended to be precise estimates. Given the known flux, the mammalian extinctions were abrupt, environmental change probability that a NEO with a kinetic energy of 107 Mt will caused the collapse of the Clovis culture, and continental- collide with the Earth in any given 13,000 year period is scale wildfires took place, all of which are lacking evidence. about 0.001 (once every 14 million years on average), and The most frequent criticism of the evidence for an impact is the probability of a comet impact with the same kinetic based on irreproducibility and misinterpretations of marker energy is about 0.00001 [NASA NEO Science Definition evidence as preserved in black mat deposits, which may not Team, 2003]. The ratio of the area of the Laurentide Ice Sheet correlate chronostratigraphically across regions. The most target to the Earth’s surface area was about 0.01. The fraction comprehensive review is that of Pinter et al. [2011a]. Be- of comets that are broken is less than 0.01 (A. Harris, Space cause of space limitations, we only address the key criteria Science Institute, personal communication, 2011, estimates related to first-order impact evidence. 0.001 based on crater chains on the Galilean satellites). The fraction of time the fragments of a broken comet remain 3.1. No Impact Markers or Crater within an Earth diameter is roughly 0.001 of the broken comet’s lifetime (based on the dispersion of Shoemaker- Various versions of the 2007 hypothesis give the following Levy 9 after it fragmented in 1992). The fraction of objects reasons for the absence of an impact crater for the hypothe- that collide with the Earth at a grazing angle less than 4° is sized YD impact event: (1) it was similar to, but larger than about 0.01, based on the distribution of isotropic collisions the 1908 Tunguska explosion, which did not produce a with a sphere with a cumulative probability of 1 À cos (2θ). crater; (2) the crater or craters were in ice, which melted Therefore, the a priori probability of the putative YD impact away and destroyed any impact features in the substrate; and of a comet in the defined configuration on the Laurentide Ice (3) the Chicxulub crater was not recognized for 10 years after Sheet during the prescribed time period is about 10À15. The the Alvarez et al. [1980] impact hypothesis, and so there is no expected recurrence interval for such an event using current expectation of immediate discovery of a YD crater. For flux rates is about 1019 years or about 109 times the age of the reasons outlined earlier, however, it is not physically possible universe. As such, the probability of this occurring is infin- for a 4 km comet to generate a Tunguska-like airburst, itesimal and is an event that is so improbable it can be called eliminating the first reason. The estimated kinetic yield of “statistically impossible.” the Tunguska explosion is 3–5Mt[Boslough and Crawford, It is worth noting that the Cretaceous-Tertiary (K/T) impact 1997], and the putative 4 km, 107 Mt comet would need to scenario, using the same line of argument, is not a low- break into around a million fragments to avoid producing a probability event. The mean recurrence interval for a 10 km single crater-forming event. However, in the case of a disin- diameter asteroid impact is about 100 million years, and there tegrating comet, the individual fragments would not be sep- are no special requirements (fragmentation state, impact an- arated by enough lateral distance to avoid reinforcing one gle, or a priori location or time interval) that make the K/T another. They would effectively become a single large low- event unusual within the last 100 million years. A subset of density crater-forming impact, instead of one that was capa- the new YD impact scenarios recently published by Israde- ble of affecting areas across multiple continents. Alcántara et al. [2012] are likewise unremarkable in terms of It should be noted that the Chicxulub crater has an age of probability because of the much smaller size and lack of a 65 Ma, more than 5000 times older than the YD interval. A prescribed unusual configuration and composition. YD impact structure should be “fresh” because erosion processes have not had sufficient time to operate during the short 3. EVIDENCE-BASED ARGUMENTS time that has elapsed since the YD. A good example of a relatively fresh impact crater is Barringer (Meteor Crater), Many papers presenting arguments against the YD impact Arizona, which at 49 ka is more than three times as old as the hypothesis are based on lack of evidence [e.g., Pinter et al., proposed YD impact and was formed by a solid impact of 2011a]. The most striking lack of physical evidence for the less than one millionth the energy [Melosh and Collins, 2005] YD impact is the absence of any impact crater of the appro- of the event postulated by YD impact proponents. Meteor priate age. Regardless of which version of the hypothesis is Crater is an unambiguous example of a well-preserved im- being considered, multiple small impacts and airbursts across pact structure [Grieve and Therriault, 2004; French and two continents or large impacts into ice, the lack of a crater or Koeberl, 2010], but not a single similar geomorphic fea- other geomorphic evidence is fundamentally problematic. ture from an impact event is known for YD time [Grieve, Another core problem is the presupposition that there are 1997; Earth Impact Database, Planetary and Space Science events that an impact hypothesis is needed to explain. A Centre, unpublished data, 2011, http://www.passc.net/Earth prerequisite set of hypotheses is that the YD was unique, ImpactDatabase/index.html]. The proposed YD impactor of 20 YOUNGER DRYAS IMPACT EVENT

Firestone et al. [2007] would have had enough mass to 3.2.2. Extinctions. Environmental change in the late Pleis- produce more than a million Meteor Craters, and even the tocene was a very complex process. The extinction of mega- much smaller Israde-Alcántara et al. [2012] YD impactor fauna was one component of this process, but the relationship would be capable of producing thousands. of the extinctions to climate change is far from clear [e.g., Another argument used to support the YD impact hypoth- Barnosky et al., 2004; Grayson, 2007; Faith and Surovell, esis is the claim of evidence for cratering in the Great Lakes 2009]. For example, some late Pleistocene mammal popula- basin. Firestone et al. [2007] state that “if multiple 2 km tions were in decline before the YD, and the timing of the objects struck the 2 km thick Laurentide Ice Sheet at <30°, extinction varied among some species and across continents they may have left negligible traces after deglaciation. . . [Grayson and Meltzer, 2003]. The dung fungus (Sporormiel- [perhaps] limited to enigmatic depressions or disturbances in la) proxy for megafauna presence indicates that some Pleis- the Canadian Shield (e.g., under the Great Lakes or Hudson tocene megaherbivores declined from 14.6 to 13.7 ka, well Bay).” However, by 12.9 ka, only the Lake Superior basin before the proposed impact; the precise chronology of ex- was still under glacial ice [Dyke et al., 2003]. Firestone et al. tinction remains unresolved [Gill et al., 2009]. [2010] suggested that “deep holes” beneath four of the Great Lakes could represent impact craters. They dismiss the pos- 3.2.3. Stratigraphy. Intensive study of the chronostrati- sibility that these holes were the result of glacial erosion, graphic and sedimentological details of the black mat citing nineteenth century research [Dawson, 1891], despite layers and associated paleontological evidence of multiple more than a century of glacial and Quaternary geological YD sites reveal that there are no continent-wide or region- research on the Great Lakes. If these holes were caused by al stratigraphic marker beds, and there are no other indian impact at 12.9 ka (and Firestone et al. [2010] provide no cators of widespread geomorphic or ecological instability evidence the holes are that old), then the impacts produced at 12.9 ka [Haynes, 2008; Meltzer and Holliday, 2010; elongated craters at different orientations. However, each hole Straus and Goebel, 2011, papers]. The fidelity of the YD is parallel to local ice flow in the up-ice end of its lake basin. time interval in black mats is questionable because they Israde-Alcántara et al. [2012] also mention a “crater are time-transgressive units, especially in western North named Corossol in the Gulf of St. Lawrence, containing America [Quade et al., 1998; Haynes, 2008; Haynes et al., basal sedimentary fill dating to 12.9 ka.” The cited study 2010] and can result from a wide array of environmental [Higgins et al., 2011], however, based the date of ~12.9 ka drivers [Meltzer and Holliday, 2010]. More relevant here, on extrapolation from dates higher in the section and suggest stratigraphic analyses conducted by Surovell et al. [2009] that the crater formed sometime between 12.9 ka and the end and Paquay et al. [2009] have failed to reproduce patterns of the Ordovician, a dating uncertainty that spans hundreds of purported YD impact marker indicators in documented of millions of years. stratigraphic contexts, including the same sections reported by Firestone et al. [2007]. More recently, Pigatietal. 3.2. Nonunique and Misinterpreted Events [2012] found that similar markers are ubiquitous within or at the base of black mats of non-YD age. One of the major flaws with the impact hypothesis is the leading premise that climate changes, faunal extinctions, and 3.2.4. Archaeology. The late Pleistocene, in general, and cultural transitions occurring during the YD require a com- the YD interval, in particular, were times of rapid environ- mon explanation. mental changes, but the timing, magnitude, and direction of those environmental changes varied across North America 3.2.1. Climate change. There was indeed an abrupt cli- [Meltzer and Holliday, 2010; see Straus and Goebel, 2011]. mate change around 12.9 ka, but the abruptness of the YD is Some researchers speculate that environmental shifts during not unique, and the cooling is of no greater magnitude or the YD affected Paleoindian populations, especially the Clo- climatological consequence than many other events occur- vis culture [e.g., Haynes, 1991, 2007]. Archaeological, geo- ring throughout the Quaternary [Broecker et al., 2010]. chronological, and stratigraphic data, however, do not Dozens of similar rapid climate-change events, such as provide strong evidence linking YD changes with significant Dansgaard-Oeschger and Heinrich events have taken place adaptations or to a demographic collapse of Clovis popula- over the last 100 ky, yet none of these are associated with tions [Holliday and Meltzer, 2010]. impacts. Such events recur at a much greater frequency than large (km-scale) impacts and do not require impacts as a 3.2.5. Paleoclimate/Paleoecology. The YD impact hy- causal mechanism. The YD is notable simply because it is pothesis proposes that continental-scale wildfires resulted. recent and the last such climate event before the Holocene. North American charcoal and pollen records do show rapid BOSLOUGH ET AL. 21 changes in environment, vegetation, and fire regimes, con- However, subsequent efforts by others to replicate claims of sistent with the climate changes during the last glacial- high Ir concentrations and high magnetic grain concentra- interglacial transition ~15 to 10 ka [Meltzer and Holliday, tions have failed. Surovell et al. [2009] demonstrated that 2010; Marlon et al., 2009; Pinter et al., 2011b]. The records magnetic grains are ubiquitous throughout the same strati- over the YD interval do not, however, indicate continental- graphic sections. There is no “Ir spike” in concentration at scale wildfire at any time. Instead, the record demonstrates the YD; the highest Ir concentrations measured by Paquay et that large scattered fires occurred asynchronously through- al. [2009] was ~100 ppt, which is well below the ppb levels out the transition, peaking around 13.2 ka and again at reported by Firestone et al. [2007]. In addition to making Ir 11.6 ka (when no comet impact is proposed) [Marlon et al., measurements, Paquay et al. [2009] also measured osmium 2009]. The pattern of increased biomass burning in response (Os) isotopes on bulk sample splits collected from the same to both abrupt warming and cooling is replicated by data on measured sections presented in the Firestone et al. [2007] biomass-burning variations during Greenland interstadials paper, and large quantities were used to avoid nugget effects (Dansgaard/Oeschgar events), stadials, and Heinrich events (artificially high apparent concentrations that arise when a over the interval from 85 to 15 ka [Daniau et al., 2010]. small sample includes an enriched grain). Os is a very sensitive tracer for extraterrestrial components in sediments, yet 3.3. Irreproducibility and Misinterpretations of Evidence all samples showed Os isotopic ratios similar to average crustal values. From these results, it was concluded that there Firestone et al. [2007] list seven professed markers that they is no significant meteoritic Os contribution to these YD interpret as evidence for extraterrestrial impact and associated sediments, therefore excluding the involvement of all mete- biomass burning: magnetic grains with Ir, magnetic micro- orite classes of chondritic nature [Paquay et al., 2009]. spherules, charcoal, soot, carbon spherules, glass-like carbon containing nanodiamonds, and fullerenes with extraterrestrial 3.3.2. Magnetic Microspherules. Magnetic microspherule 3 He. Surovell et al. [2009] and Paquay et al. [2009] conducted abundance results published by the impact proponents have independent assessments to identify some of these markers in not been reproducible by other workers. Analyses of the samples from key YD sites; they could not reproduce the same YD site stratigraphy by Surovell et al. [2009] could findings published by Firestone and other YD impact propo- not replicate observations for two of the impact markers nents. In addition to this problem reproducing the results, the published by Firestone et al. [2007]. The study by Surovell “putative markers for impact” and those proxies related to et al. [2009] found no peaks of abundance unique to the YD wildfire are conflated. An impact does not necessarily cause time interval. significant biomass burning, and evidence for a fire (such as charcoal and soot) is not a diagnostic indicator of an impact. 3.3.3. Carbonaceous Spherules. Microspherules are fairly Subsequent papers by the proponents of the YD impact abundant and occur sporadically throughout late Quaternary event include the reported discovery of hexagonal diamonds sediments. At least some of the supposed impact-derived (lonsdaleite) in sediments from Santa Rosa Island, California spherules are in fact fungal sclerotia [Scott et al.,2010].In [Kennett et al., 2009b], from the Greenland ice sheet [Kur- addition, new radiocarbon dates on carbon spherules cast doubt batov et al., 2010], and from lake sediments in Mexico on the provenance of YD boundary samples (see next section). [Israde-Alcántara et al., 2012]. However, Pinter et al. [2011a] reviewed most of this evidence in detail, as well as related 3.3.4. Charcoal and Soot. Evidence for fire exists at the claims that had already been largely rejected by the scientific YD, but fire is not a unique evidence for impact [van der community, including particle tracks in archeological chert, Hammen and van Geel, 2008; Bowman et al., 2009]. Peros et meteoritic particles embedded in Pleistocene bones [Hagstrum al. [2008] demonstrate that there were wide fluctuations in et al., 2010], an impact origin for Carolina Bays, and enrich- the Pleistocene paleoecology and fire history, including the ments of radioactive materials. Pinter et al. [2011a] evaluated continental-scale vegetation response to rapid climate the 12 original lines of evidence invoked to support an extrater- change, competition, and disturbance. The idea of the char- restrial impact [Firestone et al., 2007] and concluded that 7 of coal data providing evidence of a high-intensity fire is also the 12 are irreproducible, and the other five are consistent with flawed. No temperature data based upon charcoal reflectance noncatastrophic and/or terrestrial processes, as outlined below. [Scott et al., 2010] is provided. In addition, for reasons discussed above, it is highly implausible that there would be 3.3.1. Magnetic Grains With Iridium. Firestone et al. a continent-wide wildfire. Continental-scale wildfires have [2007] reported elevated concentrations of Ir in bulk sediments been dismissed for the K/T impact [Belcher et al., 2003, and in magnetic grains only from YD boundary sediments. 2005, 2009; Belcher, 2009]. 22 YOUNGER DRYAS IMPACT EVENT

3.3.5. Fullerenes with 3He. Fullerenes (carbon allotropes al. [2009a, 2009b] failed to find nanodiamonds [Daulton et in the topological form of closed caged structures) have been al., 2010]. While it is possible that cubic nanodiamonds were reported in YD black mat deposits. The fullerenes, them- heterogeneously distributed within carbonaceous materials selves, do not provide a diagnostic indicator of an impact in YD boundary “black mat” sediments and not present in the event since they form terrestrially, e.g., from wildfire [Hey- limited samples available to Daulton et al. [2010], the pres- mann et al., 1994], and they have been identified in candle ence of cubic diamonds is irrelevant to the YD impact soot [Su et al.,2011].Furthermore,claimsofmeteoritic hypothesis as discussed above. fullerenes isolated from stratigraphic impact horizons have Lonsdaleite, on the other hand, is often associated with been repeatedly challenged [e.g., Taylor and Abdul-Sada, shock pressures related to impacts where it has been found to 2000; Braun et al., 2001; Buseck, 2002]. Nevertheless, it occur naturally, [see Bundy and Kasper, 1967; Hanneman et was not the fullerenes, themselves, but what they contained, al., 1967; Frondel and Marvin, 1967; Erlich and Hausel, that offered intriguing evidence. It was reported that full- 2002]. Therefore, the presence of lonsdaleite in sediments erenes contained an extraterrestrial-trapped noble gas signa- suggests (but does not necessary prove) that materials have ture enriched in 3He relative to terrestrial compositions been shocked. The Russian literature reports the occurrence [Firestone et al., 2007]. However, these results have never of lonsdaleite within metamorphosed and metasomatically been replicated, and the original study [Becker et al., 1999] modified rocks of the Kumdykol diamond deposit in North has been criticized for a number of years for methodological Kazakhstan [Shumilova et al., 2011], as well as in titanium shortcomings and nonreproducible results [Farley and Mu- placers of the Ukrainian shield, diamond placers in Yakutiya, khopadhyay, 2001; Buseck, 2002; Farley et al., 2005]. and eclogites in Sal’niye Tundra, Kola Peninsula, and the Urals (for a review, see the works of Kaminsky [1994] and 3.3.6. Nanodiamonds. With many of the proposed impact Erlich and Hausel [2002]). markers encountering strong skepticism, proponents of a YD Relevant to the YD impact hypothesis is whether or not impact have increasingly focused upon reporting the pres- lonsdaleite is present in YD-aged materials (while absent in ence of abundant nanometer-sized (2 to 300 nm) diamonds overlying and underlying sediments). Daulton et al. [2010] (cubic and hexagonal) in purported YD boundary sediments demonstrated that previous studies of YD boundary sediments and carbon spherules at multiple localities across North [Kennett et al., 2009b] misidentified graphene/graphane ag- America and in Greenland ice [Firestone et al., 2007; Ken- gregates (ubiquitous in several types of carbon-rich materials nett et al., 2009a, 2009b; Kurbatov et al.,2010;Israde- from sediments) as lonsdaleite. Further, Tian et al. [2011] Alcántara et al., 2012]. While cubic and hexagonal (lonsda- found no evidence of lonsdaleite in Belgium YD boundary leite) diamond have been found in shock metamorphosed sediments. The high-resolution (HR)-lattice image of a na- meteorites and are associated with terrestrial impact struc- nocrystal from residues of Greenland ice that was used to tures, cubic diamonds are well known to occur in terrestrial identify lonsdaleite by Kurbatov et al. [2010] is crystallo- rocks that have no association with impact processes. Sub- graphically inconsistent with lonsdaleite (or cubic diamond, micron and smaller-sized cubic diamond crystals have been graphite, graphene, and graphane) and must be a nondia- recently demonstrated to exist in carbon spherules within mond (and possibly noncarbon) mineral. The published surface soils sampled from various sites in Germany and HR-lattice image of the nanocrystal identified as lonsdaleite Belgium [Yang et al., 2008]. While the origin of these dia- displays lattice fringes with two sets of 0.206 nm spaced monds remains unclear, they were evidently not produced by planes (crossing one another at 62° ± 2° as measured from impact processes because they are present in modern soil and the work of Kurbatov et al. [2010], Figure 6) and one set of lack any links to impact structures. Consequently, the value 0.193 nm spaced planes. For lonsdaleite, these lattice spacings of cubic diamonds as impact markers is highly suspect. correspond to the {002} and {101} planes, respectively. Israde-Alcántara et al. [2012] quote Tian et al. [2011] as However, no zone axis (i.e., crystallographic direction) of lons- independent confirmation of cubic nanodiamonds in YD daleite exists that can display two different sets of 0.206 nm boundary sediments; however, they do not mention that only spaced {002} planes. Further, the electron diffraction pattern a limited range of sediment horizons above and below the of Kurbatov et al. [2010, Figure 6], identified as lonsdaleite, Belgium YD boundary were studied in that work. Further, lacks a spatial calibration scale and is also consistent with they do not mention the results of Yang et al. [2008], which, graphite. taken together with Tian et al. [2011], suggest that nanodia- The work of Israde-Alcántara et al. [2012], coauthored by monds may be distributed throughout the Belgium sediments. members of the previous studies in which lonsdaleite was Another independent study of YD boundary sediments misidentified, reported the presence of lonsdaleite in purported sampled from the same collection sites as that of Kennett et YD-aged lake sediments in Mexico (although the dating has BOSLOUGH ET AL. 23 been challenged, see next section). This identification is problematic in that it is based on a fast Fourier transform (FFT) of an HR-lattice image of a nanocrystal that is not imaged along a high-symmetry zone axis. Only one set of lattice planes is discernible in the HR-lattice image [Israde- Alcántara et al., 2012, Figure 8]. Provided the weak ~2.16 Å peak in the FFT is not an artifact, the FFT is consistent with the lonsdaleite structure. However, a nonexhaustive search of the American Mineralogist Crystal Structure Database and Materials Data Incorporated (MDI) JADE database yielded the following materials (unit cell parameters are in parenthe- ses; zone axis, plane spacings, and angle between planes are in wavy brackets) largely consistent with the FFT: Achavalite- FeSe (a: 3.61 Å, c: 5.87 Å) {[2-21], 2.14 Å, 1.81 Å, 53.6°}; Algodonite-Cu6As (a: 2.6 Å, b: 4.23 Å) {[101], 2.24 Å, 1.98 Figure 4. Carbon spherules collected from Gainey, Michigan, pro- Å, 63.8°}; Mn2AsSb (a: 3.84 Å, c: 5.78 Å) {[2-21], 2.18 Å, vided by A. West. 1.92 Å, 55.4°}; and CrSb0.5As0.5 (a: 3.81 Å, c: 5.718 Å) {[2- 21], 2.16 Å, 1.91 Å, 55.5°}, all of the P63/mmc (194) space group. We are not suggesting that Israde-Alcántara et al. No reason is given for rejecting these, other than that they are [2012] mistook these particular phases as lonsdaleite; we cite older dates than predicted by their age model. The authors these examples to demonstrate that other materials (oriented invoked reworking of the deposits to explain the anomalous along various zone axes) are consistent with any single HR- radiocarbon ages. The age model is, in part, anchored by a lattice image. tephra layer at 4.7–4.5 m depth, identified as the Cieneguillas rhyolitic tephra, which has been dated elsewhere at 31 ka cal. 3.4. Radiocarbon Dating However, no evidence is provided that it is actually the Cieneguillas tephra. The tephra could be reworked, or it Precise dating of the stratigraphic record of the purported could be another one of a series of older rhyolitic tephra that YD impact is crucial in making the case for such a single, are reported in the region [Pradal and Robin, 1994]. continent-wide catastrophic event as argued by Firestone et Finally, one of us (MB) acquired carbon microspherules al. [2007]. The geochronologic record is far from precise, (Figure 4) collected from the Gainey site in Michigan from however. Of the nine sites used to make the initial argument one of the original YD impact proponents (A. West). Gainey for an impact, only three, Murray Springs, Arizona; Daisy is one of the nine key YD sites, and one of the undated ones, Cave, California; and Lake Hind, Manitoba, have robust presented by Firestone et al. [2007]. To verify the age of the numerical age control constrained by radiocarbon dating. samples, we submitted one set of spherules for accelerator Establishing a putative impact zone that spans several sites mass spectrometry radiocarbon dating at the University of requires direct ages that enable chronostratigraphic correla- Arizona. Only one microspherule has been dated thus far and tion and ties with the YD interval. is 207 ± 87 years BP (AMS lab number AA92197). This A key aspect of the Israde-Alcántara et al. [2012] study is result suggests that there are geochronology problems. One dating of the boundary layer with purported impact indica- key problem is that particles identified as diamond-containing tors. The authors unambiguously state that they recovered carbon microspheres and presumed to be related to the pur- such indicators from a 10 cm thick zone dating to 12.9 ka, ported YD impact may actually be younger than the YD, but in fact, they provide no direct numerical age control at or unrelated to the YD or to an impact, and might be modern near that date. They cite the original study of the core, in contaminants. which 16 14C dates were reported. The zone in question was bracketed by calibrated dates of 18.8 ka (3.35m) and 9.9 ka 4. CONCLUSIONS (1.95m) and then dated on the basis of a linear extrapolation. Israde-Alcántara et al. [2012] present an additional six An impact event as proposed by Firestone et al. [2007] is radiocarbon dates for the 3.35–1.95 m interval. These six not consistent with conventional understanding of the phys- dates were rejected because they are significantly older: 37.8 ics of impacts and airbursts. We conclude that the YD to 17.2 ka, bottom to top. They are consistent stratigraphi- impact hypothesis is not supportable, either physically or cally both internally and in comparison with overlying dates. statistically. Much of the putative evidence for a YD impact 24 YOUNGER DRYAS IMPACT EVENT is irreproducible. It is highly improbable that a significant Academy of Sciences, vol. 822, Near-Earth Objects: The United impact event happened during YD, as conceived by Fire- Nations Conference, edited by J. L. Remo, pp. 236–282, The stone et al. [2007]. Although the works published by the N. Y. Acad. of Sci., New York. proponents of an impact event vary in description about the Boslough, M. B. E., and D. A. Crawford (2008), Low-altitude air- – impactor, consideration of basic laws of physics indicate bursts and the impact threat, Int. J. Impact Eng., 35, 1441 1448. that such a fragmentation or high-altitude airburst event Bowman, D. M. J. S., et al. (2009), Fire in the Earth system, – would not conserve momentum or energy, would lie outside Science, 324, 481 484. any realistic range of probability, and therefore did not Braun, T., E. Osawa, C. Detre, and I. Toth (2001), On some analytical aspects of the determination of fullerenes in samples occur during the YD as described by Firestone et al. from the Permian/Triassic boundary layers, Chem. Phys. Lett., [2007]. This conclusion is supported by the present work, 348, 361–362. as well as a broad review of all the other lines of evidence Broecker, W. S., G. H. Denton, R. L. Edwards, H. Cheng, R. B. critiqued by Pinter and Ishman [2008a, 2008b], Surovell et Alley, and A. E. Putnam (2010), Putting the Younger Dryas cold al. [2009], and Pinter et al. [2011a, 2011b]. event into context, Quat. Sci. Rev., 29, 1078–1081, doi:10.1016/j. quascirev.2010.02.019. Acknowledgments. Sandia is a multiprogram laboratory operated Brown, P., R. E. Spalding, D. O. ReVelle, E. Tagliaferri, and S. P. by Sandia Corporation, a Lockheed Martin Company, for the United Worden (2002), The flux of small near-Earth objects colliding States Department of Energy under Contract DE-AC04-94AL85000 with the Earth, Nature, 420, 294–296. (M.B.). This work was funded in part by the Research Foundation Bundy, F. P., and J. S. Kasper (1967), Hexagonal diamond—A new Flanders (P.C.), and the Center for Materials Innovation at Washing- form of carbon, J. Chem. Phys., 46, 3437–3446. ton University (T.L.D.). Support for the AMS radiocarbon dating of Buseck, P. R. (2002), Geological fullerenes; review and analysis, the Gainey sample was provided by the Argonaut Archaeological Earth Planet. Sci. Lett., 203, 781–792. Research Fund (V.H., University of Arizona Foundation). The Daniau, A. L., S. P. Harrison, and P. J. Bartlein (2010), Fire regimes authors wish to thank Allen West for providing carbon spherule during the Last Glacial, Quat. Sci. Rev., 29, 2918–2930. samples collected from Gainey, Michigan. Daulton, T. L., N. Pinter, and A. C. Scott (2010), No evidence of nanodiamonds in Younger-Dryas sediments to support an impact REFERENCES event, Proc. Natl. Acad. Sci. U. S. A., 107, 16,043–16,047. Dawson, J. W. (1891), Acadian Geology: The Geological Structure, Alvarez, L. W., W. Alvarez, F. Asaro, and H. V. Michel (1980), Organic Remains, and Mineral Resources of Nova Scotia, New Extraterrestrial cause for the Cretaceous-Tertiary extinction, Sci- Brunswick, and Prince Edward Island, 4th ed., MacMillan, ence, 208, 1095–1108. London, U. K. Barnosky, A. D., P. L. Koch, R. S. Feranec, S. I. Wing, and A. B. Dyke, A. S., A. Moore, and L. Robertson (2003), Deglaciation of Shabel (2004), Assessing the causes of late Pleistocene extinc- North America, Open file 1574, Geol. Surv. of Canada, Ottawa. tions on the continents, Science, 306,70–75. Erlich, E. I., and W. D. Hausel (2002), Diamond Deposits: Origin, Becker, L., R. J. Poreda, and T. E. Bunch (1999), Fullerenes: An Exploration, and History of Discovery, Soc. for Min., Metal., and extraterrestrial carbon carrier phase for noble gases, Proc. Natl. Explor., Littleton, Colo. Acad. Sci. U. S. A., 97, 2979–2983, doi:10.1073/pnas.97.7.2979. Faith, J. T., and T. A. Surovell (2009), Synchronous extinction of Belcher, C. M. (2009), Reigniting the Cretaceous-Palaeogene fire- North America’s Pleistocene mammals, Proc. Natl. Acad. Sci. storm debate, Geology, 37, 1147–1148. U. S. A., 106, 20,641–20,645, doi:10.1073/pnas.0908153106. Belcher, C. M., M. E. Collinson, A. R. Sweet, A. R. Hildebrand, and Farley, K. A., and S. Mukhopadhyay (2001), An extraterrestrial A. C. Scott (2003), Fireball passes and nothing burns–The role of impact at the Permian–Triassic boundary?, Science, 293, 2343, thermal radiation in the Cretaceous-Tertiary event: Evidence doi:10.1126/science.293.5539.2343a. from the charcoal record of North America, Geology, 31, 1061– Farley, K. A., P. Ward, G. Garrison, and S. Mukhopadhyay (2005), 1064, doi:10.1130/G19989.1. Absence of extraterrestrial 3He in Permian–Triassic age sedimen- Belcher, C. M., M. E. Collinson, and A. C. Scott (2005), Constraints tary rocks, Earth Planet. Sci. Lett., 240, 265–275. on the thermal energy released from the Chicxulub impactor: Firestone, R. (2009), The case for the Younger Dryas extraterrestrial New evidence from multi-method charcoal analysis, J. Geol. impact event: Mammoth, megafauna, and Clovis extinction, Soc., 162, 591–602, doi:10.1144/0016-764904-104. 12,900 years ago, J. Cosmology, 2, 256–285. Belcher, C. M., P. Finch, M. E. Collinson, A. C. Scott, and N. V. Firestone, R. B., and W. Topping (2001), Terrestrial evidence of a Grassineau (2009), Geochemical evidence for combustion of nuclear catastrophe in Paleoindian times, Mammoth Trumpet, hydrocarbons during the K-T impact event, Proc. Natl. Acad. 16(2), 9–16. Sci. U. S. A., 106,4112–4117, doi:10.1073/pnas.0813117106. Firestone, R. B., A. West, and S. Warwick-Smith (2006), The Cycle Boslough, M. B. E., and D. A. Crawford (1997), Shoemaker-Levy 9 of Cosmic Catastrophes: Flood, Fire, and Famine in the History and plume forming collisions on Earth, in Annals of the New York of Civilization, Bear, Rochester, Vermont. BOSLOUGH ET AL. 25

Firestone, R. B., et al. (2007), Evidence for an extraterrestrial Higgins, M. D., P. Lajeunesse, G. St-Onge, J. Locat, M. Duchensne, impact 12,900 years ago that contributed to the megafaunal and R. Sanfaçon (2011), Bathymetric and petrological evidence extinctions and the Younger Dryas cooling, Proc. Natl. Acad. for a young (Pleistocene?) 4-km diameter impact crater in the Sci. U. S. A., 104, 16,016–16,021. Gulf of Saint Lawrence, Canada, paper presented at the 42nd Firestone, R. B., A. West, Z. Revay, J. T. Hagstrum, T. Belgya, S. S. Lunar and Planetary Science Conference, The Woodlands, Tex. Que Hee, and A. R. Smith (2010), Analysis of the Younger Dryas Holliday, V. T., and D. J. Meltzer (2010), The 12.9 ka ET impact impact layer, J. Sib. Fed. Univ. Eng. Technol., 3,30–62. hypothesis and North American Paleoindians, Curr. Anthropol., French, B. M., and C. Koeberl (2010), The convincing identifica- 51, 575–607. tion of terrestrial meteorite impact structures: What works, what Hut, P., W. Alvarez, W. P. Elder, T. Hansen, E. G. Kauffman, G. doesn’t, and why, Earth Sci. Rev., 98, 123–170. Keller, E. M. Shoemaker, and P. R. Weissman (1987), Comet Frondel, C., and U. B. Marvin (1967), Lonsdaleite, a hexagonal showers as a cause of mass extinctions, Nature, 329,118–126, polymorph of diamond, Nature, 214, 587–589. doi:10.1038/329118a0. Gill, J. L., J. W. Williams, S. T. Jackson, K. B. Lininger, and G. S. Israde-Alcántara, I., et al. (2012), Evidence from central Mexico Robinson (2009), Pleistocene megafaunal collapse, novel plant supporting the Younger Dryas extraterrestrial impact hypothesis, communities, and enhanced fire regimes in North America, Sci- Proc. Natl. Acad. Sci. U. S. A., 109, E738–E747, doi:10.1073/ ence, 326, 1100–1103. pnas.1110614109. Glasstone, S., and P. J. Dolan (1977), The Effects of Nuclear Kaminsky, F. V. (1994), Carbonado and yakutite: Properties and Weapons, U.S. Gov. Print. Off., Washington, D. C. possible genesis, in Proceedings of the Fifth International Kim- Grayson, D. K. (2007), Deciphering North American Pleistocene berlite Conference, vol. 2, Diamonds: Characterization, Genesis, extinctions, J. Anthropol. Res., 63, 185–213. and Exploration, edited by H. O. A. Meyer and O. H. Leonardos, – Grayson, D. K., and D. J. Meltzer (2003), A requiem for North pp. 136 143, Companhia de Pesquisa de Reeursos Minerais, Rio American overkill, J. Archaeol. Sci., 30, 585–593. de Janeiro, Brazil. Grieve, R. A. F. (1997), Extraterrestrial impact events: The record in Kennett, D. J., J. P. Kennett, G. J. West, J. M. Erlandson, J. R. the rocks and the stratigraphic column, Palaeogeogr. Palaeocli- Johnson, I. L. Hendy, A. West, B. J. Culleton, T. L. Jones, and fi matol. Palaeoecol., 132,5–23. T. W. Stafford Jr. (2008), Wild re and abrupt ecosystem disruption ’ – Grieve, R. A. F., and A. M. Therriault (2004), Observations at on California s Northern Channel Islands at the Ållerød Younger – – terrestrial impact structures: Their utility in constraining crater Dryas boundary (13.0 12.9 ka), Quat. Sci. Rev., 27, 2530 2545. formation, Meteorit. Planet. Sci., 39,199–216, doi:10.1111/j. Kennett, D. J., J. P. Kennett, A. West, C. Mercer, S. S. Que Hee, L. 1945-5100.2004.tb00336.x. Bement, T. E. Bunch, M. Sellers, and W. S. Wolbach (2009a), Hagstrum, J. T., R. B. Firestone, A. West, Z. Stefanka, and Z. Revay Nanodiamonds in the Younger Dryas boundary sediment layer, (2010), Micrometeorite impacts in Beringian mammoth tusks and Science, 323, 94. a bison skull, J. Sib. Fed. Univ. Eng. Technol., 3, 123–132. Kennett, D. J., et al. (2009b), Shock-synthesized hexagonal dia- Hanneman, R. E., H. M. Strong, and F. P. Bundy (1967), Hexagonal monds in Younger Dryas boundary sediments, Proc. Natl. Acad. – diamonds in meteorites: Implications, Science, 155, 995–997. Sci. U. S. A., 106, 12,623 12,638. Harris, A. W. (2002), A new estimate of the population of small Kurbatov, A. V., et al. (2010), Discovery of a nanodiamond-rich – NEAs, Bull. Am. Astron. Soc., 34, 835. layer in the Greenland ice sheet, J. Glaciol., 56, 749 759. fi Haynes, C. V., Jr. (1991), Geoarchaeological and paleohydrological Marlon, J. R., et al. (2009), Wild re responses to abrupt climate change – evidence for a Clovis age drought in North America and its in North America, Proc. Natl. Acad. Sci. U. S. A., 106, 2519 2524. bearing on extinction, Quat. Res., 35, 438–450. Melosh, H. J. (1989), Impact Cratering: A Geologic Process, Ox- Haynes, C. V., Jr. (2007), Murray Springs: A Clovis Site With ford Univ. Press, New York. Multiple Activity Areas in the San Pedro Valley, Arizona, edited Melosh, H. J., and G. S. Collins (2005), Meteor crater formed by a by C. V. Haynes Jr. and B. B. Huckell, pp. 229–239, Univ. of low-velocity impact, Nature, 434, 157. Ariz. Press, Tucson. Meltzer, D. J., and V. T. Holliday (2010), Would North American Haynes, C. V., Jr. (2008), Younger Dryas “black mats” and the Paleoindians have noticed Younger Dryas age climate changes?, – Rancholabrean termination in North America, Proc. Natl. Acad. J. World Prehist., 23,1 41. Sci. U. S. A., 105, 6520–6525. NASA Near-Earth Object Science Definition Team (2003), Study to Haynes, C. V., Jr., J. Boerner, K. Domanik, D. Lauretta, J. Ballen- determine the feasibility of extending the search for near-earth ger, and J. Goreva (2010), The Murray Springs Clovis site, objects to smaller limiting diameters, Off. of Space Sci., Sol. Pleistocene extinction, and the question of extraterrestrial impact, Syst. Explor. Div., Washington, D. C. Proc. Natl. Acad. Sci. U. S. A., 107, 4010–4015. Paquay, F. S., S. Goderis, G. Ravizza, F. Vanhaeck, M. Boyd, T. A. Heymann, D., L. P. F. Chibante, R. R. Brooks, W. S. Wolbach, and Surovell, V. T. Holliday, C. V. Haynes Jr., and P. Claeys (2009), R. E. Smalley (1994), Fullerenes in the Cretaceous-Tertiary Absence of geochemical evidence for an impact event at the – boundary layer, Science, 265, 645–647, doi:10.1126/science. Bølling Allerød/Younger Dryas transition, Proc. Natl. Acad. Sci. – 265.5172.645. U. S. A., 106, 21,505 21,510. 26 YOUNGER DRYAS IMPACT EVENT

Paterson, W. S. B. (1972), Laurentide ice sheet: Estimated volumes terrestrial impact hypothesis, Proc. Natl. Acad. Sci. U. S. A., 106, during late Wisconsin, Rev. Geophys., 10(4), 885–917. 18,155–18,158. Peros, M. C., K. Gajewski, and A. E. Viau (2008), Continental-scale Taylor, R., and A. K. Abdul-Sada (2000), There are no fullerenes in tree population response to rapid climate change, competition, the K–T boundary layer, Fullerene Sci. Technol., 8,47–54. and disturbance, Global Ecol. Biogeogr., 17, 658–669. Tian, H., D. Schryvers, and P. Claeys (2011), Nanodiamonds do not Pigati, J. S., C. Latorre, J. A. Rech, J. L. Betancourt, K. E. Martinez, provide unique evidence for a Younger Dryas impact, Proc. Natl. and J. R. Budahn (2012), Accumulation of impact markers in Acad. Sci. U. S. A., 108,40–44, doi:10.1073/pnas.1007695108. desert wetlands and implications for the Younger Dryas impact Toon, O. B., K. Zahnle, D. Morrison, R. P. Turco, and C. Covey hypothesis, Proc. Natl. Acad. Sci. U. S. A., 109, 7208–7212, (1997), Environmental perturbations caused by the impacts of doi:10.1073/pnas.1200296109. asteroids and comets, Rev. Geophys., 35(1), 41–78. Pinter, N., and S. E. Ishman (2008a), Impacts, mega-tsunami, and van der Hammen, T., and B. van Geel (2008), Charcoal in soils of other extraordinary claims, GSA Today, 18,37–38. the Allerød-Younger Dryas transition were the result of natural Pinter, N., and S. E. Ishman (2008b), Reply to comments on “Im- ﬁres and not necessarily the effect of an extra-terrestrial impact, pacts, mega-tsunami, and other extraordinary claims”, GSA To- Neth. J. Geosci., 87, 359–361. day, 18(6), e14. Werner, S. C., A. W. Harris, G. Neukum, and B. A. Ivanov (2002), Pinter, N., A. C. Scott, T. L. Daulton, A. Podoll, C. Koeberl, R. S. The near-Earth asteroid size-frequency distribution: A snapshot Anderson, and S. E. Ishman (2011a), The Younger Dryas impact of the lunar impactor size-frequency distribution, Icarus, 156, hypothesis: A requiem, Earth Sci. Rev., 106, 247–264, doi:10. 287–290. 1016/j.earscirev.2011.02.005. Yang, Z. Q., J. Verbeeck, D. Schryvers, N. Tarcea, J. Popp, and W. Pinter, N., S. J. Fiedel, and J. E. Keeley (2011b), Fire and vegetation Rosler (2008), TEM and Raman characterisation of diamond shifts in the Americas at the vanguard of Paleoindian migration, micro- and nanostructures in carbon spherules from upper soils, Quat. Sci. Rev., 30, 269–272, doi:10.1016/j.quascirev.2010.12. Diamond Relat. Mater., 27, 937–943. 010. Pradal, E., and C. Robin (1994), Long-lived magmatic phases at Los Azufres volcanic center, Mexico, J. Volcanol Geotherm. Res., 63, 201–215. P. Bartlein, Department of Geography, University of Oregon, Quade, J., R. M. Forester, and W. L. Pratt (1998), Black mats, Eugene, OR 97403-1251, USA. spring-fed streams, and late-glacial-age recharge in the southern M. Boslough, Sandia National Laboratories, Albuquerque, NM Great Basin, Quat. Res., 49, 129–148. 87185, USA. ([email protected]) Scott, A. C., N. Pinter, M. E. Collinson, M. Hardiman, R. S. P. Claeys, Earth System Science, Vrije Universiteit Brussel, Anderson, A. P. R. Brain, S. Y. Smith, F. Marone, and M. Pleinlaan 2, 1050 Brussels, Belgium. Stampanoni (2010), Fungus, not comet or catastrophe, accounts T. L. Daulton, Department of Physics and Center for Materials for carbonaceous spherules in the Younger Dryas “impact lay- Innovation, Washington University, St. Louis, MO 63130, USA. er”, Geophys. Res. Lett., 37, L14302, doi:10.1029/2010GL J. Gill and J. Marlon, Department of Geography, University of 043345. Wisconsin, 550 N. Park St., Madison, WI 53706, USA. D. Grayson, Department of Anthropology, University of Shumilova, T. G., E. Mayer, and S. I. Isaenko (2011), Natural Washington, Box 353100, Seattle, WA 98195, USA. monocrystalline lonsdaleite, Dokl. Earth Sci., 441, 1552–1554. V. Holliday, School of Anthropology and Department of Shuvalov, V. V. (2003), Numerical modeling of the Eltanin impact, Geosciences, University of Arizona, Tucson, AZ 85721, USA. paper presented at Lunar and Planetary Science XXXI, Houston, A. J. T. Jull, AMS Radiocarbon Facility, University of Arizona, Tex. Tucson, AZ 85721, USA. Shuvalov, V. V., and I. A. Trubetskaya (2007), Numerical modeling D. Meltzer, Department of Anthropology, Southern Methodist of the formation of the Eltanin submarine impact structure, Sol. University, Dallas, TX 75275-0336, USA. – Syst. Res., 41(1), 56 64, doi:10.1134/S0038094607010066. K. Nicoll, Department of Geography, University of Utah, Salt Straus, L. G., and T. Goebel (Eds.) (2011), Humans and Younger Lake City, UT 84112, USA. Dryas: Dead end, short detour, or open road to the Holocene?, F. Paquay, Department of Geology and Geophysics, University – Quat. Int., 242, 259 584. of Hawai‘iatMānoa, Honolulu, HI 96822, USA. Stuart, J. S. (2001), A near-Earth asteroid population estimate from N. Pinter, Department of Geology, Southern Illinois University, the LINEAR survey, Science, 294, 1691–1693. Carbondale, IL 62901-4324, USA. Su, Z., W. Zhou, and Z. Yang (2011), New insight into the soot A. C. Scott, Department of Earth Sciences, Royal Holloway nanoparticles in a candle ﬂame, Chem. Commun., 47,4700– University of London, Egham TW20 0EX, UK. 4702, doi:10.1039/c0cc05785a. T. Surovell, Department of Anthropology, University of Surovell, T. A., V. T. Holliday, J. A. M. Gingerich, C. Ketron, C. V. Wyoming, Laramie, WY 82070, USA. Haynes Jr., I. Hilman, D. P. Wagner, E. Johnson, and P. Claeys C. Whitlock, Department of Earth Sciences, Montana State (2009), An independent evaluation of the Younger Dryas extra- University, Bozeman, MT 59715, USA. Environment and Agency in the Ancient Maya Collapse

James J. Aimers

Department of Anthropology, State University of New York at Geneseo, Geneseo, New York, USA

For nearly a century, hypotheses about the causes of the Terminal Classic lowland Maya collapse of the ninth century A.D. have reﬂected contemporary concerns. Recently, a number of environmental scientists and archaeologists have evoked climate change and drought to explain the Maya collapse and a number of other periods of decline or abandonment in Maya history. In this paper, I suggest that many of the arguments about drought and collapse simplify or misrepresent the archaeological data. Closer collaboration between physical scientists and archaeologists is needed to better align the paleoclimate data with current archaeological evidence.

1. INTRODUCTION scientists and archaeologists is needed to better align the climate data with the archaeological evidence. For nearly a century, hypotheses about the causes of the The idea that drought caused a number of crises in Maya Terminal Classic lowland Maya collapse have reflected con- society has been espoused most influentially by the archae- temporary concerns about issues as diverse as deforestation ologist Richardson Gill in several publications including his and political revolution [Huntington, 1917; Wilk, 1985]. book “The Great Maya Droughts” [2000]. Gill’s ideas, like Since the mid-1990s, so many people have evoked climate those of Jared Diamond [e.g., Diamond, 2005], have become change and drought to explain a number of periods of decline influential beyond academia in part because they present a or abandonment in Maya history that the list of relevant tidy and dramatic narrative. In a 2001 British Broadcasting references is now too long to be cited here (for a fairly Corporation documentary that sums up his argument, Gill is comprehensive list of sources see Iannone [2012, Table presented as a maverick, bucking the skepticism of closed- 2.1]). Although the importance of drought to these discus- minded archaeologists who, like me, argue that his conclusions of collapse varies from extreme [Gill, 2000] to mea- sions about the Maya collapse are simplistic. Nevertheless, sured [Hodell et al., 1995], many of these arguments long-lasting droughts are now routinely cited as the estab- simplify or misrepresent archaeological data. Treatments of lished driving factor in a supposedly pandemic Maya col- drought in relation to the Maya collapse also frequently lapse, reaching fields outside of archaeology and in neglect issues of human adaptability, variability, and agency overviews presented in important scientific contexts [deMe- (used here to mean the ability of people to act on the world nocal, 2001]. The overall impact of this popular and scien- around them). There is little doubt that droughts of varying tific interest is that the influence of drought on the ancient scale and intensity were important factors in ancient Maya Maya has now been investigated by a large number of life, but continued collaboration between environmental archaeologists and scientists in other fields. The results of these investigations are surely important and provocative, but they are by no means definitive or clear.

Climates, Landscapes, and Civilizations 2. THREE KEY POINTS Geophysical Monograph Series 198 © 2012. American Geophysical Union. All Rights Reserved. In the introduction to an edited volume about what is now 10.1029/2012GM001204 generally called the Maya drought hypothesis [Aimers and 27 28 ENVIRONMENT AND AGENCY IN ANCIENT MAYA COLLAPSE

Iannone, 2012], Gyles Iannone and I discuss what archaeologists the Terminal Classic period. To summarize: some sites and now know about the long sweep of Maya and Mesoamerican even regions did indeed exhibit what can reasonably be prehistory. For the diverse readership of the present volume, called collapse in the form of rapid abandonment. Other sites three points are especially relevant. and regions, however, show nothing of the sort. I work at one of these sites: Lamanai, in northern Belize (Figure 1). While 2.1. The Dynamic Model sites as close as Colha, 60 km away, were in turmoil as evinced by remains such as skull pits and site abandonment First, the story of ancient Maya civilization is not a simple [Barrett and Scherer, 2005], Lamanai undertook one of the evolutionary progression from growth to death. Maya and most ambitious construction programs in its history and the Mesoamerican states continually oscillated between periods site, like some others in the Maya world, was a thriving city of integration and disintegration, florescence and collapse, in the subsequent Postclassic period (circa A.D. 900–1540) and this is characteristic of states in general. Joyce Marcus [Pendergast, 1985]. Lamanai, like the sites of Tayasal and calls this the “dynamic model” of Mesoamerican cultural Tipu, was an important Maya center right up to the seven- development [Marcus, 1989, 1992, 1994, 1998]. teenth century. Tayasal is just one of several sites in the Petén Lakes region that was occupied throughout the Postclassic 2.2. Maya Diversity, Not Unity period. The Terminal Classic collapse is typically dated from fi Second, the Maya were never a single uni ed state and about A.D. 800 to 1000, but the great temporal variability nothing like an empire, as they are sometimes called in the that has been revealed archaeologically complicates simplis- nonarchaeological literature on Maya drought [Webster et tic narratives of Maya florescence and collapse. This impor- al., 2007]. Maya civilization consisted of a patchwork of tant fact is often ignored by nonarchaeologists, probably fl city-states of varying sizes and in uence in varied physical because they are using outdated overviews of Maya prehis- environments (some much drier than others, e.g., the north- tory, which place the collapse entirely in the ninth century, a ern lowlands), jostling for power in complex ways that are at notion based mostly on early research in the central Petén least partially discernible as hieroglyphic decipherment con- region at sites like Tikal. We now know that sites in the tinues. By the Classic period (circa A.D. 250 to 800) (see Petexbatun region of Guatemala (e.g., Dos Pilas and Agua- Table 1), we see kings and courts with intrigue and treachery teca) were abandoned as early as the mid-eighth century worthy of Shakespeare, but also tiny rural villages [Sharer [Demarest, 2004], while others such as Chichen Itza (located and Traxler, 2006]. Even today, 7 million or so living Maya in the relatively dry northern lowlands) were not abandoned fi speak 31 languages and exhibit signi cant cultural diversity until the mid-eleventh century [Cobos, 2004], and in the [Campbell and Kaufman, 1985]. The idea of a homogenous Mopan Valley of Guatemala, the Terminal Classic may ex- fi fi and uni ed Maya is a colonial and postcolonial ction tend into the thirteenth century [Laporte, 2004]. Another [Grofe, 2005; Sachse, 2006]. major problem relevant to this issue is the imprecision of many archaeological dating techniques including radiocar- 2.3. Terminal Classic Cultural and Temporal Variability bon dating and ceramic chronologies, and the difficulty of aligning these with paleoclimate data sets [Yaeger and Third, the idea of the Maya collapse, like the idea of the Hodell, 2008]. Maya, is a simplification. In previous publications [Aimers, 2004, 2007], I have discussed the bewildering variability of 3. WHAT IS COLLAPSE?

a Is it reasonable to call a process of abandonment a collapse Table 1. Basic Maya Chronology if it did or did not happen, over a period of 400 years or Period Time Frame more? Aimers and Iannone [2012] reviewed various defini- Postclassic A.D. 900–1500 tions of collapse, and there is variability in those, too. We Terminal Classic A.D. 800–900 believe a collapse should involve a significant loss of socio- Late Classic A.D. 600–800 political complexity over a relatively rapid period of time, Early Classic A.D. 250–600 but what is “significant” and what is “rapid” are dependent Late Preclassic 250 B.C. to A.D. 250 upon scale. A slow collapse of a city might take 100 years, Middle Preclassic 1000 – 250 B.C. but for a civilization, this would be considered rapid by many Early Preclassic 2000 – 1000 B.C. people. We should probably abandon the term collapse for aThere are many versions for different sites and regions. the events of the Terminal Classic, but the term is now so AIMERS 29

Figure 1. Maya area major sites and sites mentioned in the text. The site of Tikal, in the central Maya area, is located at 17.2219°N, 89.6228°W. From The Electronic Atlas of Ancient Maya Sites. © copyright 2008 Clifford T. Brown and Walter R. T. Witschey. 30 ENVIRONMENT AND AGENCY IN ANCIENT MAYA COLLAPSE established in the literature that many Mayanists now use it in long-term statistical relationships at very large scales and simply as a sort of shorthand for the Terminal Classic period. tend to favor “elegant” explanations (i.e., ones with few Even if the term is ill-advised, it is entrenched, and anyone variables). In the last 50 years, many archaeologists have writing about the Maya collapse should be clear about what attempted to model archaeology on the natural sciences, and they mean by collapse and the scale under consideration. there has been a corresponding tendency toward external, What actually collapsed: a polity, a federation or alliance, an structurally oriented explanations such as climate change. empire, a society, a tradition, or a civilization? This is often There has always been, however, a strong countervailing not specified. appeal in archaeology to interpretations of the past which stress the agency of human actors despite the constraints or 4. WHAT IS DROUGHT? challenges posed by structures of all sorts (social, environmental, political, etc.). In archaeology, these tensions have The number and variety of definitions of collapse pale in played out in what are called Processual and Postprocessual comparison to the many definitions of drought, yet scholars method and theory. Most archaeologists today accept the in various disciplines often neglect to state which one they constraining effects of the physical environment, but also are using when discussing Maya drought. In an often-cited recognize that humans respond to these constraints in active article, Wilhite and Glantz [1985, p. 118] reviewed 150 and varied ways. This is not just a result of a theoretical definitions of drought and concluded that “there cannot trend; it has been documented archaeologically. (and should not) be a universal definition of drought.” In the case of drought, climate scientists generally empha- Wilhite and Glantz created a four-part typology of drought, size factors beyond human control (usually rainfall), but for showing that it can be defined (1) meteorologically by most archaeologists, it is strange to see that they often fail to changes in precipitation, (2) hydrologically based on consider that one of the conditions associated with drought changes in lakes and streams, (3) agriculturally based on could be people and what they do. For me, this leads to two insufficient water for plant growth, (4) or socioeconomical- more points. First, if there is evidence of drought, was it ly in terms of the impact of decreased water supplies on the natural or anthropogenic? In some cases, increasing aridity supply of goods. Wilhite and Glantz [1985, p. 114] specif- has been caused by increased population, agricultural over- ically point out that meteorological drought, which is what production, and deforestation. If this leads to abandonment, environmental scientists can best detect, does not always is it not clearer to say that abandonment was linked to coincide with agricultural drought, which is what would overpopulation and deforestation rather than drought, since have been relevant to the ancient Maya and is of interest to drought at least implies, to most people, natural causes? archaeologists when discussing the events of the Terminal Second, just as we must be more explicit about how human Classic. One of their conclusions is that “. . . weather, or activities might cause drought, we must consider human drought, affects far more than just crop yields and . . . responses to drought. Much of the existing literature on social factors can be equally significant in determining drought in the Maya lowlands ignores human agency almost society’s vulnerability to drought and, thus, the type and entirely. Gill et al. [2007, p. 299] exemplify this in typically magnitude of drought impacts.” [Wilhite and Glantz, 1985, dramatic fashion: “There was nothing they could do or could p. 116] This highlights the importance of considering both have done. . . and they died.” This may have been true in environment and agency in discussions of drought and the some times and places, but it simply does not correspond to Maya collapse. the evidence from about a century of research in the Maya lowlands. 5. ENVIRONMENT, ADAPTATION, AND AGENCY 6. THE TERMINAL CLASSIC COLLAPSE Dualisms have been strongly marked in Western scholar- ship since at least Descartes and are exemplified by the Archaeologists use the term “horizon” to describe a phe- distinction between the so-called “hard” physical sciences nomenon with a broad geographical distribution over a rel- compared to the social sciences as well as differential em- atively short period of time. The Terminal Classic period phasis on agency (the human ability to act on the world) and cannot be considered a collapse horizon because it involved structure (patterns or processes that restrict choice) in inter- site abandonment at different times in different places, and it pretations of the human past. Those in the sciences such as frequently did not involve abandonment at all. Even within geology tend to emphasize structure and external forces the small area of what is now Belize, abandonment happened because they are trained to seek patterns that may approach within a few generations (e.g., at the site of Xunantunich) or may even be laws. Paleoscientists also tend to be interested [LeCount et al., 2002], sometimes over centuries (e.g., at AIMERS 31

Santa Rita) [Chase and Chase, 1988], and sometimes it did 7. PROXIMATE CAUSES AND RESPONSES not occur at all (e.g., at Tipu) [Jones et al., 1986]. This temporal variability is detectable even in smaller, compara- Although sites were abandoned throughout Maya history tively homogenous landscapes like the Usamacinta River and droughts certainly occurred, drought is not an explana- basin (e.g., the sites of Piedras Negras and Yaxchilan) tion of the abandonment of a site; it may be one of the [Scherer and Golden, 2012]. proximate causes, a condition that may foster abandonment, Although debate about the details continues, a consensus and drought is well-documented in only a handful of places is emerging among Mayanists that drought alone is not (e.g., at Calakmul [see Gunn et al., 2002]). Archaeologists enough to explain the abandonment of Maya sites. No one have always placed the people they study in the context of doubts that devastating droughts occurred in Maya prehistory, their physical environment, but the assumption that all hu- as they have historically, but if we fail to consider and seek man organizations react in the same way to challenges pre- out the cultural response to drought, we come close to sented by their environment ignores what archaeologists, environmental determinism. Lamanai prospered during the historians, and anthropologists know about cultural variation Terminal Classic as Altun Ha, only 40 km away, was aban- and adaptation. Agricultural innovation, population move- doned. Proponents of drought point to Lamanai’s location on ment, as well as social, economic, and political change are the New River but many sites on rivers were abandoned, just some of the strategies Maya groups may have used to including sites on other large rivers like the Belize River and adapt to drought. Although such strategies may not always the Usumacinta. In western Belize, riverine centers occupied have been successful, people would have reacted as knowl- into the Postclassic include Baking Pot on the Belize River edgeable and strategic agents, not like plants or animals. and Tipu on the Macal River, but the other excavated sites Economic and political factors are high on the list of were abandoned (e.g., Blackman Eddy, Floral Park). There elements we must consider when considering the differential does not seem to be a direct correlation between access to changes of the Terminal Classic, especially as evidence water and survival through the Terminal Classic; other fac- accumulates that climate change was characteristic of the tors must have been involved. era. Resource diversity would have favored some centers This emerging consensus on the dramatic changes of the over others. Examples include aquatic resources in addition Terminal Classic is not only important in itself but also to hunting and agriculture for food as at Lamanai or valued because Gill and others [deMenocal, 2001; Rosenmeier et resources for trade as in the case of cacao at Tipu [Muhs et al., 2002; Webster et al., 2007] have extended the drought al., 1985]. Less centralized political organization and more hypothesis to a number of periods in Maya prehistory in sharing of power may also be characteristic of the Terminal which evidence of decline is much more debatable. For Classic (e.g., at Chichen Itza), and this may have facilitated example, much has been made of a supposed sixth century more nimble responses to crises than possible in the more Maya hiatus, which some call a collapse. Archaeologists are hierarchical systems characteristic of the Classic period. now in general agreement that there was no Maya hiatus for the lowlands as a whole. There was a gap in monument 8. CONCLUSION dedication at some sites that we are now confident was linked to political instability, and the timing of these gaps As early as 1973 in The Classic Maya Collapse, Andrews varied considerably [Moholy-Nagy, 2003]. There is, in fact, IV [1973, p. 243] pointed out that there was no collapse of very little evidence that drought played a role in the varied the Maya as a whole, and since then, many people have hiatuses and no evidence that we know of from the centers documented the temporal, spatial, and cultural variation of themselves. Nevertheless, author after author uncritically the Terminal Classic [Aimers, 2007; Demarest et al., 2004; accepts the hiatus as a pervasive crisis in the Maya region, Marcus, 1998; Webster, 2002]. Still it has been nearly im- often simply by citing Gill’s book or his outdated sources, possible to dispel the myth of a pandemic Maya collapse rather than referencing a growing literature that questions among academics and the general public. The drought hy- the very existence of a Maya hiatus. There is no room here pothesis, especially in its extreme forms (e.g., Gill’s work), is to even address widely repeated ideas about supposed Pre- just the most recent of many attempts to find order in the classic and Postclassic period Maya collapses except to say chaos of the Terminal Classic archaeological record. As such, that the disconnection between researchers in the physical it says as much about our contemporary concerns about sciences and archaeology is notable in these cases. It may climate change and the appeal of an easily identifiable expla- not be fair to ask climate scientists to keep up with a large nation as it does about the lives of the ancient Maya. The and rapidly changing literature on the Maya, but closer Maya lowlands were diverse enough that Mayan speakers interdisciplinary consultation is clearly warranted. from different regions would have not been able to understand 32 ENVIRONMENT AND AGENCY IN ANCIENT MAYA COLLAPSE one another’s languages, and we have evidence of signifi- Chase, D. Z., and A. F. Chase (1988), A Postclassic Perspective: cantly different customs and traditions across the area, as is Excavations at the Maya Site of Santa Rita Corozal, Belize, Pre- the case today. The environmental diversity of the Maya Columbian Art Res. Inst., San Francisco, Calif. lowlands is also considerable and extremely relevant to all- Cobos, R. (2004), Chichén Itzá: Settlement and hegemony during encompassing ideas about the impact of drought. the Terminal Classic Period, in The Terminal Classic in the Maya Overall, I would like to see the notion of the Maya collapse Lowlands: Collapse, Transition, and Transformation, edited by A. Demarest, P. M. Rice and D. S. Rice, pp. 517–544, Univ. Press (whether in the Preclassic, Middle Classic, Terminal Classic, of Colo., Boulder. or Postclassic) shoveled into the backdirt pile. Aside from the Demarest, A. A. (2004), After the maelstrom: Collapse of the fact that it is inaccurate, it perpetuates the idea that the Maya Classic Maya kingdoms and the Terminal Classic in western “ ” mysteriously vanished or that they were incapable of adapt- Petén, in The Terminal Classic in the Maya Lowlands: Collapse, ing to environmental crises. This further disconnects the living Transition, and Transformation, edited by A. Demarest, P. M. Maya, who may in fact be more unified today than ever before, Rice and D. S. Rice, pp. 102–124, Univ. Press of Colo., Boulder. from a past that can be a source of pride and power [Bastos, Demarest, A. A., P. M. Rice, and D. S. Rice (2004), The Terminal 2006]. The task ahead of us is to compare localized data sets Classic in the Maya Lowlands: Collapse, Transition, and Trans- from across the Maya world, work on our chronological formation, Univ. Press of Colo., Boulder. correlations, and consider how the Maya in different environ- deMenocal, P. (2001), Cultural responses to climate change during ments may have responded to drought where it is evident. I the Late Holocene, Science, 292, 667–673. would like to encourage readers working in climate science, Diamond, J. (2005), Collapse: How Societies Choose to Fail or on the issue of ancient drought, to more fully engage with Survive, Allen Lane, London. current research by archaeologists working in the field, rather Gill, R. (2000), The Great Maya Droughts: Water, Life, and Death, than uncritically repeating ideas that are out of date, specula- Univ. of N. M. Press, Albuquerque. tive, or demonstrably wrong. All of our research will be Gill, R. B., P. A. Mayewski, J. Nyberg, G. H. Haug, and L. C. stronger with this sort of collaboration, and I am happy that Peterson (2007), Drought and the Maya Collapse, Ancient Me- soamerica, 18, 283–302. the American Geophysical Union conference from which this Grofe, M. J. (2005), Ethnic identities: The construction of Maya volume derives shows that this is already happening. identity, Curr. Anthropol., 46(1), 1–2. Gunn, J. D., J. E. Foss, W. J. Folan, R. D. Carrasco, and B. B. Faust REFERENCES (2002), Bajo sediments and the hydraulic system of Calakmul, Campeche, Mexico, Ancient Mesoamerica, 13, 297–315. Aimers, J. J. (2004), Cultural change on a temporal and spatial Hodell, D. A., J. H. Curtis, and M. Brenner (1995), Possible role of frontier: Ceramics of the Terminal Classic to early Postclassic climate in the collapse of Classic Maya civilization, Nature, 375, transition in the upper Belize River Valley, BAR Int. Ser. 1325, 391–394. 541 pp., Br. Archaeol. Rep., Oxford, U. K. Huntington, E. (1917), Maya civilization and climate change, in Aimers, J. J. (2007), What Maya collapse? Terminal Classic varia- Proceedings of the 19th Congress of Americanists, pp. 150–164. tion in the Maya lowlands, J. Archaeol. Res., 15, 329–377. Iannone, G. (Ed.) (2012), The Great Maya Droughts in Cultural Aimers, J. J., and G. Iannone (2012), The dynamics of ancient Maya Context, Univ. Press of Colo., Boulder. developmental history, in The Great Maya Droughts in Cultural Jones, G. D., R. B. Kautz, and E. Graham (1986), Tipu: A Maya Context, edited by G. Iannone, Univ. Press of Colo., Boulder, in town on the Spanish colonial frontier, Archaeology, 39,40–47. press. Laporte, J. P. (2004), Terminal Classic settlement and polity in the Andrews, E. W., IV (1973), The development of Maya civilization Mopán Valley, Petén, Guatemala, in The Terminal Classic in the after the abandonment of the southern cities, in The Classic Maya Maya Lowlands: Collapse, Transition, and Transformation, edi- Collapse, edited by T. P. Culbert, pp. 243–265, Sch. of Am. Res., ted by A. Demarest, P. M. Rice and D. S. Rice, pp. 195–230, Univ. of N. M. Press, Albuquerque. Univ. Press of Colo., Boulder. Barrett, J. W., and A. K. Scherer (2005), Stones, bones, and LeCount, L. J., J. Yaeger, R. M. Leventhal, and W. Ashmore (2002), crowded plazas: Evidence for Terminal Classic Maya warfare at Dating the rise and fall of Xunantunich, Belize: A Late and Colhá, Belize, Ancient Mesoamerica, 16(1), 101–118. Terminal Classic lowland Maya regional center, Ancient Meso- Bastos, S. (2006), Ser maya en el siglo XXI: El proceso de america, 13(1), 41–63. construcción y difusión de una identidad politica, in Maya Eth- Marcus, J. (1989), From centralized systems to city states: Possible nicity: The Construction of Identity From Preclassic to Modern models for the Epiclassic, in Mesoamerica After the Decline of Times, Acta Mesoam., vol. 19, edited by F. Sachse, pp. 219–233, Teotihuacán A.D. 700–900, edited by R. A. Diehl and J. C. Berlo, Anton Saurwein, Markt Schwaben, Germany. pp. 201–208, Dumbarton Oaks, Wash. Campbell, L., and T. Kaufman (1985), Mayan linguistics: Where Marcus, J. (1992), Dynamic cycles of Mesoamerican states, Natl. are we now?, Annu. Rev. Anthropol., 14, 187–198. Geogr. Res. Explor., 8, 329–411. AIMERS 33

Marcus, J. (1994), The collapse of Maya states: A dynamic process, in The Great Maya Droughts in Cultural Context, edited by G. paper presented at the 93rd annual meeting of the Am. Anthropol. Iannone, Univ. Press of Colo., Boulder, in press. Assoc., Atlanta, Ga. Sharer, R. J., and L. P. Traxler (2006), The Ancient Maya, 6th ed., Marcus, J. (1998), The peaks and valleys of ancient states: An Stanford Univ. Press, Stanford, Calif. extension of the dynamic model, in Archaic States, edited by Webster, D. L. (2002), The Fall of the Ancient Maya, Thames and G. M. Feinman and J. Marcus, pp. 59–94, Sch. of Am. Res., Hudson, London, U. K. Santa Fe, N. M. Webster, J. W., G. A. Brook, L. B. Railsback, H. Cheng, R. L. Moholy-Nagy, H. (2003), The Hiatus at Tikal, Guatemala, Ancient Edwards, C. Alexander, and P. P. Reeder (2007), Stalagmite Mesoamerica, 14,77–83. evidence from Belize indicating signiﬁcant droughts at the time Muhs, D. R., R. R. Kauts, and J. J. MacKinnon (1985), Soils and the of the Preclassic Abandonment, the Maya hiatus, and the location of Cacao Orchards at a Maya Site in Western Belize, Classic Maya Collapse, Paleogeogr. Paleoclimatol. Paleoecol., J. Archaeol. Sci., 12, 121–137. 250,1–17. Pendergast, D. M. (1985), Stability through change: Lamanai, Be- Wilhite, D. A., and M. H. Glantz (1985), Understanding the drought lize, from the ninth to the seventeenth century, in Late Lowland phenomenon: The role of deﬁnitions, Water Int., 10(3), 111–120. Maya Civilization: Classic to Postclassic, edited by J. A. Sabloff Wilk, R. (1985), The ancient Maya and the political present, and E. W. Andrews V, pp. 223–249, Sch. of Am. Res., Univ. of J. Anthropol. Res., 41(3), 307–326. N. M. Press, Albuquerque. Yaeger, J., and D. A. Hodell (2008), The collapse of Maya civili- Rosenmeier, M. F., D. A. Hodell, M. Brenner, J. H. Curtis, and T. P. zation: Assessing the interaction of culture, climate, and environ- Guilderson (2002), A 4000-year Lacustrine record of environment, in El Niño, Catastrophism, and Culture Change in Ancient mental change in the southern Maya lowlands, Petén, Guatemala, America, edited by D. H. Sandweiss and J. Quilter, pp. 188–242, Quat. Res., 57(2), 183–190. Dumbarton Oaks, Washington, D. C. Sachse, F. (Ed.) (2006), Maya Ethnicity: The Construction of Ethnic Identity From Preclassic to Modern Times, Acta Mesoamericana, vol. 19, Verlag Anton Saurwein, Markt Schwaben, Germany. J. J. Aimers, Department of Anthropology, State University of Scherer, A. K., and C. W. Golden (2012), Water in the West: New York at Geneseo, Geneseo, NY 14454, USA. (jim_aimers@ Chronology and collapse of the western Maya River kingdoms, yahoo.com)

Rainfall Variability and the Rise and Collapse of the Mississippian Chiefdoms: Evidence From a DeSoto Caverns Stalagmite

Paul Aharon and David Aldridge

Department of Geological Sciences, University of Alabama, Tuscaloosa, Alabama, USA

John Hellstrom

School of Earth Sciences, University of Melbourne, Melbourne, Victoria, Australia

A newly acquired, absolute U/Th dated, δ18O record archived in a stalagmite from DeSoto Caverns in Alabama renders highly resolved time series for the past four millennia. Two principal states of variability are discerned in the δ18O record: (1) stable state spanning the intervals before Common Era 2350 to A. D. 400 and A.D. 1700 to 2008 that exhibits signiﬁcant periodicities of 30 and 60 years and (2) unstable state in between containing six major discontinuities alternating with rapid δ18O positive excursions. The two contrasting states are likely the manifestations of extreme rainfall events established on the basis of the imprints of anomalously high/low drip ﬂow rates discerned in the fabrics of the discontinuities. The proxy rainfall record offers valuable insights on whether climate variability may have been implicated in the rise and demise of the Mississippian chiefdoms (A.D. 800 to 1700) in the southeastern United States. The time of emergence and growth of the Mississippian chiefdoms and their subsistence transition to a dependency on corn, a warm and wet weather crop, coincides with a period of increased rainfall over the A.D. 450–1000 interval. Overall decline in rainfall between A.D. 1000 and 1500, alternating with droughts, is contemporaneous with evidence of abandonment of towns and villages and downstream movement of populations. Thus, warmer, wetter conditions than present may have promoted corn agriculture during the rise and growth of the chiefdoms, whereas food shortages, caused by failed corn crops under drought conditions, may have played a much greater role in the demise of the Mississippian chiefdoms than previously recognized.

1. INTRODUCTION successful books [e.g., Diamond, 2005 among others]. Two opposing views concerning the cause and duration of the The question of why and how societies collapse has long past societal crises were recently summarized by Lawler been a subject of debate and constitutes a topic of highly [2010]. One view emphasizes that societal changes involve a complex long decline and transformation. The opposing views point to environmental disasters as the primary cause Climates, Landscapes, and Civilizations of abrupt societal falls. The two most cited examples of the Geophysical Monograph Series 198 role of abrupt regional droughts in the collapse of ancient © 2012. American Geophysical Union. All Rights Reserved. civilizations are the fall of the Akkadian Empire in Meso- 10.1029/2012GM001203 potamia in the third millennium B.C. [Weiss et al., 1993] 35 36 RAINFALL VARIABILITY AND RISE-FALL OF MISSISSIPPIAN CHIEFDOMS and the classic Maya collapse around 900 A.D. [Hodell 1985) of spring-rainfall reconstruction in the Southeast et al., 1995]. United States. The tree ring-based records exhibit large Archeological records from the Southeast United States changes in interannual rainfall variability during the growth attest to the rise and growth of the Mississippian chiefdoms and collapse of the Mississippian polities thus prompting between A.D. 800 and 1000 and their demise between A.D. the suggestion of a cause and effect relation [Burnett and 1500 and 1700 [Anderson, 1994; Smith, 2000]. Ceremonial Murray, 1993]. mounds, the most characteristic artifacts of the chiefdoms, Here we report a newly acquired proxy rainfall record of ceased by the end of the sixteenth century (Figure 1). The the last four millennia archived in a stalagmite from DeSoto factors that precipitated the abrupt decline of the Mississip- (Kymulga) Caverns in Alabama (Figure 1) and address the pian polities continue to be a subject of heated debate among question of whether climate variability may have been im- researchers. Warfare, resource depletion, climate change, and plicated in the rise, growth, and collapse of the Mississippian epidemic diseases brought by the European expeditions in polities. the sixteenth century are the most prominent factors invoked [Burnett and Murray, 1993; Hudson, 1997]. 2. STUDY SETTING The hypothesis that climate change may have impacted the fate of the Mississippian chiefdoms has so far received little DeSoto Caverns (Childersburg, Alabama, Figure 1 inset) attention because of the scarcity of contemporaneous local encompass a karst system developed in the Upper Ordovi- and regional paleoclimate records of sufﬁcient time span and cian dolomite. The cave, ~150 m in length, consists of a resolution. Exceptional are the paleorainfall studies of bald- frontal “cathedral”-size chamber with a 36 m high ceiling cypress tree ring chronologies by Stahle and Cleveland and interconnected small chambers in the back. Active and [1992, 1994] who rendered over a millennium (A.D. 933– fossil massive stalagmites and stalactites, composed predominantly of aragonite, give the cave a richly decorated appearance. Native Americans used the cave as a burial ground at least 2000 years ago during the Woodland period and also witnessed the arrival of Hernando DeSoto’s expedition in A.D. 1540 [Lambert and Aharon, 2010]. An interannual monitoring study of the regional rainfall, the cave’s ambient environment, and dripwater isotope chemistry was carried out between 2005 and 2008 in order to assess the suitability of the cave stalagmites to serve as archives of past rainfall variability using oxygen and carbon isotope rainfall proxies. The results of the monitoring study at DeSoto Caverns are reported by Lambert and Aharon [2010, 2011].

3. METHODS

A 20 cm tall, actively depositing, stalagmite (DSSG-4) was cored from the front chamber of the DeSoto Caverns on August 2008 using a water-cooled electric drill (Figure 2). The 5 cm diameter core was returned to the speleothem laboratory at the University of Alabama and sliced along the axis of maximum growth using a water-cooled diamond saw. The stalagmite is made of predominantly white-pearly aragonite as confirmed by X-ray diffraction pattern. The upper Figure 1. Map showing the extent of the Mississippian Chiefdoms 4 cm of the stalagmite displays mm-size thick-dark layers in North America (A.D. 800–1700) (modified from Cobb [2003] and separated by thin light-colored layers. Observations of thin http://en.wikipedia.org/wiki/Southeastern_Ceremonial_Complex). Towns and villages are marked by large and small circles, respec- sections under a petrographic microscope have unraveled the fi tively, while DeSoto Caverns, located in the middle of the Southeast occurrence of nine discontinuities in the upper fth of the Mississippian communities, are marked by a triangle. Inset map cored stalagmite (Figure 2a). shows the location of DeSoto Caverns (star) in the Gulf Coast region The chronology of the stalagmite was obtained from ~20 mg (33°18′26″N, 86°16′36″W). aragonite material extracted along the axis of maximum AHARON ET AL. 37

Figure 2. (a) Cored stalagmite DSSG4 and (b) its age model. The age model is based on 19 234U/230Th radiometric dates by MC-ICP- MS and is reported in calendar years relative to A.D. 1950; 2σ errors are indicated by vertical bars. The nine discontinuities in the upper fifth of the stalagmite are marked by white lines on the slab image Figure 3. Oxygen isotope time series acquired from stalagmite and by gray bars on the age model; bars show the location of DSSG4. Negative and positive δ18O excursions above and below samples drilled for radiometric dating. the mean value of À3.3‰ (thin lines in Figures 3a and 3b) are interpreted as hydroclimate events representing elevated and re- growth using a dental drill (Figure 2a). Nineteen precise duced annual rainfall amounts, respectively (see text). (a) The δ18 234U/230Th age determinations (2σ error typically <1% of O values graphed against depth below the top (DBT) of the value) (Figure 2b) were acquired by multiple collector–in- stalagmite cored in 2008. Crosses mark the 1459 sample measurements. Vertical dashed lines indicate the position of the nine dis- ductively coupled plasma–mass spectrometer (MC-ICP-MS) continuities. (b) The δ18O time series graphed against calendar age as described by Hellstrom [2003]. Analytical details of the 234 230 (years B.P.) and year (A.D./before the Common Era). Circles denote U/ Th age measurements will be reported elsewhere. the 234U/230Th radiometric dates, and horizontal bars are 2σ errors. Samples for stable isotope analysis of about 50–80 µg by The apparent duration of the discontinuities was established on the weight were acquired at 100 µm intervals along the axis of assumption that growth rates were invariable in the first 1500 years maximum growth using a manual micromilling device. of stalagmite deposition (4.1 mm/100 years). On this basis, we Oxygen isotope determinations of samples and NBS-19 estimate that about X cm of the stalagmite was removed by back standard powders were made in a continuous flow mode dissolution. Xs mark times of stalagmite deposition during droughts using a Gasbench coupled to a modified Delta-plus{©} iso- interspersed between episodes of excess rain. Time lines of succes- tope ratio mass spectrometer at the University of Alabama sive Native American cultural periods in the Southeast United Stable Isotope Laboratory. The precision, accuracy, and States are from Hudson [1976], Cobb and Garrow [1996], and Burnett and Murray [1993]. Squares marked by “S” and “F” show reproducibility of the isotope determinations are estimated ‰ the time of Spanish, followed by French, expeditions to the South- to be ±0.1 on the basis of multiple sample and standard east. The time gap between the departure of the Spanish and the repeats. The results (Figure 3), reported in the customary arrival of the French (1543 to 1673) represents the protohistoric 18 delta notation (δ O ‰ Vienna Peedee belemnite (VPDB)), “Dark Ages” [Burnett and Murray, 1993]. Mississippian chiefdom were corrected for the fractionation resulting from differ- evolution phases, rise, growth, and collapse, are abbreviated by the ences between aragonite samples and calcite NBS-19 stan- initials r, g, and c, respectively. dard during reaction with the orthophosphoric acid and release of CO2 according to Kim et al. [2007a]. 38 RAINFALL VARIABILITY AND RISE-FALL OF MISSISSIPPIAN CHIEFDOMS

4. RESULTS

Oxygen isotope compositions acquired from the DSSG4 stalagmite are graphed against their depth below top (DBT) in Figure 3a. The δ18O values have a mean of À3.3 ± 0.5‰ (n = 1459) with the most negative and positive excursions from the mean of 1.6‰ and 1.3‰, respectively. The growth rates of the DSSG4 stalagmite are remarkably uniform and range from 4.0 mm/100 years in the upper 1.2 cm, 4.2 mm/100 years in the 2–4 cm interval, and 5.3 mm/100 years in the 4–15 cm interval, excluding the section punctuated by discontinuities and the bottom 4 cm of the cored stalagmite containing the transition from the stalagmite to a flowstone pedestal (Figure 2). Hence, given a sampling interval of 0.1 mm, the overall resolving power of the δ18Otimeseries(Figure 3b) is interannual, ranging from 2 to 5 years. Oxygen isotope time series are graphed against calendar age in Figure 3b using the age model developed in Figure 2b. The time series exhibit two principal modes of isotope variability: (1) a stable nonlinear mode and (2) an unstable, irregular mode. The stable mode is practically continuous (except for the short-lived unconformity number 9), spans the time intervals 4200 to 1400 calendar years B.P. (cal B.P.) Figure 4. Power spectra analysis of δ18O time series shown in (before the Common Era (B.C.E.) 2350 to A.D. 400) and 200 Figure 3b. (a) Continuous δ18O data are restricted to the time cal B.P. to the present (A.D. 1700 to 2008) and exhibits interval 2000–4200 calendar years B.P. (cal B.P.) preceding the multidecadal variability (Figure 4a) with statistically signif- earliest unconformity (nine in Figure 3b). Dashed line represents icant periodicities of 30 and 60 years (Figure 4b). In sharp the mean value of the time series. (b) Relative power spectra density plotted against the frequency ( f ) exhibits prominent periodicities at contrast, the interval between 1400 to 200 cal B.P. (A.D. 400 30 ± 2 and 60 ± 6 years (black wiggle) that exceeds the red noise to 1700) contains six major discontinuities alternating with spectrum (black line) and are statistically significant at 95% confi- δ18 rapid O positive excursions. We argue that the two con- dence level (dashed line). The periodicity peaks are in good agree- 18 trasting states of δ O (Figure 3b) are primarily driven by ment with the 30 and 60 year periodicities of the Atlantic rainfall variability and support our contention with the fol- multidecadal oscillation (AMO) presently governing the interdeca- lowing arguments. dal rainfall variability in the continental United States [Enfield et al., 1. Using the latest isotope temperature equation for arago- 2001; Kerr, 2005]. The coherence between the stalagmite proxy nite, inclusive of the quoted uncertainties [Kim et al., 2007b], rainfall spectra and AMO suggests that the latter had likely been the a δ18O range of À4.4 to À5.1 (‰ Vienna SMOW principal controlling factor of the quasi-regular variability observed (VSMOW)) for multiple dripwaters sampled over 3 years at prior to the onset of the phase change to a climate instability interval δ18 DeSoto Caverns and a constant cave air temperature of at ~1500 cal B.P. (~A.D. 450). Power spectra analysis of O output variables (N = 435) at a time step of Δt = 5 years was 17.9°C [Lambert and Aharon, 2010], we obtain a predicted performed using the ARAND software [Howell et al., 2006]. equilibrium δ18O aragonite values in the range of À4.7 to À5.6 (‰ VPDB). Twelve aragonite samples scrapped with a fine scalpel from the tops of active stalagmites at DeSoto 2. Three years of continuous regional rainfall monitoring Caverns yield a mean δ18O value of À4.6 ± 0.5 (‰ VPDB) (2005–2008) yield an annual weighted mean of À4.7 (‰ that is within the range of the uppermost ten δ18O determina- VSMOW) with significant departures from the mean being tions at the top of the stalagmite (À3.9 to À4.6 ‰, Figure 3) associated with anomalously high rainfall from tropical and in agreement with the predicted δ18O range. Hence, it is storms (weighted δ18O mean of À7.0 ‰ VSMOW) and reasonable to assume that aragonites forming the stalagmites 18O-enrichments of up to 2.7‰ relative to the mean associ- at DeSoto Caverns are deposited in isotope equilibrium with ated with drought years [Lambert and Aharon, 2010]. cave drips and air temperature and the observed variability 3. If the observed positive δ18O excursions of up to 2.9‰ likely tracks changes in rainfall δ18O values [Lambert and above the present value (Figure 3) were caused by a fall in Aharon, 2010]. cave air temperature, then an overall drop of 14.5°C below AHARON ET AL. 39

We interpret the repetitive sequence of the six discontinuities as being the imprints of metastable aragonite dissolution by unsaturated drips with high flow rates caused by excessive rainfall episodes and support this interpretation with the following observations: (1) truncation of the aragonite botryoids at the base of the discontinuity resulting from dissolution and infill of the voids by fine, dark-color, detritus lag (2 and 3 in Figure 5b), (2) growth of randomly oriented, large (>1 mm), aragonite botryoids (4 in Figure 5b) likely deposited from dripwater ponded on stalagmite tops, and (3) negative δ18O excursions at the initiation of the discontinuities that are succeeded by rapid shifts to 18O-enrichments at their termination. The six δ18O positive excursions (marked by X in Figure 3b) are interpreted as being the manifestation of rainfall deficits (droughts) on the basis of our 3 year monitoring study [Lambert and Aharon, 2010]. Similar fabrics associated with discontinuities caused by dissolution of calcite stalagmites were reported by Turgeon and Lundberg [2001] from the southwest Oregon caves. The remaining Figure 5. Photomicrographs of aragonite fabrics in DeSoto cored three discontinuities (1, 2, and 9 in Figure 3b) are poorly stalagmites. (a) Typical stalagmite fabric in thin section consisting developed, exhibit incomplete dissolution sequences (Figure of densely packed rows of aragonite botryoids with their long axis 4b), and are likely of brief, subdecadal, duration. oriented in the direction of growth (plane polarized light). (b) Fabric 18 of discontinuity number 4 (Figures 2 and 3) showing a succession of Whereas the timing of the drought-driven O-enrichments distinct sequences from bottom to top: aragonite botryoids (1) show- at the termination of the major discontinuities can be estab- ing dissolution marks and cavities (2); cavities infilled by dark, fine lished with relative accuracy (circa A.D. 485, 660, 835, detritus (3); randomly oriented, large aragonite botryoids likely 1020, 1290, and 1520), the timing and duration of the six deposited from dripwater ponded on stalagmite tops (4); resumption anomalous wet years can only be approximated (circa A.D. of normal aragonite deposition (5). The fabrics restricted to the six sixth, ninth, eleventh, thirteenth, sixteenth, and seventeenth major discontinuities (three to eight in Figure 3b) are interpreted as centuries) because of back dissolution that removed older being the manifestation of anomalous high rainfalls causing fast stalagmite layers. drips and aragonite dissolution (plane polarized light). 5. DISCUSSION AND CONCLUSIONS the present value of 17.9°C is predicted (i.e., 0.2‰ °CÀ1). In view of the fact that global air temperatures in the past 1000 Three successive Native American cultural periods in the years have changed up or down by no more than 0.5°C past 4000 years are recognized in the southeastern United relative to the twentieth century [Crowley and Lowery, States [Hudson, 1976]: (1) Late Archaic (B.C.E. 3000– 2000], it can be concluded that the δ18O excursions (Figure 1000), (2) Woodland (B.C.E. 1100–A.D. 800), and (3) Mis- 3) are unlikely to be caused by air temperature shifts either sissippian (A.D. 800–1700). Here we are interested in the within or outside the cave but, instead, are tracking rainfall reasons for the cultural degradations that occurred at the variability changes through time. transition between the Woodland and Mississippian periods The nature of the nine discontinuities shown in Figure 2a (A.D. 450–800) and the factors that gave rise to the growth and their durations in Figure 3b are important in the inter- and ultimate collapse of the Mississippian chiefdoms. pretation of the δ18O record as rainfall variability proxy and The archeological record in the Southeast attests that the require further explanation. Late Woodland was characterized by poorly understood cul- The typical fabric of DeSoto Caverns stalagmites exhibits tural decline including food shortages, interruption of long- tightly packed rows of botryoidal aragonite, about 0.5 mm in distance trade, and halting of burial ground construction length, whose fibrous long axis is oriented upward in the [Cobb and Garrow, 1996]. Equally confounding is the sud- direction of growth (Figure 5a). In contrast, the common den appearance throughout the Southeast of Mississippian denominator of six major discontinuities (three to eight in sites within the river floodplains at circa A.D. 850 [Smith, Figure 3b) is disruption of the typical depositional fabric and 2000]. The two cultures differed significantly in their subsis- its replacement by a complex upward sequence (Figure 5b). tence economies and mode of governance. Whereas the 40 RAINFALL VARIABILITY AND RISE-FALL OF MISSISSIPPIAN CHIEFDOMS

Woodland subsistence was based on hunting/gathering, es- the encounter with the Spanish expeditions [Hudson, sentially similar to the Late Archaic period, leading to sea- 1997]. Our rainfall proxy data show that the most severe sonal nutritional stress, the Mississippian was an agricultural drought conditions occurred between A.D. 1300 and 1650 economy relying primarily on maize and a sociopolitical and may have played a much greater role in the demise of the governance of elite chiefs who were responsible for the Mississippian society than previously recognized. storage and distribution of food during times of crop failures The episodicity of wet events alternating with droughts [Cobb and Garrow, 1996]. during the Mississippian period, unfolded in the stalagmite A number of causes have been proposed to explain the δ18O time series, complements the findings from rainfall- decline of the Woodland culture and the rise and growth of sensitive tree ring study of baldcypress trees in Southeast the seemingly successful Mississippian chiefdoms [Ander- United States [Stahle and Cleavelend, 1992, 1994]. The tree son, 1994]: (1) nutritional stress caused by population in- ring-reconstructed rainfall over the Southeast United States crease, (2) overexploitation of large game animals, (3) colder shows large shifts in interannual variability during the Mis- climate conditions affecting gathering of wild plants, and (4) sissippian interval with dry cycles persisting during the rainfall variability that caused shortages and famine. twelfth and sixteenth centuries and substantial rainfall vari- The continuous δ18O time series in the early interval of the ance during the thirteenth, seventeenth, and twentieth centu- stalagmite growth correspond to the Late Archaic (B.C.E. ries. The comparison between the tree rings and stalagmite 2340–1000) and most of the Woodland period (B.C.E. 1000 data, however, must be viewed with caution. Whereas the to A.D. 800), whereas the stalagmite section containing the baldcypress tree rings are particularly sensitive to spring pronounced six discontinuities coincides with the terminal rainfall variance (March–June), the stalagmite δ18Otime Woodland and the rise, growth, and collapse of the Missis- series integrate interannual water budget changes, a fact that sippian chiefdoms (Figure 3b). A phase change in the inter- must be considered when comparing tree rings and stalag- annual rainfall state occurred around A.D. 450 (i.e., oldest X mite rainfall proxy records. mark in Figure 3b) that separates between an interval of Climate change and its societal impact during prehistoric stable climate with periodicities of 30 and 60 years and an and early historic times has become a fertile research topic of unstable, irregular, climate-encompassing periods of excess great practical significance because of implications rain alternating with droughts over the time interval A.D. concerning how present global warming may affect our own 450–1700. society’s future [Anderson, 2001]. The Southeast United Here we contend that the observed change in the rainfall States contains a trove of archeological sites discerning pre- pattern occurring at around A.D. 450 may have been the historic cultural changes and organizational complexities of principal cause of the Late Woodland slide into poverty and the Native Americans but scarce contemporaneous paleocli- decline and the rise and growth of the Mississippian chief- mate reconstructions. Karst is extensive on a regional scale, doms. This is because the Late Woodland society whose and cave deposits such as stalagmites could offer unmatched subsistence economy was based primarily on exploitation of archives of past climate changes that are amenable to precise wild plants and animals was lacking the tools and organiza- chronologies based on high-resolution radiometric dating tion to cope with extreme rainfall events. In contrast, the and tool kits containing multiple climate proxies consisting emerging Mississippian chiefdoms have developed the skills of stable isotopes and elemental chemistry. to engage in large-scale cultivation of corn, a warm and wet In order to corroborate the interpretations advanced here, weather crop, and initiate innovative technology to store food the DeSoto Caverns results need to be duplicated from an- surpluses and distribute it during times of crop failure. Our other stalagmite in the same cave and karst paleoclimate contentions agree with Smith’s view [Smith, 1990] that the studies be expanded to other caves in the Southeast United Mississippian sweeping emergence in the Southeast repre- States that offer a fertile ground for testing proposed links sents responses to climate changes and population growth. between past climate changes and contemporaneous societal It has long been assumed that the fall and disappearance of transitions. the heavily populated Late Mississippian chiefdoms (A.D. 1500–1700) was initiated by epidemic diseases and enslave- Acknowledgments. The study of late Holocene stalagmites from ment brought by a string of Spanish expeditions to the – DeSoto Caverns was funded by grants to the senior author from the Southeast (i.e., DeSoto, A.D. 1539 1543; Tristan de Luna Alabama State Climatologist Office (State Climatologist: John – – y Arellano, A.D. 1559 1561; Juan Pardo, A.D. 1566 1568) Christy) and a faculty grant from the Gulf Coast Association of [Burnett and Murray, 1993]. However, archeological evi- Geological Societies (GCAGS). David Aldridge was the recipient dence seems to indicate that a large number of Late Missis- of graduate student grants from the Geological Society of America sippian settlement abandonment occurred centuries before (GSA) and GCAGS. We thank members of the SPELEOTEAM AHARON ET AL. 41 group (Joe Lambert, Rajesh Dhungana, and Hunter Phillips) for Kerr, R. A. (2005), Atlantic climate pacemaker for millennia past, assistance with the mass spectrometry and fruitful discussions. decades hence?, Science, 309,41–43. Book editor Peter Clift and two anonymous reviewers contributed Kim, S.-T., A. Mucci, and B. E. Taylor (2007a), Phosphoric acid constructive comments on the submitted manuscript. The senior fractionation factors for calcite and aragonite between 25 and author dedicates this contribution to the memory of his late wife, 75 °C: Revisited, Chem. Geol., 246, 135–146. Sylvia Steiner Aaron, whose love and support were unwavering Kim, S.-T., J. R. O’Neil, C. Hillaire-Marcel, and A. Mucci (2007b), throughout her life. Oxygen isotope fractionation between synthetic aragonite and water: Influence of temperature and Mg2+ concentration, Geo- REFERENCES chim. Cosmochim. Acta, 71, 4704–4715. Lambert, W. J., and P. Aharon (2010), Oxygen and hydrogen isotopes Anderson, D. G. (1994), The Savannah River Chiefdoms: Political of rainfall and dripwater at DeSoto Caverns (Alabama, USA): Key change in the Late Prehistoric Southeast, 480 pp., The Univ. of to understanding past variability of moisture transport from the Ala. Press, Tuscaloosa. Gulf of Mexico, Geochim. Cosmochim. Acta, 74,846–861. Anderson, D. G. (2001), Climate and culture change in prehistoric Lambert, W. J., and P. Aharon (2011), Controls on dissolved inor- and early historic eastern North America, Archeol. East. North ganic carbon and δ13C in cave waters from DeSoto Caverns: Am., 29, 143–186. Implications for δ13C assessments, Geochim. Cosmochim. Acta, Burnett, B. A., and K. A. Murray (1993), Death, drought and 75, 753–768. DeSoto: The bioarcheology of depopulation, in Proceedings of Lawler, A. (2010), Collapse? What collapse? Societal change re- the DeSoto Symposia, edited by G. A. Young and M. P. Hoffman, visited, Science, 330, 907–909. pp. 227–254, The Univ. of Arkansas Press, Fayetteville. Smith, B. D. (1990), Introduction: Research on the origins of Cobb, C. R. (2003), Mississippian chiefdoms: How complex?, Mississippian chiefdoms in eastern North America, in The Mis- Annu. Rev. Anthropol., 32,63–84. sissippian Emergence, edited by B. D. Smith, pp. 1–8, Smithson. Cobb, C. R., and P. H. Garrow (1996), Woodstock culture and Inst. Press, Washington, D. C. the question of Mississippian emergence, Am. Antiquity, 61(1), Smith, M. T. (2000), Coosa: The Rise and Fall of a Southeastern 21–37. Mississippian Chiefdom, 176 pp., Univ. Press of Fla., Gainesville. Crowley, T. J., and T. S. Lowery (2000), How warm was the Stahle, D. W., and M. K. Cleaveland (1992), Reconstruction and Medieval Warm Period?, Ambio, 29(1), 51–54. analysis of spring rainfall over the Southeastern U.S. for the past Diamond, J. (2005), Collapse: How Societies Choose to Fail or 1000 years, Bull. Am. Meteorol. Soc., 73, 1947–1961. Succeed, 592 pp., Viking Press, New York. Stahle, D. W., and M. K. Cleaveland (1994), Tree-ring recon- Enfield, D. B., A. M. Mestas-Nuñez, and P. J. Trimble (2001), The structed rainfall over the southeastern USA during the Medieval Atlantic Multidecadal Oscillation and its relation to rainfall and Warm Period and Little Ice Age, Clim. Change, 26, 199–212. river flows in the continental U.S., Geophys. Res. Lett., 28(10), Turgeon, S., and J. Lundberg (2001), Chronology of discontinuities 2077–2080. and petrology of speleothems as paleoclimatic indicators of the Hellstrom, J. (2003), Rapid and accurate U/Th dating using parallel Klamath Mountains, Southwest Oregon, USA, Carbonates Eva- ion-counting multi-collector ICP-MS, J. Anal. At. Spectrom., 18, porites, 16(2), 153–167. 1346–1351. Weiss, H., M. A. Courty, W. Wetterstrom, F. Guichard, L. Senior, R. Hodell, D. A., J. H. Curtis, and M. Brenner (1995), Possible role of Meadow, and A. Curnow (1993), The genesis and collapse of climate in the collapse of Classic Maya civilization, Nature, 375, third millennium North Mesopotamian Civilization, Science, 391–394. 261, 997–1004. Howell, P., N. Pisias, J. Balance, J. Baughman, and L. Ochs (2006), Time Series Analysis Software, Brown Univ., Providence, R. I. P. Aharon and D. Aldridge, Department of Geological Sciences, Hudson, C. (1976), The Southeastern Indians, 592 pp., Univ. of University of Alabama, Box 870338, Tuscaloosa, AL 35487, USA. Tenn. Press, Knoxville. ([email protected]) Hudson, C. (1997), Knights of Spain, Warriors of the Sun, 561 pp., J. Hellstrom, School of Earth Sciences, University of Mel- The Univ. of Ga. Press, Athens. bourne, Melbourne, VIC 3010, Australia. Building Resilience in Island Communities: A Paleotempestological Perspective

Jago Cooper

UCL Institute of Archaeology, London, UK

A review of past hurricane activity over the past 8000 years in the Caribbean has been correlated with a comparative study of changing settlement locations, food procurement strategies, and household architecture traditions over time. The framework for human decision making is evaluated, and the multitemporal and multi- spatial complexity of social change is put into a regional context. This study highlights the conceptual friction between linking thresholds in climate change, tipping points of sudden environmental change, and divergent pathways through societal progression. The relative resilience of different community lifeways is assessed in light of known climatic and environmental hazards in the region, and key lessons about threat and mitigation are learned that have important implications for modern-day communities living in the Caribbean.

1. INTRODUCTION: LIVING WITH GLOBAL CHANGE and paleoenvironmental data for these events are explored at different spatial scales in order to reveal the multitemporal Humans have always lived with the impacts of extreme nature of “change,”“impact,” and “adaptation” concepts. weather events and sudden environmental change. There- Central to this study is the idea that social lifeways and fore, archaeology can provide long-term perspectives on human decisions, in the face of climatic and environmental how different human communities have experienced the hazards, are the most important factors in determining the physical manifestations of climate variability and environ- relative impact of climate variability upon past human com- mental change. The research that I shall discuss in this paper munities and long-term social development [Kohler and van emerges from a study of the impacts of precipitation varia- der Leeuw,2007;Redman and Kinzig, 2003; Sheets and tion, periodic intensification of hurricane activity and rising Cooper, 2012]. relative sea levels on pre-Columbian populations in the Ca- ribbean, and the relative advantages of different settlement 2. CARIBBEAN REGION OVERVIEW locations, food procurement systems, and household architecture designs in the face of these climatic and environmen- In the Caribbean, a sense of foreboding currently perme- tal hazards [Cooper and Peros, 2010; Cooper and ates all levels of society from the individual fisherman in Boothroyd, 2011]. I will provide a brief overview of the northern Cuba to the large interregional governing bodies human experience of extreme weather events and the impor- [Caribbean Community Climate Change Centre, 2009, p. 6]. tance of understanding the temporal parameters of climatic This sense of vulnerability arises from the perceived threat of variability. This work explores how human communities in climate change and the prospect of increased hurricane in- the Caribbean have lived with the impacts of hurricane tensity, changing rainfall patterns, and rising pressures of events during the mid to late Holocene. The paleoclimatic coastal flooding and erosion due to relative sea level rises [Maul, 1993]. The pressures of climate variability are seen as the greatest threat to the sustainable development of the Climates, Landscapes, and Civilizations region, and reactions to recent extreme weather events, such Geophysical Monograph Series 198 as hurricane Ike in 2008 causing over $29,520,000,000 USD © 2012. American Geophysical Union. All Rights Reserved. of damage [Blake et al., 2011, p. 9], show how this connec- 10.1029/2012GM001212 tion between global climate change and local weather events 43 44 BUILDING RESILIENCE IN ISLAND COMMUNITIES is being created in the Caribbean consciousness. Archaeology [IPCC, 2012, p. 19]. Therefore, using historical and SST can make an important contribution by revealing the nature proxy data to assess hurricane variability cannot be reliably of this threat and informing ongoing discussions of potential retrodicted back into the past to provide a deeper time per- mitigation strategies [Puig Gonzales et al., 2010]. This can spective on hurricane frequency. Consequently, improved pre- be achieved by comparatively studying the human experi- historic data collation at the regional scale is required, in ence of climate variability and environmental change over addition to the modeling of potential taphonomic and sam- decadal, centennial, and millennial timescales. The long- pling bias, to provide a long-term perspective on the human term perspectives developed through such regional interdis- experience of hurricanes in the Caribbean. This work on ciplinary collaborations can help (1) establish the known Caribbean paleotempestology is currently being carried out parameters of Holocene climate variability in the region; by a number of scholars in the region [Frappier et al.,2007] (2) contextualize the relationship between climate variability, and provides archaeologists with a useful paleotempestologi- sudden environmental change and the human experience of cal data set with which to discuss the human experience of extreme weather events; (3) reveal the spatially and tempo- hurricane impacts and variability in the Caribbean [Cooper rally scaled nature of impact upon individuals, communities, and Peros, 2010]. and societies; and (4) explore how different lifeways mitigate or exacerbate hazards and create the relative vulnerability/ 4. CARIBBEAN PALEOTEMPESTOLOGY robustness of different communities. The first historical accounts of hurricanes in the Caribbean, 3. PARAMETERS OF CLIMATE VARIABILITY such as the devastation wrought on Santo Domingo in A.D. 1508 [de Oviedo y Valdés, 1851, p. 168], are recorded by People in the Caribbean believe that hurricanes are becom- Christopher Columbus and Gonzalo Fernandez de Oviedo y ing more frequent and more intense in recent years [Pages, Valdés. These accounts of observed hurricanes recount the 2005], but whether this is accurate is hard to determine human experience and devastating impact of hurricane activ- [Elsner, 2007]. A long-term perspective of the parameters of ity in the Caribbean and have also been used to discuss the Holocene climate variability that includes the entire human nature of impact and possible variation in landfall patterns of occupation of the islands (beginning in circa 5000 B.P.) can hurricanes [Saunders and Lea, 2008]. However, criticisms provide perspective, knowledge, and context for modern-day that highlight the sampling bias and unreliability of observed communities and inform perceptions of risk and vulnerabil- hurricanes in the historical archives, particularly as a means of ity. The collection, collation, and analysis of paleotempesto- quantifying hurricane frequency or qualifying hurricane inten- logical (reconstruction of ancient storms) data for hurricane sity, have led to a research focus on material evidence for past frequency and intensity are an excellent source for develop- hurricanes using geomorphological analyses of sediment ing this long-term perspective. The primary sources of pa- cores and isotopic data as paleotempestological proxies leotempestological data from the Caribbean come from [Woodruff et al., 2008a]. Various studies have created a series historical archives [Nyberg et al., 2007], studies of geomor- of local paleotempestological records that, when brought to- phological sediments [Donnelly and Woodruff, 2007], and gether, form a useful overview of regional mid-to-late Holo- proxy isotopic records [Greer and Swart, 2006]. cene hurricane activity. These data have a broad spatial and In recent years, there has been some lively discussion temporal range coming from the Bahamas [Bourrouilh-Le surrounding perceived links between increases in hurricane Jan, 2007], Belize [Gischler et al., 2008; McCloskey and activity recorded in historical records and increasing sea Keller, 2009], Cuba [Peros et al., 2007], Dominican Republic surface temperatures (SST) in the North Atlantic [Elsner, [Kennedy et al., 2009], Jamaica [McFarlane et al., 2002], St. 2007; Hetzinger et al., 2008; Nyberg et al., 2007, p. 698; Martin [Bertran et al., 2004; Malaize et al., 2011], Vieques Saunders and Lea, 2008]. These modern and historical data [Donnelly and Woodruff,2007;Woodruff et al., 2008b], and have some important flaws, such as the variable data resolu- U.S. Virgin Islands [Jessen et al., 2008] among other location for modern and historic paleotempestological records. tions. These records extend the time depth of hurricane activ- Furthermore, the argument for a connection between anthro- ity in the Caribbean beyond the historic and provide a regional pologically induced global warming and increased storm paleotempestological data set with which to discuss changing frequency and intensity makes modern data a problematic patterns of frequency and intensity of hurricane activity. proxy for the past. Certainly, the latest IPCC report would An overview of these data suggests that frequency of argue that while it is highly likely that increased SSTs will hurricanes on individual islands in the Caribbean is directly lead to increased hurricane intensity (wind speeds), it is still linked to variation in the North Atlantic climate systems. uncertain whether it will lead to increased hurricane frequency These conditions include changing SSTs and North Atlantic COOPER 45 wind vectors, North Atlantic oscillation patterns, El Niño– Consequently, understanding the scaled temporality of Southern Oscillation effects and the shifting position of hurricane impacts, frequency, and intensity requires an ap- the Intertropical Convergence Zone [Chang et al., 2000; proach that uses the full time depth of human experience in Handoh et al., 2006; Haug et al., 2001]. This has led scholars the region to contextualize changing patterns of impact to hypothesize how the impacts of global climate change through time [Cooper and Peros, 2010]. Such a long-term affect local hurricane patterns in the Caribbean [Frappier et approach shows that there is not a linear relationship between al., 2007; Haug et al., 2001]. The argument for linking the scale of destructive force of a hurricane and the resultant hurricane variability in the Holocene to changes in hurricane devastation on human communities; rather, all impacts are formation conditions has been validated through compari- contingent on the contemporary local social and environ- sons of modern SST data with historical hurricane records. mental context [Cooper, 2012]. This is certainly a lesson Such research has led Black et al. [2007, p. 7], among others, from the past that has been well established in other geo- to argue for a clear correlation between Caribbean SST and graphical areas such as the Ancestral Puebloan case studies increased storm/hurricane activity with more tropical storms of precipitation variation in the Southwest United States and hurricanes when SSTs are warmer. This proxy data [Hegmon et al., 2008; Nelson et al., 2012]. linkage is interesting and holds intriguing possibilities for future research, as there are a number of SST records in the 6. LIFE CYCLES OF THE BUILT ENVIRONMENT Caribbean. Therefore, the paleotempestological data in the Caribbean as well as the substantial data on North Atlantic Ethnohistorical sources can reveal the social context of climate variability suggest that hurricane frequency and in- past hurricanes in the Caribbean. These accounts highlight tensity have fluctuated in the mid-to-late Holocene. how pre-Columbian communities had a sophisticated under- The key factor that arises from my study of paleotempes- standing of hurricane events that temporally sequenced the tology in the mid-to-late Holocene is that hurricanes have hurricane between the coming of the winds, the destructive always presented a significant hazard for human communi- force of the hurricane, and the posthurricane event impacts ties in the Caribbean; however, their frequency and intensity that can ensue. Each of these stages of the hurricane event on different islands varies on annual, decadal, centennial, were represented by different deities, Gatauba, Guabancex, and millennial timescales. This poses important questions and Coatrisque, respectively [Pané, 1990], indicating that regarding how pre-Columbian communities lived with the this ecological knowledge was deeply ingrained in the cul- destructive forces of the hurricane in the Caribbean and how tural matrix of pre-Columbian societies in the Caribbean people coped with the variability in frequency and intensity [Newsom and Wing, 2004, p. 8]. of hurricane landfalls. This understanding of the scaled temporality of hurricane impact shows how potential mitigation of the hazard might 5. NATURE OF HUMAN EXPERIENCE AND IMPACT have been achieved. For example, recent studies of pre- Columbian household architecture and the effects of hurri- Hurricanes in the Caribbean embody some of the most cane impacts in the Caribbean reveal some interesting ideas destructive extreme weather events on the planet. From a on vulnerability, particularly when compared to modern-day human perspective, however, it is not just a single event of architectural designs. Detailed excavation of well-preserved destruction created by wind speeds above 72 mph; it is also house structures at the pre-Columbian site of Los Buchil- the associated coastal storm surges, posthurricane flooding, lones in northern Cuba provide an opportunity to examine and disruption to flora and fauna that have a long-term effect the relative resilience of these structures to hurricane impacts and prevent returns to prior conditions. This idea of multiple in more detail [Cooper et al., 2010; Valcárcel Rojas et al., timescales of consequence from a single weather event is key 2006]. Pre-Columbian houses, exemplified by those at Los to the human perspective, as mitigation strategies need to Buchillones, were built around a robust wooden post skele- take into account this scaled temporality of impact [Petitjean ton firmly embedded into the ground soil or bedrock, aver- Roget, 2001]. The wider research project from which this aging between 6 and 8 m [Samson, 2010]. The houses at Los paper emerges has identified that hurricanes often represent Buchillones were stilted houses that appear to have had their tipping points in longer-term processes of climatic and envi- floors suspended above water or seasonally wetland condi- ronmental change. For example, hurricane storm surges tions. These predominantly round or oval structures were breach thresholds where the pressures of long-term relative dressed in a lightweight skin of wooden rafters, stringers, sea level rise are manifested in abrupt coastal storm surges and thatched roof that provides an opportunity to reflect upon that inundate coastal areas and permanently alter coastal comparative issues of architectural resilience in the Carib- ecology [Cooper and Boothroyd, 2011]. bean (Figure 1). During a hurricane, there is no doubt that the 46 BUILDING RESILIENCE IN ISLAND COMMUNITIES

Figure 1. Photograph of reconstructed pre-Columbian house structures built in the El Baga National Park, Cayo Coco, Cuba.

Figure 2. Comparative perspectives on the relative resilience of pre-Columbian and modern household architecture designs perceived through the framework of the adaptive cycle. COOPER 47 lightweight superstructure of these houses were lifted off and with a suspended dwelling area above ground that allow for scattered by the high wind speeds. However, the structural storm surge floods and posthurricane runoff to flow beneath posts, ~30 cm in diameter and made of mahogany, are firmly the houses. Importantly, the broader theoretical mentality embedded in the soil, and these withstand the highest wind that underlies this approach to resilient architecture is that it speeds of a category 5 hurricane. Radiocarbon dating of is judged on speed of recovery rather than on robustness of these structures at Los Buchillones and associated artifacts resistance. These features are absent from many of the mod- show the house posts remain in situ and in use for over 300 ern concrete structures currently built in northern Cuba and years with repeated redressing of the lightweight superstruc- the wider Caribbean and remain one of the main reasons that ture [Cooper and Thomas, 2012; Pendergast et al., 2002]. posthurricane reconstruction costs remain so high. Therefore, these houses are not the robust shelter ideally The incorporation of traditional ecological knowledge suited for humans during the hours of the hurricane impact, within pre-Columbian belief systems shows how pre-Colum- but there is archaeological and ethnohistorical evidence to bian communities had the capacity to deal with the annual, suggest that pre-Columbian populations anticipated the com- decadal, or centennial periodicity of hurricane return rates. ing of the hurricane and took shelter in the cave systems This depth of social memory is essential for dealing with found near settlements [Cooper, 2012]. climate variability and is clearly a lesson that has not been Having survived the initial hurricane event, the scale of learned in the modern-day Caribbean, where modern settle- impact upon human communities is dependent on the speed ment planning is often focused on short-term economic with which human communities are able to recover to a “life development. This short-term focus has led to a fundamental as before” [Gunderson and Holling, 2002]. Certainly, re- change in attitude toward how to mitigate extreme weather building or rather redressing the existing house posts with events. By adopting a long-term perspective, this research easily available materials from the surrounding area shows suggests that a change of focus is required in the Caribbean how “resilient” pre-Columbian house structures are. As illus- that moves disaster management strategies away from static trated in Figure 2, the potential speed of reorganization robustness aimed at withstanding extreme weather events. stands in stark contrast to modern-day use of poor-quality Instead, dynamic recovery plans should be developed that concrete and ceramic tiles that not only require large-scale appreciate how the life cycles of the built environment need waste disposal programs but also rely on further importation to operate in synchronization with the multitemporal cycles of expensive nonlocal materials that contribute to the exces- of extreme weather events in the Caribbean region. sive costs of posthurricane reconstruction [Blake et al., 2011]. Acknowledgments. This research has been made possible through funding from the Leverhulme Trust (6/SRF/2008/0267), 7. BUILDING RESILIENCE University of Leicester School of Archaeology and Ancient History, Arts and Humanities Research Council (AH/1002596/1), Natural This discussion of the relative resilience of pre-Columbian Environment Research Council (P17233-60). I would also like to house structures poses some interesting questions surround- thank Roberto Valcárcel Rojas for his long-standing codirection of ing the benefits of long-term social memory and the preemp- this project and colleagues in Cuba who have collaborated and tive planning for swift adaptive change rather than short-term contributed to this research, including Celso Paso Alberti, Jorge á social development that attempts to provide robustness in the Calvera Rosés, Gabino la Rosa Corzo, Raúl Gómez Fern ndez, Vicente O. Rodríguez, Elena Guarch, Odalys Brito Martínez, Pedro face of external stress. By using a resilience perspective of Cruz Ramírez, Juan Carlos Mayo Rodríguez, Yanelis Buchillón the adaptive cycle, it highlights the importance of under- López, Idania Buchillón Jorge, Iosvany Martinez, Lazaro Calvo standing life cycles from a long-term perspective and illus- Iglesias, Indira Mayea, Dayessy Rojas, Daima Morel, Marcos Lab- trates how different human communities and their built rada Achoa, José Calvera, Nelson Torna, and Adrián García Lebroc. environment respond to external stress (Figure 2). This archaeological discussion surrounding the relative resilience REFERENCES of household architecture supports the argument for recent innovations in disaster management architecture that focus Amaratunga, D., and R. Haigh (2011), Reconstruction of the Built on resilient architecture [Amaratunga and Haigh, 2011]. Environment: Rebuilding for Resilience, Wiley-Blackwell, Ox- Interestingly, the pre-Columbian house structures recov- ford, U. K. ered and studied during this research exhibit many of the Bertran, P., D. Bonnissent, D. Imbert, P. Lozouet, N. Serrand, and “new” resilient house architecture features exhorted by cur- C. Stouvenot (2004), Paléoclimat des Petites Antilles depuis rent climate change resilient design planning [Watson and 4000 ans BP: l’enregistrement de la lagune de Grand-Case à Destefano, 2010, p. 131]. These include stilted structures Saint-Martin, Geoscience, 336, 1501–1510. 48 BUILDING RESILIENCE IN ISLAND COMMUNITIES

Black, D. E., M. A. Abahazi, R. C. Thunell, A. Kaplan, E. J. Tappa, the Blue Hole, Lighthouse Reef, Belize, J. Coastal Res., 24(6), and L. C. Peterson (2007), An 8-century tropical Atlantic SST 1495–1505. record from the Cariaco Basin: Baseline variability, twentieth- Greer, L., and P. K. Swart (2006), Decadal cyclicity of regional mid- century warming, and Atlantic hurricane frequency, Paleoceano- Holocene precipitation: Evidence from Dominican coral proxies, graphy, 22, PA4204, doi:10.1029/2007PA001427. Paleoceanography, 21, PA2020, doi:10.1029/2005PA001166. Blake, E. S., C. W. Landsea, and E. J. Gibney (2011), The deadliest, Gunderson, L. H., and C. S. Holling (2002), Panarchy: Under- costliest, and most intense United States tropical cyclones from standing Transformations in Human and Natural Systems, Island 1851 to 2010, report, Natl. Hurricane Cent., Miami, Fla. Press, Washington, D. C. Bourrouilh-Le Jan, F. (2007), Very high energy sedimentation (su- Handoh, I. C., A. J. Matthews, G. R. Bigg, and D. P. Stevens (2006), pratidal hurricane deposits) and Mid-Holocene highstand on Interannual variability of the tropical Atlantic independent of and carbonate platforms, Andros, Bahamas: An alternative view, associated with ENSO: Part 1. The north tropical Atlantic, Int. J. Sediment. Geol., 199,29–49. Climatol., 26, 1937–1956. Caribbean Community Climate Change Centre (2009), Climate Haug, G. H., K. A. Hughen, L. C. Peterson, D. M. Sigman, and U. change and the Caribbean: A regional framework for achieving Rohl (2001), Southward migration of the Intertropical Conver- development resilient to climate change (2009–2015), report, gence Zone through the Holocene, Science, 293, 1304–1308. Belmopan, Belize. Hegmon, M., M. A. Peeples, A. P. Kinzig, S. Kulow, C. Meegan, Chang, P., R. Saravanan, L. Ji, and G. C. Hegerl (2000), The effect and M. C. Nelson (2008), Social transformation and its human of local sea surface temperature on atmospheric circulation over costs in the prehispanic U.S. southwest, Am. Anthropol., 110(3), the tropical Atlantic sector, J. Clim., 13, 2195–2216. 313–324. Cooper, J. (2012), Fail to prepare then prepare to fail: Re-thinking Hetzinger, S., M. Pfeiffer, W.-C. Dullo, N. Keenlyside, M. Latif, threat, vulnerability and mitigation in the pre-Columbian Carib- and J. Zinke (2008), Caribbean coral tracks Atlantic Multidecadal bean, in Surviving Sudden Environmental Change: Answers from Oscillation and past hurricane activity, Geology, 36(1), 11–14. Archaeology, edited by J. Cooper and P. Sheets, pp. 91–114, IPCC (2012), Managing the Risks of Extreme Events and Disasters Univ. Press of Colo., Boulder. to Advance Climate Change Adaptation, 582 pp., Cambridge Cooper, J., and R. Boothroyd (2011), Living islands of the Carib- Univ. Press, Cambridge, U. K. bean: A view of relative sea level change from the waters edge, in Jessen, C. A., J. B. T. Pedersen, J. Bartholdy, M.-S. Seidenkrantz, Communities in Contact: Essays in Archaeology, Ethnohistory and A. Kuijpers (2008), A late Holocene palaeoenvironmental and Ethnography of the Amerindian Circum-Caribbean, edited record from Altona Bay, St Croix, US Virgin Islands, Dan. by C. L. Hofman and A. van Duijvenbode, pp. 393–406, Side- J. Geogr., 108(2), 59–70. stone Press, Leiden, The Netherlands. Kennedy, L. M., K.-B. Liu, D. Kar, T. McCloskey, A. Leblanc, and Cooper, J., and M. C. Peros (2010), The archaeology of climate A. Desjardins (2009), Sedimentary records of hurricane landfalls change in the Caribbean, J. Archaeol. Sci., 37(6), 1226–1232. in coastal lakes of the Dominican Republic, paper presented at Cooper, J., and K. Thomas (2012), Constructing Caribbean chro- the 105th Annual Meeting of the Association of American Geo- nologies: Comparative radiocarbon dating of shell and wood graphers, Las Vegas, 23–27 Mar. artefacts from pre-Columbian sites in Cuba, Archaeometry, Kohler, T. A., and S. E. van der Leeuw (2007), Introduction: Histor- 54(2), 401–425. ical socionatural systems and models, in The Model-Based Archae- Cooper, J., R. Valcárcel Rojas, and J. Calvera Rosés (2010), Recent ology of Socionatural Systems, edited by T. A. Kohler and S. E. archaeological ﬁeldwork from the region around Los Buchil- van der Leeuw, pp. 1–12, Sch. for Adv. Res. Press, Santa Fe, N. M. lones, in Beyond the Blockade: New Currents in Cuban Archae- Malaize, B., P. Bertran, P. Carbonel, D. Bonnissent, K. Charlier, D. ology, edited by S. Kepecs, A. Curet and G. De La Rosa, pp. 89– Galop, D. Imbert, N. Serrand, C. Stouvenot, and C. Pujol (2011), 105, Univ. of Ala. Press, Tuscaloosa. Hurricanes and climate in the Caribbean during the past 3700 de Oviedo y Valdés, G. F. (1851), La Historia General y Natural de years BP, The Holocene, 21(6), 911–924, doi:10.1177/ las Indias: Islas y Tierra-Firme del Mar Océano, Imprenta de la 0959683611400198. Real Acad. de la Hist., Madrid, Spain. Maul, G. A. (1993), Implications of the future climate on the Donnelly, J. P., and J. D. Woodruff (2007), Intense hurricane activ- ecosystems and socio-economic structure in the marine and ity over the past 5000 years controlled by El Nino and the West coastal regions of the Intra-Americas Sea, in Climatic Change in African Monsoon, Nature, 447, 465–468. the Intra-Americas Sea, edited by G. A. Maul, pp. 1–28, Hodder & Elsner, J. B. (2007), Tempests in time, Nature, 447, 647–649. Stoughton, London, U. K. Frappier, A., T. Knutson, K.-B. Liu, and K. Emanuel (2007), Per- McCloskey, T. A., and G. Keller (2009), 5000 year sedimentary spective: Coordinating paleoclimate research on tropical cy- record of hurricane strikes on the central coast of Belize, Quat. clones with hurricane-climate theory and modelling, Tellus, Ser. Int., 195(1–2), 53–68. A, 59(4), 529–537. McFarlane, D. A., J. Lundberg, and A. G. Fincham (2002), A late Gischler, E., E. A. Shinn, W. Oschmann, J. Fiebig, and N. A. Buster Quaternary paleoecological record from caves of southern Jamaica, (2008), A 1500-year Holocene Caribbean climate archive from West Indies, J. Cave Karst Stud., 64(2), 117–125. COOPER 49

Nelson, M. C., et al. (2012), Long-term vulnerability and resilience: Redman, C. L., and A. P. Kinzig (2003), Resilience of past land- Three examples from archaeological study in the southwestern scapes: Resilience theory, society, and the Longue Durèe, Con- US and northern Mexico, in Surviving Sudden Environmental serv. Ecol., 7(1), 14. Change: Answers From Archaeology, edited by J. Cooper and P. Samson, A. V. M. (2010), Renewing the House: Trajectories of Sheets, pp. 197–221, Univ. Press of Colo., Boulder. Social Life in the Yucayeque (Community) of El Cabo, Higüey, Newsom, L. A., and E. S. Wing (2004), On Land and Sea: Native Dominican Republic, AD 800 to 1504, Sidestone Press, Leiden, American Uses of Biological Resources in the West Indies, The The Netherlands. Univ. of Ala. Press, Tuscaloosa. Saunders, M. A., and A. S. Lea (2008), Large contribution of sea Nyberg, J., B. A. Malmgren, A. Winter, M. R. Jury, K. H. Kil- surface warming to recent increase in Atlantic hurricane activity, bourne, and T. M. Quinn (2007), Low Atlantic hurricane activity Nature, 451, 557–560. in the 1970s and 1980s compared to the past 270 years, Nature, Sheets, P., and J. Cooper (2012), Living with the dangers of sudden 447, 698–701. environmental change, in Surviving Sudden Environmental Pages, R. (2005), The Caribbean Sea is very hot, Granma Int., Change: Answers from Archaeology, edited by J. Cooper and P. 18 Aug. [Available at http://www.granma.cu/ingles/2005/agosto/ Sheets, pp. 1–18, Univ. Press of Colo., Boulder. juev18/34temperatura.html.] Valcárcel Rojas, R., J. Cooper, J. Calvera Rosés, O. Brito, and M. Pané, F. R. (1990), Relacion Acerca de las Antiguedades de los Labrada (2006), Postes en el Mar: Excavación de una estructura Indios, Ed. de Cienc. Soc., La Habana, Cuba. constructiva aborigen en Los Buchillones, El Caribe Arqueoló- Pendergast, D., E. Graham, J. Calvera, and J. Jardines (2002), The gico, 9,76–88. houses in which they dwelt: The excavation and dating of Watson, D., and J. Destefano (2010), Flood-resistant design, in Taino wooden structures at Los Buchillones, Cuba, J. Wetland Design for Flooding: Architecture, Landscape, and Urban Archaeol., 2,61–75. Design for Resilience to Climate Change, edited by D. Watson Peros, M. C., E. G. Reinhardt, and A. M. Davis (2007), A 6000 cal yr and M. Adams, pp. 131–150, John Wiley, Hoboken, N. J. record of ecological and hydrological changes from Laguna de la Woodruff, J. D., J. P. Donnelly, K. Emanuel, and P. Lane (2008a), Leche, north coastal Cuba, Quat. Res., 67,69–82. Assessing sedimentary records of paleohurricane activity using Petitjean Roget, H. (2001), Contribution à l’Etude du Troumassoıde modeled hurricane climatology, Geochem. Geophys. Geosyst., 9, et du Saladoide (600–1200 AD). Une hypothèse sur les causes de Q09V10, doi:10.1029/2008GC002043. la regression du Saladoide aux Petites Antilles, in Proceedings of Woodruff, J. D., J. P. Donnelly, D. Mohrig, and W. R. Geyer the 19th International Congress for Caribbean Archaeology, (2008b), Reconstructing relative ﬂooding intensities responsible edited by L. Alofs and R. A. C. F. Dijkhoff, pp. 227–238, Museo for hurricane induced deposits from Laguna Playa Grande, Arqueologico de Aruba, Oranjestad, Aruba. Vieques, Puerto Rico, Geology, 36(5), 391–394. Puig Gonzales, M. A., J. E. Betancourt Lavastida, and R. A. Cedeño (2010), Fortalezas Frente a Huracanes, Ed. Cient.-Téc., La J. Cooper, UCL Institute of Archaeology, 31–34 Gordon Square, Habana, Cuba. London, WC1H 0PY, UK. ([email protected]) Recent Seasonal Variations in Arid Landscape Cover and Aeolian Sand Mobility, Navajo Nation, Southwestern United States

Amy E. Draut

U.S. Geological Survey, Santa Cruz, California, USA

Margaret Hiza Redsteer and Lee Amoroso

U.S. Geological Survey, Flagstaff, Arizona, USA

The socioeconomic impacts of climate change pose problems not only in developing countries but also to residents of arid lands in the United States among marginalized societies with limited economic means. In the Navajo Nation, warming temperatures and recent drought have increased aeolian sediment mobility such that large, migrating sand dunes affect grazing lands, housing, and road access. Dust derived from this region also affects albedo and longevity of the Rocky Mountains snowpack, located downwind. We present initial results from a study that monitors sand transport and vegetation within a 0.2 km2 site in the Navajo lands, measuring the effects of drought on landscape stability since 2009. Sand mobility decreased substantially as 1 year with near-normal monsoon rainfall (2010) somewhat abated a decade-long drought, temporarily doubling vegetation cover. Vegetation that grew during 2010, with adequate rain, died off rapidly during dry conditions in 2011. Short-term increases in rainfall that promote annual, but not perennial, plant growth will not improve landscape stability in the long term. Climate projections suggest that a warmer, drier climate and potentially enhanced sediment supply from ephemeral washes will further increase aeolian sand transport and dune activity, worsening the present challenges to people living in this region. Connections among climate, vegetation, and aeolian sediment erodibility in this region are highly relevant to other areas of the world with similar environmental problems.

1. INTRODUCTION wealthiest nations there are societies whose environmental and socioeconomic resources render them particularly vul- Human communities most vulnerable to the effects of nerable to twenty-ﬁrst century climate change. The Navajo climate change are those with limited natural resources for Nation is the largest tribal reservation within the United subsistence, limited economic means, and rapid population States, and its location is presently restricted to arid and growth. Although such conditions are more commonly asso- semiarid lands within the Navajo ancestral homeland. These ciated with developing countries, even within some of the lands are stressed by drought, land-use practices, and a population that grew rapidly during the twentieth century. Climates, Landscapes, and Civilizations Like many other Native American communities, the Navajo Geophysical Monograph Series 198 people face economic disadvantage and, owing to livestock- This paper is not subject to U.S. copyright. management regulations and cultural ties, cannot readily Published in 2012 by the American Geophysical Union. relocate either their grazing lands or settlements [Redsteer 10.1029/2012GM001214 et al., 2010a]. 51 52 VARIABILITY IN ARID LANDSCAPE COVER AND AEOLIAN SAND MOBILITY

Among the environmental challenges faced by the Navajo With an area of 67,000 km2 spanning parts of Arizona, Nation are shifting vegetation patterns, including the spread New Mexico, and Utah (Figure 1), the Navajo reservation of invasive species, and aeolian sediment mobilization. lands range in elevation from 1200 to 3000 m, with average Warming temperatures and recent drought have contributed annual rainfall varying regionally from 100 to 300 mm. This to reduced streamﬂow and vegetation loss. These factors part of North America has a cultural history spanning have increased aeolian sediment mobility such that regional thousands of years, but aridity and scarcity of arable and dust storms are common and large, and migrating sand dunes grazing land have challenged past and present civilizations affect grazing lands, housing, and transportation. Here we there. Drought and depletion of natural resources are widely present recent sand transport and vegetation measurements believed to have contributed to abandonment of large An- to elucidate and quantify some of the factors affecting land- cestral Puebloan settlements in the twelfth and thirteenth scape stability in the Navajo Nation and discuss implications centuries, in lands that overlap with the modern Navajo of aeolian sediment mobilization in this part of the Colorado Nation; unfavorable hydrologic conditions also impacted the Plateau for albedo on the Rocky Mountains snowpack. large Hohokam civilization of southern Arizona beginning in

Figure 1. (a) Map of the southwestern United States, with shaded area indicating reservation boundaries of the Navajo Nation. The Hopi reservation is surrounded by Navajo lands. Regional wind direction over much of the Colorado Plateau is from southwest toward northeast. (b) Aerial photograph of the study area showing aeolian dune field downwind of ephemeral wash. (c) Sites where vegetation surveys were conducted. Sand transport, wind, and rainfall have been measured at site 7 since 2009. Images courtesy of Google Earth, © Google, Inc., 2007. DRAUT ET AL. 53 the twelfth century and led to their abandoning large settle- seen declining precipitation in the Navajo lands (Figure 2) ments by the fourteenth century [e.g., Euler et al., 1979; and more rapid warming in the southwestern United States Crown and Judge, 1991; Meko et al., 2007]. than elsewhere in North America, with associated ecological As of 2010, more than 169,000 of the 332,000 Navajo changes [Westerling et al., 2006; Seager, 2007; Weiss et al., tribal members lived on the reservation [Norris et al., 2010]. 2009]. Declining precipitation, a shift from snowfall to rain- The population grew substantially in the late twentieth cen- fall, and increased potential evapotranspiration have contrib- tury; in 2000, the median age of the Navajo reservation uted to reduced surface water in the Navajo lands; at least 30 population was 24 years, compared to 35.3 years for the streams and lakes that were perennial in the 1920s are now general U.S. population. Forty-three percent of the Navajo dry or ephemeral [Redsteer et al., 2010a]. Climate models population lived below the poverty line then, compared to project increasingly drier and warmer conditions throughout 12% for the U.S. population as a whole [U.S. Census Bureau, the southwestern United States [Seager et al., 2007; Solo- 2000]. Historically, the Navajo subsistence resources and mon et al., 2007; Dominguez et al., 2010]. economy were tied closely to livestock. Political and eco- Sand dunes have formed and enlarged substantially in the nomic forces encouraged the Navajo to expand sheep and Navajo lands since the mid-twentieth century. Nearly one goat herds in the mid-1800s, but by the 1880s, rangelands third of the reservation area has loose, sandy soil that is easily were being overgrazed to the point where native vegetation entrained by wind and shaped into dunes [Hack, 1941; was diminishing [Bailey and Bailey, 1986]. Around 1900, Redsteer et al., 2010a]. Particularly after a 1950s drought, regional vegetation communities also changed as the Eur- loss of perennial streamflow left riverbeds dry, providing a asian annual plant Russian thistle (Salsola spp.) was intro- source from which the wind mobilizes sediment [Redsteer et duced to the southwestern United States, where it spread al., 2010b]. Aerial photographs show evidence for aeolian quickly on loose, sandy soils. In its dry form known as sand transport downwind of many streambeds in the Navajo tumbleweed, Russian thistle disperses great distances aided Nation, commonly forming well-developed dune fields (Fig- by wind and is one of the most widespread invasive plants in ure 1b), similar to source-bordering dunes found near other the Navajo lands today. Although cattle may eat its shoots desert streambeds globally [Bullard and McTainsh, 2003]. when the plant is young and small, mature Salsola is thorny Comparable sediment mobilization and dune activation has and unpalatable. also occurred in other field settings in response to climatic Precipitation occurs bimodally in this region, during winter change and fluvial sediment supply [Muhs and Holliday, storms (December–March) and the North American mon- 1995; Lancaster, 1997; Clarke and Rendell, 1998; Cornelis, soon season (July–September), with 45% of annual precipi- 2006; Draut, 2012]. Aeolian sediment transport reduces air tation falling during the summer monsoon [Redsteer et al., quality, causing health risks; sand dunes in the Navajo lands, 2010a]. These wetter seasons are separated by a dry, windy migrating tens of meters per year, frequently impede road spring [e.g., Draut and Rubin, 2006; cf. Jewell and Nicoll, access and have destroyed houses (Figure 3) [Redsteer et al., 2011; Munson et al., 2011]. The past several decades have 2010a, 2010b].

Figure 2. Rainfall on the southern Navajo reservation, as percentage of the 1915–2010 average. Plot shows combined data from Winslow and Ganado, the two stations nearest the study site with long-term records. 54 VARIABILITY IN ARID LANDSCAPE COVER AND AEOLIAN SAND MOBILITY

program there in 2005 that deployed new weather stations and recently expanded to measure aeolian sand transport and vegetation. We present some initial results of that study, demonstrating the inﬂuence of seasonal weather patterns on sand mobility and vegetation. Aeolian sediment mobility is a useful indicator of dryland environmental conditions: the air quality and landscape stability issues it represents are vital to evaluating the habitability of the Navajo reservation and longevity of the Rocky Mountains snowpack as twenty-ﬁrst century climate becomes warmer and drier.

2. DATA COLLECTION

Seasonal variations in aeolian sand transport and vegetation were measured in a study area that, although small compared with the vast size of the Navajo Nation, is representative of many settlements. Some 15,000–20,000 Navajo people reside in landscapes with similar ground cover to that of the 0.2 km2 study area: sandy soils near a dune field and an ephemeral wash, in an area with livestock use and occasional off-road vehicle traffic (Figure 1). Wind velocity, rainfall, and aeolian sand transport were measured beginning in March 2009 at the site marked “7” in Figure 1c [Draut et al., 2012]. The weather station and sand traps were surrounded by fenced enclosures to prevent livestock damage, necessary considerations that precluded de- ploying a more extensive array of sand traps. Wind and rain data were recorded on a digital data logger using a 4 min sampling interval. An anemometer measured wind speed with a resolution of 0.2 m sÀ1, at a height of 2 m. A tipping bucket gauge measured precipitation with a resolution of 0.2 mm. Windblown sand was collected in four passive sampling Big Springs Number Eight (BSNE) traps [Fryrear, 1986] mounted on a vertical pole 10 m upwind of the weather station, with sand trap orifices 0.1, 0.4, 0.7, and 1.0 m above the ground. Every 4 weeks, sand was emptied from the traps and weighed. Transport rates were calculated by dividing the total mass of sand in the traps by the number of days over which it accumulated. To avoid introducing uncertainty, measured sand fluxes were not extrapolated down to the bed; thus, although measurements do not ac- Figure 3. Photographs taken on the Navajo reservation in 2008. count for the absolute total mass flux, trends in relative (a) Dunes migrating from left to right across a gravel road (road amounts of sand transport with time still are resolvable. Sand occupies foreground and is oriented toward and away from photog- transport rates were normalized to show sand mobility with- rapher). (b) Haze from airborne dust. Vertical fencepost in the center out the effects of variable wind speed in different time inter- of the photograph is ~1 m tall. (c) Russian thistle dominates the vals; each sand flux measurement was divided by the vegetation in a dune field. cumulative flux predicted for that interval by the Dong et al. [2003] transport equation. This formulation, a modification Because the Navajo lands are fairly remote, quantitative of that by O’Brien and Rindlaub [1936], was chosen because measurements of environmental parameters are scarce. The it treats wind strength as a function of velocity rather than U.S. Geological Survey began a pilot drought-monitoring shear velocity; to extrapolate shear velocity from having DRAUT ET AL. 55 measured wind speed at only one height would introduce area wash: 38 mm of rain (29 mm within 20 min) caused an unwanted uncertainty. estimated 10 m3 sÀ1 flow in the wash, and storm runoff In 2010 and 2011, March and August vegetation cover incised >1 m into nearby gravel roads. was measured at 11 sites (Figure 1c). At each site, five 3 m Vegetation cover in 2010 increased substantially between radius circles were outlined, one at the junction of two surveys in late winter (March) and those during the summer orthogonal 20 m long transects and the other four at the monsoon season (August) (Figure 5a). No similar increase transect ends [cf. Draut and Gillette, 2010]. Within each occurred from winter to summer during the much drier year circle, the proportion of space occupied by vegetation was of 2011; both annual and perennial plant cover decreased measured. Four substrate classes also were measured: sand, between March and August 2011, even though warm tem- rock, biologic soil crust, and leaf litter. Along each transect, peratures and longer daylight hours typically favor plant gap lengths were recorded where the measuring tape growth then (Figure 5a). Changes in cumulative canopy gap crossed bare sand without rocks, biologic soil crust, leaf length (Figure 5b) reflect the increase and subsequent de- litter, or overhanging plant canopy. This method was mod- crease in vegetation abundance during 2010–2011, as sub- ified from the work of Herrick et al. [2005] to measure strate measurements indicated no substantial change in spacing and abundance of elements that affect aeolian sand biologic crust, rock, or accumulated leaf litter that factor into mobility [Buckley, 1987; Belnap and Lange, 2003]. Gap the cumulative gap-length calculation; most study sites con- lengths between plant bases also were measured [Herrick sistently had >95% sand substrate. et al., 2005]. Vegetation growth during the 2010 summer monsoon was accompanied by disproportionately greater increase in an- 3. RESULTS nual plants relative to perennials (Figure 5a). Such growth also was reflected in the decreasing basal gap length be- The greatest aeolian sand transport in the study area tween winter and summer 2010 (Figure 5c): transects inter- occurred from springtime winds, with spring sand transport sect more plant bases after new annual plants have in 2009 and 2010 ten times greater than in other seasons germinated. Basal gap length decreased in 2011 relative to (Figure 4). As important factors in aeolian activity, seasonal 2010 as annual plants died and did not increase during 2011 trends in vegetation cover (Figure 5) are related to precipita- because annuals did not germinate extensively during the tion (Figure 4d). The study interval spanned a range of very dry summer of 2011. Invasive Russian thistle is by far rainfall conditions. During 2009 and 2011, the El Niño– the most abundant annual plant in the study area; in the four Southern Oscillation (ENSO) cycle was in a La Niñaphase; successive surveys, it composed 100%, 70%, 96%, and 93% 2010 was a weak El Niño year. In 2009, Arizona experienced of the annual plant cover. After the 2010 monsoon rains, its fourth driest year in 117 years on record, and the preced- total vegetation cover increased by a factor of 2.3, whereas ing 3 years also had been abnormally dry [National Climatic Russian thistle increased by a factor of nearly 15. The Data Center, 2011]. Wetter conditions returned in 2010, abundance and rapid increase of Russian thistle during which was Arizona’s 85th driest year on record with above- 2010 likely interfered with the capacity of the BSNE traps normal rainfall statewide, although northeastern Arizona to collect aeolian sediment. By winter 2011, Russian thistle received below-average rainfall (Figure 2). Dry conditions had died and produced abundant wind-mobile tumbleweed prevailed in 2011, the 23rd driest on record in Arizona, with that accumulated in thick piles at the upwind side of the the annual rainfall total in mid-August 2011 (the time of the fence enclosing the sand traps, apparently inhibiting sand fourth vegetation survey) being only 20% of the 2010 rainfall movement locally until the tumbleweed was removed dur- for the same time period and the summer monsoon having ing our monthly maintenance visits. These conditions likely brought only 14% of the rain as had fallen by mid-August in affected the relation between measurements of vegetation 2010. cover and sand mobility. Canopy gap length, at first, corre- Sand mobility (normalized sand transport) (Figure 4a) sponded well with measured sand mobility as plant cover varied over more than two orders of magnitude during the grew between March and August 2010 and as several dry study interval and was greatest in winter and spring 2009 and years transitioned into the wetter year of 2010 (Figures 4 2010 when windy weather followed abnormally dry condi- and 5). Even though plant cover in summer 2011 was only tions. Sand mobility decreased to its lowest measured values half of that in summer 2010, measured sand transport did in fall 2010, coincident with vegetation having increased not increase correspondingly. This was likely an artifact of over that summer (Figure 5). The 2010 monsoon season antecedent Russian thistle abundance, such that tumbleweed included a rain event on July 31 (year day 212) (Figure 4d) that grew during the 2010 monsoon rains interfered with that produced the only known recent discharge in the study sand collection in 2011. 56 VARIABILITY IN ARID LANDSCAPE COVER AND AEOLIAN SAND MOBILITY DRAUT ET AL. 57

4. DISCUSSION

The data presented here reflect conditions at the upwind edge of a dune field. In other nearby areas with rapidly migrating sand dunes almost or entirely devoid of vegetation, sand mobility and transport likely are much greater than in our study area. We have not included such terrain in the study because rapid dune migration (e.g., 30 m during spring 2011) makes it impossible to deploy equipment in the field or to reoccupy the same location for vegetation measurements. Instead, we focus on conditions within a landscape where dunes are presently stable enough for people to still live, work, and raise livestock. Relationships discussed here among vegetation cover and aeolian sediment erodibility on this Navajo Nation landscape are highly relevant also to other areas of the world with similar environmental problems [e.g., Kurosaki et al., 2011; Okin et al., 2011]. Relative abundance of annual and perennial plants exerts an important control on aeolian sand mobility and thus on landscape stability [Urban et al., 2009; Munson et al., 2011]. Annual plants can germinate, mature, and disperse seeds with only one season of good rainfall, but to maintain and increase perennial plant cover in the long term would require sufficient moisture in multiple consecutive seasons, not only during the summer monsoon. Perennial plants tend to have stiffer, more durable stems and roots and to accumulate more leaf litter around their bases, protecting the land surface from wind erosion more efficiently than do annual plants [Belnap et al., 2009; Okin et al., 2011]. Therefore, although a year with good monsoon rains and abnormally dry conditions in other seasons would cause short-term plant growth, it would not reduce Figure 5. Vegetation measurements, winter and summer 2010 and aeolian sediment movement in the long term. The dispropor- 2011. (a) Area covered by vegetation (measurements from circular tionate increase of annuals such as Russian thistle from a good plots at all 11 sites combined). (b) Cumulative gap length where summer monsoon is unlikely to increase landscape stability. bare, open sand was present on linear transects (percent of total Because the windy season occurs in early spring before most transect length). Boxes span the interquartile range of data collected annual growth, and because invasive exotic plants crowd out at the 11 study sites; horizontal line through each box is the median value. Circles show outlier points with values more than 1.5 times slower-growing native plants and use water that otherwise the interquartile range, and whiskers show highest and lowest non- could be available to perennials, the proliferation of Russian outlier points. (c) Mean basal gap length (distance between plant thistle may actually decrease landscape stability. Although bases) on linear transects. both annual and perennial plants increased during the wetter

Figure 4. (opposite) Sand transport, wind speed, and rainfall at the study site, 2009–2011. (a) Aeolian sand mobility, represented as dimensionless values obtained by normalizing the sand transport measurements of Figure 4b against cumulative sand flux predicted for each time interval, as a function of wind speed, by the Dong et al. [2003] transport equation. Gray vertical bars indicate times of vegetation surveys. Beginning in winter 2011, sand transport measurements probably were artificially low owing to tumbleweed accumulating at the upwind side of the sand trap enclosure. (b) Direct measurements of aeolian sand transport, in grams per cm width, obtained using sand mass collected in BSNE traps divided by the number of days over which it accumulated. Error bars factor in a 70%–130% efficiency range for BSNE traps [Goossens et al., 2000]. (c) Wind speed, expressed as daytime (06:00–18:00) and nighttime (18:00–06:00) averages of 4 min data. (d) Daily rainfall totals summed from 4 min measurements. 58 VARIABILITY IN ARID LANDSCAPE COVER AND AEOLIAN SAND MOBILITY summer of 2010, decreasing sand mobility then, the subse- conditions prompt a shift from snowfall toward rainfall and quent loss of even perennial plant cover in a dry year such as also increase aeolian sediment transport from the Colorado 2011 (Figure 5a) poses a substantial risk to landscape stability Plateau. This is a particular concern during the La Niña in a setting prone to wind erosion. phases of the ENSO cycle, when northeastern Arizona tends Climate projections indicate that the southwestern United to have higher wind velocities and less rainfall than during States will become increasingly warm and dry during the the El Niño phases [Enloe et al., 2004; Phillips and twenty-first century [e.g., Seager et al., 2007]. Models also Doesken, 2011]. suggest that intense storms and associated flooding may Previous studies indicate a link between strong winds in increase [Trenberth, 1998]. In such a scenario, perennial northeastern Arizona and dust events in the southern streamflow would continue to decline in the Navajo lands Rockies, with a daily average wind velocity above 6.7 m sÀ1 [Redsteer et al., 2010a], whereas stronger, more frequent having been proposed as a threshold value for dust events episodic floods are possible. Because dry stream beds are [Phillips and Doesken, 2011]. High springtime sand trans- sources of aeolian sediment, flash floods during storms could port measured at our study site in 2009 and 2010 is broadly supply additional material that would be entrained by wind consistent with dust-on-snow monitored in the San Juan the following spring [cf. Muhs and Holliday,1995;Lan- range of the southern Rocky Mountains; the dust-on-snow caster,1997;Han et al., 2007]. Such altered patterns of mass there was greater in spring 2009 than in any of the rainfall and streamflow could further increase sediment sup- previous 7 years [Skiles et al., 2011]. Making more specific ply into the aeolian dunes downwind of dry streambeds that connections between trends in aeolian sand mobility in our have become common in the Navajo Nation since the 1950s study area and dust deposition in the southern Rocky Moun- [Redsteer et al., 2010b]. Increased aeolian sand transport in tains is challenging because, over the 350 km between the springtime also could destabilize vegetation that initially study area and the nearest snowpack downwind where dust thrived under the previous fall monsoon rain because wind- events are recorded, the source area is so large that even with blown sand damages plants by abrasion and burial [Okin et a substantial reduction in sediment mobility, as apparently al., 2006]. We propose that a negative feedback cycle could occurred in late 2010, the wind still supplies ample material develop whereby the summer-fall monsoon rain that pro- to the San Juan range. A longer record from the study area motes plant growth also supplies ephemeral washes with new and other locations may clarify connections between source- sand that damages those plants once it is mobilized by wind. area aeolian events and dust deposition on mountain snow- Increased aeolian sand transport and dune activity in such pack downwind. a future climate regime would further compromise living conditions for the Navajo and other cultures in the south- 5. CONCLUSIONS western United States. Because many Navajo residents not only have strong traditional ties to these lands but also do not Problems faced by populations living in marginal land- have the economic means to relocate elsewhere, a future scapes against a backdrop of climate change are not restricted combination of climate change, greater sediment mobility, to developing countries or to prehistoric cultures in the desert and growth of invasive plants that neither stabilize dunes nor southwest of North America; they also occur today in this serve a purpose for people or livestock would negatively same region among marginalized societies with limited eco- impact residents over the long term. nomic means. In the Navajo Nation, decreasing perennial Aeolian sediment mobility in northeastern Arizona has streamflow and warming, drying trends lead to aeolian sed- consequences not only for human health and infrastructure iment mobilization that affects air quality, housing, and but also for the snowpack in the Rocky Mountains of Color- transportation and potentially reduces albedo of the Rocky ado, which supplies water to millions of people. Satellite Mountains snowpack. Initial results of an intended long-term imagery has shown aeolian dust plumes originating repeat- monitoring program in a 0.2 km2 site show that sand mobil- edly from the Navajo lands in and near the study area during ity decreased substantially as 1 year with near-normal the time interval covered by this study [U.S. Geological monsoon rainfall (2010) somewhat abated a decade-long Survey, 2012]. Wind-borne dust from the southwestern Col- drought, temporarily doubling vegetation cover. Vegetation orado Plateau settles on the snow in the southern Rockies, that grew during 2010, with adequate rain, died off rapidly reducing its albedo and leading to earlier spring melting during subsequent dry conditions in 2011. Short-term in- [Painter et al., 2007, 2010; Center for Snow and Avalanche creases in rainfall that promote annual but not long-term Studies, 2011; Phillips and Doesken, 2011]. Accelerated perennial plant growth will not improve landscape stability snowmelt poses a concern for resource managers who antic- because annual plants are less efficient than perennials at ipate that demand for water will increase as warmer, drier stabilizing sand against wind erosion. Climate projections DRAUT ET AL. 59 suggest that warmer, drier conditions and potentially en- Canyon, Utah, U.S. Geol. Surv. Open File Rep., 2010-1273,61 hanced sediment supply from flash floods in ephemeral pp. [Available at http://pubs.usgs.gov/of/2010/1273/.] washes will combine to increase sediment transport and dune Draut, A. E., and D. M. Rubin (2006), Measurements of wind, activity, worsening the present challenges to people living in aeolian sand transport, and precipitation in the Colorado River — this region. corridor, Grand Canyon, Arizona January 2005 to January 2006, U.S. Geol. Surv. Open File Rep., 2006-1188, 88 pp. [Avail- Acknowledgments. This study was supported by the U.S. Geo- able at http://pubs.usgs.gov/of/2006/1188/.] logical Survey. We thank the Navajo Nation, and the Biggambler Draut, A. E., M. H. Redsteer, and L. Amoroso (2012), Vegetation, family, in particular, for allowing this study to occur in their grazing substrate, and aeolian sediment transport at Teesto Wash, lands. R.L. Reynolds, editor K. Nichols, and three anonymous Navajo Nation, 2009 to 2012, U.S. Geol. Surv. Sci. Invest. Rep., reviewers provided constructive review comments that improved 2012-5095, in press. the manuscript. Enloe, J., J. J. O’Brien, and S. R. Smith (2004), ENSO impacts on peak wind gusts in the United States, J. Clim., 17, 1728–1737. REFERENCES Euler, R. C., G. J. Gumerman, T. N. V. Karlstrom, J. S. Dean, and R. H. Hevly (1979), The Colorado plateaus: Cultural dynamics Bailey, G., and R. G. Bailey (1986), A History of the Navajos: The and paleoenvironment, Science, 205, 1089–1101. Reservation Years, 2nd ed., 360 pp., Sch. of Am. Res. Press, Fryrear, D. W. (1986), A field dust sampler, J. Soil Water Conserv., Santa Fe, N. M. 41,117–119. Belnap, J., and O. L. Lange (Eds.) (2003), Biological Soil Crusts: Goossens, D., Z. Offer, and G. London (2000), Wind tunnel and Structure, Function, and Management, Ecol. Stud. Ser., vol. 150, field calibration of five aeolian sand traps, Geomorphology, 35, Springer, Berlin. 233–252. Belnap, J., R. L. Reynolds, M. C. Reheis, S. L. Phillips, F. E. Hack, J. T. (1941), Dunes of the western Navajo country, Geogr. Urban, and H. L. Goldstein (2009), Sediment losses and gains Rev., 31, 240–263. across a gradient of livestock grazing and plant invasion in a Han, G., G. Zhang, and Y. Dong (2007), A model for the active cool, semi-arid grassland, Colorado Plateau, USA, Aeolian Res., origin and development of source-bordering dunefields on a 1,27–43. semiarid fluvial plan: A case study from the Xiliaohe Plain, Buckley, R. (1987), The effect of sparse vegetation on the transport Northeast China, Geomorphology, 86, 512–524. of dune sand by wind, Nature, 325, 426–428. Herrick, J. E., J. W. Van Zee, K. M. Havstad, L. M. Burkett, and Bullard, J. E., and G. H. McTainsh (2003), Aeolian-fluvial interac- W. G. Whitford (2005), Monitoring Manual for Grassland, tions in dryland environments: Examples, concepts, and Austra- Shrubland, and Savanna Ecosystems, 36 pp., Univ. of Ariz. lia case study, Prog. Phys. Geogr., 27, 471–501. Press, Tucson. Center for Snow and Avalanche Studies (2011), Colorado Dust-on- Jewell, P. W., and K. Nicoll (2011), Wind regimes and aeolian Snow Program (CDOS), Silverton, Colo. [Available at http:// transport in the Great Basin, U.S.A., Geomorphology, 129, www.snowstudies.org/codos1.html.] 1–13. Clarke, M. L., and H. M. Rendell (1998), Climate change impacts Kurosaki, Y., M. Shinoda, and M. Mikami (2011), What caused a on sand supply and the formation of desert sand dunes in the recent increase in dust outbreaks over East Asia?, Geophys. Res. south-west USA, J. Arid Environ., 39, 517–531. Lett., 38, L11702, doi:10.1029/2011GL047494. Cornelis, W. M. (2006), Hydroclimatology of wind erosion in arid Lancaster, N. (1997), Response of eolian geomorphic systems to and semi-arid environments, in Dryland Ecohydrology, edited by minor climate change—Examples from the southern California P. D’Odorico and A. Porporato, pp. 141–159, Springer, Dor- deserts, Geomorphology, 19, 333–347. drecht, The Netherlands. Meko, D. M., C. A. Woodhouse, C. A. Baisan, T. Knight, J. J. Crown, P. L., and W. J. Judge (1991), Chaco and Hohokam: Lukas, M. K. Hughes, and M. W. Salzer (2007), Medieval Prehistoric Regional Systems in the American Southwest, 384 pp., drought in the upper Colorado River Basin, Geophys. Res. Lett., Sch. of Am. Res. Press, Santa Fe, N. M. 34, L10705, doi:10.1029/2007GL029988. Dominguez, F., J. Cañon, and J. Valdes (2010), IPCC-AR4 climate Muhs, D. R., and V. T. Holliday (1995), Evidence of active dune simulations for the southwestern U.S.: The importance of future sand on the Great Plains in the 19th century from accounts of ENSO projections, Clim. Change, 99, 499–514. early explorers, Quat. Res., 43, 198–208. Dong, Z., X. Liu, H. Wang, and X. Wang (2003), Aeolian sand Munson, S. M., J. Belnap, and G. S. Okin (2011) Responses transport: A wind tunnel model, Sediment. Geol., 161,71–83. of wind erosion to climate-induced vegetation changes on Draut, A. E. (2012), Effects of river regulation on aeolian land- the Colorado Plateau, Proc. Natl. Acad. Sci. U. S. A., 108, scapes, Colorado River, southwestern USA, J. Geophys. Res., 3854–3859. 117, F02022, doi:10.1029/2011JF002329. National Climatic Data Center (2011), State of the Climate National Draut, A. E., and E. R. Gillette (2010), Vegetation and substrate on Overview, http://www.ncdc.noaa.gov, Natl. Oceanic and Atmos. aeolian landscapes in the Colorado River corridor, Cataract Admin., Washington, D. C. 60 VARIABILITY IN ARID LANDSCAPE COVER AND AEOLIAN SAND MOBILITY

Norris, T., P. L. Vines, and E. M. Hoeffel (2010), The American Seager, R., et al. (2007), Model projections of an imminent transi- Indian and Alaska Native population: 2010, 21 pp., U.S. Census tiontoamorearidclimateinsouthwesternNorthAmerica, Bureau, Suitland, Md. [Available at http://www.census.gov.] Science, 316, 1181–1184. O’Brien, M. P., and B. D. Rindlaub (1936), The transport of sand by Skiles, M., T. H. Painter, J. S. Deems, and A. P. Barrett (2011), wind, Civ. Eng., 6, 325–327. Interannual variability in dust deposition, radiative forcing, and Okin, G. S., D. A. Gillette, and J. E. Herrick (2006), Multi-scale snowmelt rates in the Colorado River basin, Abstract C41F-05 pre- controls on and consequences of aeolian processes in landscape sented at 2011 Fall Meeting, AGU, San Francisco, Calif., 5–9Dec. change in arid and semi-arid environments, J. Arid Environ., 65, Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. 253–275. Averyt, M. Tignor, and H. L. Miller (Eds.) (2007), Climate Okin, G. S., J. E. Bullard, R. L. Reynolds, J.-A. C. Ballantine, K. Change 2007: The Physical Science Basis, Contribution of Work- Schepanski, M. C. Todd, J. Belnap, M. C. Baddock, T. E. Gill, ing Group I to the Fourth Assessment Report of the Intergovern- and M. E. Miller (2011), Dust: Small-scale processes with global mental Report on Climate Change, 996 pp., Cambridge Univ. consequences, Eos Trans. AGU, 92(29), 241, doi:10.1029/2011E Press, Cambridge, U. K. O290001. Trenberth, K. (1998), Atmospheric moisture residence times and Painter, T. H., A. P. Barrett, C. C. Landry, J. C. Neff, M. P. cycling: Implications for rainfall rates and climate change, Clim. Cassidy, C. R. Lawrence, K. E. McBride, and G. L. Farmer Change, 39, 667–694. (2007), Impact of disturbed desert soils on duration of moun- Urban, F. E., R. L. Reynolds, and R. Fulton (2009), The dynamic tain snow cover, Geophys. Res. Lett., 34, L12502, doi:10.1029/ interaction of climate, vegetation, and dust emission, Mojave 2007GL030284. Desert, USA, in Arid Environments and Wind Erosion, edited by Painter, T. H., J. S. Deems, J. Belnap, A. F. Hamlet, C. C. Landry, A. Fernandez-Bernal and M. A. De la Rosa, pp. 243–267, Nova and B. Udall (2010), Response of Colorado River runoff to dust Sci., Hauppauge, N. Y. radiative forcing in snow, Proc. Natl. Acad. Sci. U. S. A., 107, U. S. Census Bureau (2000), Table DP-1—Profile of general 17,125–17,130. demographic characteristics, Navajo Nation reservation and Phillips, M., and N. Doesken (2011), Continental wind patterns off-reservation trust land, AZ-NM-UT, Suitland, Md. [Available associated with Colorado alpine dust deposition—An application at http://www.census.gov.] of the BLM/USFS RAWS network, J. Serv. Climatol., 5,1–11. U. S. Geological Survey (2012), Dust monitoring and dust events, Redsteer, M. H., K. B. Kelley, H. Francis, and D. Block (2010a), Southwest Geographic Science Team, Reston, Va. [Available at Disaster risk assessment case study: Recent drought on the Na- http://sgst.wr.usgs.gov/dust_monitoring/dust-events/.] vajo Nation, southwestern United States, Contributing case study Weiss, J. L., C. L. Castro, and J. T. Overpeck (2009), Distinguishing to the 2011 United Nations Global Assessment Report on Disas- pronounced droughts in the southwestern United States: Season- ter Risk Reduction, annexes and papers, 16 pp., U. N. Int. ality and effects of warmer temperatures, J. Clim., 22, 5918– Strategy for Disaster Reduct., Geneva, Switzerland. [Available 5932. at http://www.preventionweb.net/english/hyogo/gar/2011/en/ Westerling, A. L., H. G. Hidalgo, D. R. Cayan, and T. W. Swetnam home/annexes.html.] (2006), Warming and earlier spring increase western U.S. forest Redsteer, M. H., R. Bogle, J. Vogel, D. Block, and B. Middleton wildfire activity, Science, 313, 940–943. (2010b), The history and growth of a recent dune field at Grand Falls, Navajo Nation, NE Arizona, Geol. Soc. Am. Abstr. Programs, 42(5), 416. L. Amoroso and M. Hiza Redsteer, U.S. Geological Survey, Seager, R. (2007), The turn of the century North American 2255 North Gemini Drive, Flagstaff, AZ 86001, USA. drought: Global context, dynamics, and past analogs, J. Clim., A. E. Draut, U.S. Geological Survey, 400 Natural Bridges Drive, 20, 5527–5552. Santa Cruz, CA 95060, USA. ([email protected]) Utilization of Distal Tephra Records for Understanding Climatic and Environmental Consequences of the Youngest Toba Tuff

Emma Gatti and Clive Oppenheimer

Department of Geography, University of Cambridge, Cambridge, UK

The most recent supereruption of the Toba caldera, Sumatra, approximately 74 ka, is the largest known of the Quaternary. It has been implicated in global and regional climate deterioration with widespread ecological effects, including dramatic reduction of genetic diversity in Homo sapiens. Since the first major studies of Youngest Toba Tuff (YTT) in the 1980s, several new ash deposits have been discovered in South Asia. In the light of these new findings, it is timely to review the available data in order to reexamine the significance and impact of the Toba supereruption. This paper examines the particle sizes and thicknesses of the YTT tephra fall deposits and correlates these with distance from the vent. We show that there are no correlations between the distance and finest ash component, suggesting that the ash distribution does not follow a traditional exponential decay. We compare two techniques to calculate the ash volume, obtaining estimates between 770 and 2000 km3 (dense rock equivalent). Although these parameters are keys to understanding the climatic and environmental impact of the eruption, there remain uncertainties in other critical factors such as the season during which the eruption took place and the local short-term impacts of the ash fallout.

1. INTRODUCTION acteristics of tephra deposits, and their implications for global climatic and environmental impact are still not fully under- Tephra fallout from the ~74 ka Youngest Toba Tuff (YTT) stood [Oppenheimer, 2002; Herzog and Graf, 2010]. eruption of Toba, Sumatra, produced regional-scale ash de- Characteristics of tephra deposits can provide insights into posits that extend up to 4200 km from the volcano (Figure 1). key physical parameters of eruptions [Sarna-Wojcicki and Williams et al. [2009] suggested that the effect of the ash Davis, 1991], some of the most important being their forma- deposits on the local environment following the YTT eruption processes, duration, the time that ash remains in the tion lasted on the scale of a few centuries. A similar conclu- atmosphere, and the total volume of ash ejected. Speciﬁcally, sion was reached by Zielinski et al. [1996a] who suggested particle size distributions and thickness are indicators of the that a decennial permanence of sulfur aerosol in the strato- intensity of the eruption and its magnitude, parameters that sphere led to centennial-scale climatic anomalies that gener- are routinely entered into climate models and are used in ated environmental disruption for ~200 years. However, the paleoenvironmental research in order to investigate global exact amount of sulfur ejected by Toba, the dispersal char- and local posteruption impacts. Understanding what ash deposits can tell us about the physical parameters of eruptions is therefore crucial to draw meaningful conclusions on Climates, Landscapes, and Civilizations the resulting environmental impacts. Geophysical Monograph Series 198 The purpose of this chapter is to analyze the most © 2012. American Geophysical Union. All Rights Reserved. recent discoveries of YTT tephra, in order to inform our 10.1029/2012GM001216 present understanding of the environmental impacts of the 63 64 UTILIZATION OF DISTAL TEPHRA RECORDS

Figure 1. Youngest Toba Tuff (YTT) tephra distribution in Asia and YTT sulfur signal in Greenland. Sites with insufficient information are not presented. supereruption. Specifically, we summarize particle size and Quaternary [Chesner and Rose, 1991; Chesner, 1998, 2012] thickness data from the literature and newly discovered sites (Figure 1). The eruption ejected an estimated ~2800 km3 in a unique database, and we use this updated database to dense rock equivalent of rhyolitic magma, equivalent to a adjust the areal and volume calculation, comparing our re- magnitude M8.8 eruption [Rose and Chesner, 1990; Chesner sults with previous studies. and Rose, 1991; Chesner et al., 1991; Chesner, 2012]. This is equivalent to ejecting the volume of a solid cube of rock 2. THE TOBA SUPERERUPTION measuring 14 km in each dimension. The Toba caldera is 100 km Â 30 km and covers an area of ~2270 km2. The “supereruption” of Toba in northern Sumatra, ~74 ka, YTT fall deposits have been reported spanning an area referred to as the YTT, is the largest identified eruption of the exceeding 7 Â 106 km2 in southern and southeastern Asia GATTI AND OPPENHEIMER 65

(Figure 1), the South China Sea, and the central Indian Ocean between the two Dansgaard-Oeschger events [Zielinski et al., basin [Rose and Chesner, 1987]. Terrestrial tephra fall de- 1996a]. posits from the YTT eruption have been identiﬁed in several As a consequence of this proposed climatic shock, some sites on the Indian Peninsula [Korisettar et al., 1988; Achar- researchers theorized that ecosystems and hominid populations yya and Basu, 1993; Kale et al., 1993; Shane et al., 1995; over a wide region of South Asia would have been devastated. Westgate et al., 1998], Malaysia [Ninkovich et al., 1978b; The climatic simulation by Robock et al. [2009] suggested that Rose and Chesner, 1990; Shane et al., 1995], and possibly 4 years after the 300 times Pinatubo eruption, cooling would Bangladesh [Acharyya and Basu, 1993]. YTT deposits are have led to the disappearance of both broadleaf evergreen and also preserved in the Indian Ocean, Bengal Fan, and South tropical deciduous trees. Earlier, Rampino and Ambrose China Sea marine cores [Ninkovich,1979;Pattan et al., [2000] suggested that there would have been reduced photo- 1999; Buhring et al., 2000; Gasparotto et al., 2000; Liang synthesis and limited ocean surface productivity, due to tephra et al., 2001; Schulz et al., 2002; Liu et al., 2006]. sedimentation. They further argued that the consequent millennium of cold climate generated by 6 years of “volcanic 2.1. Climatic Impact winter” may have induced a severe population bottleneck, with consequent strong selective pressures. In particular, Since volcanic sulfate aerosols are the principal cause of Ambrose [1998] argued that Neanderthals might have fared volcanic forcing of climate, several studies have attempted better in these climatic conditions due to their raised cold to estimate the amount of sulfur emitted by Toba. Studies tolerance. Others have argued that such catastrophic human of the chemistry of minerals and glasses in pumices from scenarios of the YTT eruption remain unsubstantiated and the YTT suggested the eruption ejected a minimum of improbable [Oppenheimer, 2002] and that such dramatic Â 15 Â ~3.5 10 gofH2S[Rose and Chesner, 1990] and 1 environmental impacts are not supported by faunal evidence 16 10 gofH2SO4 aerosol into the stratosphere [Chesner et [Gathorne-Hardy and Harcourt-Smith, 2003]. al., 1991]. A model by Robock et al. [2009] assumed that Toba ejected at least 300 times the sulfur emission of the 2.2. Environmental Impact Â 15 eruption of Mount Pinatubo (i.e., ~6 10 gofSO2). However, recent melt inclusion analyses suggested about Analysis of pollen in a marine core from the Bay of Bengal 14 10 gofH2SO4 aerosols, 2 orders of magnitude less than [Williams et al., 2009] shows a distinct change from tree- and previous petrologic estimates [Scaillet et al., 1998; Chesner shrub-dominant vegetation to more open vegetation, sug- and Luhr, 2010]. gesting a millennium of dry conditions. Similarly, oscilla- The postulated exceptional mass of sulfur released tions in pollen counts in a core from northwest Sumatra [van prompted several hypotheses about the effect of the super- der Kaars et al., 2012] provide evidence for a devastating eruption on climate. It has been suggested that the YTT effect on pine forests after the eruption. eruption was responsible for the extended cooling period and Conversely, paleoclimatic marine proxies retrieved from a ice sheet advance immediately succeeding it. Rampino and piston core in the Arabian Sea [Schulz et al., 2002; Von Rad Self [1992, 1993] proposed that the YTT eruption led to a et al., 2002], an area particularly sensitive to monsoon inten- “volcanic winter,” an immediate surface cooling effect of up sity alterations, show no particular impact of the YTT erup- to 5°C caused by high atmospheric opacity due to the volca- tion, and the authors concluded that the eruption had only a nic dust, similar to the one proposed for postnuclear war minor impact on low-latitude monsoonal climate on the scenarios. This global cooling effect, they argued, would centennial to millennial scale. have lasted for several years, generating up to 12°C of cooling in summer temperatures and a millennium of cold climate 3. METHODOLOGY [Rampino and Self, 1993]. However, later measurements of sulfate in the Greenland Ice Sheet Project 2 (GISP2) ice core The distal ash layer from the ~74 ka Toba supereruption, 4À [Zielinski et al., 1996a] show a distinctive SO2 spike at presumed to be of coignimbrite cloud origin, is one of the ~71 ± 5 ka, a period of cooling transition, but major ions and most extensively sampled deposits from a prehistoric erup- oxygen isotopes indicate that the extended ~1000 years of tion. In this study, we have summarized and compared par- stadial cooling between the interstadial 20 (~74.5 ka) and ticle sizes and thickness identiﬁed in recently surveyed YTT interstadial 19 (~69 ka) was already underway before the sites with previously published information on the distal YTT. These data suggest that the eruption did not initiate a deposits (Table 1). In the case of the sites reported by the major glacial period but suggest instead that the eruption may authors, particle size analyses were measured using a Mal- have accelerated stadial conditions during the millennium vern Instruments Mastersizer 2000. 66 UTILIZATION OF DISTAL TEPHRA RECORDS

Table 1. Marine and Terrestrial YTT Cores Reporting Primary Ash Thickness and Particle Size Distributionsa Coordinates Primary Distance Core Thickness Average From Core ID NEType (cm) PS (µm) Location Toba (km) Reference GISP2 76.6 38.5 ice Greenland 11,000 Zielinski et al. [1996b] SO130-289 KL 23.096 66.483 marine 0.00 70 northern Arabian Sea 4,500 Schulz et al. [2002], Von Rad et al. [2002] SO93-47KL 11.831 88.851 marine 0.10 31–88 central Bay of Bengal 1,524 Gasparotto et al. [2000] SO93-115KL 17.716 89.493 marine 0.06 44–62.5 northern Bay of Bengal 2,055 Gasparotto et al. [2000] SO93-124KL 19.814 90.000 marine 0.02 11–25 Bengal upper fan 2,213 Gasparotto et al. [2000] SO93-28KL 4.701 84.967 marine 0.12 22–63 southern Bengal Fan 1,573 Gasparotto et al. [2000] ODP758 5.384 90.361 marine 0.34 50–150 Ninetyeast Ridge 1,064 Dehn et al. [1991] (layer A) MD01-2393 10.502 110.061 marine 0.04 20–500 South China Sea 1,526 Liu et al. [2006] 17962-4 7.181 112.081 marine 0.04 100 South China Sea 1,623 Buhring et al. [2000] RC14-37 1.643 89.927 marine 0.15 125–176/62– Ninetyeast Ridge 953 Ninkovich et al. 88 [1978a, 1978b], Ninkovich [1979] JWP3 15.320 78.134 land 0.04 70 south India 2,600 this work Morgaon 18.305 74.330 land 0.12 64.50 western central India 3,262 this work Perak 4.801 101.162 land no primary 109/41.96 North Peninsular 380 this work ash available Malaysia Bori 19.626 74.633 land no primary 51 western central India 3,300 this work ash available Tejpur 21.892 73.487 land 0.10 62 western India 3,563 this work/Raj [2007] aParticle sizes divided by a long dash indicate ranges reported by the authors. Values separated by a slash indicate two layers of ash within the core.

Thicknesses at each location were retrieved from the orig- The second technique applies a Voronoi tessellation, a inal publications, field work by the authors, and personal well-known method of spatial analysis that can be defined communications with colleagues. Data were excluded from as the partitioning of the plane such that, for any set of total volume calculations if the reported thickness did not distinct data points, the cell associated with a particular data differentiate between primary and reworked units (in the case point contains all spatial locations closest to that point of terrestrial deposits) or undisturbed and bioturbated layers [Bonadonna and Houghton, 2005]. The tephra thickness (in the case of marine deposits). data are transformed in Voronoi cells with an algorithm Several techniques have been utilized to determine the total based on Delaunay triangulation. The total thickness vol- volume of tephra deposits. We present here two volume- ume is calculated by summing the average thickness for the calculation techniques. The first technique is based on a weighted area of each Voronoi cell. We calculated the total proposed exponential dependence between thickness and dis- volume for two extents: a smaller area, delimited by the tance from the vent [Pyle, 1989; Pyle et al., 2006]. This maximum extent of the available field data, and a larger approach considers tephra distribution as an exponentially area, determined by statistical calculation. thinning sheet characterized by two parameters: T0, the ash thickness at the origin, and bt, the thickness half-distance (the 4. RESULTS linear distance over which the isopach thickness falls by half). Isopach areas (areas with the same thickness) are ex- 4.1. Size Characteristics of Distal Fallout trapolated and plotted against the known thicknesses in a ln (thickness) À (isopach area)1/2 plot. This allows to extrapolate Particle size from marine and terrestrial cores (Table 1) empirically the thickness T0 (maximum thickness at area = 0). indicate that the YTT tephra is a medium silt, with a unim- Â The rate of thinning is bt = ln2/k √π [Pyle, 1989], where k is odal distribution, all particles at distal localities in the slope on the ln(thickness) À (isopach area)1/2 plot. The the selected cores being <500 μm in diameter, including total volume of an exponentially thinning tephra sheet is then significant fractions of fine (<63 μm) and very fine (<32 μm) 2 calculated with the formula V = 13.08 T0bt . ash. GATTI AND OPPENHEIMER 67

Vertical size grading has been noted in the proximal tephra diameter correlation, with a regression coefficient (R)= in marine cores. Within 2000 km in the main dispersal À0.9, and exponential decay (R2 = 0.87), thus suggesting direction (west over the Bay of Bengal), the tephra samples that the coarsest particles decay with distance [Pyle, 1989]. show a coarser tephra layer at the base and a finer layer at the By contrast, the finest particles (Figure 2b) show weak neg- top. This double layer has been explained by Ninkovitch et ative correlation (R = À0.2) and weak exponential decay al. [1978b] as an indicator of two major phases in the (R2 = 0.2). This confirms that the size of the finest particles eruption, a first Plinian and a second coignimbrite. The is constant over distances >390 km [Ledbetter and Sparks, presence of the lower coarser unit appears to be dependent 1979; Rose and Chesner, 1987]. on both the distance from the vent and the main dispersal direction; distal cores >2000 km do not preserve the two 4.2. Thickness Variations phases nor do deposits closer to the source but more easterly (such as in the South China Sea). The coarser component The total coignimbrite ash volume erupted by YTT was (Figure 2a) is characterized by negative distance/particle estimated to be ~800 km3 dense rock equivalent (DRE) by

Figure 2. Plots of the (a) coarsest and (b) ﬁnest particle size components versus distance. 68 UTILIZATION OF DISTAL TEPHRA RECORDS

Table 2. Comparison of Different Volume Calculation Techniquesa Minimum Maximum Volume Volume Method (km3) (DREb) (km3) (DREb) Exponential decay 1200 - Voronoi 770 2000 Atmospheric model 1500 1900 [Matthews et al., 2012] Integrated volume-estimate routine 800 - based upon area/thickness relations [Rose and Chesner, 1987] aMatthews et al. [2012] calculated the total volume based on an atmospheric diffusion model. bDense rock equivalent.

Rose and Chesner [1987], determined by an integrated volume-estimate routine based upon area/thickness relations. They approximated (from YTT deposits in the eastern Indian Ocean) an average YTT tephra thickness of 0.1 m covering 7 Â 106 km2. Figure 3 shows a statistical approximation of areal thickness using the updated thickness database and variability of the ash layer, calculated using thickness exponential decay (A) and Voronoi tessellation (B). The total area covered by the known data reporting YTT is 13 Â 106 km2, while the total statistical area calculated by the techniques is 24 Â 106 km2. The exponential decay volume calculation (approximated using spline surface interpolation) showed a statistically signiﬁcant interpolation only within the surface covered by the ﬁeld data. Figure 3a shows that the isopachs do not resemble a classic elliptical pattern, suggesting that several wind systems affected the distribution of the tephra. The calculation of the volume through the formula V = 2 13.08 T0bt results in a total DRE minimum volume of ~1200 km3 (Table 2). This is similar to the calculation of Matthews et al. [2012], who obtained a total volume of 1500 km3 (Figure 3c). The Voronoi tessellation technique (Figure 3b) indicates a minimum ash volume of 770 km3 DRE (corrected to dense rock equivalent from bulk volume Â 0.5, as suggested by Rose and Chesner [1987] and a maximum ash volume of ~2000 km3 DRE). This estimate is consistent with Rose and

Figure 3. Thickness distribution map for YTT obtained using (a) exponential decay (expressed in cm intervals) and (b) Voronoi tessellation. Average thickness values were calculated on the basis of the single thickness value, weighted by the area of each Voronoi cell. (c) For comparison, the thickness results obtained by atmospheric modeling by Matthews et al. [2012]. GATTI AND OPPENHEIMER 69

Chesner’s original estimate of 800 km3 minimum volume around ~900 km [Bonadonna and Phillips, 2003; Herzog and is similar to the dispersal simulation proposed by Mat- and Graf, 2010]. It appears at the moment that no simula- thews et al. [2012] (Table 2). tions are able to realistically model the tephra distribution of a supereruption such as Toba. 5. DISCUSSION The choice of the Voronoi tessellation appears more suitable to calculate the total ash volume. The values (minimum 3 3 5.1. Volume of the YTT Coignimbrite Ashfall 770 km and maximum 2000 km ) closely resemble the value of 800 km3 suggested by Rose and Chesner [1987] Matthews et al. [2012] used particle size distributions from and the most recent model by Matthews et al. [2012]. Nev- Malaysia and India to calibrate a YTT tephra dispersal model. ertheless, it is important to highlight that the techniques Their results suggested that at least 60% of the mass fell as presented here are not models of tephra dispersal, but instead particles <32 µm. This indicates that the finest component is the statistical calculations of total volume of ash ejected and that principal part of the YTT distal and superdistal deposits. The the results should be interpreted considering some limits of particle size distribution in Figure 2 shows no apparent corre- the Voronoi technique, related to scarcity of data points and lation between the finest particle size and the distance traveled uneven data point distribution. The terrestrial thicknesses by the ash. Similarly, Figure 4 shows a weak exponential used for the calculation are less reliable in quantifying the correlation between the ash thickness measured in the field and original thickness of the ash over the continents, since those the distance from the vent (R2 = 0.5). This suggests that any sites are subjected to heavy post depositional reworking. volume calculation based on an exponential decay of the Toba Malaysian sites have an average thickness between 0.18 and ash sheet (Figure 3a) is unrealistic. 0.25 m thick, but this is potentially skewed by the Serdang Figure 2 clearly shows the lack of data within ~350 km of ash [Stauffer et al., 1980], near Kuala Lumpur (Malaysia), the vent (the ash is not found in Sumatra, and there are no reported to be ~0.85 m thick. Similarly, the absence of data available in the Strait of Malacca) and indicates that the thickness data between Malaysia and the South China Sea ash sheet extends as far as 4000 km from the source. This is influences the result. important when considering the validity of 2- and 3-D simu- Additionally, the Voronoi tessellation can be executed in a lations of tephra distribution. Several simulations modeling geographic information system only within a projected Plinian and coignimbrite eruptions assume T0 is within coordinate system, thus introducing an area error when trans- ~70 km of the vent and set a maximum “distal” distance forming the sphericity of the Earth to a flat plane. The error

Figure 4. Plot of ash thickness against distance from the vent. Reworked thicknesses and bioturbated layers (when indicated) were not included in the volume calculation. 70 UTILIZATION OF DISTAL TEPHRA RECORDS increases with the increase in the area, and this could account grains were found to substantiate the correlation. The anom- 2À for overestimation of the maximum volume. Finally, the alous SO4 values recorded in the high GISP2 ice layers minimum volume obtained with the Voronoi tessellation show 6–7 years fallout of volcanic aerosol, and further (770 km3) is almost half of the value estimated by Matthews analyses reveal concentrations of chloride, nitrate, and the et al. [2012]. However, the authors reported an overestima- ratio of Cl2+ to Na+ in the GISP2 YTT layer, demonstrating tion of the modeled thicknesses between 0.2 and 5 times the a strong impact of the eruption on tropospheric chemistry observed data that could account for these differences. [Yang et al., 1996]. Oppenheimer [2002], nevertheless, argued that the apparent perturbations to ice core chemistry 5.2. Limitations and Uncertainties could reflect postdeposition reactions occurring in the highly acidic YTT ice horizon and pointed out that no other ice Although the study of ash deposits and dispersal has core has yet been reported with a similar peak at ~70 ka. provided useful insights about the YTT eruption, which can Noticeably, the smaller sulfur spikes retrieved at ~72 ka and provide data for climatic and environmental models, major 73.6 ka (Figure 5) have been largely ignored, due to size- issues still exist when trying to model the global climatic and biased interpretation. However, the most recent studies on environmental impact of major eruptions (Table 3). YTT melt inclusions [Chesner and Luhr, 2010] encourage a Sulfate aerosol is the main volcanic output responsible for new comparison between the smaller spikes identified by global climatic posteruption effects. The database here pre- Zielinski et al. [1996a] and the sulfur output determined from sented yields no new information about the amount of sulfur the YTT melt inclusion data. emitted by YTT. Tephra deposits cannot provide direct information about The only two proxies currently available are the crystal- seasonal or latitudinal distributions of volcanic sulfate aero- hosted melt inclusions, which suggested about 1014 gof sols. For instance, simulation of multiple-season eruptions H2SO4 aerosol production [Chesner and Luhr, 2010], and based on the 1991 eruption of Mt. Pinatubo showed that the the glaciochemical records in the GISP2 ice core from Sum- same Pinatubo eruption in January would have led to a ~25% mit, Greenland [Zielinski et al., 1996a, 1996b]. The ice core larger global mean aerosol optical depth than eruptions in 2À record revealed a strong SO4 spike at 71 ± 5 ka and several other months [Toohey et al., 2011]. Similarly, continuous smaller spikes at 69.4 ka, 72 ka, and 73.6 ka (Figure 5). The measurements of SO2 concentrations at the summit of Mt. ~71 ka spike was identified as YTT fallout, although no ash Fuji [Igarashi et al., 2006] showed that a major factor in

Table 3. Summary of Tephra Proxies Relevant for the Assessment of the YTT Supereruption Impact and Present State of Knowledge Regarding Its Potential Environmental and Climatic Effectsa Long-Term Long-Term Short-Term Environmental Short-Term Climatic Environmental Climatic Tephra Proxy Index Impact Impact Impact Impact Particle size fragmentation, column yes (indirectly) yes (indirectly) no no dynamics, eruption style Thickness ejected volume, yes (indirectly) yes (indirectly) no no total grain size distribution Sulfur in melt sulfur production yes yes yes (indirectly) yes inclusion Areal distribution wind direction no no yes (indirectly) yes (indirectly) Suggested impact disruption of vegetation between 2 years to unknown ~200 years of Toba growth and hydrological few decades of enhanced cooling networks, due to ash cooling due to [Zielinski et al., fallout [Williams et al., sulfate aerosol 1996b] 2009] [Robock et al., 2009] aTephra physical and chemical characteristics can be directly linked to eruption dynamics, but sulfur is the key to understand the impact of the eruption on climate. GATTI AND OPPENHEIMER 71

Figure 5. Image from the sulfur spike presented by Zielinski et al. [1996b]. Smaller volcanic spikes are visible at 69.4 ka, 72 ka, and 73.6 ka (dotted lines). The authors chose to use the ~71 ka spike as indicator of the YTT. However, they highlighted as “the low concentrations of these other peaks do not preclude the possibility that one of them is from Toba” [Zielinski et al., 1996b, p. 838].

controlling the seasonal amount of SO2 transported far from set of ash particle sizes and thicknesses, concluding the the vent is the reduced removal of SO2 from the plume in the following: cold and dry air of winter. 1. The finest component of the distal ash deposits appears homogeneous and its thickening is independent of the dis- 6. CONCLUSIONS tance from Toba. This confirms the early interpretation of Rose and Chesner [1987] and highlights the complexity of Since the end of the 1980s, several new YTT ash deposits modeling ash thickness with an exponential decay model. have been discovered in Asia. A re-analysis of the signifi- 2. The most reliable and simple statistical interpolation to cance of the YTT tephra deposits and an update of the current assess the total volume of coignimbrite ash ejected is the data was therefore timely. We presented here an updated data Voronoi tessellation, which indicates between 770 km3 72 UTILIZATION OF DISTAL TEPHRA RECORDS

(DRE) and 2000 km3 (DRE). The results are similar to the Chesner, C. A., W. I. Rose, A. Deino, R. Drake, and J. A. Westgate first estimates provided by Rose and Chesner [1987] and the (1991), Eruptive history of Earth’s largest Quaternary caldera most recent values provided by Matthews et al. [2012]. (Toba, Indonesia) clarified, Geology, 19(3), 200–203. 3. Important information relating the eruption to the gen- Dehn, J., J. W. Farrel, and H. U. Schmincke (1991), Neogene eral climatic context and the season of the eruption (and thus tephrochronology from site 758 on northern Ninetyeast Ridge: the local environmental and climatic impacts) remains Indonesian arc volcanism of the past 5 Ma, Proc. Ocean Drill. Program Sci. Results, 121, 273–295. unresolved. Gasparotto, G., E. Spadafora, V. Summa, and F. Tateo (2000), In order to parameterize models of the eruption’s possible Contribution of grain size and compositional data from the Ben- global consequences, tangible data on the YTT eruption gal Fan sediment to the understanding of Toba volcanic event, based on high-temporal resolution paleoclimate records, Mar. Geol., 162(2–4), 561–572. along with advances in understanding the contemporary Gathorne-Hardy, F. J., and W. E. H. Harcourt-Smith (2003), The climate and atmospheric dynamics and their response to the super-eruption of Toba, did it cause a human bottleneck?, eruption, will be essential. J. Human Evol., 45(3), 227–230. Herzog, M., and H.-F. Graf (2010), Applying the three- Acknowledgments. E.G. thanks the British Geological Society dimensional model ATHAM to volcanic plumes: Dynamic of and the International Association of Sedimentology for the support large co-ignimbrite eruptions and associated injection heights provided to attend the Chapman Conference on climate, past land- for volcanic gases, Geophys. Res. Lett., 37, L19807, doi:10.1029/ scapes, and civilization; H. Kudrass for providing particle size data 2010GL044986. of the Bay of Bengal sites; and A. Durant for useful discussions Igarashi, Y., Y. Sawa, K. Yoshioka, H. Takahashi, H. Matsueda, and about the ash sites in India. We also acknowledge S. Boreham and Y. Dokiya (2006), Seasonal variations in SO2 plume transport C. Rolfe from the Geography Science Laboratories of the Univer- over Japan: Observations at the summit of Mt. Fuji from winter to sity of Cambridge for assistance with the particle size analyses. The summer, Atmos. Environ., 40(36), 7018–7033. authors thank Peter Clift, Stephen Self, and the two anonymous Kale, V. S., D. N. Patil, N. J. Powar, and S. N. Rajaguru (1993), reviewers for the valuable suggestions. Discovery of a volcanic ash bed in the alluvial sediments at Morgaon, Maharashtra, Man Environ., 18, 141–143. REFERENCES Korisettar, R., S. Mishra, S. N. Rajaguru, V. D. Gogte, R. K. Ganjoo, T. R. Venkatesan, S. K. Tandon, B. L. K. Somayajulu, Acharyya, S. K., and P. K. Basu (1993), Toba ash on the Indian and V. S. Kale (1988), Age of the Bori volcanic ash and Lower subcontinent and its implications for correlation of Late Pleisto- Palaeolithic culture of the Kukdi Valley, Maharashtra, Bull. Dec- cene alluvium, Quat. Res., 40(1), 10–19. can Coll. Postgrad. Res. Inst., 48, 135–138. Ambrose, S. H. (1998), Late Pleistocene human population bottle- Ledbetter, M. T., and R. S. J. Sparks (1979), Duration of large- necks, volcanic winter, and differentiation of modern humans, magnitude explosive eruptions deduced from graded bedding in J. Human Evol., 34(6), 623–651. deep-sea ash layers, Geology, 7, 240–244. Bonadonna, C., and B. F. Houghton (2005), Total grain-size distri- Liang, X., G. Wei, L. Shao, X. Li, and R. Wang (2001), Records of bution and volume of tephra-fall deposits, Bull. Volcanol., 67(5), Toba eruptions in the South China Sea, Sci. China, Ser. D, 441–456. 44(10), 871–878. Bonadonna, C., and J. C. Phillips (2003), Sedimentation from Liu, Z., C. Colin, and A. Trentesaux (2006), Major element geochem- strong volcanic plumes, J. Geophys. Res., 108(B7), 2340, istry of glass shards and minerals of the Youngest Toba Tephra in the doi:10.1029/2002JB002034. southwestern South China Sea, J. Asian Earth Sci., 27(1), 99–107. Buhring, C., M. Sarnthein, and Leg 184 Shipboard Scientific Party Matthews, N. E., V. C. Smith, A. Costa, A. J. Durant, D. M. Pyle, (2000), Toba ash layers in the South China Sea: Evidence of and N. J. G. Pearce (2012), Ultra-distal tephra deposits from contrasting wind directions during eruption ca. 74 ka, Geology, super-eruptions: Examples from Toba, Indonesia and Taupo Vol- 28(3), 275–278. canic Zone, New Zealand, Quat. Int., 258,54–79. Chesner, C. A. (1998), Petrogenesis of the Toba Tuffs, Sumatra, Ninkovich, D. (1979), Distribution, age and chemical composition Indonesia, J. Petrol., 39(3), 397–438. of tephra layers in deep-sea sediments off western Indonesia, Chesner, C. A. (2012), The Toba caldera complex, Quat. Int., 258, J. Volcanol. Geotherm. Res., 5(1–2), 67–86. 5–18. Ninkovich, D., R. Sparks, and M. Ledbetter (1978b), The excep- Chesner, C. A., and J. F. Luhr (2010), A melt inclusion study of the tional magnitude and intensity of the Toba eruption, Sumatra: An Toba Tuffs, Sumatra, Indonesia, J. Volcanol. Geotherm. Res., example of the use of deep-sea tephra layers as a geological tool, 197(1–4), 259–278. Bull. Volcanol., 41(3), 286–298. Chesner, C. A., and W. I. Rose (1991), Stratigraphy of the Toba Oppenheimer, C. (2002), Limited global change due to the largest Tuffs and the evolution of the Toba Caldera Complex, Sumatra, known Quaternary eruption, Toba ~74 kyr BP?, Quat. Sci. Rev., Indonesia, Bull. Volcanol., 53(5), 343–356. 21(14–15), 1593–1609. GATTI AND OPPENHEIMER 73

Pattan, J. N., P. Shane, and V. K. Banakar (1999), New occurrence Shane, P., J. Westgate, M. Williams, and R. Korisettar (1995), New of Youngest Toba Tuff in abyssal sediments of the Central Indian geochemical evidence for the Youngest Toba Tuff in India, Quat. Basin, Mar. Geol., 155(3–4), 243–248. Res., 44(2), 200–204. Pyle, D. M. (1989), The thickness, volume and grainsize of tephra Stauffer, P. H., S. Nishimura, and B. C. Batchelor (1980), Volcanic fall deposits, Bull. Volcanol., 51(1), 1–15. ash in Malaya from catastrophic eruption of Toba, Sumatra, Pyle, D. M., G. D. Ricketts, V. Margari, T. H. van Andel, A. A. 30,000 years ago, in Physical Geology of Indonesian Island Arcs, Sinitsyn, N. D. Praslov, and S. Lisitsyn (2006), Wide dispersal edited by S. Nishimura, pp. 156–164, Kyoto Univ., Kyoto, Japan. and deposition of distal tephra during the Pleistocene Campanian Toohey, M., K. Kruger, U. Niemeier, and C. Timmreck (2011), The Ignimbrite/Y5 eruption, Italy, Quat. Sci. Rev., 25(21–22), inﬂuence of eruption season on the global aerosol evolution and 2713–2728. radiative impact of tropical volcanic eruptions, Atmos. Chem. Raj, R. (2007), Occurrence of volcanic ash in the Quaternary Phys. Disc., 11(8), 22,443–22,481. alluvial deposits, lower Narmada basin, western India, J. Earth van der Kaars, S., M. A. J. Williams, F. Bassinot, F. O. Guichard, E. Syst. Sci., 117(1), 41–48. Moreno, F. Dewilde, and E. J. Cook (2012), The inﬂuence of the Rampino, M. R., and S. H. Ambrose (2000), Volcanic winter in ~73 ka Toba super-eruption on the ecosystems of northern Suma- the Garden of Eden: The Toba supereruption and the late Pleis- tra as recorded in marine core BAR94-25, Quat. Int., 258,45–53. tocene human population crash, in Volcanic Hazards and Disas- Von Rad, U., K.-P. Burgath, M. Pervaz, and H. Schulz (2002), ters in Human Antiquity, edited by F. W. McCoy and G. Heiken, Discovery of the Toba Ash (c. 70 ka) in a high-resolution core Spec. Pap. Geol. Soc. Am., 345,71–82. recovering millennial monsoonal variability off Pakistan, Geol. Rampino, M. R., and S. Self (1992), Volcanic winter and acceler- Soc. Spec. Publ., 195(1), 445–461. ated glaciation following the Toba super-eruption, Nature, 359 Westgate, J. A., P. A. R. Shane, N. J. G. Pearce, W. T. Perkins, R. (6390), 50–52. Korisettar, C. A. Chesner, M. A. J. Williams, and S. K. Acharyya Rampino, M. R., and S. Self (1993), Climate-volcanism feedback (1998), All Toba tephra occurrences across peninsular India and the Toba eruption of ~74,000 years ago, Quat. Res., 40(3), belong to the 75,000 yr B.P. eruption, Quat. Res., 50(1), 269–280. 107–112. Robock, A., C. M. Ammann, L. Oman, D. Shindell, S. Levis, and G. Williams, M. A. J., S. H. Ambrose, S. van der Kaars, C. Ruehle- Stenchikov (2009), Did the Toba volcanic eruption of ~74 ka B.P. mann, U. Chattopadhyaya, J. Pal, and P. R. Chauhan (2009), produce widespread glaciation?, J. Geophys. Res., 114, D10107, Environmental impact of the 73 ka Toba super-eruption in South doi:10.1029/2008JD011652. Asia, Palaeogeogr. Palaeoclimatol. Palaeoecol., 284(3–4), Rose, W. I., and C. A. Chesner (1987), Dispersal of ash in the great 295–314. Toba eruption, 75 ka, Geology, 15(10), 913–917. Yang, Q., P. A. Mayewski, G. A. Zielinski, M. Twickler, and K. C. Rose, W. I., and C. A. Chesner (1990), Worldwide dispersal of Taylor (1996), Depletion of atmospheric nitrate and chloride as a ash and gases from Earth’s largest known eruption: Toba, Suma- consequence of the Toba volcanic eruption, Geophys. Res. Lett., tra, 75 ka, Palaeogeogr. Palaeoclimatol. Palaeoecol., 89(3), 23(18), 2513–2516. 269–275. Zielinski, G. A., P. A. Mayewski, L. D. Meeker, S. Whitlow, and Sarna-Wojcicki, A. M., and J. O. Davis (1991), Quaternary tephro- M. S. Twickler (1996a), A 110,000-yr record of explosive vol- chronology, in The Geology of North America, Quaternary Non- canism from the GISP2 (Greenland) ice core, Quat. Res., 45(2), glacial Geology: Conterminous U.S., edited by R. B. Morrison, 109–118. pp. 93–116, Geol. Soc. of Am., Boulder, Colo. Zielinski, G. A., P. A. Mayewski, L. D. Meeker, S. Whitlow, M. S. Scaillet, B., B. Clemente, B. W. Evans, and M. Pichavant (1998), Twickler, and K. Taylor (1996b), Potential atmospheric impact of Redox control of sulfur degassing in silicic magmas, J. Geophys. the Toba Mega-Eruption ~71,000 years ago, Geophys. Res. Lett., Res., 103(B10), 23,937–23,949. 23(8), 837–840. Schulz, H., K.-C. Emeis, H. Erlenkeuser, U. von Rad, and C. Rolf (2002), The Toba volcanic event and interstadial/stadial climates E. Gatti and C. Oppenheimer, Department of Geography, at the marine isotopic stage 5 to 4 transition in the northern Indian University of Cambridge, Downing Place, Cambridge CB2 3EN, Ocean, Quat. Res., 57(1), 22–31. UK. ([email protected]) An Abrupt Shift in the Indian Monsoon 4000 Years Ago

M. Berkelhammer,1 A. Sinha,2 L. Stott,3 H. Cheng,4,5 F. S. R. Pausata,6,7,8 and K. Yoshimura9,10

The emergence of high-resolution proxy records from the Asian monsoon region suggests that the monsoon system is bistable and can abruptly transition between a suppressed and active state. This observation is critical in considering how the monsoon system may have inﬂuenced the development of societies across South and East Asia during the Holocene. Using a new high-resolution (~5 years/sample) speleothem stable isotope record from northeast India that spans the early and mid- Holocene, a number of abrupt changes in the oxygen isotopic composition of 18 precipitation (δ Op) are documented. The most dramatic of these events occurred ~4000 years ago when, over the course of approximately a decade, isotopic values abruptly rose above any seen during the early to mid-Holocene and remained at this anomalous state for almost two centuries. This event occurs nearly synchronously with climatic changes documented in a number of proxy records across North Africa, the Middle East, the Tibetan Plateau, southern Europe, and North America. We hypothesize that the excursion could represent a shift toward an earlier Indian Summer Monsoon withdrawal or a general decline in the total amount of monsoon precipitation. The new record provides a very signiﬁcant advance with respect to age control and sample resolution of terrestrial climate change over South Asia during this period when a number of major societal changes occurred. While evidence of a causal relationship between climate and the reorganization of the Indus Valley and Old Kingdom Nile civilizations is beyond the scope of this study, the tight age constraints of the record show with a high degree of certainty that much of the documented deurbanization of the Indus Valley at 3.9 kyr B.P. occurred after multiple decades of a shift in the monsoon’s character but before the monsoon returned to its previous mid-Holocene state.

1Department of Atmospheric and Oceanic Sciences and Coop- 4Institute of Global Environmental Change, Xian Jiaotong Uni- erative Institute for Research in Environmental Sciences, University versity, Shanghai, China. of Colorado, Boulder, Colorado, USA. 5Department of Earth Sciences, University of Minnesota, 2Department of Earth Sciences, California State University, Minneapolis, Minnesota, USA. Dominguez Hills, Carson, California, USA. 6Bjerknes Center for Climate Research, Bergen, Norway. 3Department of Earth Sciences, University of Southern Califor- 7Geophysical Institute, University of Bergen, Bergen, Norway. nia, Los Angeles, California, USA. 8Joint Research Centre, Institute for Environment and Sustain- ability, European Commission, Ispra, Italy. Climates, Landscapes, and Civilizations 9Atmosphere and Ocean Research Institute, University of Geophysical Monograph Series 198 Tokyo, Tokyo, Japan. © 2012. American Geophysical Union. All Rights Reserved. 10Scripps Institution of Oceanography, University of California, 10.1029/2012GM001207 San Diego, La Jolla, California, USA. 75 76 ISM AT 4000 YEARS AGO

1. INTRODUCTION cords of previous monsoon failures are therefore useful both for testing the dynamics of monsoon suppression and also for Indian Summer Monsoon (ISM) precipitation during the understanding the role that climate and water availability twentieth century is characterized by a biennial oscillation, played in previous societal changes across Asia [Graham et such that ISM precipitation varied between singularly strong al., 2011; Pausata et al., 2011; Zhang et al., 2008; Sinha et and weak years but rarely deviated far from its mean state for al., 2007]. For example, evidence of widespread famine and consecutive years [Meehl and Arblaster, 2002]. This obser- societal unrest across India during the major droughts of the vation has engendered a hypothesis that dynamic feedbacks fourteenth and seventeenth centuries illustrates the cultural internal to the monsoon system ballasts it against sustained impacts of such shifts in the monsoon system [Sinha et al., anomalies. Meehl [1993] and Loschnigg and Webster [2000] 2007] and also provides a test of the capacity for hindcast argue that the Indian Ocean (IO) provides the inertia to carry general circulation model simulations to capture the dynam- the influence of 1 year’s monsoon circulation on to the next. ics of monsoon change [Graham et al., 2011]. According to this hypothesis, warm ocean temperatures in In this paper, we present a new record of the δ18O of calcite the IO are conducive to enhanced convection, which through from a speleothem that spans 3600 to 12,500 years B.P. The air-sea interactions and large-scale wind fields during strong proxy is located in Cherrapunji, India, which is among the monsoon circulation, cool the surface ocean and ultimately wettest locations on Earth with an annual average precipita- suppress convection the following year. tion in excess of 11,000 mm, 70% of which falls during the The development of high-resolution paleoclimatic records summer monsoon months (June through September) [Mu- from regions in the heart of the monsoon domain provide a rata et al., 2007]. While the site is not located in the core long-term perspective on monsoon precipitation variability monsoon zone of India [Hoyos and Webster, 2007], it is [Buckley et al., 2007; Sinha et al., 2011a; Buckley et al., 2010; situated in a location that is highly sensitive to the northward Cook et al., 2010; Borgaonkar et al., 2010; Sano et al., 2009], propagating convective systems that originate in the Bay of which allows for an empirical test of the theory that the ISM Bengal and penetrate into the Tibetan Plateau. The record is a self-regulating system. These proxy records from the provides one of the first high-resolution terrestrial-based monsoon region, which expand in many cases through the proxies of the monsoon over South Asia through the early last millennium, clearly show the presence of monsoon to mid-Holocene and, therefore, yields opportunities both to anomalies that may last for decades or longer [Buckley et al., capture the response of the monsoon to shifts in large-scale 2010; Cook et al.,2010;Sinha et al., 2011a]. The most boundary conditions and to consider how the monsoon sys- pronounced and best replicated of these events occurred tem influenced the spatiotemporal development of civiliza- during the mid- to late fourteenth century when monsoon tion across the Indian subcontinent. In the context of this precipitation was reduced to 80% of its mean for over 30 multimillennia record, we focus our analysis and discussion years [Berkelhammer et al., 2010]. Levermann et al. [2009] on the presence of a singular abrupt and sustained change in and Schewe et al. [2011] use a simple dynamical model to the monsoon that occurred approximately 4000 years B.P. show the capacity for the monsoon system to bifurcate between and lasted for almost two centuries (Figure 1). an active and suppressed state due to feedbacks between The monsoonal shift documented here occurs temporally moisture advection and latent heating from condensation close to the 4.2k Event, which was characterized by a series of during monsoon precipitation. Based on this model, sustained abrupt and nearly concomitant climatological changes that monsoon suppression, as observed in the paleorecord, arises occurred in the Middle East [Weiss et al., 1993; Cullen et al., from the influence that external boundary conditions such as 2000], and nearly synchronous with large hydrological radiative forcing or low-frequency large-scale ocean modes changes in East Africa [Gasse, 2000; Thompson et al.,2002; have on IO sea surface temperatures (SSTs). These changes in Arz et al., 2006], southern Europe [Drysdale et al., 2006; SSTs influence the moisture-carrying capacity of the marine Wagner et al., 2009], and southern China [Hong et al., atmosphere and reduce moisture advection onto the Indian 2003] as well as in the tropical IO [Rijsdijk et al., 2011]. subcontinent during monsoon circulation. In a paleoclimate There exists a standing controversy regarding the role that context, these changes appear large enough to override the changes in the monsoon at this time may have played in the internal stability that occurs from the quasi-biennial monsoon large-scale deurbanization of the Indus Valley Civilization feedback cycle [i.e., Meehl, 1993]. that occupied a region presently part of northwest India and The capacity for the system to bifurcate into a suppressed Pakistan [Madella and Fuller, 2006; Misra, 1984; Clift et al., state is a source of prevailing concern in light of the con- 2012]. A clear linkage between a shift in either the strength sequences this would impose on municipal and agricultural of the monsoon or winter season precipitation regimes and systems across the Indian subcontinent. Paleoclimatic re- the documented societal changes at the time have long been BERKELHAMMER ET AL. 77

Figure 1. The KM-A isotope record from NE Indian is shown both as raw data (purple) and smoothed using a Lanczos filter (black) where the isotopic ratios are reported relative to Vienna PeeDee Belemnite and the y axis is reversed. Dots on the x axis are used to delineate the locations where U/Th dates were used to generate the age model. The size of the dot is proportional to the uncertainty. Please see Table 1 of supplementary material and Figure 2 for additional information on the dates. The location of the record is marked on the map as a star. A stack of additional records whose locations are represented with color coding on the map are shown for comparison. (top to bottom) Gulf of Oman dolomite record [Cullen et al., 2000], the Shabban Deep foraminifera record [Arz et al., 2006], the Mount Kilimanjaro dust record [Thompson et al., 2002], southern Italy speleothem record [Drysdale et al., 2006], Tibetan Plateau δ13C sedimentary cellulose [Hong et al., 2003], and lake level record from central United States [Booth et al., 2005]. hampered by (1) poor age constraints on many of the regional climatic changes during this period [Fleitmann et al., 2003]. proxy records [Singh et al., 1990; Enzel et al., 1999] and We aim to shed light on this controversy by providing a (2) that many of the proxies such as lake pollen or sedimen- direct terrestrial climate proxy from the Indian subcontinent, tation rates may be influenced by nonclimatic factors such as which has unprecedented age constraints (20 years) and tectonics or land use change [Madella and Fuller, 2006; Clift resolution (5 years) during the period in which signifi- et al., 2012]. Further adding to the enigma is the fact that cant deurbanization occurred. The new record presented some key regional proxies show no obvious evidence for here suggests that the period of large-scale Indus Valley 78 ISM AT 4000 YEARS AGO deurbanization [Madella and Fuller, 2006] occurred while 2.2. Age Model the Indian Monsoon system was experiencing an anomaly larger than any other during the early to mid-Holocene. The age-depth relationship along the sample was constrained with 12 U/Th dates measured at the University of 2. METHODS Minnesota following the methods of Cheng et al. [2009] (see Table 1) (refer to Appendix A for details of supple- 2.1. Analytical Analysis mentary material). The age model was developed using the approach described by Scholz and Hoffmann [2011], where In 2005, a calcitic stalagmite (KM-A) was collected from a linear interpolation between depth and age is made through Mawmluh Cave (25°15′44″N, 91°52′54″E,elevation1290m) each progressive triplet of adjacent U/Th dates (Figure 2). located in Cherrapunji, Meghalaya, in northeast India. KM- This process is repeated for 10,000 iterations using the two- A was collected from a large chamber of the cave about 1500 sigma uncertainty of the U/Th date as constraints in “jig- m from the entrance. Continuous temperature measurements ging” the age model. This procedure affords a quantitative over the course of a year from inside the chamber suggest a method to continuously assess the age uncertainty along the remarkably stable ambient environment with temperatures record despite having analytical constraints only at locations varying between 18.0°C and 18.5°C [Breitenbach, 2009]. where the U/Th dates exist. Furthermore, this iterative pro- Repeated spot measurements of the relative humidity cedure minimizes the influence imparted on the age model showed values in excess of 95% even during the dry season. by subjective decisions regarding the depths U/Th measure- High relative humidity, minimal temperature fluctuations, ments are made. and the deep location of the sampling spot represent theoret- Although the full data set from KM-A is presented in ically optimal conditions for calcite to form in isotopic equi- Figure 1, we focus the discussion primarily on the time librium with the percolating precipitation. While no period between 3800 and 5500 B.P. when the largest singular mineralogical analyses were done on the sample, there were event is observed. During this period, the age uncertainty no obvious visual changes in the color or texture during the is <30 years, and the record shows linear growth rates, 4 kyr B.P. excursion. Furthermore, because this region is providing confidence in the timing (onset and duration) of among the wettest locations on Earth and has experienced the event. We emphasize that due to large age uncertainty only minimal interannual variations during the instrumental and a rapid change in the growth rate during the middle part period, we have confidence that the vapor pressure in the of this record (i.e., ~6–12 kyr B.P., Figure 2), this section is chamber would remain close to saturated even during peri- presented only preliminarily until additional constraints on ods when the monsoon was relatively weak. This suggests the ages are available. While the growth rate obtained from that variations in the δ18O of the calcite reflect, to a first the age-depth relationship could also be used as an additional 18 order, changes in the precipitation-weighted δ Op. The sample was split along its growth axis, and continuous powdered calcite samples were removed using both a Table 1. U/Th Dates Used to Develop the KM-A Age Model Dremel tool and automated drilling system. The isotopic Following the Analytical Methods of Cheng et al. [2009]a composition of individual powdered calcite samples was Depth (mm) Age (B.P.) Error measured using dual-inlet isotope ratio mass spectrometry, where the calcite is reacted with phosphoric acid, and the 0 3,654 20 6 4,112 30 resulting CO2 is measured relative to a reference gas that has been calibrated against a series of known isotopic standards. 29 5,084 32 48 5,725 115 The analyses were performed using an online automated 83 6,058 31 carbonate preparation system linked to a VG Prism II isotope 115 6,518 59 ratio mass spectrometer housed at the University of Southern 128 7,285 21 ’ California s stable isotope lab. Repeat measurements of stan- 139 7,946 21 dards and samples reveal a procedural uncertainty of 167 9,706 284 <0.15‰. A total of 1128 isotopic measurements were made 178 10,551 480 principally during 2009. The samples spanning the 3.8 to 4.5 193 11,216 86 kyr B.P. period were rerun on the same instrument in 2011 to aBold font delineates the dates that bracket the 4.2 kyr event. ensure the analytical uncertainty during this critical period of Table 1 of supplementary material includes raw U and Th measure- the record were representative of the system’s typical ments and some additional dates that were excluded from the age behavior. model because of extraordinarily large errors. BERKELHAMMER ET AL. 79

Figure 2. Age model for KM-A stalagmite based on the age-depth relationship of 12 U/Th dates. The age model was generated using StalAGE Scholz and Hoffmann [2011] where a Monte Carlo simulation is used to generate linear fits between each progressive subset (3) of dates. The two-sigma analytical uncertainty of each measured date (dots) is represented by the error bars, whereas the 95% uncertainty assessed through the Monte Carlo simulation is represented by the light gray lines. The algorithm produces a realistic representation of age uncertainty through regions of the stalagmite where no radiometric dates are available. The highlighted region in gray represents the time period in which this paper principally is focused. Through this section, the age error is typically less than 30 years. Table 1 of supplementary material contains some additional dates from KM-A through the middle section of the record that were excluded from the age model due to very large age uncertainty. This section of the age model is thus considered only preliminary. climate proxy, we have no direct empirical evidence or falling at a given location (“amount effect”)[Dansgaard, strong theoretical guidance to transfer changes in growth rate 1964] or changes in the moisture source region [Pausata et into an environmental signal. Thus, with respect to climate al., 2011]. A number of pointed criticisms in the recent variability, all discussion will focus on variations in the δ18O literature have required a reconsideration of the appropriate of the calcite. interpretation of the speleothem records from the inland regions of the Asian monsoon suggesting they are not in fact 2.3. Interpreting the Proxy proxies for regional/local precipitation changes, but rather, reflect changes in seasonal precipitation balances and shifts Speleothem proxies from the Asian monsoon region have in the isotopic composition of water vapor as a consequence been interpreted to largely reflect changes in monsoon of “upstream” processes over India [LeGrande and strength because the isotopic composition of the speleothems Schmidt, 2009; Pausata et al., 2011]. Both modeling and is highly correlated with changes in local summer insolation observational studies, however, indicate the presence of a due to the precession of the equinoxes [Wang et al., 2001]. consistently robust “amount effect” over the Indian subcon- Orbital variations affect the δ18O in speleothems by changing tinent and uniformly suggest that isotopic records from this local precipitation processes and/or amounts [Wang et al., region yield information on monsoon precipitation amounts 2001; Cheng et al., 2009], such as the ratio of summer-to- [Dayem et al., 2010; Pausata et al., 2011; Vuille et al., 2005]. spring precipitation (summer rainfall has more negative The region of NE India, from where the sample in this 18 δ Op than spring precipitation), the intensity of precipitation study was acquired, is outside of the core Indian monsoon 80 ISM AT 4000 YEARS AGO zone [Berkelhammer et al., 2010] where most previous mod- IsoGSM reproduces the observed event-scale isotopic var- eling and observational studies of the isotopic composition iability from Breitenbach et al. [2010] with skill (r2 = 0.81, of ISM precipitation have been done [Vuille et al., 2005; p << 0.01) (Figure 3) and the year-to-year variability in Sengupta and Sarkar, 2006]. Breitenbach et al. [2010] con- precipitation amounts (r2 = 0.41, p << 0.01), which gives ducted a 2 year high-resolution sampling campaign in Cher- confidence that the monsoon dynamics in the model are 18 18 rapunji of δ Op, which is likely too brief and sporadic to represented sufficiently to generate a time series of δ Op 18 18 adequately test the climatic controls on δ Op. We thus test that can be used to test the climatic controls on δ Op.As 18 the climatic controls on δ Op using an isotope-enabled suggested from previous work, we find an inverse relation- 18 general circulationmodel(IsoGSM)[Yoshimura et al., ship between δ Op of monsoon season (May–October) pre- 2008], which provides an extended (multidecadal), albeit cipitation and monsoon season precipitation amount (Figure synthetic, data set of regional isotopic values and precipita- 3). A significant relationship is, however, only present when tion amount (Figure 3). a seasonal-averaging window that includes September and

18 Figure 3. An analysis of the amount effect in NE India: (a) Comparison of modeled precipitation rate and δ Op averaged for MJJASO from 1970 to 2008 from the isotope-enabled general circulation model (IsoGSM) [Yoshimura et al., 2008]. (b) The linear fit between the time series shown in Figure 3a (green) and the gray dots (filled) shows the results of a robust regression analysis, and the open gray dots are the one-sigma uncertainty around this fit. From this, we infer that changes in precipitation rate can account for up to 30% of the observed interannual variability in monsoonal precipitation at this site. 18 (c and d) A comparison between daily observed [Breitenbach et al., 2010] and modeled δ Op in Cherrapunji. This direct daily comparison between model and observation is feasible because of the use of a spectral nudging procedure in the simulation, which forces the model at each time step toward Reanalysis II atmospheric climate fields [Yoshimura et al., 2008]. From here, we infer the model is capable of reproducing the isotopic composition of precipitation at this site and is 18 therefore useful for extrapolation and assessing the climate controls on δ Op as in Figures 3a and 3b. BERKELHAMMER ET AL. 81

October is used, suggesting that the presence of an “amount- rainout and is therefore more depleted in heavy isotopes. In effect” in this region largely reflects the close relationship addition, a change in land surface processes that would shift between total monsoon precipitation and the timing of mon- the proportion of transpiration and bare-soil evaporation 18 soon withdrawal [Fasullo and Webster, 2003]. Thus, during would influence δ Op because the former seeds the atmo- strong monsoon years, precipitation extends later into au- sphere with relatively enriched vapor and the latter with tumn [Fasullo and Webster, 2003], and the isotopic compo- relatively depleted vapor. Last, there is a strong seasonality 18 sition of the accumulated precipitation is relatively depleted as in δ Op in this region (Figure 3), thus a change in the timing a consequence of these late monsoon season rains, which are of monsoon onset (which is not particularly sensitive to total the isotopically lightest rains of the year [Sengupta and monsoon rainfall) [Fasullo and Webster, 2003] would pro- 18 Sarkar, 2006; Breitenbach et al., 2010]. duce a change in integrated δ Op value that may not reflect a Further empirical evidence of the precipitation signal in change in monsoon strength. The multitude of processes 18 the KM-A record is garnered through isotopic analysis of a influencing δ Op would require a transfer function between 18 nearby speleothem record (WS-B from Sinha et al. [2011b]), δ Op and climate that would be multivariate [Johnson and which grew throughout the instrumental era and can thus be Ingram, 2004], likely nonlinear and, ultimately, nonstation- calibrated directly against instrumental precipitation records. ary. Furthermore, changes in the temperature of the chamber The record suggests the presence of a negative correlation and humidity would impart effects on thermodynamic and between δ18O of the calcite and local precipitation amount kinetic fractionation between water and calcite that could 18 and therefore that the inverse relationship between δ Op and produce some small isotopic variability not directly tied to precipitation amount as observed in IsoGSM is carried monsoon processes. Until model simulations with isotope through to the speleothem record. tracers and representative 4000 year B.P. boundary condi- 18 The relationship between precipitation amount and δ Op tions are available, we will not attempt to cast the isotopic however, only accounts for between 20% and 30% of the changes directly into quantitative climate terms. total isotopic variability. Thus, while some of the variability in the KM-A record may be a direct proxy for monsoon 3. RESULTS strength, it is critical to acknowledge that changes in δ18O reflect a myriad of potential climatic processes. Breitenbach The complete isotopic record (Figure 1) is characterized by et al. [2010] and Sengupta and Sarkar [2006] discuss how a a long-term parabolic trend, which is punctuated by a number southerly shift in the locus of convective activity in the Bay of multicentennial excursions. The isotopic values in this of Bengal lead to precipitation that had experienced greater speleothem are similar to, but more depleted than, modern

Figure 4. Probability distribution functions of the δ18O of the KM-A (medium gray) and WS-B (light gray) [Sinha et al., 2011b] stalagmites during selected time windows. The PDFs were created using a nonparametric ﬁt with a normal kernel. The width of the kernel was optimized following the default algorithm in the Matlab Software package. The data for the twentieth century distribution used all the data from the WS-B stalagmite from Sinha et al. [2011b], whereas the 4.2 kyr data used all the data points between 4.1 and 3.8 kyr B.P., and the mid-Holocene data is that taken from 7–8 kyr B.P. 82 ISM AT 4000 YEARS AGO

18 calcite values measured from an actively growing sample in a an increase in δ Op, including a more northerly locus for the nearby cave (Figure 4) [Sinha et al., 2011b]. The most isoto- convective systems that form in the Bay of Bengal and rain- pically enriched values of the entire record occur between out over northeast India [Sengupta and Sarkar, 2006] or a 4071 B.P. (±18 years) and 3888 B.P. (±22 years) during general weakening of monsoon-related convection, which which the calcite remained enriched by ~0.8‰ relative to enriches the precipitation by increasing postcondensational modern values (1.5‰ relative to the background values of the evaporation [Lee and Fung, 2008]. time) for a period of 183 years. The isotopic changes at this The strength and/or seasonality of the monsoon is inti- time manifested as a two-step process where values experi- mately tied to large-scale tropical ocean-atmosphere dynam- enced a small steplike rise between ~4315 and 4303 years ics with secondary influences associated with land surface B.P. and experienced a second and more precipitous rise processes (i.e., latent heating and albedo) [Webster, 1987]. between ~4071 and 4049 years B.P. The abrupt shift occurred Abram et al. [2009] and Rashid et al. [2011] have both over approximately two decades, after which the values sta- reviewed SST records from the western Tropical Pacific and bilized at this relatively enriched state for ~180 years before IO from this time period and have found that SST patterns in rapidly returning to previous background values at 3888 years the region likely underwent a formative change with warm- B.P. Comparably enriched isotopic values were periodically ing in both the Indo-Pacific Warm Pool and eastern IO and reached during the instrumental era for single years but never cooling on the western margin of the IO. The SST pattern remained sustained at these enriched levels for more than a thus may have resembled something akin to a negative year or two [see Sinha et al., 2011b] (see also Figure 4). Thus, Indian Ocean Dipole (IOD) mode, which would have in- the monsoon over NE India at 4000 years B.P. could be creased divergence over India and reduced the overall considered analogous to end members of the modern mon- strength of monsoon circulation [Ashok et al., 2004]. A soon. The only other section of the KM-A record that shows similar mechanism involving changes in regional SSTs and comparably enriched values are those from ~12,000 B.P., their influence on the dynamics of the ITCZ have been called which may correspond to the Younger Dryas, a period that upon to explain a number of other regional climatic changes has been identified from marine and speleothem records as around 4000 years B.P. including severe droughts in East being characterized by having notably diminished monsoon Africa, Mauritius, and across the Mediterranean [Gasse, precipitation [Sinha et al., 2005; Rashid et al., 2011]. 2000; Rijsdijk et al., 2011; Thompson et al., 2002; Cullen et al., 2000; Drysdale et al., 2006]. 4. DISCUSSION In addition to dynamics within the IO basin, changes in El Niño–Southern Oscillation (ENSO) would have influenced 4.1. Climate Changes in NE India at 4 kyr B.P. both the seasonality and strength of the monsoon vis-à-vis a zonal shift in the subsiding limb of the Walker cell. Fasullo The monsoon over northeast India appears to have expe- and Webster [2003] find that the timing of monsoon onset is rienced an abrupt excursion at 4000 years B.P., the magni- strongly correlated with ENSO such that La Niña conditions tude of which, in terms of both amplitude and length, lead to an earlier onset. Thus, a change in ENSO would exceeds any other event during either the most recent 600 produce a climatic and isotopic influence over this region, years [Sinha et al., 2011a] or throughout the early to mid- and this effect could be modulated (i.e., exacerbated) in the Holocene. Modeling of the isotopic composition of precipi- presence of a persistently positive IOD [Ashok et al., 2004]. tation in this region suggests that a shift toward a consistently It may be that positive interactions between conditions in the earlier monsoon onset or earlier withdrawal could both fea- ENSO domain and regional changes in the IO are needed to sibly explain the observed isotopic enrichment during this produce sustained and significant climate changes in this time period. The earlier onset enriches isotopic values in the region. This interpretation is consistent with the findings of precipitation by including a greater percentage of spring Graham et al. [2011] who show that in order to effectively precipitation in the annually weighted accumulation, while model the hydroclimatic changes in Asia during the Medie- the earlier withdrawal reduces the contribution of depleted val Climate Anomaly, the combined influences of both late-season precipitation, also leading to a more enriched ENSO and the IO must be considered. A short coral record annually weighted precipitate. Early monsoon withdrawal from Vanuatu in the western Tropical PacificbyCorrège et would suggest that the monsoon was weakened at the time al. [2000] hint at the possibility of some extreme ENSO [Fasullo and Webster, 2003], which is consistent with the events during this period, but the record is too short to findings of Wang et al. [2005]. However, as noted earlier in properly assess the conditions at the time. The Ecuadorian this study and previously by Breitenbach et al. [2010], there lake record of Moy et al. [2002], however, indicates nothing are a number of additional mechanisms that could have led to anomalous about ENSO conditions during this time period. BERKELHAMMER ET AL. 83

Land surface processes could also have been a critical dent reconstruction of monsoonal changes documented here driver in influencing the strength and seasonality of the suggests that changes in river flow must still be considered in monsoon through changes in the land-sea thermal gradient understanding the deurbanization of the Indus Valley or continental recycling of water. The most proximal high- Civilization. resolution terrestrial record through this time period is the The Mawmluh record also provides a significant improve- lacustrine cellulose record from the Tibetan Plateau by Hong ment with respect to sample resolution and age control, et al. [2003]. This record depicts a significant enrichment in yielding a refined perspective on the duration and onset time both δ18O and δ13C of lake sedimentary cellulose, which of this climate event. As can be seen in Figure 1, a number of reflects either a severe warming and/or drying episode. The records seem to cluster close to one another (e.g., Gulf of multiproxy review of Tibetan Plateau temperatures by Mis- Oman, Shabban Deep, and W. North America), but there are chke and Zhang [2010], based on a suite of lake sediment also some records, which owing to either hysteresis effects in records and Himalayan ice cores, suggests, contrary to the the proxy (e.g., long karstic residence times), nonlinear findings of Hong et al. [2003], that there was an abrupt proxy response functions (e.g., dust spikes), or lower sam- cooling event around 4 kyr B.P. that exceeded in magnitude pling resolution that tend to blur the exact sequence of any other event during the Holocene. These studies together change. Madella and Fuller [2006] highlight how dating suggest that the Tibetan Plateau experienced an abrupt cool errors in low-resolution proxies can produce marked differ- [Mischke and Zhang, 2010] and dry [Hong et al., 2003] ences in understanding whether climate events observed in period in near synchronicity with the isotopic changes in NE distal sites are causally related and how reconstructed climate India. Cool conditions on the Plateau would be conducive to events might be related to societal change. For example, the weakened monsoon circulation, and the dry conditions [i.e., choice of how to apply radiocarbon corrections to the pollen Hong et al., 2003] suggest that the anomalous conditions we sequences of northwest India will change whether the deur- report from NE India were regionally pervasive. banization of the Indus Valley Civilization occurred during an anomalously wet or dry period [Madella and Fuller, 4.2. Societal and Dynamical Implications 2006]. In other words, low-resolution records with poor age control can produce antipodal hypotheses regarding climate- It is increasingly clear based on a growing number of cultural linkages. So while understanding the regional cul- records spanning across the Mediterranean, Middle East, tural responses to climate extends beyond the scope of the North Africa, North America, and China [Weiss, 2012] that study, the Mawmluh cave record cannot be “wiggled” more there was a significant large-scale climate excursion at ap- than a few years in any direction, which reduces any future proximately 4 kyr B.P. The new record presented here from ambiguities that may be associated with dating errors of Mawmluh Cave in NE India forwards work on this event by monsoonal changes at 4 kyr B.P. providing the first terrestrial record over the Indian subcon- Because of its uniquely large signature, in terms of both tinent, which confirms that climate changes at this time were magnitude and spatial scope, the climatic changes at this not restricted to regions influenced by the westerlies but also time bring about a number of questions regarding the global to monsoonal domains. Previous paleoclimate work in the climate at this period. Ding and Wang [2005] and Krishnan Indus Valley has relied heavily on reconstructions of the flow et al. [2009] have shown that changes in the ISM can from the Indus River, which integrates both summer and influence much of the Northern Hemisphere by producing winter precipitation [Staubwasser et al., 2003]. It is hypo- a Rossby wave dispersion pattern that affects the location thesized that changes in river flow would provide the most and strength of the midlatitude westerlies. The change in tangible link between climate and societal change in this midlatitude westerlies, in turn, enhances the intrusion of dry region [Clift et al., 2012; Staubwasser and Weiss, 2006]. extratropical air into the monsoon system, which further Recent studies of sediment provenance suggest that signifi- suppresses the monsoon. In this way, a positive feedback cant changes in the drainage pattern of the major Ghaggar- loop between the monsoon and westerlies occurs (i.e., weaker Hakra river system (i.e., the loss of Himalayan snowmelt) monsoon ⇒ change in westerlies ⇒ even weaker monsoon) significantly predate Indus Valley deurbanization [Clift et al., that can generate circumglobal pressure anomalies and pro- 2012]. These results imply that when the banks of the river vide a theoretical link between proxies that would be sensi- system were densely occupied 4000 years ago, the river was tive to monsoonal changes such as in NE India and those small (perhaps ephemeral) and fed principally by monsoonal records that would be influenced also by westerly flow such precipitation. Because the river had lost much of its Himala- as drainage from the Indus River or the ice core and peat yan drainage source, it would have had no buffer against records from the western United States. The trigger to initi- monsoonal precipitation changes. Therefore, the indepen- ate this feedback loop could either arise from a perturbation 84 ISM AT 4000 YEARS AGO of the westerlies from a high-latitude forcing (i.e., the North tions that begin with realistic 4 kyr B.P. boundary conditions Atlantic) [Bond et al., 1997; Hong et al., 2003] or from and are perturbed with a variety of forcings would be a useful dynamics within the tropics. The latter was discussed previ- starting point to produce hypotheses for the dynamical under- ously by Booth et al. [2005], who suggested that a strong pinnings of this event. Similar efforts have been successfully SST gradient in the eastern Tropical Pacific coupled with undertaken to understand potential mechanisms that lead to anomalous warmth in the IO could produce global climate the 8.2 kyr event [LeGrande et al., 2006]. Ultimately, the anomalies akin to those experienced between 1998 and 2002 model simulations would need to be either systematically (i.e., The Perfect Ocean for Drought)[Hoerling and Kumar, linked offline to individual proxy process models or simula- 2003]. However, the challenge in applying the dynamics tions that include tracers would be needed to place the di- described explicitly by either Krishnan et al. [2009] or verse proxies into common quantitative climate terms. This Hoerling and Kumar [2003] toward an understanding of the technique has been effective in understanding the complex changes at 4000 years B.P. is that while these feedback signal across the Last Glacial Maximum monsoon spe- operate effectively on short timescales, it has never been leothem proxies [Pausata et al., 2011] and could help in this tested as to whether they could produce an anomalous cli- context to shed light on why the signature of the 4 kyr B.P. mate mode for decades to centuries. signal in Omani [Fleitmann et al., 2003], Chinese [Wang et Despite what appears to be growing evidence for a wide- al., 2005], Middle Eastern [Bar-Matthews et al., 1999], and spread climate event at this time, it also must be acknowl- Indian (this study) speleothem records have such distinct edged that a number of locations, which theoretically should differences. This effort is needed before the proxy records show a signature of this event, appear complacent. For ex- can be successfully operationalized by archaeologists in such ample, the Qunf cave record from Oman [Fleitmann et al., a way as to make a substantive comment on the role that 2003] and the Dongge record from China [Wang et al., hydrological changes played in the cultural changes across 2005], which are both touted as proxies of the ISM, lack a the Middle East, India, and North Africa at this time [Weiss signature that is congruous with the magnitude of change et al., 1993]. observed in NE India. It may be that because neither record is situated directly in the monsoon domain, they lack the sen- APPENDIX A: sitivity to the ISM of the NE India record. Further, with SUPPLEMENTARY MATERIAL respect to an arid region such as Oman, the presence of an overlooked hiatus or period of very slow growth may plau- Supplementary data (M. Berkelhammer et al., sibly explain the absence of a 4 kyr B.P. excursion. It is also Mawmluh Cave, India Holocene Oxygen Isotope Data, clear both in the paleo [Sinha et al., 2011b] and modern http://www.ncdc.noaa.gov/paleo/speleothem.html, IGBP [Conroy and Overpeck, 2011] contexts that ISM precipita- PAGES/World Data Center for Paleoclimatology, Data Con- tion is characterized by complex spatial heterogeneity. For tribution Series 2010-011, NOAA/National Climatic Data example, NE India can exhibit antiphased behavior with Center Paleoclimatology Program, Boulder, Colorado, 2012) central India. In this respect, reliance on a single record from have been permanently archived and are freely accessible. 18 NE India to characterize the ISM is tenuous, and therefore, it Table 1 contains δ O data from a calcitic stalagmite (KM- is acknowledged that additional records from the Indian A) from Mawmluh Cave, located in Cherrapunji, Meghalaya continent are urgently needed to provide a conclusive picture in northeast India. Dating is based on eight U/Th series dates. of monsoonal changes at 4000 years B.P. Acknowledgments. The authors would like to thank Miguel Rincon for the assistance with the isotopic measurements and Brian 5. CONCLUSIONS Khapran-Daly, Robert Berkelhammer, and Kevin Cannariato for assistance with the field work. This work was supported, in part, There are some tempting hints that the climate system at by the National Science Foundation grants to A.S. (ATM: 0823554 4000 years B.P. may serve as a remarkable instance of a and AGS: 1103360) and the National Science Foundation of China large-scale, abrupt Holocene climate event. The event is grant to HC (41140016). particularly interesting because the climate at the time was, ’ in many ways, comparable to today s[Corrège et al., 2000]. REFERENCES However, there still exist a number of major impediments with respect to quantitatively understanding this climate Abram, N., H. McGregor, M. Gagan, W. Hantoro, and B. Suwargadi event. Much of the uncertainty stems from the fact that most (2009), Oscillations in the southern extent of the Indo-Pacific of the available proxy data, including the record presented Warm Pool during the mid-Holocene, Quat. Sci. Rev., 28(25–26), here, are only qualitative. General circulation model simula- 2794–2803. BERKELHAMMER ET AL. 85

Arz, H., F. Lamy, and J. Patzold (2006), A pronounced dry event Corrège, T., T. Delcroix, J. Récy, W. Beck, G. Cabioch, and F. Le recorded around 4.2 ka in brine sediments from the northern Red Cornec (2000), Evidence for stronger El Niño-Southern Oscilla- Sea, Quat. Res., 66(3), 432–441. tion (ENSO) events in a mid-Holocene massive coral, Paleocea- Ashok, K., Z. Y. Guan, N. H. Saji, and T. Yamagata (2004), Individual nography, 15(4), 465–470. and combined influences of ENSO and the Indian Ocean Dipole on Cullen, H., P. DeMenocal, S. Hemming, G. Hemming, F. Brown, T. the Indian summer monsoon, J. Clim., 17(16), 3141–3155. Guilderson, and F. Sirocko (2000), Climate change and the Bar-Matthews, M., A. Ayalon, A. Kaufman, and G. Wasserburg collapse of the Akkadian empire: Evidence from the deep sea, (1999), The Eastern Mediterranean paleoclimate as a reflection of Geology, 28(4), 379–382. regional events: Soreq cave, Israel, Earth Planet. Sci. Lett., 166 Dansgaard, W. (1964), Stable isotopes in precipitation, Tellus, 16, (1–2), 85–95. 436–468. Berkelhammer, M., A. Sinha, M. Mudelsee, H. Cheng, R. Edwards, Dayem, K., P. Molnar, D. Battisti, and G. Roe (2010), Lessons and K. Cannariato (2010), Persistent multidecadal power of the learned from oxygen isotopes in modern precipitation applied to Indian Summer Monsoon, Earth Planet. Sci. Lett., 290(1–2), interpretation of speleothem records of paleoclimate from eastern 166–172. Asia, Earth Planet. Sci. Lett., 295(1–2), 219–230. Bond, G., W. Showers, M. Cheseby, R. Lotti, P. Almasi, P. Deme- Ding, Q., and B. Wang (2005), Circumglobal teleconnection in the nocal, P. Priore, H. Cullen, I. Hajdas, and G. Bonani (1997), A Northern Hemisphere summer, J. Clim., 18(17), 3483–3505. pervasive millennial-scale cycle in North Atlantic Holocene and Drysdale, R., G. Zanchetta, J. Hellstrom, R. Maas, A. Fallick, glacial climates, Science, 278(5341), 1257–1266. M. Pickett, I. Cartwright, and L. Piccini (2006), Late Holocene Booth, R., S. Jackson, S. Forman, J. Kutzbach, E. Bettis, J. Kreigs, drought responsible for the collapse of Old World civilizations is and D. Wright (2005), A severe centennial-scale drought in recorded in an Italian cave flowstone, Geology, 34(2), 101–104. midcontinental North America 4200 years ago and apparent Enzel, Y., L. Ely, S. Mishra, R. Ramesh, R. Amit, B. Lazar, S. global linkages, The Holocene, 15(3), 321–328. Rajaguru, V. Baker, and A. Sandler (1999), A high-resolution Borgaonkar, H., A. Sikder, S. Ram, and G. Pant (2010), El Niño and Holocene environmental changes in the Thar Desert, north- related monsoon drought signals in 523-year-long ring width western India, Science, 284, 125–128. records of teak (Tectona grandis LF) trees from south India, Fasullo, J., and P. Webster (2003), A hydrological definition Palaeogeogr. Palaeoclimatol. Palaeoecol., 285(1–2), 74–84. of Indian monsoon onset and withdrawal, J. Clim., 16(19), Breitenbach, S. (2009), Changes in monsoonal precipitation and 3200–3211. atmospheric circulation during the Holocene reconstructed from Fleitmann, D., S. J. Burns, M. Mudelsee, U. Neff, J. Kramers, A. stalagmites from northeastern India, Ph.D. thesis, Dtsch. Geo- Mangini, and A. Matter (2003), Holocene forcing of the Indian ForschungsZentrum GFZ, Potsdam, Germany. monsoon recorded in a stalagmite from Southern Oman, Science, Breitenbach, S., J. Adkins, H. Meyer, N. Marwan, K. Kumar, and G. 300(5626), 1737–1739. Haug (2010), Strong influence of water vapor source dynamics Gasse, F. (2000), Hydrological changes in the African tropics since on stable isotopes in precipitation observed in Southern Megha- the Last Glacial Maximum, Quat. Sci. Rev., 19(1–5), 189–211. laya, NE India, Earth Planet. Sci. Lett., 292(1–2), 212–220. Graham, N., C. Ammann, D. Fleitmann, K. Cobb, and J. Luterba- Buckley, B., K. Palakit, K. Duangsathaporn, P. Sanguantham, and P. cher (2011), Support for global climate reorganization during the Prasomsin (2007), Decadal scale droughts over northwestern “Medieval Climate Anomaly”, Clim. Dyn., 37(5–6), 1217–1245. Thailand over the past 448 years: Links to the tropical Pacific Hoerling, M., and A. Kumar (2003), The perfect ocean for drought, and Indian Ocean sectors, Clim. Dyn., 29(1), 63–71. Science, 299(5607), 691–694. Buckley, B., K. Anchukaitis, D. Penny, R. Fletcher, E. Cook, M. Hong, Y., et al. (2003), Correlation between Indian Ocean summer Sano, L. Nam, A. Wichienkeeo, T. Minh, and T. Hong (2010), monsoon and North Atlantic climate during the Holocene, Earth Climate as a contributing factor in the demise of Angkor, Cam- Planet. Sci. Lett., 211(3–4), 371–380. – bodia, Proc. Natl. Acad. Sci. U. S. A., 107(15), 6748 6752. Hoyos, C. D., and P. J. Webster (2007), The role of intraseasonal Cheng, H., R. Edwards, W. Broecker, G. Denton, X. Kong, Y. variability in the nature of Asian monsoon precipitation, J. Clim., Wang, R. Zhang, and X. Wang (2009), Ice age terminations, 20(17), 4402–4424. – Science, 326(5950), 248 252. Johnson, K. R., and B. L. Ingram (2004), Spatial and temporal Clift, P., et al. (2012), U-Pb zircon dating evidence for a Pleistocene variability in the stable isotope systematics of modern precipita- Sarasvati River and capture of the Yamuna River, Geology, 40(3), tion in China: Implications for paleoclimate reconstructions, 211–214. Earth Planet. Sci. Lett., 220(3–4), 365–377. Conroy, J., and J. Overpeck (2011), Regionalization of present-day Krishnan, R., V. Kumar, M. Sugi, and J. Yoshimura (2009), Internal precipitation in the Greater Monsoon Region of Asia, J. Clim., 24 feedbacks from monsoon-midlatitude interactions during droughts (15), 4073–4095, doi:10.1175/2011JCLI4033.1. in the Indian summer monsoon, J. Atmos. Sci., 66(3), 553–578. Cook, E., K. Anchukaitis, B. Buckley, R. DArrigo, G. Jacoby, and Lee, J. E., and I. Fung (2008), “Amount effect” of water isotopes W. Wright (2010), Asian monsoon failure and megadrought and quantitative analysis of post-condensation processes, Hydrol. during the last millennium, Science, 328(5977), 486–489. Processes, 22(1), 1–8. 86 ISM AT 4000 YEARS AGO

LeGrande, A. N., and G. A. Schmidt (2009), Sources of Holocene Scholz, D., and D. Hoffmann (2011), StalAge-An algorithm variability of oxygen isotopes in paleoclimate archives, Clim. designed for construction of speleothem age models, Quat. Geo- Past, 5, 1133–1162. chronol., 6(3–4), 369–382. LeGrande, A. N., G. A. Schmidt, D. T. Shindell, C. V. Field, R. L. Sengupta, S., and A. Sarkar (2006), Stable isotope evidence of dual Miller, D. M. Koch, G. Faluvegi, and G. Hoffmann (2006), Con- (Arabian Sea and Bay of Bengal) vapour sources in monsoonal sistent simulations of multiple proxy responses to an abrupt climate precipitation over north India, Earth Planet. Sci. Lett., 250(3–4), change event, Proc. Natl. Acad. Sci. U. S. A., 103(4), 837–842. 511–521. Levermann, A., J. Schewe, V. Petoukhov, and H. Held (2009), Basic Singh, G., R. Wasson, and D. Agrawal (1990), Vegetational and mechanism for abrupt monsoon transitions, Proc. Natl. Acad. Sci. seasonal climatic changes since the last full glacial in the Thar U. S. A., 106(49), 20,572–20,577. Desert, northwestern India, Rev. Palaeobot. Palynol., 64(1–4), Loschnigg, J., and P. Webster (2000), A coupled ocean-atmosphere 351–358. system of SST modulation for the Indian Ocean, J. Clim., 13(19), Sinha, A., K. G. Cannariato, L. D. Stott, H. C. Li, C. F. You, H. 3342–3360. Cheng, R. L. Edwards, and I. B. Singh (2005), Variability of Madella, M., and D. Fuller (2006), Palaeoecology and the Harappan Southwest Indian summer monsoon precipitation during the civilisation of South Asia: A reconsideration, Quat. Sci. Rev., 25 Boiling-Allerod, Geology, 33(10), 813–816. (11–12), 1283–1301. Sinha, A., K. G. Cannariato, L. D. Stott, H. Cheng, R. L. Edwards, Meehl, G. (1993), A coupled air-sea biennial mechanism in the M. G. Yadava, R. Ramesh, and I. B. Singh (2007), A 900-year tropical Indian and Pacific regions: Role of the ocean, J. Clim., (600 to 1500 A.D.) record of the Indian summer monsoon pre- 6,31–41. cipitation from the core monsoon zone of India, Geophys. Res. Meehl, G., and J. Arblaster (2002), The tropospheric biennial os- Lett., 34, L16707, doi:10.1029/2007GL030431. cillation and Asian-Australian monsoon rainfall, J. Clim., 15(7), Sinha, A., M. Berkelhammer, L. Stott, M. Mudelsee, H. Cheng, 722–744. and J. Biswas (2011a), The leading mode of Indian Summer Mischke, S., and C. Zhang (2010), Holocene cold events on the Monsoon precipitation variability during the last millennium, Tibetan Plateau, Global Planet. Change, 72(3), 155–163. Geophys. Res. Lett., 38, L15703, doi:10.1029/2011GL047713. Misra, V. (1984), Climate, a factor in the rise and fall of the Indus Sinha, A., L. Stott, M. Berkelhammer, H. Cheng, R. Edwards, B. Civilization: Evidence from Rajasthan and beyond, in Frontiers Buckley, M. Aldenderfer, and M. Mudelsee (2011b), A global of the Indus Civilization, edited by B. B. Lal and S. P. Gupta, context for megadroughts in monsoon Asia during the past mil- pp. 461–490, Books & Books, New Delhi, India. lennium, Quat. Sci. Rev., 30,47–62. Moy, C., G. Seltzer, D. Robdell, and D. Anderson (2002), Variabil- Staubwasser, M., and H. Weiss (2006), Holocene climate and cul- ity of El Niño/Southern Oscillation activity at millennial time- tural evolution in late prehistoric–early historic West Asia, Quat. scales during the Holocene epoch, Nature, 420, 162–165, doi:10. Res., 66(3), 372–387. 1038/nature01194. Staubwasser, M., F. Sirocko, P. M. Grootes, and M. Segl (2003), Murata, F., T. Hayashi, J. Matsumoto, and H. Asada (2007), Climate change at the 4.2 ka BP termination of the Indus Rainfall on the Meghalaya plateau in northeastern India: One valley civilization and Holocene south Asian monsoon vari- of the rainiest places in the world, Nat. Hazards, 42(2), ability, Geophys. Res. Lett., 30(8), 1425, doi:10.1029/2002GL 391–399. 016822. Pausata, F., D. Battisti, K. Nisancioglu, and C. Bitz (2011), Thompson, L., et al. (2002), Kilimanjaro ice core records: Evidence Chinese stalagmite δ18O controlled by changes in the Indian of Holocene climate change in tropical Africa, Science, 298 monsoon during a simulated Heinrich event, Nat. Geosci., 4(7), (5593), 589–593. 474–480. Vuille, M., M. Werner, R. S. Bradley, and F. Keimig (2005), Rashid, H., E. England, L. Thompson, and L. Polyak (2011), Late Stable isotopes in precipitation in the Asian monsoon region, glacial to Holocene Indian summer monsoon variability based J. Geophys. Res., 110, D23108, doi:10.1029/2005JD006022. upon sediment, Terr. Atmos. Ocean. Sci., 22(2), 215–228. Wagner, B., A. Lotter, N. Nowaczyk, J. Reed, A. Schwalb, R. Rijsdijk, K., et al. (2011), Mid-Holocene (4200 kyr BP) mass Sulpizio, V. Valsecchi, M. Wessels, and G. Zanchetta (2009), mortalities in Mauritius (Mascarenes): Insular vertebrates resil- A 40,000-year record of environmental change from ancient ient to climatic extremes but vulnerable to human impact, The Lake Ohrid (Albania and Macedonia), J. Paleolimnol., 41(3), Holocene, 21(8), 1179–1194. 407–430. Sano, M., B. Buckley, and T. Sweda (2009), Tree-ring based hydro- Wang, Y. J., H. Cheng, R. L. Edwards, Z. S. An, J. Y. Wu, C. C. climate reconstruction over northern Vietnam from Fokienia Shen, and J. A. Dorale (2001), A high-resolution absolute-dated hodginsii: Eighteenth century mega-drought and tropical Pacific Late Pleistocene monsoon record from Hulu Cave, China, influence, Clim. Dyn., 33(2), 331–340. Science, 294(5550), 2345–2348. Schewe, J., A. Levermann, and H. Cheng (2011), A critical humid- Wang, Y. J., H. Cheng, R. L. Edwards, Y. Q. He, X. G. Kong, ity threshold for monsoon transitions, Clim. Past Discuss., 7, Z. S. An, J. Y. Wu, M. J. Kelly, C. A. Dykoski, and X. D. 1737–1765. Li (2005), The Holocene Asian monsoon: Links to solar BERKELHAMMER ET AL. 87

changes and North Atlantic climate, Science, 308(5723), Zhang, P. Z., et al. (2008), A test of climate, sun, and culture 854–857. relationships from an 1810-year Chinese cave record, Science, Webster, P. (1987), The elementary monsoon, in Monsoon, 322(5903), 940–942. edited by J. Fein and P. Stephens, pp. 3–32, John Wiley, New York. M. Berkelhammer, Cooperative Institute for Research in Weiss, H. (2012), Altered trajectories: The intermediate Bronze Environmental Sciences, University of Colorado, Boulder, CO Age, in The Oxford Handbook of the Archaeology of the Levant, 80309, USA. ([email protected]) edited by A. Killebrew and M. Steiner, Oxford Univ. Press, H. Cheng, Institute of Global Environmental Change, Xian Oxford, U. K., in press. Jiaotong University, Shanghai, China. Weiss, H., M. Courty, W. Wetterstrom, F. Guichard, L. Senior, R. F. S. R. Pausata, Bjerknes Center for Climate Research, Bergen, Meadow, and A. Curnow (1993), The genesis and collapse of Norway. third millennium north Mesopotamian civilization, Science, 261 A. Sinha, Department of Earth Sciences, California State (5124), 995–1004. University, Dominguez Hills, Carson, CA, USA. Yoshimura, K., M. Kanamitsu, D. Noone, and T. Oki (2008), L. Stott, Department of Earth Sciences, University of Southern Historical isotope simulation using Reanalysis atmospheric California, Los Angeles, CA, USA. data, J. Geophys. Res., 113, D19108, doi:10.1029/2008JD K. Yoshimura, Atmosphere and Ocean Research Institute, 010074. University of Tokyo, Tokyo, Japan. Late Holocene Drought Pattern Over West Asia

Michael Staubwasser

Institute of Geology and Mineralogy, University of Cologne, Cologne, Germany

A 5000 year long temperature-corrected δ18O record from planktonic foraminifera off shore from the Indus river delta reconstructs fluvial discharge variability that closely mimics a record of northern Red Sea water column stratification, derived from the difference in δ18O between benthic and planktonic foraminifera. Multi- centennial drought (pluvial) conditions over the Indus region correspond to high (low) stratification in the northern Red Sea. The well-described present-day pattern of atmospheric subsidence and uplift over west Asia provides a probable physical mechanism responsible for the observed correlation of the two widely separated paleoclimate records. The pattern is known to respond directly to the upper tropospheric thermal structure in both seasons of the South Asian monsoon domain and exerts a fundamental control of northern Red Sea turnover through large-scale eastern Mediterranean advection and subsidence of cold air from the north.

1. INTRODUCTION methods, may compromise the establishment of true coincidence of climate events between two different regions or The Middle East and the Indian subcontinent have under- mask true antiphasing and lead-lag causality. As a conse- gone considerable climate change during the Holocene quence, a synoptic-scale pattern of climate change recon- [Staubwasser and Weiss, 2006]. Several century-long struction for recognized climatic events is currently missing droughts have occurred across much of the region over the for the region. If the underlying mechanism responsible for last 6000 years and appear to have coincided with cultural climate change and the regional response pattern were changes. However, even pronounced events, such as the known, the likelihood of causality between seemingly coin- 4.2 ka B.P. drought of west and South Asia, are sometimes cident climate and cultural change across the region may be difficult to detect within paleoclimate proxy records. Besides assessed more convincingly even in the light of slight chro- the more technical aspect of proxy reliability, this raises the nological uncertainty. question whether the same event even occurred everywhere. Paleoclimatic and paleoceanographic reconstruction on Another well-known problem is dating accuracy. If recorded large spatial scales, despite dating uncertainty, have been at all, Holocene climate events often lasted only a few most successful when analyzing a repetitive climate change centuries, a length of time often comparable to radiocarbon pattern. Prominent examples are the linking of MIS 3 marine dating uncertainty. Even despite high analytical precision, ice-rafting events to the Dansgaard-Oeschger oscillations chronological accuracy is always difficult to verify, e.g., observed in Greenland ice cores [Bond et al., 1993] through- 14 because of uncertain C reservoir corrections or smoothing out marine isotope stage 3. In the study presented here, a of the sediment age (and proxy signal) due to bioturbation. correlated repetitive pattern of multicentennial climate These, and comparable corrections applied to other dating change is shown between two distant regions across west Asia for the last 5000 years. A reconstruction of Indus River outflow into the northern Arabian Sea is derived from a δ18 Climates, Landscapes, and Civilizations temperature-corrected O record of planktonic foraminif- Geophysical Monograph Series 198 era Globigerinoides ruber. This is compared to a water © 2012. American Geophysical Union. All Rights Reserved. column stratification record from the Gulf of Aqaba, the 10.1029/2012GM001223 northernmost extension of the Red Sea, derived from δ18O 89 90 LATE HOLOCENE DROUGHT PATTERN OVER WEST ASIA difference between planktonic and benthic foraminifera [Arz seasonal flow reversals generally of a much larger velocity et al., 2003]. Coincident climate change across the entire (and volume) than the apparent average seasonal flow [Via- region is discussed within the frameworks of the well de- lard et al., 2009; Shenoi, 2010]. scribed present-day pattern of large-scale vertical flow in the troposphere [Rodwell and Hoskins, 1996]. Finally, regions 2.2. Validation with a high probability for a causal relationship between climate events and cultural change are defined. At the present day, the effect on surface ocean δ18O changes due to Indus River discharge will unfortunately be 2. ESTIMATION OF INDUS RIVER DISCHARGE impossible to measure directly because of the construction of major dams on the river Indus and substantial reduction in 2.1. General Approach outflow [Milliman et al., 1984]. A number of factors add considerable complexity to the climatic meaning of δ18Oin Stable oxygen isotope ratios of calcite, such as tests of the Arabian Sea mixed layer beside simple dilution from G. ruber, depend on the temperature and the isotopic value of Indus outflow, but can, to a reasonable degree, be assumed the ambient water during precipitation [Epstein et al., 1953; to have remained relatively constant over the last 5000 years. Erez and Luz, 1983]. In ocean regions with high freshwater These are (1) the seasonality distribution of the planktonic discharge from a river with a continental, i.e., low δ18O value foraminifera species G. ruber, (2) the foraminifera depth such as the northeastern Arabian Sea and Indus River, mixed habitat, (3) evaporation from the northern Arabian Sea, layer seawater δ18O will be affected by the discharge plume (4) the mixed layer depth, (5) the carbonate ion concentration and vary in response to the amount of discharge. The shallow of surface seawater, and (6) the δ18O composition of Indus dwelling G. ruber [Curry et al., 1992] is ideally suited to water. Factors 1 to 5 fall largely under the control of seasonal record the δ18O of the mixed layer. If the sea surface tem- changes due to the South Asian monsoon winds [Curry et al., perature (SST) is known, calculation of ambient seawater 1992; Schott et al., 2009]. The most potent influence, in δ18O of the past from δ18O measured in G. ruber is possible principle, comes from variations in the mixed layer depth, in principle, and the variations observed may be interpreted which ranges from under 20 m in summer and up to 120 m in as changes in river outflow. This approach has been applied winter across the entire northern Arabian Sea [Kumar and for the late Holocene by Doose-Rolinski et al. [2001], on Prasad, 1999]. In effect, the annual mixed layer deepening northern Arabian Sea core 56KA using alkenone-derived buffers changes in surface water δ18O due to evaporation. SST and a correlation of δ18O with salinity (for a thorough Despite a significant influence of the mixed layer dynamics discussion of the application of the proxy in this high SST on present-day hydrographic properties across the entire environment see Doose-Rolinski et al. [2001]). These northern Arabian Sea, the eastern and western parts of the authorsassumednosignificant influence of Indus River basin record opposite trends in G. ruber δ18O across the discharge at their core site and interpreted their calculated 4.2 ka B.P. drought event [Staubwasser et al., 2003]. The car- sea surface salinity changes exclusively as the result of bonate system is also unlikely to have changed over the last changing evaporation. However, the potential impact of In- 5000 years because of relatively constant late Holocene 14C dus discharge on the northern Arabian Sea oxygen isotopic reservoir ages [Staubwasser et al., 2002, 2003]. In general, composition can be demonstrated by a simple mass balance large changes in South Asian monsoon winds occurred dur- calculation. The recorded annual Indus discharge volume ing the early and mid-Holocene [Sirocko et al., 1993; Gupta prior to dam construction was 208 km3, mostly concentrated et al., 2003], and trends during the last 5000 years recorded over the summer monsoon season, with an interannual var- in the monsoon wind-dominated upwelling regime off Oman iability of up to 50% [Milliman et al., 1984]. The difference are not observed in the northern Arabian Sea [Staubwasser, in δ18O between the Indus River and the Arabian Sea is about 2006]. These arguments support the notion to attribute any 12‰ [Delaygue et al., 2001; Karim and Veizer, 2002]. As- changes in G. ruber δ18O unrelated to sea surface variations suming homogeneous mixing into a 25 m mixed layer over primarily to Indus River discharge rather than factors 1 to 5 the northeastern Arabian Sea quadrant north of the latitude of listed above. Mumbai, ~18° N, and east of 65° E, the resulting interannual Although the majority of discharge takes place in the change of mixed layer δ18O would amount to 0.28‰. Even summer monsoon, the δ18O of Indus River water is heavily though the general circulation regime is described as clock- affected by melting snow (of very low δ18O) in the high wise around the Arabian Sea, and thus would potentially altitude regions of the Karakoram and the Ladakh [Karim steer the discharge plume away from the core sites, a wide and Veizer, 2002]. Much of the snow is from winter precip- dispersal of the plume is ensured because of frequent intra- itation [Wake, 1989]. However, available reconstruction of STAUBWASSER 91

Figure 1. Comparison of proxy records including original data (thin lines) and 210 years smoothed data (heavy lines). (a) The 56KA Alkenone SST, northern Arabian Sea [Doose-Rolinski et al., 2001]. (b) The 63KA Globigerinoides ruber δ18O, 18 18 northern Arabian Sea [Staubwasser et al., 2003], and calculated temperature-only effect (δ Ocalcite-δ Owater, see next 18 plot). (c) Calculated δ Owater of the northern Arabian Sea with interpretation (calculation for smoothed data only and 18 18 using the equation T (°C) = 14.9 À 4.8 (δ Ocalcite À δ Owater)byBemis et al. [1998]). (d) Comparison of Figures 1c and 1e, with Figure 1e shifted by À80 years. Stippled lines are for visualization of the À80 year shift of the Red Sea record; the chronologic uncertainty of calibrated 14C ages is in the range of ±40–150 years. (e) Gulf of Aqaba (northern Red Sea) 18 stratification, δ Oplankton-benthos with interpretation [Arz et al., 2003]. 92 LATE HOLOCENE DROUGHT PATTERN OVER WEST ASIA glacier advances for the Holocene suggest comparably stable ern Red Sea water flows into the Gulf and during wintertime glaciologic conditions in the headwater regions of the Indus cooling, deep convection leads to near or even full water and its tributaries during the last 5000 years [Phillips et al., column homogeneity [Wolf-Vecht et al., 1992]. Measure- 2000]. In summary, to the best of current knowledge, Indus ments of recently growing corals in the Gulf show a seasonal River discharge from stable oxygen isotopes in planktonic variation of oxygen isotopes of over 1‰ in δ18O and reflect foraminifera rests largely on quantifying the temperature wintertime cooling and seasonal overturning [Felis et al., aspect of the oxygen isotope equation. 1998]. Eshel et al. [2000] compared a thermodynamic analysis of Red Sea meteorologic and hydrographic observation- 18 2.3. Synchronizing Arabian Sea Sediment Core Chronologies al data to a ~20 year δ O record of a recent coral. They were 18 and Calculation of δ Owater in the Arabian Sea Mixed Layer able to demonstrate the coincidence of anomalously cold and dry winter conditions throughout the northern Red Sea with This reconstruction of Arabian Sea surface δ18O is done by anomalously high wintertime δ18O values in the coral (re- using high-resolution data from two cores from the northern presenting cold and saline conditions) and recorded times of Arabian Sea continental slope, the 56KA alkenone-derived deep convection in 2 years during the mid-1970s to 1990s. SST record by Doose-Rolinski et al. [2001] and the 63KA The origin of anomalous conditions was an enhanced cold air δ18O record on G. ruber of Staubwasser et al. [2003]. The advection across the lower and middle troposphere from the chronology of core 56KA is based on counted seasonal north, subsidence over the Red Sea and, subsequently, en- couplets of sediment layers, which are not sufficiently re- hanced evaporation as a result of descending dry air. A solvable in 63KA. To synchronize records, the two cores are similar conclusion was reached by Felis et al. [2000] with correlated on the basis of radiocarbon ages, with the 56KA a longer record. Transferring the observation on recent corals data recalibrated using the same reservoir correction as in to the 5000 year foraminifer record is straight forward: Times 63KA (565 years). In order to minimize noise, both records of strong or expanded winter convection in the Gulf of were smoothed with a 210 years sliding window in the time Aqaba lead to longer intervals with a homogenous water domain. Because of the irregular spacing, a result of small column, thus reducing the difference of δ18O between sur- variations in sedimentation rate or simply lack of foraminifer face- and bottom-dwelling foraminifera. specimen, the smoothing averages between 5 and 25 data The basin water stratification on the seasonal scale corre- points at any time in each of the records. The calculation of sponds to air temperature, and strong convection is triggered mixed layer δ18O is done using the equation established for by extended winter cooling [Wolf-Vecht et al., 1992]. Ex- G. ruber [Bemis et al., 1998; Russell and Spero, 2000]. The change between the 1830 m deep Gulf of Aqaba and the Red smoothed original data are plotted on top of the measured Sea is restricted by a sill of 252 m depth. Because of the data together with the calculated Arabian Sea δ18O on the basin’s geometry and deep overturning in winter, salinity 63KA time scale (Figures 1a–1c). varies very little with depth, and water column stratification responds very sensitively to surface temperature changes. On 3. GULF OF AQABA WATER COLUMN the decadal scale of the foram record, temperature is also the STRATIFICATION more likely controlling parameter. At 21°C–26°C water temperature, very large salinity changes are required to change The proxy for water column stratification in the Gulf of seawater density significantly. Aqaba is a composite of planktonic and benthic foraminifera oxygen isotopes measured in core GEOB-5804-4 [Arz et al., 4. PACING OF CLIMATE CHANGE IN WEST ASIA 2003]. Oxygen isotopes reflect SST and salinity and there- OVER THE LAST 5000 YEARS fore ambient water density. The difference between oxygen isotope ratios of surface planktonic foraminifera G. ruber The original foraminifera δ18O data, calculated northern 18 and benthic foraminifera Cibicidoides mabahethi is a mea- Arabian Sea δ Owater, and smoothed records for both cores sure of density difference between surface and deep water, are shown together in Figure 1. A common pacing of change averaged over several decades. Because there is virtually no between Indus outflow and Gulf of Aqaba stratification is freshwater inflow into the northern Red Sea, salinity is di- clearly observable. Times of drought in the Indus region, rectly controlled by evaporation. Total annual evaporation of such as after 4200 years ago, correspond to intervals of over 2.5 m and winter cooling results in the establishment of enhanced stratification in the Gulf of Aqaba. Moister condi- a thermohaline overturning cell with deep water formation in tions in the Indus region correspond to enhanced water the northern Red Sea [Eshel and Naik, 1997]. The Gulf of column turnover. The direct conclusion from this 5000 year Aqaba essentially behaves similarly. During summer, north- coherence is that these seemingly very different climate STAUBWASSER 93 parameters from two spatially very separate regions appear Positive eastern Mediterranean rainfall anomalies occur at to share the same underlying mechanism responsible for the times when anomalous southerly flow and uplift cancels the quasi-cyclic change pattern throughout the last 5000 years. overall subsidence [Eshel, 2002]. Over the Levant, however, The best match is achieved when the northern Red Sea rainfall occurs when the midtropospheric pressure field fa- stratification record is shifted by À80 years. This would be vors cold advection over the eastern Mediterranean [Ziv et the equivalent of increasing the 14C reservoir age by 80 al., 2006]. Drought occurs when southwesterly winds im- years, which is within the typical uncertainty range of such pede cold advection. This may explain the coincidence of estimates. The amount of the shift introduced to the Red Sea high Red Sea stratification (absence of subsidence) and record appears more reasonable than lowering the upwelling- drought east of the Mediterranean, as recorded by the Dead dominated Arabian Sea 14C reservoir age by the same Sea level changes between 4200 and 3300 years ago [Mi- amount. Alternatively, the shift could be the result of biotur- gowski et al., 2006]. However, the correlation between Le- bation at the Red Sea core site (the Arabian Sea sediments vant rainfall and northern Red Sea δ18O as recorded in a are laminated). Bioturbation would smear out and mix with subrecent ~250 year coral record is statistically weak and ambient material any event recorded by tests of foraminifera apparently nonstationary [Felis et al., 2000; Rimbu et al., deposited on the sediment surface, thus changing the event’s 2006]. Reduced subsidence should not necessarily have re- average sediment age. Regardless, the applied shift is signif- sulted in drought over the eastern Mediterranean at 4200 icantly lower than the average 700 year period of the quasi- years B.P. More records with very thorough dating will be cyclic pattern [Staubwasser et al., 2003] and largely within required to test, if the present-day thermodynamic mechan- dating precision of ±40 – 150 years. Thus, the correlation as isms leading to drought are valid also for the past of that shown appears to be robust with respect to the applied 80 region. years offset in chronologies. The large-scale subsidence/uplift pattern over the eastern It is currently not possible to discriminate between a chro- Mediterranean and west Asia is the result of high tropospher- nologic inaccuracy and a true lead-lag relationship as cause ic Rossby waves, which originate from thermal anomalies, for the apparent 80 year offset between the two records. such as the top of the summer monsoon convection cell in the Taken at face value, the apparent shift would mean that the Bay of Bengal, or the snow and ice covered central Asian changing Red Sea stratification led a shift in the Asian high altitudes. The Rossby waves may propagate against the monsoon domain. This is not easy to accommodate physi- wind, depending on wind speed and latitude [Rodwell and cally, given that the Asian monsoon is one of the largest Hoskins, 1996; Gordon et al., 1998] across west Asia where climate systems on Earth. There are two general aspects to they induce an alternating pattern of subsidence and uplift discuss, in principle, as possible causes for leads and lags. (Figure 2, not shown for summer). The coherence of Red Sea Thresholds inherent in the system may cause the analyzed stratification (or eastern Mediterranean/Red Sea subsidence) paleoclimate parameter, i.e., the δ18O of Indus water (con- with humidity in the Indus region (Figure 1d) suggests a trolled by the ratio of winter to summer precipitation) to be regionwide response to changes in the subsidence and uplift transferred into foraminifera δ18O (recording Arabian Sea pattern. A detailed thermodynamic analysis as for the eastern mixed-layer properties) with a delay. The other aspect is Mediterranean is not available for the Indus region, but a large-scale meteorologic causes. These, however, must be direct winter/spring influence is likely because present-day plausible with respect to the underlying atmospheric physical Indus outflow and its isotopic composition responds to win- causes of drought and pluvial in the region. Currently, the ter and spring snowfall in the high-altitude headwater region influence of North Atlantic Oscillation (NAO) and El Niño– [Karim and Veizer, 2002]. The winter/spring subsidence cell Southern Oscillation (ENSO) are discussed, but correlation over the eastern Mediterranean/northern Red Sea corre- coefficients are small [Ziv et al., 2006]. The best correlation sponds to a large cell of uplift over the Indus headwater was found with polar stratospheric jets and western Pacific region (Figure 2). This may provide a physical explanation sea surface temperatures. An answer to these questions may for the observed coherence of Red Sea stratification and be provided by future numeric simulations. Indus outflow of this study and of coherent drought in the Levant and Indus regions at 4200 years ago. 4.1. Subsidence and Drought 4.2. Mid to Late Holocene Climate and Cultural Change, The analysis of Eshel and Farrell [2000] suggests that the Open Questions general winter and early spring subsidence pattern over the Red Sea, which leads to enhanced turnover in the northern Based on the causality between subsidence and drought at Red Sea, extends across the entire eastern Mediterranean. the margins of west Asia, the cause of late Holocene drought 94 LATE HOLOCENE DROUGHT PATTERN OVER WEST ASIA

Figure 2. The 500 hPa vertical winds for winter (December–February) and spring (March–May) [Kalnay et al., 1996]. Positive numbers represent downward flow. Stars indicate the location of core sites 56KA and 63KA, both close to each other in the Arabian Sea, and GEOB-5804-4 in the Gulf of Aqaba. events, in general, and elsewhere in the region may be linked subsidence/uplift or in the magnitude of vertical flow. In to expanding or shifting cells of subsidence. Sites with a addition to the hemispheric-scale systems affecting local clear record of drought at 4200 years B.P. in west Asia, such climate (NAO, ENSO, polar jets, Pacific SST, see discussion as the Dead Sea and Levant [Bar-Matthews and Ayalon, above), a regional-scale monsoon dominated control on the 2011; Migowski et al., 2006], northern Mesopotamia [Cullen west Asian Rossby wave pattern by the extent of central et al., 2000; Weiss, 1993a, 1993b], eastern and central Tur- Asian high-altitude glaciation has been convincingly dem- key [Lemcke and Sturm, 1997; Kuzucuoglu et al., 2011], and onstrated in numerical sensitivity studies for the mid to late the source region of Indus River [Staubwasser et al., 2003] Holocene transition [Bush, 2000]. The anomalous wave pat- appear to be situated inside cells of present-day uplift or near tern following an expanded cold anomaly above a prescribed the boundaries between cells of subsidence and uplift (Figure 2). glaciated Himalaya and Karakoram about 6–5kabefore Sites in the vicinity of these cell boundaries would be gen- [Phillips et al., 2000] led to significant changes in the tropo- erally more susceptible to changes in the spatial pattern of spheric subsidence and uplift pattern between the Himalayas STAUBWASSER 95 and the eastern Mediterranean. It remains to be tested which Bush, A. B. G. (2000), A positive climatic feedback mechanism for of the above causes led to the variation in subsidence and Himalayan glaciation, Quat. Int., 65–66,3–13. uplift recorded during the last 5000 years. Cullen, H. M., P. B. deMenocal, S. Hemming, G. Hemming, F. H. There is abundant archaeological evidence for major cul- Brown, T. Guilderson, and F. Sirocko (2000), Climate change tural changes at 4200 years ago, particularly in the Indus and the collapse of the Akkadian empire: Evidence from the deep – region, Mesopotamia, and Palestine [cf. Staubwasser and sea, Geology, 28(4), 379 382. Weiss, 2006]. The Red Sea–Indus correlation presented here Curry, W. B., D. R. Ostermann, M. V. S. Guptha, and V. Ittekkot (1992), Foraminiferal production and monsoonal upwelling in suggests that this event may be an extreme manifestation of a the Arabian Sea: Evidence from sediment traps, in Upwelling time-repetitive pattern. However, the underlying spatial pat- Systems: Evolution Since the Early Miocene, edited by C. P. tern of climate change suggests that this change may not be Summerhayes, W. L. Prell and K. C. Emeis, Geol. Soc. Spec. recorded everywhere in the region. While the Indus culture Publ., 64,93–106. itself was centered along the Indus River plain, well inside a Delaygue, G., E. Bard, C. Rollion, J. Jouzel, M. Stiévenard, J.-C. region of dominating subsidence, the source region of Indus Duplessy, and G. Ganssen (2001), Oxygen isotope/salinity rela- water lies in the mountain ranges to the northeast, at the tionship in the northern Indian Ocean, J. Geophys. Res., 106(C3), margin of a center of present-day uplift. Sites under centers 4565–4574. of cells of subsidence where water availability responds Doose-Rolinski, H., U. Rogalla, G. Scheeder, A. Lückge, and U. directly to local atmospheric conditions may be more com- von Rad (2001), High-resolution temperature and evaporation placent to change or may not even record climate change at changes during the Late Holocene in the northeastern Arabian – all. From regions of overall subsidence, a severe drought at Sea, Paleoceanography, 16(4), 358 367. 4200 B.P. is so far only recorded at the eastern Persian Gulf Epstein, S., R. Buchsbaum, H. A. Lowenstam, and H. C. Urey [Parker et al., 2004, 2006]. (1953), Revised carbonate-water isotopic temperature scale, Geol. Soc. Am. Bull., 64(11), 1315–1326. The question of causality between cultural change and Erez, J., and B. Luz (1983), Experimental paleotemperature equa- climate events is often claimed on the basis of far-field tion for planktonic foraminifera, Geochim. Cosmochim. Acta, correlations blurred by chronological uncertainty. In the 47(6), 1025–1031. absence of complementary archives from the same locality, Eshel, G. (2002), Mediterranean climates, Isr. J. Earth Sci., 51, such potential causality may be addressed more convinc- 157–168. ingly when supported by the regionwide climate change Eshel, G., and B. F. Farrell (2000), Mechanisms of eastern pattern suggested here. On the basis of a coherent, physi- Mediterranean rainfall variability, J. Atmos. Sci., 57(19), cally plausible regional climate pattern of the past, con- 3219–3232. firmed by climate modeling, it may become possible to Eshel, G., and N. H. Naik (1997), Climatological coastal jet colli- model the adaptation of ancient agricultural societies in sion, intermediate water formation, and the general circulation of direct response to climate change or, when unable to adapt, the Red Sea, J. Phys. Oceanogr., 27(7), 1233–1257. model the habitat tracking of such societies as a result of Eshel, G., D. P. Schrag, and B. F. Farrell (2000), Troposphere- climate change. planetary boundary layer interactions and the evolution of ocean surface density: Lessons from Red Sea corals, J. Clim., 13, REFERENCES 339–351. Felis, T., J. Pätzold, Y. Loya, and G. Wefer (1998), Vertical water Arz, H. W., F. Lamy, J. Pätzold, P. J. Müller, and M. Prins (2003), mass mixing and plankton blooms recorded in skeletal stable Mediterranean moisture source for an early-Holocene humid carbon isotopes of a Red Sea coral, J. Geophys. Res., 103(C13), period in the northern Red Sea, Science, 300,118–121. 30,731–30,739. Bar-Matthews, M., and A. Ayalon (2011), Mid-Holocene climate Felis, T., J. Pätzold, Y. Loya, M. Fine, A. H. Nawar, and G. Wefer variations revealed by high-resolution speleothem records from (2000), A coral oxygen isotope record from the northern Red Sea Soreq Cave, Israel and their correlation with cultural changes, documenting NAO, ENSO, and North Pacific teleconnections on The Holocene, 21, 161–173. Middle East climate variability since the year 1750, Paleoceano- Bemis, B. E., H. J. Spero, J. Bijma, and D. W. Lea (1998), Reeval- graphy, 15(6), 679–694. uation of the oxygen isotopic composition of planktonic forami- Gordon, A., W. Grace, P. Schwerdtfeger, and R. Byron-Scott nifera: Experimental results and revised paleotemperature (1998), Dynamic Meteorology: A Basic Course, John Wiley, equations, Paleoceanography, 13(2), 150–160. New York. Bond, G., W. Broecker, S. Johnsen, J. Mcmanus, L. Labeyrie, J. Gupta, A. K., D. M. Anderson, and J. T. Overpeck (2003), Abrupt Jouzel, and G. Bonani (1993), Correlations between climate changes in the Asian southwest monsoon during the Holocene records from North Atlantic sediments and Greenland ice, and their links to the North Atlantic Ocean, Nature, 421, Nature, 365, 143–147. 354–357. 96 LATE HOLOCENE DROUGHT PATTERN OVER WEST ASIA

Kalnay, E., M. Kanamitsu, and R. Kistler (1996), The NCEP/NCAR Schott, F. A., S.-P. Xie, and J. P. McCreary Jr. (2009), Indian Ocean 40-year reanalysis project, Bull. Am. Meteorol. Soc., 77,437–471. circulation and climate variability, Rev. Geophys., 47, RG1002, Karim, A., and J. Veizer (2002), Water balance of the Indus River Basin doi:10.1029/2007RG000245. and moisture source in the Karakoram and western Himalayas: Shenoi, S. S. C. (2010), Intra-seasonal variability of the coastal Implications from hydrogen and oxygen isotopes in river water, J. currents around India: A review of the evidences from new Geophys. Res., 107(D18), 4362, doi:10.1029/2000JD000253. observations, Indian J. Mar. Sci., 39(4), 489–496. Kumar, S. P., and T. G. Prasad (1999), Formation and spreading of Sirocko, F., M. Sarnthein, H. Erlenkeuser, H. Lange, M. Arnold, Arabian Sea high-salinity water mass, J. Geophys. Res., 104(C1), and J. C. Duplessy (1993), Century-scale events in monsoonal 1455–1464. climate over the past 24,000 years, Nature, 364, 322–324. Kuzucuoglu, C., W. Dörﬂer, S. Kunesch, and F. Goupille (2011), Staubwasser, M. (2006), An overview of Holocene South Asian Mid- to late-Holocene climate change in central Turkey: The monsoon records – Monsoon domains and regional contrasts, Tecer Lake record, The Holocene, 21(1), 173–188. J. Geol. Soc. India, 68(3), 433–446. Lemcke, G., and M. Sturm (1997), δ18O and trace element measure- Staubwasser, M., and H. Weiss (2006), Holocene climate and cul- ments as proxies for the reconstruction of climate changes at tural evolution in late prehistoric-early historic West Asia – Lake Van, in Third Millenium BC Climate Change and Old World Introduction, Quat. Res., 66(3), 372–387. Collapse, NATO ASI Ser., Ser. 1, vol. 49, edited by H. N. Dalfes, Staubwasser, M., F. Sirocko, P. M. Grootes, and H. Erlenkeuser G. Kukla, and H. Weiss, pp. 653–678, Springer, New York. (2002), South Asian monsoon climate change and radiocarbon in Migowski, C., M. Stein, S. Prasad, J. F. W. Negendank, and A. the Arabian Sea during early and middle Holocene, Paleoceano- Agnon (2006), Holocene climate variability and cultural evolu- graphy, 17(4), 1063, doi:10.1029/2000PA000608. tion in the Near East from the Dead Sea sedimentary record, Staubwasser, M., F. Sirocko, P. M. Grootes, and M. Segl (2003), Quat. Res., 66(3), 421–431. Climate change at the 4.2 ka BP termination of the Indus valley Milliman, D. J., G. S. Quraishee, and M. A. A. Beg (1984), Sedi- civilization and Holocene south Asian monsoon variability, Geo- ment discharge from the Indus River to the ocean: Past, present, phys. Res. Lett., 30(8), 1425, doi:10.1029/2002GL016822. future, in Marine Geology and Oceanography of Arabian Sea Vialard, J., S. S. C. Shenoi, J. P. McCreary, D. Shankar, F. Durand, and Coastal Pakistan, edited by B. U. Haq and J. D. Milliman, V. Fernando, and S. R. Shetye (2009), Intraseasonal response of pp. 65–70, Reinhold, New York. the northern Indian Ocean coastal waveguide to the Madden- Parker, A. G., L. Eckersley, M. M. Smith, A. S. Goudie, S. Stokes, Julian Oscillation, Geophys. Res. Lett., 36, L14606, doi:10. S. Ward, K. White, and M. J. Hodson (2004), Holocene vegeta- 1029/2009GL038450. tion dynamics in the northeastern Rub’ al-Khali desert, Arabian Wake, C. (1989), Glaciochemical investigations as a tool for deter- Peninsula: A phytolith, pollen and carbon isotope study, J. Quat. mining the spatial and seasonal variation of snow accumulation Sci., 19(7), 665–676. in the central Karakoram, northern Pakistan, Ann. Glaciol., 13, Parker, A. G., A. S. Goudie, S. Stokes, K. White, M. J. Hodson, M. 279–284. Manning, and D. Kennet (2006), A record of Holocene climate Weiss, H. (1993a), The genesis and collapse of 3rd millennium change from lake geochemical analyses in southeastern Arabia, north Mesopotamian civilization, Science, 261, 995–1004. Quat. Res., 66(3), 465–476. Weiss, H. (1993b), Correction to “The genesis and collapse of Phillips, W. M., V. F. Sloan, J. F. Shroder Jr., P. Sharma, M. L. 3rd millennium north Mesopotamian civilization”, Science, Clarke, and H. M. Rendell (2000), Asynchronous glaciation at 262, 1358. Nanga Parbat, northwestern Himalaya Mountains, Pakistan, Wolf-Vecht, A., N. Paldor, and S. Brenner (1992), Hydrographic Geology, 28(5), 431–434. indications of advection/convection effects in the Gulf of Elat, Rimbu, N., T. Felis, G. Lohmann, and J. Päatzold (2006), Winter Deep Sea Res., Part A, 39(7–8), 1393–1401. and summer climate patterns in the European-Middle East during Ziv, B., U. Dayan, Y. Kushnir, C. Roth, and Y. Enzel (2006), recent centuries as documented in a northern Red Sea coral Regional and global atmospheric patterns governing rainfall in record, The Holocene, 16(3), 321–330. the southern Levant, Int. J. Climatol., 26(1), 55–73. Rodwell, M. J., and B. J. Hoskins (1996), Monsoons and the dynamics of deserts, Q. J. R. Meteorol. Soc., 122, 1385–1404. Russell, A. D., and H. J. Spero (2000), Field examination of the M. Staubwasser, Institute of Geology and Mineralogy, Uni- oceanic carbonate ion effect on stable isotopes in planktonic versity of Cologne, Zülpicher Strasse 49a, D-50674 Cologne, foraminifera, Paleoceanography, 15(1), 43–52. Germany. ([email protected]) Geomorphological Constraints on the Ghaggar River Regime During the Mature Harappan Period

Hideaki Maemoku,1 Yorinao Shitaoka,2 Tsuneto Nagatomo,3 and Hiroshi Yagi4

The course of the Ghaggar ephemeral stream has been proposed as a possible location for the mythical Sarasvati River. Here we examine “the Lost Sarasvati” hypothesis from a geomorphological perspective with the aid of a new chronology of floodplains and sand dunes in the Ghaggar river basin. First, we compare the floodplains of the large glacial-fed Himalayan Rivers with the Ghaggar floodplain to show that the size of floodplains is directly proportional to the volume of annual river discharges. Thus, the Ghaggar floodplain is too small to explain large discharges along this stream in the past or under the present ephemeral flows. Next, we use optically stimulated luminescence to determine the age of sand dunes developed on either side of Ghaggar floodplain. These dunes have started to form prior to the Mature Harappan period and continue to grow until present. In our conclusions, we argue that the Ghaggar was not a large river as “the mighty Sarasvati” has been described at least during the Mature Harappan period.

1. INTRODUCTION: THE GHAGGAR Large urban centers, such as Mohenjo-Daro and Harappa, AND THE LOST SARASVATI flourished in the Indus Valley between 2500 and 1900 before the Common Era (B.C.E.), but most cities of the In this study, we aim to examine “the Lost Sarasvati” Indus Civilization were abandoned during the following hypothesis [Gupta et al., 2004] put forward to explain the Post-Urban Harappan period [Possehl, 2002]. Since the decline and fall of the Indus Civilization. Toward this goal, discovery of ruins of these cities in the nineteenth century, we use geomorphological analysis in combination with a many reasons have been proposed for the decline and fall of new absolute floodplain and sand dune chronology to assess the Harappan culture, including invasion of hostile ethnic if the Ghaggar River in northwestern India was indeed the groups, earthquakes, devastating floods, climate change, “mighty Sarasvati” during the peak of the Indus Civilization. etc. [Wheeler, 1947; Mackay, 1943; Sahni, 1956; Rao, 1979; Lahiri, 2000]. “The Lost Sarasvati” hypothesis proposes that one of the major triggers for the decline of the Indus civilization lies in 1Graduate School of Education, Hiroshima University, Higashi- the drying of the Sarasvati River [Yashpal et al., 1984]. The Hiroshima, Japan. “mighty Sarasvati” is supposed to have been a powerful river 2 Institute for Geothermal Sciences, Kyoto University, Beppu, during the Vedic period [Kalyanaraman, 2008], although Japan. 3 there is still a controversy among Sanskritists about when Nara University of Education, Nara, Japan. the period started and how long it lasted. Traditional scholars 4Faculty of Education, Art and Science, Yamagata University, suggest that the inception of the Vedic period be around 1200 Yamagata, Japan. B.C.E., while Witzel proposes that it lasted from 1500 to 500 Climates, Landscapes, and Civilizations B.C.E. [Deshpande, 1995; Hock, 1999; Witzel, 2006]. Those Geophysical Monograph Series 198 who hold “the Lost Sarasvati” hypothesis believe that the © 2012. American Geophysical Union. All Rights Reserved. river shrunk drastically in response to climate change or river 10.1029/2012GM001218 capture, which presumably caused severe water shortages 97 98 GEOMORPHOLOGICAL CONSTRAINTS ON GHAGGAR RIVER REGIME during the peak period of the Harappan development [Old- B.P. Using satellite image analysis, Yashpal et al. [1980] also ham, 1893; Marshall, 1931; Stein, 1942; Mughal, 1990; concluded that the Ghaggar used to be connected to the Raikes, 1968]. One of the leading candidates for the lost glacial area of the Higher Himalayas via the Sutlej or Yamuna Sarasvati is the Ghaggar-Hakra Valley [Sharma et al., River. Valdiya [2002] proposed uplift or rejuvenation of ad- 2008] that originates in the Lower Himalayas of northwest- jacent rivers as possible reasons for the piracy of the Yamuna ern India. The Ghaggar is ephemeral and, during the summer or Sutlej to the Ganges or the Indus watersheds, respectively. monsoon, flows from the Punjab Plain toward the west, Based on geomorphological research on river terraces, disappearing into the Cholistan Desert as the valley turns Rajaguru and Badam [1999] reported instead that the Ghag- into a dry riverbed. Many Harappan sites are distributed gar did not experience drastic changes in water discharge along the river including major sites such as Kalibangan or during the Harappan period. Tripathi et al. [2004] deduced Rakhigarhi [Possehl, 2002] (Figure 1). The past lower course that the Ghaggar alluvium did not include sediments from of the Ghaggar-Hakra cum Saravati in Pakistan is yet unclear glaciated regions resulting from analyzing their Sr and Nd although several options have been proposed using satellite isotopic composition and consequently proposed that the imagery [Roy and Jakhar, 2001; Gupta et al., 2004], includ- Ghaggar was never a glacial-fed river, at least during the late ing a possible continuation via the Nara Valley to as far south Quaternary. as the Arabian Sea. We reanalyze the geomorphology along the Ghaggar Val- Naruse [1974, 1985] investigated topographical maps and ley to clarify if the river might have been glacial-fed similar found old channels and riverbank dunes on the interfluves to its larger neighbors, the Sutlej and the Yamuna. We assess between the Sutlej and the Ghaggar. He further proposed that the history of the river on the basis of new chronologies for the Ghaggar was once connected to the Sutlej before 4000 floodplain and dune deposits in the Ghaggar river basin.

Figure 1. Geographical features and major archaeological sites of Harappan Culture in and around the Indus Valley. Site locations are after Possehl [2002]. MAEMOKU ET AL. 99

2. GEOLOGICAL AND GEOMORPHOLOGICAL Siwalik Formation underlying these alluvial deposits is OUTLINE OF THE STUDIED AREA folded with a wavelength of 40 km and a NW-SE axis [Naruse, 1974]. The alluvial deposits are also deformed The Indus, which originates in the Tibetan Plateau, is one by folding that causes discontinuity of groundwater level in of the largest rivers in Asia. The river joins other glacial-fed the Bist Doab located between the Sutlej and the Beas rivers, including the Jhelum, the Chenab, the Ravi, the Beas, [Fujiwara, 1978]. and the Sutlej in the western part of Punjab Plain, and flows eventually into the Arabian Sea in the vicinity of Karachi, 3. PRESENT FLOODPLAIN AND WATER FLOW Pakistan. The Punjab Plain is alluvial, formed by sediments OF MAJOR RIVERS AND THE GHAGGAR transported by these major tributaries of the Indus. The Ghaggar river basin is located on the Indo-Gangetic Satellite imagery shows that the present floodplain of the alluvial plain between the Indus and the Ganges watersheds. Ghaggar is at most 8 km wide in Rajasthan [Yashpal et al., The Ghaggar River originates in ephemeral streams in the 1984]. One of the reasons why the Ghaggar is thought to Lower Himalayas and continues into Hakra River flowing have been a glacial-fed mighty river is that the width of its westerly between the Sutlej Valley and the Thar Desert until floodplain was assumed to be larger than present flows it turns into a dry riverbed in the Cholistan Desert. would justify [Yashpal et al., 1980]. Sridhar et al. [1999], The Punjab and Haryana plains are composed of dissected however, suggested that ephemeral rivers often have wider alluvial fans built at the foot of the Himalayas, alluvial up- floodplains because they can be flooded easily due to their lands on the interfluves between major tributaries of the shallow riverbed. The Ghaggar River experienced severe Indus, as well as floodplains along these rivers [Naruse, floods in 1988, 1993, and 1995 as marked on the pier of 1985]. Sand dunes, several meters in height, occur in places, Sirsa Bridge, although it usually shows dry except for the whereas stabilized dunes higher than 10 m cover the southern summer monsoon season. part of the plain adjacent to the Thar Desert. To examine the geomorphological differences between The Punjab-Haryana plain is not stable tectonically during Ghaggar and modern perennial rivers in the region, we the recent geological period. The ~200 m thick alluvial analyzed the relationship between the size of the floodplain deposits of the plain are relatively thin compared to the and the annual flow of the Indus and its major tributaries in Himalayan front with a thickness >2000 m because the Punjab. If the Ghaggar once had its origin in the glacial area

Figure 2. Principal rivers and fluvial microrelief of the Punjab Plain. Elevation data are based on Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model (one arc second). Elevation is tinted every 5 m. The bold line with ticks on both sides indicates the floodplain width. 100 GEOMORPHOLOGICAL CONSTRAINTS ON GHAGGAR RIVER REGIME

Figure 3. Geomorphological map of the Punjab Plain and northern part of the Thar Desert. After Naruse [1985], reprinted with permission.

of the Higher Himalayas and the fluvial landforms formed then were preserved until today, the size of the floodplain must be comparable to that of present perennial rivers such as the Sutlej, for example. The floodplain is defined as the part of a fluvial plain that can be inundated by floodwater. Thin loamy sand or silt is usually deposited on it. Hydrologically, the floodplain is defined as a regularly flooded area. It is not necessary that the floodplain is inundated entirely by every flood; however, the lower floodplain that is inundated by most floods is defined as the active floodplain. We utilized satellite-derived elevation data (Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation Model (GDEM): one arc second) to analyze fluvial landforms. ASTER GDEM is a DEM data set, which is acquired by a satellite-borne sensor “ASTER” to cover all the land on earth developed by The Ministry of Economy, Trade and Industry of Japan (METI) in collaboration with NASA (see http://www.ersdac.or.jp/ GDEM/E/index.html). Figure 2 shows in detail the relief of Figure 4. Relationship between the floodplain width and elevation the Punjab Plain tinted at every 5 m in elevation. We can for the seven principal rivers of the Indus watershed. MAEMOKU ET AL. 101 distinguish the floodplains along all major rivers from inter- suggest that the width of its floodplain is narrower relative fluves by the dislocation color pattern by more than 10 m in to its annual flow. relative height in concordance with the geomorphological The average width of the Ghaggar floodplain is ~5 km map produced by Naruse [1985] (Figure 3). Naruse [1985] and significantly smaller than the floodplains of the Hima- showed that the abandoned floodplain was ~3 m higher than layan Rivers. Furthermore, it is much accordant with its the active one and could be inundated by the largest floods present discharge (Figure 5). If the Ghaggar had been as occurring once in several decades. Floodplain is separated by large as the Sutlej, for example, the width of its floodplain steep scarp of 1.5 to 12 m high from the alluvial upland or should have been several times wider than the present one. interfluve. For perspective, the abandoned floodplains along We conclude that fluvial landforms along the present Ghag- the middle and lower course of the Indus were inundated gar must have developed under processes similar to the totally in the August 2010 severe flood (see e.g., NASA present river regime, not of a past hydrology dominated by Photo Journal: PIA13337, http://photojournal.jpl.nasa.gov/ large discharges. catalog/PIA13337). Figure 4 shows the relationship between the width of the 4. WHEN WAS THE PRESENT FLOW REGIME floodplain as measured at locations indicated in Figure 2 ESTABLISHED ON THE GHAGGAR? and elevation at the same locations, which is used as a proxy for an average gradient. Major rivers except for The answer to the above question could provide a solution Ghaggar have floodplains of 10–20 km wide in front of the for “the Lost Sarasvati” hypothesis. If the present hydrologic sub-Himalayas at elevations of ~250 m. The floodplain regime of the Ghaggar started before the Mature Harappan width increases proportionally as the elevation decreases. period, the hypothesis should be removed from the list of Figure 5 shows that the average width of the floodplain of possible causes for the decline of the Indus Civilization. each river increases with the average annual flow of the The Ghaggar flows across the Punjab Plain starting from river. Because the Indus, flowing along the foot of the the sub-Himalayas in the vicinity of Chandigarh. The Sulaiman Ranges, is prone to be covered by debris trans- ephemeral river flows southwestward down to Sirsa, then ported by many small tributary streams from that side, we it turns to the west. The Chautang River, one of the large tributaries of the Ghaggar, also originates in the sub-Hima- layas in northeastern Haryana. The course of the Chautang is evident as a continuous depression only when large floods occur. Yashpal et al. [1980] suggested that the Chau- tang was, in fact, the old Yamuna course flowing into the Ghaggar River. The series of depressions continue through Hisar, where the Chautang is obvious again when supplied by overflow water from the Ghaggar drain canal after Rawatsar (near Bhakrawala). The Chautang joins the Ghag- gar east of Suratgarh (Figure 6). In the northern area of the Thar Desert, along the courses of the Ghaggar and Chautang Rivers, numerous dunes occur. Their wavelength varies between several hundred meters and several kilometers, and their height is several tens of meters. These dunes occupy the northern part of the dune fields surrounding the Aravalli Range and have been formed by strong summer monsoon winds [Kar, 1993]. In general, in the Thar Desert, most dunes are activated when precipitation falls below 200–250 mm per year [Goudie et al., 1973; Singhvi and Kar, 2004]. Since the annual precipitation in the northern part of the Thar is higher (i.e., 250– 500 mm; Figure 7), almost all dunes are inactive and Figure 5. Relationship between the average floodplain width and stabilized, particularly along the Ghaggar and Chautang average annual water flows from Johnson [1979a] and Subba [Juyal et al., 2003]. [2001]. There is a strong positive correlation between them (R = We mapped dunes using CORONA satellite images, 0.892 except the main Indus). ASTER GDEM data, and Google Map images (Figure 6) 102 GEOMORPHOLOGICAL CONSTRAINTS ON GHAGGAR RIVER REGIME

Figure 6. Distribution of desert sand dunes in the northern Thar Desert and the location of the Ghaggar and the Chautang. Contours are at 20 m intervals in the alluvial plain and at 100–500 m in the mountain area. Box indicates the area of Figure 8. in order to assess the age of the dune fields on either side OSL dating is suitable for eolian materials because the of the rivers Ghaggar and Chautang. We can recognize record of natural radiation stored in its minerals is apt to be which dunes limit the edge of floodplain of the Ghaggar reset during transportation in the air [Aitken, 1998]. Sub- and Chautang and choose some locales for sampling for samples of the collected samples were tested with a grain dating considering position, size, and accessibility from the size analyzer to check whether they were eolian or fluvial in distribution map. The rationale is that the floodplain could origin. We confirmed that all samples were composed of not have been wider than it is today earlier than the dune eolian sand because they showed unimodal distribution in fields starting ages. grain size rather than bimodal (Figure 9). We carried out field surveys in 2008 and 2010 to sample We measured OSL ages based on the single aliquot dune sands for optically stimulated luminescence (OSL) regenerative-dose (SAR) protocol [Murray and Wintle, dating. We selected sampling sites along both sides of the 2000] using an OSL/TL reader (NRL-99-OSTL) [Nagatomo Ghaggar and Chautang as close as possible to the rivers at et al., 2009] equipped with an array of 32 blue LEDs (470 intervals of tens kilometers covering an entire area of dune nm) at Nara University of Education (NUE), Japan. Dates are fields. Fortunately, local people uses dune sand for brick shown in Table 1 [Shitaoka et al., 2012]. manufacture so that we could find several quarries of dune OSL dating indicates that most of the dunes adjacent to the sand along the Ghaggar Valley. The samples were collected Ghaggar floodplain on either side are older than ~10 to 15 ka. from eight locales selected based on the dunes distribution Only one sample (43GB) is younger at 5.1±0.3 ka. We did map (Figure 6) as well as accessibility to the outcrops (Fig- not detect dunes aged 9 to 6 ka, which agrees with previous ure 8). We sampled outcrops that showed original dune research that suggested a pluvial for that time period [Singhvi sedimentary structures and from positions as low as possible and Kar, 2007; Chawla et al., 1992]. to get the oldest age of sedimentation. Samples were collected Thus, sand dunes limiting both sides of the Ghaggar flood- in opaque bags covered in opaque cloth to protect them from plain have begun developing prior to the Mature Harappan exposure to light. period (i.e., 4.5–3.9 ka), and they were never washed away MAEMOKU ET AL. 103

Figure 7. Average annual precipitation in and around the Indus Valley based on Johnson [1979a, 1979b]. Unit of isopleth is millimeter. by much larger water ﬂows of a “mighty Sarasvati.” We Our results are supported by two undeniable facts. First, conclude that the Vedic Sarasvati cannot be equated with a we found habitation layers on the dunes adjacent to the larger Ghaggar River similar to the Indus or its Himalayan Ghaggar River (e.g., 43GB) listed as sites of the Mature tributaries at least since the beginning of the Holocene. Harappan period [Joshi et al., 1984]. Thus, dunes should

Figure 8. Optically stimulated luminescence ages of sampled locales and major archaeological sites in the Ghaggar river basin. See Shitaoka et al. [2012] for details on dating. 104 GEOMORPHOLOGICAL CONSTRAINTS ON GHAGGAR RIVER REGIME

Figure 9. Typical grain size histograms of dune sand and fluvial mud in the Ghaggar river basin. Vertical axis shows the volume percentage analyzed by laser diffraction SHIMADZU particle size analyzer (SALD-3100). have already been established at that time. Second, many astating flood since Harappan people settled there. It would small Harappan sites, including many belonging to the Ma- be unreasonable to imagine that small sites could withstand ture period, occur in the present Ghaggar floodplain [Joshi et complete washouts when huge floods from a glacial-fed river al., 1984] indicating that this area never suffered from dev- hit them periodically.

Table 1. Optically Stimulated Luminescence (OSL) Ages for Dune Sands Sampled in the Ghaggar River Basina Location Coordinate Elevation (m) Sample Paleodose (Gy) Annual Dose (mGy aÀ1) OSL Age (ka) Sisarkhas N28°56′41.32″ 217 upper 38.3±2.5 2.50±0.08 15.3±1.1 E76°14′11.75″ lower 32.0±1.6 2.66±0.09 12.1±0.7 Mayer N29°05′54.11″ 208 upper 30.2±2.2 2.34±0.10 12.9±1.1 E75°50′49.07″ lower 34.1±2.0 2.47±0.09 13.8±1.0 Bhodasar N29°10′34.00″ 192 upper 13.4±1.0 2.57±0.25 5.2±0.6 E75°10′41.89″ lower 0.5b 2.45±0.11 0.2b Langmahar N29°20′01.65″ 188 upper 26.4±1.3 2.01±0.08 13.1±0.8 E73°57′39.44″ lower 34.1±2.3 2.37±0.08 14.4±1.1 Bhakrawala N29°16′25.23″ 189 upper 11.2±0.4 2.27±0.09 4.9±0.3 E74°22′45.48″ lower 28.2±1.1 2.14±0.07 13.2±0.7 Lalpurasikhan1 N29°16′25.23″ 169 upper ––– E74°22′45.48″ lower 33.4±2.3 3.03±0.11 11.0±0.9 Lalpurasikhan2 N29°16′25.23″ 168 upper ––– E74°22′45.48″ lower 32.4±1.7 3.15±0.09 10.3±0.6 43GB N29°10′47.80″ 151 upper 11.7±0.7 2.33±0.08 5.0±0.3 E73°28′27.77″ lower 11.5±0.6 2.23±0.08 5.1±0.3 aSee Shitaoka et al. [2012] for details on dating. bProbably due to modern contamination in sampling. MAEMOKU ET AL. 105

5. CONCLUSION The Indo-Aryans of Ancient South Asia, edited by G. Erdosy, pp. 67–84, De Gruyter, Berlin. We carried geomorphological and geochronological re- Fujiwara, K. (1978), Behavior of the hydrologic cycle in the Punjab, search on the floodplains and dunes to explore whether the India [in Japanese with English abstract], J. Geogr., 87, 130–151. Ghaggar River fits the description of “mighty” glacial-fed Goudie, A., B. Allchin, and K. T. M. Hegde (1973), The Sarasvati of the Vedas. Our conclusions can be summarized former extensions of the great Indian sand desert, Geogr. J., as follows: 139, 243–257. 1. The width of the present Ghaggar floodplain is much Gupta, A. K., J. R. Sharma, G. Sreenivasan, and K. S. Srivastava smaller than that of glacial-fed Indus or its tributaries, sug- (2004), New Findings on the course of river Saraswati, J. Indian – gesting that the floodplain is not a fossil landform that was Soc. Remote Sens., 32,1 24. developed by a larger river. Hock, H. H. (1999), Out of India? The linguistic evidence, in Aryan and Non-Aryan in South Asia: Evidence, Interpretation and 2. Most sand dunes accompanying the Ghaggar and the fl Ideology, Harvard Oriental Ser. Opera Minora, vol. 3, edited by Chautang on either side of their oodplains began to develop J. Bronkhorst and M. Deshpande, pp. 1–18, Harvard Univ., Cam- – around 15 10 ka; however, some dunes were dated at circa bridge, Mass. 5 ka. Johnson, B. L. C. (1979a), India: Resources and Development, 3. Dunes already existed during the Mature Harappan Heinemann Educ. Books, London, U. K. period (4.5–3.9 ka) along the Ghaggar and were not washed Johnson, B. L. C. (1979b), Pakistan, Heinemann Educ. Books, away in any significant degree since then. London, U. K. 4. Our results are supported by habitation layers on the Joshi, J. P., M. Bala, and J. Ram (1984), The Indus civilization: sand dunes dating back to the Mature Harappan period and A reconsideration on the basis of distribution maps, in Frontiers by the many Harappan sites occurring in the present Ghaggar of the Indus Civilization: Sir Mortimer Wheeler Commemoration floodplain. Volume, edited by B. B. Lal and S. P. Gupta, pp. 510–530, Books 5. The mythical Sarasvati did not exist as described in the & Books, New Delhi, India. Vedas in the Ghaggar basin as a glacial-fed large river such Juyal, N., A. Kar, S. N. Rajaguru, and A. K. Singhvi (2003), as the Indus or its tributaries at least during the Mature Luminescence chronology of Aeolian deposition during the late Harappan period. Quaternary on the southern margin of Thar desert, India, Quat. Int., 104,87–98. 6. Saini and Mutaba [2010] reported that the Ghaggar flowed Kalyanaraman, S. (Ed.) (2008), Vedic River Sarasvati and Hindu regularly between 5.9 and 4.3 ka along the Haryana Plain. These Civilization, Aryan Books Int., New Delhi, India. events, however, could have occurred even in the present time ñ – Kar, A. (1993), Aeolian processes and bedforms in the Thar desert, due to global atmospheric oscillation such as El Ni o Southern J. Arid Environ., 25,83–96. Oscillation or Indian Ocean Dipole, and they have not resulted Lahiri, N. (Ed.) (2000), The Decline and Fall of the Indus in extensive washing out of attendant sand dunes. Civilization, Permanent Black, Delhi, India. Mackay, E. J. H. (1943), Chanhu-Daro Excavations 1935–36, Am. Acknowledgments. This research work is a part of the project of Oriental Ser., vol. 20, Am. Oriental Soc., New Haven, Conn. the Research Institute for Humanity and Nature (RIHN), Kyoto, Marshall, J. (1931), Mohenjo-Daro and the Indus Civilization, “ ” Japan, entitled Environmental Change and the Indus Civilization Asian Educ. Serv., New Delhi, India. led by T. Osada (RIHN). We express our appreciation to Osada, Mughal, M. R. (1990), The decline of the Indus civilization and who provided us with the resources necessary for the execution of the Late Harappan period in the Indus Valley, Lahore Mus. J., 3, this research. We also appreciate the help provided in the field by V. 1–22. Shinde and P. Shirvalkar (Deccan College, Pune, India), M. Kumar Murray, A. S., and A. G. Wintle (2000), Luminescence dating of and V. Dangi (University, Rohtak, India), and their many energetic quartz using an improved single-aliquot regenerative-dose students. protocol, Radiat. Meas., 32,57–73, doi:10.1016/S1350-4487 (99)00253-X. REFERENCES Nagatomo, T., Y. Shitaoka, H. Namioka, M. Sagawa, and Q. Wei (2009), OSL dating of the Strata at Paleolithic sites in the Nihe- Aitken, M. J. (1998), An Introduction to Optical Dating, Oxford wan Basin, China, Acta Anthropol. Sin., 28, 276–284. Univ. Press, Oxford, U. K. Naruse, T. (1974), Sand dunes in the Punjab Plains, in Geograph- Chawla, S., R. P. Dhir, and A. K. Singhvi (1992), Thermolumi- ical Field Research in Northwestern India: A Progress Mono- nescence chronology of sand profiles in the Thar desert and their graph, edited by H. Ishida, pp.122–128, Dep. of Geogr., implications, Quat. Sci. Rev., 11,25–32. Hiroshima Univ., Hiroshima, Japan. Deshpande, M. M. (1995), Vedic Aryans, non-Vedic Aryans, and Naruse, T. (1985), Aeolian geomorphology of the Punjab plains and non-Aryans: Judging the linguistic evidence of the Veda, in the north Indian desert, Ann. Arid Zone, 24, 267–280. 106 GEOMORPHOLOGICAL CONSTRAINTS ON GHAGGAR RIVER REGIME

Oldham, C. F. (1893), The Saraswati and the lost river of the Indian Stein, A. (1942), A survey of ancient sites along the “lost” Saras- desert, J. R. Asiatic Soc., 34,49–76. wati River, Geogr. J., 99, 173–182. Possehl, G. L. (2002), The Indus Civilization, AltaMira Press, Subba, B. (2001), Himalayan Waters: Promise and Potential, Lanham, Md. Problems and Politics, Panos South Asia, Kathmandu, Nepal. Raikes, R. L. (1968), Kalibangan: Death from natural causes, Tripathi, J. K., B. Bock, V. Rajamani, and A. Eisenhauer (2004), Is Antiquity, 42, 281–286. river Ghaggar, Saraswati? Geochemical constraints, Curr. Sci., Rajaguru, S. N., and G. L. Badam (1999), Evolutionary story of 87, 1141–1145. a lost river of northwestern India, Mem. Geol. Soc. India, 42, Valdiya, K. S. (2002), Saraswati: The River that Disappeared, 143–151. Univ. Press, Hyderabad, India. Rao, S. R. (1979), Lothal: A Harappan port town, 1955-62, Mem. Wheeler, M. (1947), Harappa 1946: The defences and cemetery Archaeol. Soc. India, 1,1–77. R-37, Ancient India, 3,58–130. Roy, A. B., and S. R. Jakhar (2001), Late Quaternary drainage Witzel, M. (2006), Central Asian roots and acculturation in South disorganization, and migration and extinction of the Vedic Asia: Linguistic and Archaeological evidence from western Cen- Saraswati, Curr. Sci., 81, 1188–1195. tral Asia, the Hindukush and northwestern South Asia for early Sahni, M. R. (1956), Bio-geological evidence bearing on the Indo-Aryan language and religion, in Indus Civilization: Text and decline of the Indus Valley Civilization, J. Paleontol. Soc. India, Context, edited by T. Osada, pp. 61–185, Manohar, Delhi, India. 1, 101–107. Yashpal, S. B., R. K. Sood, and D. P. Agrawal (1980), Remote Saini, H. S., and S. A. I. Mutaba (2010), Luminescence dating of sensing of the lost Saraswati River, Proc. Indian Acad. Sci., 89, the sediments from a buried channel loop in Fatehabad area, 317–337. Haryana: Insight into Vedic Saraswati River and its environ- Yashpal, S. B., R. K. Sood, and D. P. Agrawal (1984), Remote ment, Geochronometria, 37,29–35, doi:10.2478/v10003-010- sensing of the lost Saraswati river, in Frontiers of the Indus 0021-5. Civilization: Sir Mortimer Wheeler Commemoration Volume, Sharma, D. P., M. Sharma, and K. Pandey (2008), The Lost edited by B. B. Lal and S. P. Gupta, pp. 491–497, Books & Saraswati Civilization, Bharatiya Kala Prakashan, Delhi, India. Books, New Delhi, India. Shitaoka, Y., H. Maemoku, and T. Nagatomo (2012), Quartz OSL dating of sand dunes in Ghaggar Basin, northwestern India, H. Maemoku, Graduate School of Education, Hiroshima Uni- – Geochronometria, 39(3), 221 226. versity, 1-1-1 Kagamiyama, Higashi-Hiroshima, Japan. (maemoku@ Singhvi, A. K., and A. Kar (2004), The aeolian sedimentation hiroshima-u.ac.jp) record of the Thar desert, J. Earth Syst. Sci., 113(3), 371–401. T. Nagatomo, Nara University of Education, Takabatake-cho, Singhvi, A. K., and A. Kar (2007), The history of sand dunes in the Nara, Japan. Thar desert, ISG Newslett., 13,4–15. Y. Shitaoka, Institute for Geothermal Sciences, Kyoto Uni- Sridhar, V., S. S. Merth, and J. N. Malik (1999), Late Quaternary versity, 3088-176 Noguchibaru, Beppu, Japan. drainage disruption in northwestern India: A geoarchaeological H. Yagi, Faculty of Education, Art and Science, Yamagata enigma, Mem. Geol. Soc. India, 42, 187–204. University, 1-4-12 Kojirakawa-machi, Yamagata, Japan. A Simulation of the Neolithic Transition in the Indus Valley

Carsten Lemmen and Aurangzeb Khan

Institute of Coastal Research, Helmholtz-Zentrum Geesthacht, Geesthacht, Germany

The Indus Valley Civilization (IVC) was one of the first great civilizations in prehistory. This Bronze Age civilization flourished from the end of the fourth millennium B.C. It disintegrated during the second millennium B.C.; despite much research effort, this decline is not well understood. Less research has been devoted to the emergence of the IVC, which shows continuous cultural precursors since at least the seventh millennium B.C. To understand the decline, we believe it is necessary to investigate the rise of the IVC, i.e., the establishment of agriculture and livestock, dense populations, and technological developments in 7000–3000 B.C. Although much archaeological information is available, our capability to investigate the system is hindered by poorly resolved chronology and by a lack of fieldwork in the intermediate areas between the Indus Valley and Mesopotamia. We thus employ a complementary numerical simulation to develop a consistent picture of technology, agropastoralism, and population developments in the IVC domain. Results from this global land use and technological evolution simulator show that there is (1) fair agreement between the simulated timing of the agricultural transition and radiocarbon dates from early agricultural sites, but the transition is simulated first in India then Pakistan; (2) an independent agropastoralism developing on the Indian subcontinent; and (3) a positive relationship between archeological artifact richness and simulated population density that remains to be quantified.

1. INTRODUCTION emergence of this great civilization, which shows continuous cultural precursors at least since the seventh millennium The Indus Valley Civilization (IVC), often termed Harap- B.C., such that some authors prefer to speak of the Indus pan civilization after its major type site, ﬂourished along the Valley cultural tradition [e.g., Kenoyer, 2006]. banks of the river Indus and its tributaries, including the The Indus Valley is one of the two great river basins on the adjacent coastal areas, between the fourth and second mil- Indian subcontinent, separated from the Ganges Valley to the lennium B.C. The IVC is characterized by urban centers, east by the Aravalli Mountain Range and the Thar Desert bronze technology and seals, trade networks with Mesopo- (Figure 1b). The region has been occupied by anatomically tamia and Arabia, and an as yet undeciphered writing system modern humans for at least 34,000, but possibly >80,000 [Shaffer, 1992; Possehl, 1998; Kenoyer, 2008]. Much re- years [Petraglia et al., 2010]. The late Paleolithic and the search has been devoted to the disintegration of the IVC Mesolithic are visible in stone and blade industries; microlith during the second millennium B.C. and much less to the blades indicate composite tools, differentiated occupation levels, and a broad-spectrum diet of foraging people from 26,000 years before present until at least the second millen- Climates, Landscapes, and Civilizations nium B.C., well into the Bronze Age [Petraglia et al., 2009]. Geophysical Monograph Series 198 Neolithic subsistence is based on a combination of barley- © 2012. American Geophysical Union. All Rights Reserved. and wheat-dominated agriculture and herding of sheep, 10.1029/2012GM001217 goats, and cattle. Domesticates are of mixed origin: barley 107 0 IUAINO NU ALYNEOLITHIC VALLEY INDUS OF SIMULATION 108

Figure 1. Temporal and spatial domain for this study: (a) Neolithic chronology of Baluchistan, the Indus Valley, and Gujarat based on the work of Possehl [2002] with updates from Fuller [2006]. Relative resource shortage is based on palaeoproxy evidence from Lemmen and Wirtz (submitted paper, 2012) and a transient Holocene climate simulation by Brovkin et al. [2002]. (b) Geography, topography, and site distribution of artifacts [Law, 2007] typologically associated with the Neolithic. LEMMEN AND KHAN 109 and wheat have no local wild progenitor, but zebu cattle mediate areas between the Indus Valley and Mesopotamia. bears a clear signature of local domestication [Chen et al., We thus employ a complementary numerical simulation to 2010; Fuller, 2011a]. The beginning of the Neolithic on the the IVC domain. Ackland et al. [2007] simulated the Neo- Indian subcontinent is regionally diverse and ranges from lithic transition in the IVC with an advancing frontier spread- 6500 B.C. in Baluchistan [Jarrige, 1995] to 3000 B.C. or ing from Southwest Asia and a competition between later on the central Indian Plateau and in the Himalaya migrating original farmers and resident converts. After Mountain Range [Samuel, 2000; Boivin et al., 2008]. Our 5000 simulation years (9000–4000 B.C.), their simulated focus here is on the period that spans from the Neolithic to spread of agriculture into the Indian subcontinent exhibits a the beginning of the Bronze Age of the IVC (7000–3200 demarcation line along the Indus River, separating original B.C.), i.e., on the early food production and regionalization farmers to the west from converts to the east. eras or the Mehrgarh phase and parts of the early Harappan A local South Asian agricultural center was considered in phase according to the different chronologies [Sharma et al., the numerical model by Patterson et al. [2010]. They as- 1980; Fuller, 2006; Kenoyer, 2008]. sumed a background hunting-gathering population and ini- Possehl [2002] separated these phases and eras of the tial farmer populations near Mehrgarh and at the eastern edge pre-Harappan Neolithic into two stages (Figure 1a): Early of their model domain. After 2000 simulation years, the Neolithic and Developed Neolithic. The Early Neolithic is population in the Indus Valley is dominated by expanding divided in the Kili Ghul Mohammad (KGM) and the Burj original farmers, a lesser share of converts, and a tiny frac- Basket-Marked (BBM) periods. Sites attributed to these peri- tion of remaining hunter-gatherers. ods are found in the western borderlands of the greater Indus We go beyond the biophysical approaches by Ackland et region and cluster in the hills and piedmont of Baluchistan al. [2007] or Patterson et al. [2010]; in our global land use (Figure 1b, phases 1 and 2). During the BBM, soft pottery and and technological evolution simulator (GLUES) [e.g., Lem- handmade pottery were introduced. Deposits of trash with men et al., 2011], we include sociotechnological innovation burnt pebbles, ash, animal bones, bone tools, hammer stones, in addition to migration, population growth, and subsistence polishers colored with red ochre, and a large collection of change to develop a consistent picture of the dynamics of blades, cores, and flint debris have been discovered and point technology, agropastoralism, and population developments to leather making, basket making, or weaving. Domesticated during the Mesolithic-Neolithic transition. This simulated wheat, barley, and cotton were used, pulses may have been data is subsequently compared to a representative archaeo- present; the use of copper is visible in the form of arrowheads logical data set [Law, 2007], where the chronology of Neo- and beads [Moulherat et al., 2002; Fuller, 2011a]. lithic sites is used to detect the regional transition from The Developed Neolithic is characterized by improved foraging to farming and where the site occurrence frequency pottery, developed agropastoral communities, growth, conti- is used to detect the Neolithic demographic transition. nuity, and geographical expansion. During its first period (here defined as both Togau and Sheri Khan Tarakai (SKT) 2. DATA AND MODEL DESCRIPTION (Figure 1b, phase 3), the settlement patterns show growth of village life; technological innovations include the use of gold, 2.1. Global Land Use and Technological the manufacture of compartmented seals, glazed steatite, and Evolution Simulator beads. An eastward expansion is visible in sites emerging along the ancient Ghaggar Hakra River. The second period of GLUES is a sociotechnological model, which hindcasts the Developed Neolithic (Figure 1b, phase 4) comprises technological evolution, potential population density, and the several cultural complexes. In the Kechi Beg complex in timing of the transition to agropastoralism for 685 world Baluchistan, wheat replaces barley; an irrigation canal near regions based on the geoenvironmental and cultural contexts, Mehrgarh may date to this period [Kenoyer, 2008]. On the and innovation and adaptation of regional populations. Each Punjab Plain, a change in the assemblage of pottery types local population in an approximately country-size region characterizes the Hakra Ware complex: Microlithic tools are utilizes its natural resources and interacts with its geograph- abundant, and sites are represented by a light scatter of ical neighbors via trade and migration. The full model is pottery without a buildup of a midden point to seasonal described by Wirtz and Lemmen [2003], with refinements of occupations [Possehl, 2002]. At the same time, the Anarta geography [Lemmen, 2010], migration and knowledge loss complex appears in Gujarat with distinct pottery styles. [Lemmen and Wirtz, 2010], and climate events (C. Lemmen Our capability to investigate the system more closely and and K. W. Wirtz, On the sensitivity of the simulated Euro- consistently in the regional context is hindered by the poorly pean Neolithic transition to climate extremes, submitted resolved chronology and by a lack of fieldwork in the inter- to Journal of Archaeological Science, 2012, hereinafter 110 SIMULATION OF INDUS VALLEY NEOLITHIC referred to as Lemmen and Wirtz, submitted paper, 2012); it 2.2. Artifact Database has been validated against archaeological data by Lemmen et al. [2011]. Key concepts are summarized below. We used the Indus Google Earth Gazetteer (version Au- Local societies are defined by their population density (P) gust 2008) compiled by Law [2007] for the geolocation of and by three characteristic traits: (1) technology (T), which artifacts relating to the Indus Valley. From this database of describes efficiency of food procurement; (2) share of agro- 21,687 metal and stone artifacts, we use here the cultural pastoral activities (Q) as the allocation pattern of manpower attribution and location of those 2028 artifacts and 368 sites to farming and herding as opposed to foraging; and (3) eco- that are in the spatial and temporal domain of our study. For nomic diversity (N), a technology-related trait, which repre- KGM, 18 sites contribute 90 artifacts to the database; for sents the different subsistence styles of a community. The BBM, there are 159 artifacts at 32 sites. We combined the temporal change of traits follows the direction of increased sites attributed to the Togau period with sites from the SKT benefit for success (i.e., growth) of its associated population (both with starting date 4300 B.C.) [Fuller, 2006] and obtain with a speed related to the trait’s variance [Fisher, 1930; 317 artifacts from 65 sites. Kisdi and Geritz, 2010]. Population growth rate depends on The subsequent period is represented by the Hakra Ware, resource availability (E), subsistence intensity, overexploita- Kechi Beg, and Anarta complexes. Combined, there are tion of resources, administrative overhead, and health im- 1250 artifacts (396, 670, and 184, respectively) from 253 provements; the latter four terms are functions of P, T, N, and sites (99, 112, and 42, respectively). While the main chro- Q, while the resource availability E is provided externally. E nology is based on the work of Possehl [2002], sites from the is estimated from the net primary productivity, which is in Anarta complex were included in this period based on the turn based on Climber-2 [Claussen et al., 1999] transient work of Fuller [2006]. To our knowledge, this gazetteer is temperature and precipitation anomalies from the Interna- the most representative data set of lithic and metal artifacts of tional Institute for Applied Systems Analysis (IIASA) clima- the Indus Valley tradition. It has been used to investigate the tological database [Leemans and Cramer, 1991]. trade and distribution networks of the IVC by its author Abrupt climate deteriorations modulate E; they are incor- [Law, 2011]. porated by using the extreme event anomalies found in 124 In archaeological or palaeoenvironmental archives, the globally distributed palaeoclimate proxies collected and an- decrease in the number of dates further back in time that is alyzed by Wirtz et al. [2010]; the estimation of resource caused by gradual destruction of the archive by postdeposi- failure from proxy climate is described by Lemmen and tional processes results in a taphonomic bias [Surovell et al., Wirtz (submitted paper, 2012). Figure 1a shows the average 2009]. We applied to the artifact density data a bias correc- relative decline of E in the Indus Valley region; these anoma- tion derived from the comparison of volcanic eruption lies are derived from regional abrupt changes found in oxy- layers in the Greenland Ice Sheet Project 2 (no taphonomic gen isotope anomalies from Kyrgyzstan [Ricketts et al., bias) ice core with radiocarbon-dated ash deposits; the correc- 2001], Tibet [Fontes et al., 1996], the Arabian Sea [Doose- tion factor c depends on time t (in years before present): 6 À1.39 Rolinski et al., 2001; Gupta et al., 2003], and Oman [Fleit- c(t)=5.73· 10 (2175.4 + t) [Surovell et al., 2009]. This mann et al., 2007]. method was successfully validated with Canadian coin mint- We set up the eight global model parameters and initial age data compared to a coin collection and applied to North values such that the simulation is able to hindcast an accu- American radiocarbon dates by Peros et al. [2010]. rate timing and location of the early farming centers in Southwest Asia, northern China, and Mesoamerica [Smith, 3. RESULTS AND DISCUSSION 1998] and a reasonable global pattern of the subsequent Neolithic transition. The simulation is started at 9500 sim- The simulated transition to predominantly agropastoralism- ulation years (sim) B.C. (equivalent to calendar years B.C., based subsistence occurs in the spatial domain of the Indus but emphasizes the artificial model time scale) [Lemmen et Valley and surrounding areas between 6300 and 3800 sim B.C. al., 2011). All of the 685 biogeographically defined regions, (Figure 2a). There is a general east-west trend in the simu- including 32 around the Indus Valley, are initially set with lated transition dates. Contemporary with the KGM, the farming activity at 4%, one fourth established agropastoral earliest transitions are simulated in India (before 5600 economies and unit technology; this setup represents a low- sim B.C.), including Rajasthan, Gujarat, and the Ganges density Mesolithic population relying on a broad spectrum Valleys, and early transitions (before 5000 sim B.C.) in foraging lifestyle with low unintentional farming activity. southern Sindh and northern Pakistan, including parts of the The regional simulated transition to agropastoralism is re- Punjab and Kashmir. During this first period, 15 out of 32 corded when Q exceeds 0.5. simulation regions undergo the transition to agropastoralism. LEMMEN AND KHAN 111

Figure 2. (a) Timing of the Mesolithic-Neolithic transition from a global land use and technological evolution simulator (GLUES) simulation (background color, simulation years B.C.) and inferred from the presence of Neolithic sites (foreground triangles, calendar years B.C.) in the greater Indus Valley. (b) Recovered artifact number in relation to simulated population size. Histogram of artifact density in four pre-Harappan periods of the Indus Valley and Baluchistan based on Law’s [2007] Indus Google Earth Gazetteer and corrected for taphonomic bias (bars, annotated with n Â c, where n is the raw number of artifacts, and c is the taphonomic correction factor); GLUES-simulated population size in all areas within 200 km from where Neolithic pre-Harappan artifacts were discovered (solid line).

Most of the remaining regions (Pakistan, eastern Afghani- use of (Mesolithic) microlithic tools, transhumance life style, stan, eastern Iran, and Makran coast) transition in the BBM i.e., the seasonal back-and-forth movement of herders with period before 4300 sim B.C. Of these, agropastoralism is their livestock and the continuation of foraging practice well simulated first in the Punjab and northern Baluchistan (be- into the Neolithic [Mughal, 1990; Fuller, 2011a]; it is not fore 4700 sim B.C.), then in northern Sindh (before 4500 resolved in the model, where agropastoral lifestyle is usually simB.C.). The simulated transition occurs latest, during the dominant within less than 500 years after its emergence. Togau period, in the eastern model domain, from central The model simulates first agropastoralism in northern Ba- Afghanistan to the Pakistan-Afghanistan-Iran border trian- luchistan and southern Sindh in the KGM period consistent gle (4300–3800 sim B.C.). with earliest sites in these areas. For the major part of A detailed and quantitative comparison is limited foremost Baluchistan, including the Mehrgarh site, the model places by the inherent timing uncertainties in the typologically the transition in the BBM period, which is later than the dated artifacts. This is closely related to the secondary com- earliest site dates, but agrees with dates from many other parison difficulty, which is created by the lack of temporal locations in Baluchistan. The most obvious mismatch is the continuity of samples from the Mesolithic to the Neolithic. early simulated (KGM and BBM) transition along the Indus, While we have shown how these issues can be addressed in the Ghaggar-Hakra, and Punjab Rivers, where site dates are the context of plenty and precise dates for Europe [Lemmen absent until the Kechi Beg and Hakra periods. et al., 2011], the comparison with dates from the Indian While evidence for early archaeological sites may have subcontinent can only be qualitative at this moment (see been lost in the geomorphologically dynamic Indus River also the work of Patterson et al. [2010], who make a floodplain, the lack of early sites in Rajasthan is striking, qualitative comparison not with single sites, but with considering that the model sees a biogeographically favor- approximate timing contours). able environment and cultural setting for an early Neolithic: In addition, the coexistence of Mesolithic and Neolithic one should ask whether an agricultural center on the Indo- subsistence in the archaeological record complicates the Gangetic divide is undetected so far. First evidence for comparison between model and data because the degree of northern Indian agriculture could come from further east, both subsistence styles cannot be quantified with the current from Lahuredawa in the middle Ganges plains, where evi- data. This coexistence is evident in the long-term continued dence for domesticated rice appears as early as 6500 B.C. 112 SIMULATION OF INDUS VALLEY NEOLITHIC

[Tewari et al., 2008] (but see the work of Fuller [2011a] for a the first approximation, the increases in simulated population debate on the rice morphology). Less favored by the model size and in reconstructed artifact density correspond: the are the valleys along the Indo-Iranian Plateau, where broad model correctly represents the emergence of dense settle- subsistence possibilities are seen as one precondition for the ments. Statistically, however, no meaningful relationship rise of the IVC and where agropastoralism arose before 6500 should be derived from only four different periods. B.C. [Jarrige, 1995]. The model might underestimate the In summary, there is fair qualitative agreement between the potential for agropastoralism in this area because of its coarse reconstructed and the simulated transition to agropastoral- spatial scale. ism, both in the spatiotemporal distribution of agropastoral From a regional perspective, the simulated east-west direc- sites and simulation regions, as well as the overall artifact tion of agropastoralism contradicts the eastward trend seen in and population density. Important lessons from the model are the site data. The westward direction had been suggested in that it corroborates an independent agropastoral center and earlier literature on the Indus Valley tradition: Wheeler that it suggests potential for earlier Indian agropastoralism. It [1959] wrote of “movement of the Neolithic from Burma if will be left to further studies to quantify the agreement and to not behind.” In their simulation of the spread of agriculture elaborate on the reasons, both in archaeology and in the into the Indian continent, Patterson et al. [2010] assume such simulation, of discrepancies. a Southeast Asian center in addition the Mehrgarh location. While Wheeler’s [1959] earliest dates (1000 B.C.) have been 4. CONCLUSION pushed further back by more recent datings, this westward view cannot be upheld from either artifacts [Law, 2007] or We presented a novel numerical simulation of the pre- the description of sites in central India (3000 B.C.) [Boivin et Harappan Neolithic of the Indus Valley tradition in the al., 2008], the Indus plain (4000 B.C.) [Mughal, 1990], context of representative radiocarbon dates from material northern India (5000 B.C.) [Sharma et al., 1980], or Baluchi- culture. Simulated population size is qualitatively reflected stan (6500 B.C.) [Jarrige, 1995]; all of these demonstrate the in artifact frequency in four pre-Harappan periods. Within opposite eastward and then southward trend. the uncertainties of the coarse chronology, the model pre- The simulated earliest agropastoralism in the Indus Valley dicts the spatiotemporal structure of the Neolithic transition region is independent of the two prescribed early centers of of this area fairly well. Our simulation points to a possible agropastoralism in Eurasia, i.e., China and Southwest Asia; earlier Neolithic in northern India than seen in the data, and toward the end of KGM, however, a continuous band of early it corroborates an independent South Asian Neolithic. For agropastoralism ranges from the northern Indian subcontinent quantitative model-data comparisons and clarification of the through the northern Southeast Asia to China (not shown). connection to Southwest Asia, better chronologic control of This South and East Asian agricultural area connects to the the pre-Harappan material as well as more evidence from the Southwest Asian center during the BBM period. The simulat- intervening Iranian Plateau are needed. ed independent emergence of South Asian agropastoralism corresponds well to evidence for independent domestication APPENDIX A: of cattle and rice [Chen et al., 2010; Fuller, 2011b]. Though SUPPLEMENTARY MATERIAL not completely detached from Southwest Asian domesticates assemblage in the data, this separation is visible in the delayed The simulated data have been permanently archived and simulated transition to agropastoralism in the area between the are freely accessible from PANGAEA (Data Publisher Indo-Gangetic and the Southwest Asian founder centers. for Earth & Environmental Science, Bremerhaven, Building on the earlier idea of Sanders et al. [1979], we Germany, http://www.pangaea.de) as a netCDF data set relate the number of artifacts in the database (corrected for (C. Lemmen and A. Khan, Simulated Transition to taphonomic bias) to population estimates from our simula- Agropastoralism in the Indus Valley 7500-3000 BC, tion (Figure 2b). We summed the total simulated population http://doi.pangaea.de/10.1594/PANGAEA.779706, 2012). in each half-degree model grid cell within 200 km distance to GLUES is open source software and can be downloaded fi the nd sites and obtain a population trajectory, which shows from http://glues.sourceforge.net. an increase from 0.2 million at 6000 sim B.C. to 2.5 million at 3500 sim B.C.; this corresponds to a population density of Acknowledgments. We thank R.W. Williams for publicly provid- –2 0.1 and 1.3 km , respectively. At 3500 sim B.C., our esti- ing his extensive data set for further analysis; we thank three anony- mate is higher than the one by Patterson et al. [2010] (0.2 mous reviewers and the editor for their critical and helpful comments – km 2), as should be expected from our calculation from the on earlier versions of this manuscript. C.L. acknowledges financial surroundings (within 200 km) of mostly settlement sites. To support from the DFG priority program Interdynamik (SPP 1266), LEMMEN AND KHAN 113

A.K. acknowledges financial support from the Higher Education Gupta, A., D. Anderson, and J. Overpeck (2003), Abrupt changes in Commission of Pakistan and the PACES program of the Helmholtz the Asian southwest monsoon during the Holocene and their Gemeinschaft. This paper benefitted greatly from ideas presented and links to the North Atlantic ocean, Nature, 421, 354–356. discussed at the AGU Chapman Conference on Climates, Past Land- Jarrige, C. (1995), Mehrgarh: Field Reports 1974–1985, From scapes, and Civilizations, held in Santa Fe, New Mexico, USA, Neolithic Times to the Indus Civilization, Dep. of Cult. and March 2011. Maps were prepared with Natural Earth free vector and Tourism, Gov. of Sindh, Sindh, Pakistan. raster map data available from http://naturalearthdata.com. Kenoyer, J. M. (2006), Cultures and societies of the Indus Tradition, in Historical Roots in the Making of ‘The Aryan’, edited by R. REFERENCES Thapar, pp. 21–49, National Book Trust, New Delhi, India. Kenoyer, J. M. (2008), The Indus Civilization, in Encyclopedia of Ackland, G. J., M. Signitzer, K. Stratford, and M. H. Cohen (2007), Archaeology, edited by D. M. Pearsall, pp. 715–733, Academic, Cultural hitchhiking on the wave of advance of beneficial tech- New York. nologies, Proc. Natl. Acad. Sci. U. S. A., 104(21), 8714–8719, Kisdi, E., and S. A. H. Geritz (2010), Adaptive dynamics: A doi:10.1073/pnas.0702469104. framework to model evolution in the ecological theatre, J. Math. Boivin, N., D. Q. Fuller, R. Korisettar, and M. D. Petraglia (2008), Biol., 61(1), 165–169, doi:10.1007/s00285-009-0300-9. First farmers in south India: The role of internal processes and Law, R. W. (2007), Building a South Asian archaeological site and external influences in the emergence and transformation of south rock & mineral resource “Gazetteer” in Google Earth, paper India’s earliest settled societies, Pragdhara, 18, 179–200. presented at Indus Archaeological Technical Training Workshop, Brovkin, V., J. Bendtsen, M. Claussen, A. Ganopolski, C. Kubatzki, Global Heritage Fund, Baroda, India, 17–20 March. V. Petoukhov, and A. Andreev (2002), Carbon cycle, vegetation, Law, R. W. (2011), Linguistics, Archaeology, and the Human Past, and climate dynamics in the Holocene: Experiments with the vol. 11, Inter-Regional Interaction and Urbanism in the Ancient CLIMBER-2 model, Global Biogeochem. Cycles, 16(4), 1139, Indus Valley, Res. Inst. for Hum. and Nat., Kyoto, Japan. doi:10.1029/2001GB001662. Leemans, R., and W. P. Cramer (1991), The IIASA database for Chen, S., et al. (2010), Zebu cattle are an exclusive legacy of the mean monthly values of temperature, precipitation and cloudi- South Asia Neolithic, Mol. Biol. Evol., 27(1), 1–6, doi:10.1093/ ness of a global terrestrial grid, Int. Inst. of Appl. Syst. Anal., molbev/msp213. Laxenburg, Austria. Claussen, M., C. Kubatzki, V. Brovkin, A. Ganopolski, P. Hoelz- Lemmen, C. (2010), World distribution of land cover changes mann, and H. Pachur (1999), Simulation of an abrupt change in during Pre- and Protohistoric Times and estimation of induced Saharan vegetation in the Mid-Holocene, Geophys. Res. Lett., 26 carbon releases, Géomorphologie, 4, 303–312. (14), 2037–2040. Lemmen, C., and K. W. Wirtz (2010), Socio-technological revolu- Doose-Rolinski, H., U. Rogalla, G. Scheeder, A. Lückge, and U. von tions and migration waves re-examining early world history with Rad (2001), High-resolution temperature and evaporation changes a mathematical model, in The Spread of the Neolithic to Central during the Late Holocene in the northeastern Arabian Sea, Paleo- Europe, RGZM Tagungen, vol. 4, edited by D. Gronenborn and ceanography, 16(4), 358–367, doi:10.1029/2000PA000511. J. Petrasch, pp. 25–38, Römisch-Germanisches Zentralmuseum, Fisher, R. A. (1930), The Genetical Theory of Natural Selection, Mainz, Germany. Dover, New York. Lemmen, C., D. Gronenborn, and K. W. Wirtz (2011), A simulation Fleitmann, D., et al. (2007), Holocene ITCZ and Indian monsoon of the Neolithic transition in Western Eurasia, J. Archaeol. Sci., dynamics recorded in stalagmites from Oman and Yemen (Soco- 38(12), 3459–3470, doi:10.1016/j.jas.2011.08.008. tra), Quat. Sci. Rev., 26(1–2), 170–188, doi:10.1016/j.quascirev. Moulherat, C., M. Tengberg, J.-F. Haquet, and B. Mille (2002), First 2006.04.012. evidence of cotton at Neolithic Mehrgarh, Pakistan: Analysis of Fontes, J.-C., F. Gasse, and E. Gibert (1996), Holocene environmen- mineralized fibres from a copper bead, J. Archaeol. Sci., 29(12), tal changes in Lake Bangong basin (Western Tibet). Part 1: Chro- 1393–1401. nology and stable isotopes of carbonates of a Holocene lacustrine Mughal, M. (1990), The decline of the Indus civilization and the core, Palaeogeogr. Palaeoclimatol. Palaeoecol., 120,25–47. Late Harappan period in the Indus valley, Lahore Mus. Bull., 3(2), Fuller, D. Q. (2006), Agricultural origins and frontiers in South 1–17. Asia: A working synthesis, J. World Prehist., 20(1), 1–86, doi:10. Patterson, M. A., G. R. Sarson, H. Sarson, and A. Shukurov (2010), 1007/s10963-006-9006-8. Modelling the Neolithic transition in a heterogeneous environ- Fuller, D. Q. (2011a), Finding plant domestication in the Indian ment, J. Archaeol. Sci., 37(11), 2929–2937, doi:10.1016/j.jas. subcontinent, Curr. Anthropol., 52(S4), S347–S362, doi:10.1086/ 2010.07.003. 658900. Peros, M. C., S. E. Munoz, K. Gajewski, and A. E. Viau (2010), Fuller, D. Q. (2011b), Pathways to Asian civilizations: Tracing the Prehistoric demography of North America inferred from radio- origins and spread of rice and rice cultures, Rice, 4(3–4), 78–92, carbon data, J. Archaeol. Sci., 37(3), 656–664, doi:10.1016/j.jas. doi:10.1007/s12284-011-9078-7. 2009.10.029. 114 SIMULATION OF INDUS VALLEY NEOLITHIC

Petraglia, M. D., et al. (2009), Population increase and environmen- Sharma, G. R., J. D. Clark, B. K. Thapar, and D. Mandal (Eds.) tal deterioration correspond with microlithic innovations in South (1980), History to Prehistory: Archaeology of the Ganga Valley Asia ca. 35,000 years ago, Proc. Natl. Acad. Sci. U. S. A., 106(30), and the Vindhyas, Univ. of Allahabad, Dep. of Ancient Hist., 12,261–12,266. Cult., and Archaeol., Allahabad, India. Petraglia, M. D., M. Haslam, D. Q. Fuller, N. Boivin, and C. Smith, B. D. (1998), The Emergence of Agriculture, 231 pp., Clarkson (2010), Out of Africa: New hypotheses and evidence Freeman, New York. for the dispersal of Homo sapiens along the Indian Ocean rim, Ann. Surovell, T. A., J. Byrd Finley, G. M. Smith, P. J. Brantingham, and Hum. Biol., 37(3), 288–311, doi:10.3109/03014461003639249. R. Kelly (2009), Correcting temporal frequency distributions for Possehl, G. L. (1998), Sociocultural complexity without the state: taphonomic bias, J. Archaeol. Sci., 36(8), 1715–1724, doi:10. The Indus Civilization, in The Archaic States, edited by G. M. 1016/j.jas.2009.03.029. Feinman and J. Marcus, pp. 261–291, Sch. of Am. Res. Press, Tewari, R., R. Srivastava, K. K. Sing, I. B. Sing, and K. Saraswat Santa Fe, N. M. (2008), Early farming at Lahuradewa, Pragdhara, 18, 347–373. Possehl, G. L. (2002), The Indus Civilization: A Contemporary Wheeler, M. (1959), Early India and Pakistan: To Ashoka, 190 pp., Perspective, 276 pp., AltaMira Press, Walnut Creek, Calif. Praeger, New York. Ricketts, R. D., T. C. Johnson, E. T. Brown, K. A. Rasmussen, and Wirtz, K. W., and C. Lemmen (2003), A global dynamic model for V. V. Romanovsky (2001), The Holocene paleolimnology of the Neolithic transition, Clim. Change, 59(3), 333–367. Lake Issyk-Kul, Kyrgyzstan: Trace element and stable isotope Wirtz, K. W., G. Lohmann, K. Bernhardt, and C. Lemmen (2010), composition of ostracodes, Palaeogeogr. Palaeoclimatol. Pa- Mid-Holocene regional reorganization of climate variability: laeoecol., 176, 207–227. Analyses of proxy data in the frequency domain, Palaeogeogr. Samuel, G. (2000), The Indus Valley civilization and early Tibet, Palaeoclimatol. Palaeoecol., 298(3–4), 189–200, doi:10.1016/j. Senri Ethnol. Rep. 15,pp.651–670, Natl. Mus. of Ethnol., palaeo.2010.09.019. Osaka, Japan. Sanders, W., J. Parsons, and R. Santley (1979), The Basin of Mexico, Academic, New York. Shaffer, J. (1992), The Indus Valley, Baluchistan and Helmand traditions: Neolithic through Bronze Age, in Chronologies in Old A. Khan and C. Lemmen, Institute of Coastal Research, World archaeology, 3rd ed., edited by R. Ehrich, pp. 441–464, Helmholtz-Zentrum Geesthacht, Max-Planck-Str. 1, 21205 Geesthacht, Univ. of Chicago Press, Chicago, Ill. Germany. ([email protected]; [email protected]) Mid-to-Late Holocene Agricultural System Transformations in the Northern Fertile Crescent: A Review of the Archaeobotanical, Geoarchaeological, and Philological Evidence

Simone Riehl,1,2 Konstantin Pustovoytov,3,4 Aron Dornauer,3 and Walther Sallaberger5

The region of the northern Fertile Crescent experienced dramatic changes in the political and cultural life of its societies during the mid-late Holocene period (approximately 3000–1000 calibrated years B.C.). The range of these changes in terms of agricultural production, as well as their interrelationship with climate, is poorly understood. We review and highlight the transformations of agricultural systems, and what might have triggered them, through an interdisciplinary approach based on archaeobotanical, geoarchaeological, and philological data from a series of archaeological sites in northern Mesopotamia. The archaebotanical record suggests changes in crop cultivation at the transition from the Early Bronze Age to the Middle Bronze Age (MBA). The general pattern of the MBA manifests itself also in the Late Bronze Age sites, whereas the Iron Age sites reveal changes such as the extension of free-threshing wheat occurrence and a return of flax. These changes have been set in comparison to fluctuations of the stable carbon isotope composition in seeds and to the dynamics of irrigation networks documented by pedosediment profiles in the field, as well as by textual sources. According to the evidence, a number of reasons can be considered responsible for societal change during the Bronze Age, such as changing climatic and environmental conditions, increasing societal, political, and economic complexity, population growth and related problems of sustainability and warfare, which were all interwoven through feedback mechanisms. Whether we call these developments “collapse” or “recommencement” remains a matter of opinion.

1. INTRODUCTION 1Institute for Archaeological Science, University of Tübingen, Tübingen, Germany. Transformations of agricultural systems may have oc- 2Senckenberg Center for Human Evolution and Palaeoecology, curred for many reasons, the most commonly established of Tübingen, Germany. which is climate change, which has been intensively dis- 3Altorientalische Philologie, Institut für Archäologische Wis- cussed for the past two decades regarding questions on the senschaften, Albert-Ludwigs-Universität Freiburg, Freiburg, Germany. sustainability of our world’s economy (e.g., Stern [2007] and 4 Institute of Soil Science and Land Evaluation, University of Ho- others). Transformations of human systems are, however, henheim, Stuttgart, Germany. inextricably related to changes in human society and culture. 5Institut für Assyriologie und Hethitologie, Ludwig-Maximilians- There is common agreement upon the role of historical and Universität München, Munich, Germany. sometimes archaeological information in the expansion of Climates, Landscapes, and Civilizations consciousness and, in particular, in the understanding of Geophysical Monograph Series 198 societal and cultural changes and their laws of causality. In © 2012. American Geophysical Union. All Rights Reserved. western civilizations, causality research includes determin- 10.1029/2012GM001221 isms, which are generally coined by the disciplines involved 115 116 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT in the study [Riehl, 2009b, 2012a, 2012b]. As recent studies, tion in specific ecological regions and simultaneous socio- however, on agricultural change in the past suggest, causality economic restrictions, disabling adaptation to changing research has to take multiple causes, feedback mechanisms, environmental conditions through increased social complex- and reciprocal amplification into account to develop realistic ity [cf. Wilkinson, 1997]. Despite the existence of a large models [Riehl, 2009b]. When considering agricultural amount of historical information, a systematic investigation change today and in the past, we need to abandon the search of the interrelationship of agricultural development and po- for single causes in the development of ancient societies and litical change is still pending. instead integrate environmental, economic, and social factors into the causal chain. This involves an integrative and inter- 1.1.1. Flourishing and declining of city-state networks disciplinary approach that we present below. during the Early Bronze Age. A period of ruralization with Our approach of using multiple lines of evidence includes only small administrative centers in northern Syria and (1) the analysis of archaeobotanical crop data from 114 southeast Anatolia, marked through the end of the Uruk archaeological sites in Upper Mesopotamia in order to re- expansion (the final prehistoric cultural sequence of Meso- construct ancient crop production patterns, as well as (2) potamia between 3900/3700 and 3100/2900 B.C.), has been stable carbon isotope analysis of ancient cereal grains as a described by Akkermans and Schwartz [2003] for the begin- proxy of past water stress on crop species which would have nings of the Early Bronze Age (EBA) (around 3000 B.C.). affected yields, (3) the evaluation of geoarchaeological re- Local traditions reemerged, and rural communities, charac- cords to detect ancient land use, and (4) the textual evidence terized by economic specialization, were integrated into a from cuneiform sources on ancient crop management. Taken larger economic network. Archaeologists refer to high cul- together, these information sources permit a well-balanced tural diversity as occurring throughout the entire region, a reconstruction of the interplay of environmental and social situation that some researchers attribute to the north-south factors of agricultural transformation processes. gradient of precipitation resulting in rain-fed agriculture in The historical sequence considered here roughly dates be- the north and irrigation in the south [cf. Jas, 2000]. Accord- tween 3000 and 1000 calibrated (cal) years B.C. When attrib- ing to some scholars, this led to the existence of smaller cities uting absolute dates to historical events mentioned in different in the north with higher populations in the outlying villages, textual sources, we follow the “Middle Chronology” (MC), compared to the south [Van De Mieroop, 2003]. which most likely needs to be lowered by circa 50 years after Generally speaking, cities of various rank evolved after 2100 B.C., depending on the assumed length of the “Dark roughly 2900 B.C. in Upper Mesopotamia and Syria. City- Age,” but only by about 20 years or less before circa 2150 state indicates an accumulation of political power through B.C. It is assumed that MC more or less equals radiocarbon agricultural surplus production. Cooperation between inde- dates. Throughout the text, we differentiate between radiocar- pendent cities has been documented [Van De Mieroop, bon data as B.C. placed behind the year and historical data 2003]. Early administrative structures and the appearance of according to the “Middle Chronology” as MC placed before cuneiform writing in the middle of the third millennium B.C. the year (for a detailed discussion over the dispute of middle are indicators of a complex organization within the city-state, versus short chronology in ancient Near Eastern Archaeology, as defined by the largest capital cities (like Tell Brak, Tell see the works of Pruzsinzsky [2009] and Mebert [2011]). Khuera, or Tell Hariri/Mari). The network of city-state Another terminology note: We avoid using the term “col- stretched from lowland Sumer to Upper Mesopotamia and lapse” in our own interpretation due to any of its vague or even to Syria (Tell Mardikh/Ebla) [Sallaberger, 2007, negative connotations [cf. Tainter, 1988; McAnany and Yof- 2012a]. fee, 2010; Riehl, 2012a]. The texts discovered at Tell Beydar reveal the best evidence for the internal communal organization of the city- 1.1. Changes in the Political, Cultural, and Economic Life state. Studies of these documents show that the central (Circa 3000–1000 cal years B.C.) organization included all or at least the large majority of the inhabitants of the city. Agriculture was organized as collec- Textual evidence and numerous finds from archaeological tive labor of the urban center and of dependent smaller excavations directly or indirectly emphasize the role of po- settlements [Sallaberger and Ur, 2004]. This organization litical and cultural factors in the development of agricultural corresponds largely to the one known from lowland Meso- production. Some archaeologists even attribute causality to potamia, despite the differences in irrigation, which are well agriculture for the decline of early societies. Agricultural documented in the south. Collective labor included the cen- mismanagement through overexploitation may be assumed tral allocation of human resources, draft animals (donkeys, [Hole, 2007], which followed an agglomeration of produc- oxen), and tools for work in the fields. At the time of harvest, RIEHL ET AL. 117 practically the whole population of a settlement was obliged Shimanum, which was probably located in the Upper Tigris to cooperate. The grain was stored in communal granaries region at the Tur Abdin. Mari and Ebla continued to be the and distributed to the population according to status, profes- dominant centers of the region, even after the fall of the Third sion, sex, and age. The central organization also controlled Dynasty of Ur (MC 2003). They were able to maintain good the large animal flocks of sheep and goats that were handed relations with Ur’s successor Isin [Sallaberger, 2007]. over to herdsmen [Sallaberger, 2004]. Communal organization included craftsmen who were given their wages in grain, 1.1.2. Competition for power between small states and whereas some human resources were directly linked to the agricultural underproduction in the Middle Bronze Age. royal palace (e.g., metal and textile production, military Upper Mesopotamia and Syria are characterized by contin- officials, royal court). The secondary settlements contributed uous change in both political regimes and economic struc- to the upkeep of the state by delivering grain, caring for royal tures throughout the Middle Bronze Age (MBA) (circa animal herds (especially equids), and offering labor service 2000–1500 B.C.), the period of the Amorite kingdoms. in state projects or military expeditions. The communal Rivalry and political conflict were as numerous as before organization documented also at Ebla or pre-Sargonic Mari [Klengel, 1992; Heimpel, 2003], and a general trend had (before 2350 B.C.) can be seen as a typical feature of the developed within the entire area, namely, competition for society and economy of early Mesopotamia that lasted power between small states gathered around the more in- through the third millennium. fluential powers: Qatna, Yamkhad (Aleppo), Mari, Upper The system of city-state came to an end when the accu- Mesopotamia or Assur, Babylon, Ešnunna, Larsa, and Elam mulation of power in a few centers such as Mari, Ebla, (from Syria in the northwest to Iran in the southeast). In Nagar, or Akkad led to wars of unprecedented intensity. central Anatolia, the Old Hittite kingdom was gaining in Centers like Ebla, Mari, or Nagar were destroyed around power. MC 2330–2320 and paved the way for the military success Whether these processes changed the economic organiza- of Sargon of Akkad (MC 2353–2314) [Sallaberger, 2007, tion cannot directly be detected in the archaeological or the 2012a]. textual record. It has been speculated that under Hattusili I, For Upper Mesopotamia, the end of the early city-state the Hittites expanded southward with the aim of acquiring meant a strong reduction in urban settlements for the last larger cereal fields in northern Syria than available in their third of the third millennium, with only a few centers of own country [Van De Mieroop, 2003]. In the west, new continuous importance at the northern fringe of the Jezirah influential cities, such as Qatna, developed during the plain [cf. Ristvet and Weiss, 2005]. The few remains of MBA. The geopolitical role of Ebla seems to have shifted administrative archives from these sites (Tell Mozan/Urkesh, the seat of the rulers of Yamkhad in the early MBA to the Tell Brak/Nagar, Tell Leilan/Shekhna) indicate the continu- old cultic center of Aleppo. For Babylonia, there is no ation of a similar communal organization of resources and indication of economic decline in the first three centuries labor as in the preceding period, at least at the urban centers. of the second millennium. Mari, in the middle Euphrates This, however, has to be evaluated against the general reduc- region, maintained its economic and political power [cf. tion of urban settlement in Upper Mesopotamia in the late Charpin and Ziegler, 2003]. Its economy was strongly third millennium and, at the same time, the appearance of influenced by interaction with nomadic pastoralists [cf. nomadic people, the Amorites. The evidence “suggests an Fleming, 2004]. Mari also played an important role in trade ethnogenesis of Amorite nomads meaning that a changing [Stol, 2004], but its subsistence economy was mainly based lifestyle of the former urban inhabitants of Upper Mesopo- on extensive irrigation agriculture in the Euphrates and tamia towards nomadism also included the adoption of the Khabur region. language of the nomads, Amorite” [Sallaberger, 2007, The archives of the palace of Mari during the reign of p. 450]. Zimri-Lim (early MC eighteenth century) inform us on in- Mari, located on the Middle Euphrates, was able to recover stitutional practices of agriculture based on thousands of after its destruction by Sargon of Akkad, becoming, like legal and administrative texts and letters from the viewpoint Ebla, a leading power during the period of the Third Dynasty of the palace [van Koppen, 2001]. The dominant role of of Ur, which ruled in Mesopotamia (MC 2110–2003). The palace estates and households of officials, and the absence regions of Upper Mesopotamia and Syria covered by the of temple households, is a feature that the institutional agri- study presented here never became parts of the state of Ur, culture of Mari shares with other Upper Mesopotamian states though good diplomatic relations existed with the dominant of this time period (e.g., Tell al-Rimah, Shekhna). The offi- centers, foremost among them being Mari and Ebla, fol- cials in charge were personally and financially accountable to lowed by Urshu, probably in the region of Gaziantep, and the palace for successful production. Officials were also 118 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT responsible for managing access rights to irrigation water. 1.1.3. Globalization, warfare, and political decline during The ikkaru, “farmer; leader of a plough team,” managed the the Late Bronze Age (LBA). During the LBA (circa 1500– basic work units in order to meet production goals and was 1200 B.C.), Syria, the region between the Mediterranean Sea responsible for communication with the administration. and the Euphrates, became a focus of interest for three large Generally, as evident from the texts as well as from the political powers: first, the Hurrian state of Mittani, which archaeobotanical remains, the main cereal was barley flourished between the sixteenth and the early fourteenth (še’um), though evidence also exists for wheat species (kib- century; second, the Hittite state of Hatti from Anatolia, tum and possibly burrum). Mari’s agriculture did not so especially after the successful conquests of Suppiluliuma I much lack arable land, as it did sufficient manpower. This in the mid-thirteenth century B.C.; and third, the Egyptian point, along with the issue of contemporary political con- pharaohs who entered inner Syria from the Mediterranean flicts, led to local underproduction [Lafont, 2000], forcing coast between the fifteenth and thirteenth centuries. Before the palace to acquire barley on the market. This explains why 1340 B.C., Egypt and Mitanni competed for control over the military commanders were responsible for the organization region; after that date, when Mitanni was turned into a vassal of additional personnel for the harvest [van Koppen, 2001]. of the Hittite king, Egypt and Hatti competed for control over Shortages in the labor force are also known for other Meso- the Mitanni state. By 1300 B.C., the Hittites had expanded potamian societies, e.g., the Middle Assyrian Empire in the their system of vassal states as far south as Qadesh, culmi- north or the Ur III period in southern Mesopotamia. nating first in the battle with Egypt in 1274 B.C. and finally The end of the MBA is marked by the Hittite attack on with the peace treaty between Hatti and Egypt in 1259 B.C. Yamkhad in northwest Syria and the sack of Babylonia in [Van De Mieroop, 2003]. MC 1595. Although this led to a complete transformation of We can consider this from a wider perspective, since political structures, with a power vacuum opening up in contacts overseas were extended to also include islands like formerly dominant regions, some scholars have pointed to Cyprus or Crete after 1500 B.C. The cultures of Syro- a continuous transition to the LBA in the material culture Mesopotamia were seemingly linked to a maritime trading [Akkermans and Schwartz, 2003]. This is especially true for system, which enabled far-reaching economic and cultural the region we focus on here, where states like Qatna, exchange. This was both a sign of, and a means for, the Yamkhad/Halab (Aleppo), Hana, Mittani, and Hatti contin- international character of a large number of territorial states ued to exist, even gaining in importance. Mittani, the state in from western Iran to the Aegean, from Anatolia to Nubia the region with a primarily Hurrian population along the [Van De Mieroop, 2003]. Van De Mieroop describes five Upper Tigris and in Upper Mesopotamia, became the domi- zones of political unification and centralization of power that nant power in northern Mesopotamia and Syria. Hana, the experienced a period of prosperity during the LBA. Besides kingdom at the Middle Euphrates, lost its power, and the Egypt, Hatti, Mittani-Assyria, and Babylonia, he also includes region became part of the long-lived Kassite state of Baby- the Middle Elamite Kingdom in western Iran. Within these lonia that had arisen after the fall of Babylon (MC 1595, but regions, several cities and their hinterlands were tied together later in absolute chronology) in the so-called “Dark Age,” by economic integration, while the regions, themselves, seem when, according to the texts, resources from Babylonia came to have maintained a coherent trade network that resembles almost to an end. The economic and agricultural conse- the later aims of internationalization and globalization, of quences of these political upheavals are difficult to grasp due course not free of rivalry. An exception to this trend toward to the almost total disappearance of relevant written sources. larger regional political units was Syro-Palestine where the Apparently, the loss of the connection to Babylonia in its basic system of small states concentrated around a single city “Dark Age” played an important role, since Babylonian persisted [Van De Mieroop, 2003]. These small states sur- scholars were vital for the dissemination of writing in the vived as allies of the larger political powers. regions to the west. One may recall that Babylonian scribes The decline of Hatti and the important cities under its were active especially in the early periods at Hattusa or that control around 1200 B.C. has been explained through the writing was apparently little used in the state of Mittani, attacks by the so-called “Sea Peoples,” by rebellion in the which flourished in the “Dark Age” of Babylonian culture territories in the west and by possible crop failures. Similarly, between MC 1595 and the fourteenth century. The spread of the end of the Mycenaean culture in the eastern Mediterra- cuneiform in the fourteenth to thirteenth centuries both at nean has been attributed to internal conflict, population palaces (like Alalakh IV) and in private households (Mes- movement, and warfare. It has been assumed that economic kene/Emar, Ekalte) may not only be taken as evidence for a decline may have played a causal role in the collapse of LBA flourishing economy but also for renewed contacts with civilizations. It is most likely that complex interactions be- Kassite Babylonia. tween political and ecological factors were responsible for RIEHL ET AL. 119 these massive supraregional changes. How these changes however, clear that regional diversity played a role in the influenced agricultural organization still remains relatively particular climate effects in the different geographic regions. unclear. A bioarchaeological specification of economic de- Anthropogenic impact during the Bronze Age played a cline has so far not been determined. greater role in vegetation development than it did during the The “Dark Age” following the end of the LBA varied in early Holocene, a fact which makes interpretation of the duration within the different regions. In Assyria, which palaeoclimate proxy data more complex [Roberts et al., survived the 1200/1180 B.C. breakdown of the LBA sys- 2004]. This has been overcome by including stable isotope tems, a period of decline lasted from the early eleventh data [e.g., Wick et al., 2003], allowing for differentiation century until circa 935 B.C. Little is known about this of climatic and human factors in the degradation of the period, but it seems clear that an almost complete restructur- vegetation. ing of the society must have taken place due to large-scale Rapid climate change events (RCC) considered to be global, nomadization, migration, internal population movements, occur in a larger number of distant regions and generally can and technological and trade network changes, while there be related to Bond cycles. For our study here, two RCCs are is also evidence for continuity, e.g., the maintenance and relevant, the “4200 B.P. event” (Bond cycle number 3), restoration of the Lower Khabur and Middle Euphrates which has been related in the Near East with the transition canals by local rulers [Fales, 2008]. With the beginning of from the Early to the MBA [Staubwasser and Weiss, 2006], the first millennium B.C., the political situation in the Near and the climate changes between 1200 and 900 B.C., which East had become more stable, and a new network of states correlates with the “Dark Age” between the LBA and the arose, which becomes tangible in the ninth century B.C. by Iron Age (IA). written records again. In the most relevant palaeoclimate proxy archives, includ- Beginning with the campaigns of Ashur-Dan II (934–912 ing the speleothems from Soreq Cave [Bar-Matthews and B.C.) and Adad-nerari II (911–892 B.C.), Assyria success- Ayalon, 1998, 2011] and pollen and stable oxygen data from fully started the reconquest of Upper Mesopotamia. Shalma- Lake Van [Wick et al., 2003], for the geographic area con- neser III (822–811 B.C.) has finally conquered all of Upper sidered here, the 2200 B.C. climate change indicates more a Mesopotamia. Phoenicia, Israel, the Aramean, and the Neo- long-term shift to more arid conditions than an event. At Van, Hittite states, Urartu and Babylonia, were forced to pay change in the data is rather visible in the pollen, with a strong tribute. Tiglat-Pileser III (744–727) terminated a short period reduction in oak and other tree species, than in the oxygen of stagnation. He and his successors subjugated nearly all of isotope record [Litt et al., 2009]. As discussed above, this the ancient Near East including Babylonia, the Levante, and could be the result of anthropogenic impact on the vegeta- parts of Egypt, Anatolia, and western Iran. tion. However, occurring simultaneously with the “4200 B.P. Features of the Neo-Assyrian agriculture are the planned event,” the lake level of Kinneret decreased [Hazan et al., distribution of rural labor forces, the installation of large- 2005]. The lake level at Van also decreased after 2000 B.C. scale irrigation systems, and a centralized program of rural [Lemcke and Sturm, 1997; Wick et al., 2003], and carbonate land settlement policy [Bernbeck, 1993; Bagg, 2000; Mor- cutans on objects from southern Anatolia stopped growing andi Bonacossi, 2008]. The dense rural population was able around this time [Pustovoytov et al., 2007a, 2007b]. to produce more agricultural goods to feed the dramatically Chronological discrepancy in the appearance of drier con- increased urban population in central Assyria [cf. Wilkinson ditions, generally linked to internal methodological problems et al., 2005]. of dating, could alternatively be related to time differences in the abandonment of EBA settlements, which started earlier 1.2. Environmental Change (Circa 3000–1000 cal years B.C.) in the Euphrates than in the Khabur region [Kuzucuoglu and Marro, 2007; Riehl and Bryson, 2007]. Despite general sparseness of pollen reservoirs and low In the Levant, modern precipitation values were within preservation rates for pollen in the Near East, palaeoclimate a similar range of 350–580 mm as those since 5000 B.C., proxy data sets have created a solid framework of Holocene as calculated by Bar-Matthews et al. [1997]. According to climate history. Bond’s data, a global cooling event between 1200 and 700 A period of increasing moisture, with the beginning of the B.C. (Bond cycle no. 2) [Bond et al., 2001] should have Holocene, ends around 5000 B.C., when a general trend of affected wider areas and might also have had an impact in increasing aridity starts which has, among others, been the Near East between the end of the LBA and the IA II. linked to orbital parameters and was probably associated This corresponds well with the stable oxygen isotope data with a southward shift of the Intertropical Convergence Zone from Lake Van where increasing aridity is visible within [Wanner et al., 2008]. Despite this global component, it is, this sequence [Litt et al., 2009]. Aside from probable 120 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT chronological problems, the data imply regional variability tions regarding water stress on crop plants during their grain- with more stable moisture conditions in the Levant and filling period [Araus et al., 1999, 2001; Ferrio et al., 2005; stronger fluctuations inland in southern Turkey during the Riehl et al., 2008; Fiorentino et al., 2008]. Despite complex transitional period from the Bronze to the IA. causes of variation [Dawson et al., 2002], increased water The “1200 B.C. hypothesis” of increased aridity in north- stress accounts for the largest part of δ13C variation in arid to ern Mesopotamia between 1200 and 900 B.C. of Neumann semiarid environments [Tieszen, 1991]. and Parpola [1987] goes back to Carpenter’s local model on Carbon isotopes of the archaeological barley remains were the discontinuation of Greek civilization [Carpenter, 1966], measured in δ13C Vienna Peedee belemnite ‰. The changes 13 which was supported by archaeoclimatological models [Bry- over time in atmospheric CO2 (δ Cair) composition had to son et al., 1974; Bryson, 2005]. Neumann and Parpola’s be considered when samples from different periods and, in model [1987], which includes a causal argument for drought, particular, when ancient and modern sample values were famine, and social problems leading to political collapse, compared. Discrimination of δ13C(Δ13C) was conducted correlates with the Van record. Recently, Rohling et al. with a transfer function [Farquhar et al., 1982, 1989]. Values 13 [2009] suggested that the cooling period at the end of the of δ Cair are available from ice-core projects in Greenland LBA adversely affected the agricultural quality in the north- and Antarctica [Barnola et al., 2003]. According to ancient 13 13 ern and northeastern regions of the Aegean, thus triggering δ Cair, the following δ Cair values were used for the calcu- end-of-Bronze Age migrations. Rohling et al. [2009] empha- lation of discrimination: for the EBA, À6.3/À6.4, for the size that the Minoan eruption, which has been radiocarbon MBA and LBA, À6.5, and for the modern samples, À8. dated to 1627–1600 B.C. [Friedrich et al., 2006], is not A relatively comprehensive series of geoarchaeological causally related to their RCC, which they date to 1500–500 studies, including geographic information system (GIS)-based B.C. While the former date would be more indicative of a analysis of satellite imagery and modeling, has been published chronological relationship with the end of the MBA, the and helped with the identification of settlement and land use latter covers major parts of the LBA and the IA. patterns as aspects of early economic systems in some regions of the Near East [Goldhausen and Ricci,2005;Wilkinson, 2. METHODS 1997, 2003; Wilkinson et al., 2007; Deckers and Riehl, 2008; Altaweel, 2008]. The selective nature of such studies requires We reviewed multifaceted evidence from archaeobotany, continuous and systematic research on land use changes by geoarchaeology, and philology with the goal of developing a integrating available environmental data, the results from holistic view of the economic developments between 3000 geoarchaeological fieldwork, GIS-based analyses of satellite and 1000 B.C. (Figure 1). images, and the application of agronomic models, by combin- Results on crop cultivation patterns are heavily based on ing these data with palaeoclimatological, archaeological an archaeobotanical database in which more than 16 million (environmental and architectural), and philological data. seed records were gathered from archaeobotanical publica- The rich textual record of cuneiform tablets, which has tions on archaeological sites in the Eastern Mediterranean been estimated to number more than 500,000 texts [Streck, and the Near East. A reduced version of this database can be 2010], covers a broad range of information related to agri- found at www.ademnes.de. Crop proportions and ubiquities culture that has not yet been systematically analyzed. For our have been calculated within the database, and archaeobota- study, it was necessary to focus on grain, wine, fiber, and oil nical distribution maps were created with ArcGis. Further species, which are the most important crops mentioned in data derives from our own archaeobotanical work at a num- texts written in Sumerian and Akkadian. ber of sites, such as Tell Atchana [Riehl, 2010a], Tell Mozan The cuneiform record usually provides indirect informa- [Riehl, 2010c], Tell Halaf [Riehl and Deckers, 2008], Emar tion about agriculture. No guidelines exist for the farmer, at [Riehl, 2010b], Qatna, Tell Fadous [Riehl and Deckers, least not for Upper Mesopotamia (for the southern alluvium 2011], Zeraqon [Riehl, 2004], and Troy [Riehl, 1999]. see the work of Civil [1994]). The most important sources are As the nature of archaeobotanical data carries with it a administrative documents, i.e., lists of incoming and outgo- number of interpretative problems to crop distribution pat- ing goods, and tabulations of the goods in stock that were terns [Riehl, 2011], an independent method was considered issued by communal organization. The very existence of necessary to reveal whether climate fluctuations may have administrative texts is a sign of complex organization affected the crop yields and thus adaptive human behavior in concerned with the distribution of goods or the management crop production. of labor within its specific sphere of responsibility. Such Stable carbon isotope analysis has been in use for some documents stem from communal organizations within the time at archaeological sites in order to facilitate investiga- city-state of the third millennium [Ismail et al., 1996; Milano RIEHL ET AL. 121

Figure 1. Overview of the study area with archaeological sites from different periods that produced archaeobotanical data. Toponyms from the cuneiform sources and regions where geoarchaeological research has been conducted are shown. Isolines represent isohyets; information on geoarchaeological surveys mostly derives from Wilkinson [2000] and our own investigation. Visible site codes represent the following localities: ADA, ’Ain Dara; ALA, Tell Atchana; ARST, Arslan- tepe; ASHA, Tell Ashara; BEY, Tell Beydar; CHU (Tell Khuera), EM (Emar), GV (Gre Virike), HAL (Haleb/Aleppo), HAR (Tell Hariri/Mari), HH (Hassek Höyük), HI (Hajji Ibrahim), and HAD (Tell Hadidi; hidden), HOR (Horum Höyük), HWES (Tell Hwes), IMA (Imamoglu), KEL (Kamid el-Loz), KIH (Kinet Höyük), KOR (Korucutepe), KT (Kenan Tepe), KUH (Kurban Höyük), LEIL (Tell Leilan), MH (Mezraa Hoyük), NIM (Nimrud), RAW (Tell al-Rawda), SHAM (Tell Ras Shamra), SLK (Tell Selenkahiye), SWE (Tell es-Sweyhat), TAF (Tell Afis), TAR (Tell al-Raqa’i)], TB (Tilbeshar), TBD (Tell Bderi), TBR (Tell Brak), TEA (Tell el-Abd), TEMI (Tell Mishrifeh/Qatna), TH (Titriş Höyük), THAL (Tell Halaf), THT (Tell Hammam et-Turkman), TJT (Tell Jerablus Tahtani), TKA (Tell Karrana), TKE (Tell Kerma), TM (Tell Mozan), TMA (Tell Matsuma), TMUN (Tell Munbāqa),TNM (Tell Nebi Mend), TQQ (Tell Qara Quzaq), TQR (Tell Qarqur), TSABA (Tell Sabi Abyad), TSF (Tell Shiukh Fawqani), TSH (Tell Shekh Hamad), TUTU (Tell Bi’a), UEM (Umm el- Marra), UMB (Umbashi), YOR (Yorgan Tepe/Nuzi), and ZAG (Tell Zagan). et al., 2004] such as the organization of a palace like in MBA documents, though, presuppose a certain degree of knowl- Mari (MC eighteenth century) (cf. references in the work of edge of the available structures, of the terminology used in Charpin [2004]) or in MBA and LBA Alalakh [Dassow, the texts, of the function of the persons, who are listed mostly 2008; Zeeb, 2001] or from the provincial administration of by their names only, or of the role of various settlements. Assyria [Röllig, 2008; Radner, 2004; Jakob, 2003]. Many Furthermore, the administrative documents were issued for administrative texts provide exact data, e.g., the precise specific purposes like the calculation of grain for seeding or amount of grain delivered to a granary from various persons. for the feeding of animals. Researchers today face two es- Within an archive, both the numbers and the types of pro- sential problems. They must tackle the difficult terminology ducts are consistent and can be used to reconstruct a larger of Akkadian, the Semitic language of Mesopotamia and picture of a city-state’s agriculture. Common, everyday adjacent regions from the late third to the first millennium, 122 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT and of Sumerian, the language of the first script, which 3. RESULTS strongly influenced the cuneiform writing especially in its earlier periods. They must not only reconstruct the adminis- 3.1. Agricultural Transformations in the Archaeobotanical trative structure behind the names of the documents, but also and Stable Carbon Isotope Record use the documents in such a way as to contribute to current investigations. Despite these imperfections, administrative Considering the two transitional sequences that have been texts are the best textual sources for uncovering details into related to major climate fluctuations, the Early to MBA the economies and societies of the ancient world. transition and the LBA to IA transition, results on agricul- These sources, however, are absent in regions and/or in tural transformations during the former sequence have al- periods where writing was less widespread. The states located ready been published [Riehl, 2008, 2009a, 2011; Riehl et al., at greater distances from Mesopotamia often used writing 2008]. only or mainly within the palace, and thus, only the central The overall result of these studies demonstrates a reduction palace administration, including the management of the royal in the presence, proportion, and ubiquity of water-demanding treasury, is documented (e.g., Middle Elamite writing). Fur- species comparing EBA and MBA data sets. Other, more thermore, even in the case of widespread writing such as at stress-tolerant species, such as barley (Hordeum vulgare), LBA Meskene/Emar [Arnaud, 1986] or Tall Munbaqa/Ekalte increase in abundance throughout time. Today, a dominance [Mayer, 2001], we are missing administrative texts, perhaps of barley is a common feature in arid and semiarid environ- because large sectors of the economy were not organized ments due to its short reproduction cycle, which protects the collectively or because the documents were simply not crop from extensive drought stress. Although the ubiquity of found. We do have at our disposal private legal documents free-threshing wheat (Triticum durum/aestivum) is generally that only indirectly hint at the agricultural context, e.g., the high in the Bronze Age, suggesting routine usage in ancient acquisition or heritage of vineyards or fields [Mori, 2006]. households as a highly esteemed component in human diet, As the historical survey has made clear, periods of dramatic proportions are low in contrast to barley. In the MBA, free- change are followed by “Dark Ages” with an absence of threshing wheat in Syria is recorded with even lower propor- written sources. The historical and philological investigation tions than in the EBA sites. This is notable in view of the fact will therefore never permit a direct investigation of ongoing that free-threshing wheat has comparatively higher water changes, and even if sources were to exist, they would not requirements than barley. Garden pea (Pisum sativum), document the change of a system, but only a particular which has comparatively high water requirements, is wide- situation. In terms of philological and historical research, on spread and occurs with relatively high proportions during the changing agricultural systems in the Syro-Mesopotamian EBA, particularly in northern Syria, i.e., in an area with realm, describing these sites or regions and these periods as higher mean annual precipitation (>400 mm). During the exactly as possible becomes our most important task. Only a MBA, it occurs particularly in the part of Syria west of the bird’s eye perspective will allow us to discover long-term Euphrates in a probably wetter area, better suiting the water trends. Inclusion of the geoarchaeological and the archaeo- demand of the pulse crop. Stress-tolerant bitter vetch (Vicia botanical records helps us to estimate regional differences of ervilia), which is reported as a crop with considerable the various archives more precisely. For a few sites, both drought adaptation, but is not a part of the human diet today, philological and archaeobotanical evidence has been pub- played a comparatively minor role in EBA plant production. lished, e.g., for the thirteenth/twelfth century. Tell Sheikh However, it occurs with higher proportions in at least some Hamad (texts [Röllig, 2008], archaeobotany [van Zeist, of the MBA sites. Linseed (Linum usitatissimum) also oc- 2001; Frey and Kürschner, 1991]). In some cases, relevant curred more often and with higher frequencies during the texts and palaeobotanical samples may stem from the same EBA farther to the east in northern Mesopotamia. It virtually site and period, but have not been published with equal disappeared in the MBA. Seed flax is raised under a wide measure. Examples include the twenty-fourth century Tell range of conditions, but it is particularly intolerant of salinity Beydar/Nabada with published texts [Ismail et al., 1996; and high temperatures in terms of normal seedling growth, Milano et al., 2004], but archaeobotany is still to be pub- while fiber flax requires abundant moisture during the lished or Tell Khuera (texts [Jakob, 2003], archaeobotany to growing season. It has been argued that flax disappeared be published). In most cases, however, there is no textual with the end of the EBA because yields no longer justified equivalent to the archaeobotanical study of a site. the production costs [Riehl, 2009a]. Grape (Vitis vinifera L. Despite these methodological problems, administrative ssp. vinifera) was cultivated over a large area during the texts are invaluable documents for any research on the econ- EBA, although it occurs in larger proportions mainly in omies or societies of a given time period. areas with modern mean annual precipitation above 400 RIEHL ET AL. 123 mm. Cultivation during the EBA seems to have been prac- The rich new spectrum, as well as the more water-demanding ticed south of the natural distribution of the wild progenitor species already established in the LBA, was possible most (V. vinifera ssp. sylvestris). Allowing for fewer analyzed likely due to the use of irrigation. That irrigation was an MBA sites, grape cultivation generally seems to have been important factor in agricultural production is not only known reduced during this period, particularly in the central part of from the textual evidence but also indicated in the archae- northern Syria. obotanical remains. At Emar, for example, the wild plant In summary, the archaeobotanical crop patterns indicate a remains contain an increased number of salinization indica- disappearance or reduction in drought-susceptible species, tors during the LBA, implying continuous irrigation of crops particularly in flax, garden pea, and grape in many areas of [Riehl, 2010b]. the Near East from the EBA to the MBA, while drought- This argument correlates with the stable carbon isotope tolerant species tend to increase or remain unchanged. These record available from a couple of sites, such as Troy and Tell results correlate with increased water stress during the MBA Shiukh Fawqani, where there is only a slight difference in as has been recognized in the stable carbon isotope data from stable carbon isotope values from LBA and IA samples, barley remains in a number of sites [Riehl et al., 2008] which might also be related to irrigation. It has to be noted (Figure 2). that at Troy, the settlement was abandoned between roughly Transformations in agricultural patterns can also be ob- 1000 and 700 B.C. At Qatna, on the other hand, increased served between the LBA and the IA in many places through- water stress could be recognized in the IA barley grains, out the Near East by new crop species appearing on the compared to the LBA remains. scene, such as cotton (Gossypium sp.), pomegranate (Punica granatum), and cucumber (Cucumis sativus)[Riehl and Nes- 3.2. Evidence of Transformations of Agricultural Systems in bitt, 2003]. Sesame (Sesamum indicum) appears more regu- the Geoarchaeological Record larly in the archaeobotanical record than before, but as we discuss below, there is a discrepancy between the archaeo- In theory, transformations of agricultural systems over time botanical and the philological record. should be found documented in the geoarchaeological re- Changes in persisting crop species are particularly visible cord; in practice, however, their identification in the modern in free-threshing wheat and grape ubiquities, which are gen- landscape is problematic. Their context can be destroyed by erally increased in the IA compared to the previous period. erosion and overlain by accumulating sediments [Wilkinson,

Figure 2. (top) Stable carbon isotope values for archaeobotanical barley grains from different sites and (bottom) stable carbon isotopes from Soreq Cave (values have been converted into Δ13C to enable direct comparison between both records). 124 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT

2003]. Furthermore, they can become less recognizable stream downcutting and drying out of swamps [Rosen, through diagenetic alteration. When they are found in the 1997]. Similar results have been obtained for the nearby site landscape, the chronological context might still remain un- of Titriş Höyük [Rosen and Goldberg, 1995]. In Wadi Jagh- clear. Nevertheless, two lines of geoarchaeological evidence jagh (North Syria), a decrease in stable flow is assumed after are worth mentioning in connection with the mid-Holocene 2500 B.C. [Deckers and Riehl, 2007]. Furthermore, at Tell to late Holocene transformations of agricultural systems in Brak (North Syria), the early to mid-Holocene channels, the northern Fertile Crescent (Table 1). terminating presumably in the second millennium B.C., pre- First, multiple sedimentological findings at archaeological sent a contrast to deeper and narrow channels after 1000 B.C. sites in northern Mesopotamia, as well as within a broader [Wilkinson, 2003]. Studies in the wadis al-Walla and ash- context of the eastern Mediterranean, indicate a distinct Shallalah (Jordan) have demonstrated a high water stand, change in the regime of water streams at the transition from valley aggradation, and floodplain stability at 3500–2500 the third to the second millennium B.C. Relatively stable, B.C., followed by floodplain destabilization and incision at moderately strong water flows gave way to increasingly rare, 2500–2000 B.C. [Cordova, 2008]. Sedimentological evi- erratic, and stronger flows, which suggests an aridification dence from the Jordanian plateaus of Madaba and Dhiban tendency, in general, and/or more frequent droughts. At reveals an almost identical picture [Cordova et al., 2005]. An Kazane Höyük in the Harran Plain (SE Turkey), the sediment incision of streams occurring at the same time has also been profiles indicate relatively high moisture availability, stable found in the plains surrounding the Dead Sea [Donahue et water streams, and even swampy environments during the al., 1997]. All these observations correspond to a trend of Chalcolithic and the EBA, whereas the MBA contexts show relatively rapid aridification trend roughly around 2000 B.C.

Table 1. List of Mid-Holocene Geoarchaeological Phenomena Discussed in the Text Authors Location Period in B.C. Observed Phenomena Cordova et al. [2005] wadis al-Walla and ash-Shallalah 3500–2000 high-water stand, valley aggradation and floodplain stability 2500–2000 floodplain destabilization, incision Frumkin and Elitzur Dead Sea 2500–2000 rapid drop of water level [2002] Donahue et al. [1997] wadis in the Dead Sea plain 2500–2000 incision Cordova et al. [2005] river terraces on the Madaba 2500–2000 incision and Dhiban Plateaus Parker and Goudie lake sediments and sand dunes before 4000 lake sediments, C3-dominated savannah [2008] in the SE Arabian Gulf 4000–2000 lake sediments, C4-dominated grassland after 2000 active sand dunes Rosen [1997] alluvial sediments, channel fills, Chalcolithic-EBA steady channel flows, swampy environments Kazane Höyük MBA stream downcutting, drying out of swamps Kühne [1990] ancient canal system, Middle New Assyrian, and Lower Khabur seventh century Wilkinson et al. [2010] hollow way fills, Tell Brak circa 2000 until today filling since 2/3 ka B.C., wearing down before that, most probably in the third millennium B.C. or slightly earlier Wilkinson [1999] cultural layers and sediments, 4000 spring context Kurban Höyük mid-third millennium well structure Mari texts early second millennium inhabitants of Tuttul (near Raqqa) complain about high water withdrawals at Zalpa upstream Rosen and Goldberg sediment sections, Titriş Höyük late third/early flow becomes more erratic [1995] second millennium Deckers and Riehl sediments, Wadi Jaghjagh circa 2500 decrease in stable flow [2007] RIEHL ET AL. 125 and/or, more specifically, the “4200 B.P. event.” The latter several layers of fill, the oldest of which contained pottery has been suggested as possibly triggering the late EBA shards from the third millennium B.C. [Wilkinson et al., societal collapse in northern Mesopotamia [Weiss et al., 2010]. 1993] and has since become the subject of debate [Dalfes et The later layers contained ceramic fragments of succes- al., 1997] with an array of additional proxy evidence having sively decreasing age toward the modern land surface. This been found for this environmental change in the Near East suggests that the hollow ways, having once been intensively [Bar-Matthews et al., 1997; Cullen and DeMenocal, 2000; used (most probably as roads), lost their original importance Staubwasser et al., 2003; Pustovoytov et al., 2007b; Riehl et after the turn from the third to second millennium B.C. The al., 2008]. A broader scale of the aridification trend is dem- continuous, nonabrupt formation of fills suggests that the onstrated by the simultaneous disappearance of lakes and surrounding fields and presumably the hollow ways, them- sand dune activation on the Arabian Peninsula [Parker et selves, were still in use at later times rather than having been al., 2006]. Although landscape changes at the EBA to MBA completely abandoned. Another type of negative relief fea- transition that are visible in the geoarchaeological and sedi- ture is enclosed depressions around tells. They represent a mentological record in the northern Fertile Crescent mostly typical element of a tell landscape and are assumed to have involve natural features (channel fills, etc.), they do suggest a been the mud-brick quarries for extracting structural material restructuring of the regime of landscape moisture supply, for the tell [Wilkinson, 2003]. The depressions, otherwise which should have had a significant impact on agricultural called “city ditches,” might also have had secondary func- systems. tions, for which a number of hypotheses have been put The second line of geoarchaeological evidence comprises forward (see Pustovoytov et al. [2011] for a review). For one the finds of artificial, land use-related structures in the of those city ditches, around Tell Mozan (NE Syria), a similar landscape and their evolution over time. At Kurban Höyük pattern of the development as for the hollow ways discussed (SE Turkey), it has been established that the water tables above have been found [Pustovoytov et al., 2011]. were higher in the mid-fourth millennium B.C., while a The youngest radiocarbon date from the depression base water hole or a spring provided water supply to the settle- immediately below the fill was about 2.5 ka cal years B.C., ment. This water source most probably became exhausted suggesting that the fill accumulation started about that time due to dropping groundwater level, and an artificial well or shortly thereafter. The formation of the horizon of sec- was then dug in the mid-third millennium B.C. [Wilkinson, ondary carbonate accumulation in the exposed soil profile 1999]. Furthermore, the development of some artificial neg- started as early as 1 ka cal years B.C., which marks the phase ative relief features obviously ceased around the late third of stabilization of the land surface. Although more data are and early second millennium B.C. and gradually filled with needed to deduce the accumulation rate of the fill, it appears sediments. One of the features represents the so-called that its formation took place between 2.5 and 1 ka B.C. The “hollow ways” (synonyms: holloways, sunken lanes). A evolution of the city ditch bears testimony to an end of loam hollow way is defined as a “road or track running in a extraction (at least at the study location), which, in turn, may natural or man-made hollow deepened through wear caused be linked to a decline in settlement growth. Tell Mozan, by prolonged usage or the raising of the ground on each however, survived well into the LBA, with the formation of side” [Darvill, 2008]. This term established itself in archae- fill in the depression possibly indicating continuous agricul- ological literature over the last decade or two to describe the tural activities. linear structures, primarily in Mesopotamia, leading radially In general, geoarchaeological evidence, though promising, from one common center (usually a tell) and extend through remains very local in terms of landscape reconstruction. themodernlandscapeovermanykilometers[Wilkinson, Additionally, it is highly intriguing to find parallels (or contra- 1993, 2003; Ur, 2003]. It is important to emphasize that for dictions) between the geoarchaeological record and textual most of Mesopotamia today, there are no or very few and iconographic sources. However, the uneven distribution distinct hollow ways in the sense of definition given above. of the sources does not allow us to produce straightforward To the observer on the ground, the relief between tells conclusions. Our sources for the third millennium originate appears almost perfectly flat, and the long linear depressions from the Khabur region, while documents from the eighteenth can be detected only indirectly (i.e., by vegetation or stand- century city of Mari at the Middle Euphrates prevail where ing snowmelt water) or in satellite images [Wilkinson, irrigation was a commonly used devise. The inhabitants of 2003]. However, these landscape features seem to have been Tuttul (near modern Raqqa), for example, complained about true hollow ways in the past. high water withdrawals at Zalpa, lying upstream on the Eu- The hollow ways at Tell Brak experienced a certain evo- phrates in the early second millennium B.C. [Wilkinson, lution, that is, they first were incised and then replenished by 2003], an indication of a legal conflict that occurred 126 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT concerning water rights. The representations of shaduf (a based on Sumerian še, which was read differently in the local seesaw-like water extraction tool consisting of a bucket and Akkadian dialect, probably še’um). Two interpretations are a counterweight fixed at the ends of a long swinging pole) are possible: Either SIG15 is simply another term used locally for found on Mesopotamian seals starting at circa 2400 B.C. “wheat,” which was called kìb/kibtu in southern Mesopota- [Bagg, 2000], pointing to the importance of irrigation technol- mia, or SIG15 identifies another cereal other than “wheat.” In ogy in southern Mesopotamia. the latter case, a remarkable variation in the standard cereal species would then be evident. Considering the archaeobo- 3.3. The Transformations of Agricultural Systems in the tanical evidence, the most common cereal species during the Textual Record: The Presence and Absence of Crops period under consideration are barley, emmer, and free- (Cereals, Wine, Sesame, and Linseed) threshing wheat. Emmer (ZÍZ) was mainly used as fodder for animals in MC Cereals form the basis of ancient Near Eastern agriculture twenty-fourth century Beydar. Besides barley, emmer was and thus figure prominently both in the textual and in the given to sheep, nanny goats, and oxen, which were fed archaeobotanical record. The terms for cereals in Sumerian before slaughter, and to “plough oxen” and even birds as and Akkadian were identified in the groundbreaking work by well. Donkeys, however, were always fed barley, and espe- Hrozný [1913]. Since then, further philological research has cially, the donkeys of the ruler’s entourage were richly fed largely corroborated his identifications (see the articles in the with barley. Bulletin on Sumerian Agriculture, e.g., cereal crops: Powell In the archaeobotanical record of the second millennium [1984]; oil plants: Postgate [1985], Waetzoldt [1985]; sesa- B.C., emmer strongly decreases in proportions, although it me: Stol [1985b]; beans, peas, lentils, and vetches: Stol remains ubiquitous within many archaeological sites. There [1985a]; legumes: Maekawa [1985]). is very little textual evidence for “emmer” (ZÍZ) from the Excavations have uncovered more textual material since Middle Assyrian (fourteenth to thirteenth century) and from the times of Hrozný [1913], with the new texts including the Neo-Assyrian period (eighth to seventh century), an other terms for grains as well. Hrozný had identified the three indication of its continual decline in use. main cereal crops, “barley” (Sumerian še, Akkadian še’u), SIG15 appears only in the single cited text from Tell Beydar “emmer” wheat (Sumerian zíz, Akkadian kunāšu), and (twenty-fourth century), and there is no information on its “wheat” (Sumerian kìb, Akkadian kibtu), which is consid- use. The term turns up more often in the slightly later “bread ered as free-threshing wheat. The dominance of barley in all and beer texts’” from Palace G at Ebla (twenty-fourth cen- periods corresponds with the archaeobotanical record. The tury), which document the expenditure of food for persons textual evidence generates a model of Mesopotamian socie- present in the palace. According to the Beydar text, it must be ties that are basically barley-producing societies, due to a cereal species and not a specific preparation type. In con- climatic and sociopolitical factors. Collective agricultural sideration of the archaeobotanical evidence, SIG15 may de- production within a “top to down” decision-making structure note a specific free-threshing wheat, such as T. durum. may have favored monocultures, i.e., barley monocultures. SIG15 does not appear as such in second millennium In addition, the single farmer was expected to produce barley sources from the region dealt with here. Ancient Mesopota- to fulfill his liabilities even in periods without a collective mian vocabularies inform us that SIG was read hišlētu, 15 ˘ mode of production. literally “the crushings, grindings,” in Akkadian. In the The philological work necessary for a proper understand- mid-second millennium, texts from Nuzi/Yorgan Tepe situ- ing of the sources, so that they can be used in this interdis- ated east of the Tigris in the region of Assyria, hišlētu ˘ ciplinary study, may be demonstrated by the following appears in the same texts together with kibtu “wheat.” Thus, example on the terminology of wheat species. Generally we are dealing with two different sorts of wheat at Nuzi, and speaking, the argument deals with the difficult question given the equation of SIG = hišlētu, it is reasonable to 15 ˘ whether different terms denote different crops or are simply assume that SIG15 denotes another type of wheat other than local expressions for the same crop. Any differentiation, kìb. In this case, kìb would have been cultivated in southern however, lies on the basis to detect large-scale developments. Mesopotamia, SIG15 in northern Mesopotamia, perhaps An administrative document from Tell Beydar (approxima- corresponding to T. aestivum (kìb/kibtu)andT. durum tely MC twenty-fourth century [Ismail et al., 1996]) lists (SIG /hišlētu), respectively. 15 ˘ numbers (perhaps indicating the number of days for work in Whereas Presargonic Tell Beydar and Ebla (both MC the field or for the harvest) together with three crop species: twenty-fourth century) know the term SIG15, another cereal, 5400 ŠE “barley,” 1240 SIG15, and 1240 ZÍZ “emmer” (NB: the burrum, appears in Old Babylonian (mainly first half of MC font type ŠE indicates the use of a logogram for “barley” eighteenth century) texts of the region, from Mari and Tuttul RIEHL ET AL. 127 in the middle Euphrates Valley, to Chagar Bazar in the 10% of the produced textiles, linen appears with much lower Khabur triangle and to Tell al-Rimah southeast of the Jebel frequency in the written record than does woolen products Sinjar. The term also does not appear in southern Mesopota- [Potts, 1997], and linen textiles never seem to have acquired mian texts discussed by Hrozný [1913]. Burrum appears the same importance as they did in Egypt [Waetzoldt, 1983]. together with ŠE “barley” and ZÍZ “emmer” in Tell al-Rimah, Although the textual evidence is limited and deriving and it may plausibly be identified as a free-threshing wheat particularly from southern Mesopotamia, a decrease of flax species; furthermore, burrum is perhaps the word that stands cultivation seems likely to have occurred in the second behind the logogram SIG15 in the third millennium texts or at millennium B.C. The rare presence of a few linen garments least later replaces the archaic SIG15. in lists of textiles in the early second millennium (e.g., at Tell The investigation has led to the identification of two hith- al-Rimah), or the import of linen in the first millennium, for erto unidentified words for “wheat” (archaic SIG15, burrum); example, as tribute from Syria to the Neo-Assyrian empire they may well denote another species than the one used in the (934–609 B.C.), does not necessarily imply local production. south (kìb/kibtu). However, the archaeobotanical evidence of Developing a more refined methodology for quantifying the hexaploid (T. aestivum) and tetraploid free-threshing wheats proportion of linen textiles within various contexts is essen- (T. durum) is problematic due to the fact that the two species tial before we can draw definitive conclusions on the devel- are only discernible on the basis of their chaff remains, which opment of flax production from the third to the second were often not considered in early archaeobotanical work millennium. (for details, see the work of Riehl [2011, p. 155]). The The plant cultivated for oil production in lowland Mesopo- available evidence indicates a marked concentration of tet- tamia was sesame, a term related to Akkadian šamaššammū, raploid free-threshing wheat in the upper Khabur region for and Sumerian (še-)ŋeš-ì, literally “(seeds of the) oil plant” the third millennium B.C. (thus = SIG15, burrum?), while [Powell, 1991]. Before the IA, sesame is archaeobotanically hexaploid species occur everywhere else (thus = kìb/kibtu?). only recorded from EBA Abu Salabikh [Charles, 1993] The second millennium evidence is too scarce from which to and from LBA Tall Sheikh Hamad [van Zeist, 2001]. It draw any valid conclusions. More detailed research is needed appears, however, in the cuneiform sources as having been to confirm the new identifications. The problem of the pres- cultivated in Mesopotamia as early as the third millennium ence of einkorn, for which no native term has been proposed, B.C. [Stol, 1985b, 2010; Waetzoldt, 1985; Bedigian, 1998, must also be addressed. Only carefully controlled old and, if 2004]. possible, new identifications will allow a much better use of Although sesame oil was used regularly for body care, and the cuneiform documentation from the region. thus appears frequently in the textual record (e.g., at Old Another crop category relevant for environmental change Babylonian Mari, Tell al-Rimah, etc.), the sources from includes oil plants, linseed, and sesame, respectively. Upper Mesopotamia and Syria are practically silent about its Linseed occurs regularly in the Near Eastern archaeobota- cultivation. nical record starting at Pre-pottery Neolithic A Mureybet The textual evidence for olives corresponds with the nat- (circa 10.000–8.800 B.C.). Written evidence for linseed, ural distribution of the species, which is limited to the coastal called gu in Sumerian and qû in Akkadian, derives mainly areas of the Mediterranean [Zohary and Hopf, 2000]. In from southern Mesopotamian sources [Waetzoldt, 1983]. Mesopotamia, olive oil appears therefore as a good imported Linen, the textile made from linseed fibers, which is attested from the west [Stol, 2003–2005]. Recently, morphometric since around 7000 B.C. in Anatolia [Potts, 1997], is gada in and biomolecular analyses have provided more insight into Sumerian (third millennium) and kitû in Akkadian (second to the emergence of cultivars [Terral et al., 2004; Breton et al., first millennium). But there is a single writing that proves 2009], though the local beginnings of systematic use of that kitû in the thirteenth century B.C. Tall Sheikh Hamad/ domesticated species are still debated. The plausible argu- Dūr-Katlimmu means the flax plant [Cancik-Kirschbaum, ment of increased find numbers as an indicator of established 1996, pp. 106–111]. That both qû and kitû are Sumerian olive cultivation places the beginnings into the EBA or in loanwords in Akkadian is an expression of the cultural con- some places even the Chalcolithic. tinuity in Mesopotamia. Although linseed is mentioned in Written evidence of large-scale olive cultivation comes the textual sources, e.g., at seventeenth century B.C. Mari, its from Ebla/Tell Mardikh in the MC twenty-fourth century, consumption is not documented, and linseed oil was appar- where the term used is GIŠ.Ì, literally “wood of oil.” In the ently unknown in ancient Mesopotamia [Waetzoldt, 1983]. second millennium, the Akkadian term was serdu. Interest- The rarity of linseed is directly related to the fact that wool ingly, Ebla seems to have been at the eastern fringe of the was the dominant raw material for textile production as early region where olives were cultivated, and in remarkable quan- as the Chalcolithic period [Algaze, 2008]. With less than tities (an annual production of 170,000 L [after Archi, 1991, 128 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT p. 219]). Ebla would send olive oil in small quantities to the cultivation during the EBA and MBA was concentrated on courts of allied states, selling or delivering it as tithe for the the banks of the Euphrates south of Karkemish, the hilly Euphrates trade to the mighty city of Mari. regions south of the Taurus Mountains, the Upper Khabur, The common beverage in Mesopotamia was beer pro- and the region of Jebel Sinjar, which corresponds well with duced from barley. However, in Upper Mesopotamia, there the natural distribution of the wild progenitor species [Zoh- is plenty of evidence for the consumption of wine (for the ary and Hopf, 2000]. cuneiform evidence, see Powell [1996]). “Wine” is ŋeštin in Wine cultivation is known from LBA I Alalakh [Dietrich Sumerian, karānu in Akkadian, but at Emar, the word hamru and Loretz, 1969] and from LBA II Ashtata with its capital ˘ was used, written KAŠ.GEŠTIN (the signs meaning “the beer of Emar (Tell Meskene) [Mori, 2006], Karkemish, and Ugarit. the grape”). The overall proportions of wine and raisin In the Middle Assyrian Empire, wine continues to be used as (muziqu)[seePostgate, 1987] production are, however, the beverage of the upper class. There is one single Middle unknown. At MC twenty-fourth century Tell Mardikh/Ebla, Assyrian text in a private collection that mentions a vineyard vineyards existed both near Ebla and in the surrounding presumably somewhere in the Lower Khabur valley [Fales, countryside [Milano, 1996]. 2010, p. 77]. Interestingly, wine had to be bought at the contemporary The situation radically changes in the first millennium cities of Nabada/Tell Beydar [Ismail et al., 1996] and at Mari documentation from the Neo-Assyrian Empire. Textual evi- [Sallaberger, 2012b]. This could indicate that wine was not dence points to large vineyards in Upper Mesopotamia with produced at Ebla. Alternatively, it could demonstrate that the tens of thousands of plants in the regions of the Balikh Plain organizations from which the available documents derive did and the Jebel Sinjar. Wine production occurred in the region not dispose of their own vineyards, since these were, for of Guzana (Tell Halaf) in the Khabur triangle, in the hilly example, kept by the palace. regions of the Jebel Sinjar and the landscape of Izalla (pre- In the Old Babylonian period, the MBA, especially MC sumablytheDibekDagiintheeasternedgeoftheTur late nineteenth/eighteenth centuries B.C., we find plenty of Abdin). Zamua, in the vicinity of modern Kirkuk, was fa- evidence for wine consumption in Upper Mesopotamia, for mous for its wine as well [Powell, 1996, p. 115]. Still wine instance from the archives at Mari (Tell Hariri), Tell al- was the beverage of the upper social strata, although becom- Rimah, and Shekhna (Tell Leilan). Since it was significantly ing a regular part of cultic ceremonies [Powell, 1996, p. 119] more expensive to produce grapes than barley, wine was and the everyday beverage of professional soldiers. The per the beverage of the upper social strata. The wine jars found capita consumption of wine, however, is documented as listed in the texts, however, could have been imported having been low [Powell, 1996, p. 121]. from elsewhere. At Mari, wine trade became a source of The renewed interest in wine cultivation in the Balikh and income since the palace took 10% of every shipment as a Khabur region is surely a sign of a more dense agricultural tax on the Euphrates trade [Finet, 1974–1977, pp. 123– exploitation of the region in Neo-Assyrian times in order to 124; Michel, 1996, pp. 407–408]. At Mari, various quali- fulfill the rising demand for wine in the center of Assyria. ties of wine are documented, such as wine “of good Since there is evidence that the state was involved in the quality,” wine “of second quality,”“old” wine or “red” cultivation of the countryside, and with evidence from the wine. The latter one implies that different phenotypes of late seventh century Harran census [Johns, 1901] showing grapes were cultivated. the probable concern of royal estates with large vineyards in Most wine with origins explicitly named did not come the region of Harran, it seems likely that the rise in cultiva- from the Middle Euphrates, though we do have evidence for tion of grapes was commissioned by the crown itself. viniculture in the region upstream of Terqa (Tell Ashara) [Stol, 2004, p. 870]. 4. DISCUSSION The main center of wine production in Upper Mesopota- mia should have been Karkemish (Tell Jerablus). We do not Looking at the geoarchaeological, archaeobotanical, and have explicit evidence for local cultivation of grape, but the philological data, we find clear patterns linking evidence to low prices for wine from Karkemish suggest its cultivation known changes in the environment and the political and there [Powell, 1996, p. 108]. Other centers of wine produc- cultural life of the region under study. tion include Ugarit (Ras Shamra), Shubat-Enlil (Shekhna, The evolution of city-state after 2900 B.C. in the Khabur Tell Leilan), the land of Apum (the greater region of Tell region, which was accompanied by the accumulation of Leilan), and the landscape of Ida-Maras (region on the Upper political power through agricultural surplus production and Khabur or in the Khabur triangle [Vincente, 1991, pp. 299, cooperation between independent cities and the continuous 305, 310; Powell, 1996, pp. 115–116]. To summarize, grape centralization of power, was followed by the fall of the RIEHL ET AL. 129

Akkad Dynasty shortly before 2100 B.C., which is frequently assemblages. This correlates with the geoarchaeological re- discussed in relation with the climate change event of 4200 cord of a change in the regime of water streams at the B.P. [Weiss et al., 1993; Weiss and Courty, 1993; deMenocal, transition from the third to the second millennium B.C. from 2001; Bar-Matthews and Ayalon, 2011; Kuzucuoglu and relatively stable, moderately strong water flows to stronger, Marro, 2007]. The chronological resolution of palaeoclimate more rare, and erratic flows in a number of places, suggesting proxy data, however, differs from historical time resolution, a trend toward aridification, in general, and/or more frequent and the regional variations in the dating of climate fluc- droughts. These trends suggest a restructuring of the regime tuations render the relationship between climate effects of landscape moisture supply, which should have had a and population developments into a problematic issue [cf. significant impact on agricultural systems. Roberts et al., 2008; Riehl and Bryson, 2007]. We need Regional differences in these patterns exist (e.g., between to keep in mind that the label of “4200” for the mid- the Khabur and the Euphrates, and the western, more Med- Holocene climate fluctuation is not an exact dating, as the iterranean regions), but can only be discussed within more correlation of different palaeoclimate proxy archives is limited detailed studies [Riehl and Bryson, 2007; Riehl, 2012a]. due to the methodological problems arising from differences The absence of the term KÌB/kibtu for the hexaploid form of in the material dated (see various contributions of Battarbee free-threshing wheat in the Beydar evidence may be ex- and Binney [2008]). This is particularly the case when we plained by a lack of large-scale irrigation, whereas the tetra- consider that climate fluctuations are trends between two ploid form would have generated greater yields than the peaks, which were experienced by ancient populations more hexaploid in this area. This would correspond with the ar- as a relatively long sequence of minor fluctuations (“creeping chaeobotanical record of the third millennium, where tetra- normalcy” as introduced by Diamond [2005]). Chronological ploid free-threshing wheat occurred in larger amounts in the resolution of historical state development is much higher, Khabur region though hexaploid free-threshing wheat is which makes a probable correlation between climate fluctua- found particularly in places were irrigation is assumed. tions and historical events difficult. Phases of economic Despite a general correlation of the archaeobotanical with growth and increasing population, such as during the reign of the philological record, discrepancies occur as well between Ur III, as well as various examples of settlement abandon- both records, as in the case of sesame, which according to the ment, are within the time frame of the “4200 B.P. event.” texts was already an important crop in the EBA, though not Despite the problem of linking historical development to preserved in most archaeological sites due to the physiolog- climate fluctuations other than catastrophes, the general cli- ical character of the seeds. There is, however, no textual mate trend of increasing aridity after 5000 B.C. [Wanner et evidence that any adaptations in sesame cultivation were al., 2008], which was superposed by regional climate effects conducted throughout time. Differently for linseed, although starting at around 2200 B.C., has often been considered also an oil crop, it seems to have been used only in textile responsible for state collapse. Within the whole historical production, but even there in a very restricted manner. sequence, warfare, destruction of prominent political centers, Between 2000 and 1500 B.C., internal political change was opposition, and rebellion are frequently reported and are greatly caused by the Amorites, who seized power over a generally accepted as signs of collapse. Famine, inflation, number of urban centers. It seems probable that Amorite and problems with taxation are also among the most fre- ways of life held consequences for animal husbandry. Other quently mentioned reasons. Generally, high-energy needs for earlier centers regained economic and political power, and surplus production and development of technological and even new centers, such as Qatna, developed during the MBA. organizational complexity [Tainter, 1995; McIntosh et al., The end of the MBA is accompanied by warfare in 2000] are considered as important problems for maintaining many regions, particularly the Hittite attack on Yamkhad in the societal status quo, and this seems to be reflected in the northwest Syria and the sack of Babylonia. Many powerful agricultural developments observed in the archaeobotanical cities disappeared, and there was also a discontinuation of and the philological data. administrative and scribal practices as the levels of economic Due to fieldwork methodology, the archaeobotanical evi- and cultural activities decreased. Textual evidence becomes dence provides only low chronological solution, usually scare, reflecting the emergence of a “Dark Age.” Fewer ar- showing only general differences between the EBA and the chaeological sites and less archaeobotanical data are available. MBA assemblages on the supraregional level. A reduction in There is a trend, however, of greater regional water stress the presence, proportion, and ubiquity of water-demanding on crop species and, consequently, a greater concentration on crop species is visible when comparing EBA and MBA more drought-tolerant species, as outlined above. The com- assemblages. At the same time, water stress signals inferred paratively small body of textual evidence, especially from through δ13C in barley grains are more frequent in MBA southern Mesopotamia, unequivocally confirms the decrease 130 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT of flax cultivation in the second millennium B.C. compared gation in areas of irregular and low rainfall. Changes in with the earlier periods. Although there is almost no written persisting crop species are particularly visible in free- evidence for the cultivation of flax, and generally woven threshing wheat and grape ubiquities, which occur generally textiles were made of wool, the available sources support more often in the archaeobotanical record of the IA, and the assumption that linen and linen textiles were imported in which is confirmed by the textual evidence of numerous small quantities, by trade or tribute during the second mil- vineyards in the IA of Upper Mesopotamia. The larger lennium B.C. The import of the end product was thus more body of textual evidence from the first millennium also efficient than local production of the source crop, supporting reveals more linen textiles than in the second millennium. earlier interpretations of the disappearance of linseed from Still, the evidence is almost exclusively for the import of MBA archaeobotanical assemblages [Riehl, 2011]. A climate linen as tribute from Syria to the Neo-Assyrian Empire shift toward more arid conditions would have led to decreas- (934–609 B.C.). Only later, during the Neo-Babylonian ing yields of flax, which has high water requirements. An period, do we find evidence for more frequent linen culti- additional factor may have been autotoxicity, which is the vation [Waetzoldt, 1983]. Olive is not frequently mentioned fact that plants of the same species react negatively to the in ancient texts, but the import of olive oil from the west is metabolic products of the previous generation. In flax, this attested [Faist, 2001; Stol, 2003–2005, p. 33], corresponding leads to a cultivation interval of several years on the same with the archaeobotanical record. plot of earth before reasonable yields can be expected again. The importance of irrigation is known from the textual It is unknown whether ancient farmers were aware of this evidence and is indicated in the archaeobotanical remains, problem. The effect, however, should have been clearly where wild plant remains contain an increased number of recognized as strongly reduced yields in the second year, salinization indicators [Riehl, 2010b]. A local increase in the leading to either a shift to other cultivable land or, in the water supply, either through precipitation or irrigation, is worst case, to an abandonment of the crop, in general. In this visible in the IA δ13C values. At many other sites, however, case, climate would have only acted as an additional factor. there is no such signal. Another correlation of the archaeobotanical with the philological data is the development of second millennium em- 5. CONCLUSION mer production, which appears significantly regressive in the textual evidence and is economically meaningless since Late Available archaeobotanical, philological, and geoarchaeo- Bronze II. There is contemporary evidence, though, for irri- logical data of northern Mesopotamia show that climatic gated wheat production (kibtu; presumably aestivium;see fluctuations at the end of the EBA correlate with the regional above). This has often been related to the development of abandonment of settlements, the reduction of crop species markets and the mass production of barley [Nesbitt and with higher water requirements, and an increased water stress Samuel, 1996]. in barley. At the transition from the LBA to the IA, there are Most of the settlement centers disintegrate with the end of locally higher water stress signals in the IA, but also increas- the LBA due to political conflict and economic decline (e.g., ingly more water-demanding species, which can best be crop failures, rebellion against taxation). It is most likely that explained by an improvement in agricultural technology and complex interactions between political, ecological, and cli- irrigation techniques. matic (Bond event no. 2) factors were responsible for these The archaeobotanical, geoarchaeological, and philological massive supraregional changes. Another “Dark Age” follow- evidence confirm that a number of reasons, including feeding the end of the LBA was of varying duration and lasted in back mechanisms, must have been responsible for the devel- Assyria until 935 B.C. Very little is known about the centu- opments occurring at the end of the EBA and the LBA. The ries between 1100 and 900 B.C., but it seems clear that an complex organization, which is necessary to sustain redistri- almost complete restructuring of society must have taken bution systems, such as in EBA northern Mesopotamia, was place due to migration, internal population movements, and challenged through climate fluctuations and increasing arid- technological and trade network changes [Van De Mieroop, ity. Surplus production required high-energy input through 2003]. But there is also textual evidence for continuity, e.g., labor, probably increasing throughout time by population the restoration of the irrigation systems of the Lower Khabur growth and thus becoming increasingly difficult to sustain, and Middle Euphrates [Fales, 2008] or the maintenance of as documented through shortages in the labor force for many urban centers in Upper Mesopotamia under Aramean agricultural production. The endeavor to maintain such sys- supremacy. tems was related to the accumulation of power, which often During the IA, a number of new crops appear in the ended in warfare between the different centers, with all its archaeobotanical assemblages, most of them requiring irri- negative consequences. RIEHL ET AL. 131

Despite many difﬁcult methodological issues, the follow- Altaweel, M. (2008), Investigating agricultural sustainability and strat- ing synthesis can be developed, based on our body of evi- egies in northern Mesopotamia: Results produced using a socio- dence for observable agricultural transformations in northern ecological modeling approach, J. Archaeol. Sci., 35, 821–835. Mesopotamia between 3000 and 1000 B.C. Araus, J. L., A. Febrero, M. Catala, M. Molist, J. Voltas, and I. During the EBA, the aim of surplus production appears Romagosa (1999), Crop water availability in early agriculture: together with increasingly drier conditions toward the end of Evidence from carbon isotope discrimination of seeds from a tenth millennium BP sites on the Euphrates, Global Change the EBA and the beginning of the MBA, consistent with the Biol., 5, 201–212. 4200 B.P. event. This led to decreasing yields, opening up Araus, J. L., G. A. Slafer, I. Romagosa, and M. Molist (2001), new perspectives with probable economic change through an Estimated wheat yields during the emergence of agriculture ethnogenesis of Amorite nomads. The continued reliance on based on the carbon isotope discrimination of grains: Evidence surplus production during the MBA coupled with increased from a 10th millennium BP site on the Euphrates, J. Archaeol. water stress, as indicated in the stable carbon isotope and Sci., 28, 341–350. geoarchaeological evidence, was answered by a relative shift Archi, A. (1991), Culture de l’olivier et production de l’huile à Ebla, in from demanding and labor-intensive crop species to more Marchands, Diplomates et Empereurs. Etudes sur la Civilisation stress-tolerant species, barley in particular. However, a short- Mésopotamienne Offertes à Paul Garelli, edited by D. Charpin and age in human resources resulted in continuous economic F. Joannès, pp. 211–222, Ed. Rech. sur les Civ., Paris, France. decline on the political level, consequently leading to polit- Arnaud, D. (1986), Recherches au Pays d’Aštata. Emar VI.3. Textes ical instability, the competition for resources between states, Sumériens et Accadiens, vol. 1, Ed. Rech. sur les Civ., Paris, and occasional warfare. This seems to have continued into France. the LBA. With the new network of states emerging at the Bagg, A. (2000), Assyrische Wasserbauten. Landwirtschaftliche beginning of the IA, agricultural production rejuvenated, Wasserbauten im Kernland Assyriens zwischen der 2. Hälfte des 2. und der 1. Hälfte des 1. Jahrtausends v. Chr., Philipp von with a considerable amount of new crop species being intro- Zabern, Mainz, Germany. duced and a comeback of more demanding crops occurring, Bar-Matthews, M., and A. K. Ayalon (1998), Middle to late Holo- such as grape. The combined archaeobotanical and stable cene (6500 yr. period) paleoclimate in the Eastern Mediterranean carbon isotope evidence suggests that this can be traced to region from stable isotopic composition of speleothems from advances in irrigation technology for some places in northern Soreq Cave, Israel, in Water, Environment and Society in Times Syria. For other, more southern, places, such as Qatna, in- of Climatic Change, edited by A. B. Issar, pp. 203–214, Kluwer creased water stress indicated problems in agricultural pro- Acad., Dordrecht, The Netherlands. duction, which may have been related to locally increased Bar-Matthews, M., and A. K. Ayalon (2011), Mid-Holocene climate aridity. variations revealed by high-resolution speleothem records from These developments illustrate the difﬁculty in using the Soreq Cave, Israel and their correlation with cultural changes, term “collapse” in relation to changing conditions. The end The Holocene, 21, 163–171. of one system almost always brings options for evolution or Bar-Matthews, M., A. K. Ayalon, and A. Kaufman (1997), Late transformation. Despite climate changes and the visible re- Quaternary paleoclimate in the Eastern Mediterranean region structuring of the landscape, particularly in relation to water from stable isotope analysis of speleothems at Soreq Cave, Israel, Quat. Res., 47, 155–168. resources and management, the development of agricultural Barnola, J. M., D. Raynaud, C. Lorius, and N. I. Barkov (2003), systems spanning the cultural sequence considered here de- Historical CO2 Record From the Vostok Ice Core, Trends: A monstrates continuous adaptation rather than collapse. Compendium of Data on Global Change, Carbon Dioxide Inf. Anal. Cent., U.S. Dep. of Energy, Oak Ridge, Tenn. Acknowledgments. The basic project work was funded by the Battarbee, R., and H. A. Binney (2008), Natural Climate Variability German Research Foundation (DFG), project Ri 1193/6-1. and Global Warming: A Holocene Perspective, Wiley-Blackwell, Hong Kong. REFERENCES Bedigian, D. (1998), Early history of sesame cultivation in the Near East and beyond, in The Origins of Agriculture and the Domes- Akkermans, P. M. M. G., and G. M. Schwartz (2003), The Archae- tication of Crop Plants in the Near East. The Harlan Symposium, ology of Syria. From Complex Hunter-Gatherers to Early Urban edited by A. B. Damania and J. Valkoun, pp. 93–101, Int. Cent. Societies (ca. 16000–300 BC), Cambridge Univ. Press, Cam- for Agric. Res. in Dry Areas (ICARDA), Aleppo, Syria. bridge, U. K. Bedigian, D. (2004), History and lore of sesame in Southwest Asia, Algaze, G. (2008), Ancient Mesopotamia at the Dawn of Civiliza- Econ. Bot., 58, 329–353. tion: The Evolution of an Urban Landscape, Univ. of Chicago Bernbeck, R. (1993), Steppe als Kulturlandschaft, Dietrich Reimer, Press, Chicago, Ill. Berlin. 132 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT

Bond, G., B. Kromer, J. Beer, R. Muscheler, M. N. Evans, W. Deckers, K., and S. Riehl (2008), Resource exploitation of the Showers, S. Hoffmann, R. Lotti-Bond, I. Hajdas, and G. Bonani upper Khabur basin (NE Syria) during the 3rd millennium BC, (2001), Persistent solar influence on North Atlantic climate dur- Paléorient, 34, 173–189. ing the Holocene, Science, 294, 2130–2136. deMenocal, P. B. (2001), Cultural responses to climate change Breton, C., J.-F. Terral, C. Pinatel, F. Médail, F. Bonhomme, and A. during the late Holocene, Science, 292, 667–672. Bervillé (2009), The origins of the domestication of the olive tree, Diamond, J. (2005), Collapse. How Societies Choose to Fail or C. R. Biol., 332, 1059–1064. Succeed, Penguin, New York. Bryson, R. A. (2005), Archaeoclimatology, in The Encyclopedia of Dietrich, M., and O. Loretz (1969), Die soziale Struktur von Alalah World Climatology, edited by J. E. Oliver, pp. 58–63, Springer, und Ugarit (V) Die Weingärten des Gebietes von Alalah im 15. Dordrecht, The Netherlands. Jahrhundert, Ugarit Forsch., 1,37–64. Bryson, R. A., H. H. Lamb, and D. L. Donley (1974), Drought and Donahue, J., B. Peer, and R. T. Shaub (1997), The southeastern the decline of Mycenae, Antiquity, 48,46–50. Dead Sea plain: Changing shorelines and their impact on settle- Cancik-Kirschbaum, E. C. (1996), Die mittelassyrischen Briefe aus ment patterns through historical periods, in Studies in the History Tall Šēh Hamad. Berichte der Ausgrabung Tall Šēh Hamad, Dūr- and Archaeology of Jordan VI: Landscape Resources and Hu- ˘ ˙ ˘ ˙ Katlimmu 4, Reimer, Berlin. man Occupation in Jordan Throughout the Ages, edited by G. Carpenter, R. (1966), Discontinuity in Greek Civilization, Cam- Bisheh et al., pp. 127–136, Dep. of Antiquities, Amman, Jordan. bridge Univ.Press, Cambridge, U. K. Faist, B. (2001), Alter Orient und Altes Testament, vol. 265, Der Charles, M. (1993), Botanical remains, in Abu Salabikh Excava- Fernhandel des assyrischen Reiches zwischen dem 14. und 11. tions, vol. 4, The 6G Ash-Tip and Its Contents: Cultic and Jahrhundert v. Chr., Ugarit, Münster, Germany. Administrative Discard from the Temple?, edited by A. Green, Fales, M. (2008), Canals in the Neo-Assyrian rural landscape: A pp. 203–207, Br. Sch. of Archaeol., Bagdad, Iraq. view from the Khabur and Middle Euphrates, in Umwelt und Charpin, D. (2004), Histoire politique du Proche-Orient Amorite Subsistenz der assyrischen Stadt Dur-Katlimmu am Unteren – Khabur, Berichte der Ausgrabung Tall Šēh Hamad/Dūr-Katlimmu (2002 1595), in Mesopotamien. Die altbabylonische Zeit, edited ˘ ˙ by P. Attinger et al., pp. 25–480, Universitäts-Verlag/Vanden- 8, edited by H. Kühne, pp. 181–186, Harrassowitz, Wiesbaden, hoeck & Ruprecht, Freiburg, Switzerland. Germany. Charpin, D., and N. Ziegler (2003), Mari et le Proche-Orient à Fales, M. (2010), Production and consumption at Dur-Katlimmu: A l’époque amorrite. Essai d’histoire politique. Florilegium Mar- survey of the evidence, in Dur-Katlimmu 2008 and Beyond, edited ianum5 (Mémoires de NABU 6), SEPOA, Paris, France. by H. Kühne, pp. 67–85, Harrasowitz, Wiesbaden, Germany. Civil, M. (1994), The Farmer’s Instructions. A Sumerian Agricul- Farquhar, G. D., M. H. O’Leary, and J. A. Berry (1982), On the tural Manual., Ed. Ausa, Barcelona, Spain. relationship between carbon isotope discrimination and the inter- Cordova, C. (2008), Floodplain degradation and settlement history cellular carbon dioxide concentration in leaves, Aust. J. Plant in Wadi al-Wala and Wadi ash-Shallalah, Jordan, Geomorphology, Physiol., 9, 121–137. 101,443–457. Farquhar, G. D., J. R. Ehleringer, and K. T. Hubick (1989), Carbon Cordova, C., C. Foley, A. Nowell, and M. Bisson (2005), Land- isotope discrimination and photosynthesis, Annu.Rev.Plant. forms, sediments, soil development and prehistoric site settings Physiol. Plant. Mol. Biol., 40, 503–537. on the Madaba-Dhiban Plateau, Jordan, Geoarchaeology, 20, Ferrio, J. P., J. L. Araus, R. Buxó, J. Voltas, and J. Bort (2005), 29–56. Water management practices and climate in ancient agriculture: Cullen, H. M., and P. B. DeMenocal (2000), North Atlantic influ- Inferences from the stable isotope composition of archaeobota- ence on Tigris-Euphrates streamflow, Int. J. Climatol., 20, nical remains, Veg. Hist. Archaeobot., 14, 510–517. 853–863. Finet, A. (1974–1977), Le vin à Mari, Arch. Orientforsch., 25, Dalfes, H. N., G. Kukla, and H. Weiss (1997), Third Millennium BC 122–131. Climate Change and Old World Collapse, NATO ASI Ser., vol. 1, Fiorentino, G., V. Caracuta, L. Calcagnile, M. D’Elia, P. Matthiae, 728 pp., Springer, Berlin. F. Mavelli, and G. Quarta (2008), Third millennium B.C. climate Darvill, T. (2008), The Concise Oxford Dictionary of Archaeology, change in Syria highlighted by carbon stable isotope analysis of 547 pp., Oxford Univ. Press, Oxford, U. K. 14C-AMS dated plant remains from Ebla, Palaeogeogr. Palaeo- – Dassow, E. V. (2008), State and Society in the Late Bronze Age: climatol. Palaeoecol., 266,51 58. Alalah Under the Mittani Empire, 593 pp., Capital Decisions Fleming, D. E. (2004), Democracy’s Ancient Ancestors: Mari and Ltd., Bethesda, Md. Early Collective Governance, Cambridge Univ. Press, Cam- Dawson, T. E., S. Mambelli, A. H. Plamboeck, P. H. Templer, and bridge, U. K. K. P. Tu (2002), Stable isotopes in plant ecology, Annu. Rev. Frey, W. J., and H. Kürschner (1991), Holzkohlefunde in Dur- Ecol. Syst., 33, 507–559. Katlimmu/Tall Seh Hamad, in Die rezente Umwelt von Tall Seh Deckers, K., and S. Riehl (2007), Fluvial environmental contexts Hamad und Daten zur Umweltrekonstruktion der Assyrischen – for archaeological sites in the Upper Khabur basin (northeastern Stadt Dur-Katlimmu, edited by H. Kühne, pp. 137 161, Dietrich Syria), Quat. Res., 67, 337–348. Reimer, Berlin. RIEHL ET AL. 133

Friedrich, W. L., B. Kromer, M. Friedrich, J. Heinemeier, T. Pfeif- International Continental Scientiﬁc Drilling Program (ICDP): fer, and S. Talamo (2006), Santorini eruption radiocarbon dated Site survey results and perspectives, Quat. Sci. Rev., 28, 1555– to 1627–1600 B.C., Science, 312, 548. 1567. Frumkin, A., and Y. Elitzur (2002), Historic Dead Sea level ﬂuctua- Maekawa, K. (1985), Cultivations of legumes and Mun-gazi plants tions calibrated with geological and archaeological evidence, in Ur III Girsu, Bull. Sumerian Agric., 2,97–118. Quat. Res., 57, 334–342. Mayer, W. (2001), Tall Munbaqa-Ekalte II. Die Texte, Harrassowitz, Goldhausen, M., and A. Ricci (2005), Political centralisation in the Wiesbaden, Germany. Syrian Jezira during the 3rd millennium: A case study in settle- McAnany, P. A., and N. Yoffee (2010), Questioning Collapse: ment hierarchy, Altorientalische Forsch., 32, 132–157. Human Resilience, Ecological Vulnerability and the Aftermath Hazan, N., M. Stein, A. Agnon, S. Marco, D. Nadel, J. F. W. of Empire, Cambridge Univ. Press, Cambridge, U. K. Negendank, M. J. Schwab, and D. Neev (2005), The late Qua- McIntosh, R. J., J. A. Tainter, and S. K. McIntosh (2000), Climate, ternary limnological history of Lake Kinneret (Sea of Galilee), history, and human action, in The Way the Wind Blows. Climate, Israel, Quat. Res., 63,60–77. History, and Human Action, edited by R. J. McIntosh et al., Heimpel, W. (2003), Letters to the King of Mari: A New Transla- pp. 1–44, Columbia Univ. Press, New York. tion, With Historical Introduction, Notes, and Commentary, Ei- ī s Mebert, J. (2011), Die Venustafeln des Amm -˙aduqa und ihre senbrauns, Winona Lake, Indiana. Bedeutung für die astronomische Datierung der Altbabylonischen Hole, F. (2007), Agricultural sustainability in the semi-arid Near Zeit, Institut für Orientalistik der Universität Wien, Vienna. East, Clim. Past, 3, 193–203. Michel, C. (1996), Le commerce dans les textes de Mari, in Mari, Hrozný, F. (1913), Das Getreide im alten Babylonien. Ein Beitrag Ebla et les Hourrites, dix ans de travaux, Ière partie, Actes zur Kultur- und Wirtschaftsgeschichte des alten Orients. I. Teil, colloque international Paris 1993. Amurru 1, edited by J.-M. A. Hölder, Vienna. Durand, pp. 385–426, Ed. Rech. sur les Civ., Paris, France. Ismail, F., W. Sallaberger, P. Talon, and K. Van Lerberghe (1996), Milano, L. (1996), Ébla: Géstion des terres et géstion des ressources Administrative Documents From Tell Beydar (Seasons 1993– alimentaires, in Mari, Ebla et les Hourrites, dix ans de travaux, 1995), Brepols, Turnhout, Belgium. Ière partie, Actes colloque international Paris 1993. Amurru 1, Jakob, S. (2003), Mittelassyrische Verwaltung und Sozialstruktur: edited by J.-M. Durand, pp. 135–171, Ed. Rech. sur les Civ., Untersuchungen, Brill, Leiden, The Netherlands. Paris, France. Jas, R. M. (Ed.) (2000), Rainfall and Agriculture in Northern Milano, L., W. Sallaberger, P. Talon, and K. Van Lerberghe (2004), Mesopotamia, Ned. Inst. voor het Nabije Oosten, Leiden, The Third Millennium Cuneiform Texts From Tell Beydar (Seasons Netherlands. 1996–2002), Brepols, Turnhout, Belgium. Johns, C. H. W. (1901), An Assyrian Doomsday Book, or Liber MorandiBonacossi, D. (2008), Betrachtungen zur Siedlungs- und Censualis of the District Round Harran in the Seventh Century B. Bevölkerungsstruktur des Unteren Habur-Gebietes in der neuas- C., J. C. Hinrichs, Leipzig. syrischen Zeit, in Umwelt und Subsistenz der neuassyrischen Klengel, H. (1992), Syria—3000 to 300 BC. A Handbook of Polit- Stadt Dūr-Katlimmu am unteren Habur, edited by H. Kühne, ical History, Akademie, Berlin. pp. 189–214, Harrassowitz, Wiesbaden, Germany. Kühne, H. (1990), The effects of irrigation agriculture: Bronze and Mori, L. (2006), Information on landscape from the Emar legal Iron Age habitation along the Khabur, Eastern Syria, in Man’s texts, in The City of Emar Among the Late Bronze Age Empires. Role in the Shaping of the Eastern Mediterranean Landscape, History, Landscape, and Society. Proceedings of the Konstanz edited by S. Bottema et al., pp. 15–30, A. A. Balkema, Rotter- Emar Conference, edited by L. d’Alfonso et al., pp. 113–128, dam, The Netherlands. Ugarit, Münster, Germany. Kuzucuoglu, C., and C. Marro (2007), Sociétés Humaines et Chan- Nesbitt, M., and D. Samuel (1996), From staple crop to extinction? gement Climatique à la Fin du Troisième Millénaire: Une Crise The archaeology and history of the hulled wheats, in Hulled A-t-elle eu Lieu en Haute Mésopotamie?, Varia Anatolica, vol. Wheats. Promoting the Conservation and Use of Underutilized 19, Diffusion De Boccard, Paris, France. and Neglected Crops, Workshop on Hulled Wheats, 21–22 July Lafont, B. (2000), Irrigation agriculture in Mari, in Rainfall and 1995, Castelvecchio Pascoli, Tuscany, edited by S. Paludosi et – Agriculture in Northern Mesopotamia,editedbyR.M.Jas, al., pp. 41 100, Int. Plant Genetic Resour. Inst., Rome, Italy. pp. 129–146, Ned. Inst. voor het Nabije Oosten, Leiden, The Neumann, J., and S. Parpola (1987), Climatic change and the 11th– Netherlands. 10th century eclipse of Assyria and Babylonia, J. Near East. Lemcke, J., and M. Sturm (1997), Delta 18 O and trace element Stud., 6, 161–162. measurements as proxy for the reconstruction of climate changes Parker, A. G., and A. S. Goudie (2008), Geomorphological and at Lake Van (Turkey): Preliminary results, in Third Millennium palaeoenvironmental investigations in the southeastern Arabian BC Climate Change and Old World Collapse, edited by H. N. Gulf region and the implication for the archaeology of the region, Dalfes et al., pp. 653–678, Springer, Berlin. Geomorphology, 101, 458–470. Litt, T., S. Krastel, M. Sturm, R. Kipfer, S. Örcen, G. Heumann, Parker, A. G., A. S. Goudie, S. Stroke, K. White, M. J. Hodson, M. S. O. Franz, U. B. Ülgen, and F. Niessen (2009), ‘PALEOVAN’, Manning, and D. Kennet (2006), A record of Holocene climate 134 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT

change from lake geochemical analyses in southeastern Arabia, Riehl, S. (2010a), Flourishing agriculture in times of political insta- Quat. Res., 66, 465–476. bility—The archaeobotanical and isotopic evidence from Tell Postgate, J. (1985), The “oil plant” in Assyria, Bull. Sumerian Atchana, in Excavations in the Plain of Antioch. Tell Atchana, Agric., 2, 145–152. Ancient Alalakh, a Bronze Age Capital in the Amuq Valley, Postgate, J. (1987), Notes on fruits in cuneiform sources, Bull. Turkey. The 2003–2004 Excavation Seasons, edited by K. A. Sumerian Agric., 3,115–144. Yener, pp. 123–136, Zero, Istanbul, Turkey. Potts, D. T. (1997), Mesopotamian Civilisation: The Material Riehl, S. (2010b), Maintenance of agricultural stability in a chang- Foundations, Cornell Univ. Press, Ithaca, N. Y. ing environment—The archaeobotanical evidence at Emar, in Powell, M. A. (1984), Sumerian cereal crops, Bull. Sumerian Ag- Emar After the closure of the Tabqa Dam. The Syrian-German – ric., 1,48–72. excavations 1996 2002, vol. 1, Late Roman and Medieval Cem- Powell, M. A. (1991), Epistemology and Sumerian agriculture: The eteries and Environmental Studies, edited by U. Finkbeiner and – strange case of sesame and linseed, in Ancient Near Eastern F. Sakal, pp. 177 224, Brepols, Turnhout, Belgium. — Studies in Honor of Miguel Civil on the Occasion of His Sixty- Riehl, S. (2010c), Plant production in a changing environment Fifth Birthday, edited by P. Michalowski, pp. 155–164, Ed. Ausa, The archaeobotanical remains from Tell Mozan, in Ausgrabun- – Barcelona, Spain. gen 1998 2001 in der Zentralen Oberstadt von Tall Mozan/ š Powell, M. A. (1996), Wine and the vine in ancient Mesopotamia: Urke : The Development of the Environment, Subsistence and š The cuneiform evidence, in The Origins and Ancient History of Settlement of the City of Urke and its Region, edited by K. – Wine, edited by S. J. Fleming, pp. 97–122, Gordon and Breach, Deckers et al., pp. 13 158, Harrassowitz, Wiesbaden, Germany. Amsterdam, The Netherlands. Riehl, S. (2011), Climate and agricultural decision-making: Envi- Pruzsinszky, R. (2009), Contributions to the Chronology of the ronmental constraints and economic development in Near East- – Eastern Mediterranean, vol. 22, Mesopotamian Chronology of ern sites between 5000 3500 cal BP, in Between Sand and Sea. the 2nd Millenium B.C.: An Introduction to the Textual Evidence Archaeology and Human Ecology in Southwestern Asia. Inter- and Related Chronological Issues, Österreichischen Akad. der national Symposium in Honor of Hans-Peter Uerpmann, edited – Wiss., Vienna. by N. J. Conard et al., pp. 147 166, Kerns, Tübingen, Germany. Pustovoytov, K., K. Schmidt, and H. Parzinger (2007a), Radiocar- Riehl, S. (2012a), Variability in ancient Near Eastern environmental bon dating of thin pedogenic carbonate laminae from Holocene and agricultural development, J. Arid Environ., doi:10.1016/j. archaeological sites, The Holocene, 17, 835–843. jaridenv.2011.09.014, in press. Pustovoytov, K., K. Schmidt, and H. Taubald (2007b), Evidence Riehl, S. (2012b) Understanding the reasons for non-sustainability for Holocene environmental changes in the northern Fertile Cres- in past agricultural systems, in Climate and Ancient Societies: cent provided by pedogenic carbonate coatings, Quat. Res., 67, Causes and Human Responses. The Stine Rossel Memorial Con- 315–327. ference, edited by S. Kerner, Univ. of Copenhagen, Copenhagen, Pustovoytov, K., K. Deckers, and P. Goldberg (2011), Genesis, age Denmark, in press. and archaeological significance of a pedosediment in the depres- Riehl, S., and R. A. Bryson (2007), Variability in human adaptation sion around Tell Mozan, Syria, J. Archaeol. Sci., 38, 913–924. to changing environmental conditions in Upper Mesopotamia Radner, K. (2004), Das mittelassyrische Tontafelarchiv von Giricano/ during the Early to MBA transition, in Sociétéshumaineset Dunnu-Ša-Uzibi. Ausgrabungen in Giricano 1, Brepols, Turnhout, changement climatique à la fin du troisième millénaire: Une crise Belgium. a-t-elle eu lieu en Haute-Mésopotamie?, Varia Anatolica, vol. 19, Riehl, S. (1999), Bronze Age Environment and Economy in the edited by C. Marro and C. Kuzucuoglu, pp. 523–548, Diffusion Troad: The Archaeobotany of Kumtepe and Troy,MoVince, De Boccard, Paris, France. Tübingen, Germany. Riehl, S., and K. Deckers (2008), Vorbericht zu einigen eisenzeitli- Riehl, S. (2004), Archaeobotany at the Early Bronze Age settlement chen und mittelalterlichen Pflanzenresten vom Tell Halaf, in Aus- of Hirbet ez-Zeraqon: A preliminary report, Z. Dtsch. Palästina- grabungen auf dem Tell Halaf 2006 und 2007. Vorbericht über die Ver., 120, 102–122. 1. und 2. syrisch-deutsche Grabungskampagne, edited by A. Riehl, S. (2008), Climate and agriculture in the ancient Near East: A Baghdo et al., pp. 105–118, Harrassowitz, Wiesbaden, Germany. synthesis of the archaeobotanical and stable carbon isotope evi- Riehl, S., and K. Deckers (2011), The botanical finds from the 2007 dence, Veg. Hist. Archaeobot., 17,43–51. and 2008 seasons of excavations, Baal, 13,110–116. Riehl, S. (2009a), Archaeobotanical evidence for the interrelation- Riehl, S., and M. Nesbitt (2003), Crops and cultivation in the Iron ship of agricultural decision-making and climate change in the Age Near East: Change or continuity?, in Identifying Changes— ancient Near East, Quat. Int., 197,93–114. The Transition From Bronze to Iron Ages in Anatolia and Its Riehl, S. (2009b), A cross-disciplinary investigation of cause-and- Neighbouring Regions. Proceedings of the International Work- effect for the dependence of agro-production on climate change shop Istanbul, Nov. 8–9, 2002, edited by B. Fischer, pp. 301–312, in the ancient Near East, in Knochen pflastern ihren Weg. Fest- Zero, Istanbul, Turkey. schrift for Hans-Peter and Margarete Uerpmann, edited by S. Riehl, S., R. A. Bryson, and K. Pustovoytov (2008), Changing Münzel et al., pp. 217–226, Marie Leidorf, Rahden, Germany. growing conditions for crops during the Near Eastern Bronze RIEHL ET AL. 135

Age (3000–1200 BC): The stable carbon isotope evidence, Stern, N. (2007), The Economics of Climate Change. The Stern J. Archaeol. Sci., 35, 1011–1022. Review, Cambridge Univ. Press, Cambridge, U. K. Ristvet, L., and H. Weiss (2005), The Habur Region in the late third Stol, M. (1985a), Beans, peas, lentils and vetches in Akkadian texts, and early second millennium BC, in The History and Archaeol- Bull. Sumerian Agric., 2, 127–139. ogy of Syria, edited by W. Orthmann, pp. 1–26, Saarbrücken, Stol, M. (1985b), Remarks on the cultivation of sesame and the Saarbrücken, Germany. extraction of its oil, Bull. Sumerian Agric., 2,119–126. Roberts, N., T. Stevenson, B. Davis, R. Cheddadi, S. Brewster, and Stol, M. (2003–2005), Öl, Ölbaum, in Reallexikon der Assyriolo- A. Rosen (2004), Holocene climate, environment and cultural gie und Vorderasiatischen Archäologie, vol. 10, edited by D. O. change in the Circum-Mediterranean region, in Past Climate Edzard and M. P. Streck, pp. 32–33, W. DeGruyter, Berlin. Variability Through Europe and Africa, edited by R. W. Battar- Stol, M. (2004), Wirtschaft und Gesellschaft in altbabylonischer bee, pp. 343–362, Springer, Dordrecht, The Netherlands. Zeit, in Mesopotamien. Die altbabylonische Zeit, edited by P. Roberts, N., et al. (2008), Stable isotope records of Late Quaternary Attinger et al., pp. 643–1006, Universitäts-Verlag/Vandenhoeck climate and hydrology from Mediterranean lakes: The ISOMED & Ruprecht, Freiburg, Switzerland. synthesis, Quat. Sci. Rev., 27,2426–2441, doi:10.1016/j.quascirev. Stol, M. (2010), Sesam, in Reallexikon der Assyriologie und Vor- 2008.09.005. derasiatischen Archäologie, vol. 12, edited by M. P. Streck, Rohling, E. J., A. Hayes, P. A. Mayewski, and M. Kucera (2009), pp. 400–404, W. DeGruyter, Berlin. Holocene climate variability in the eastern Mediterranean, and Streck, M. (2010), Großes Fach Altorientalistik: Der Umfang des the end of the Bronze Age, in Forces of Transformation: The End keilschriftlichen Textcorpus, Mitt. Dtsch. Orientgesellschaft, of the Bronze Age in the Mediterranean, edited by C. Bachhuber 142,35–58. and G. Roberts, pp. 2–5, Oxbow, Oxford, U. K. Tainter, J. A. (1988), The Collapse of Complex Societies, Cam- Röllig, W. (2008), Land- und Viehwirtschaft am Unteren Kābūrin bridge Univ. Press, Cambridge, U. K. Mittelassyrischer Zeit, Berichte der Ausgrabung Tall Šēikh Hamad/ Tainter, J. A. (1995), Sustainability of complex societies, Futures, Dūr-Katlimmu 9, Harassowitz, Wiesbaden, Germany. 27, 397–407. Rosen, A. (1997), The geoarchaeology of Holocene environments Terral, J.-F., et al. (2004), Historical biogeography of olive domes- and land use at Kazane Hoyuk, S.E. Turkey, Geoarchaeology, 12, tication (Olea europaea L.) as revealed by geometrical morphom- 395–416. etry applied to biological and archaeological material, J. Biogeogr., Rosen, A., and P. Goldberg (1995), Paleoenvironmental investiga- 31,63–77. tions, in Titriş Hoyuk, A Small EBA Urban Center in Southeast- Tieszen, L. L. (1991), Natural variations in the carbon isotope ern Anatolia: The 1994 Season,editedbyG.Algazeetal., values of plants: Implications for archeology, ecology, and pa- Anatolica, 21,32–37. leoecology, J. Archeol. Sci., 18, 227–248. Sallaberger, W. (2004), A note on the sheep and goat flocks— Ur, J. (2003), CORONA satellite photography and ancient road Introduction to texts 151–167, Subartu, XII,13–21. networks. A northern Mesopotamian case study, Antiquity, 295, Sallaberger, W. (2007), From urban culture to nomadism. A history 102–115. of Upper Mesopotamia in the late third millennium, in Sociétés Van De Mieroop, M. (2003), A History of the Ancient Near East ca. humaines et changement climatique à la fin du troisième millén- 3000–323 BC, Blackwell, Malden, Mass. naire: Une crise a-t-elle eu lieu en Haute-Mésopotamie?, Varia van Koppen, F. (2001), The organization of institutional agriculture Anatolica, vol. 19, edited by C. Marro and C. Kuzucuoglu, in Mari, J. Econ. Soc. Hist. Orient, 44, 451–504. pp. 417–456, Diffusion De Boccard, Paris, France. van Zeist, W. (2001), Third to first millennium BC plant cultivation Sallaberger, W. (2012a), History and philology, in Jezirah (AR- on the Khabur, North-Eastern Syria, Palaeohistoria, 41/42(1999/ CANE 1), edited by M. Lebeau, pp. 319–334, Brepols, Turnhout, 2000), 111–125. Belgium. Vincente, C. (1991), The 1987 Tell Leilan Tablets Dated by the Sallaberger, W. (2012b), Urban organizations for offerings, overland Limmu of Habil-kinu, vol. I and II, Fac. of the Grad. Sch., Yale traffic and the Euphrates trade at pre-Sargonic Mari, in Mari, ni est Univ., Ann Arbor, Mich. ni ouest? 75 ans de découvertes à Tell Hariri, edited by P. Butterlin, Waetzoldt, H. (1983), Leinen (Flachs), in Reallexikon der Assyrio- J.-C. Margueron, B. Muller, IFPO, Paris, France, in press. logie und Vorderasiatischen Archäologie, edited by D. O. Sallaberger, W., and J. Ur (2004), Tell Beydar/Nabada in its regional Edzard, pp. 583–594, W. DeGruyter, Berlin. setting, Subartu, XII,51–71. Waetzoldt, H. (1985), Ölpflanzen und Pflanzenöle im 3. Jahrtau- Staubwasser, M., and H. Weiss (2006), Holocene climate and cul- send, Bull. Sumerian Agric., 2,77–96. tural evolution in late prehistoric-early historic West Asia, Quat. Wanner, H., et al. (2008), Mid- to Late Holocene climate change: Res., 66, 372–387. An overview, Quat. Sci. Rev., 27, 1791–1828. Staubwasser, M., F. Sirocko, P. M. Grootes, and M. Segl (2003), Weiss, H., and M.-A. Courty (1993), The genesis and collapse of Climate change at the 4.2 ka BP termination of the Indus valley the Akkadian empire: The accidental refraction of historical law, civilization and Holocene south Asian monsoon variability, Geo- in Akkad: The First World Empire,editedbyM.Liverani, phys. Res. Lett., 30(8), 1425, doi:10.1029/2002GL016822. pp. 131–155, Sargon, Padua, Italy. 136 AGRICULTURAL SYSTEM TRANSFORMATIONS IN FERTILE CRESCENT

Weiss, H., M.-A. Courty, W. Wetterstrom, F. Guichard, L. Senior, R. Wilkinson, T. J., H. J. Christiansen, J. Ur, M. Widell, and M. Meadow, and A. Curnow (1993), The genesis and collapse of Altaweel (2007), Urbanization in a dynamic environment: Mod- third millennium North Mesopotamian civilization, Science, 261, elling Bronze Age communities in Upper Mesopotamia, Am. 995–1004. Anthropol., 109,52–68. Wick, L., G. Lemcke, and M. Sturm (2003), Evidence of Lateglacial Wilkinson, T., C. French, J. Ur, and M. Semple (2010), The and Holocene climatic change and human impact in eastern geoarchaeology of route systems in northern Syria, Geoarchaeol- Anatolia: High-resolution pollen, charcoal, isotopic and geo- ogy, 25, 745–771. chemical records from the laminated sediments of Lake Van, Zeeb, F. (2001), Die Palastwirtschaft in Altsyrien nach den spätalt- Turkey, The Holocene, 13, 665–675. taltbabylonischen Getreidelieferlisten aus Alalah (Schicht VII), Wilkinson, T. J. (1993), Linear hollows in the Jazira, Upper Meso- Ugarit, Münster, Germany. potamia, Antiquity, 67, 548–562. Zohary, D., and M. Hopf (2000), Domestication of Plants in the Old Wilkinson, T. J. (1997), Environmental ﬂuctuations, agricultural World: The Origin and Spread of Cultivated Plants in West Asia, production and collapse: A view from Bronze Age upper Meso- Europe, and the Nile Valley, 3rd ed., Oxford Univ. Press, Oxford, U. K. potamia, in Third Millennium BC Climate Change and Old World – Collapse, edited by H. N. Dalfes et al., pp. 67 106, Springer, A. Dornauer, Altorientalische Philologie, Institut für Archäolo- Berlin. gische Wissenschaften, Albert-Ludwigs-Universität Freiburg, Platz ﬁ Wilkinson, T. J. (1999), Holocene valley lls of Southern Turkey der Universität 3, D-79085 Freiburg, Germany. (aron.dornauer@ and Northwestern Syria: Recent geoarchaeological contributions, orient.uni-freiburg.de) – Quat. Sci. Rev., 18, 555 571. K. Pustovoytov, Institute of Soil Science and Land Evaluation, Wilkinson, T. J. (2000), Regional approaches to Mesopotamian University of Hohenheim, Emil-Wolff-Str. 27, D-70599 Stuttgart, archaeology: The contribution of archaeological surveys, J. Ar- Germany. ([email protected]) chaeol. Res., 8, 219–267. S. Riehl, Institute for Archaeological Science, University of Wilkinson, T. J. (2003), Archaeological Landscapes of the Near Tübingen, Rümelinstraße 23, D-72070 Tübingen, Germany. East, Univ. of Ariz. Press, Tucson. ([email protected]) Wilkinson, T. J., J. Ur, E. Barbanes Wilkinson, and M. Altaweel W. Sallaberger, Institut für Assyriologie und Hethitologie, (2005), Landscape and settlement in the Neo-Assyrian Empire, Ludwig-Maximilians-Universität München, eschwister-Scholl- Bull. Am. Sch. Oriental Res., 340,22–56. Platz 1, D-80539 München, Germany. ([email protected]) Late Holocene Evolution of the Fuzhou Basin (Fujian, China) and the Spread of Rice Farming

Barry V. Rolett

Department of Anthropology, University of Hawai‘iatMānoa, Honolulu, Hawaii, USA

In ancient China, productive lowlands were vital in the development and spread of rice-dependent economies centered on paddy field farming. This paper compares and analyzes two independent lines of evidence documenting the late Holocene formation of lowlands suitable for paddy field systems in the Fuzhou Basin (Fujian, China). One paleogeographic reconstruction is based on the analysis of sediment cores from the Fuzhou Basin. Stage one of the paleoenvironmental model is marked by early Holocene sea level rise and the mid-Holocene sea level highstand. Stage two is defined by a fall in sea level, at around 1900 B.P., from the mid- Holocene highstand to modern levels. The paleoenvironmental model suggests that the floodplain and other lowlands suitable for irrigated rice agriculture formed after 1900 B.P., prior to which a large paleoestuary filled the Fuzhou Basin. Do ancient Chinese textual records support the paleoenvironmental model? Are the ancient texts relevant in understanding the anthropogenic contribution to environmental change in the Fuzhou Basin? Textual records covering nearly 2000 years of Chinese history reveal close agreement among the paleoenvironmental and text- based geographic models. Agricultural systems based on rain-fed fields may have existed during the mid-Holocene, but lowlands suitable for paddy field systems did not exist until after 2000 B.P.

1. INTRODUCTION Fuller et al., 2011]. As rice farming spread from the Yangtze region, one trajectory of dispersal was along the southeast Domesticated rice (Oryza sativa japonica) originated in the coast of China [Bellwood, 2011] (Figure 1). The rate of spread Yangtze River Delta area, where it is associated with early was influenced by environmental, as well as cultural, factors rice farming during the mid-Holocene [Zong et al., 2007; [Bellwood, 2005]. Because rice is a semiaquatic plant, one Fuller et al., 2010; Zheng et al., 2007]. Rice farming spread key environmental factor was the availability of lowlands from this homeland zone to cover much of Asia, and today, with abundant fresh water; early varieties of rice grew best rice feeds about half of the world’s population [Chang, 2000]. in wet rather than dry fields [Chang, 2000]. The importance Irrigated rice farming is highly productive and can support of water is such that a Song Dynasty (A.D. 960–1279) Chi- large populations. Methane emissions from the paddy fields nese treatise on agriculture declares: “Whether the land be may have begun anthropogenic global warming during the good or poor, if the water is clear then the rice will be good” Neolithic era, up to 5000 years ago [Ruddiman et al., 2008; [Bray, 1984, p. 498]. Mid-late Holocene evolution of the southeast China coast generally created favorable conditions for rice farming, as fl Climates, Landscapes, and Civilizations sea level change, anthropogenic activities, and other in u- Geophysical Monograph Series 198 ences increased the availability of land suitable for paddy © 2012. American Geophysical Union. All Rights Reserved. fields. Unraveling the interrelationships of these various 10.1029/2012GM001224 factors and understanding the history of advances in rice 137 138 LATE HOLOCENE EVOLUTION OF THE FUZHOU

contemporaneous sites including Zhuangbianshan are notable for sedentary villages supported by a maritime-oriented economy, with only slight evidence for agriculture [Chen, 2004; G. Lin, 2008]. Tanshishan culture displays affinities to the early Neolithic cultures of Taiwan, suggesting that Fujian contributed to the Neolithic colonization of Taiwan [Chang,1995; Bellwood,2005;G. Lin, 2008]. Millennia later, Fuzhou City became the capital of Fujian, a province that thrived through international trade, especially in tea, silk, and porcelain. This article examines the recently proposed Fuzhou Basin paleoenvironmental model in light of early textual records kept by Han Chinese government officials and settlers. His- torical texts documenting Fujian reach back nearly 2 millennia. Although much of the information that these texts contain is approximate or qualitative, they are nevertheless valuable for documenting population growth and various aspects of human activity [e.g., Perkins, 1969; Clark, 2009]. I draw upon previous studies of these texts to address two questions: (1) Do ancient Chinese textual records sup- Figure 1. Map of southeast China showing locations of provinces, port the core-based paleoenvironmental model? and cities, and rivers mentioned in the text. The political boundaries for (2) How do the ancient texts contribute to our understanding Fujian and the other provinces have changed somewhat over time; of anthropogenic aspects of environmental change? Finally, I those shown are the modern boundaries. Latitude and longitude for Fuzhou City are 26.0739°N and 119.2967°E. farming requires a multidisciplinary approach involving the natural sciences, archaeology, and history. Multidisciplinary research on the emergence and spread of rice farming in the Yangtze River Delta area reveals the domestication of rice around 8000 years ago [Zong et al., 2007; Fuller et al., 2010], followed by the rise of early rice-dependent Neolithic cultures around 6500 cal B.P. [Bellwood, 2005; Fuller and Qin, 2010], and finally, the expansion of rice farming through hydraulic engineering and land reclamation [Elvin, 2004]. Some 750 km south of the Yangtze River Delta, an ongoing paleoenvironmental study of the Fujian coast traces the evolution of the Fuzhou Basin [Rolett et al., 2011], with significant implications for understanding the spread of rice farming. The Fuzhou Basin, which drains the mouth of the Min River, supported dense rice-dependent urban settlements since early in Chinese history. Paleoenvironmental reconstruction reveals that the floodplain and other lowlands suitable for irrigated rice formed after 2000 B.P., prior to which the Fuzhou Basin was dominated by an estuary that extended more than 75 km inland [Rolett et al., 2011] (Figure 2). Although the Fuzhou Basin is only one point in the rice farming spread zone, it played a prominent role in the rise of Chinese civilization. The basin contains the Tanshishan ar- Figure 2. Map of the Fuzhou Basin showing (a) the present-day chaeological site, the most intensively investigated Neolithic environmental setting and (b) a paleogeographic reconstruction for site in Fujian and type site for the Tanshishan culture (5000– the time period around 6000 cal B.P. [after Rolett et al., 2011]. Latitude 4300 cal B.P.) [G. Lin, 2008] (Figure 3). Tanshishan and and longitude for Fuzhou City are 26.0739°N and 119.2967°E. ROLETT 139

Figure 3. Map of the Fuzhou Basin showing the locations of archaeological and coring sites mentioned in the text. The light gray zone indicates the floodplain and lowlands below 50 m in elevation. Map created by S. Allen. © Google Maps, 2012. bring attention to the significance of these issues in modeling of sediments from recently collected cores [Rolett et al., the spread of rice farming. 2011], but also supported by preliminary palynological data (Y. Yue and Z. Zheng, personal communication, 2012). The 2. MODERN ENVIRONMENTAL SETTING cores sample deposits from near the mouth of the Min River, as well as from areas upstream of Fuzhou City. Core FZ5 was The modern coast of southeast China is characterized by collected in Minhou County, 75 km from the sea, at the base narrow coastal plains, large estuaries and ocean bays, wind- of the Neolithic era Zhuangbianshan archaeological site ing shorelines, and thousands of offshore islands. The Fuzhou (Figure 3). The following summary of the paleoenvironmen- Basin was formed during regional volcanic activity and tec- tal model is based on the work of Rolett et al. [2011]. tonic uplift [Lu, 2004]. It is surrounded by low mountains and In stage one, early Holocene sea level rise began to inun- is more than 20 km wide in the Fuzhou City area (Figure 3). date the Fuzhou Basin around 9000 cal B.P., and continued As the Min River flows through the basin, it splits into two sea level rise during the approach to the mid-Holocene high- large branches just upstream of Fuzhou. This split in the stand created a large estuary (Figure 2). Neolithic settlers river represents the present-day maximum inland reach of with a marine-oriented way of life founded settlements on the tides. The branches of the river reunite before flowing small islands in the upper estuary. Sediment transported to into the modern Min Estuary, which is tightly constrained by the coast by the Min River accumulated mainly within the mountains as it approaches the sea. The modern estuary is Fuzhou Basin itself. Core FZ4 records 30 m of Holocene much smaller than the paleoestuary that reached around sediment accretion in the lower estuary, but without large- 75 km inland from the modern coast, to the present-day scale delta formation. This history of landscape evolution location of Minhou, during the mid-Holocene highstand. contrasts with that for the Lower Yangtze River area, where the mid-Holocene slowing of sea level rise is linked with 3. THE FUZHOU BASIN shoreline progradation and the formation of an expansive PALEOENVIRONMENTAL MODEL delta plain well suited for rice agriculture [Chen et al., 2008]. Stage two of the paleoenvironmental model is marked by The paleoenvironmental reconstruction is a two-stage a fall in sea level from the mid-Holocene highstand to near model based mainly on chemical analysis and AMS dating modern levels. This occurred close to 1900 cal B.P., as 140 LATE HOLOCENE EVOLUTION OF THE FUZHOU dated by peat deposits in core FZ5. As the ocean retreated, p. 222]. By this time, Chinese towns were established on the the area surrounding FZ5 was transformed to a brackish floodplain near Minhou, around 20 km upstream from Fuz- swamp within a tidal zone environment close to the river hou City. However, much of the floodplain still consisted of mouth. Based on our reconstruction, the transition from an wetlands. estuary-dominated to a mainly fluvial environment began Fuzhou City continued to expand throughout the Song rather abruptly around 1900 years ago. The Fuzhou Basin Dynasty (A.D. 960–1279). Ships no longer had direct access cores show significant sedimentation during centuries fol- to the city because of the accumulated sediment in the lowing the final Holocene marine regression, especially ancient bay. Large sand bars began to develop by A.D. near the urban center of Fuzhou City; however, estimates 1050 [T. Lin, 2008, pp. 222–223]. Sedimentation and land of sediment accumulation rates will require higher chrono- emergence around the city steadily progressed during the logical resolution than is presently available. Sedimentation Ming Dynasty (A.D. 1368–1644). The area of the south bay, after 1900 cal B.P., during the stage two fluvial setting, which today consists of terraced land, was wetlands. Large created the first lowlands suitable for irrigated rice agricul- sand bars filled much of the river channel. The present-day ture in Fuzhou Basin. configuration of the Min River and its banks began to take shape [T. Lin, 2008, pp. 223–224]. 4. A TEXT-BASED GEOGRAPHIC HISTORY In the Qing Dynasty (A.D. 1644–1911), sand bars in the OF THE FUZHOU BAY river channel joined together to form land masses, complet- ing a process that began almost 1000 years earlier. Near For comparison of the paleoenvironmental model with the Fuzhou City, islands in the ancient bay, including those that historical record, I draw upon the research of historical separate the northern and southern branches of the Min geographer Lin Tingshui, who examined early Chinese texts River, were consolidated into larger land masses [T. Lin, to compile a geographic history of Fuzhou City [T. Lin, 2008, p. 226]. 2008]. Lin’s study is supplemented by maps created for the Fujian geographical atlas [Lu, 2004]. My summary below of 5. ACCORDANCE OF THE PALEOENVIRONMENTAL fi his ndings, based on these Chinese language publications, AND TEXT-BASED GEOGRAPHIC MODELS is the first English account of Lin’s text-based geographic reconstruction. Dates for the Chinese dynasties are recorded A number of conclusions can be drawn by comparing the in calendar years, as is standard practice among historical paleoenvironmental model, based on sediment cores, with sources. T. Lin’s [2008] historical model based on early Chinese texts. The earliest written description of Fuzhou Bay is from the First, it is significant to note that the peat deposit sea level Han Dynasty (206 B.C. to A.D. 220). At this time, the bay indicator in core FZ5 dates the final Holocene marine regres- shoreline reached nearly to the foot of the mountains lying sion to 1900 cal B.P. Thus, the core-based paleoenvironmen- directly north of Fuzhou City. According to Han government tal model suggests the emergence of inhabitable Fuzhou records, an indigenous king of the Yue culture moved from Basin lowlands around this time [Rolett et al., 2011]. This the mountains in northern Fujian to Fuzhou during the early finding coincides closely with historical records that date Han times [T. Lin, 2008, p. 220]. The king and his followers Fuzhou City origins to the Han Dynasty. Han era descrip- settled on a peninsula in the north bay. This area was sur- tions portray the city setting as a peninsula surrounded by rounded by swamp and marshland. The location is near the swamp and marshland, suggesting that establishment of the modern center of Fuzhou City [T. Lin, 2008, p. 220]. first settlement occurred even as the shoreline of the paleoes- During the Jin Dynasty (A.D. 265–420), as in earlier his- tuary continued to migrate seaward. Prior to emergence of toric times, Fuzhou City was situated on a peninsula reaching the Fuzhou Basin lowlands, indigenous peoples inhabited into the bay [T. Lin, 2008, p. 220]. The East and West Lakes islands and tablelands, especially in the upper paleoestuary were created to drain wetlands near the city center, allowing [Rolett et al., 2011]. Evidence of these earlier settlements is expansion of the city to the south. Another water feature, a documented by archaeological sites such as Zhuangbianshan wide canal, was added along the southern fringe of the city to and Tanshishan, both of which date back to Neolithic times provide direct access to Fuzhou Bay and the sea. [G. Lin, 2008]. During the Tang Dynasty (A.D. 618–907), Fuzhou Bay The paleoenvironmental and text-based geographic models continued to reach the southern part of the city. The land in also agree that the Min River floodplain, together with other the city’s southern district remained swampy, and an impor- lowlands in the Fuzhou Basin, formed during the late Holo- tant temple was moved to higher ground as a result of cene. This point is clear in the historical texts, which record problems caused by the wet conditions [T. Lin, 2008, Chinese towns on the floodplain near Minhou (Figure 3) ROLETT 141 during the Tang Dynasty (A.D. 618–907) and final closing up of the paleoestuary beginning in the Ming (A.D. 1368– 1644). By this time, more than 1500 years after the arrival of the Han Chinese, evolution of the Fuzhou Basin was strongly impacted by anthropogenic activities.

6. TEXT-BASED EVIDENCE FOR THE ANTHROPOGENIC CONTRIBUTION TO ENVIRONMENTAL CHANGE IN THE FUZHOU BASIN

Population statistics are a good starting point for examining the record of human impact on the landscape. Histor- ical research by Perkins [1969] and So [2000] compiles early population data for Fujian, including Fuzhou and other coastal areas. These data derive from official counts of the ethnic Han Chinese population. The Han records do not count the indigenous Yue peoples, as the two groups generally lived apart. The counts are based mainly on village surveys in which the number of houses was used to calculate the number of households. Historians [e.g., Per- kins, 1969] usually assume the average household contained five family members. Archaeological evidence [e.g., G. Lin, 2008] suggests that the native Yue peoples lived in small, dispersed villages prior to Han colonization. The Han empire reached Fujian and Guangdong by 100 B.P. [Clark, 2009, p. 13]. Following Figure 4. Population and arable acreage data for Fuzhou City and – colonization, the Yue population decreased through extermi- Fujian Province A.D. 742 1950. Black lines plot population and acreage data for Fujian Province [Perkins, 1969]. Dashed lines plot nation, displacement, and assimilation. However, census re- population and acreage data for Fuzhou City, the capital of Fujian cords for A.D. 450 indicate that the Han population was still [So, 2000]. small at this time, numbering only 2843 households for the entire coast of Fujian [Clark, 1991, p. 15]. Within a few centuries, the demographic situation had changed dramati- during the same period was the introduction of new plow cally; Clark [2009] notes that by A.D. 750, the remaining technology century [So, 2000]. These changes stimulated Yue were fully subjugated to Han rule. The Han Chinese increasing food production, as reflected by the number of population grew rapidly from less than 100,000 households acres under cultivation, which rose steadily with the growing in the A.D. 742 census to nearly 1,000,000 by A.D. 1080 population (Figure 4). (Figure 4). In turn, rapid increases in population and expanding agri- Along with a population boom, the eleventh century cultural systems began to leave their mark on the landscape. brought far-reaching cultural and economic transformation Han records indicate that even as the slopes surrounding to the Fujian coast. Coastal cities, especially Quanzhou and Fuzhou were cleared and terraced for rice cultivation, during Fuzhou, entered an era of international influence through the Song Dynasty (A.D. 960–1279) the Chinese also began trade and cultural exchange. Fujian merchants established to exploit these same areas for mining and tea plantations prosperous ventures with direct ties to Korea and Southeast [T. Lin, 2008]. The result was a sharp increase in human- Asia, while communities of expatriate Muslim traders took induced erosion. Indeed, Song records state that sedimenta- root in Quanzhou and Fuzhou [Clark, 2009]. Agriculture and tion linked to increased erosion of the Fuzhou Basin slopes, as land use patterns were revolutionized by the introduction of well as those along the neighboring coast, noticeably contrib- Champa rice from Southeast Asia during the late tenth or uted to the creation of arable lowlands [T. Lin, 2008, p. 222]. early eleventh century [So, 2000]. Champa rice is early Around 1200, however, Fujian’s international trade began maturing, allowing double cropping in paddy fields, and to collapse, initiating a cycle of emigration and declining because it is drought-resistant, it could also be grown as a population; this downturn lasted around 200 years, until the drycropinhillyareas[Bray, 1984]. Another innovation Ming Dynasty when a new boom began [Clark, 2009]. 142 LATE HOLOCENE EVOLUTION OF THE FUZHOU

7. DISCUSSION AND CONCLUSIONS Chang, T. T. (2000), Rice, in The Cambridge World History of Food, edited by K. F. Kiple and K. C. Ornelas, pp. 132–149, Both the paleoenvironmental and text-based geographic Cambridge Univ. Press, Cambridge, U. K. models indicate that in the Fuzhou Basin region, and quite Chen, Z. S. (2004), Diet and subsistence strategies of the Tan- likely other areas of the Fujian coast, the final marine regres- shishan culture [in Chinese with English abstract], Fujian Wenbo, sion occurred after around 2000 years ago. This finding is 48,41–46. consistent with variability in sea level data for the southeast Chen, Z., Y. Zong, Z. Wang, H. Wang, and J. Chen (2008), Migra- coast of China analyzed by Zong [2004]. It is also significant tion patterns of Neolithic settlements on the abandoned for the hypotheses proposed to explain the spread of rice Yellow and Yangtze River deltas of China, Quat. Res., 70, 301–314. farming and the development of Neolithic and later cultures Clark, H. R. (1991), Community, Trade, and Networks: Southern in southeast China. Fujian Province From the Third to the Thirteenth Century, Cam- It is widely suggested that rice farming spread from the bridge Univ. Press, Cambridge, U. K. Yangtze River Delta homeland area to the Fuzhou Basin, and Clark, H. R. (2009), Frontier discourse and China’s maritime fron- even to the Pearl River by 5000 B.P. [e.g., Bellwood, 2005; tier: China’s frontiers and the encounter with the sea through Zhang and Hung, 2010]. However, direct archaeobotanic early imperial history, J. World Hist., 20,1–33. evidence for Neolithic rice farming is scarce in coastal areas Elvin, M. (2004), The Retreat of the Elephants: An Environmental to the south of the Yangtze [Zhao, 2011]. Furthermore, there History of China, Yale Univ. Press, New Haven, Conn. is no evidence at all to ascertain that the earliest rice farming Fuller, D. Q. (2011), Pathways to Asian civilizations: Tracing the in coastal Fujian and Guangdong was based on irrigated origins and spread of rice and rice cultures, Rice, 4,78–92. paddy fields rather than hillside dryland agricultural systems Fuller, D. Q., and L. Qin (2010), Declining oaks, increasing artistry, [Fuller, 2011]. The results presented here from two indepen- and cultivating rice: The environmental and social context of the dent lines of evidence show that, for the Fuzhou Basin, emergence of farming in the Lower Yangtze Region, Environ. lowlands suitable for paddy field systems formed only within Archaeol., 15, 139–159. the last 2 millennia and mainly during the last 1000 years. Fuller, D. Q., Y. Sato, C. Castillo, L. Qin, A. R. Weisskopf, E. J. Agricultural systems based on rain-fed fields may have ex- Kingwell-Banham, J. Song, S. Ahn, and J. van Etten (2010), isted during the mid-Holocene, but large-scale irrigated rice Consilience of genetics and archaeobotany in the entangled history of rice, Archaeol. Anthropol. Sci., 2,115–131. agriculture was not feasible until after around 1900 B.P., Fuller, D. Q., J. van Etten, K. Manning, C. Castillo, E. Kingwell- when falling sea level dramatically transformed the coast Banham, A. Weisskopf, L. Qin, Y. Sato, and R. J. Hijmans and created emergent lowlands. (2011), The contribution of rice agriculture and livestock pastoralism to prehistoric methane levels: An archaeological assess- Acknowledgments. I thank Hui-Lin Lee for assistance in trans- ment, Holocene, 21, 743–759. lating the Chinese language publications on the Fuzhou Basin. Lin, G. (2008), Introduction: A brief review of studies on the Tingshui Lin of Xiamen University introduced me to the historical Tanshishan site and Tanshishan culture [in Chinese with English geography of southeast China, and I am grateful for his encourage- abstract], in Collection of Papers on the Tanshishan Culture, ment and inspiration. I am also grateful to my Chinese collaborators edited by G. Lin, pp. 2–11, Fujian Prov. Tanshishan Site Mus., Zhuo Zheng and Yuanfu Yue. Figures 1 and 3 were created by Scott Fuzhou, China. Allen. My work in China was supported by the Andover Founda- Lin, T. (2008), The historic period transformation of the Fuzhou tion for Archaeological Research (through grants from L. T. Clay). Bay[inChinesewithEnglishabstract],Hist. Geogr., 23, 220–226. REFERENCES Lu, M. (Ed.) (2004), Fujian Geographical Atlas [in Chinese], Fujian Atlas, Fuzhou, China. Bellwood, P. (2005), First Farmers: The Origins of Agricultural Perkins, D. H. (1969), Agricultural Development in China 1368– Societies, Blackwell, Oxford, U. K. 1968, Aldine, Chicago, Ill. Bellwood, P. (2011), The checkered prehistory of rice movement Rolett, B. V., Z. Zheng, and Y. Yue (2011), Holocene sea-level southwards as a domesticated cereal—From the Yangzi to the change and the emergence of Neolithic seafaring in the Fuzhou Equator, Rice, 4,93–103. Basin (Fujian, China), Quat. Sci. Rev., 30, 788–797. Bray, F. (1984), Science and Civilisation in China, vol. 6, Biology Ruddiman, W. F., Z. Guo, X. Zhou, H. Wu, and Y. Yu (2008), Early and Biological Technology Part II: Agriculture, edited by rice farming and anomalous methane trends, Quat. Sci. Rev., 27, J. Needham, Cambridge Univ. Press, Cambridge, U. K. 1291–1295. Chang, K. C. (1995). Taiwan Strait archaeology and Proto-Austro- So, B. K. L. (2000), Prosperity, Region, and Institutions in Mari- nesian, in Austronesian Studies Relating to Taiwan, edited by P. J. time China: The South Fukien Pattern, 946–1368, Harvard Univ. Li et al., pp. 161–183, Academia Sinica, Taipei, Taiwan. Asia Cent., Cambridge, Mass. ROLETT 143

Zhang, C., and H. Hung (2010), The emergence of agriculture in Zong, Y., Z. Chen, J. B. Innes, C. Chen, Z. Wang, and H. Wang southern China, Antiquity, 84,11–25. (2007), Fire and flood management of coastal swamp enabled Zhao, Z. (2011), New archaeobotanic data for the study of the first rice paddy cultivation in east China, Nature, 449, 459–462. origins of agriculture in China, Curr. Anthropol., 52, S295–S306. Zheng, Y., G. Sun, L. Qin, C. Li, X. Wu, and X. Chen (2007), Rice fields and modes of rice cultivation between 5000 and 2500 BC in east China, J. Archaeol. Sci., 36, 2609–2616. B. V. Rolett, Department of Anthropology, University of Hawai‘i Zong, Y. (2004), Mid-Holocene sea-level highstand along the at Mānoa, 2424 Maile Way, Honolulu, HI 96822, USA. (rolett@ southeast coast of China, Quat. Int., 117,55–67. hawaii.edu) Climate-Induced Changes in Population Dynamics of Siberian Scythians (700–250 B.C.)

Irina P. Panyushkina

Laboratory of Tree-Ring Research, University of Arizona, Tucson, Arizona, USA

Demography and population dynamics of Siberian Scythians are fundamentally underexplored topics. Scarcity of known settlements has hampered our ability to fully understand the Siberian Scythians inhabiting the Altai Mountains, known as the Pazyryk Culture. However, expanded investigation of mortuary complexes by means of dating with radiocarbon and tree rings may reveal the dynamics of Scythian pastoralists. Generally, population dynamics recorded in archaeological data trace major social and cultural changes, which relate directly to the economic sustainability of past societies. Climate has been an indirect but important factor affecting demographic trends, triggering changes in supply of important ecosystem services. This paper examines the synergy of environmental change and behavioral adaptation of Scythian horsemen in the Russian Altai. We developed well-replicated tree ring chronologies from archaeological timbers of the Pazyryk Culture (sixth to third century B.C.) to describe past temperature variability in the region. Three distinctive episodes of Altai climate change appear to be tracking three major cultural phases of the Siberian Scythians advancement: (1) 700–480 B.C., cold and highly variable climate; (2) 480–360 B.C., mild warm climate and stable environmental conditions; and (3) 360–250 B.C., turbulent cold climate with ampliﬁed decadal variability. Demographic study of mortality linked to the Sibe- rian Scythian chronology shows population growth during the cold periods. The evidence demonstrates successful behavioral adaptation of Siberian Scythians to colder climates. The trend in population dynamics suggests an increase of mobility and possible intensiﬁcation of Pazyryk interactions with the outside world during warm times and a subsequent decrease in occupancy of the Altai.

1. INTRODUCTION 800 years. It was a very dynamic time across this vast geographical region designated the Eurasian Steppe and The term “Siberian Scythians” refers to seminomadic dominated by the economic strategies of mobile pastoralism. tribes occupying the heart of Eurasia: Altai-Sayan Mountains Royal families and local elites controlled and facilitated during the first millennium B.C. Broadly, these tribes were a south-north and west-east trading routes on a truly global part of the Scythian world reaching from the Black Sea scale [Jacobson, 1995]. The elite Siberian Scythians in the region to Lake Baykal over 4000 km and thriving for about remote outskirts of the Eurasian Steppe had access to fine Chinese silk, Persian rugs, and Greek pottery [Rudenko, 1970; Polosmak, 2001; Parzinger, 2006]. Their prosperity fi Climates, Landscapes, and Civilizations allowed them to have slaves, possess lavish golden out ts Geophysical Monograph Series 198 weighing many kilograms, and elaborate their burials with © 2012. American Geophysical Union. All Rights Reserved. mummified bodies, dozens of horses, and wooden chambers 10.1029/2012GM001220 replicating dwellings. The chronology of the Pazyryk has 145 146 POPULATION DYNAMICS OF SIBERIAN SCYTHIANS been dated using wiggle matching of radiocarbon dates and However, highly dense clusters of ordinary and small-size tree ring cross dating [Sljusarenko et al., 2004; Hajdas et al., burials are present mostly at high elevations in the southern 2004; Panyushkina et al., 2007] and appears to be in agree- part of the Russian Altai. The distribution of the Siberian ment with the Eurasian Scythian radiocarbon chronology Scythian population is highly uneven and correlates with the [Zaitseva et al., 2005]. deforested landscape of steppe and alpine meadows that have The archaeological artifacts decorated with distinctive the most abundant fodder resources for animal stock. Is it animal-style art of Scythians found in burial sites remain the possible that climate variability and deforestation were im- best evidence confirming the existence of interconnected portant factors impacting the demography and population cultural communities across the Eurasian Steppe. Mobility dynamics of the Siberian Scythians? associated with horseback riding was the key element to the As previously observed, cold and dry climatic conditions wealth and social development of these tribes [Levine et al., corresponded to reduced regional woody coverage in south- 2003]. The Scythian economy heavily depended on livestock ern Siberia at circa 1200 B.C., but seemed to promote woody breeding (horses, cattle, sheep, and goats). The Siberian plant growth in other parts of central Asia, for example, Scythians inhabited mountainous landscapes that offered di- central Kazakhstan [Kleinen et al., 2011]. Bioclimatic verse seasonally used vertical pastures. Multiple lines of pollen-based modeling finds replacement of wet dark taiga archaeological evidence demonstrate vertical seasonal migra- (fir) with light taiga forest (pine and larch) and birch stands in tion of Scythian horse breeders in central Asia based on the northern parts of the Altai-Sayan Mountains [Tchebakova settlement pattern (e.g., Saka in the Tian-Shan Mountains and et al., 2009]. The long-term tendency of cold and dry climate Dzungaria, Liushui in the Kunlun Mountains, Xiongnu in the persisted until circa 200–100B.C.intheRussianAltai Khangai Mountains) [Chang, 2008; Frachetti, 2008; Houle, [Mikhaylov and Redkin, 2000]. Archaeological data of Sibe- 2010; Wagner et al., 2011]. The kin-based groups moved rian Scythians show intensive development of pastoral from winter pastures at lower elevations to spring and summer mobility in the Altai-Sayan mountains corresponding to the pastures at higher elevations as the vegetation seasons pro- Early Scythian period 800–600 B.C. [Jacobson-Tepfer, 2008]. gressed. Distances of seasonal migrations varied greatly from Given easier access and exploitation of mountain pastures as a as little as 3–5to30–50 km. The higher productivity of result of natural deforestation, the transhumant mode of pas- mountain pastures over the grasslands in plains is widely toralism (seasonal movement of livestock) appeared in the known from historical livestock production [Masanov, Altai Mountains in the Classic period of the Siberian 1995]. At the same time, the mountain pastures challenge Scythians between 500 and 300 B.C. [Levine et al., 2003]. herders more often with highly variable weather patterns, for The final period of Siberian Scythians associated with the example, when cold springs delay snowpack melting and later Pazyryk Culture brings mobile pastoralists deeper and higher create flood surges or when ice crust forms on grass surfaces into the mountains. The peak of Pazyryk inhabitation of Altai due to freezing temperatures in midsummers and early fall. In only lasted half of a century [Panyushkina et al., 2007], the Russian Altai, success of transhumant pastoralism de- followed by a massive withdrawal of the Siberian Scythians pends more on temperature and less on precipitation because from the Altai at circa 240 B.C. as dated by various inde- of favorable topography, orography, glaciation, and relict pendent sources [Jacobson, 1995; Zaitseva et al., 2005; permafrost, which compensate moisture deficit in dry and Panyushkina et al., 2007]. Migration pathways of Siberian warm climates and offer alternative pastures unstressed by Scythians withdrawal are much more complex than can be drought [Bolortsetseg and Tuvaansuren, 1996]. explained by the current Eurasian Steppe archaeology [Kor- Demography of the Siberian Scythians is vastly under- yakova and Epimakhov, 2007]. Indeed, generally, Siberian studied. To date, no clear evidences of Pazyryk dwellings Scythian demography responded to the deforestation and have been found in the Altai Mountains [Dashkovskiy, 1998]. long-term climatic fluctuations, but how did short-term cli- Mortuary complexes, largely excavated since the late matic variability impact the Siberian Scythian population nineteenth century, remain the only source of information on size and residential pattern? The main objective of this study occupational patterns. In cases where pathways of seasonal is to investigate the impact of changes in climate variance on pastoral migration are short, distribution and relative density the population size and population dynamics of mobile pas- of settlement sites correlate with mortuary and ritual sites toralists in Altai using tree rings. [Chang, 2008; Houle, 2010]. Burials are typically located along riverbanks or on river floodplains. The most important 2. DATA AND METHODS and largest burials tend to occur at low elevations of the northern part of the Altai-Sayan Mountains and Minusink The study area is located in the southeastern Russian Altai Basin (e.g., Arjan-1 and -2; Pazyryk-5, Tuekta-1, Berel-1). Mountains near the Russian borders with Kazakhstan, China, PANYUSHKINA 147 and Mongolia (Figure 1). Geographically, the area comprises Polosmak [Rudenko, 1970; Kubarev, 1987, 1991; Derevyanko the Ukok Plateau and Chuya and Ulagan Steppe delineating and Molodin, 2000; Polosmak, 2001]. basins between high mountain ranges covered with glaciers. The tree ring width measurements and cross dating were Tree ring records from five locations of the Pazyryk culture done with TSAP (software by www.rinntech.de) at the Lab- clustering at elevations of 1400, 1800, 2100, and 2300 m oratory of Tree-Ring Research, University of Arizona. Based above sea level (asl) were assembled with the help of archae- on a high correlation of overlapping tree ring series between ologists from the SB RAS Institute of Archaeology and kurgans and cemeteries, two composite tree ring records of Ethnography (Novosibirsk, Russia). Size of the burial fields larch (Larix sibirica) were developed. A 453 year tree ring varies and may contain a dozen or many dozens of kurgans. chronology at high elevation (1800–2350 m asl) combines A kurgan is a burial mound made of earth and stones ranging data sets of nine cemeteries/29 kurgans (Figure 2) scattered in diameter from 3 to 20 m and usually piled to a depth across four areas in the Chuya River Basin and Ukok Plateau between 0.3 and 0.5 m. Large kurgans built for warriors and (about 120 km apart). A 295 year tree ring chronology at low chiefs extend up to 80 m in diameter and rise 2 m above the elevation (1400 m asl) combines 36 wood specimens from ground (e.g., Arjan-2, Pazyryk-5 kurgans). The Altai’s cli- five kurgans of the Pazyryk cemetery in the Ulagan Valley, mate, cold and dry, has contributed to exceptional preserva- 200 km southward from the Chuya Basin. Both chronologies tion of archaeological timbers and other organic matter have strong common signal strength with intraseries correla- buried in kurgans for the last 2800–2250 years. Beams and tion coefficients of 0.6 and higher. As shown in Figure 2, planks of kurgan chambers provide complete cross-sections sample size per each location (subset) is remarkably high with pith and outer rings. Length of the archaeological tree (>17 trees), and the chronology sample depth (number of ring series varies from 90 to 390 years. A total of 190 wood trees contributing to the chronology at any date) is consis- specimens used in this study came from original excavations tently high with the exception of the first couple of decades. of S.I. Rudenko, V.D. Kubarev, V.I. Molodin, and N. V. Radiocarbon wiggles derived from the cross-dated tree rings

Figure 1. Location of the study area centered on the Russian Altai region in central Asia. During the first millennium B.C. (Iron Age), the region was occupied by culturally similar tribes of mobile pastoralists whose networks encompassed the Scythian World in this part of Eurasian Steppe: Siberian Scythians in the Altai-Sayan Mountains, Saka in the Tian Shan Mountains, Xiongnu in the Khanui Mountains, and the Gushi in the Tarim Basin and Xinjiang. The light gray lines indicate major rivers; black areas are lakes; gray shaded areas are mountains. 148 POPULATION DYNAMICS OF SIBERIAN SCYTHIANS indicate that the two developed tree ring records overlap [Sljusarenko et al., 2004; Hajdas et al., 2004]. The calendar age of the floating chronology span begins at circa 720 B.C. and extends to circa 250 B.C. We used temporal distribution of cemetery ages derived from our data set and other published information to approximate trends in population size of the studied area. The archaeological evidence of Pazyryk population size and occupation density was drawn from published cemetery data and archaeological materials dated with tree rings that inform us about timing of kurgan construction. The Scythian cemetery Ulandryk-I was nearly completely excavated back in the 1980s [Kubarev, 1987] and reexcavated recently to assemble a complete collection of burial timbers [Panyushkina et al., 2007] suitable for a Figure 3. Smoothed curves of detrended and standardized tree ring case study of Pazyryk demography. width measurements of archaeological timbers from high elevations, which were averaged into a composite tree ring record. The For climatic assessment, variance of the cross-dated tree calendar age of this record (circa 700 to 250 B.C.) was determined ring width series was stabilized through detrending with by radiocarbon wiggle matching of an overlapped Ulandryk-IV either Hugershoff growth spline or alternatively the Fried- series [Sljusarenko et al., 2004; Panyushkina et al., 2007]. Gray man supersmoother at alpha 5 [Friedman, 1984]. The de- line is the mean. trending removes an individual age-growth-related variance and other background noise while preserving high-frequency climate-related variance in the tree ring series (Figure 3). all, a 450 year span composed of the two overlapped records Calculation of tree ring indices as residuals of the estimated was used for climatic interpretation. age curve and raw measurements stabilizes tree ring variance and produces unbiased tree ring indices for environmental 3. RESULTS AND DISCUSSION studies [Cook and Peters, 1997]. Additionally, careful screening of samples and evaluation of strength of intertree The Sayan-Altai dendroclimatic network has identified relationships with the expressed population signal (EPS) summer temperature as the main factor limiting radial tree statistic resulted in the removal of a number of tree ring growth from the upper tree line at 2200–2400 m asl to the series and exclusion of a few decades at the beginning and lower tree line at 1400 m asl [Magda et al., 2011]. Larch a few years at the end of the records. Thus, the new high- tree ring records at low elevations covered with mountainous elevation record has 445 years and 117 series, and the final forest-steppe vegetation have a mixed temperature-precipitation low-elevation record is 238 years in length (Figure 4). Over- signal. The precipitation signal weakens significantly during cold periods during which coherence of temperature-related tree ring variance from low and high elevations increases [Magda et al., 2011]. Well-replicated tree ring records from upper tree lines retain a regional climatic signal in their variance. The modern upper tree line chronologies of larch in the Russian Altai and Sayan Mountains correlate significantly (R = 0.25–0.37 at p < 0.05) proving common response to synoptic-scale circulation over south Siberia. In the vicinity of the gathered archeological timbers, tree ring width Figure 2. Numbers of archaeological tree ring specimens in four records at modern tree lines (Kurayskiy and Severo-Chuys- fl studied subsets of oating chronologies of Siberian Scythians from kiy Ranges) show a strong positive relationship with the the Russian Altai. Shaded area indicates replication of low-elevation June–July temperature [Panyushkina et al., 2005]. Variance samples for the Pazyryk cemetery in the Ulagan Steppe (subset 1). and means of tree ring indices of these records and the Lines show sample sizes of cemetery tree ring records from high elevation. Gray line indicates timbers from subset 2: the Ulandryk-I, archaeological site chronologies are remarkably similar (Fig- Ulandryk-IV and Tashanta-I cemeteries; black dashed line indicates ure 5) suggesting the presence of a strong summer tempera- subset 3: Barbugazy-I and Yustiyd-I cemeteries in the Chuya Steppe; ture signal in the Scythian archaeological tree rings. The black solid line indicates subset 4: cemeteries of Ak-Alakha-II, -III, composite record of archaeological timbers collected from and -V. Kaldjin-I, -II on the Ukok Plateau. burial fields above the modern tree line to 100–400 m below PANYUSHKINA 149

Figure 4. Scythian tree ring chronologies from (a) high and (b) low elevation associated with summer temperature anomalies. Thick line indicates bidecadal variability of tree ring indices smoothed with a 20 year Hamming weights low- pass filter. Positive z scores of the high-elevation record (Figure 4a) refer to warm summer temperature, and negative scores indicate cold summers. Variance of the low-elevation record most likely contains mixed temperature and precipitation signal commonly found in living larch chronologies from the same area [Magda et al., 2011]. (c) Discrepancy (difference) between decadal variability of these two records. The difference in sign and amplitude of the records reveals periods when the moisture signal surpasses the temperature signal. This gives insight into precipitation variability for those periods of the studied records when the discrepancy is clearly seen. For example, climate during 370–350 B.C. was cold and wet, and the 430–410 B.C. period was warm and dry. Labels below Figure 4c delineate three periods of Scythian archaeological chronology in the Eurasian Steppe including south Siberia. the tree line is designated here as the high-elevation chronol- The Siberian Scythian high-elevation record reveals three ogy. Abundant frost rings and partially formed or one- or multidecadal-scale periods of summer temperature change: two-cell rings provide additional strong evidence for temper- (1) 700–480 B.C., cold and high variance climate; (2) 480– ature signal in the high-elevation archaeological record (Fig- 360 B.C., mild warm climate and the most stable environ- ure 6). The second composite record of the Pazyryk cemetery mental conditions; and (3) 360–250 B.C., turbulent cold located at 1400 m asl is termed the low-elevation chronolo- climate with the amplified decadal variability (Figure 4). gy. Both variance and tree ring widths of this record are The first period was the coldest time of the Siberian Scyth- much greater; supporting the presence of a common signal ian occupation of the Altai. To be fair, the two warmest related to both temperature and precipitation as seen in the decades (620–610 and 580–570 B.C.) are documented with- living trees (see outlier in Figure 5). in this period as well. Two multidecadal periods (650–600 150 POPULATION DYNAMICS OF SIBERIAN SCYTHIANS

through the Minusink Basin [Parzinger, 2006]. Most radiocarbon dates for the Aldy-Belsk Culture kurgans fall on the Hallstatt Plateau of the 14C calibration curve, which places them between 700 and 500 B.C. with reduced precision [Zaitzeva et al., 2005]. The next period (480–360 B.C.) had a much warmer climate. The tree ring record shows reduced variance during this time, which accounts for stable environmental conditions most suitable for pastoralist practices. Sparse dates for kurgans from this time prevent detailed interpretations of occupational patterns. However, the transitional nature of demographic patterns is suggested. The elite Tuekta-I kurgan is dated with tree rings to circa 430 B.C. The Scythian radiocarbon chronology offers half as many dates of archaeological occurrences in the Atai-Sayan region than in the previous period [Zaitzeva et al., 2005]. Possibly, the many Figure 5. Scatterplot of mean and standard deviation of composite fewer kurgans dated around this time indicate population tree ring width chronologies from modern larch growing at upper contraction and lower occupational density. This surprising tree lines (2150–2250 m above sea level (asl)) and the studied archaeological tree rings from the Russian Altai. The outlier is a inverse relationship between climate and Scythian popula- – low-elevation record from the Ulagan Steppe (1400 m asl). This tion (warmer climate less habitation) will be discussed plot provides evidence for a common summer temperature signal in further. the variance of the high-elevation sites. Variance of the low-elevation The cold tendency returns around 340 B.C. and charac- chronology is different (see Figure 4). terizes the last 100 years of Scythian inhabitation of the Altai associated with the Pazyryk Culture. Two decades of prominent warming and low-variance climate (circa 340–310 B.C.) and 540–480 B.C.) of cold summers were certainly the most were interrupted by sudden cooling of summers, lasting for a critical times for herd survival. The highest numbers of miss- couple of decades (circa 310–290 B.C.). It is worth reiterat- ing rings in the archaeological records seemingly suggest a ing that the occupational pattern of the third period is very century of frequent extremely cold springs between 550 and different from the previous periods. Burial sites are found 450 B.C. (Figure 6). Formation of narrow rings and missing both at low steppe elevations (1400–1600 m asl) and subal- rings (partially formed radial growth) in larch correlates pine meadows of high ranges or plateaus (up to 2400 m asl). positively with late springs, cold early summer temperature, Initially, the Bronze Age and Early Scythian burials tend to and delay of snowpack melt [Panyushkina et al., 2003; be found at mid and low elevations [Bourgeois et al., 2007]. Kirdyanov et al., 2003]. Late springs and shorter summers The kurgans at the upper tree line appeared around 400 B.C. not only force the pastoralists to alter seasonal patterns of [Derevyanko and Molodin, 2000]. The expansion of the roaming but can also easily result in loss of large quantities Pazyryk people in the high Altai is a result of their progres- of livestock. Ampliﬁed variance observed in the record for sive success in adaptation to colder climates. this period suggests highly variable weather conditions from The UNESCO project of the Golden Mountains of Altai one year to the next. The coldest summers alternated with the surveyed several thousand Scythian kurgans in southeast warmest summers on a millennium scale. This made weather less predictable. Livestock management under such stressful environmental change required rapid changes in decision- making strategies. Failure in livestock management should induce intensiﬁcation of social dynamics in the population for which archaeological evidence remains to be found. In all likelihood, seasonal migration of the population was significantly disturbed by changes in climate seasonality and cool- Figure 6. Distribution of missing rings (230 missing rings or 1.35% ing from 700 to 480 B.C. Similarity of burial rituals of the of total ring number) in a 445 year tree ring record from high widespread Aldy-Belsk Culture (Tuva in the Sayan Moun- elevation. The locally absent or partially formed rings indicate years tains) from that time provides strong evidence for active of extremely low larch tree ring growth driven by late cold spring zonal migration between the Altai and Sayan Mountains and/or cold summer with abnormally short growth season. PANYUSHKINA 151

kurgans in southeastern Altai belong to the Pazyryk Culture (I. Sljusarenko, personal communication, 2011). Forty kurgans of Pazyryk Culture from 16 cemeteries are dated to 350–275 B.C., which corresponds to the low-variance cold and warm decades (Figure 4). Two completely excavated burial grounds in the Chuya River Basin reveal an age and gender pattern of the Pazyryk population (Table 1). The cemeteries Yustiyd-XII and Ulan- dryk-I were established by Pazyryk commoners at 1800 and 2100 m asl. Tree rings and radiocarbon date the cemeteries to circa 290 to 300 B.C. [Sljusarenko et al., 2004; Panyushkina et al., 2007]. Every 2–3 years, an Ulandryk I extended family Figure 7. Demographic case study of the Ulandryk-I cemetery in constructed a kurgan from 8 to 13 m in diameter and 0.5 m the Russian Altai. (a) Sketch of linear cluster of kurgans at the high (Figure 7b). Thirteen kurgans with 21 buried people Ulandryk-I cemetery. Black circles represent kurgans, open circles were erected over an 11 year period as determined from tree indicate ritual stone rings, and gray circles indicate intrusive Turkic ring dates on wooden burial structures from each kurgan. burials. Smaller symbols indicate smaller kurgans of infants and Mortality among females and children were slighter higher adolescents. The numbers next to each kurgan indicate the order of than in males. The total percent of buried children and kurgan construction (by year) determined from cross-dated tree women is twice that of buried male adults (30% versus rings. Missing numbers indicate woodcutting dates that could not 70% and 24% versus 76%, respectively). Most likely, the be determined. Twenty-one members of the Ulandryk-I family other four cemeteries of Ulandryk Valley were used simulta- (Table 1) were buried during the 11 year period around 300 B.C. neously and overlapped with each other as shown in the tree [Panyushkina et al., 2007]. (b) Burial fields of Scythian commoners along riverbanks at 2150 m asl. Half-circle symbols show positions ring dating of Ulandryk-IV and -I cemeteries. Age and gen- of five Ulandryk cemeteries (U1-U5) and three Tashanta cemeteries der statistics of this small cemetery match those of a larger (T1–T3), each composed of 6–18 kurgans ranging approximately cemetery on the Yustiyd River (Table 1). The demography of 0.3 to 0.5 m in height and 4 to 13 m in diameter. The burial fields the studied cemetery indicates the growth of pastoralist represent a kin grouping of people occupying this area between 320 population. and 280 B.C., who also used the land as family summer pasture. Overall, the burial grounds of Siberian Scythians are more Modified after Kubarev [1987]. spatially dense than burials of any other groups inhabiting Altai from 5000 B.C. to the present day. The high number of Scythian kurgans suggests a large population size. The mod- Altai (Chuya River Basin and Ukok Plateau) [Bourgeois et ern rural population of the Russian Altai [RF-FSSS Statistics, al., 2007]. Scythian kurgans significantly outnumber burial 2011] is 149,409 people with 2.2 per km2 population density monuments of Neolithic Age, Bronze Age, Hun-Sarmatian, (Russian Census 2002). This is 50,000 people (one third) Turkic, and even Altaic ethnographic periods. The total more than a century ago (Russian Census 1923) near the end number of Scythian structures surveyed within Altai’s river of the Little Ice Age and long before modern technological basins varies between 64% and 45% of total registered impact on the Altaic nomadic population. If we assume that archaeological and historical structures. This is strong evi- settlement patterns of the historic Altai population are similar dence for high occupational density of Siberian Scythians in to the Scythian pastoralists, as was demonstrated in studies of the studied area. Burial grounds of Siberian Scythians follow a common landscape pattern: rows of kurgans and stone enclosures associated with them were established on grass- Table 1. Example of Pazyryk Demography From Two Completely lands overlooking rivers (Figure 7A). A typical cemetery Excavated Cemeteries in the Chuya River Basina would have over a dozen kurgans organized in a single row, Adults extended family assembly (Figure 7). Each kurgan has a dual or group burial (three to four human skeletons); single bur- Cemetery/Date Kurgans People Male Female Children ials are a less common feature of the cemeteries [Kubarev, Yustiyd-XII/circa 29 47 14 15 18 1991; Derevyanko and Molodin, 2000]. Besides kurgans of 350 B.C. commoners, this part of the Russian Altai has large kurgans Yustiyd-XII/circa 13 21 5 9 7 of Pazyryk warriors (Ak-Alakha-3) and higher noble elite 350 B.C. (Pazyryk-5). It is believed that most of surveyed Scythian aAfter Kubarev [1987, 1991]. 152 POPULATION DYNAMICS OF SIBERIAN SCYTHIANS

Bronze-Iron Age pastoralists of Kazakhstan and Mongolia and in northwest China reported for period 0 to A.D. 400 [Frachetti, 2008; Houle, 2010], a feasible approximation of [Sheppard et al., 2004; Wanner et al., 2008] would have been the lower bounds of Siberian Scythian population size is less attractive to animal breeders. Humidity excess correlates roughly 100,000 people (comparable to Altai nomadic pop- with high-frequency epizootic livestock diseases and rodent- ulation in the early twentieth century). However, because the born plague outbreaks (including both great gerbil and hu- Scythian burials far outnumber the modern and historical man populations) [Kausrud et al., 2010]. nomadic cemeteries, the upper bounds are more realistic in The trend in population dynamics of Siberian Scythians this case: ~260,000 people and higher (100,000 multiplied suggests increasing mobility and possible intensification of by 2.6, the average number of people buried in Scythian Pazyryk interactions with the outside world during warm kurgans). intervals and a subsequent decrease in occupancy of the In conclusion, the demography of mortality linked to the Altai. The described human-environment interactions and chronology of Siberian Scythians and particularly to the the inverse relationship between climate and Siberian Scyth- Pazyryk culture chronology shows population growth during ian population dynamics (warmer climateless habitation) the first and third cold periods. In contrast, warmer climate requires further investigation to illuminate the complex so- correlates with population contraction in the studied area. cial and political relationships among mobile pastoralists Increased climatic variance suggests instability and large- across the Eurasian Steppe. scale changes of environment [Büntgen et al., 2011]. The amplification of climatic variance in Altai during the first Acknowledgments. This work has been funded through NSF cold period is hypothesized here to have increased mobility BCS 0207654 and BCS 1122359 awards. Thanks to anonymous of the Pazyryk population and possibly resulted in develop- reviewers and the editor of this volume, R. Flad, for the helpful ment of transhumant pastoralism by 400 B.C. Seasonal mo- comments improving the manuscript. bility as an economic strategy increases natural carrying capacity of landscapes and allows further population growth REFERENCES [Honeychurch and Amartuvshin, 2007]. High adaptation to climate combined with high mobility may have motivated Bolortsetseg, B., and G. Tuvaansuren (1996), The potential impacts dispersal of the Pazyryk people to explore and conquer new of climate change on pasture and cattle production in Mongolia, – environments. Water Air Soil Pollut., 92(1-2), 95 105. Overall, climate variability reliably tracks Pazyryk popu- Bourgeois, J., W. Gheyle, R. Goossens, and A. De Wulf (2007), The frozen tombs of the Altai Mountains: Phase I 2005–2006, UGent lation growth between 750 and 520 B.C. and then again from Archeol. Rapp. 2007-4, 84 pp., UGent-Vakgroep Archeol. en 340 to 275 B.C. Enhanced climate variance leads to dispersal Oude Geschiedenis van Europa, Ghent, Belgium. of the population and southward migration across the Altai. Büntgen, U., et al. (2011), 2500 Years of European climate vari- – A brief cold episode at 360 350 B.C. resulted in relocation ability and human susceptibility, Science, 331, 578–582, doi:10. and concentration of the Pazyryk population in the south- 1126/science.1197175. eastern Altai. The last contraction of Pazyryk population Chang, C. (2008), Mobility and sedentism of the Iron Age agropas- occurred in warm decades before 250 B.C., after which toralists of southeast Kazakhstan, in The Archaeology of Mobil- mortuary evidence of Pazyryk population disappeared from ity, edited H. Barnard and W. Wendrich, pp. 329–342, Cotsen the Altai landscape. The decrease in density of kurgans dated Inst. of Archaeol. at Univ. of Calif., Los Angeles. with tree rings may point to the dispersal of the Pazyryk Cook, E. R., and K. Peters (1997), Calculating unbiased tree-ring population from the Altai, which began during warm decades indices for the study of climatic and environmental change, The of the first millennium B.C. (280–240 B.C.). Holocene, 7, 361–370. There may be more than one plausible scenario of Scythian Dashkovskiy, P. K. (1998), Osnovniye problemi izucheniya pose- routes for withdrawal from the Altai. Regional climates leniy skifskoy epokhi Gornogo Altaya, in Poselenia: Sreda, – changed differently after circa 200 B.C. Territory south and Kultura, Sozium, pp. 183 185, St. Petersburg Univ. Press, St. Petersburg, Russia. east of the Russian Altai (Mongolia) between 200 B.C. and Derevyanko, A. P., and V. I. Molodin (Eds.) (2000), Phenomenon of A.D. 200–400 was rapidly warming and drying as documen- the Altai Mummies [in Russian], 318 pp., Inst. of Archaeol. and ted by multiproxy reconstructions of temperature and pre- Ethnography Press, Novosibirsk, Russia. cipitation [Yang et al., 2009]. More northerly steppe Frachetti, M. (2008), Variability and dynamic landscapes of mo- landscapes may have continued to expand northward at the bile pastoralism in ethnography and prehistory, in The Archae- expense of taiga, providing attractive newly available graz- ology of Mobility: Nomads in the Old and in the New World, ing areas of the south Siberia. On the other hand, more humid edited by H. Barnard and W. Wendrich, Cotsen Adv. Seminar and warm summers toward the southwest in the Tian-Shan Ser., 4, 366–396. PANYUSHKINA 153

Friedman, J. H. (1984), A variable span smoother, Tech. Rep. LCS 5, Altai Mummies, edited by A. P. Derevyanko and V. I. Molodin, Lab. for Comput. Stat., Dep. of Stat., Stanford Univ., Stanford, pp. 13–29, Inst. of Archaeol. and Ethnography Press, Novosi- Calif. birsk, Russia. Hajdas, I., G. Bonani, I. Y. Slusarenko, and M. Seifert (2004), Panyushkina, I. P., M. K. Hughes, E. A. Vaganov, and M. A. R. Chronology of Pazyryk 2 and Ulandryk 4 kurgans based on high Munro (2003), Summer temperature in northeastern Siberia since resolution radiocarbon dating and dendrochronology – A step 1642 reconstructed from tracheid dimensions and cell numbers of towards more precise dating of Scythian burials, in Impact of the Larix cajanderi, Can. J. For. Res., 33(10), 1905–1914, doi:10. Environment on Human Migration in Eurasia, edited by E. M. 1139/x03-109. – Scott, A. Alekseev and G. Zaitseva, pp. 107 122, Springer, Panyushkina, I. P., D. V. Ovtchinnikov, and M. F. Adamenko New York. (2005), Mixed response of decadal variability in larch tree-ring Honeychurch, W., and C. Amartuvshin (2007), Hinterlands, urban chronologies from upper tree lines of Russian Altai, Tree Ring centers, and mobile settings: The “New” Old World archaeology Res., 61(1), 33–42. – from the Eurasian Steppe, Asian Perspect., 46(1), 36 64. Panyushkina, I., I. Sljusarenko, N. Bikov, and E. Bogdanov (2007), Houle, J.-L. (2010), Emergent complexity of the Mongolian Steppe: Floating larch tree ring chronologies from archaeological timbers Mobility, territoriality, and the development of early nomadic in the Russian Altai between 800 BC and 800 AD, Radiocarbon, polities, Ph.D. thesis, 240 pp., Univ. of Pittsburgh, Pittsburgh, Pa. 49(2), 693–702. Jacobson, E. (1995), The Art of Scythians: The Interpretation of Parzinger, H. (2006), The Early People of Eurasia, 1045 pp., C. H. Cultures at the Edge of the Hellenic World, 305 pp., Brill Acad., Beck, Munich, Germany. Leiden, The Netherlands. Polosmak, N. V. (2001), Vsadniki Ukoka, 334 pp., Infolio-Press, Jacobson-Tepfer, E. (2008), The emergence of cultures of mo- Novosibirsk, Russia. bility in the Altai Mountains of Mongolia, in The Archaeol- RF-FSSS Statistics (2011), 2004-05-21: Population of Russia, Its ogy of Mobility, edited by H. Barnard and W. Wendrich, pp. Federal Districts, Federal Subjects, Districts, Urban Localities, 200–229, Cotsen Inst. of Archaeol. at Univ. of Calif., Los Rural Localities—Administrative Centers, and Rural Localities Angeles. with Population of Over 3,000 [in Russian], vol. 3, Fed. State Kausrud, K. L., et al. (2010), Modeling the epidemiological history Stat. Serv. of Russ. Fed., Moscow, Russia. [Available at http:// of plague in central Asia: Palaeoclimatic forcing on a disease www.gks.ru/wps/wcm/connect/rosstat/rosstatsite/main/population/ – system over the past millennium, BMC Biol., 8,112 116, doi:10. demography/#]. 1186/1741-7007-8-112. Rudenko, S. I. (1970), Frozen Tombs of Siberia: The Pazyryk Kirdyanov, A., M. Hughes, E. Vaganov, F. Schweingruber, and P. Burials of Iron Age Horsemen, 340 pp., Univ. of Calif. Press, Silkin (2003), The importance of early summer temperature and Berkeley. date of snow melt for tree growth in the Siberian Subarctic, Trees, Sheppard, P. R., P. E. Tarasov, L. J. Graumlich, K. U. Heussner, M. 17(1), 61–69, doi:10.1007/s00468-002-0209-z. Wagner, H. Osterle, and L. G. Thompson (2004), Annual precip- Kleinen, T., P. Tarasov, V. Brovkin, A. Andreev, and M. Stebich itation since 515 BC reconstructed from living and fossil juniper (2011), Comparison of modeled and reconstructed changes in growth of northeastern Qinghai Province, China, Clim. Dyn., 23, forest cover through the past 8000 years: Eurasian perspective, 869–881, doi:10.1007/s00382-004-0473-2. The Holocene, 21(5), 723–734, doi:10.1177/0959683610386980. Sljusarenko, I. Y., Y. V. Kuzmin, J. A. Christen, G. S. Burr, A. J. T. Koryakova, L. N., and A. V. Epimakhov (2007), The Urals and Jull, and L. A. Orlova (2004), 14C wiggle-matching of the Western Siberia in the Bronze and Iron Ages, 408 pp., Cambridge Ulandryk-4 (Early Iron Age, Pazyryk cultural complex) ﬂoating Univ. Press, Cambridge, U. K. tree-ring chronology, Altai Mountains, Siberia, in Radiocarbon Kubarev, V. D. (1987), Kurgani Ulandryk, 218 pp., Nauka, Novo- and Archaeology: Fourth International Symposium, St. Cathe- sibirsk, Russia. rine’s College, Oxford, 9–14 April 2002, edited by T. Higham, Kubarev, V. D. (1991), Kurgani Uystida, 186 pp., Nauka, Novosi- C. B. Ramsey and C. Owen, pp. 177–185, Oxford Univ. Sch. of birsk, Russia. Archaeol., Oxford, U. K. Levine, M., C. Renfrew, and K. Boyle (Eds.) (2003), Prehistoric Tchebakova, N. M., T. A. Blyakharchuk, and E. I. Parfenova Steppe Adaptation and the Horse, 428 pp., McDonald Inst. for (2009), Reconstruction and prediction of climate and vegetation Archaeol. Res., Cambridge, U. K. change in the Holocene in the Altai–Sayan mountains, central Magda, V. N., J. Block, O. C. Oydupaa, and E. A. Vaganov (2011), Asia, Environ. Res. Lett., 4,1–11, doi:10.1088/1748-9326/4/4/ Extraction of moisture signal from tree ring chronologies in 045025. forest–steppe of the Altai-Sayan Region [in Russian], For. Sci., Wagner, M., X. Wu, P. Tarasov, A. Aisha, C. B. Ramsey, M. Schultz, 1,28–37. T. Schmidt-Schultz, and J. Gresky (2011), Radiocarbon- Masanov, N. E. (1995), Kochevaya Tsivilizatsia Kazakhov [Nomadic dated archaeological record of early ﬁrst millennium B.C. Civilization of Kazakhs], 320 pp., SotsInvest, Almaty, Kazakhstan. mounted pastoralists in the Kunlun Mountains, China, Proc. Mikhaylov, N. N., and A. G. Redkin (2000), Ploskogoriye v poz- Natl. Acad. Sci. U. S. A., 108(38), 15,733–15,738, doi:10.1073/ nelednokovoe e poslelednikovoe vremya, in Phenomenon of the pnas.1105273108. 154 POPULATION DYNAMICS OF SIBERIAN SCYTHIANS

Wanner, H., et al. (2008), Mid- to Late Holocene climate change: carbon and Archaeological Chronologies [in Russian], 289 pp., An overview, Quat. Sci. Rev., 27(19-20), 1791–1828, doi:10. Thesa, St. Petersburg, Russia. 1016/j.quascirev.2008.06.013. Yang, B., J. Wang, A. Bräuning, Z. Dong, and J. Esper (2009), Late Holocene climatic and environmental changes in arid central Asia, Quat. Int., 194(1-2), 68–78, doi:10.1016/j.quaint.2007.11.020. I. P. Panyushkina, Laboratory of Tree-Ring Research, University Zaitseva, G. I., N. A. Bokovenko, A. Y. Alekseev, K. V. Chugunov, of Arizona, 105 W. Stadium, Tucson, AZ 85720, USA. and E. M. Scott (Eds.) (2005), Eurasia in Scythian Time: Radio- ([email protected]) Geoarchaeological Perspectives on Holocene Climate Change as a Civilizing Factor in the Egyptian Sahara

Kathleen Nicoll

Department of Geography, University of Utah, Salt Lake City, Utah, USA

Throughout prehistory, rapid hydroclimatic changes (“wet-dry” cycles) played a key role in landscape habitability and resource availability in the Egyptian Sahara. As water sources waned during episodes of Holocene drought, people developed various subsistence strategies, including opportunistic hunting of small animals, and food production by gathering, transhumance, and livestock rearing. The geoarchaeological record of the Neolithic culture at Nabta Playa ~100 km west of the Nile Valley suggests that Holocene droughts were a civilizing factor; migrations toward water-fostered acculturation and social complexity, which is evident in burials, ceremonial centers, and solar calendars built at Nabta. The ultimate resilience strategy of the desert dwellers was relocation to the Nile Valley and the transition to agriculture after 5300 calendar years B.P.

1. INTRODUCTION crisis” is the failure of a society to adapt, there has been less emphasis on looking at the prehistoric people them- Key milestones in human history have been linked with selves, their material records, and their cultural responses to various multiproxy records of Holocene climate change, severe and/or abrupt climate changes [Clare and Weninger, including the collapse of agricultural communities and 2010]. ancient empires in the Near and Middle East [e.g., Tainter, This chapter briefly evaluates the geoarchaeological re- 1990; Cullen et al., 2000; Staubwasser and Weiss, 2006; cords of prehistoric people in the Egyptian Sahara, which Weninger et al., 2009]. A growing emphasis of inquiry has been affected by rapid, decadal-to-centennial-scale hy- addresses the prehistoric human response due to large-scale droclimatic changes throughout the Holocene. Recurrent or abrupt variations in global and regional hydroclimate droughts promoted cultural adaptation, which was an im- [e.g., McIntosh et al., 2000; Burroughs, 2005; Diamond, portant civilizing factor in the sense of Rosen [2006]. The 2005; Rosen, 2006]. Most hypotheses that link climate Neolithic of Egypt witnessed many critical junctures in change, evolution, and human behavior relate how large- prehistory, including the development of pottery, initial do- scale shifts or sudden variations in climate altered the mestications, the advent of agriculture, a shift from nomadic landscape ecology, which, in turn, presented specific adap- lifeways toward sedentism, the first cities, and the rise of tive or speciation pressures leading to genetic selection, empires. Neolithic remains at Nabta Playa provide a valu- adaptation, and innovation. Although an “environmental able case study for interpreting the social-adaptive behaviors of foragers and nomadic pastoralists as its cultures faced environmental degradation and the onset of the modern hyperarid Sahara. Moreover, the Neolithic sites near Nabta Climates, Landscapes, and Civilizations Playa are well dated and are among the best-studied prehis- Geophysical Monograph Series 198 toric archives in North Africa [e.g., Wendorf and Schild, © 2012. American Geophysical Union. All Rights Reserved. 1980; Wendorf and Schild, 2001, and references therein] 10.1029/2012GM001219 (see Figure 1). 157 158 CIVILIZING CLIMATE CHANGE IN EGYPTIAN SAHARA

gregated was related to water resource availability over time as environments ﬂuctuated between a habitable savanna and an inhospitable desert [Nicoll, 2008]. The region west of the Nile Valley was occupied after 12,000 calendar (cal) years B.P. when occasional rainstorms sustained a sparse vegetation of savanna grasses, trees, and bushes, enough to feed hares, gazelle, and a few small carnivores [Haynes, 2001]. Most of the terminal Paleolithic-Neolithic assemblages are located near ephemeral waterways and rainwater-fed “pluvial” features, including wadis and shallow ponds or playas.

3. HOLOCENE NEOLITHIC PLUVIALS AT NABTA PLAYA AND THE EMERGENCE OF SOCIAL COMPLEXITY

Nabta Playa is a large site located in southern Egypt ~100 km west of the Nile Valley [Wendorf and Schild, Figure 1. Location of Nabta Playa. 2001]. More than 115 published radiocarbon dates constrain the cultural occupation sequence in the area, and converging lines of evidence suggest that this area experienced three 2. INSIGHTS FROM GEOARCHAEOLOGICAL main wet-dry oscillations during the Holocene [Nicoll, RECORDS IN THE NE SAHARA 2004]. Occupation was during wet phases, as constrained by radiocarbon dates on organic material in stratigraphic con- “ ” 2.1. Regional Hydroclimatic Setting text. The timing of the droughts is bracketed by windows or lacunae, during which aridity is manifest by cultural Today, the Egyptian Sahara is highly continental, virtually abandonment and/or depositional hiatuses, aeolian activity, rainless, and extremely inhospitable, with few potable sur- and erosion [Nicoll, 2001]. face water sources. However, an abundance of recent fossil The Nabta environs preserve dwelling structures, storage remains, artifacts, water-lain sediments, and landforms attest pits, wells, and various material culture remains preserved in that conditions were markedly different in the past. Mon- stratified contexts, including assemblages of characteristic soons influenced this portion of the hyperarid African conti- tools and microliths, grinding tablets, pottery, and ostrich nental interior, bringing convective rainfall that briefly eggshell beads. The local archaeological record reflects crit- enabled surface runoff and water storage as a function of ical adaptations and transitions from foraging to food pro- astronomical precession forcing during the early Holocene. duction strategies, domestication, and animal husbandry, A review of radiocarbon dates and multiproxy records which suggest emerging traditions, structure, and social indicates a broadly synchronous pattern of recent environ- complexity [Close, 1984; Wendorf and Schild, 1998; Wen- mental change across the region of Egypt and Sudan [Nicoll, dorf and Schild, 2001]. 2001, and references therein]. Aridity was pervasive during Initial settlements at Nabta were established ~11,000– the late Pleistocene; wet conditions existed after 10,000 B.P.; 9300 cal years B.P. The inhabitants of Nabta seem to have water persisted at the landscape surface through droughts, herded cattle, but it is not clear whether these were domes- but diminished through ≈7000 B.P.; and a period of desicca- ticated. The people carried distinctive ceramic vessels deco- tion ensued, with unabated hyperaridity since ~4500 B.P. rated with impressions and wavy patterns, and this pottery is [Nicoll, 2004, and references therein] (see Figure 2). Further among the oldest known in the region. In a similar manner discussion of the specific climate drivers for, and effects of practiced by modern pastoralist peoples in Africa, Nabtans rapid variation and stasis during, the Holocene [i.e., Mayewski may have regarded their cattle as economic units of power, et al., 2004] is beyond the scope of this chapter. social status, and prestige; while proper forage is available, cattle may have provided a regular supply of milk and 2.2. Water-Centric Prehistoric Site Patterns blood, which is a more sustainable practice than slaughtering animals for meat [Close, 1987; Close and Wendorf, 1992]. Spatial patterns of prehistoric occupation in the northeast- However, such modern-day ethnographic comparisons may ern Sahara suggest that the primary loci where people con- not be relevant for reconstructing the Neolithic. Evidence Figure 2. Reconstructions of late Quaternary air temperatures from isotopic analyses of the Greenland Ice Sheet Project 2 (GISP2) ice core, the 8.2 event 159 NICOLL [Rohling and Pälike, 2005], and the Younger Dryas [Alley et al., 1997]. Major variants (italicized) and cultural transitions are described in the text. Modified after Nicoll [2004]. 160 CIVILIZING CLIMATE CHANGE IN EGYPTIAN SAHARA suggests that the intervening phase of playa desiccation at and Malville, 2001]. Within the structures, bone remains of Nabta after 9300 cal years B.P. forced people to migrate to butchered cattle, goats, and sheep attest to animal offerings. areas with more water, causing consequent population in- One tumulus site dating to 7400 years ago contained sacrifi- creases along the Nile Valley in the east and in areas further cial remains of a calf buried in an elaborate chamber covered south. with branches. After 9000 years ago, larger settlements were established Some interpretations suggest that the arrangements and at Nabta; small huts were arranged in straight lines, and groupings of megalith structures served as astronomical cal- walk-in water wells were dug. Sustenance included a number endars [Applegate and Zedeño, 2001]. Their locations are of edible plants, including variants of millets, sorghum, along sightlines marking the position of important stars and legumes, tubers, and fruits, and small animals such as hares the solstice along the Tropic of Cancer during the time period and gazelles. Around 8800 cal years ago (~7800 B.P., uncal- ~6800–6000 B.P. [Malville et al., 2007, 2008]. The stelae of ibrated), pottery was produced locally. There is some evi- the longest alignment at Nabta are oriented to point toward dence around 8100 years ago that suggests the domestication the brightest star in the Big Dipper [Malville et al., 2008]. of larger animals, goats, and sheep and development of This star (Dubhe) as well as the entire constellation Ursa animal husbandry [Close, 1990; Wendorf and Schild, 2001]. Majoris was quite important in ancient Egypt and figured in Between 8000 and 7000 years ago, Nabta Playa became various symbols and myths combining cattle, funeral rituals, dry and was abandoned during two major drought periods kingship, and celestial order. Another alignment near Nabta bracketed by radiocarbon dates. As hyperarid conditions Playa points to the two brightest stars in the “belt” of Orion, developed, the water table dropped, the savanna grassland the constellation associated with Osiris, the Egyptian god of diminished, and wind deflation intensified, creating impos- the afterlife, the underworld, and the dead. A third alignment sible living conditions. The people returning to Nabta after indicated the position where Sirius, the brightest star in the these droughts appear to be a more complex and organized night sky, returned after a 70 day absence and appeared society; archaeological remains suggest the practice of a prominently in the dawn sky just before the summer solstice. ritualistic belief system associated with livestock. Excavated The reappearance of this star heralded the onset of the annual items supporting this inference include sacrifices of young flooding of the Nile and had deep symbolic meaning to cows and their burial in clay-lined and roofed chambers ancient Egyptians, during the First Dynasty, who ritually covered by stone slabs. The inhabitants of Nabta also con- celebrated the event as the start of their calendar year [David, structed ~25 complex building structures with surface and 2002]. subterranean features, including storage pit chambers and a The ceremonial center at the settlement of Nabta reflected shaped stone that might possibly represent the oldest sculp- the cosmological underpinnings of belief systems central to ture in this part of Egypt [Wendorf and Schild, 1998, 2001]. ancient Egyptian magic and religion, and the material record A significant find in the Nabta region is a late Neolithic of the Nabtan culture reflects a regional connectedness of the cemetery [Kobusiewicz et al., 2004]. More than 30 graves Saharan people. It is clear that the Nabtans had affiliations were discovered, some of which were multiple and symbolic with the incipient Predynastic Egyptian civilization emerg- burials, and containing the remains of 69 individuals in all. ing in the Nile Valley [Hoffman, 1991; Rice, 2003]. Were the There were also two secondary burials in which the skeleton desert people at Nabta the original celestial observers, the eye sockets were filled with teeth [Irish et al., 2003]. A series progenitors of profound ideologies that shaped the later of richly furnished graves date to 6400 and 6000 uncalibrated civilization in the dynasties of Egypt? The available chro- radiocarbon years B.P. Associated grave goods (e.g., carne- nologies for the Neolithic settlement at Nabta substantiate lian beads, shell bracelets, polished stones, pottery, ochre, the deep antiquity of a connection with the skies and deriv- needles, and stone tools) demonstrate that these local pasto- ative symbolic ideologies. Recurrent themes that seem to ralists likely practiced transhumance and had trade connec- have been important to these nomadic pastoralists include, tions across the region [Kobusiewicz et al., 2009]. water, death, stars, cattle, and the sun, all of which persisted The Nabtans erected monuments and megaliths made of as important motifs in Dynastic Egypt. worked stones [Applegate et al., 2001], and these are among The people of Nabta may have been the last dwellers of the oldest shrines and ceremonial centers in Africa [Wendorf this marginal environment. As intense drought conditions and Królik, 2001]. Many of these archaeological structures persisted, water sources dried up, and the grassland disap- involve large stones (stelae or tumuli) placed in various peared ~6000 years B.P.; the area of Nabta was inhospitable arrangements including mound complexes, carvings in bed- after 5300 years ago, which correlates to 3350 B.C.E. (before rock buried several meters below the desert surface, and in the Common Era). The “terminal” date for final occupation at both circular (i.e., cromlech) and linear orientations [Wendorf Nabta is around 4780 B.P., as hyperaridity prevailed, and the NICOLL 161

Sahara was fully established. This profound environmental Assessing “what people did in the past” is ambiguous by change precipitated migration, an “Exodus event” in which nature because actions and intentions are often intangible, humans left the desert locales for reliable water sources, as obscure parts of the prehistoric record. Even today, human evinced by the rising population along the Nile [Midant- nature and its collective experience are difficult to define, and Reynes, 1992; Malville et al., 1998]. As the Nabtan people it is rarely tidy: responses and actions may be alternately and relocated, they inevitably contributed their own culture and simultaneously nuanced, vague, relative, flexible, and non- beliefs to the birth of ancient Egyptian religion and the linear. Although the interpretations expressed in this chapter Pharonic civilization, which organized its empire around might be inherently lacking in concreteness and certainty, the irrigation agriculture within the overpopulated confines of the behaviors of the Saharan Neolithic desert dwellers as they Nile Valley [Krzyżaniak, 1991; Nicoll, 2004]. transitioned toward a well-watered settlement were fundamental in shaping history and the characteristics of people 4. CONCLUSIONS living today. Perhaps one of the core adaptations of human beings since prehistory is the ability to respond to, or miti- The archaeological record for the Egyptian Sahara demon- gate, certain challenges associated with rapid climate strates prehistoric human interaction with their surroundings; variations. the Nabta culture developed resilient strategies, technologies, ritual practices, long-distance exchange networks of Acknowledgments. My research in the Egyptian Sahara has been resources and goods, and methods to harvest and store foods financially supported by grants from NASA, the Smithsonian Insti- and water [Nicoll, 2004]. As the population at Nabta coped tution, the Royal Society (U.K.), and Oxford University. I thank with profound droughts during the Holocene, they experienced Philip Van Peer, Dorian Fuller, and other reviewers for their helpful significant transitions in lifeways from foraging-gathering ± feedback. hunting to herding pastoralism, and they developed complex social interactions that favored survival. Perhaps the ultimate REFERENCES resilience strategies of the Nabtans were interaction with others afield and their eventual migration to favorable locations to Alley, R. B., P. A. Mayewski, T. Sowers, M. Stuiver, K. C. Taylor, practice irrigation agriculture. The latter proved to be a pro- and P. U. Clark (1997), Holocene climatic instability: A prominent, widespread event 8200 years ago, Geology, 25, 483–486. foundly successful strategy, enabling nomadic desert dwellers Applegate, A., and N. Zedeño (2001), Site E-92-9: A possible Late to transition location so that they could flourish and “civilize” Neolithic solar calendar, in Holocene Settlement of the Egyptian the Nile Valley. Sahara, vol. 1, The Archaeology of Nabta Playa, edited by F. The geoarchaeological records at Nabta enable us to link Wendorf et al., pp. 463–467, Kluwer Acad., New York. fi droughts, deserti cation, and trends of societal transforma- Applegate, A., A. Gautier, and S. Duncan (2001), The north tumuli tion in deep prehistory. Collectively, the people of Nabta of the Nabta Late Neolithic ceremonial complex, in Holocene spent several generations coping with drought. While tradi- Settlement of the Egyptian Sahara, vol. 1, The Archaeology of tional archaeological theory views the evolution of social Nabta Playa, edited by F. Wendorf et al., pp. 468–488, Kluwer complexity as an internally generated and linear process that Acad., New York. has progressively unfolded through time, the record of the Burroughs, W. J. (2005), Climate Change in Prehistory: The End of prehistoric culture at Nabta demonstrates the iterative resil- the Reign of Chaos, 372 pp., Cambridge Univ. Press, Cambridge, ience of an adapted desert people for whom migration was a U. K. strategic opportunity that fostered acculturation with people Clare, L., and B. Weninger (2010), Social and biophysical vulner- in the Nile Valley. With this social context in mind, it is worth ability of prehistoric societies to rapid climate change, Doc. – reconsidering notions of climate determinism and the Holo- Praehist., XXXVII, 269 284. cene evolution of the eastern Sahara (in the sense of Kuper Close, A. E. (Ed.) (1984), Cattle-Keepers of the Eastern Sahara: The Neolithic of Bîr Kiseiba, 438 pp., South. Method. Univ. and Kröpelin [2006]). Press, Dallas, Tex. Integrative geoarchaeological studies at Nabta demon- Close, A. E. (Ed.) (1987), Prehistory of Arid North Africa, 357 pp., strate the dynamic interplay of hydroclimatic and environ- South. Method. Univ. Press, Dallas, Tex. mental change and social-adaptive responses, including Close, A. E. (1990), Living on the edge: Neolithic herders in the population migration and technological transitions. As in- eastern Sahara, Antiquity, 64,79–96. triguing as these apparent temporal coincidences might be, Close, A. E., and F. Wendorf (1992), The beginnings of food- however, demonstrating direct causal linkage between cli- production in the Eastern Sahara, in Transitions to Agriculture mate changes and cultural responses poses significant chal- in Prehistory, edited by A. B. Gebauer and T. D. Price, pp. 63–72, lenges to the various disciplines that reconstruct antiquity. Prehistory Press, Madison, Wis. 162 CIVILIZING CLIMATE CHANGE IN EGYPTIAN SAHARA

Cullen, H. M., P. B. deMenocal, S. Hemming, G. Hemming, F. H. Nicoll, K. (2001), Radiocarbon chronologies for prehistoric human Brown, T. Guilderson, and F. Sirocko (2000), Climate change occupation and hydroclimatic change, Egypt and Sudan, and the collapse of the Akkadian empire: Evidence from the deep Geoarchaeology, 16,47–64. sea, Geology, 28, 379–382. Nicoll, K. (2004), Records of recent environmental change and David, R. (2002), Religion and Magic in Ancient Egypt, 480 pp., prehistoric human activity in Egypt and northern Sudan, Quat. Penguin, London, U. K. Sci. Rev., 23, 561–580. Diamond, J. (2005), Collapse: How Societies Choose to Fail or Nicoll, K. (2008), Geographic Overviews: Africa—Central and Succeed, 575 pp., Penguin, New York. Nilotic Sudan, in Encyclopedia of Archaeology,editedbyD. Haynes, C. V., Jr. (2001), Geochronology and climate change of the Pearsall, pp. 5–9, Academic, New York. Pleistocene-Holocene transition in the Darb el’Arba’in Desert, Rice, M. (2003), Egypt’s Making. The Origins of Ancient Egypt Eastern Sahara, Geoarchaeology, 16,119–141. 5000–2000 BC, 360 pp., Routledge, London, U. K. Hoffman, M. A. (1991), Egypt Before the Pharaohs—The Prehis- Rohling, E. J., and H. Pälike (2005), Centennial-scale climate cool- toric Foundations of Egyptian Civilization, 409 pp., Univ. of Tex. ing with a sudden cold event around 8,200 years ago, Nature, Press, Austin. 434, 975–979. Irish, J. D., M. Kobusiewicz, R. Schild, and F. Wendorf (2003), Rosen, A. (2006), Civilizing Climate. Human Responses to Envi- Neolithic tooth replacement in two secondary burials from south- ronmental Change in the Ancient Near East, 209 pp., Altamira ern Egypt, J. Archaeol. Sci., 30, 281–285. Press, Lanham, Md. Kobusiewicz, M., J. Kabaciński, R. Schild, J. D. Irish, and F. Staubwasser, M., and H. Weiss (2006), Holocene climate and cul- Wendorf (2004), Discovery of the ﬁrst Neolithic cemetery in tural evolution in late prehistoric–early historic West Asia, Quat. Egypt’s Western Desert, Antiquity, 78, 566–578. Res., 66, 372–387. Kobusiewicz, M., J. Kabaciński, R. Schild, J. D. Irish, and F. Tainter, J. A. (1990), The Collapse of Complex Societies, 290 pp., Wendorf (2009), Burial practices of the Final Neolithic pastoral- Cambridge Univ. Press, Cambridge, U. K. ists at Gebel Ramlah, Western Desert of Egypt, Br. Mus. Stud. Wendorf, F., and H. Królik (2001), Site E-96-1: The complex Ancient Egypt Sudan, 13, 147–174. structures or shrines, in Holocene Settlement of the Egyptian Krzyżaniak, L. (1991), Early farming in the Middle Nile Basin: Sahara,vol.1,The Archaeology of Nabta Playa, edited by Recent discoveries at Kadero, Antiquity, 65, 515–532. F. Wendorf et al., pp. 503–520, Kluwer Acad., New York. Kuper, R., and S. Kröpelin (2006), Climate-controlled Holocene Wendorf, F., and J. M. Malville (2001), The megalith alignments, in occupation in the Sahara: Motor of Africa’s evolution, Science, Holocene Settlement of the Egyptian Sahara, vol. 1, The Archae- 313, 803–807. ology of Nabta Playa, edited by F. Wendorf et al., pp. 489–502, Malville, J. M., F. Wendorf, A. A. Mazar, and R. Schild (1998), Kluwer Acad., New York. Megaliths and Neolithic astronomy in Southern Egypt, Nature, Wendorf, F., and R. Schild (1980), Prehistory of the Eastern Sa- 392, 488–491. hara, 414 pp., Academic, New York. Malville, J. M., R. Schild, F. Wendorf, and R. Brenmer (2007), Wendorf, F., and R. Schild (1998), Nabta Playa and its role in Astronomy of Nabta Playa, Afr. Sky, 11,2–7. northeastern African prehistory, J. Anthropol. Archaeol., 17, Malville, J. M., R. Schild, F. Wendorf, and R. Brenmer (2008), 97–123. Astronomy of Nabta Playa, in African Cultural Astronomy— Wendorf, F., and R. Schild (2001), Holocene Settlement of the Current Archaeoastronomy and Ethnoastronomy Research in Africa, Egyptian Sahara, vol. 1, The Archaeology of Nabta Playa, edited by J. C. Holbrook, J. O. Urama and R. T. Medupe, Astrophys. 707 pp., Kluwer Acad., New York. Space Sci. Proc.,131–143, doi:10.1007/978-1-4020-6639-9_11. Weninger, B., et al. (2009), The impact of rapid climate change on Mayewski, P. A., et al. (2004), Holocene climate variability, Quat. prehistoric societies during the Holocene in the Eastern Mediter- Res., 62, 243–255. ranean, Doc. Praehist., XXXVI,7–59. McIntosh, R. J., J. A. Tainter, and S. K. McIntosh (Eds.) (2000), The Way the Wind Blows—Climate, History and Human Action, 413 pp., Columbia Univ. Press, New York. Midant-Reynes, B. (1992), The Prehistory of Egypt, 344 pp., Black- K. Nicoll, Department of Geography, University of Utah, Salt well, Oxford, U. K. Lake City, UT 84112, USA. ([email protected]) Heavy Rainfalls in a Desert(ed) City: A Climate-Archaeological Case Study From Sudan

Jonas Berking,1 Janina Körper,2 Sebastian Wagner,3 Ulrich Cubasch,2 and Brigitta Schütt1

The rise and fall of ancient cultures in drylands is mainly controlled by the availability of water. Where no perennial water sources are available, (ancient) cultures in drylands often depend on the availability of water by effective rainfall as a source for water-harvesting measures. These settlements are susceptible to climatic changes. The reconstruction of such climate-human interactions is constrained, as paleoclimatic proxy archives provide climatic information at most with annual or seasonal resolutions, which is often not a sufﬁcient resolution to relate them to the reactions or adaptations of societies to climatic shifts. For the city of Naga (16°N 33°E) in the dry hinterland of the middle Nile River, we present precipitation data for the last 6500 years employing two different downscaling approaches. In time slice experiments, we simulated selected episodes using a spatially high-resolved atmosphere general circulation model. Furthermore, combining observed precipitation data and a simulation of a coupled ocean-atmosphere general circulation model, we attained statistically downscaled precipitation data for the last 6500 years. Our results indicate that the mean precipitation and the frequency of runoff-generating rainfall events decreased from 6000 until 2000 years B.P. Since then, no signiﬁcant changes occurred. Thus, the foundation of Naga about 2500 years B.P. coincided with a time when rainfall was still more reliable and less variable. Alterations in orbital parameters (according to Milanko- vic theory) may have been a major reason for the abandonment of the city.

1. INTRODUCTION both present-day and palaeoclimatic conditions [Wilby and Wigley, 1997; Intergovernmental Panel on Climate Change, Global circulation models (GCMs) are, owing to their low 2007]. In this study, we use a combination of simulated and spatial resolution, not suited to provide local-scale informa- observed atmospheric conditions to obtain an estimate of the tion. Dynamical and statistical downscaling approaches have probability of rainfall events. Our main hypothesis is that been developed to derive data more applicable in terms of the Meroitic city of Naga was founded in the hinterland of the trend analyses or stochastic approaches on a regional scale for middle Nile Valley during a time when the interannual rainfall availability was more stable and reliable than at present. We 1 Institute of Physical Geography, Freie Universität Berlin, also address the question, discussed by a number of authors, Berlin, Germany. ﬁ 2 whether aridi cation in the sub-Saharan regions was abrupt or Institute for Meteorology, Freie Universität Berlin, Berlin, continuous [Claussen and Gayler, 1997; Liu et al., 2007]. Germany. The settlement history of the Nile Valley is linked with the 3Helmholtz Zentrum Geesthacht, Geesthacht, Germany. immigration from central North Africa coinciding with the Climates, Landscapes, and Civilizations climatic deterioration after the Holocene Optimum [Hoelz- Geophysical Monograph Series 198 mann et al., 2001]. © 2012. American Geophysical Union. All Rights Reserved. In the ﬁrst millennium before the Common Era (B.C.E.), 10.1029/2012GM001208 prominent civilizations populated the middle and upper Nile 163 164 CLIMATE-ARCHAEOLOGY

Valley. Great cities and monuments were built not only in The ancient city of Naga, located at 16°N and 33°E Egypt but also in Sudan, along the middle reaches of the Nile. (Figure 1), dates back to the fourth century B.C.E. and Along the Egyptian Nile Valley, water availability relied on existed for about 800 years until the fourth century Com- upstream sources for the river and floods. By contrast, in mon Era (C.E.). As shown in a previous study, the city had Sudan, the monsoonal rainy season is a dominant source of to rely on short and heavy rainfall events for flood water water. The transitional zone corresponds to the dry savanna harvesting and irrigation purposes [Berking et al., 2010; spreading from 17°N to15°N between the semiarid Sahel and Berking and Schütt, 2011]. The collected surface water was the hyperarid Sahara. Here, owing to the lack of reliable stored in artificial basins, known as hafirs (Arabic: dig). The rainfall, permanent settlements or rainfed agriculture are pres- local hand-dug “Great Hafir” of Naga had a reservoir vol- entlyalmostabsentinthehinterlandoftheNile[Nicoll, 2004]. ume of 37,000 m3 (Figures 1b to 1d).

Figure 1. The study site (red star) and its location in northeast Africa. (a) The northern of the modern (red dotted line) and Holocene Maximum (blue dotted line) of the northern core of the Intertropical Convergence Zone, derived from the time slice experiments, calculated analogously by Braconnot et al. [2007]. Gray arrows symbolize major wind directions (scaling of the arrows according to mean wind speed in June, July, August, and September) derived from ERA-40 reanalysis [Uppala et al., 2005]. (b) The “Great Hafir” of Naga from an eastern perspective. (c) Four times spatially exaggerated 3-D image of the “Great Hafir.” (d) Contour level map of the “Great Hafir.” BERKING ET AL. 165

At present, the expected rain amount per year is around local precipitation. In a ﬁrst step, a downscaling model is 100 mm and particularly variable owing to the interannual calibrated and validated in the observational period linking variances of the Intertropical Convergence Zone (ITCZ), changes in sea level pressure with precipitation changes at occasionally reaching the sub-Saharan regions (Figure 1). Naga. The statistical method employed to link the large scale Among the processes known to inﬂuence the northward with the local scale is the principal component regression extension of the ITCZ are the El Niño–Southern Oscillation, described by Luterbacher et al. [2002] and Wagner et al. explaining variances on interannual to decadal time scales, [2007]. Due to the semiarid climate in Naga, the downscaling and orbital forcings, explaining variances on millennial time models only show moderate performance for the Naga region. scales [Braconnot et al., 2007]. The physics behind the statistical model do, however, resemble physically meaningful processes and therefore are used for 2. METHODS AND DATA SETS further analysis. The downscaling model is eventually applied to the large-scale information derived from a multimillennial 2.1. Statistical Downscaling Holocene simulation with Echo-G [Wagner et al., 2007].

GCMs are too coarsely resolved to be directly used for local 2.2. Time Slice Experiments hydrological variables such as precipitation. Therefore, we employed statistical downscaling of precipitation for the study To increase the horizontal resolution, the atmospheric site. Changes in the large-scale atmospheric circulation that model ECHAM5 is run at a resolution corresponding to are better represented in GCMs are linked with the observed ~1° Â 1° grid cells, for the selected time slices. For the time

Figure 2. Model data comparison. (a) Annual cycle of precipitation derived from Beck et al. [2004], as fraction of annual precipitation (%). Blue bars represent the annual monsoonal rainy season; beige bars represent the annual dry season. (b) Running 51-year mean of statistically downscaled July precipitation of the last 6500 years, displayed as the anomaly relative to 0–100 Common Era mean (mm dÀ1). Vertical gray lines are drawn for the periods of the time slice experiments (25 years each). The red area displays the dating of Naga. The frequency of rainfall events exceeding 10 mm dÀ1 is given at the top. (c) Continental Humidity Index of core GeoB7920-2 [Tjallingii et al., 2008]. 166 CLIMATE-ARCHAEOLOGY slice technique [see, e.g., Cubasch et al., 1995], sea surface 2. The simulated recurrence interval (RI) of runoff-generating temperatures and sea ice fractions from a coupled simulation rainfall events exceeding 10 mm dÀ1 changed significantly are prescribed as lower-boundary forcing. With this setup, over the investigated time slices (Figure 3). The present-day six time slice simulations are carried out. The maximum of RI of runoff-generating rainfall totals less than once a year accumulated daily precipitation from the nearest four grid (frequency: 0.65 yearÀ1). Two thousand years ago, the rain- boxes colocated with Naga are employed for the reconstruc- fall character was quite similar (frequency: 0.7 yearÀ1). tion of the rainfall events. Three thousand years ago, such a rainfall event was expected to occur more than once per year (frequency: 1.5 yearÀ1). 2.3. Hydrological Model During the mid-Holocene, around 6000 years ago, events with magnitudes of ≥10 mm dÀ1 are simulated to occur To validate and rate our results, we incorporated the results nearly three times a year (frequency: 2.8 yearÀ1). from a hydrological model. This model approximates the The graphs for the different time slice experiments show rainfall that is needed either to induce any runoff or to similar slopes but differ in their absolute precipitation produce runoff exceeding the threshold for the Hafir of Naga amounts (Figure 3). to be filled [Berking et al., 2010]. The respective subcatch- The high-resolution downscaled July precipitation shows ment of the hafir has a size of 14.2 km2 and is a tributary to similar behavior as the Continental Humidity Index for NW the wadi Awatib with a size of 2360 km2. For the water Africa [Tjallingii et al., 2008] (Figure 2c), indicating trends harvesting strategies, the so-called Hortonian or infiltration in Naga that reflect the supraregional trends (similar to Krö- excess overland flow is initiated when rainfall intensities pelin et al. [2008] and Mawson and Williams [1984]). Since exceed the infiltration capacity of the soils. The rainfall we did not include any vegetation feedback in our model, character in drylands is, in turn, dominated by erratic, high- comparisons of the precipitation regimes are problematic. intensity short-duration rainfall events with small spatial Hence, one should not overinterpret our results from the extension [Kidron and Yair, 2001; Schütt et al., 2010]. southern desert margin because comprehensive studies for Therefore, and owing to computer memory limitations, the the central Sahara and their surroundings with continuous daily rainfall sums refer to hourly rainfall events [Berking et al., 2010]. In arid and semiarid landscapes, intense rainfall events, exceeding the soil infiltration capacities, lead to surface runoff. For the generation of such runoff, we assume a threshold of ≥10 mm hÀ1 as frequently cited in literature [Kidron and Yair, 2001; Esteves and Lapetite, 2003].

3. RESULTS

To estimate the precipitation conditions since the Holo- cene Optimum, we use a continuous time series of statistically downscaled summer precipitation of the last 6500 years (Figure 2c) and time slice experiments at different periods during the Holocene (Figures 2c and 3). We evaluated the precipitation data in two ways: (1) long-term trends, annual distribution, and intensities and (2) the recurrence interval of runoff-generating rainfall events: 1. The general precipitation regime is characterized by a rainy season from June to September, peaking in August with more than a third of the annual precipitation falling in Figure 3. Magnitude-frequency distributions for the recurrence of rainfall events in Naga. Diamonds represent modeled data. Solid this month (Figure 2a). The statistically downscaled long- lines show the logarithmic fit of the data (colors indicate the differ- term analysis of the central rainy month of July shows a À1 À1 ent time slices between 6 ka B.P. and present day). Horizontal general decline of 1.3 mm d (40 mm month ,respectively) dotted line indicates the threshold of 10 mm hÀ1 for a rainfall event from 6000 years before present to 2000 years before present. that generates runoff. In the light green area (rainfall events), runoff For the last 2000 years, significant changes in the mean July is generated at least once per year (from 6 to 3 ka B.P.). In the light precipitation did not occur (Figure 2c). rose area, runoff occurs less than once a year. BERKING ET AL. 167 and high-resolution records are challenging and largely miss- Claussen, M., and V. Gayler (1997), The greening of Sahara during ing [Hoelzmann et al., 2001; Renssen et al., 2003; Schütt the mid-Holocene: Results of an interactive atmosphere-biome et al., 2010]. model, Global Ecol. Biogeogr. Lett., 6, 369–377. Cubasch, U., J. Waszkewitz, G. Hegerl, and J. Perlwitz (1995), 4. CONCLUSIONS Regional climate changes as simulated in time-slice experiments, Clim. Change, 31, 273–304. The regional precipitation character has changed particu- Esteves, M., and J. M. Lapetite (2003), A multi-scale approach of larly since the time when the ancient and now-abandoned runoff generation in a Sahelian gully catchment: A case study in – settlement of Naga was founded, and the Meroitic engineers Niger, Catena, 50, 255 271. coped with the rare annual runoff events by building the Hoelzmann, P., B. Keding, H. Berke, and S. Kröpelin (2001), Environmental change and archaeology: Lake evolution and “Great Hafir” of Naga, a convenient flood water-harvesting human occupation in the Eastern Sahara during the Holocene, measure. We suggest that the abandonment of the city during Palaeogeogr. Palaeoclimatol. Palaeoecol., 169, 193–217. fi the rst centuries C.E. is associated with the decreased avail- Intergovernmental Panel on Climate Change (2007), Climate ability of surface runoff as a response to orbital forcing, at Change 2007: The Physical Science Basis. Contribution of Work- least as one of the causes. The regular summer rainy season ing Group I to the Fourth Assessment Report of the Intergovern- by itself does not consecutively imply sufficient rainfall for mental Panel on Climate Change, edited by S. Solomon et al., (agri-)cultural needs because the sum of rainfall is not crucial, Cambridge Univ. Press, Cambridge, U. K. but its intensity. Effective precipitation and runoff are avail- Kidron, G. J., and A. Yair (2001), Runoff-induced sediment yield able in such arid environments only when a low-pressure over dune slopes in the Negev desert. 1: Quantity and variability, system produces a convective storm with short and heavy Earth Surf. Processes Landforms, 26, 461–474. rainfall [Mulligan, 1998]. Kröpelin, S., D. Verschuren, and A.-M. Lezine (2008), Climate- We simulated sufficient rainfall for flood water harvesting driven ecosystem succession in the Sahara: The past 6000 years, – and irrigation purposes at Naga that occurred more often until Science, 322, 765 768, doi:10.1126/science.1154913. 3000 years before present, owing to small atmospheric Liu, Z., Y. Wang, R. Gallimore, F. Gasse, T. Johnson, and J. Adkins changes rather than to a broad climate change. Consequently, (2007), Simulating the transient evolution and abrupt change of Northern Africa atmosphere–ocean–terrestrial ecosystem in the the Meroitic city of Naga was probably founded during a time Holocene, Quat. Sci. Rev., 26(13–14), 1818–1837, doi:10.1016/ when runoff-generating rainfall events appeared more reli- j.quascirev.2007.03.002. fl able, and therefore, ood water harvesting was more feasible. Luterbacher, J., E. Xoplaki, D. Dietrich, R. Rickli, J. Jacobeit, C. Beck, D. Gyalistras, C. Schmutz, and H. Wanner (2002), Recon- Acknowledgments. This study was supported by the Cluster of struction of sea level pressure fields over the Eastern North Excellence Exc264 TOPOI “The Formation and Transformation of Atlantic and Europe back to 1500, Clim. Dyn., 18(7), 545–561, Space and Knowledge in Ancient Civilizations.” We thank the doi:10.1007/s00382-001-0196-6. editor and two anonymous reviewers for their friendly and helpful Mawson, R. M., and A. Williams (1984), Awetter climate in eastern comments. Sudan 2,000 years ago?, Nature, 309,49–51. REFERENCES Mulligan, M. (1998), Modelling the geomorphological impact of climatic variability and extreme events in a semi-arid environ- Beck, C., J. Grieser, and B. Rudolf (2004), A new monthly precip- ment, Geomorphology, 24,59–78. itation climatology for the global land areas for the period 1951 to Nicoll, K. (2004), Recent environmental change and prehistoric 2000, Climate Status Report, pp. 181–190, Deutscher. Wetter- human activity in Egypt and Northern Sudan, Quat. Sci. Rev., dienst, Offenbach, Germany. 23, 561–580, doi:10.1016/j.quascirev.2003.10.004. Berking, J., and B. Schütt (2011), Late Quaternary morphody- Renssen, H., V. Brovkin, T. Fichefet, and H. Goosse (2003), Holo- namics in the area of the Meroitic settlement of Naga, Central cene climate instability during the termination of the African Sudan, Z. Geomorphol., 55(3), suppl., 1–24. Humid Period, Geophys. Res. Lett., 30(4), 1184, doi:10.1029/ Berking, J., B. Beckers, and B. Schütt (2010), Runoff in two semi- 2002GL016636. arid watersheds in a geoarcheological context: A case study of Schütt, B., K. Ducke, and J. Krause (2010), A cluster-analysis-based Naga, Sudan, and Resafa, Syria, Geoarchaeology, 25, 815–836, climate classification for NE Africa, Palaeoecol. Afr., 30, 199–225. doi:10.1002/gea.20333. Tjallingii, R. I. K., M. Claussen, J.-B. W. Stuut, J. Fohlmeister, A. Braconnot, P., et al. (2007), Results of PMIP2 coupled simulations Jahn, T. Bickert, F. Lamy, and U. Röhl (2008), Coherent high- of the Mid-Holocene and Last Glacial Maximum – Part 2: Feed- and low-latitude control of the northwest African hydrological backs with emphasis on the location of the ITCZ and mid- and balance, Nat. Geosci., 1, 670–675, doi:10.1038/ngeo289. high latitudes heat budget, Clim. Past, 3(2), 279–296, doi:10. Uppala, S. M., et al. (2005), The ERA-40 re-analysis, Q. J. R. 5194/cp-3-279-2007. Meteorol. Soc., 131(612), 2961–3012, doi:10.1256/qj.04.176. 168 CLIMATE-ARCHAEOLOGY

Wagner, S., M. Widmann, J. Jones, T. Haberzettl, A. Lücke, C. J. Berking and B. Schütt, Institute of Geographical Sciences, Freie Mayr, C. Ohlendorf, F. Schäbitz, and B. Zolitschka (2007), Universität Berlin Malteserstr. 74-100, D-12249 Berlin, Germany. Transient simulations, empirical reconstructions and forcing me- ([email protected]; [email protected]) chanisms for the Mid-Holocene hydrological climate in Southern U. Cubasch and J. Körper, Institute for Meteorology, Freie Patagonia, Clim. Dyn., 29, 333–355. Universität Berlin, D-12249 Berlin, Germany. (cubasch@zedat. Wilby, R. L., and T. M. L. Wigley (1997), Downscaling general fu-berlin.de; [email protected]) circulation model output: A review of methods and limitations, S. Wagner, Helmholtz Zentrum Geesthacht, D-21502 Prog. Phys. Geogr., 21, 530–548. Geesthacht, Germany. ([email protected]) The Influence of Transgressive Paleogeography on the Development and Decline of Cardium Pottery Culture (Mediterranean Neolithic)

A. Amorosi and A. Morelli

Dipartimento di Scienze della Terra e Geologico-Ambientali, University of Bologna, Bologna, Italy

Cardium Pottery is a decorative style which developed in the Mediterranean area during the Early Neolithic. The distinctive feature of Cardium Pottery Culture is the imprinting of the clay with the shell of the mollusc Cardium edule. Owing to its rapid expansion and relatively short duration (from a few centuries to about one millennium), this culture represents a powerful archeological marker for the Neo- lithic of the Mediterranean area. Despite the key role played by this archeological facies, the reasons for its appearance and disappearance are strongly debated in archeological circles and still far from a solution. Through the combination of stratigraphic and sedimentological data from the subsurface of modern Mediterra- nean coastal plains, this study puts a geological perspective into Cardium Pottery Culture, suggesting early to mid-Holocene changes in paleogeography across the Mediterranean as factors that influenced its origin, development, and (possibly) demise. Integration of radiocarbon and cultural ages shows that Cardium Pottery Culture grew during the early Holocene in concomitance with the widespread development of brackish environments within transgressive, barrier-lagoon-estuary systems, in which huge amounts of shells were made available. The landward migration of the shoreline, which occurred in response to the Holocene sea level rise, likely induced the progressive penetration of this culture toward more internal regions. The decline of Cardial Culture, among other (sociocultural) factors, was possibly favored by lagoon infilling in response to generalized mid-Holocene coastal progradation, which very rapidly made Cardium shells unavailable for ceramic decoration. 1. A GEOLOGICAL FRAMEWORK FOR THE Mediterranean area during the Early Neolithic [Perlès, MEDITERRANEAN NEOLITHIC 2001]. The most important ceramic production of this period is Cardium Pottery Culture (or Cardial Culture), which is The “Neolithic Revolution” is commonly defined by the based upon decorations by shell impressions of Cardium presence of sedentary villages concurrently with the domes- edule [Guilaine, 1994]. Pots were adorned, rather than tication of plants (agriculture) and animals (pastoralism) painted, with incised, impressed, or stamped patterns. This [Childe, 1925; Camps, 2001]. As for pottery, the Initial culture developed specifically in the Mediterranean area Neolithic was characterized by absence or scarcity of pottery, during a comparatively short-duration time window, cen- whereas significant ceramic production is recorded in the tered in the VI Millennium B.C. The Cardial Culture is traditionally subdivided into two separate archeological facies [Bernabeu Aubán, 1989; Martì Oliver, 2000, 2002; Climates, Landscapes, and Civilizations Guilaine and Manen, 2002]. During the Early Neolithic Geophysical Monograph Series 198 phase, ceramic impression was restricted almost entirely to © 2012. American Geophysical Union. All Rights Reserved. the use of shells of C. edule, while in the subsequent, late 10.1029/2012GM001205 Early Neolithic (epicardial) phase this culture evolved 171 172 TRANSGRESSIVE PALEOGEOGRAPHY AND CARDIUM POTTERY CULTURE through a more varied impression technique [Guilaine, 2000; progradation of deltaic and coastal systems occurred on a Zilhão, 2000]. global scale. The relationship between sea level change, delta construc- This paper examines the available literature on Cardium tion, and Neolithic culture is not a new issue. General hy- Pottery and makes a comparison between archeological find- potheses about how changes in Holocene paleogeography ings and well-known changes in paleogeography in the may have affected human societies have been offered from Mediterranean area. While high-resolution stratigraphic both a geological [Stanley and Warne, 1994] and archaeo- studies are available from the subsurface of the major Med- logical [Camps, 2001] perspective, but arguments and evi- iterranean deltas, such as the Nile [Coutellier and Stanley, dence supporting this appealing theory have not been 1987; Stanley and Warne, 1994], Ebro [Somoza et al., 1998], advanced in detail. The aim of this paper is to provide a Rhône [Oomkens, 1970; Tesson et al., 1990; Gensous and geological framework to the research on the Early Mediter- Tesson, 1996; Boyer et al., 2005; Berné et al., 2007], Po ranean Neolithic through integration of the archeological [Trincardi et al., 1994; Amorosi et al., 1999, 2008; Cattaneo record with detailed stratigraphic and sedimentological data. and Trincardi, 1999; Correggiari et al., 2005; Stefani and The specific objective is to document to what extent changes Vincenzi, 2005], Tiber [Bellotti et al., 1995], and Arno in Holocene transgressive paleogeography may have influ- [Amorosi et al., 2009], archeological data of the Early Neo- enced the spread and decline of Cardium Pottery across the lithic age from the Mediterranean are quite spotty and lack a Mediterranean area. The development of Cardial Culture comprehensive correlation scheme. Most Cardial archeolo- took place during the final stages (Holocene transgression) gical sites are domestic sites, such as caves, and their foothill of the post-Last Glacial Maximum (LGM) sea level rise, location is probably due to higher preservation of topograph- which occurred between about 19 and 7 calendar (cal) ka ically elevated areas. The common aspect of all these arche- B.P. This transgressive phase was discontinuous and punc- ological sites is their marked preference for locations very tuated by episodes of rapid sea level rise alternating with close to coastal systems [Guilaine, 1994; Atiénzar, 2007] short phases of stillstand or quasi-stillstand [Bard et al., (Figure 1). 1996; Hanebuth et al., 2000]. During this period, the lowstand (glacial) prograding bodies were drowned, and world- 2. CARDIUM EDULE OR CERASTODERMA GLAUCUM? wide landward migration of the shoreline occurred [Penland et al., 1988], turning fluvial mouths into backstepping, wave- The common feature of archeological and geological stud- dominated estuaries [Dalrymple et al., 1992; Amorosi and ies focusing on the Early Neolithic (or early mid-Holocene) Milli, 2001]. With marked deceleration in sea level rise is the shell used for pottery decoration, although species around 7000 cal B.P. [Bird et al., 2010], sediment supply attributions diverge for these two types of studies. Cardium outpaced the rate at which accommodation was created, and (= Cerastoderma) edule Linnaeus (1758), widely reported in

Figure 1. Map showing distribution of Cardial Culture in the Mediterranean. Note concentration of archeological sites (black dots) along coastal areas. AMOROSI AND MORELLI 173 the archeological literature, is a typical Atlantic species oc- B.C. From these earliest settlements, the colonization grad- curring from Mauritania to Northern Europe. Despite very ually moved to the west [Guilaine and Manen, 2007], set- slight differences in terms of salinity and temperature condi- tling in western Italy (Latium, Tuscany, Liguria, and tions [Brock, 1991], this species displays a niche overlap Sardinia) and Corsica, between 6000 and 5900 cal B.C. with Mediterranean species Cerastoderma glaucum Poiret Progressively younger ages document a rapid spread of Car- (1789) [Mariani et al., 2002]. The latter is a soft-bottom dial and related cultures to southern France (5800 cal B.C.), bivalve, widely reported from the Holocene geological re- SE Spain (5600 cal B.C.), and then Portugal (5500 cal B.C.), cord, which is typically found in confined and partly en- thus documenting a general expansion from east to west closed coastal environments, such as lagoons, salt marshes, during a time span of about seven centuries (Figure 2). The and brackish lakes [Kingston, 1974]. Owing to its intolerance end of Cardial Culture is not firmly constrained from a chro- to subaerial exposure [Russell, 1971] and high-energy wave nological viewpoint, owing to its gradual transition to the regimes [Boyden and Russell, 1972], C. glaucum cannot be epicardial culture. According to the most reliable data from encountered in nearshore (shoreface) or tide-dominated en- southern France and Corsica, it seems comprised between vironments and for this reason represents a powerful marker 5000 cal B.C. and 4500 cal B.C. of brackish paleoenvironments. Given the mutually exclu- The origin and history of Cardium Pottery Culture in the sive (Mediterranean versus Atlantic) modern geographic dis- Mediterranean appear to be intimately related to major tribution of these two species and their reduced genetic changes in coastal paleogeography, which took place in diversity, which makes their separation very difficult on a response to the twofold history of Holocene sea level rise morphological ground, misinterpretation of one (C. edule) and following stillstand. If matched against well-established for the other (C. glaucum) by the archeological literature Holocene sea level curves, the rapid onset of Cardial Culture, seems a reasonable possibility, although this hypothesis de- which implies sudden and widespread availability of C. serves additional paleontological investigation. edule/C. glaucum shells in their preferred habitat (brackish environments), appears to be correlative with the mid-Holocene 3. DEVELOPMENT AND DECLINE OF CARDIUM (Atlantic) phase of rapid sea level rise that followed the 8200 POTTERY: POSSIBLE CONTROLLING FACTORS cal B.P. cold event. This phase, also known as meltwater pulse 1d [Liu et al., 2004], is assigned to 8000–7500 cal B.P. A comprehensive analysis of the archaeological record and, in the Mediterranean, drove sea level around its present shows that the onset of Cardium Pottery in the western position [Lambeck et al., 2004]. During this interval of time, central Mediterranean area took place first in Greece and sea level rose of about 10 m and the interfluves flanking the southern Italy during the last centuries of VII Millennium LGM incised valleys were drowned. Owing to the very high

Figure 2. East to west expansion (arrows) of Cardium Pottery Culture in the Mediterranean. Dates (calendar year B.C., summarized from archaeology research) indicate the onset of Cardial Culture in the different regions. 174 TRANSGRESSIVE PALEOGEOGRAPHY AND CARDIUM POTTERY CULTURE rate of sea level rise, the Mediterranean coastal plains could On the other hand, cultural and demographic factors influ- not keep pace with rising sea level and were progressively enced the strongly diachronous spread of Cardium Pottery submerged. The landward migration of the shoreline promoted observed across the Mediterranean area (Figure 2). Particu- the widespread formation of backstepping barrier-lagoon- larly, the 700 years delayed onset of Cardial Culture in the estuary systems [Dalrymple et al., 1992]. In that period, the Iberian Peninsula compared to southern Italy, does not ap- Neolithic culture was centered on a highly dynamic land- pear to reflect an external (sea level, climate) control, but is scape, including wide coastal systems and large paleoestu- interpreted here as the simple response to human migration aries, the morphology of which was continuously re-shaped and diffusion along the Mediterranean coastal fringe, with by rising sea level. The observed inland expansion of Cardium foundation of newly established settlements in progressively Pottery from the coastal regions to the interior is interpreted more western locations. This human dispersal pattern, which to reflect the generalized backstepping of coastal deposition- is strongly consistent with the great maritime navigation al environments and fluvial mouths, which occurred on a capabilities of the Early Neolithic communities [Camps, variable distance of 10 to 50 km. Under a geological 2001], was probably stimulated and catalyzed by the remark- perspective, the supposed inland migration of ceramic pro- able physiographic changes of the Mediterranean coasts in duction “along fluvial paths” [Beeching, 1999; Martì response to the early Holocene sea level rise [Bird et al., Oliver, 2000; Guilaine and Manen, 2002] is much more 2010]. These changes probably led humans to explore new likely to reflect colonization of progressively more internal territories, settle the increasing number of islands made zones through a network of channels within wave-dominated available by the Holocene transgression, and thus colonize (Italy, France, Spain) to tide-influenced (Portugal) estuarine larger regions of the Mediterranean coasts. systems. There is evidence for concomitant maximum development REFERENCES of Cardium Pottery Culture across the whole Mediterranean area between 5500 and 5300 cal B.C. and the phase of Amorosi, A., and S. Milli (2001), Late Quaternary depositional maximum marine ingression, which is recorded in the same architecture of Po and Tevere river deltas (Italy) and worldwide region between 7800 and 7000 cal B.P. The achievement of comparison with coeval deltaic successions, Sediment. Geol., – nearly stable sea level conditions between about 7000 and 144, 357 375. 6500 cal B.P. [Bird et al., 2010; Lambeck et al., 2011], Amorosi, A., M. L. Colalongo, G. Pasini, and D. Preti (1999), among other factors, may have contributed to the rapid Sedimentary response to Late Quaternary sea-level changes in the Romagna coastal plain (northern Italy), Sedimentology, 46, decline of Cardial Culture. When the rate of sea level rise 99–121. decreased, a generalized phase of coastal and deltaic progra- Amorosi, A., E. Dinelli, V. Rossi, S. C. Vaiani, and M. Sacchetto dation rapidly took place due to overwhelming sediment (2008), Late Quaternary palaeoenvironmental evolution of the fi supply [Stanley and Warne, 1994]. This led to signi cant Adriatic coastal plain and the onset of Po River Delta, Palaeo- reduction (and eventually, disappearance) of brackish, shallow- geogr. Palaeoclimatol. Palaeoecol., 268,80–90. water environments suitable for Cardium/Cerastoderma Amorosi, A., M. Ricci Lucchi, V. Rossi, and G. Sarti (2009), development. Ultimately, lagoons and bays were filled dur- Climate change signature of small-scale parasequences from ing coastal progradation and then replaced by freshwater Lateglacial-Holocene transgressive deposits of the Arno valley swamps and floodplain environments. Although it cannot be fill, Palaeogeogr. Palaeoclimatol. Palaeoecol., 273, 142–152. stated that the demise of Cardium Pottery, which is still an Atiénzar, G. G. (2007), Occupazione e sfruttamento del territorio unsolved problem, was direct consequence of this phase of nel Neolitico: L’alto e medio bacino del fiume Serpis (Alicante, decelerated sea level rise, the influence of coastal evolution Spagna), Boll. Paleoetnol. It., 96,17–36. on the last stage of culture cannot be ruled out. In this Bard, E., B. Hamelin, M. Arnold, L. Montaggioni, G. Cabioch, G. respect, additional radiometric dates on shells are needed to Faure, and F. Rougerie (1996), Deglacial sea-level record from better constrain the archeological data into a high-resolution Tahiti corals and the timing of global meltwater discharge, Na- ture, 382, 241–244. chronologic framework. Beeching, A. (1999), Les premières étapes de circulation et de peu- plement dans les Alpes françaises au Néolithique. Apport de la 4. CONCLUSIVE REMARKS céramique, in Circulations et Identités Culturelles Alpines à la Fin de la Préhistorie, edited by A. Beeching, pp. 427–480, Travaux du The synchronous response of Cardium Pottery Culture to Centre d’Archéologie Préhistorique de Valence, Valence, France. the Early Neolithic environmental changes across different Bellotti, P., S. Milli, P. Tortora, and P. Valeri (1995), Physical geodynamic settings of the Mediterranean area constitutes an stratigraphy and sedimentology of the Late Pleistocene-Holocene example of cultural response to Holocene sea level change. Tiber Delta depositional sequence, Sedimentology, 42, 617–634. AMOROSI AND MORELLI 175

Bernabeu Aubán, J. (1989), La tradición cultural de las cerámicas Guilaine, J., and C. Manen (2007), From Mesolithic to early impresas en la zona oriental de la Península Ibérica, Trabajos Neolithic in the western Mediterranean, in Going Over: The varios del SIP, 86, Diputación Provincial de Valencia, Valencia, Mesolithic-Neolithic Transition in North-West Europe, edited Spain. by A. Whittle and V. Cummings, Proc. Brit. Acad., 144, Berné, S., G. Jouet, M. A. Bassetti, B. Dennielou, and M. Taviani 21–51. (2007), Late Glacial to Preboreal sea-level rise recorded by the Hanebuth, T., K. Stattegger, and P. M. Grootes (2000), Rapid flood- Rhône deltaic system (NW Mediterranean), Mar. Geol., 245, ing of the Sunda shelf: A late-glacial sea-level record, Science, 65–88. 288, 1033–1035. Bird, M. I., W. E. N. Austin, C. M. Wurster, L. K. Fifield, M. Kingston, P. F. (1974), Studies on the reproductive cycles of Car- Mojtahid, and C. Sargeant (2010), Punctuated eustatic sea-level dium edule and C. glaucum, Mar. Biol., 28, 317–323. rise in the early mid-Holocene, Geology, 38, 803–806. Lambeck, K., F. Antonioli, A. Purcell, and S. Silenzi (2004), Sea Boyden, C. R., and P. J. C. Russell (1972), The distribution and level change along the Italian coast for the past 10,000 yrs, Quat. habitat range of the brackish water cockle Cardium (Cerasto- Sci. Rev., 23, 1567–1598. derma) glaucum in the British Isles, J. Animal Ecol., 41, Lambeck, K., F. Antonioli, M. Anzidei, L. Ferranti, G. Leoni, G. 719–734. Scicchitano, and S. Silenzi (2011), Sea level change along the Boyer, J., C. Duvail, P. Le Strat, B. Gensous, and M. Tesson (2005), Italian coast during the Holocene and projections for the future, High resolution stratigraphy and evolution of the Rhône delta Quat. Int., 232, 250–257. plain during Postglacial time, from subsurface drilling data bank, Liu, J. P., J. D. Milliman, S. Gao, and P. Cheng (2004), Holocene Mar. Geol., 222–223, 267–298. development of the Yellow River’s subaqueous delta, North Brock, V. (1991), An Interdisciplinary Study of Evolution in the Yellow Sea, Mar. Geol., 209,45–67. Cockles Cardium (Cerastoderma) Edule, C. glaucum and C. Mariani, S., V. Ketmaier, and E. de Matthaeis (2002), Genetic lamarcki, 31 pp., Vestjydsk, Vinderup, Denmark. structuring and gene flow in Cerastoderma glaucum (Bivalvia: Camps, G. (2001), Il Neolitico Mediterraneo. Tecniche e Modi di Cardiidae): Evidence from allozyme variation at different geo- Vita, 95 pp., Jaca Book SpA, Milan, Italy. graphic scales, Mar. Biol., 140, 687–697. Cattaneo, A., and F. Trincardi (1999), The Late Quaternary trans- Martí Oliver, B. (2000), La Cerámica Cardial. El Neolítico de la gressive record in the Adriatic epicontinental sea: Basin widening Península Ibèrica, Monografías de Arte Roca, Barcelona, Spain. and facies partitioning, in Isolated Shallow Marine Sand Bodies: Martí Oliver, B. (2002), La Ceramica Impressa della Penisola Sequence Stratigraphic Analysis and Sedimentological Interpre- Iberica, in Le Ceramiche Impresse nel Neolitico Antico. Italia e tation, edited by K. M. Bergman and J. W. Snedden, SEPM Spec. Mediterraneo, edited by M. A. Fugazzola Delpino, A. Pessina, Publ., 64, 127–146. and V. Tinè, pp. 51–62, Istituto Poligrafico e Zecca dello Stato, Childe, V. G. (1925), The Dawn of European Civilization, 392 pp., Rome, Italy. Routledge and Kegan Paul, London, U. K. Oomkens, E. (1970), Depositional sequences and sand distribution Correggiari, A., A. Cattaneo, and F. Trincardi (2005), Depositional in the post-glacial Rhône delta complex, in Deltaic Sedimenta- patterns in the Holocene Po Delta system, in River Deltas: Con- tion – Modern and Ancient, edited by J. P. Morgan, SEPM Spec. cepts, Models and Examples, edited by J. P. Bhattacharya and L. Publ., 15, 198–212. Giosan, SEPM Spec. Publ., 83, 365–392. Penland, S., R. Boyd, and J. R. Suter (1988), Transgressive depo- Coutellier, V., and D. J. Stanley (1987), Late Quaternary stratigra- sitional system of the Mississippi Delta Plain: A model for barrier phy and paleogeography of the eastern Nile delta, Egypt, Mar. shoreline and shelf sand development, J. Sediment. Petrol., 58, Geol., 77, 257–275. 932–949. Dalrymple, R. W., B. A. Zaitlin, and R. Boyd (1992), Estuarine Perlès, C. (2001), The Early Neolithic in Greece. Cambridge World facies models: Conceptual basis and stratigraphic implications, Archaeology, Cambridge Univ. Press, Cambridge, U. K. J. Sediment. Petrol., 62, 1130–1146. Russell, P. J. C. (1971), A reappraisal of the geographical distribu- Gensous, B., and M. Tesson (1996), Sequence stratigraphy, seismic tions of the cockles Cardium edule L. and C. glaucum Bruguière, profiles, and cores of Pleistocene deposits on the Rhône conti- J. Conchol., 27, 225–234. nental shelf, Sediment. Geol., 105, 183–190. Somoza, L., A. Barnolas, A. Arasa, A. Maestro, J. G. Rees, and F. J. Guilaine, J. (1994), La Mer Partagée. La Méditerranée Avant Hernandez-Molina (1998), Architectural stacking patterns of the L’écriture, 910 pp., Hachette, Paris, France. Ebro delta controlled by Holocene high-frequency eustatic fluc- Guilaine, J. (2000), Premiers Paysans du Monde, la Naissance des tuations, delta-lobe switching and subsidence processes, Sedi- Agricultures, 320 pp, Errance, Paris, France. ment. Geol., 117,11–32. Guilaine, J., and C. Manen (2002), La Ceramica Impressa della Stanley, D. J., and A. G. Warne (1994), Worldwide initiation of Francia Meridionale, in Le Ceramiche Impresse nel Neolitico Holocene marine deltas by deceleration of sea level rise, Science, Antico. Italia e Mediterraneo, edited by M. A. Fugazzola Delpi- 265, 228–231. no, A. Pessina, and V. Tinè, pp. 37–50, Istituto Poligrafico e Stefani, M., and S. Vincenzi (2005), The interplay of eustasy, Zecca dello Stato, Rome, Italy. climate and human activity in the late Quaternary depositional 176 TRANSGRESSIVE PALEOGEOGRAPHY AND CARDIUM POTTERY CULTURE

evolution and sedimentary architecture of the Po Delta system, Zilhão, J. (2000), From the Mesolithic to the Neolithic in the Iberian Mar. Geol., 222–223,19–48. peninsula, in Europe’sFirstFarmers,editedbyT.D.Price, Tesson, M., B. Gensous, G. P. Allen, and C. Ravenne (1990), Late pp. 144–182, Cambridge Univ. Press, Cambridge, U. K. Quaternary deltaic lowstand wedges on the Rhône continental shelf, France, Mar. Geol., 91, 325–332. Trincardi, F., A. Correggiari, and M. Roveri (1994), Late Quaternary A. Amorosi and A. Morelli, Dipartimento di Scienze della Terra transgressive erosion and deposition in a modern epicontinental e Geologico-Ambientali, University of Bologna, Bologna IT- shelf: The Adriatic semienclosed Basin, Geo-Mar. Lett., 14,41–51. 40126, Italy. ([email protected]) A Dynamic Human Socioecology of Prehistoric and Protohistoric Ulster

Tina Thurston

Department of Anthropology, University at Buffalo, State University of New York, Buffalo, New York, USA

Gill Plunkett

School of Geography, Archaeology and Palaeoecology, Queen’s University, Belfast, UK

Remains of past human activity in Northern Ireland’s Bronze Age, Iron Age, Medieval, and Early Modern eras are normally invisible under continuous pasture cover. By combining scientific indicators, archaeological methods, and historical documents, we suggest new ways of understanding “invisible” change throughout this period. Political activity at “royal” sites correlates with increased farming seen in lake core pollen, while soil geochemical mapping reveals normally unseen, undiscovered farming settlements. Such multidisciplinary data indicate that climate and environmental conditions influenced high-level political outcomes in a variety of critical ways, while challenging farmers with unusually difficult livelihood choices.

1. INTRODUCTION 2. A TRANSDISCIPLINARY VENTURE

Since 2008, researchers from the Department of Anthro- While paleoecology studies environmental events and pology at The University at Buffalo, State University of New trends, and archaeology focuses on the record of past human York and Queen’s University Belfast, School of Geography, activity, scholars in both ﬁelds already have a long history of Archaeology and Palaeoecology, have participated in a col- combining these approaches to study humans in long-term laborative project on the political ecology of prehistoric and ecological contexts from the very earliest appearance of our protohistoric Ulster, linking paleoecological, archaeological, species until the present era. Yet at one time, not so long ago, and ethnohistoric perspectives (Figure 1). such study could be described as deterministic and unidirec- In our study of the former kingdom of Ulster, now largely tional in nature [Berglund et al., 1991; Berglund, 2003; comprising Northern Ireland, we consider the results of Butzer, 1982; Cronk, 1991; Winterhalder and Smith, 2000], our individual work while simultaneously undertaking new decidedly devoid of the social, adopted almost directly from collaborative research, in an attempt to study the role of animal-oriented studies in ﬁelds such as evolutionary ecol- human-environment relationships in the political and eco- ogy and economic optimization studies. This grew from a nomic processes of the Late Bronze Age, Iron Age, Medieval, marriage of sociobiology and functionalism in the last quar- and Early Modern eras, from about 1200 before the Common ter of the twentieth century and has elicited a growing cri- Era (B.C.E.) to 1700 Common Era (C.E.). tique [Boone and Smith, 1998; Gremillion, 2002; Ingold, 1996; Pauketat, 2004]. We do not desire to enter this debate, only to offer studies that emphasize the possibility that human ecological research need not inordinately privilege “sci- ” “ ” Climates, Landscapes, and Civilizations ence or culture. We also hope to argue that environmental Geophysical Monograph Series 198 archaeology and human paleoecology should not only be © 2012. American Geophysical Union. All Rights Reserved. conducted at the local or site-based level, that is, reconstruct- 10.1029/2012GM001227 ing human-environment interactions in the context of a 177 178 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

through the use of resilience theory, the notion that the environmental troubles (whatever their cause, social or natural) of one cultural iteration may form the thriving basis for its next phase or for later cultures in the same regions. Both seminal theorists and authors more recently initiated into this genre have pointed out that local conditions are often partly predicated upon nonlinear, planetwide complex adaptive systems that are largely unpredictable. Within these systems, people often only indirectly abuse the environment, often in ways neither they, in the past, nor even we today, could predict. Moreover, human impacts are indeed significant contributors to large- and small-scale conditions, but they are often beneficial impacts, not only destructive [Cha- pin et al., 2010:58; Costanza et al., 2007:522]. Today, new studies of human-environment relationships focus on the nonlinear dynamism of such relationships, in which human impacts and creative strategies for dealing with environmental and climatic challenges are given equal con- Figure 1. Britain, Ireland, and the study region within Northern sideration alongside climate-driven processes, and the rela- Ireland. Study area location is 54°32′N, 6°33′W. tionship is assumed to be in constant state of change. Another problem identified by critics of early historical particular settlement or catchment area. Given the growing ecology, when practiced in historic or protohistoric periods, focus on broader phenomena, many at the leading edge of was that too much was often assumed from the textual such studies advocate that the most beneficial “scale” at record. The historic record offers significant resources for which to study such issues is, in fact, multiscalar: from the recovering data and studying relational interactions between site, to the landscape, to the region, and beyond [Crumley, people, their social institutions, and environment and cli- 1994, 2000, 2007; Lock and Molyneaux, 2007]. mate, yet texts alone, especially those prior to the seven- For some time, researchers have been seeking appropriate teenth century C.E., rarely offer a complete picture of human alternative theoretical perspectives for such work. Since the and environmental impacts. Ecological, archaeological, and late 1980s, the approach known as historical ecology has historic data must be considered together in the light of social attempted to link “historicist” data (both historic and prehis- theory, to assess critically each other’s accuracy and to create toric) about specific cultures and time periods together with a data set that is more comprehensive than any single-source ecological data in an effort to shed light on these relation- resource. ships [Balée, 1998, 2006; Crumley, 1994]. Early iterations of historical ecology were critiqued as displaying deterministic 3. A DYNAMIC HISTORICAL SOCIOECOLOGY tendencies despite claims to the contrary, and when human OF ANCIENT ULSTER agency was credited with impacts, early practitioners were too focused on the negative impacts of human beings on the Our ongoing project attempts to discern first how human- environment [Whitehead, 1998], often drawing fairly direct environment relationships in the Late Bronze Age, Iron Age, connections between past human activities and the destruc- Medieval, and Early Modern eras (from about 1200 B.C.E. tive consequences of their actions. to 1700 C.E.) impacted the lives of ordinary people and, Yet from this critique, the last two decades have seen the second, how it aided or hindered the ambitions of expansion- growth of an historical ecology infused with a dynamic ary polities and colonial states. socioecology, an outgrowth from the “new ecology” per- It is true that the region’s political trajectory from the late spective [Balée and Erickson, 2006; Erickson, 1993, 1999]. prehistoric to the modern day has been a varied and tumul- This paradigm has been put forward to compensate for the tuous one, with an apparently abrupt social reorganization at issues raised in the critique and includes the explicit idea of the transition from Bronze Age to Iron Age, an Iron Age chaotic flux as a normal state to which humans are actually marked by internecine chiefly competition and warfare, the well adapted, a view that humans can equally impact and be aggressive fifth century C.E. Christian proselytizing mission, impacted by the environment in a recursive cycle (sometimes ninth century Viking colonization, twelfth century Anglo- simultaneously), that not all human impacts are negative, and Norman invasion, and finally, the Tudor and Jacobean THURSTON AND PLUNKETT 179

English conquest and plantations of Ulster in the sixteenth- climate and environmental change are considered particularly seventeenth-eighteenth centuries. Each regime had different useful. Its oceanic bogs, reflecting influence from the oce- goals and made different demands on the public. anic climate zone’s mild, moist conditions, are responsive The paleoclimate record shows a region impacted by many to shifts in precipitation and temperature, and this is re- major and minor climate events over three millennia. In flected by the humification (decomposition) of the peat, the recent years, several well-dated palaeohydrological recon- plant assemblages, and the subfossiled fauna and microfauna structions have been obtained from bogs across the North of within the peat [Barber et al., 2004; Swindles et al., 2007b, Ireland that suggest a Late Bronze Age climate deterioration, 2010]. Comparison of bog proxy data and instrumental circa 750 B.C.E., a “Roman warm period” beginning in the records spanning the last century or so suggest that bog third century B.C.E., followed in the late fifth century C.E. surface wetness in western Europe is mainly determined by by the onset of the Dark Age climate deterioration that lasted the effective summer precipitation (i.e., the relationship until the Medieval Warm Period at about 800 C.E. During the between temperature-controlled evaporation and precipita- last millennium, a pronounced wet/cold phase at circa 1650 tion) [Charman et al., 2004; Charman, 2007]. Hence, dry C.E. corresponds to the most severe stage of the Little Ice shifts in the North of Ireland bogs have been interpreted as Age, which was brought to an end, circa 1850 C.E., by a shift periods of increased summer drought [Swindles et al., to warmer/drier conditions [Plunkett, 2006; Swindles et al., 2010]. Other paleoenvironmental proxies spanning the late 2007a, 2007b, 2010]. Parallel with these changes, a number postglacial period in Ireland include isotopic and chemical of shifts in economic and agricultural foci are seen in the data from a speleothem record in SW Ireland [McDermott et archaeological and paleoecological records, pertaining to the al., 2001] and lake sediments [e.g., Schettler et al., 2006]. livelihoods of farmers and herders. What was the cause of Climatologists also routinely use Ireland’s unusually plenti- such shifts? Political change? Climate? Some combination of ful “Dark Age” textual record to glean phenological obser- these effects or something completely different? vations for comparison with scientific data, as they often note years of unusual plenty or famine due to weather and 4. BACKGROUND AND CONTEXT environmental conditions. As noted, an important component for linking human Environmental archaeology has as long a history in Ire- behavior with climate is the production of environmental land as anywhere else in the world [Murphy and White- data on a temporal scale matching human lifespans rather house, 2010], aided by the fact that conditions across the than geological processes; Irish bogs deliver fine-scale data, island make it highly fruitful for North Atlantic and general often at decadal resolution [Plunkett et al., 2004]. The European climatological research. Northern Ireland lies be- dating and cross correlation of paleoclimate reconstructions tween the 54th and 56th parallels, comparable in Europe to from Irish bogs, in particular, has been greatly facilitated by Denmark and Sweden, and in North America with central- the frequent occurrence of volcanic ash (tephra) layers northern Labrador and lower Hudson’sBay.Irelandis [Hall and Pilcher, 2002], the dates of which are established considerably more temperate than some areas at the same either through geochemical correlation of historically re- latitude, due to its proximity to the Gulf Stream, yet the corded eruptions (mainly Icelandic eruptions since the ninth impacts of climate change are still marked. As the western- century) or through 14C dating [Pilcher et al., 1996; Plun- most point of Atlantic Europe (excluding Iceland and kett et al., 2004]. In the context of this study, tephras of Greenland), Ireland has been among the first places in Eur- Late Bronze Age and Early-Late Medieval date, in partic- ope to experience the effects of rapid climate changes: ular, provide useful tie-points between paleoclimatic and deglaciation, sea level rise, cooling, and warming; while on palynological studies across the island [Hall and Bunting, a geological time frame, such changes seem to occur simul- 2000; Hall, 2005; Plunkett, 2006, 2009; Swindles et al., taneously across northern Europe, on a human time scale, 2010]. Paleoenvironmental data are achievable through ex- there are large differences in the timing of these processes cellent pollen preservation, testate amoebae-derived water [Bradwell et al., 2008; Greenwood and Clark, 2009; Hub- table reconstruction, peat humification changes, and plant bard et al., 2009; Knight, 2003; McCabe and Clark, 2003; macrofossil analyses (Figure 2). In comparison with oceanic McCabe et al., 2005, 2007; Sejrup et al., 2005; Finlayson et zone syntheses, Northern Ireland specifically saw major al., 2010]. The region remains highly sensitive to distur- wet/cold shifts at circa 1510 B.C.E., 750 B.C.E., and bances in the Atlantic conveyer, which sits virtually right off 470 C.E. [Swindles et al., 2007b]. Smaller magnitude but the coast [Glick et al., 2004], making northwest Ireland one still significant shifts are also recorded at circa 380 B.C.E., of modern climatology’s most studied regions. In terms of 150 B.C.E., 180 C.E., and 690 C.E. [Swindles et al., 2007a, prehistoric and protohistoric trends, Irish proxy data for 2007b]. 180 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

Figure 2. Historical and paleoecological indicators of climate change [after Kerr et al., 2009].

Both the individual and collaborative work of our project’s As noted briefly above, Ireland lies almost entirely in participants is predicated upon the unification of environ- pasture, grass, or other rough grazing cover, making the mental data with archaeology and, in some instances, history. residential and industrial sites of those who farmed, herded, As can be imagined, the historic texts are relatively plentiful, and produced in support of their changing systems of ruler- there is no lack of environmental data or of archaeological ship, difficult to discern. The issue concerning low-visibility research projects; the primary issue rather is the problem of sites is most pertinent during the protohistoric period, a scale. While some questions can be answered by examining situation recently summed up nicely by Bradley [2007], single “known” sites, since the 1990s [Bender, 1992, 1993; without any words at all (Figure 3) in his recent book on Crumley and Marquardt, 1990; Deetz, 1990; Dunnell, 1992], British and Irish prehistory. The issue is equally problematic there has been an acknowledgment across the discipline that for earlier and later periods, an exception being the earth or a large number of potential research questions about processes stone built “raths” (enclosed or defended homesteads) that of change or longitudinal developments within society can proliferate in the Early Medieval period and that dominate only be answered by looking at regional and cultural land- the present understanding of past settlement patterns. These scape trends across time. This, in turn, is another area of our sites suggest a dispersed form of settlement, but whether they specialization. Understanding human activities, land clear- represent the only form of settlement of the time is not ance, farming, land abandonment, in pollen records is simi- known. Bronze Age settlement was similarly deemed to have larly best achieved through the adoption of a wider been of a dispersed nature until the fortuitous discovery of perspective, in which these activities are assessed in terms what can only be described as a village comprising more than of their spatial and temporal dimensions [Plunkett, 2009]. 70 structures, a so far unique example for its time, at THURSTON AND PLUNKETT 181

political, economic, and social trajectories of politically complex societies past and present, an idea with a long history in diverse disciplines [Caseldine and Turney, 2010; DeMeno- cal, 2001; Gallet et al., 2006; Stothers, 2000; Turney et al., 2006; Warrick and Riebsame, 1981; Young and Lipton, 2006; Huang and Su, 2009]. Ireland presents a valuable case study for long-term investigation of such socionatural relationships. Two examples from Medieval Ireland and closely connected Gaelic Scotland provide a window on this phenomenon. Historians have long noted that the “traditional” Celtic cattle complex, in which prestige and wealth was invested in animals, shifted in the Early Middle Ages to a system of land-based wealth. In the eighth and ninth century Ireland, paleoenvironmental proxy data suggest a shift toward increased cereal production at that time [Hall, 2005; “ Figure 3. The character of Iron Age settlement in Britain and the Lomas-Clarke and Barber, 2004; Newman et al., 2007; ” near-continent [from Bradley, 2007, Figure 5.14]. Latitude/longi- Weir, 1995]. It has been suggested that this was due to tude range for depicted area is 54°41′N, 6°56′W for Northern Ire- ′ ′ ′ ′ population increase; despite the preponderance of raths gen- land; 58°27 N, 4°44 W for northern Scotland; 50°54 N, 1°34 W for fi southern England; and 49°11′N, 1°18′W for the coast of France. erally attributed to this period, without a rm chronological perspective of their occupation and an understanding of their representativeness of settlement and society, the theory Corrstown, County Londonderry [Conway et al., 2005]. of population increase cannot be tested. It has also been Beyond Ireland, fieldwalking, examining the surface for suggested that as the Uí Néill dynasty took over the High artifacts, is still the primary mode of prospection for non- Kingship of Ireland in the mid-ninth century, they “effec- monumental sites, usually undertaken both by archaeologists tively relegated the earlier system of kingship and hierar- and amateurs, or elsewhere in Europe, by proxy from the chies to the position of a primitive survival” (Byrne [2001] large collections of farmers from before World War II, who cited by Kerr et al. [2009, p. 2,870]. Others suggest that easily spotted artifacts riding behind a horse-drawn plow. In establishment of Viking trading centers with a “money” Ireland, this practice is rarely carried out, and when it has economy eroded the cattle-based system [Valante, 2008]. been, it has usually been with the express purpose of exam- Yet by using multiple data threads, historical, archaeologi- ining earlier prehistoric remains [e.g., Green and Zvelebil, cal, and climatological, Kerr et al. [2009] demonstrate a 1990]. With the advent of contract archaeology in compli- climatic downturn during which more severe winters may ance with preservation laws, more sites have been found, but have prevented traditional outwintering of cattle, and since not enough in any one area to reveal patterns, except the path haymaking for animal fodder was not yet practiced, a shift of the modern highways with which they are associated. to increased crop production occurred (Figure 2). This is Recently, attempts at syntheses of archaeological and cli- amply seen archaeologically in the sudden construction of mate data has begun [Hall, 2003, 2005; Kerr et al., 2009; grain-drying kilns all over the island, a technology that Stout, 1997; Plunkett, 2009] only because it seems apparent made increased cereal production tenable in the cold, wet to many researchers that significant climate changes and period. Kerr et al. [2009] argue that the climate shift and the related environmental shifts are in some way intertwined processes it precipitated were factors in the ultimate disso- with social, economic, and political changes that altered the lution of both the economic and political system. Archaeo- course of Irish culture [Kerr et al., 2009]. Although few in logically, in the North of Ireland at least, the period saw the number, these hint already at the rich data that could be demise of settlement associated with earlier raths predicated revealed by such collaborations. on cattle production, in favor of a new type of rath in a pattern seemingly related to arable farming [McCormick, 5. THE STATE AND NATURE 1992, 2008; Kerr, 2009]. The role of, and impact on, the less wealthy during this change is unknown for either period. We approach our topic, the idea of the state and nature, Nevertheless, this example serves to illustrate social and through the idea that natural forces such as climate, weather, political adaptation to ostensibly adverse environmental environmental conditions, and topography play a role in the conditions. 182 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

A second example comes from the twelfth to fourteenth disconnected from the indigenous population. Historians, centuries, when both Irish and Scots-Irish communities saw who are largely humanists, have been strongly critiqued by the creation of highly militarized small polities, replete with scholars in the social sciences for the practice of extrapolat- mercenary buannachan armies in Ireland, and ceatharn mer- ing from such textual records that are confined to one place cenaries in Scotland [Oram and Adderley, 2008]. For Ireland, and even one year, to make broad statements about the past this is often attributed to the impact of the 1167 C.E. Anglo- of which, in reality, little is known [Richardson, 2001]. Art- Norman invasion and subsequent colonization, precipitating works such as paintings and illuminated manuscripts, which the transformation of territorially expansive kingdoms ruled are highly contextualized within aristocratic ideals and sym- by rí (kings), to something more like warring chiefdoms bolism, still provide more information on the life of peasants [FitzPatrick, 2004]. In “Irish” Scotland, it is usually seen as and how they were perceived than do any written texts an outgrowth of the Anglo-Scottish wars of the thirteenth [Alexander, 1990]. Records did not become detailed until century, when political destabilization prompted local lords the conquest of Ireland by the English Plantation colonists in to reinforce their power and extend their domains using the sixteenth century, whose systems of land tenure, rural mercenary warbands [Oram and Adderley, 2008], some of law, and regulation of social and economic relationships which were exchanged with leaders across the Irish Sea. In between farmers and nobles was entirely different and alien both Ireland and Gaelic Scotland, such feuding and raiding to those understood for Ireland. At that time, the conditions are traditionally seen as representing political competition of indigenous people, as in all colonial systems, were dis- among militarized kin groups aiming to subordinate neigh- rupted and altered, again in ways not fully understood by bors. Oram and Adderley [2008] have used historic records historians. Were the Irish pushed off their land and replaced and proxy climate data for the emergent Little Ice Age (LIA), by planters? Or did they occupy differing landscapes already which together record massive impacts on human and animal due to different livelihoods? Did their health or nutritional populations: terrible winters, sustained climatic disturbances, status change, did the paths and projects of their daily lives the growing season shortened by a month, and decreased transform or persist? Archaeological and paleoenvironmen- arable and pasturage: from the ability to farm even in a broad tal data are among the only windows into the lives of farmers, high altitude range in the Medieval Warm Period, to a much herders, and other nonelite indigenous Irish. It is with the smaller range in the LIA, as conditions at higher elevations goal of resolving these questions that we collaborate in the were no longer able to support a full growing season for grain study of our latest phase, the Early Modern era, which itself crops. At the same time, a cattle plague recorded in both must be considered against the environmental backdrop of Scotland and Ireland wiped out up to 90% of the herds. This the most severe phase of the Little Ice Age. soonledtothelocalexpressionofthe“Great European During these and other episodes of shifting conditions, a Famine” of 1315–1322. range of climate change effects are apparent, and more than Increased militarism is contextualized inside of competi- once, their impact may have altered the course of the socio- tion within this framework: competition not only for “polit- economic system, propelled change in almost every sector, ical dominion” as usually held, but for actual arable land, perhaps contributing to both protection from and ultimate pasture, and animals to rectify losses. At the same time, to vulnerability to expansionist colonial powers. If climate feed mercenaries, tenant taxation increased, and many re- changes do indeed correlate specifically to parts of Ulster’s vised herd management practices are seen in faunal profiles. political trajectory, did they produce conditions that sus- Some elite kin groups succeeded with these strategies, while tained or undermined local agropastoral support for indige- others did not. nous governments? Did this, in turn, facilitate or inhibit the The impact of militarization, increased taxes, starvation, political ends of Ulster’s external would-be rulers? These and new modes of livelihood on ordinary people in Ireland is relationships between the state and nature did not operate at unknown. It is easy to confuse generalized top-down records a conscious level in the past; rather, nature formed a substrate of field systems, taxation goals, and how markets were for political development, with chiefdoms, kingdoms, and administered as representing the conditions of daily life for empires instrumentalizing natural conditions to coalesce, individual people, families, and communities. Before circa guide policy, and govern, but also to falter, fragment, and 1200 C.E., such records of local economic conditions do not reorganize in cycles of dissolution and resilience. While we generally exist, and the only “data” comes from occasional propose to incorporate study of “elite” actions and reactions random notations in annual monastic chronicles or docu- vis-à-vis ecological change, our primary goal is elucidating ments like law codes created at the behest of nobles and what changes are apparent for the subjects of these various royalty. After that, what little documentation there is comes regimes, farmers, herders, craft producers, and other ordinary from England or the Anglo-Normans in Ireland who were people, in order to navigate both what was thrust on them by THURSTON AND PLUNKETT 183 their rulers, as well as to deal individually and communally In contrast, the same chronological imprecision can hinder with climate and environmental change in their own daily the recognition of causal relationships in archaeological and lives, by altering practices, making decisions, and imple- paleoenvironmental records through the process of “smear- menting new ideas. Ethnohistoric texts and the “visible” ing” (in the sense of Baillie [1991]). This is a problem that is archaeological remains most studied in Ireland lend some likely to remain with us for some time, but attempts to power to interpretation of elites; we focus on new methods mitigate it are achieving some success through the use of aimed specifically at revealing populations that are heavily Bayesian and other quantitative analyses of chronological underrepresented both historically and archaeologically. data [e.g., Gearey et al., 2009; Bayliss et al., 2011]. Insofar as long-term perspectives of human activities can be gleaned 6. CHALLENGES IN PREHISTORY from pollen records, all the more so if study sites are well- positioned in relation to archaeological sites of interest, A word of caution is required, however, when dealing with palaeoenvironmental investigations can address questions of increasingly less well-dated evidence further back in time. climate impacts on subsistence strategies, human adaptation Paleoclimatic inferences for the Bronze and Iron ages must techniques in the face of climate change, and human impacts rely on proxy data, which come with varying levels of on the landscape. uncertainty in terms of their representation of climate change, their spatial relevance, and the timing of “events.” 7. ENVIRONMENT AS AGENT AND ACTOR IN There are no contemporary commentaries with which to PREHISTORIC AND PROTOHISTORIC ULSTER check their reliability or precision. There have been tempta- tions then to overlook unyielding chronological imprecision Based on prior work across Ireland, we have focused our and to draw correlations between events, archaeological and/ collaborative work on two select landscapes in northern or paleoenvironmental, that have closely comparable dates. Armagh and southern Tyrone. The project, entitled Environ- For instance, environmental deterioration has been invoked ment as Agent and Actor in Iron Age, Medieval, and Early to explain the emergence of hillforts in Late Bronze Age Modern Ulster studies change from the Late Bronze Age Ireland, arising from the comparable, albeit imprecise, tim- through Early Modern eras, via palynological studies that ing of the construction of such sites and a precisely dated, identify changing levels of agropastoralism during this longue extreme environmental event recognized in the dendrochro- durée when the region underwent several dramatic shifts in its nological record [Baillie, 1989]. This theory has since been sociopolitical regimes. Plunkett [2009] has noted that levels of undermined by the realization that the construction of Late agricultural activity around “elite” sites such as high-elevation Bronze Age hillforts was in fact diachronic [Bradley, 2007] enclosures vary through time, yet this may stem alternately and cannot therefore be explained by a single, simplistic from overall increases in production via an increase in the deterministic model, which in this case seems to be a classic number of people farming, or it may represent a nucleation of example of what Baillie [1991] terms “suck-in.” It might extant farmers around sites to which they have political or also be tempting to identify cause-and-effect from a phase of economic connections. woodland regeneration (implying land abandonment) and a To assess this, the change or stability of the regional decline in the settlement record roughly around the time of pattern of settlement through time must be understood. Using the well-dated climate deterioration recorded in Ireland circa geochemical survey, conducted between 2008 and 2011, 750 B.C.E. In sites where the palaeoenvironmental and land Thurston has located a number of “invisible” settlements of use records derive from the same cores, however, land numerous periods, addressing the root of many problems within abandonment can be seen to take place prior to the climate Irish archaeology. Fieldwork in Armagh has so far focused event (G. M. Plunkett, Environmental change in the Late on several square kilometers around the civic-ceremonial Bronze Age in Ireland (1200–600 cal. BC), unpublished landscape known as the Navan Complex (Figure 4): a PhD thesis, Queen’s University, Belfast, 1999). The appar- Bronze Age ceremonial enclosure called Haughey’sFort, ent demise of the Irish Bronze Age in the seventh century Navan Fort, or Emain Macha, the ritual center of Ulster’s B.C.E. is another event oftentimes attributed to economic Iron Age elites from about 300 B.C. to A.D. 330, the setting strife in the face of a deteriorating climate [e.g., Raftery, of parts of the “Ulster Cycle” sagas, the town of Armagh, 1997], but other economic variables unrelated to subsistence, where early Christian activity began around A.D. 440 and most notably the introduction of iron technology and realign- which continued on through the Middle Ages and Early ments in trade networks in the wider region, may have played Modern era as a center for both the Catholic and later a far greater role in prompting social and political change at Protestant churches in Ireland. Loughnashade, a pool near this time. Navan Fort used for votive offerings in the Iron Age has 184 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

Figure 4. The Navan Fort within the Navan Civic-Ceremonial Complex. Partial qualitative phosphate map indicating areas of historic and prehistoric settlement. yielded a well-preserved pollen record that can shed light on Navan Fort in the respective periods [Weir, 1993]. This farming and possible demographic change over time. Previ- study also suggested periods in which the landscape was ous work on this site, which was the symbolic “capital” of abandoned, most notably during the Iron Age when wood- Iron Age Ulster (Figure 5), has provided tantalizing evi- land rapidly expanded circa 100 B.C. (D. A. Weir, An dence for intensive farming in the immediate surrounds, environmental history of the Navan area, Co. Armagh, un- which has been equated with the rise of Haughey’s Fort and published PhD thesis, 389 pp, Queen’s University, Belfast, 1993, hereinafter referred to as Weir, unpublished PhD thesis, 1993). Such a feature has been noted in pollen diagrams in other parts of Ireland, and Weir [1995; unpublished PhD thesis, 1993] interprets it as an indication of drastic population decline in the wake of a climate deterioration. Such work has often been carried out before the broad availability of far more accurate AMS 14C dating, so as part of the current project, Loughnashade was re-cored by Plunkett in 2011, revealing strata dated by AMS to the Neolithic through recent times. Preliminary pollen analyses of this new sequence reveal comparable landscape changes to those recorded by Weir (unpublished PhD thesis, 1993). We aim to establish a more reliable, independent chronology for the land use and environmental history of the area through a combined program of tephrochronological and 14C dating of the lake sequence. In Tyrone, our research focuses on the former estates of the Figure 5. Aereal view of Emain Macha, or Navan Fort, Iron Age aristocratic O’Donnelly and O’Hagen families as they ex- component of the Navan Complex. isted in the medieval through early modern eras, where THURSTON AND PLUNKETT 185

Figure 6. (a) The nineteenth century ordinance map superimposed over current orthoimage showing lough and crannog, or artificial island, dating to the Medieval and Early Modern periods. (b) Current orthoimage map with areas of 25 m grid geochemical testing superimposed. Darker areas indicate high levels of phosphate enrichment at lakeshore contemporary with Crannog site. contemporary settlements of client populations were fo- study period as well. Geochemistry reveals dense settlement cused, as well as Iron Age and other prehistoric monuments around its former shoreline as well as settlements scattered that were repurposed and used for ceremonial activities of the across the Ballydonnelly/Castlecaulfield region (Figure 6). Uí Néill kings of Ulster. Project participant Donnelly has In areas surrounding the Medieval Gaelic timber house of the focused on research on the reconstruction of aristocratic O’Donnellys, supplanted by the later Tudor castle, are many landscapes and holdings as well as excavation of important indications of settlement from a variety of periods including Medieval and Early Modern sites in southern Tyrone [Don- those contemporary with the Medieval estate (Figure 7). nelly, 2004; Donnelly and McGranaghan, 2006; Trick et al., Surface artifact collection, performed in the single plowed 2007; Donnelly et al., 2007, 2008]. We have focused our field, which correlates with high levels of phosphate in the geochemical survey around the castle and village of Castle- soil, indicates the likely density of materials at other areas of caulfield where much evidence of Medieval and Early Mod- enriched phosphate, currently invisible with ordinary archae- ern political activity is centered. The area comprised ological reconnaissance methods. Ballydonnelly, the estates of the O’Donellys, the Uí Néill’s military Marshalls [Donnelly et al., 2010]. Caulfield, an 8. CONCLUSIONS English colonial, appropriated Ballydonnelly, renamed it after himself and created, probably from the foundations or We aim ultimately to study the dynamic socioecological remains of the O’Donelly Gaelic timber house, a Tudor-style landscape as both proxy for and impress of sociopolitical castle to protect planter settlers from the Irish. Our pollen processes. In recent years, environmental scholars have ar- coring at Lough Aughlish, a small nearby lake with a Medi- gued that the successful resolution of current global pro- eval/Early Modern crannog, or artificial island usually problems lies partly in their comparative study with past viding residential space for social or political elites, gives responses to climate change and its corresponding environ- well-preserved indications of farming trends: botanical data mental turbulence [e.g., Caseldine and Turney, 2010; Cost- not only for the historic era, but the earlier portion of our anza et al., 2007; DeMenocal, 2001; Dark, 2006; Dugmore 186 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

Figure 7. Area of geochemical testing in Castlecaulfield, Tyrone, Northern Ireland. Darker areas represent highest phosphate, indicating past settlement areas. Archaeological artifact distribution is shown for the surface of the single plowed field in the tested area. Archaeological materials from other indicated sites are invisible under pasture grass. THURSTON AND PLUNKETT 187 et al., 2007; Fisher and Feinman, 2005; Hirschboeck, 2009; acknowledged “autonomous” adaptation, through bottom-up Musacchio, 2009; Redman et al., 2007; Robin and Steffen, decision making on individual and community levels, but 2007; Turney et al., 2006]. Far removed from the mid- stressed long-term administrative forethought as key. Their twentieth century idea that the normal state of nature is suggestion is for a “long-term” perspective: “a couple of “balance,” the last decades have seen the application of decades.” Archaeologists have long poked fun at the idea dynamic nonequilibrium theories [Scoones, 1999; Zimmerer, that “a couple of decades” typical of government plans [e.g., 1994], in which the environment’s normal state is seen as Mortimore, 2010; Weart, 2010] provide a long term perspec- constant flux, and humankind’s normal state is seen as tive, yet in light of our current dilemmas, it is no longer quite constant and creative negotiation of change, crisis, and so funny [Dearing, 2006; Gunderson and Folke,2003;Miller transformation in both natural and social systems. These et al., 2008; Redman and Kinzig, 2003]. interpenetrating chains of processes and events have paral- Recent publications [Crumley, 2006, Gunderson et al., lels in historic and archaeological contexts, some in which 1995; Gunderson and Holling, 2002; Holling, 2001; Janssen politically complex societies have wrought both successful et al., 2006; Peters et al., 2007] refer to the interconnected- and unsuccessful coping strategies. ness of climate and social life: impacts of warming, desert- It is common today to see assessments of twentieth century ification, insect, plant and animal habitat range alterations, climate change that project who will be underwater and who sea level rise, storm increases, and others, which in turn will be waterless in 50 years [Palmer et al., 2004; Rummu- affect food security, resource availability, disease vectors, kainen, 2010; Woth et al., 2006; Xu et al., 2005]. Given the and more, all factoring into subtle or radical alterations in accuracy of current scientific methods, this is a monumental social organization and community structure [Walther et al., yet relatively straightforward enterprise. The archaeologist 2002]. Ultimately, these uncertainties can quickly find ex- and ethnohistorian ask different questions: what relation- pression in economic and political instability, leading to ships exist, for example, between the impacts of climate leadership failures, or creating conditions in which nonlocal change in contemporary Africa and the outcomes of colo- groups can establish primacy during periods of “power vac- nialism between the sixteenth and mid-twentieth centuries? uum” or political ambiguity. Northern Ireland’s past may If African farmers had continued subsistence and local mar- offer another important case study suggesting links between ket agriculture rather than being ruled by a dizzying succes- climate, state, and empire, and the interdependence between sion of foreign colonials, each with a different agenda, and if “rulers” and “ruled.” In studying such contested landscapes, they had not been forced into the monocropping of nonfood we hope to answer more general questions about how chief- commodities for generations, and if they had not traded local doms, kingdoms, and empires instrumentalize natural condi- knowledge of food production for cash to buy European tions to coalesce, make policy, and govern, but also to falter, commodities, how different would food security today be, fragment, and reorganize in cycles of dissolution and resil- even in light of desertification, drought, and catastrophic ience. We offer it as an especially important case study. flooding? Despite attempts by UNESCO and others to bol- ster local economies and governments, the future impact of Acknowledgment. This material is based upon work supported strategies that enriched colonial homelands rather than local by the National Science Foundation under grant 1025654. peoples, will no doubt impact the human ability to manipu- late current climate conditions or adapt to them. REFERENCES In their widely cited article of a decade ago, focusing on adaptation to climate change, Fankhauser et al. [1999] ar- Alexander, J. (1990) Labeur and paresse: Ideological representa- – gued for “planned adaptations” through the adoption of long tions of Medieval peasant labor, Art Bull., 72(3), 436 452. – time horizons in order to create flexible systems to cope with Baillie, M. (1989), Hekla 3: How big was it?, Endeavour, 13(2), 78 81. abrupt or unexpected change. They envisioned an orderly Baillie, M. (1991), Suck in and smear: Two related chronological problems for the 90s, J. Theor. Archaeol., 2,12–16. process, where enormous anticipatory lifestyle changes were Balée, W. (1998), Historical ecology: Premises and postulates, in made far ahead of disaster. Resilience theorists in the natural Advances in Historical Ecology, edited by W. Balée, pp. 13–29, sciences [Folke, 2006; Groffman et al., 2006; Holling, 1973; Columbian Univ. Press, New York. Holling and Gunderson, 2002; Kinzig et al., 2006; Walker et Balée, W. (2006), The research program of historical ecology, Annu. al., 2006] and in archaeology [e.g., Barton et al.,2004; Rev. Anthropol., 35,75–98. Nelson et al., 2006; Peeples et al., 2006; Redman, 2005] Balée, W., and C. L. Erickson (2006), Time, complexity and his- have deconstructed the likelihood of preplanned anticipatory torical ecology, in Time and Complexity in Historical Ecology: change, as humans typically postpone disruptive responses Studies in the Neotropical Lowlands, edited by W. Balée and until they cannot be avoided. Fankhauser et al. [1999, p. 71] C. L. Erickson, pp. 1–17, Columbian Univ. Press, New York. 188 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

Barber, K., B. Zolitschka, P. Tarasov, and A. F. Lotter (2004), Costanza, R., L. Graumlich, W. Steffen, C. Crumley, J. Dearing, K. Atlantic to Urals: The Holocene climatic record of mid-latitude Hibbard, R. Leemans, C. Redman, and D. Schimel (2007), Sus- Europe, in Past Climate Variability Through Europe and Africa, tainability or collapse: What can we learn from integrating the edited by R. W. Battarbee, F. Gasse and C. E. Stickley, pp. 417– history of humans and the rest of nature?, Ambio, 36(7), 522– 442, Springer, New York. 527. Barton, C. M., J. Bernabeu, J. E. Aura, O. Garcia, S. Schmich, and Cronk, L. (1991), Human behavioral ecology, Annu. Rev. Anthro- L. Molina (2004), Long-term socioecology and contingent land- pol., 20,25–53. scapes, J. Archaeol. Method Theory, 11(3), 253–295. Crumley, C. L. (1994), Historical ecology: A multidimensional Bayliss, A., J. van der Plicht, C. Bronk Ramsey, G. McCormac, F. ecological orientation, in Historical Ecology Cultural Knowledge Healy, and A. Whittle (2011), Towards generational time-scales: and Changing Landscapes, edited by C. Crumley, pp. 1–16, Sch. The quantitative interpretation of archaeological chronologies, in of Am. Res. Press, Santa Fe, N. M. Gathering Time: Dating the Early Neolithic Enclosures of South- Crumley, C. L. (2000), From garden to globe: Linking time and ern Britain and Ireland, edited by A. Whittle, F. Healy and A. space with meaning and memory, in The Way the Wind Blows: Bayliss, pp. 17–59, Oxbow Books, Oxford, U. K. Climate, History, and Human Action, edited by R. J. McIntosh, Bender, B. (1992), Theorising landscapes, and the prehistoric land- J. A. Tainter and S. K. McIntosh, pp. 193–208, Columbia Univ. scapes of Stonehenge, Man, 27(4), 735–755. Press, New York. Bender, B. (1993), Landscape: Politics and Perspectives, Berg, Crumley, C. L. (2006), Archaeology in the new world order: What Oxford, U. K. we can offer the planet, in Space and Spatial Analysis in Archae- Berglund, B., M. Hjelmroos, and E. Kolstrup (1991), The Köpinge ology, edited by C. E. Robertson et al., pp. 383–396, Univ. of Area: Vegetation and landscape through time, in The Cultural Calgary Press, Calgary, Alberta, Canada. Landscape During 6000 Years in Southern Sweden: The Ystad Crumley, C. L. (2007), Historical ecology: Integrated thinking at Project, edited by B. Berglund, pp. 109–112, Munksgaard Int., multiple temporal and spatial scales, in The World System and Copenhagen, Denmark. the Earth System. Global Socioenvironmental Change and Berglund, B. E. (2003), Human impact and climate changes – Sustainability Since the Neolithic, edited by A. Hornborg and Synchronous events and a causal link?, Quat. Int., 105(1), 7–12. C. L. Crumley, pp. 15–28, Left Coast Press, Walnut Creek, Boone, J. L., and E. A. Smith (1998), Is it evolution yet? A critique Calif. of evolutionary archaeology, Curr. Anthropol., 39(S1), 141–174. Crumley, C. L., and W. H. Marquardt (1990), Landscape: A unify- Bradley, R. (2007), The Prehistory of Britain and Ireland, Cam- ing concept in regional analysis, in Interpreting Space: GIS bridge Univ. Press, Cambridge, U. K. and Archaeology, edited by K. Allen, S. Green and E. Zubrow, – Bradwell, T., et al. (2008), The northern sector of the last British Ice pp. 73 79, Taylor and Francis, London, U. K. Sheet: Maximum extent and demise, Earth Sci. Rev., 88,207–226. Dark, P. (2006), Climate deterioration and land-use change in the ﬁ Butzer, K. W. (1982), Archaeology as Human Ecology: Method and rst millennium BC: Perspectives from the British palynological – Theory for a Contextual Approach, Cambridge Univ. Press, Cam- record, J. Archaeol. Sci., 33(10), 1381 1395. bridge, U. K. Dearing, J. (2006), Climate-human-environment interactions: Re- – Byrne, F. J. (2001), Irish Kings and High-Kings, Four Courts Press, solving our past, Clim. Past Discuss., 2(4), 187 203. Dublin. Deetz, J. (1990), Landscapes as cultural statements, in Earth Pat- Caseldine, C. J., and C. Turney (2010), The bigger picture: Towards terns: Essays in Landscape Archaeology, edited by W. M. Kelso – integrating palaeoclimate and environmental data with a history and R. Most, pp. 1 4, Univ. Press of Va., Charlottesville. of societal change, J. Quat. Sci., 25(1), 88–93. DeMenocal, P. B. (2001), Cultural responses to climate change – Chapin, F. S., C. L. Crumley, C. P. Gomes, T. E. Graedel, J. Levin, during the late Holocene, Science, 292(5517), 667 673. P. A. Matson, K. Matus, S. Myers, and V. K. Smith (2010), Donnelly, C. (2004), Masshouses and meetinghouses: The archae- Managing human-environment systems for sustainability, in To- ology of the penal laws in early modern Ireland, Int. J. Hist. ward a Science of Sustainability, edited by S. A. Levin and W. C. Archaeol., 8(2), 119–132. Clark, CID Working Pap. 196, pp. 58–73, Cent. for Int. Dev. at Donnelly, C., and C. McGranaghan (2006), O’Connor’s Strong- Harvard Univ., Cambridge, Mass. hold, Derrywarragh Island, Maghery, County Armagh, Ulster J. Charman, D. J. (2007) Summer water deﬁcit variability controls on Archaeol., 63, 123–129. peatland water-table changes: Implications for Holocene palaeo- Donnelly, C., E. Murray, and P. Logue (2007), Excavating with climate reconstructions, The Holocene, 17, 217–227. Time Team at Castle Hill, Dungannon, Co. Tyrone, Archaeol. Charman, D. J., A. D. Brown, D. Hendon, and E. Karofeld (2004), Ireland, 21(4), 16–19. Testing the relationship between Holocene peatland palaeocli- Donnelly, C., E. Murray, and R. McHugh (2008), Dungannon mate reconstructions and instrumental data at two European sites, Castle: Its history, architecture and archaeology, Dúiche Néill, Quat. Sci. Rev., 23(1–2), 137–143. 17,11–24. Conway, M., A. Gahan, and S. Rathbone (2005), Corrstown: A Donnelly, C., S. McGivern, and G. McCabe (2010), Ballydonnelly large Middle Bronze Age village, Curr. Archaeol., 195, 120–123. and the estates of the O’Neill Lught Tighe in Late Medieval THURSTON AND PLUNKETT 189

Tyrone, paper presented at the 10th Annual Conference, Irish Gremillion, K. J. (2002), Foraging theory and hypothesis testing in Post-Medieval Archaeol. Group, Belfast, U. K., 5–6 Feb. archaeology: An exploration of methodological problems and Dugmore, A. J., D. M. Borthwick, M. J. Church, A. Dawson, K. J. solutions, J. Anthropol. Archaeol., 21(2), 142–164. Edwards, C. Keller, P. Mayewski, T. H. McGovern, K. A. Mairs, Groffman, P. M., J. S. Baron, T. Blett, A. J. Gold, I. Goodman, L. H. and G. Sveinbjarnardöttir (2007), The role of climate in settle- Gunderson, B. M. Levinson, M. A. Palmer, H. W. Paerl, and ment and landscape change in the North Atlantic islands: An G. D. Peterson (2006), Ecological thresholds: The key to successful assessment of cumulative deviations in high-resolution proxy environmental management or an important concept with no prac- climate records, Hum. Ecol., 35(2), 169–178. tical application?, Ecosystems, 9(1), 1–13. Dunnell, R. C. (1992), The notion site, in Space, Time, and Archae- Gunderson, L., and C. Folke (2003), Toward a science of the long ological Landscapes, edited by J. Rossignol and L. A. Wandsni- view, Conserv. Ecol., 7(1), 15. der, pp. 21–24, Plenum Press, New York. Gunderson, L. H., and C. S. Holling (2002), Panarchy: Under- Erickson, C. L. (1993), The social organization of prehispanic standing Transformations in Human and Natural Systems, edited raised field agriculture in the Lake Titicaca Basin, in Economic by L. H. Gunderson and C. S. Holling, Island Press, Washington, Aspects of Water Management in the Prehistoric New World, D. C. edited by V. Scarborough and B. Isaacs, pp. 367–424, JAI Press, Gunderson, L. H., C. Holling, and S. S. Light (1995), Barriers and Greenwich, Conn. Bridges to the Renewal of Ecosystems and Institutions, Columbia Erickson, C. L. (1999), Neo-environmental determinism and agrar- Univ. Press, New York. ian “collapse” in Andean prehistory, Antiquity, 73, 634–642. Hall, V. A. (2003), Vegetation history of mid-to western Ireland in Fankhauser, S., J. B. Smith, and R. S. J. Tol (1999), Weathering the 2nd millennium AD: Fresh evidence from tephra-dated paly- climate change: Some simple rules to guide adaptation decisions, nological investigations, Veg. Hist. Archaeobot., 12(1), 7–17. Ecol. Econ., 30(1), 67–78. Hall, V. A. (2005), The vegetation history of monastic and secular Finlayson, A., J. Merritt, M. Browne, J. Merritt, A. McMillan, and sites in the midlands of Ireland over the last two millennia, Veg. K. Whitbread (2010), Ice sheet advance, dynamics, and decay Hist. Archaeobot., 15,1–12. configurations: Evidence from west central Scotland, Quat. Sci. Hall, V. A., and L. Bunting (2000), Tephra-dated pollen studies of Rev., 29, 969–988. medieval landscapes in the north of Ireland, in Gaelic Ireland c. Fisher, C. T., and G. M. Feinman (2005), In focus: Landscapes over 1250–1650: Land, Lordship and Settlement, edited by P. J. Duffy, time: Resilience, degradation and contemporary lessons, Am. D. Edwards and E. FitzPatrick, pp. 207–222, Four Courts Press, Anthropol., 107(1), 62–69. Dublin, Ireland. FitzPatrick, E. (2004), Royal Inauguration in Gaelic Ireland c. Hall, V. A., and J. R. Pilcher (2002), Late-Quaternary Icelandic 1100–1600: A Cultural Landscape Study, Boydell Press, Wood- tephras in Ireland and Great Britain: Detection, characterization bridge, U. K. and usefulness, The Holocene, 12(2), 223–230. Folke, C. (2006), Resilience: The emergence of a perspective for Hirschboeck, K. K. (2009), Future hydroclimatology and the re- social-ecological systems analyses, Global Environ. Change, 16 search challenges of a post-stationary world, J. Contemporary (3), 253–267. Water Res. Educ., 142(1), 4–9. Gallet, Y., A. Genevey, M. Le Goff, F. Fluteau, and S. Ali Eshraghi Holling, C. S. (1973), Resilience and stability of ecological sys- (2006), Possible impact of the Earth’smagneticfield on the tems, Annu. Rev. Ecol. Syst., 4,1–23. history of ancient civilizations, Earth Planet. Sci. Lett., 246(1– Holling, C. S. (2001), Understanding the complexity of economic, 2), 17–26. ecological, and social systems, Ecosystems, 4(5), 390–405. Gearey, B. R., P. Marshall, and D. Hamilton (2009), Correlating Holling, C. S., and L. H. Gunderson (2002), Resilience and adaptive archaeological and palaeoenvironmental records using a Bayes- cycles, in Panarchy: Understanding Transformations in Human ian approach: A case study from Sutton Common, South York- and Natural Systems, edited by L. H. Gunderson and C. S. shire, England, J. Archaeol. Sci., 36, 1477–1487. Holling, pp. 25–62, Island Press, Washington, D. C. Glick, D., F. Montaigne, and V. Morell (2004), Signs from Earth: Huang, C. C., and H. Su (2009), Climate change and Zhou reloca- The big thaw: The climate is changing at an unnerving pace. tions in early Chinese history, J. Hist. Geogr., 35(2), 297–310. Glaciers are retreating, ice shelves are fracturing, sea level is Hubbard, A., T. Bradwell, N. Golledge, A. Hall, H. Patton, D. rising, permafrost is melting. What role do humans play?, Natl. Sugden, R. Cooper, and M. Stoker (2009), Dynamic cycles, ice Geogr., September, 12–15. streams and their impact on the extent, chronology and degla- Green, S., and M. Zvelebil (1990), The Mesolithic colonization and ciation of the British-Irish ice sheet, Quat. Sci. Rev., 28(7–8), agricultural transition of south-east Ireland, Proc. Prehist. Soc., 758–776. 56,57–88. Ingold, T. (1996), The optimal forager and economic man, in Nature Greenwood, S. L., and C. D. Clark (2009) Reconstructing the and Society, edited by P. Descola and G. Pålsson, pp. 25–44, last Irish Ice Sheet 2: A geomorphologically-driven model of Routledge, London, U. K. ice sheet growth, retreat and dynamics, Quat. Sci. Rev., 28, Janssen, M. A., Ú. Bodin, J. M. Anderies, T. Elmqvist, H. Ernstson, 3101–3123. R. R. J. McAllister, P. Olsson, and P. Ryan (2006), Toward a 190 DYNAMIC HUMAN SOCIOECOLOGY OF PREHISTORIC AND PROTOHISTORIC ULSTER

network perspective of the study of resilience in social-ecological zation: A case study from the Mimbres region of the US South- systems, Ecol. Soc., 11(1), 15. west, Am. Antiquity, 71(3), 403–432. Kerr, T. R. (2009), The height of fashion: Raised raths in the Newman, C., M. O. O’Connell, M. Dillon, and K. Molloy (2007), landscape of north-west Ulster, J. Irish Archaeol., 18,63–75. Interpretation of charcoal and pollen data relating to a late Iron Kerr, T. R., G. T. Swindles, and G. Plunkett (2009), Making hay Age ritual site in eastern Ireland: A holistic approach, Veg. Hist. while the sun shines? Socio-economic change, cereal production Archaeobot., 16(5), 349–365. and climatic deterioration in Early Medieval Ireland, J. Archaeol. Oram, R., and W. P. Adderley (2008), Lordship and environmental Sci., 36(12), 2868–2874. change in Central Highland Scotland c. 1300–c. 1400, J. North Kinzig, A. P., P. A. Ryan, M. Etienne, H. E. Allison, T. Elmqvist, Atl., 1(1), 74–84. and B. H. Walker (2006), Resilience and regime shifts: Assessing Palmer, T., U. Andersen, P. Cantelaube, M. Davey, M. Deque, cascading effects, Ecol. Soc., 11(1), 20. F. Doblas-Reyes, H. Feddersen, R. Graham, S. Gualdi, and Knight, J. (2003), Geomorphic and sedimentary evidence for pat- J. F. Gueremy (2004), Development of a European multi-model terns of late Midlandian ice retreat in the Tempo Valley, north- ensemble system for seasonal to inter-annual prediction central Ireland, Irish Geogr., 36(2), 127–144. (DEMETER), Bull. Am. Meteorol. Soc., 85(6), 853–872. Lock, G., and B. L. Molyneaux (2007), Introduction, in Confront- Pauketat, T. R. (2004), Archaeology without Alternatives, Anthro- ing Scale in Archaeology: Issues of Theory and Practice, edited pol. Theory, 4(2), 199–203. by G. Lock and B. L. Molyneaux, pp. 1–11, Springer, New York. Peeples, M. A., C. M. Barton, and S. Schmich (2006), Resilience Lomas-Clarke, S. H., and K. E. Barber (2004), Palaeoecology of lost: Intersecting land use and landscape dynamics in the prehis- human impact during the historic period: Palynology and geo- toric southwestern United States, Ecol. Soc., 11(2), 22. chemistry of a peat deposit at Abbeyknockmoy, Co. Gaiway, Peters, D. P. C., B. T. Bestelmeyer, and M. G. Turner (2007), Ireland, The Holocene, 14(5), 721–731. Cross-scale interactions and changing pattern-process relation- McCabe, A. M., and P. U. Clark (2003), Deglacial chronology from ships: Consequences for system dynamics, Ecosystems, 10(5), County Donegal, Ireland: Implications for deglaciation of the 790–796. British-Irish ice sheet, J. Geol. Soc. London, 160(6), 847–855. Pilcher, J. R., V. A. Hall, and F. G. McCormac (1996), An outline McCabe, A. M., P. U. Clark, and J. Clark (2005), AMS 14C dating tephrochronology for the Holocene of the north of Ireland, J. of deglacial events in the Irish Sea Basin and other sectors of the Quat. Sci., 11, 485–494. British–Irish ice sheet, Quat. Sci. Rev., 24, 1673–1690. Plunkett, G. (2006), Hekla 3, environmental downturn and Irish McCabe,A.M.,P.U.Clark,J.Clark,andP.Dunlop(2007), Late Bronze Age hillfort connections revisited, Emania, 20, Radiocarbon constraints on readvances of the British–Irish Ice 55–60. Sheet in the northern Irish Sea Basin during the last deglaciation, Plunkett, G. (2009), Land-use patterns and cultural change in the Quat. Sci. Rev., 26, 1204–1211. Middle to Late Bronze Age in Ireland: Inferences from pollen McCormick, F. (1992), Early faunal evidence for dairying, Oxford records, Veg. Hist. Archaeobot., 18(4), 273–295. J. Archaeol., 11(2), 201–210. Plunkett, G. M., J. R. Pilcher, F. G. McCormac, and V. A. Hall McCormick, F. (2008), The decline of the cow: Agricultural and (2004), New dates for first millennium BC tephra isochrones in settlement change in early medieval Ireland, Peritia, 20(1), Ireland, The Holocene, 14(5), 780–786. 209–224. Raftery, B. (1997), Pagan Celtic Ireland: The Enigma of the Irish McDermott, F., D. P. Mattey, and C. Hawkesworth (2001), Iron Age, Thames and Hudson, London, U. K. Centennial-scale Holocene climate variability revealed by a Redman, C. L. (2005), Resilience theory in archaeology, Am. high-resolution Speleothem δ18O record from SW Ireland, Anthropol., 107(1), 70–77. Science, 294, 1328–1331. Redman, C. L., and A. P. Kinzig (2003), Resilience of past land- Miller, T. R., T. D. Baird, C. M. Littlefield, G. Kofinas, F. S. Chapin scapes: Resilience theory, society, and the longue durée, Conserv. III, and C. L. Redman (2008), Epistemological pluralism: Reor- Ecol., 7(1), 14. ganizing interdisciplinary research, Ecol. Soc., 13(2), 46–62. Redman, C. L., C. Crumley, F. A. Hassan, F. Hole, J. Morais, F. Mortimore, M. (2010), Adapting to drought in the Sahel: Lessons Riedel, V. L. Scarborough, J. A. Tainter, P. Turchin, and Y. for climate change, WIREs Clim. Change, 1(1), 134–143. Yasuda (2007), Group report: Millennial perspectives on the Murphy, E. M., and N. J. Whitehouse (2010), Environmental Ar- dynamic interaction of climate, people, and resources, in Sustain- chaeology in Ireland, Oxbow, Oxford, U. K. ability or Collapse? An Integrated History and Future of People Musacchio, L. R. (2009), The scientific basis for the design of on Earth, edited by R. Costanza, L. J. Graumlich and W. Steffen, landscape sustainability: A conceptual framework for translation- pp. 115–148, MIT Press, Boston, Mass. al landscape research and practice of designed landscapes and the Richardson, G. (2001) A tale of two theories: Monopolies and craft six Es of landscape sustainability, Landscape Ecol., 24(8), 993– guilds in Medieval England and modern imagination, J. Hist. 1013. Econ. Thought, 23(2), 232–238. Nelson, M. C., M. Hegmon, S. Kulow, and K. G. Schollmeyer Robin, L., and W. Steffen (2007), History for the Anthropocene, (2006), Archaeological and ecological perspectives on reorgani- Hist. Compass, 5(5), 1694–1719. THURSTON AND PLUNKETT 191

Rummukainen, M. (2010), State-of-the-art with regional climate Walther, G. R., E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. C. models, WIREs Clim. Change, 1(1), 82–96. Beebee, J. M. Fromentin, O. Hoegh-Guldberg, and F. Bairlein Schettler, G., R. L. Romer, M. O’Connell, and K. Molloy (2006), (2002), Ecological responses to recent climate change, Nature, Holocene climatic variations and postglacial sea-level rise geo- 416(6879), 389–395. chemically recorded in the sediments of the brackish karst lake Warrick, R. A., and W. E. Riebsame (1981), Societal response to An Loch Mor, western Ireland, Boreas, 35(4), 674–693. CO2-induced climate change: Opportunities for research, Clim. Scoones, I. (1999), New ecology and the social sciences: What Change, 3(4), 387–428. prospects for a fruitful engagement?, Annu. Rev. Anthropol., 28, Weart, S. R. (2010), The idea of anthropogenic global climate 479–507. change in the 20th century, WIREs Clim. Change, 1(1), 67–81. Sejrup, H. P., B. O. Hjelstuen, K. I. T. Dahlgren, H. Haflidason, A. Weir, D. (1995), A palynological study of landscape and agricul- Kuijpers, A. Nygard, D. Praeg, M. S. Stoker, and T. O. Vorren tural development in County Louth from the second millennium (2005), Pleistocene glacial history of the NW European conti- BC to the first millennium AD, Discovery Programme Rep., 2, nental margin, Mar. Petrol. Geol., 22, 1111–1129. 77–126. Stothers, R. B. (2000), Climatic and demographic consequences of Whitehead, N. (1998), Ecological history and historical ecology: the massive volcanic eruption of 1258, Clim. Change, 45(2), Diachronic modeling versus historical explanation, in Advances 361–374. in Historical Ecology, edited by W. Balée, pp. 201–214, Colum- Stout, M. (1997), The Irish Ringfort, Four Courts Press, Dublin, bian Univ. Press, New York. Ireland. Winterhalder, B., and E. A. Smith (2000), Analyzing adaptive Swindles, G. T., G. Plunkett, and H. M. Roe (2007a), A delayed strategies: Human behavioral ecology at twenty-five, Evol. climatic response to solar forcing at 2800 cal. BP: Multi-proxy Anthropol. Issues News Rev., 9(2), 51–72. evidence from three Irish peatlands, The Holocene, 17, 177–182. Woth, K., R. Weisse, and H. von Storch (2006), Climate change and Swindles, G. T., G. Plunkett, and H. M. Roe (2007b), A multiproxy North Sea storm surge extremes: An ensemble study of storm climate record from a raised bog in County Fermanagh, Northern surge extremes expected in a changed climate projected by four Ireland: A critical examination of the link between bog surface different regional climate models, Ocean Dyn., 56(1), 3–15. wetness and solar variability, J. Quat. Sci., 22(7), 667–679. Xu, C., E. Widén, and S. Halldin (2005), Modelling hydrological Swindles, G. T., A. Blundell, H. M. Roe, and V. A. Hall (2010), A consequences of climate change: Progress and challenges, Adv. 4500-year proxy climate record from peatlands in the North of Atmos. Sci., 22(6), 789–797. Ireland: The identification of widespread summer “drought Young, K. R., and J. K. Lipton (2006), Adaptive governance and phases?”, Quat. Sci. Rev., 29(13–14), 1577–1589. climate change in the tropical highlands of western South Amer- Trick, S., C. Donnelly, and R. McHugh (2007), A geophysical ica, Clim. Change, 78(1), 63–102. survey at Tullaghoge Fort, The Bell, 11(11), 16–27. Zimmerer, K. S. (1994), Human geography and the “New ecology:” Turney, C. S. M., M. Baillie, J. Palmer, and D. Brown (2006), The prospect and promise of integration, Ann. Assoc. Am. Geogr., Holocene climatic change and past Irish societal response, J. 84(1), 108–125. Archaeol. Sci., 33(1), 34–38. Valante, M. A. (2008), The Vikings in Ireland: Settlement, Trade G. Plunkett, School of Geography, Archaeology and Palaeo- and Urbanization, Four Courts Press, Dublin, Ireland. ecology, Queen’s University, Palaeoecology Centre, Room 01035, Walker, B. H., J. M. Anderies, A. P. Kinzig, and P. Ryan (2006), Belfast BT7 1NN, UK. ([email protected]) Exploring resilience in social-ecological systems through com- T. Thurston, Department of Anthropology, University at Buffalo, parative studies and theory development: Introduction to the State University of New York, 380 MFAC, Ellicott Complex, special issue, Ecol. Soc., 11(1), 12. Buffalo NY 14261, USA. ([email protected]) Influences of Active Tectonism on Human Development: A Review and Neolithic Example

Eric R. Force

Department of Geosciences, University of Arizona, Tucson, Arizona, USA

Bruce G. McFadgen

School of Maori Studies, Victoria University of Wellington, Wellington, New Zealand

Two groups of researchers have independently studied the effects of active tectonism on human development. The first group focuses on tectonically influenced topography in providing varied food sources, security, natural hunting traps, and water resources through Paleolithic times. The second group suggests that active tectonism forced the pace of cultural change in antiquity, accelerating the development of cultural complexity in comparison to neighbors in tectonically quiescent areas. Economic, political, religious, and other modern evidence indicates that this may still be the case. Similarly, the effects of tectonically influenced topography can be traced into ancient agricultural societies and into some segments of the modern world. The effects of topography were particularly important in preurban stages of human development, and effects of the forced pace of cultural change were generally later and centered on towns. Potentially interesting overlaps in the Neolithic are addressed with a look at the Maori of New Zealand, who altered their way of life in response to earthquakes and consequent tsunami in the fifteenth century. The complexity of this example suggests that unidirectional processes should not be expected at the level of individual cultural entities. Many lines of evidence, however, show that tectonic activity should be considered along with other factors such as climate change in evaluating human activity and development.

1. INTRODUCTION tectonism can have and the interactions that these influences can show. An example from New Zealand shows how com- Recent literature presents two hypotheses that both suggest plex and unpredictable a response can be. positive reinforcement of the development of our species and In their work over more than a decade, King, Bailey, and its cultures by active tectonism. It might appear that the two their coauthors [e.g., Bailey et al., 1993, 2011; King et al., hypotheses have no common ground, despite the direction 1994; King and Bailey, 2006, 2010; Bailey and King, 2010] they share. This paper is an attempt to integrate the hypoth- have emphasized the influence that active tectonism can have eses, in order to show the variety of influences that active in providing diverse topographic environments and, hence, diverse resources. Humanity, in its various stages of development, has preferentially taken advantage of these resources, in Climates, Landscapes, and Civilizations comparison to more quiescent tectonic environments with Geophysical Monograph Series 198 less diversity. These resources include basins periodically © 2012. American Geophysical Union. All Rights Reserved. rejuvenated by tectonic activity, with sedimentation and 10.1029/2012GM001215 well-watered fertility; barriers created by tectonic activity that 195 196 INFLUENCES OF ACTIVE TECTONISM ON HUMAN DEVELOPMENT offer tactical advantage to human predators dependent on well described [see Sheets and Grayson, 1979; Fisher et al., mobile and elusive prey; and ongoing creation of a landscape 1997; Balmuth et al., 2005; Gratton and Torrence, 2008]. that is not only topographically complex, but often also These lines of evidence suggest that our cultural develop- ecologically complex, offering a variety of resources. The ment scenario, although developed for the period of antiquity, advantages resulting from tectonic activity thus include var- is not limited to a particular time period. Remote images of the ied food sources, security, natural hunting traps, and water world at night show that tectonically active plate boundaries resources. King and Bailey [2006] explored the implications are still loci for our own peculiar type of cultural complexity; of their hypothesis in delineating probable early human indeed, they suggest that the main locus has shifted from the migration routes. Tethyan to the circum-Pacific realm. The work of King, Bailey, and colleagues has so far focused on three main areas and times, the late Pleistocene 2. A SIMPLE INTEGRATION of northern Greece (Paleolithic), and the Pliocene of South BASED ON TIME PERIODS Africa and the Afar Triangle. We note that many other areas are pertinent for testing their hypothesis. Since workers A marriage of the two scenarios could have them applica- elsewhere have suggested similar themes [e.g., Molnar, ble in different stages of development of our species, and we 1990], this seems promising. The argument could be think that this simple solution is a valid part of a general strengthened with description of converse cases. We suggest integration. The topographically based model would be pow- below that the tectonic landscape approach can be extended erfully applicable in the hunter-gatherer stages of develop- in time outside the pre-Homo sapiens and Paleolithic interval ment, as it “accentuated and maintained topographic barriers that the King-Bailey group describes. that could be used to tactical advantage in the monitoring, Our previous work [Force, 2008; Force and McFadgen, trapping, and control of mobile prey species” [King and 2010] addressed the more recent period of classical antiquity, Bailey, 2010]. We can see that in the transition to agricultural beginning with transitions into the Bronze Age, (i.e., about cultures, tectonic topography would still be of great impor- 3500 B.C. to A.D. 500, or 4000 years), and increasingly, it tance, as it forms discrete basins with arable floors and has focused on the impact of active tectonism on mankind’s internal reservoirs, margined by highlands with exceptionally cultural development. This work began with the counterin- productive springs [cf. Jackson,2006],partlyduetoasetof tuitive observation that “great ancient civilizations” tended to fractures held open by stress [Force, 2008], and with erosion originate along active plate boundaries [Force, 2008, Fig- of fresh rocks contributing nutrients. ure 1], and we subsequently developed several other indica- Our tectonic boundary-culture model is most clearly ap- tions that pointed to tectonism forcing the pace of cultural plied to the period of classical antiquity, when required change, eventually resulting in greater cultural complexity rebuilding and reordering of urban areas as a result of tec- along active tectonic boundaries. These indications are tonic activity was first recorded. In some of the late Neolithic (1) continent-scale perspectives showing complexity of an- cultures that preceded this period, these requisites were al- cient cultures decreasing away from these boundaries; (2) a ready met and show unusually close spatial relations to measure of the relative stasis of cultures, which increases away tectonic boundaries [Force and McFadgen, 2010]. from tectonic boundaries [Force, 2008, Figure 2]; (3) the propagation of trade routes, including island hopping, along 3. A SECOND STEP EMPHASIZING DYNAMICS tectonic boundaries [Force and McFadgen, 2010, Figure 2]; AND TIME OVERLAPS (4) the many cultural discontinuities corresponding to ancient tectonic events [Force and McFadgen, 2010], especially A more comprehensive synthesis would address different religious discontinuities [Soren and James, 1988; Rothaus, dynamics linking tectonism and human development, and 1996; deBoer and Sanders, 2005]; and (5) evidence that focus on the time overlaps that follow. For example, even at religious and intellectual traditions molded in part by tecto- the height of the classical era in Greece when and where nism in antiquity are still being so influenced in the modern tectonically driven cultural acceleration was most apparent world. Tectonic events still force the pace of cultural change (Euripides punctuating his dramas with earthquakes, Herod- [Kant, 1790; James, 1906; Skidmore and Toya, 2002; Davis, otus and Thucydides differing on which earthquake more 2005; Vale and Campanella, 2005a, 2005b; Rozario, 2007; significantly divided history into epochs, temples incorpo- Cuaresma et al., 2008], as do other natural disasters (reviews rating increasingly innovative systems to prevent earthquake and compilations by Oliver-Smith and Hoffman [1999], Rey- damage, even as whole cites like Helike are downthrown craft and Bawden [2000], and Hoffman and Oliver-Smith beneath the waves), aspects of tectonic topography were still [2002]); cultural responses to volcanic events are particularly of importance. For example, the poet Hesiod (in “Works and FORCE AND MCFADGEN 197

Days” of about 650 B.C.), figuratively manning his plow in fifteenth century tsunami inundations were a Neolithic Boeotia, noted “in the mountain valleys, fertile lands.” This equivalent of a “creative destruction” event [Schumpeter, area in its tectonically formed basin provided an agricultural 1942; Rozario, 2007] that not only destroyed physical ele- background of subsistence for ancient Attica. Nearby, the ments of Maori culture, but through the deaths of elders, tectonic topography of the Gulf of Corinth provided the depleted their knowledge base. Maori society nevertheless potential for maritime wealth. In modern cultures based on recovered, but to what extent the creative outcomes of later agriculture in small areas, tectonic topography is still impor- Maori culture were a carry-on from earlier times, and what tant, as Jackson [2006] has shown in southern Persia. were directly stimulated by the events, requires further re- There is no reason to think that tectonism did not promote search to answer. What is clear, however, is that tectonic cultural innovation during the Neolithic and earlier, although events can have effects on the landscape that reduce the tests for this proposition are more difficult to perform. In the resources that are available, as well as enhance them, and early Neolithic, there are already indications that the spread over a period of time, the one event can have both outcomes. of village-based agricultural systems, complete with ideolog- In the northern part of the North Island (Figure 1), the ical trappings, tended to follow tectonic plate boundaries in landscape is dominated by volcanic activity [Lowe, 2010], the Levant and the northern margin of the “Fertile Crescent,” which provided the first settlers with good resources, espe- quite independently of obsidian sources (e.g., distributions cially rocks for making tools and soils for growing crops. described by Cauvin [2000]). Volcanic activity during the Maori period was generally Thus, the Neolithic period, and places and periods where minor [McFadgen, 2007]; the largest eruption of any real similar lifeways prevailed, appears to be of special interest as consequence to Maori, the Kaharoa, was close to the time of stages of probable overlap of the two basic dynamics pro- Maori settlement [Hogg et al., 2003]. The Kaharoa Tephra posed by the two groups of investigators. Were cultures Formation [Froggatt and Lowe, 1990] extends from the becoming more innovative at the same time that they were Okataina Volcanic Centre well into Northland, but only taking advantage of local tectonic topography? around the Bay of Plenty was it thick enough to kill vegetation and animals [Lowe et al., 2002]. It would have left the 4. A CASE FROM NEW ZEALAND Bay of Plenty landscape desolate, which probably explains why settlement of this part of New Zealand occurred later An opportunity to address this sort of question is provided than in adjacent areas. But the tephra was light and sandy, by cultures that retained Neolithic ways of life into recent and it generally fell on a heavier sandy loam soil. When the historic times. One of us [McFadgen, 2007] has studied the Maori occupied the Bay of Plenty landscape, they found that response of the Maori to New Zealand’s tectonic environ- the two soils, when mixed together, were most favorable ment. This example is probably one of the most detailed physically for growing kumara (sweet potato, an important available responses of a Neolithic culture to tectonic activ- food source) [Gibbs and Pullar, 1961], and these mixed soils ity. In addition, the unusual isolation of Maori culture in the Bay of Plenty form one of the largest areas of prehis- allows some focus on the local cultural response to physical toric garden soil anywhere in New Zealand. change alone without imported response lessons; some re- For the remainder of the North Island and South Island sults including ethnographic memories were still present and (Figure 1), effects of seismic activity dominate the landscape. recorded at European contact. This case shows, however, that Here a long history of active tectonism has resulted in a the long-term results of tectonism can be complex and distinctive landscape of tall, rugged, and steep mountains, unpredictable by simple models. Any one example, of hills, and ridges highly susceptible to slope failure, with course, cannot stand in for the range of responses to tecto- alluvial plains and fast-flowing rivers, bounded at the coast nism in cultures that show great variety and occupy different by sandy beaches and rocky shores. Faults rupture the land- environments. scape. The most notable is the Alpine Fault, a strike-slip fault New Zealand straddles the obliquely convergent tectonic with a displacement and length of more than 400 km, with boundary in the SW Pacific between the Australian and many major splinter faults to the northeast. The east coast of Pacific plates, some 1500 km east of Australia. It was first the North Island and northern South Island is generally being settled by Polynesians, the ancestors of the Maori, less than uplifted; to the west, between Taranaki and the northern 750 years ago, and rediscovered and settled by Europeans South Island, the land is generally subsiding (Figure 1). only about 200 years ago. Its prehistory, although short, was Major earthquakes drive environmental change [Goff and nevertheless punctuated by significant cultural change that McFadgen, 2002] in this part of New Zealand. Their imme- appears to have been strongly influenced by tectonic events. diate outcomes are landslips and, in the right circumstances, In economic and social terms, it is suggested here that the late tsunamis. Steep slopes may be denuded of soil and vegetation 198 INFLUENCES OF ACTIVE TECTONISM ON HUMAN DEVELOPMENT

Figure 1. Map of New Zealand showing dominant tectonic inﬂuences. Approximate boundary is shown by a dashed line: northwest of the line, the dominant inﬂuence is volcanic activity (V); for the rest of the country, the dominant activity is seismic activity: uplifted shorelines (U) and subsided shorelines (S). OVC is Okataina Volcanic Centre.

[e.g., Grapes and Downes, 1997], which gets washed into tion flooded settlements, enlarged lagoons and estuaries, and streams and rivers and eventually reaches the sea. Delayed caused changes to shellfish habitat. Finally, advancing dunes outcomes are the formation of aggradational terraces by rivers and accumulating river sediments buried settlements and and the accumulation of sand dunes in coastal areas. Within gardens. Earthquake activity, nevertheless, enabled the for- 2 years, half the sediment released by landslips can reach the mation of friable soils on uplifted shoreline deposits, stream coast; it is mostly sand, the gravel comes later [Adams, 1980]. fans, and river terraces and exposed hard, fine-grained meta- Within a few decades, new coastal dunes become stable and morphic rocks for tool manufacture. Earthquake shaking, in vegetated [Wells and Goff, 2006]. itself, was not directly destructive for Maori, as they had no Uplift, subsidence, dune building, and river aggradation all masonry buildings. Yet the indirect effects of earthquakes had their effect on Maori communities [McFadgen, 2007]. were important. Maori occupation was primarily oriented to the coast. Seis- The fifteenth century was the most seismically active pe- mic uplift drained lagoons and destroyed habitat, material riod during Maori occupation [McFadgen, 2007]. Sedimen- resources, and food supplies, while subsidence and compac- tary deposits along the northern South Island and North FORCE AND MCFADGEN 199

Island coasts record one or more large tsunamis that are The outcome of a major tsunami inundation is necessarily thought to have struck in the late fifteenth century. The waves speculative, but based on the effects of recent modern were highest in the north, where they washed to more than tsunamis such as the 2004 Boxing Day event in Indonesia, 30 m above sea level and traveled more than 800 m inland a reasonable scenario can be put together that accounts for [Nichol et al., 2003a, 2003b, 2004]. In the northern South the changes observed. There would have been extensive loss Island and southern North Island, the waves were somewhat of resources: food stores, shellfish beds, and gardens (smoth- lower, probably around 10 m high. Large waves are men- ered with sand and contaminated with seawater). Coastal tioned in Maori traditions, but the references are often ob- settlements would have become uninhabitable and places to scure. The best oral records, however, are reasonably clear and avoid because there was no food. Canoes and floatable describe waves inundating gardens on cliff tops [Smith,1910] artifacts would have been washed away. There would be the and wiping out entire settlements [Mitchell and Mitchell, deaths of young children, the elderly, and women (who tend 2004]. to stop and help the young and elderly; men appear to have a Archaeological evidence consistent with major tsunami in- stronger sense of self-preservation [Dudley and Lee, 1998, undation is reviewed by McFadgen [2007]. Archaeological p. 260]). People whose occupations kept them on the coast remains, including pieces of at least one smashed up canoe, would have been vulnerable, especially the craftsmen who along with other wooden objects and broken artifacts, have made the adzes and canoes, and the sailors and navigators been found driven inland, along with ripped up peat and plant who sailed the canoes. material. There are pebbles strewn over sand dunes around With the loss of canoes and navigators, long distance trade early settlement sites and shell middens and ovens spread over would have declined. In the absence of traded argillite, local the landscape, apparently by large waves washing over the stone was used for adzes, and with the loss of craftsmen, sites. Garden soils are buried beneath sand, and there are adzes became simpler. With the loss of resources, shortages changes in the shellfish content of middens, consistent with of food, and a shortage of women, warfare developed, but sand deposition, or its removal, affecting shellfish beds. There possibly the most crucial losses were the elders, who were are indications in areas where food was once plentiful of the sources of knowledge; when they died, so would much of people scratching for subsistence. On Matakana Island, a sandy the knowledge that kept their culture alive. With the loss of island at the entrance to Tauranga Harbor (Figure 1), 4 km of long distance coastal trade, the communication of ideas and the southeastern end was removed and later grew back. new innovations would have been reduced, and the moun- These seismic events coincide with a significant cultural tainous nature of the hinterland meant that overland commu- change recorded by archaeology. Before the fifteenth century, nication was more difficult than coastal interchange. It is the population appears to have been relatively peaceful: there therefore likely that for a time, parts of the country became are no clear signs of warfare, no fortifications. Then, at the isolated from each other, adding to the difficulties of recov- end of the fifteenth century, earthwork forts appeared in the ery. How long this isolation would have lasted for any region North Island and northern South Island [McFadgen et al., is unknown. Its eventual outcome, however, would be grad- 1994; Schmidt, 1996], warfare developed, and by the time of ual cultural divergence, similar to that proposed for isolated European rediscovery by James Cook in 1769, had become Polynesian islands [Irwin, 1992, p. 202]. Long distance endemic [Davidson, 1984]. Adze styles changed, from a very communication, however, had certainly been reestablished sophisticated kit that would have required trained craftsmen by the time of European contact in the late eighteenth century to make and use, to a simplified general-purpose tool that [Firth, 1929, p. 407; Beaglehole, 1968, p. 250], and artifacts potentially anyone could make [Furey, 2004]. The rock used from later period archaeological sites indicate its reestablish- for adze manufacture also changed. The D’Urville Island ment sometime earlier. metamorphosed argillite that was widely used and traded Unlike modern societies, there was no external support for during the early period fell out of use in favor of local those afflicted; the scale of the devastation was too large. (inferior) stone, implying that less trade was occurring. Gar- There is, however, a suggestion of subsequent cultural influ- dening practices changed, and stone row gardens were aban- ence from the southern South Island through the use of doned. At the same time, people deserted parts of the coast nephrite and the adoption of some artifact styles and prac- and established their settlements further inland. What ap- tices notably related to fishing [Davidson, 1984]. Changes to pears to have been a more or less uniform East Polynesian fishing technology would have been appropriate, considering culture, shared by people living in much of the North Island the impact that tsunami inundation would have had on north- and northern South Island, became the regionally distinctive ern coastal areas, but the extent, timing, and exact nature of Maori culture encountered by the late eighteenth century this influence has still to be documented. At the time of European explorers. European contact, regional differences in material culture, 200 INFLUENCES OF ACTIVE TECTONISM ON HUMAN DEVELOPMENT social organization, and linguistic dialects were apparent value of tectonic topography and lead to greater cultural [Skinner, 1921; Davidson, 1984; Harlow, 2007], but how complexity cannot be specified from our example and remain much was due to change since the fifteenth century, possibly unclear for many Neolithic situations. triggered by reduced trade and communication, and how much predates the events described here is not known. 5. CONCLUSIONS AND IMPLICATIONS In summary, New Zealand presented the Polynesian settlers with a resource-rich temperate environment that, climat- Tectonic landscapes and acceleration of cultural change, ically, was a significant contrast to the tropical environment the two hypotheses discussed above, both appear to be con- from which they had come. Their ability to adapt is well- sequences of active tectonism that have accelerated the de- demonstrated by their innovative approach to cultivation of velopment of our species. The former was, to some extent, the subtropical kumara (sweet potato), which led to tech- more important in earlier stages of our evolution, and the niques of soil modification to improve growing conditions latter is most obvious at the beginnings of history, but both and pit storage to overwinter tubers for planting the follow- factors probably were operating at any given stage, certainly ing season [Davidson, 1984]. They retained their innovative including the present day. skills throughout the pre-European period, well illustrated in Our New Zealand example proves to be an odd case of the latter part by their sophisticated system of warfare, and local cultural response to a landscape modified by tectonism, later, by readily adapting to European-introduced crops [Pet- but not in ways predictable in any simple scenario. It shows rie, 2006]. In addition, after European contact, Maori also that such response can go in a variety of directions, especially demonstrated strong entrepreneurial enterprise [Petrie, for individual events. However, tectonism clearly forced the 2006], which must have had its roots in the pre-European pace of change, and cultural response to tectonism was com- period, although how far back in time requires further re- prehensive, probably founded on an expectation of further search to ascertain. What they were unprepared for, however, events. Additional studies of historic Neolithic cultures in were the dangers of living in close proximity to a major other physical environments may shed more light on the tectonic plate boundary: the destruction that could follow dynamics of tectonically driven change. major earthquakes and tsunami in a young and fragile land- There are many things in the relation between humanity and scape. The immediate outcome inferred for Maori culture tectonic activity that we would like to know more about. was breakdown of long distance communication, regional However, there is more than enough information to conclude isolation, and warfare. The aggregate culture that subse- that effects of tectonic activity should be considered along quently developed was very different from that which had with other factors such as climate change in evaluating human prevailed before. activity and development. Tectonic activity has played a cen- The path taken by Maori cultural response to tectonic tral role in the physical and biological evolution of our planet; activity is quite distinct from that taken by some other perhaps, it is not surprising that humankind, in its evolution cultures, including Neolithic ones, where increased cultural and development, has both unwittingly taken advantage of complexity apparently accompanied active tectonism. Two Earth’s basic machinery and suffered attendant disadvantages. factors in the difference can be discerned: 1. The tsunamis that Maori endured were more destructive Acknowledgments. We appreciate the useful comments of J. of their coastal habitat than earthquakes per se would have Edward Wright, John Dohrenwend, Claudio Vita-Finzi, and been and, in some instances, led to relocation rather than Wayne Howell to ERF and Geoff Irwin, Auckland University; Brian Easton, Wellington; and Winifred Bauer, Victoria University of rebuilding. The Maori were driven away from a rich marine- Wellington to BGM. We are associated with International Geolog- terrestrial interface, and their impoverished resource base ical Correlation Project 567 “Earthquake Archaeology.” The section then reduced trade and cultural complexity. Excessive event on the NZ Maori is the development of a seminar given by BGM at magnitudes and short recurrence intervals may have discour- the invitation of Erika Guttman-Bond, Institute for Geo and aged recovery attempts. Bioarchaeology, Vrije Universiteit, Amsterdam. 2. The cultural and physical isolation of Maori limited their recovery options. This makes their response less like the REFERENCES response in cultures where innovative techniques for rebuilding in situ were available. Adams, J. (1980), Contemporary uplift and erosion of the Southern The Maori case shows that landscape modification by Alps, New Zealand, Geol. Soc. Am. Bull., 91,1–114. tectonism can be detrimental and that consequent cultural Bailey, G. N., and G. C. P. King (2010), Dynamic landscapes and change can be away from complexity. Maori response was human dispersal patterns: Tectonics, coastlines, and the recon- clearly comprehensive, however. Variables that enhance the struction of human habitats, Quat. Sci. Rev., 6,1–32. FORCE AND MCFADGEN 201

Bailey, G., G. King, and D. Sturdy (1993), Active tectonics and A 15th century New Zealand example, Quat. Sci. Rev., 21, land-use strategies: A Paleolithic example from northwest 2227–2234. Greece, Antiquity, 67, 292–303. Grapes, R., and G. Downes (1997), The 1855 Wairarapa, New Bailey, G. N., S. C. Reynolds, and G. C. P. King (2011), Landscapes Zealand earthquake—Analysis of historical data, Bull. N. Z. Soc. of human evolution: Models and methods of tectonic geomor- Earthquake Eng., 30, 271–369. phology and the reconstruction of hominin landscapes, J. Human Gratton, J., and R. Torrence (Eds.) (2008), Living Under the Shadow: Evol., 60, 257–280. Cultural Impacts of Volcanic Eruptions, Oxford Univ. Press, Balmuth, M. S., D. K. Chester, and P. A. Johnston (Eds.) (2005), Oxford, U. K. Cultural Responses to the Volcanic Landscape: The Mediterra- Harlow, R. (2007), Maori: A Linguistic Introduction, Cambridge nean and Beyond, Colloq. Conf. Pap.,vol.8,Am.Inst.of Univ. Press, Cambridge, U. K. Archaeol., Boston, Mass. Hoffman, S. A., and A. Oliver-Smith (Eds.) (2002), Catastrophe Beaglehole, J. C. (Ed.) (1968), The Journals of Captain James and Culture: The Anthropology of Disaster, pp. 187–212, Sch. of Cook: The Voyage of the Endeavour, 1768–1771, Cambridge Am. Res., Santa Fe, N. M. Univ. Press, Cambridge, U. K. Hogg, A. G., T. G. F. Higham, D. J. Lowe, J. G. Palmer, P. J. Cauvin, J. (2000), The Birth of the Gods and the Origins of Agri- Reimer, and R. M. Newnham (2003), A wiggle-match date for culture, translated by T. Watkins, Cambridge Univ. Press, Cam- Polynesian settlement of New Zealand, Antiquity, 77,116–125. bridge, U. K. Irwin, G. (1992), The Prehistoric Exploration and Colonization of Cuaresma, J. C., J. Hlouskova, and M. Obersteiner (2008), Natural the Paciﬁc, Cambridge Univ. Press, Cambridge, U. K. disasters as creative destruction? Evidence from developing Jackson, J. (2006), Fatal attraction: Living with earthquakes, the countries, Econ. Inquiry, 46, 214–226. growth of villages into megacities, and earthquake vulnerability Davidson, J. M. (1984), The Prehistory of New Zealand, Longman of the modern world, Philos. Trans. R. Soc. A, 364, 1911–1925. Paul, Auckland, N. Z. James, W. (1906), On some mental effects of the earthquake, in Davis, D. E. (2005), Reverberations: Mexico City’s 1985 earth- Writings 1902–1910, pp. 1215–1222, Libr. of Am., New York. quake and the transformation of the capital, in The Resilient City, Kant, I. (1790), Of nature regarded as might, in Critique of Judg- edited by L. J. Vale and T. J. Campanella, pp. 255–280, Oxford ment, p. 74, Nabu Press, Charleston, S. C. Univ. Press, Oxford, U. K. King, G., G. Bailey, and D. Sturdy (1994), Active tectonics and deBoer, J. Z., and D. T. Sanders (2005), Earthquakes in Human human survival strategies, J. Geophys. Res., 99(B10), 20,063– History, Princeton Univ. Press, Princeton, N. J. 20,078. Dudley, W. C., and M. Lee (1998), Tsunami!, Univ. of Hawaii Press, King, G. C. P., and G. N. Bailey (2006), Tectonics and human Honolulu. evolution, Antiquity, 80, 265–286. Firth, R. (1929), Primitive Economics of the New Zealand Maori, King, G. C. P., and G. N. Bailey (2010), Dynamic landscapes and Routledge, London, U. K. human evolution, in Ancient Earthquakes, edited by M. Sintubin Fisher, R. V., G. Heiken, and J. B. Hulen (1997), Volcanoes: et al., Spec. Pap. Geol. Soc. Am., 471,1–19. Crucibles of Change, Princeton Univ. Press, Princeton, N. J. Lowe, D. J. (2010), Quaternary volcanism, tephras, and tephra- Force, E. R. (2008), Tectonic environments of ancient civili- derived soils in New Zealand: An introductory review, in Guide- zations in the Eastern hemisphere, Geoarchaeology, 23, book for Pre-conference North Island, New Zealand ‘Volcanoes 644–653. to Oceans’ Field Tour (27–30 July). 19th World Soils Congress, Force, E. R., and B. G. McFadgen (2010), Tectonic environments of International Union of Soil Sciences, Brisbane, edited by D. J. ancient civilizations: Opportunities for archaeoseismological and Lowe et al., Soil Earth Sci. Occas. Publ. 3, pp. 7–29, Massey anthropological studies, in Ancient Earthquakes, edited by M. Univ., Palmerston North, N. Z. Sintubin et al., Spec. Pap. Geol. Soc. Am., 471,21–28. Lowe, D. J., R. M. Newnham, and J. D. McCraw (2002), Volcanism Froggatt, P. C., and D. J. Lowe (1990), A review of late Quaternary and early Maori society in New Zealand, in Natural Disasters silicic and some other tephra formations from New Zealand: and Cultural Change, edited by R. Torrence and J. Grattan, Their stratigraphy, nomenclature, distribution, volume, and age, pp. 126–161, Routledge, London, U. K. N. Z. J. Geol. Geophys., 33,89–109. McFadgen, B. G. (2007), Hostile Shores: Catastrophic Events in Furey, L. (2004), Material culture, in Change Through Time: Prehistoric New Zealand and Their Impact on Maori Coastal 50 Years of New Zealand Archaeology, N. Z. Archaeol. Assoc. Communities, Auckland Univ. Press, Auckland, N. Z. Monogr., vol. 26, edited by L. Furey and S. Holdaway, pp. 29–54, McFadgen, B. G., F. B. Knox, and T. R. L. Cole (1994), Radiocar- N. Z. Archaeol. Assoc., Auckland. bon calibration curve variations and their implications for the Gibbs, H. S., and W. A. Pullar (1961), Soils of the Bay of Plenty, interpretation of New Zealand Prehistory, Radiocarbon, 36(2), N. Z. Grassland Publ. 1802,23pp.,N.Z.GrasslandSoc., 221–236. Dunedin. Mitchell, H., and J. Mitchell (2004), Te Tau Ihu O Te Waka: A Goff, J. R., and B. G. McFadgen (2002), Seismic driving of nation- History of Maori of Nelson and Marlborough, vol. 1, Te Tangata wide changes in geomorphology and prehistoric settlement— me te Whenua: The People and the Land, Huia, Wellington, N. Z. 202 INFLUENCES OF ACTIVE TECTONISM ON HUMAN DEVELOPMENT

Molnar, P. (1990), The rise of mountain ranges and the evolution of Schumpeter, J. A. (1942), Capitalism, Socialism and Democracy, humans: A causal relation?, Irish J. Earth Sci., 10, 199–207. Harper, New York. Nichol, S. L., O. B. Lian, and C. H. Carter (2003a), Sheet-gravel Sheets, P. D., and D. K. Grayson (Eds.) (1979), Volcanic Activity evidence for a late Holocene tsunami run-up on beach dunes, and Human Ecology, Academic, New York. Great Barrier Island, New Zealand, Sediment. Geol., 155, Skidmore, M., and H. Toya (2002), Do natural disasters promote 129–145. long-term growth?, Econ. Inquiry, 40, 664–687. Nichol, S. L., J. R. Goff, and H. Regnault (2003b), Cobbles to Skinner, H. D. (1921), Culture areas in New Zealand, J. Polynesian diatoms: Facies variability in a palaeo-tsunami deposit, paper Soc., 30,71–78. presented at the 5th International Symposium on Coastal Engi- Smith, S. P. (1910), History and Traditions of the Maoris of the neering and Science of Coastal Processes, Florida, 18–23 May. West Coast North Island of New Zealand Prior to 1840, Mem. Nichol, S. L., J. R. Goff, and H. Regnault (2004), Sedimentary Polynesian Soc., vol. 1, Thomas Avery, New Plymouth, N. Z. evidence for tsunami on the northeast coast of New Zealand, Soren, D., and J. James (1988), Kourion—The Search for a Lost Geomorphologie, 1,35–44. Roman City, Anchor, New York. Oliver-Smith, A., and S. M. Hoffman (Eds.) (1999), The Angry Vale, L. J., and T. J. Campanella (2005a), Introduction: The cities Earth: Disaster in Anthropological Perspective, Routledge, rise again, in The Resilient City: How Modern Cities Recover New York. From Disaster, edited by L. J. Vale and T. J. Campanella, Petrie, H. (2006), Chiefs of Industry. Maori Tribal Enterprise in pp. 3–26, Oxford Univ. Press, Oxford, U. K. Early Colonial New Zealand, Auckland Univ. Press, Auckland, Vale, L. J., and T. J. Campanella (2005b), Conclusion: Axioms of N. Z. resilience, in The Resilient City: How Modern Cities Recover Reycraft, R. M., and G. Bawden (Eds.) (2000), Environmental from Disaster, edited by L. J. Vale and T. J. Campanella, pp. 335– disaster and the archaeology of human response, Anthropol. Pap. 356, Oxford Univ. Press, Oxford, U. K. 7, Maxwell Mus. of Anthropol., Albuquerque, N. M. Wells, A., and J. R. Goff (2006), Coastal dune ridge systems as Rothaus, R. M. (1996), Earthquakes and temples in Late Antique chronological markers of palaeoseismic activity: A 650-yr record Corinth, in Archaeoseismology, edited by S. Stiros and R. E. from southwest New Zealand, The Holocene, 16(4), 543–550. Jones, Fitch Lab. Occas. Pap. 7, pp. 105–112, Br. Sch. at Athens, Athens, Greece. Rozario, K. (2007), The Culture of Calamity: Disaster and the E. R. Force, Department of Geosciences, University of Arizona, Making of Modern America, Univ. of Chicago, Chicago, Ill. Tucson, AZ 85721, USA. ([email protected]) Schmidt, M. (1996), The commencement of pa construction in New B. G. McFadgen, School of Maori Studies, Victoria University Zealand prehistory, J. Polynesian Soc., 105(4), 441–451. of Wellington, Wellington 6140, New Zealand. Irragric Anthrosols—Artifacts of Human Adaptation to Arid Conditions: Examples From the High Himalaya and the Coastal Desert of Peru

J. Baade

Department of Geography, Friedrich-Schiller-University, Jena, Germany

Irrigation agriculture represents an ancient adaptation to the environmental conditions in dry lands. Utilizing additional ground or surface water sources enabled societies to establish arable farming systems in otherwise water-scarce environments and gain a measure of resilience against climate fluctuations already thousands of years ago. When surface water is utilized for irrigation, the diversion of water and suspended sediment to the fields results in the buildup of irragric anthrosols representing genuine artifacts of irrigation agriculture. Often these deposits contain other artifacts like charcoal and pottery fragments that can be used for age determination. Surprisingly, it is difficult to find publications using the term irragric anthrosols in a geoarchaeological context. This is most probably due to the novelty of the term in the global soil description system since it was only introduced in the late twentieth century. This paper presents examples of irragric anthrosols from two rather contrasting landscapes, the High Himalaya and the coastal desert of Peru, to advance the recognition of the term and concept. In addition, some of the far-reaching implications for the interpretation of landscape development and human impact are discussed. It is suggested that a reanalysis of soil profiles from other areas where stream discharge has been used for irrigation will probably reveal that this soil type is much more abundant than presently perceived.

1. INTRODUCTION to several thousand years ago (see the work of Li et al. [2009] for a current global overview). In many of these locations, Irrigation agriculture represents an ancient and widespread irrigation is still used today, providing evidence for the creative human adaptation to the water-scarce environmental fundamental sustainability and resilience of irrigation agri- conditions in semiarid to hyperarid regions. Utilizing addi- culture [e.g., Sandor, 1992], despite the fact that failed irri- tional ground or surface water sources enabled societies to gation systems possibly caused the decline of civilizations, establish arable farming systems under otherwise hostile and the abandonment of settlements and regions has often environmental conditions and gain a measure of resilience gained much more scientific and general attention. against annual variations in rainfall. In a number of locations When surface water is used for irrigation, the diversion of in arid Africa, North and South America, and Eurasia, the water-carrying suspended sediment to the fields eventually establishment of early irrigation systems can be traced back results in the development of what has been termed an irragric horizon by the IUSS Working Group WRB [2006]. Often these sediments contain other artifacts like charcoal Climates, Landscapes, and Civilizations and pottery fragments that can be used for age determination. Geophysical Monograph Series 198 This short chapter introduces the term and presents examples © 2012. American Geophysical Union. All Rights Reserved. of irragric anthrosols from two contrasting landscapes, i.e., 10.1029/2012GM001206 the High Himalaya and the coastal desert of Peru in order to 203 204 IRRAGRIC ANTHROSOLS advance the recognition of this rather novel term from soil water into the fields themselves, adding directly to the land sciences. In addition, some of the far-reaching implications surface” when presenting their interpretation of the develop- for the interpretation of landscape development and human ment of the up to 10 m thick “alluvium” of the Mesopotamian impact are discussed. Reanalysis of soil profiles in other plain. Further early reports on soil or sediment buildup due to irrigated context will probably reveal abundant evidence for surface runoff irrigation in an archaeological context are irragric anthrosols. available from the Yemen [Brunner, 1983; Coque-Delhuille and Gentelle, 1997], the Southwest of North America [San- 2. IRRAGRIC ANTHROSOLS: DEFINING THE TERM dor, 1992], as well as the Hetian region, Xinjiang, China [Bridges et al., 1998; Gong et al., 1999]. The term irragric anthrosol (ATir) was introduced to the revised Food and Agriculture Organization (FAO) global soil 2.1. Irragric Anthrosols in the High Himalaya description system, i.e., the World Reference Base for Soil Resources [IUSS Working Group WRB, 2006], based on the Investigations in the High Himalaya focused on the upper recognition of human impact as a key soil-forming factor. Kali Gandaki Valley (2500 to >4000 m above sea level (asl)) According to the IUSS Working Group WRB [2006, p. 24; or Thakkhola, respectively, in the Mustang District, Nepal. see also Gong et al., 1999; FAO, Topsoil characterization for Semiarid climatic conditions in this region located north of sustainable management, draft report, Rome, 1998, herein- the Himalaya’s high crest are characterized by a mean annual after referred to as FAO, draft report, 1998] an air temperature of 11°C, a mean annual precipitation of 265 mm, and a mean annual pan evaporation of approximately irragric horizon . . . is a human induced mineral surface horizon [with] 2000 mm in the valley bottom at Jomsom (28.783°N, a thickness of [at least] 20 cm that builds up gradually through continuous application of irrigation water with substantial amounts of 83.723°E, 2740 m asl). sediments, and which may include fertilizers, soluble salts, organic The Thakkhola is characterized by a mixture of small- matter, etc. scale operational and abandoned irrigation terraces located mostly along and at the mouth of the tributaries to the Once a soil is characterized by an at least 50 cm thick irragric Kali Gandaki River. In general, the irrigation terraces are horizon, the soil is termed an “irragric anthrosol” [IUSS located several meters to a few tens of meters above the Working Group WRB, 2006, p. 53]. Diagnostic properties for corresponding valley bottom. Feeder channels diverting an irragric horizon refer to the content of fines (clay, silt, and water from perennial streams often fed by glaciers reach sand), carbonates, and organic matter, all in comparison to several hundred to a thousand meters upstream. The 4 m the underlying stratum as well as abundant traces of soil thick fine deposits present on the irrigation terraces represent animal activity. The “lacking evidence of stratification owing irragric anthrosols and provide evidence for a long-lasting to continuous ploughing” [IUSS Working Group WRB, 2006, utilization of the landscape for irrigation-based agriculture. p. 25] helps to distinguish irragric horizons or anthrosols As an example, Figure 1 shows the buildup of the terraced from their natural equivalents’ alluvial deposits, fluvic ma- “alluvial” fanatthemouthoftheMuktinathValleyat terials, or fluvisols. Kagbeni (28.835°N, 83.783°E, 2790 m asl). The base is Surprisingly, publications where the term “irragric anthro- comprised of poorly sorted coarse material (gravels, cobbles, sols” has been used are difficult to find. As of September and boulders) typical for fluvial and debris-flow deposits in 2011, a Google Scholar search for the term returned 75 hits, high mountain environments. The overlying ~3.5 m thick while Scopus, Web of Knowledge, and other publisher- layer of fines (predominantly silt and sand), lacking evi- related databases yielded ≤1 hit. The scarcity of publications dence of stratification with the exception of some massive using the term “irragric anthrosols” in the context of bedding, represents an irragric anthrosol. A stone wall ex- geoarchaeological research is most probably due to the tends nearly to the base of the fines and provides further novelty of the term. It can be traced back to the draft of the support for this inference. Dates on charcoal collected from revised global FAO soil description where it is used with a the base of the irragric anthrosol in Kagbeni suggest that reference to a soil survey in Yemen from 1983 (FAO, draft irrigation and arable farming started at Kagbeni around report, 1998). Although coined as a technical term only in 1700 ± 130 calendar years B.P. (cal B.P.) (2-σ-error). This the late twentieth century, the recognition of the importance is further supported by potsherds retrieved from 2.7 to 2.0 m of the associated processes (sediment transport to and depo- below the surface of the deposits dating back to the fourth to sition on irrigated fields) can be traced back at least to the seventh and eighth to tenth centuries A.D. [Baade et al., work of Jacobsen and Adams [1958: p. 1252] who stated that 2001]. At another location, 5 km upstream of Kagbeni in the “another increment of sediment accompanies the irrigation Muktinath Valley, the sedimentation of the anthrosol started BAADE 205

2.2. Irragric Anthrosols in the Coastal Desert of Peru

In the coastal lowlands of Peru, settlements are confined to irrigated river oasis fed by runoff from the Andes. Dating of ancient irrigation channels suggests a start of irrigation agriculture in the northern coastal lowlands around 5400 or even 6700 years ago [Dillehay et al., 2005]. This study focused on the Palpa Valley located in the Rio Grande de Nazca catchment, 400 km south of Lima, representing one of the major river oases in this region. Hyperarid climatic conditions are characterized by mean annual air temperature of 22°C, mean annual precipitation of 10 mm, and mean annual pan evaporation of approximately 1650 mm in the valley bottom at Palpa (14.533°S, 75.183°W, 310 m asl) [Hesse and Baade, 2009]. The Palpa Valley, as well as the neighboring valleys, is characterized today by large-scale irrigation schemes including fruit orchards totaling ~1900 ha [Baade and Hesse, 2008]. In general, the irrigation terraces are located only several meters above the corresponding valley bottom. Feeder channels diverting water from intermittent streams draining the western slopes of the Andes and providing discharge for a few months, only reach several hundred to a few thousand meters upstream. Locally, additional groundwater sources have been tapped; since the last century, irrigation water supply is supplemented with groundwater pumped from wells. However, river discharge is the main source of irrigation water, diverting an estimated 55%–95% of mean annual runoff to the fields. The thickness of irragric anthrosols varies between <0.5 m in recently reclaimed areas and up to 3.8 m. Extensive dating of soil profiles in the Palpa Valley using charcoal (N = 70) Figure 1. Exposure of the terraced “alluvial” fan at Kagbeni, Nepal, suggests that the buildup of irragric anthrosols in this valley with coarse basal deposits overlain by an ~3.6 m thick irragric anthrosol. Black dotted line denotes the base of the anthrosol; white started at least 2960 ± 120 cal B.P. and that irrigation agri- dotted line denotes a stone wall reaching down to the bottom of the culture was widely applied around 2500 years ago [Hesse, anthrosol. 2008; Hesse and Baade, 2007b]. These findings are in good agreement with the results of archaeological investigations of different settlement sites in the Palpa Valley and along the ~2415 ± 85 cal B.P. (2-σ-error) (for details, see the work of Rio Grande de Nazca. The long-term mean sedimentation Baade [2000]). The long-term mean sedimentation rate of the rate of the irragric anthrosols varies between 0.3 and 3.0 mm À1 anthrosols, i.e., the mean rate calculated from total thickness yr for different soil profiles. and the basal dates, varies between 2 mm yrÀ1 at Kagbeni and The sedimentary record from the Palpa Valley provides 8mmyrÀ1 in other parts of the Muktinath Valley. evidence for erosion and sedimentation during extreme run- The general persistence of irrigation agriculture over more off events, but there is no indication of widespread abandon- than two millennia demonstrates the robustness and sustain- ment of irrigated fields related to the cultural changes ability of this technique in the High Himalaya. The mixture documented for this area [e.g., Reindel and Isla Cuadrado, of operational and abandoned terraces indicates that processes 2004]. This demonstrates the resilience and sustainability of of regional or global extent, like climate change, might not this technique in the hyperarid coastal desert of Peru. be accountable for the abandonment of specific sites. Instead, geomorphologic analysis of abandoned sites revealed that 3. DISCUSSION debris flows, deep-seated mass movements, and river incision disrupting feeder canals are implicated [Baade et al., The two examples of ancient irrigation systems presented 1998]. in this paper from study sites in the High Himalaya and the 206 IRRAGRIC ANTHROSOLS coastal desert of Peru provide evidence for the resilience and hand, evince the introduction of stream discharge-based irri- sustainability of irrigation agriculture under rather contrast- gation at a specific site usually in the lower part of a catch- ing environmental conditions. In both cases, long-lasting ment and the utilization of water (and sediment) discharge irrigation agriculture resulted in the slow buildup of several from the catchment and cannot be interpreted as evidence for meters thick irragric horizons due to the intentional diver- land use or climate change in the upstream catchment area sion of water and sediment from the rivers to the fields. [Hesse and Baade, 2007a]. In addition, irragric anthrosols Thus, irragric anthrosols can be considered to represent should be considered as genuine artifacts of human adapta- genuine artifacts of human impact on the environment. The tion to difficult climatic conditions. They provide evidence deposits are characterized by rather massive beds of fines of a long-lasting history of human impact on the runoff and lacking further evidence of stratification due to the field sediment transport system in arid environments. Thus, some preparation taking place prior and shortly after irrigation. of the river oasis environments around the world might Usually, irragric anthrosols contain charcoal and other arti- actually represent true cultural landscapes. facts like potsherds that can be used to date the deposits and thus the onset and duration of irrigation agriculture at a 4. CONCLUSIONS specific site. In addition, single-grain quartz optical luminescence dating might be applicable [Berger et al., 2009]. The many reports on abandoned irrigation systems might Comparison of sedimentation rates of irragric anthrosols at suggest that this technology is particularly susceptible to the two sites suggests mean long-term rates in the order of global environmental change. These two case studies of 0.5 to 10 mm yrÀ1. These rates support an initial estimate of ancient and still operational irrigation systems in rather con- the age of an irrigation system from the thickness of the trasting environmental settings (semiarid high mountains, irragric anthrosols, taking into account an error in the order hyperarid coastal lowlands) provide evidence for the poten- of a magnitude or so. Unfortunately, the number of studies tial resilience and sustainability of irrigation agriculture. This identifying and describing irragric anthrosols and providing paper further suggests that dating of irragric anthrosols is a chronologies for their development from other sites around convenient approach for establishing the onset of irrigation the world is still very low. Therefore, it is very difficult to agriculture at a specific site. Comparison of sedimentation provide comparable measurements. However, the sedimen- rates suggests that a first crude estimate of the age of an tation rate estimate provided corresponds well to sedimen- irrigation system can be derived from the thickness of the tation rates (3–5mmyrÀ1) and age determinations (1870 irragric anthrosols, taking into account an error in the order years B.P.) reported by Gong et al. [1999] for up to 5 m of a magnitude. However, in order to verify this hypothesis, thick irragric anthrosols in the Hetian region of Xinjiang, many case studies and dates for irragric anthrosols are needed. China (≈37.083°N, 79.917°E). In addition, the paper highlights the importance of a reliable Another aspect that might further explain the scarcity of differentiation of alluvial soils and irragric anthrosols be- reports on irragric anthrosols, in addition to the novelty of the cause of the far-reaching consequences for the palaeoenvir- term, is the fact that irrigation systems and irragric anthrosols onmental interpretation of Holocene landscape evolution in are usually found in an “alluvial” setting, i.e., in the valley dryland environments. bottom and on the lower terraces. In this geomorphic context, soils and sediments are often interpreted solely as floodplain Acknowledgments. Financial support of the German Science deposits even if old irrigation channels are present [e.g., Foundation (DFG) (research grants Ma 1308/5-1 to 5-4 and Ba Doolittle, 2006]. For example, in situations where ancient 1377/6-1, 6-2) is acknowledged. Three anonymous reviewers and the editor provided helpful comments to the manuscript. irrigation channels were used to divert surface runoff to fields, e.g., in the middle Gila River valley [Woodson, 2010], the soils might actually represent irragric anthrosols REFERENCES and not alluvial soils. Within the context of palaeoecological studies, a reliable distinction between alluvial soils and Baade, J. (2000), Landschaftswandel im Thakkhola. Untersuchun- gen zur Landschaftsgenese im semi-ariden Hochgebirge Nepals irragric anthrosols is more than an academic soil classifica- seit dem Jungpleistozän, Habilitationsschrift, Chem.-Geowiss. tion dispute. It has far-reaching consequences for the inter- Fakultät, Univ. of Jena, Jena, Germany. [Available at http:// pretation of landscape evolution. Alluvial soils clearly www.db-thueringen.de/servlets/DocumentServlet?id=16503.] fl indicate precipitation-induced overbank ooding events, and Baade, J., and R. Hesse (2008), An overlooked sediment trap in arid they can be used as indicators for land use and climate environments: Ancient irrigation agriculture in the coastal desert change in the upstream catchment [e.g., Ferring, 1992; Leo- of Peru, in Sediment Dynamics in Changing Environments, edi- pold and Vita-Finzi, 1998]. Irragric anthrosols, on the other ted by J. Schmidt et al., IAHS Publ., 325, 375–382. BAADE 207

Baade, J., R. Mäusbacher, G. A. Wagner, E. Heine, and R. Kostka tions of geoarchaeological evidence: A comment on B. Eitel et al. (1998), Landslides and deserted places in the semi-arid environ- (2005) Geoarchaeological evidence from desert loess in the ment of the Inner Himalaya, in Geomorphological Hazards in Nazca-Palpa region, southern Peru: Palaeoenvironmental High Mountain Areas, edited by J. Kalvoda and C. L. Rosenfeld, changes and their impact on pre-Columbian cultures, Archaeome- pp. 49–62, Kluwer Acad., Dordrecht, The Netherlands. try, 47(1), 137–158, Archaeometry, 49(3), 595–602, doi:10.1111/j Baade, J., R. Mäusbacher, and G. A. Wagner (2001), The devel- .1475-4754.2007.00322.x. opment of the Dzong Chu alluvial fan at Kagbeni, in Kagbeni – Hesse, R., and J. Baade (2007b), Early Horizon anthrosols in the Contributions to the Village’s History and Geography, Giess- Palpa Valley, southern Peru, Geoöko, 28(3–4), 160–186. ener Geogr. Schr., vol. 77, edited by P. Pohle and W. Haffner, Hesse, R., and J. Baade (2009), Irrigation agriculture and the pp. 25–41, Selbstverlag, Inst. für Geogr., Univ. Giessen, Giessen, sedimentary record in the Palpa Valley, southern Peru, Catena, Germany. 77(2), 119–129, doi:10.1016/j.catena.2008.09.003. Berger, G. W., S. Post, and C. Wenker (2009), Single and multi- IUSS Working Group WRB (2006), World Reference Base for Soil grain quartz-luminescence dating of irrigation-channel features Resources 2006. A Framework for International Classiﬁcation, in Santa Fe, New Mexico, Geoarchaeology, 24(4), 383–401, Correlation and Communication, World Soil Resour. Rep., vol. doi:10.1002/gea.20271. 103, FAO, Rome, Italy. Bridges, E. M., N. H. Batjes, and F. O. Nachtergaele (Eds.) (1998), Jacobsen, T., and R. M. Adams (1958), Salt and silt in ancient World Reference Base for Soil Resources: Atlas, ISRIC-FAO- Mesopotamian agriculture, Science, 128(3334), 1251–1258. ISSS-Acco, Leuven, Belgium. Leopold, L. B., and C. Vita-Finzi (1998), Valley changes in the Brunner, U. (1983), Die Erforschung der antiken Oase von Marib Mediterranean and Americas and their effects on humans, Proc. mit Hilfe geomorphologischer Untersuchungsmethoden, Archäo- Am. Philos. Soc., 142(1), 1–17. ologische Berichte aus dem Yemen, vol. 2, von Zabern, Mainz, Li, Y. J., P. S. Thenkabail, C. M. Biradar, P. Noopjipady, V. Dheer- Germany. avath, M. Velpuri, O. R. P. Gangalakunta, and X. L. Cai (2009), A Coque-Delhuille, B., and P. Gentelle (1997), Floods and controlled history of irrigated areas of the world, in Remote Sensing of sedimentation in ancient Yemen [in French], Géomorphologie, Global Croplands for Food Security, edited by P. S. Thenkabail 3(2), 99–109. et al., pp. 13–38, CRC Press, Boca Raton, Fla. Dillehay, T. D., H. H. Eling, and J. Rossen (2005), Preceramic Reindel, M., and J. IslaCuadrado (2004), Archäologisches Projekt irrigation canals in the Peruvian Andes, Proc. Natl. Acad. Sci. “Paracas in Palpa,” Peru. Bericht über die Grabungskampagne U. S. A., 102(47), 17,241–17,244, doi:10.1073/pnas.0508583102. 2003, in Jahresbericht der Schweizerisch-Liechtensteinischen Doolittle, W. E. (2006), Agricultural manipulation of ﬂoodplains in Stiftung für Archäologische Forschung im Ausland, Vol. 2003, the southern Basin and Range Province, Catena, 65(2), 179–199, pp. 137–156, Schweizerisch-Liechtensteinische Stift. für archäol. doi:10.1016/j.catena.2005.11.002. Forsch. im Ausland (SLSA), Zurich, Switzerland. Ferring, C. R. (1992), Alluvial pedology and geoarchaeological Sandor, J. A. (1992), Long-term effects of prehistoric agriculture on research, in Soils in Archaeology. Landscape Evolution and Hu- soils: Examples from New Mexico and Peru, in Soils in Archae- man Occupation, edited by V. T. Holliday, pp. 1–39, Smithson. ology. Landscape Evolution and Human Occupation, edited by Inst. Press, Washington, D. C. V. T. Holliday, pp. 217–245, Smithson. Inst. Press, Washington, Gong, Z., G. Zhang, and G. Luo (1999), Diversity of anthrosols in D. C. China, Pedosphere, 9(3), 193–204. Woodson, M. K. (2010), Re-drawing the Map of the Hohokam Canals Hesse, R. (2008), Fluvial Dynamics and Cultural Landscape Evo- in the Middle Gila River Valley, J. Ariz. Archaeol., 1(1), 5–19. lution in the Rio Grande de Nazca Drainage Basin, Southern Peru, BAR Int. Ser., vol. 1787, Archaeopress, Oxford, U. K. Hesse, R., and J. Baade (2007a), Palaeoenvironmental changes in J. Baade, Department of Geography, Friedrich-Schiller-Univer- the Nazca-Palpa region, Southern Peru—Alternative interpreta- sity, Löbdergraben 32, D-07743 Jena, Germany. ([email protected]) Hunter-Gatherers, Agriculturalists, and Climate: Insights From a Cross-Disciplinary Review

Daniel M. Cadzow

Department of Anthropology, State University of New York at Buffalo, Buffalo, New York, USA

Archaeologists have been studying human-caused environmental impacts for decades. Most studies have examined major sites or civilizations, often with a focus on their collapse or reactions to environmental crises. Evidence emerging from numerous disciplines suggests it was the aggregated efforts of hunter-gatherers and agriculturalists that also created some of humanity’s greatest environmental impacts. This paper addresses two paths by which these impacts can manifest: agriculture/forest management and biomass burning. This conclusion can be provocative for those who value indigenous populations as benign ecological stew- ards. However, many of these impacts cannot be neatly classiﬁed as negative or positive. Some human-caused disturbances, for example, have helped to preserve threatened species, fostered biodiversity, and increased the resilience of socioecological systems. To increase our knowledge of our collective human heritage and better support indigenous and western decision making, additional cross- disciplinary research on how hunter-gathers and agriculturalists inﬂuence socioecological systems is needed. This will require the convergence of data, insights, and theory from many disciplines.

1. INTRODUCTION ments, occurred outside the spatial and temporal boundaries of the great civilizations and archaeological sites. With the Archaeologists have long had a keen interest in the research integration of findings from numerous disciplines, it is be- of geoscientists working with paleoenvironmental reconstruc- coming increasingly likely that the primary drivers of envi- tions. However, as research proceeds in both arenas, it is ronmental impacts were hunter-gatherers and agriculturalists. becoming increasingly clear that more intensive interdisci- The term “hunter-gatherer” refers to people that acquire plinary cooperation is needed to better investigate the history most of their food from hunting and foraging. “Agricultural- of humanity’s relationships with its environments [Costanza ists,” conversely, obtain most of their food through the cul- et al., 2007]. We need an understanding that transcends tivation of domesticated or semidomesticated flora and, in simple, unicausal theories of human impacts or societal col- some cases, fauna. The terms are combined here because lapse that focus only on major sites or civilizations [e.g., they often represent two extremes on a spectrum of potential Diamond, 2006]. While such events are dramatic, they are activities for a population. Indeed, it has been shown that the interesting because they are the exceptions to the norm. Most transition from strictly hunter-gatherer to strictly agricultural of human prehistory, including some of its greatest achieve- societies took millennia, with much movement back and forth along the continuum [Harris, 2007; Fuller, 2010]. In comparison to modern environmental impacts, it can be fi Climates, Landscapes, and Civilizations dif cult to imagine how people with these modes of subsis- Geophysical Monograph Series 198 tence could have any notable influences on the environment. © 2012. American Geophysical Union. All Rights Reserved. However, humans made their first appearance in most of the 10.1029/2012GM001210 habitats they now occupy as an invasive species. Unlike 209 210 HUNTER-GATHERERS, AGRICULTURALISTS, AND CLIMATE

Asian carp or purple loosestrife, humans are relatively large, by humans also sustains populations of fauna, like red brocket socially organized, highly intelligent, and possess advanced deer, which are also exploited by people (i.e., wild-managed technologies, including fire. While the rates of change with fauna). Even after settlements are abandoned, resources from which most species adapt to external changes are regulated the fallow gardens and swiddens continue to be exploited by biology, humans also evolve culturally. This means they (e.g., for food, medicine, manufacturing, propagation of can cause shocks long after they arrive. Although the prac- plants by taking cuttings, etc.). Interwoven with these acts tices of agriculturalists and hunter-gatherers can appear triv- of exploitation are acts of maintenance that foster the devel- ial, their effects accumulate, and they have been agents in the opment of entrenched resources, like forest orchards, that environment for thousands and tens of thousands of years. persist even if maintenance ceases. The ways we think about humanity and its relation to the Over time, these activities have shaped the composition of environment are structured by traditions of thought we in- forests that can be classified as old growth. These entrenched herit, sometimes without critical appraisal. Common among resources can be thought of as an ecological form of land- these assumptions is that humans in natural settings univer- esque capital [see Blaikie and Brookfield, 1987] and would sally cause harm [see Redman, 1999]. However, there is a increase the resilience of environmental regimes that are growing body of research and data that illustrate how in- beneficial for humans. These resources also act as a form of creasing populations and human impacts can cause, for ex- externalized “memory” that has helped groups reorganize ample, deforestation or reforestation as well as losses or when confronting shocks like the collapse of states, war, gains in biodiversity. It is not the presence of humans, or slavery, or involuntary migrations. even the size of a population, that determines how ecological Human interactions with the evolution of organisms and impacts manifest; it is what people do with their resources ecosystems take many forms. For example, Brazil’s Huaor- [Fairhead and Leach, 1996; Malakoff, 2011]. ani actively cultivate many useful plants that are not domesticated through morphological or genetic modification. 2. METHODS Conversely, they also make use of many domesticated species that are dispersed throughout the forest [Rival, The clearest view of the history of humanity’s environ- 1998]. Based on interviews, groups like the Ka’apor are mental impacts arises from the amalgamation of diverse sets unaware of the roles they play in long-term processes like research. Discussed here are findings from cultural anthro- domestication. However, they can discern human activities pology (i.e., ethnology), history, archaeology, paleoenviron- in the distant past because of their influences on the forest mental reconstruction, ecology, and paleoclimatology. This [Balée, 1993, 1994]. The Huaorani similarly see a record of research addresses two related paths by which these impacts past activities in their forested land, which they describe as can manifest: agriculture/forest management and biomass a sequence of fallows. They are also aware of the effects burning. their daily practices have in the maintenance of the forest and fallows. For example, they know that their cooking 2.1. Ethnography facilitates propagation of ungurahua palm by encouraging the germination of its seeds [Rival, 1998]. These hunter- Recent ethnographic investigations of modern hunter- gatherer/agriculturalists possess a sophisticated knowledge gatherers and agriculturalists in the neotropics provide evi- about anatomy, life cycle, and techniques for manipulating, dence of beliefs, technologies, practices, and ecological harvesting, and propagating plants. This depth of knowl- modifications resulting from indigenous land management. edge of, and influences on, ecosystems does away with One investigation describes a centuries-long continuum of stereotypes of ecologically minded primitives enjoying the shifting management practices as occurring in three stages: fruits of a pristine wilderness. home garden, managed fallow, and managed forest [Peters, Human-impacted areas can also be bastions of biodiversity. 2000]. Another describes a similar process in terms of door- This is evidenced by floral inventories of eight, noncontigu- way gardens, young swiddens, old swiddens, and fallow ous, single-hectare plots of well-drained forests in the terri- [Balée, 1993, 1994]. Swiddens are small areas around a tories of the Ka’apor, Guajá, and Tembé Indians in Amazonia settlement that are cleared and burned to make room for [Balée, 1993]. Four of the study plots were “high forest” (i.e., crops. They are typically cultivated for a few years and then having no evidence of human impacts) and four of the plots left fallow (i.e., uncultivated) for as much as 20 years. In both were “fallows” (i.e., had been cultivated within the last three cases, the cultivation and care for flora near homes and centuries). The floral composition of high forest and fallows swiddens is related to the care for other nondomesticated were distinctive. Correlations between plots were much flora (i.e., wild-managed flora). Some of the flora managed stronger with relation to the presence or absence of past CADZOW 211 agriculture than with distance between plots. Both types of Despite the abundance of evidence, the use of biomass plots had similar total numbers of different species. However, burning in Native North America has gone largely unrecog- high forest study plots were dominated by a few species, with nized due to the concomitant association of Native Ameri- only rare occurrences of many other species. The fallows, in cans with the environmental movement and belief that wild contrast, were not dominated by a few species, and most land fires are universally harmful [Redman, 1999; Cadzow, species had numerous occurrences. The high forests, being 2011]. The latter point is refuted by a brief sampling of first- dominated by few species, would be more susceptible to hand accounts of biomass burning. During the mid-seven- disturbances like infestations and therefore less resilient teenth century, a Dutch colonist described biomass burning [Holling and Gunderson, 2002]. in what is now New York State [Van der Donck, 1656, A survey of biodiversity at 10 forest reserves, 25 sacred pp. 150–151]: groves, and 23 coffee plantations in the Western Ghats of India had similar results. The report concluded that indige- The Indians have a yearly custom . . . of burning the woods, plains and meadows in the fall of the year . . . Those places which are then passed nous forest management systems are as, or perhaps more, over are fired in the spring in April. This practice is . . . done for several effective at protecting threatened species and maintaining reasons; first, to render hunting easier . . . Secondly, to thin out and biodiversity as the formal forest reserves [Bhagwat et al., clear the woods of all dead substances and grass, which grow better the ensuing spring. Thirdly, to circumscribe and enclose the game within 2005]. the lines of the fire . . . and also, because the game is more easily These studies illustrate how the activities of hunter-gatherers tracked over the burned parts of the woods . . . still the green trees do and agriculturalists can promote biodiversity, foster threat- not suffer. The outside bark is scorched three or four feet high. . . ened species, and increase the resilience of forest systems. fi This is in contrast to historically negative portrayal of the so- Samuel Champlain described the use of re in clearing fi called slash-and-burn (i.e., swidden) agriculture. These dis- elds near Gloucester Harbor, Massachusetts, in 1604 – turbances, cutting, burning, cultivating, and maintenance [Champlain, 1878, pp. 43 115]: create a patchwork of managed successional forests and Some of the land was already cleared up, and they were constantly thereby refuges for species that variously thrive at different making clearings. Their mode of doing it is as follows: after cutting stages of succession. They can prevent a forest from becom- down the trees at the distance of three feet from the ground, they burn ing dominated by a few species, which lowers biodiversity as the branches upon the trunk, and then plant their corn between these stumps, in the course of time tearing up also the roots. There are well as resilience to shocks like fire or infestation [see likewise fine meadows here. . . Thompson et al., 1979]. While ethnographies provide a sense of how hunter-gatherers and agriculturalists interact with Jacques Cartier [1906] described the extent of such efforts their environs, history, archaeology, and geosciences give us in his account of Hochelaga, a village containing about 50 insights as to how far into the past we can extend this long houses, along Saint Lawrence River in Québec in 1535. knowledge. “We began to finde goodly and large fieldes, full of such corne as the countrie yeeldieth... In the midst of those fields 2.2. Ethnohistory is the citie of Hochelaga” (p. 59). Though the type of cultivation practiced in this region can be described as swidden Ethnohistoric literature is filled with descriptions of peo- agriculture, some of these fields were large and cultivated for ple, landscapes, and agricultural practices including biomass decades before villages and fields were relocated. It is also burning. The ubiquity of written evidence for protohistoric unclear what role the agricultural fitness of the fields played, human-modified landscapes and biomass burning has been if any, in deciding when to relocate [Engelbrecht, 2003]. discussed for years [e.g., Day, 1953; Stewart, 1956; Lewis, In the late eighteenth century, Henry Bouquet wrote of 1973]. For example, historic analysts for the United States Native Americans in Ohio: “In their expeditions they live Department of Agriculture Forest Service have compiled chiefly by hunting, or on wild fruits and roots, with which the roughly 1500 references for use of fire in New World eco- woods supply them almost every where” [Smith, 1765, systems management. Documented uses for biomass burning p. 39]. This was long after plagues and warfare decimated include facilitating hunting and gathering, crop management Native American populations.Manyformerlyinhabited (referring to cultivars and wild-managed species), improving areas were then used only for fur trade hunting or passage- growth yields (referring to wild-managed flora and fauna), way. These areas would have been undergoing secondary fireproofing areas, insect collecting, pest management, war- succession for a century or more. This suggests there was fare and signaling, clearing areas for travel and safety, felling persistence in the ecological landesque capital in early colo- trees, and clearing riparian areas (referring to hydrophilic nial North America similar to that in Amazonia (described wild-managed flora and fauna) [Williams, 2001]. above) [Balée, 1993]. 212 HUNTER-GATHERERS, AGRICULTURALISTS, AND CLIMATE

2.3. Archaeology 1993]. It is also becoming evident that one of humanities’ biggest achievements, agriculture, was not “invented” in a Before the advent of archaeology, what was known about few discrete centers and disseminated as once thought. Nor the history of humanity was restricted to texts and authorita- was agriculture a reaction to environmental stressors. Agri- tive pronouncements that covered only a few thousand years. culture emerged in numerous regions with different sets of Working with physical anthropologists and geoscientists, practices and cultivars in each. As archaeological and botan- archaeologists have assembled a history of humanity that ical investigations continue, it is likely that the number of spans millions of years. The materials, technologies, cultural documented regions in which domestication occurred will features, site types, and site distributions cataloged by ar- grow [Harris, 2007; Fuller, 2010]. chaeologists can help to extend what is known through ethnographies and the ethnohistoric literature further in to 2.4. Paleoenvironmental Reconstructions the past. However, geoscientists working with these data should be While ethnologists have demonstrated the effectiveness of familiar with a cardinal rule in archaeology: “the absence of indigenous groups as ecological managers, the technology evidence is not evidence of absence.” Archaeologists can say most capable of generating rapid and large-scale change is what has happened, but they will never be able to say that fire. How and why humans used fire can vary greatly. More that is all that has happened. This is because most places ancient than our species, it is a technology inherited from our have never been surveyed by archaeologists. Further, many hominid ancestors. In regions long occupied by humans, the survey methods can only detect certain kinds of sites (e.g., beginning of human-affected fire regimes is often blurred. In those with aboveground features or durable remains like more recently colonized areas, however, the delineation of stone tools). In addition, ecological features that constitute human-affected fire regimes is much clearer. This is because prime real estate can change slowly: as such, evidence of without humans, the only significant sources for fire are early occupation is often destroyed by later episodes of lightning and volcanoes. The arrival of the Aborigines in occupation. Nevertheless, anyone working in paleoenviron- Australia, the Maoris in New Zealand, the Norse in Iceland mental reconstructions should become versed in the culture and Greenland, the Malagasy in Madagascar, and the Portu- history in their area for potential insights to unexpected guese in Madeira all correlate with charcoal horizons in changes in their data. palynological studies [Pyne, 1998, 2001]. While archaeologists have been studying human-caused Several recent studies using pollen, phytoliths, Sporor- environmental impacts for decades, the degree to which hu- miella, and charcoal analyses suggest many activities de- mans have spurred environmental degradation or reacted to it scribed by ethnographers and, in the historic literature, is debated from site to site. Fine-grained analyses, aided by occurred throughout antiquity. In northeastern Thailand, the advances in analytical methods, are showing that many epi- increased occurrence of burnt phytoliths occurred at a time of sodes of environmental degradation once thought to be the increased pollen from hydrophytic vegetation (circa 10,000 cause of societal collapse, were actually the result of collapse B.P.). This means that as the environment became wetter, the or abandonment [van der Leeuw, 2005; Fisher et al., 2003]. incidents of fire increased. This makes humans the most In many cases, for example, it was not the establishment of likely sources of fire [Kealhofer and Penny, 1998]. Research terraced fields or systems of irrigation that created environ- in Central America had similar findings extending to 11,000 mental shocks; it was the cessation of their maintenance that B.P. [Piperno, 2006]. Biomass burning was also identified in spurred processes like soil degradation or erosion. Civiliza- northern Europe beginning 8000 B.P. [Edwards et al., 2000]. tions have succumbed to massive environmental shocks, like Two investigations of the Pleistocene megafaunal extinc- droughts, and have caused environmental degradation, like tions, using dung fungus Sporormiella as a proxy for mega- deforestation. However, there are no documented cases of fauna, have related the extinctions to changes in fire regimes. civilizations collapsing because of self-inflicted resource At Lynch’s Crater in northeastern Australia, Sporormiella degradation or over population [Tainter, 2006]. counts dropped to almost zero around 41,000 B.P., during a Human responses to environmental shocks cannot be pre- period of climatic stability. The first event was the decline of dicted. A society confronting a drought may collapse, or it the Sporormiella. Approximately a century later, there was a may develop a system of irrigation [Costanza et al., 2007]. marked increase in charcoal. Roughly three centuries after Further, developments like massive irrigation systems can be the decline in Sporormiella, there was an increase in grasses. the product of state-organized works or the aggregation of The shift to sclerophyllous vegetation was complete approx- efforts from agriculturalists working in societies organized imately four centuries after the decline in Sporormiella [Rule around kinship [Erickson, 1993, 2000; Lansing and Kremer, et al., 2012]. CADZOW 213

An investigation of four sites in southeastern New York future fires. Because all places do not burn equally, the result- documented a patchy and discontinuous decline in megafauna ing patchy landscape can foster increases in biodiversity. As that spanned a millennium (circa 11,000 B.P.) and did not with shifting swiddens described above, organisms that vari- proceed in lockstep with documented climatic changes. At ously thrive at different stages of successional habitat would several sites, a modest increase in charcoal preceded the have oases in which to thrive. The long-term use of fire may decline in Sporormiella, and more intensive increase fol- also have assisted with the domestication of certain flora. lowed [Robinson et al., 2005]. The first increase was inter- Many species that would otherwise be restricted to water’s preted as the result of the arrival of humans in the area and the edge or wetlands could spread across the landscape. While second as the result of an increase in available fuel due to the plants do not adapt specifically to fire, they do to a negotiated decline in grazing and trampling [see Owen-Smith, 1987]. set of pressures that include fire regimes. Adaptations to In both of these cases, the biggest increases in charcoal drought, grazing, and fire, for example, include the storing of occurred after the loss of the megafauna. It may be that the a large portion of biomass in root systems and the ability to decline of megafauna, or megaherbivores, spurred changes resprout from damaged stems [Pyne, 1998]. Biomass burning in vegetation that the humans, who had recently arrived in may have been one of many activities that were necessary both locales, found it difficult to cope with. The broadcast precursors for the emergence of agriculture in some regions. use of fire, or biomass burning, to ameliorate these condi- The long-term use of biomass burning, coupled with other tions, may have also pulled these groups across tipping sets of ecological management practices, has the potential to points into new patterns of ecological management. create ecosystems that thrive only with the continuation of As these studies, coupled with insights from ethnohistoric those practices. In the Sierra Nevadas, millennia of ecosys- literature, illustrate, interpretations of charcoal profiles in tems management by Native Americans resulted in a biotic areas occupied by humans must be cautious. A low concen- community that had become dependent on human praxis, tration of charcoal could be the result of either an infrequent- including biomass burning. Modern, rigid management or frequent-fire regime. This is because in a frequent-fire practices were causing an increased risk for catastrophic regime, like that described by Van der Donck above, there fires, loss of biodiversity, and lowered ecosystem resilience may not be enough time between fires for the accumulation [Anderson and Moratto, 1996]. of the fuel necessary to create notable amounts of charcoal in It is also possible that widespread biomass burning can sediments. As such, data from sediments excavated near affect regional weather patterns and climate. An ensemble- historically documented fires, occurring after centuries of fire based model study suggests that the vegetative changes suppression, should not be used to interpret data from sedi- associated with biomass burning in northern Australia could ments that accumulated under frequent-fire regimes. significantly impact regional hydrologic cycles. Specifically, Changes in the frequency of human-affected fire regimes biomass burning can cause decreases in precipitation, higher can result from changes in demography, adoption or aban- near-surface temperatures, and enhance atmospheric stability donment of slash-and-burn agriculture, changes in hunting [Notaro et al., 2011]. practices, migrations, extinctions of herbivores, or reactions The relatively few (compared to a thousand years ago) to a changing environment. An investigation of pre-Columbian places in which biomass burning occurs is still significant and post-Columbian fire regimes in Amazonian savannas enough to attract the attention of agencies like the United reported increased charcoal concentrations occurring after Nations (UN). The UN Assessment of Black Carbon and the collapse of indigenous populations. This was interpreted Tropospheric Ozone noted that black carbon (BC), soot as a post-Columbian increase in fire frequency [Iriarte et al., produced from numerous sources including biomass burn- 2012]. An alternative interpretation is that the collapse of ing, has a short-lived but intense effect on atmospheric indigenous populations reduced the frequency of fires, there- temperature. BC absorbs energy from sunlight and radiates by increasing fuel loads and the corresponding signature in heat in the atmosphere. It also affects snow, ice, and cloud the charcoal record. Perhaps a better strategy is to focus on formations [Hicks et al., 2011]. Simulations suggest that BC relative changes in charcoal concentrations preserved at spe- on the Tibetan Plateau can reduce snow reflectivity by as cific locations and relate them to other sets of data including much as 4% to 6%. This is enough to warm the average the archaeological record. surface air temperature by approximately 1°C [Kopacz et al., 2011]. Such impacts from BC may have been more intensely 2.5. Consequences of Biomass Burning expressed in the past where populations were smaller and fuel for biomass burning larger. Biomass burning changes moisture regimes by exposing While it is tempting to view biomass burning as a one- the surface to the sun and wind, improving conditions for way transferal of carbon from terrestrial reserves to the 214 HUNTER-GATHERERS, AGRICULTURALISTS, AND CLIMATE atmosphere, the development of certain forms of agriculture foster biodiversity [e.g., Balée, 1993, 1994]. Instead of di- included negative feedback. The making of terra preta (i.e., recting and restricting what people do from the top down and black earth) in Amazonia included the addition of large far away, it may be more useful to study and foster their amounts of fire-derived organic material, or biochar (i.e., activities as managers of ecosystems that may be most resil- charcoal created by pyrolysis). The addition of biochar to ient under particular traditions of maintenance. soils increases nutrient retention, ameliorates fluctuations in Similarly, a recent report from the Lake Constance Think precipitation, and is a means of storing carbon in the soil Tank on Global Change and Feedback from Global Carbon [Glaser et al., 2002; Hecht, 2004]. Dynamics was somewhat dismissive on the role of biochar in Working at global scales, paleoclimatologists have sequestering carbon in the soil because there is a 25% loss in cataloged past variations in climate due to periodic changes carbon every century [Preston and Schmidt, 2006; Schmidt et in the Earth’s orbit around the sun, the wobbles in its axis, al., 2011]. Compared to the efficiency with which agricultur- variations in the Earth’s magnetic dipole, and other factors alists can add biochar to the soil, these losses are inconse- that resulted in predictable patterns of change in greenhouse quential. The incorporation of biochar to programs aimed at gases recovered from ice cores. These expected variations increasing food security in developing nations might boost matched the observed trends in greenhouse gasses, but only productivity while mitigating some forms of human-affected until the advent of agriculture and animal husbandry around climate change. the beginning of the Holocene [Ruddiman, 2005]. A meta- analysis of 80 proxy records from around the world found 4. CONCLUSIONS changes in CO2 and temperature to be spatially variable. In most places, the increases in CO2 preceded temperature While it remains to be widely accepted, this cross- increases at the Pleistocene-Holocene transition by several disciplinary investigation suggests that humanity has had the centuries [Shakun et al., 2012]. These findings are in agree- capacity to effect ecological changes including weather and ment with empirical data that illustrate how variation in climate for thousands or even tens of thousands of years. regional-scale land use and weather can affect regional at- Ethnographers working with modern populations have docu- mospheric concentrations of CO2. Indeed, the gradient be- mented the ways in which simple technologies, including tween differently vegetated regions can be four times greater biomass burning, can be used to shape forest composition, than the ocean-continent gradient and 40 times greater than promote biodiversity, and increase the resilience of affected the interhemispheric gradient [Miles et al., 2012]. As human- forest stands [e.g., Balée, 1993, 1994; Peters, 2000; Rival, ity has long had the capacity to affect vegetative composition 1998]. Ethnohistoric literature documents similar activities in and terrestrial reserves of carbon (especially through the use the recent, preindustrial past [e.g., Cartier, 1906, Champlain, of fire), human prehistory should be involved in the interpre- 1878]. Archaeological investigations provide a framework of tation of proxy data, especially during the late Pleistocene sites, features, and technologies related to ethnographically and Holocene. and ethnohistorically documented activities that extend into the deep past. Paleoenvironmental reconstructions using pol- 3. DISCUSSION len, phytolith, Sporormiella, and charcoal concentrations have documented changes in fire regimes that appear to be Understanding the roles hunter-gathers and agriculturalists caused by human agency [Kealhofer and Penny, 1998; Pi- play in their socioecological settings has important implica- perno, 2006; Edwards et al., 2000]. It has also been demon- tions for regional land management as well as for establish- strated that biomass burning in modern times can impact ing responsible climate change policies. The UN report on regional weather patterns, affects near-surface temperatures, BC and tropospheric ozone calls for the “banning of field and affects snow and ice formations [Hicks et al.,2011; burning of agricultural waste” to mitigate short-term human- Kopacz et al., 2011]. It is therefore likely that biomass induced impacts to climate [Hicks et al., 2011, p. 2]. Such a burning played a role in unexpected changes in greenhouse policy, however, would frequently be in contradiction with gasses around the beginning of the Holocene. two of the reports’ stated goals: to increase food security and While there are long-term trends that connect the first decrease biodiversity losses associated with climate change. bipeds to the development of civilizations, the details of First, the activities to be banned are often directly related to prehistory are much more complex and dynamic. It happened food production. Second, many of these practices are ves- in fits and with long periods of stasis. Populations nucleated tiges of ancient land management strategies to which some and dispersed, civilizations waxed and waned, wars were ecosystems may have become adapted [e.g., Anderson and won and lost, and underwriting all of those fantastic events Moratto, 1996]. It is also possible that these practices also was the implementation of the practices related to basic CADZOW 215 subsistence. The true grandeur of prehistory may not be in Erickson, C. L. (1993), The social organization of prehispanic the great sites, the monumental architecture, conflicts, or raised field agriculture in the Lake Titicaca Basin, in Research resolutions, but in the patchy and discontinuous networks of in Economic Anthropology: Supplement, vol. 7, Economic As- disturbance that foster the biodiversity and ecological com- pects of Water Management in the Prehispanic New World, – plexity that has been the result of humanity’s existence. To edited by V. L. Scarborough and B. L. Isaac, pp. 369 426, JAI Press, Greenwich, Conn. rediscover what has been lost of that legacy of constructive Erickson, C. L. (2000), The Lake Titicaca Basin: A pre-Columbian human impacts to environmental systems that, and to, con- built landscape, in Imperfect Balance: Landscape Transforma- serve what remains should be foremost among all of our tions in the Precolumbian Americas, edited by D. L. Lentz, efforts. pp. 311–356, Columbia Univ. Press, New York. Fairhead, J., and M. Leach (1996), Misreading the African Land- REFERENCES scape: Society and Ecology in a Forest-Savanna Mosaic, Cam- bridge Univ. Press, Cambridge, U. K. Anderson, M. K., and M. J. Moratto (1996), Native American land- Fisher, C. T., H. P. Pollard, I. Israde-Alcántara, V. H. Garduño- use practices and ecological impacts, in Sierra Nevada Ecosys- Monroy, and S. K. Banerjee (2003), A reexamination of human- tem Project: Final Report to Congress, vol. II, Assessments and induced environmental change within the Lake Patzcuaro Basin, Scientific Basis for Management Options, pp. 187–206, Univ. of Michoacan, Mexico, Proc. Natl. Acad. Sci. U. S. A., 100(8), Calif. , Cent. for Water and Wildland Resour., Davis. 4957–4962. Balée, W. (1993), Indigenous transformation of Amazonian forests: Fuller, D. Q. (2010), An emerging paradigm shift in the origins of An example from Maranhão, Brazil, L’Homme, 33(126/128), agriculture, Gen. Anthropol., 17(1), 8–12. – 231 254. Glaser, B., J. Lehmann, and W. Zech (2002), Ameliorating Balée, W. (1994), Footprints of the Forest: Ka’apor Ethnobotany – physical and chemical properties of highly weathered soils in The Historical Ecology of Plant Utilization by an Amazonian the tropics with charcoal—A review, Biol. Fert. Soils, 35(4), People, Columbia Univ. Press, New York. 219–230. Bhagwat, S. A., C. G. Kushalappa, P. H. Williams, and N. D. Brown Harris, D. R. (2007), Agriculture, cultivation and domestication: (2005), The role of informal protected areas in maintaining Exploring the conceptual framework of early food production, biodiversity in the Western Ghats of India, Ecol. Soc., 10(1), 8. in Rethinking Agriculture: Archaeological and Ethnographic Blaikie, P., and H. Brookfeld (1987), Land Degradation and Soci- Perspectives, edited by T. Denham, J. Iriarte and L. Vrydaghs, ety, Methuen, London, U. K. pp. 16–35, Left Coast Press, Walnut Creek, Calif. Cadzow, D. M. (2011), Native American culture and practices, in Hecht, S. B. (2004), Indigenous soil management and the Green Society: Toward a Sustainable Future,vol.11,Green creation of Amazonian dark earths: Implications of Kayapo Culture: An A-to-Z Guide, Sage, Los Angeles, Calif., doi:10. practices, in Amazonian Dark Earths: Origin, Properties, 4135/9781412975711. Management, edited by J. Lehmann et al., pp. 355–372, Cartier, J. (1906), First relation of Jaques Carthier of S. Malo, 1534, Springer, New York. in Early English and French Voyages, Chiefly From Hakluyt, Hicks, K., F. Raes, E. Rosenthal, and M. Williams (2011), Integrated 1534–1608, edited by H. S. Burrage, pp. 3–31, Charles Scribner, assessment of black carbon and tropospheric ozone summary for New York. decision makers, edited by B. Ullstein, U. N. Off. at Nairobi, Champlain, S. (1878), Voyages of Samuel de Champlain, vol. 2, Nairobi, Kenya. edited by E. F. Slafter, translated from French by C. P. Otis, Holling, C. S., and L. H. Gunderson (2002), Resilience and adaptive 157 pp., Prince Soc., Boston, Mass. cycles, in Panarchy: Understanding Transformations in Human Costanza, R., L. Graumlich, W. Steffen, C. Crumley, J. Dearing, K. and Natural Systems, edited by L. H. Gunderson and C. S. Hibbard, R. Leemans, C. Redman, and D. Schimel (2007), Sus- Holling, pp. 25–62, Island Press, Washington, D. C. tainability or collapse: What can we learn from integrating the Iriarte, J., M. J. Power, S. Rostain, F. E. Mayle, H. Jones, J. Watling, history of humans and the rest of nature?, Ambio, 36(7), 522–527. B. S. Whitney, and D. B. McKey (2012), Fire-free land use in Day, G. M. (1953), The Indian as an ecological factor in the Pre-1492 Amazonian savannas, Proc. Natl. Acad. Sci. U. S. A., northeastern forest, Ecology, 34(2), 329–346. 109(17), 6473–6478. Diamond, J. (2006), Collapse: How Societies Chose to Fail or Kealhofer, L., and D. Penny (1998), A combined pollen and phy- Succeed, Penguin, New York. tolith record for fourteen thousand years of vegetation change Edwards, K. J., G. Whittington, and R. Tipping (2000), Multiple in Northeastern Thailand, Rev. Palaeobot. Palynol., 103(1–2), charcoal profiles in a Scottish Lake: Taphonomy, fire ecology, 83–93. human impact and inference, Palaeogeogr. Palaeoclimatol. Kopacz, M., D. L. Mauzerall, J. Wang, E. M. Leibensperger, D. K. Palaeoecol., 164(1–4), 67–86. Henze, and K. Singh (2011), Origin and radiative forcing of black Engelbrecht, W. (2003), Iroquoia: The Development of a Native carbon transported to the Himalayas and Tibetan Plateau, Atmos. World, Syracuse Univ. Press, Syracuse, N. Y. Chem. Phys., 11, 2837–2852. 216 HUNTER-GATHERERS, AGRICULTURALISTS, AND CLIMATE

Lansing, J. S., and J. N. Kremer (1993), Emergent properties of Ruddiman, W. F. (2005), Plows, Plagues, and Petroleum: How Balinese water temple networks: Coadaptation on a rugged fit- Humans Took Control of Climate, Princeton Univ. Press, Prince- ness landscape, Am. Anthropol., 95(1), 97–114. ton, N. J. Lewis, H. T. (1973), Patterns of Indian Burning in California; Rule, S., B. W. Brook, S. G. Haberle, C. S. M. Turney, A. P. Ecology and Ethnohistory, Ballena Press, Ramona, Calif. Kershaw, and C. N. Johnson (2012), The aftermath of megafau- Malakoff, D. (2011), Are more people necessarily a problem?, nal extinction: Ecosystem transformation in Pleistocene Austra- Science, 333(6042), 544–546. lia, Science, 335(6075), 1483–1486. Miles, N. L., S. J. Richardson, K. J. Davis, T. Lauvaux, A. E. Schmidt, M. W. I., et al. (2011), Persistence of soil organic matter as Andrews, T. O. West, V. Bandaru, and E. R. Crosson (2012), an ecosystem property, Nature, 478(7367), 49–56. Large amplitude spatial and temporal gradients in atmospheric Shakun, J. D., P. U. Clark, F. He, S. A. Marcott, A. C. Mix, Z. boundary layer CO2 mole fractions detected with a tower-based Liu, B. Otto-Bliesner, A. Schmittner, and E. Bard (2012), network in the U.S. upper Midwest, J. Geophys. Res., 117, Global warming preceded by increasing carbon dioxide con- G01019, doi:10.1029/2011JG001781. centrations during the last deglaciation, Nature, 484(7392), Notaro, M., K.-H. Wyrwoll, and G. Chen (2011), Did aboriginal 49–54. vegetation burning impact on the Australian summer monsoon?, Smith, W. (1765), An Historical Account of the Expedition Against Geophys. Res. Lett., 38, L11704, doi:10.1029/2011GL047774. the Ohio Indians in the Year 1764. Under the Command of Henry Owen-Smith, N. (1987), Pleistocene extinctions: The pivotal role of Bouquet. Facsimile Copy From University Microfilms, Ann Ar- megaherbivores, Paleobiology, 13(3), 351–362. bor (1966), William Bradford, Philadelphia, Pa. Peters, C. M. (2000), Precolumbian silviculture and indigenous Stewart, O. C. (1956), Fire as the first great force employed by man, management of neotropical forests, in Imperfect Balance: Land- in Man’s Role in Changing the Face of the Earth, vol. 1, edited by scape Transformations in the Precolumbian Americas, edited by W. L. Thomas Jr., pp. 115–133, Univ. of Chicago Press, Chicago, D. L. Lentz, pp. 203–224, Columbia Univ. Press, New York. Ill. Piperno, D. R. (2006), Quaternary environmental history and agri- Tainter, J. A. (2006), Archaeology of overshoot and collapse, Ann. cultural impact on vegetation in Central America, Ann. Mo. Bot. Rev. Anthropol., 35(1), 59–74. Garden, 93(2), 274–296. Thompson, W. A., C. S. Holling, D. Kira, C. C. Huang, and I. Preston, C. M., and M. W. I. Schmidt (2006), Black (pyrogenic) carbon: Vertinskf (1979), Evaluation of alternative forest system man- A synthesis of current knowledge and uncertainties with special agement policies the case of the spruce budworm in New Bruns- consideration of boreal regions, Biogeosciences, 3, 397–420. wick, J. Environ. Econ. Manage., 6(1), 51–68. Pyne, S. J. (1998), Forged in fire: History, land, and anthropogenic Van der Donck, A. (1656), Collections of the New-York Historical fire, in Advances in Historical Ecology,editedbyW.Balée, Society, vol. 1, Description of the New Netherlands, N. Y. Hist. pp. 64–103, Columbia Univ. Press, New York. Soc., New York. [Printed for the Society, 1841.] Pyne, S. J. (2001), Fire: A Brief History, Univ. of Wash. Press, Seattle. van der Leeuw, S. E. (2005), Climate, hydrology, land use, and Redman, C. L. (1999), Human Impacts on Ancient Environments, environmental degradation in the Lower Rhone Valley during the Univ. of Ariz. Press, Tucson. Roman Period, C. R. Geosci., 337(1–2), 9–27. Rival, L. (1998), Domestication as a historical and symbolic pro- Williams, G. W. (2001), References on the American Indian Use of cess: Wild gardens and cultivated forests in the Ecuadorian Fire in Ecosystems, USDA For. Serv., Washington, D. C. Amazon, in Advances in Historical Ecology, edited by W. L. Balee, pp. 323–350, Columbia Univ. Press, New York. Robinson, G. S., L. P. Burney, and D. A. Burney (2005), Landscape D. M. Cadzow, Department of Anthropology, State University of paleoecology and megafaunal extinction in Southeastern New New York at Buffalo, 380 MFAC-Ellicott Complex, Buffalo, NY York State, Ecol. Monogr., 75(3), 295–315. 14261-0005, USA. ([email protected]) AGU Category Index

Abrupt/rapid climate change, 13, 35, 75, 89, 107, 157, 163 Hydrology and fluvial processes, 203 Africa, 157 Impact phenomena, cratering, 13 Agent-based models, 27, 107 Impacts of global change, vii, 51 Agricultural systems, 115, 137, 177, 209 Irrigation, 203 Anthropogenic effects, 1, 137, 171, 203 Land cover change, 1, 51 Asia, 89, 97 Middle atmosphere dynamics, 35 Atmospheric transport and circulation, 89 Natural hazards, 43 Biodiversity, 209 New fields, 107, 115 Biosphere/atmosphere interactions, 75 North America, 13 Carbon cycling, 209 Paleoclimatology, 43 Climate dynamics, vii, 75, 115, 145, 163, 209 Paleoclimatology and paleoceanography, 145 Climate variability, 27, 43, 89, 145 Paleoecology, 1, 13, 177 Continental neotectonics, 195 Paleoseismology, 195 Desertification, 51, 157 Quaternary geochronology, 97, 171 Drought, 51 Radioisotope geochronology, 35 Earth system modeling, 163 Regional climate change, vii, 27, 43, 145 Ecosystems, structure and dynamics, 107 Sea level change, 137, 171 Europe, 177 Sedimentary geochronology, 171 Explosive volcanism, 63 Seismicity and tectonics, 195 General or miscellaneous, 27 Soils, 203 Geomorphological geochronology, 97 Soils/pedology, 177 Geomorphology: fluvial, 97 Stable isotope geochemistry, 75, 115 Global change from geodesy, 137 Tectonics and landscape evolution, 195 Greenhouse gases, 1 Tephrochronology, 63 Human impact, vii Volcano/climate interactions, 63 Human impacts, 35 Volcanoclastic deposits, 63 Hydroclimatology, 157

217 Index

Note: Page numbers with f and t refer to figures and tables

4200 B.C. event, 119, 129 anthrosols, irragric, 203–206 1200 B.C. hypothesis, 120 in coastal desert of Peru, 205 definition of, 204 A in High Himalaya, 204–205 18 active tectonism, 195–200 Arabian Sea, d Owater , 90, 91f, 92–93 influences on human development, 195–200 aragonite, 36, 38, 39f Neolithic culture and, 197–200 archaeologists, 1 New Zealand, 197–200 archaeology, 8–9 Adad-nerari, 119 Ashur-Dan II, 119 aeolian sand transport, 54–58 Assur, 117 agricultural system transformation, 115–131 Assyria, 119 changes in political, cultural and economic life, 116 asteroids. See also Younger Dryas impact event city-state networks during Early Bronze Age, 116–117 airburst altitude for, 16f competition between small states in impact probability, 18f Middle Bronze Age, 117–119 environmental change, 119–120 B evidence agricultural system transformation, 123–126 Babylonia, 117–118 globalization, warfare, and political decline during Late Baluchistan, 108f, 109 Bronze Age, 118–119 barley, 118, 122–123, 126–128 mid-late Holocene agricultural system transformations, 115–131 Barringer (Meteor Crater), Arizona, 19 presence and absence of crops, 126–128 Bay of Bengal, 76, 93 agriculturalists, 209–215 Bay of Plenty, 197 archaeology, 212 Belize, 30–31 biodiversity and, 211 biochar, 214 biomass burning, 211, 213–214 biodiversity, 211 definition of, 209 biomass burning, 211–214 ethnohistory, 211 black carbon, 213 paleoenvironmental reconstructions, 212–213 black mat, 14, 19–20, 22 agriculture, long-fallow phase of, 5 blast waves, optimum altitude for, 16 airburst, 16–17 Boserup, Ester, 5–7 Akkad, 117 Brazil, 210 Akkadian Empire, 35 Britain, early forest clearance in, 4 Aldy-Belsk Culture, 150 Bronze Age. See also Indus Valley Civilization Alpine Fault, 197 Early, 116–117, 122–123, 129–131 Altai-Sayan Mountains, 145–152 Late, 118–123, 128, 179 Amazonia, 210 Middle, 117–119, 122–123, 128–131 Amorites, 117 Burj Basket-Marked (BBM) period, 109–112. See also Indus Anatolia, 116 Valley Civilization Ancient Asteroid (documentary), 17 Burj period, 108f

219 220 Index

C Indus Valley Civilization, 107–112 calcitic stalagmite, 78 Lost Sarasvati hypothesis, 97–105 carbon dioxide, 3, 214 Mature Harappan period, 97–105 carbonaceous spherules, 21 Mississippi chiefdoms, 35–40 Cardium edule, 171–174 Naga, Sudan, 163–167 Cardium Pottery Culture, 171–174 Siberian Scythians, 145–152 Caribbean, 43–47 Terminal Classic lowland Maya, 27–32 building resilience in, 47 Ulster, 177–187 climate variability in, 44 comets, 14–19 human experience and impact, 45 airburst altitude for, 16f life cycles of built environment, 45–47 Corossol crater, 20 paleotempestology, 44–45 cotton, 123 pre-Columbian communities, 45–47 craters, 19–20 sea surface temperatures, 44–45 cucumber, 123 Carolina Bays, 15–16 Cycle of Cosmic Catastrophes (Firestone), 15 Cerastoderma glaucum, 172–173 Champa rice, 141 D charcoal, 21 Dansgaard-Oeschger event, 20 Chautang River, 101–102 Dark Age, 116, 118–119, 130 Cherrapunji, India, 76, 78 Dead Sea, 93–94 Chichen Itza, 28 DeSoto Caverns, 36 China Developed Neolithic, 109 early forest clearance in, 6 Domesday Book, 4 Fuzhou Basin, 137–142 droughts per capita land use in, 4 civilization collapse and, 212 reforestation in, 8 in Egyptian Sahara, 157–162 Chicxulub crater, 19 Fertile Crescent, 122–124 city-states, 116–117 late Holocene, 89–95 climate change Maya collapse and, 30–32 Caribbean island communities and, 43–47 Navajo Nation, 51–59 as civilizing factor in Egyptian Sahara, 157–162 in NE India at 4 kyr B.P., 82 Egyptian Sahara, 157–162 subsidence and, 93 Fertile Crescent, 119–120 droughts, Holocene, 89–95 18 Fuzhou Basin, 137–142 d Owater in Arabian Sea mixed layer, 92 Indian monsoon, 75–84 Gulf of Aqaba water column stratification, 92 late Holocene drought, 89–95 Indus River discharge and, 90–92 Mississippi chiefdoms rise-collapse and, 35–40 mid to late Holocene climate and cultural change, 93–95 Naga, Sudan, 163–167 proxy records, 91f Navajo Nation arid landscape and, 51–59 subsidence and, 93 in NE India at 4 kyr B.P., 82–83 West Asia, 92–95 Siberian Scythian population dynamics and, 145–152 dung fungus (Sporormiella), 20, 212–213 in West Asia over last 5000 years, 92–95 Younger Dryas impact event, 20 E Youngest Toba Tuff tephra fall deposits and, 63–72 Early Bronze Age (EBA), 116–117, 122–123, 129–131 climate variability, 44 Early Neolithic, 109, 171–174 Clovis culture, collapse of, 13–20 Ebla, 117 collapse, 28–30 Egyptian Sahara, 157–162 Cardium Pottery Culture, 171–174 Holocene droughts in, 157–162 Clovis culture, 13–20 Holocene Neolithic pluvials at Nabta Playa, 158–161 Fertile Crescent, 115–131 hydroclimatic setting, 158 Harappan civilization, 107–112 late Quaternary air temperatures, 159f Index 221

Egyptian Sahara (continued) global population, pre-industrial, 3f reconstructions of late Quaternary air temperatures, 159f Gloucester Harbor, Massachusetts, 211 water-centric prehistoric patterns, 158 Great Hafir, 164, 165f, 167 El Niño–Southern Oscillation (ENSO), 82, 93, 165 Great Lakes, 19 Ela, 117 greenhouse gases, Holocene period, 3, 214 Emain Macha, 183, 184f Greenland Ice Sheet Project 2 (GISP2), 65, 159f emmer wheat, 125 Guajá Indians, 210 Enceladus, 117 Gulf of Aqaba, water column stratification, 92 Ešnunna, 117 Europe, forest clearance vs. population in, 5f H extinctions, 20 Hakra-Kechi, 108f Hakra Ware complex, 109 F Han Dynasty, 140–141 Fertile Crescent, 115–131 Hana, 118 changes in political, cultural and economic life, 116 Harappan civilization, 107–112 city-state networks during Early Bronze Age, 116–117 deurbanization of, 83 competition between small states in Middle Bronze Age, and Ghaggar floodplain, 97–105 117–119 Lost Sarasvati hypothesis, 97–105 environmental change, 119–120 Mature Harappan period, 97–105 evidence agricultural system transformation, 123–126 moonsoon and, 76 globalization, warfare, and political decline during Late pre-Harappan Neolithic, 107–112 Bronze Age, 118–119 Haryana Plain, 99 mid-Holocene geoarchaeological phenomena, 124t Hatti, 118 mid-late Holocene agricultural system transformations, Haughey’s Fort, 183 115–131 Heinrich event, 20 presence and absence of crops, 126–128 High Himalaya, irragric anthrosols in, 204–205 floodplains, 99–101 Hittite, 117–118 forest clearance Hochelaga, Quebec, 211 footprint of abandoned land, 6f hollow ways, 125 vs. population, 4, 5f Holocene climate, 44 simulations of, 6, 7f Holocene droughts, 89–95. See also climate change forest transition, 7t Arabian Sea sediment core chronologies, 92 fullerenes, 22 as civilizing factor in Egyptian Sahara, 157–162 Fuzhou Basin (Fujian, China), 137–142 climate change in West Asia over last 5000 years, 92–95 18 anthropogenic contribution to environmental change in, 141 d Owater in Arabian Sea mixed layer, 92 late Holocene evolution of, 137–142 Gulf of Aqaba water column stratification, 92 map of, 138f–139f Indus River discharge and, 90–92 modern environmental setting, 139 mid to late Holocene climate and cultural change, 93–95 paleoenvironmental model, 139–141 proxy records, 91f spread of rice farming in, 137–142 subsidence and, 93 text-based geographic history of, 140 Holocene Optimum, 166 Hopi reservation, 52f G Hortonian overland flow, 166 Gaelic Scotland, 181–182 Huaorani, 210 Ghaggar-Hakra Valley, 98 Hudson Bay, 15 Ghaggar River, 97–105. See also Indus Valley Civilization hunter-gatherers, 209–215 ASTER GDEM data, 100 archaeology, 212 flood plain, 99–101 biodiversity and, 211 flow regime, 101–104 biomass burning, 211, 213–214 global circulation models (GCMs), 163, 165 definition of, 209 global land use and technological simulator (GLUES), 109–110 ethnography, 210 222 Index

hunter-gatherers (continued) irragric anthrosols, 203–206 ethnohistory, 211 in coastal desert of Peru, 205 paleoenvironmental reconstructions, 212–213 definition of, 204 hurricanes, 43–47 in High Himalaya, 204–205 hydrological model, 166 irrigation agriculture, 203–206 island communities, 43–47 I building resilience in, 47 impact markers, 19–20 climate variability and, 44 India human experience and impact, 45 early forest clearance in, 6 life cycles of built environment in, 45–47 Indus Valley Civilization, 75–84 paleotempestology, 44–45 paleoclimate in NE India at 4 kyr B.P., 75–84 pre-Columbian communities, 45–47 Indian monsoon, 75–84. See also climate change sea surface temperatures and, 44–45 during the 20th century, 75 isotope-enabled general circulation model (IsoGSM), 80 age model, 78–79 climate changes in NE India at 4 kyr B.P., 82–83 J KM-A isotope record, 77f Jin Dynasty, 140 paleoclimatic records, 76 societal and dynamical implications, 83–84 K speleothem proxies, 79–81 Ka’apor Indians, 210 Indian Ocean, 75 Kagbeni, Nepal, 204–205 Indian Ocean Dipole (IOD), 82 Kaharoa Tephra Formation, 197 Indo-Pacific Warm Pool, 82 Kali Gandaki River, 204 Indus Google Gazetteer, 110 Kalibangan, 98 Indus River, 99 Kalio Gandaki Valley, 204 discharge estimation, 90–92 Karkemish, 128 Indus culture along, 95 karst, 36 source region of, 94 Kili Ghul Mohammad (KGM) period, 109–112 Indus Valley, 97 Kili period, 108f annual precipitation, 103f kurgans, 147 climate change and, 75–84 deurbanization of, 75–84 L map of, 98f Lamanai, 31 simulation of Neolithic transition in, 107–112 land use, 6f Indus Valley Civilization (IVC), 107–112. See also collapse population and, 5t deurbanization of, 75–84 preindustrial records of, 3–8 Lost Sarasvati hypothesis, 97–105 landesque capital, 210 moonsoon and, 75–84 larch (Larix sibirica), 147 pre-Harappan Neolithic, 107–112 Larsa, 117 intertropical convergence zone (ITCZ), 82, 119, 165 Last Extinction, The (documentary), 15 Ireland, 177–187 Late Archaic period, 39–40 Celtic cattle complex, 181 Late Bronze Age (LBA), 118–123, 128, 179 environmental archaeology, 179–181 Laurentide Ice Sheet, 14–15, 17 militarization, 181 Leopold, Luna, 1–2 paleoclimatic reconstructions, 179–181 linseed, 127 prehistoric Ulster, 183–185 Little Ice Age (LIA), 182 protohistoric Ulster, 183–185 long-fallow phase, of early agriculture, 5 rapid climate change in, 179 lonsdaleite, 21–23 state and nature, 181–183 Lost Sarasvati hypothesis, 97–105. See also collapse iridium, 21 flood plain, 99–101 Iron Age, 184 flow regime on Ghaggar River, 101–104 Index 223

Lost Sarasvati hypothesis (continued) droughts and, 40 geological and geomorphological outline of area, 99 Mittani, 118 water flow of major rivers, 99–101 monsoon, Indian, 75–84 during the 20th century, 75 M age model, 78–79 magnetic grains, with iridium, 21 climate changes in NE India at 4 kyr B.P., 82–83 magnetic microspherules, 21 KM-A isotope record, 77f Maori, 197–200 paleoclimatic records, 76 Mari, 117–118, 128 societal and dynamical implications, 83–84 Mature Harappan period, 97–105 speleothem proxies, 79–81 flood plain, 99–101 Mopan Valley, Guatemala, 28 flow regime on Ghaggar River, 101–104 Mount Pinatubo, Philippines, 65 geological and geomorphological outline of area, 99 Muktinath Valley, 204–205 water flow of major rivers, 99–101 N Mawmluh Cave, 78, 83 Nabta Playa, 157–162 Maya civilization Naga (Sudan) rainfall, 163–167 chronology of, 28t hydrological model, 166 dynamic model of, 28 long-term trends, 166 major sites, 29f magnitude-frequency distributions for recurrence, 166f Maya collapse, 27–32. See also collapse recurrence interval, 166 adaptation and, 30 statistical downscaling, 165 agency and, 30 time slice experiments, 165–166 definition of, 28–30 nanodiamonds, 22–23 diversity of Maya civilization and, 28 Navajo Nation, 51–59 drought and, 30 aeolian sand transport in, 54–58 environment and, 30 map of, 52f proximate causes and responses, 31 population, 53 Terminal Classic, 28, 30–31 precipitation, 54 Medieval Climate Anomaly, 82 sand dunes in, 53, 54f Medieval Warm Period, 179 vegetation in, 51–58 Mediterranean Neolithic, 171–174 Navan Complex, 183, 184f Merritts, D., 1 Navan Fort, 183, 184f Mesolithic-Neolithic transition, 110–111 near-Earth objects (NEOs), 17–19 Mesopotamia, 116 Neolithic culture, in Egyptian Sahara, 157–162 forest clearance in, 6 Neolithic Revolution, 171 methane, 3, 137 Neolithic transition, 107–112 microspherule, magnetic, 21 New Zealand, 197–200 Middle Bronze Age (MBA), 117–119, 122–123, 128–131 dominant tectonic influences, 198f Middle Chronology, 116 Maori culture, 197–200 Middle Elamite Kingdom, 118 Neolithic culture, 197–200 Milankovic theory, 163 tectonism, 197–200 millponds, 1–3 tsunamis, 199–200 mills, 1–3 Nile Valley, 163–164 Min River, 138 North Atlantic Oscillation (NAO), 93 Ming Dynasty, 140 North Island, New Zealand, 197 Minusink Basin, 150 Mississippian chiefdoms, rise-fall of, 35–40. O See also collapse Okataina Volcanic Centre, 197, 198f archaeological records, 36 Old Hittite kingdom, 117 climate change and, 36, 40 olive, 126–127 DeSoto Caverns, 36–40 optically stimulated luminescence (OSL) dating, 102, 104t 224 Index

Orion, 160 statistically downscaled, 165f oxygen isotope time series, 38–40 time slice experiments, 165–166 Punjab Plain, 99, 100f, 101, 109 P paleoclimate, 20–21 Q Caribbean, 43–47 Qatna, 117–118, 123 Egyptian Sahara, 157–162 Qing Dynasty, 140 Fertile Crescent, 119–120 Quaternary Period, 159f Fuzhou Basin, 137–142 Qunf cave, 84 Indian monsoon 4000 years ago, 75–84 late Holocene drought, 89–95 R Mississippi chiefdoms rise-collapse and, 35–40 Rackham, Oliver, 4 Naga, Sudan, 163–167 radiocarbon dating, 23 Navajo Nation arid landscape and, 51–59 rainfall. See also climate change in NE India at 4 kyr B.P., 75–84 annual cycle precipitation, 165f Siberian Scythian population dynamics and, 145–152 hydrological model, 166 in West Asia over last 5000 years, 92–95 Indian monsoon shift 4000 years ago, 75–84 Younger Dryas impact event and, 13–24 long-term trends, 166 Youngest Toba Tuff tephra fall deposits and, 63–72 Mississippi chiefdoms rise-collapse and, 35–40 paleoclimate scientists, 1 Naga, Sudan, 163–167 paleoecology, 20–21 recurrence interval, 166 paleogeography, transgressive, 171–174 statistical downscaling, 165 paleotempestology, 44–45 statistically downscaled precipitation, 165f Palpa Valley, 205 time slice experiments, 165–166 Pazyryk Culture, 145–152. See also collapse Rakhigarhi, 98 archaeological artifacts, 146 rapid climate change (RCC), 119 burials, 146–148 Red Sea, 92–93 demography of, 146 Red Sea stratification, 89–95.See also climate change kurgans, 147–148 Arabian Sea sediment core chronologies, 92 population, 152 as civilizing factor in Egyptian Sahara, 157–162 summer temperature change, 149–152 climate change in West Asia over last 5000 years, 92–95 18 tree ring chronologies, 148–149 d Owater in Arabian Sea mixed layer, 92 Peru, irragric anthrosols in, 205 Gulf of Aqaba water column stratification, 92 Petexbatum region, Guatemala, 28 Indus River discharge and, 90–92 Polynesians, 197 mid to late Holocene climate and cultural change, 93–95 pomegranate, 123 proxy records, 91f pre-Columbian communities, 45–47 subsidence and, 93 pre-Harappan Neolithic, 107–112 reforestation, 7–8 artifacts, 110 rice farming, spread of, 137–142 land use, 109–110 Rio Grande de Nazca, 205 simulated transition, 110–112 Roche limit, 17 technological evolution, 109–110 Roman warm period, 179 precipitation, 163–167. See also climate change Rossby waves, 93–94 annual cycle, 165f Russian Altai, 145–152 hydrological model, 166 Russian thistle (Salsola spp.), 53 Indian monsoon shift 4000 years ago, 75–84 long-term trends, 166 S magnitude-frequency distributions for recurrence of, 166f Sahara, Egyptian, 157–162 Naga, Sudan, 163–167 Holocene droughts in, 157–162 recurrence interval, 166 Holocene Neolithic pluvials at Nabta Playa, 158–161 statistical downscaling, 165 hydroclimatic setting, 158 Index 225

Sahara, Egyptian (continued) Tell Beydar, 116 late Quaternary air temperatures, 159f Tell Brak, 116 reconstructions of late Quaternary air temperatures, 159f Tell Hariri/Mari, 116 water-centric prehistoric patterns, 158 Tell Jerablush, 128 Salsola spp., 53 Tell Khuera, 116 sand dunes, 54f Tell Leilan, 117 sand transport, in Navajo Nation, 54–58 Tell Mardikh/Ebla, 116 sawmills, 1–3 Tell Mozan/Urkesh, 117 Scotland, early forest clearance in, 4 Tell Shiukh Fawqani, 123 Sea Peoples, 118 Tembé Indians, 210 sea surface temperature (SST), 44–45, 76, 82 tephra fall deposits, 63–72 sesame, 123, 126–129 size characteristics, 66–67 Sheri Khan Tarakai (SKT), 109 thickness variations, 67–69 Siberian Scythians, 145–152. See also climate change volume of coignimbrite ashfall, 67–70 archaeological artifacts, 146 Terminal Classic collapse, 27–32. See also collapse burials, 146–148 adaptation and, 30 demography of, 146 agency and, 30 kurgans, 147–148 definition of, 28–30 summer temperature change, 149–152 diversity of Maya civilization and, 28 tree ring chronologies, 148–149 drought and, 30 Sierra Nevada, California, 213 environment and, 30 Sirius, 160 proximate causes and responses, 31 Siwalik Formation, 99 terra preta, 214 Song Dynasty, 140 Thakkhola, 204 soot, 21, 213 Thar Desert, 99, 100f, 101, 102f South Island, New Zealand, 197 Tibetan Plateau, 213 speleothem records, 76, 79–82, 84 Tiglat-Pileser III, 119 spherules, carbonaceous, 21 time slice experiments, 165–166 Sporormiella, 212–213 Toba supereruption (Sumatra, Indonesia), 63–72 stalagmite (DSSG-4), 36–40 climatic impact, 65 stratigraphy, 20 environmental impact, 65 subsidence, 93 size characteristics of distal fallout, 66–67 Sudan, climate-archaeological study, 163–167 sulfur spikes, 71f sulfate aerosol, 70 tephra fall deposits, 63–72 supereruption of Toba (Sumatra, Indonesia), 63–72 thickness variations in fallout, 67–69 environmental impact, 65 volume of coignimbrite ashfall, 67–70 size characteristics of distal fallout, 66–67 Togau, 109 sulfur spikes, 71f transgressive paleogeography, 171–174 tephra fall deposits, 63–72 tree rings, 147–148 thickness variations in fallout, 67–69 Troy, 123 supernova, 15 TSAP (software), 147 Sutlej River, 98 tsunamis, 199–200 swiddens, 210 Tunguska event, 17, 19 Syria, 116 U T Uí Néill dynasty, 181 Tang Dynasty, 140 Ulster, 177–187. See also collapse Tanshishan culture, 138 Celtic cattle complex, 181 tectonism, 195–200 environment as agent, 183–185 influences on human development, 195–200 environmental archaeology, 179–181 New Zealand, 197–200 historical socioecology of, 178–179 226 Index

Ulster (continued) Younger Dryas impact event, 13–24 Iron Age, 184 archaeology, 20 map of, 178f basis of impact hypothesis, 14 militarization, 181 climate change, 20 paleoclimatic reconstructions, 179–181 earliest form of impact hypothesis, 14 prehistoric, 183–185 evidence-based arguments against, 15, 19–23 protohistoric, 183–185 extinctions, 20 rapid climate change in, 179 fundamental flaws of, 15–19 state and nature, 181–183 hypothesis variants, 14 Upper Mesopotamia, 116–118 inconsistencies and contradictions, 15–16 Urshu, 117 infinitesimal likelihood, 17–18 irreproducibility and misinterpretations of evidence, 21–24 V no impact markers or crater, 19–20 vegetation, in Navajo Nation, 51–58 nonunique and misinterpreted events, 20–21 volcanic winter, 65 paleoclimate/paleoecology, 20–21 Voronoi tessellation, 69–72 physical implausibility, 16–17 radiocarbon dating, 23 W stratigraphy, 20 Wales, early forest clearance in, 4 Youngest Toba Tuff (YTT), 63–72 Walter, R., 1 climatic impact, 65 water stratification, 89–95 environmental impact, 65 wheat, 126–127 size characteristics of distal fallout, 66–67 William the Conqueror, 3 sulfur spikes, 71f wine, 128 tephra fall deposits, 63–72 Wolman, Gordon, 1–2 thickness variations in fallout, 67–69 Woodland period, 36, 39–40 Toba supereruption, 64–65 volume of coignimbrite ashfall, 67–70 Y Yamkhad (Aleppo), 117–118 Z Yamuna River, 98 Zamua, 128 Yangtze River Delta, 138 Zimri-Lim, 117