Forthcoming Current Anthropology Wenner-Gren Symposium Current Anthropology Supplementary Issues (in order of appearance)

VOLUME 58 SUPPLEMENT 16 AUGUST 2017 Human Colonization of Asia in the Late Pleistocene. Christopher J. Bae, Current Katerina Douka, and Michael J. Petraglia, eds. The Anthropology of Corruption. Sarah Muir and Akhil Gupta, eds. Cultures of Militarism. Catherine Besteman and Hugh Gusterson, eds. Anthropology

Previously Published Supplementary Issues

Working Memory: Beyond Language and Symbolism. Thomas Wynn and THE WENNER-GREN SYMPOSIUM SERIES Frederick L. Coolidge, eds.

Engaged Anthropology: Diversity and Dilemmas. Setha M. Low and August 2017 Sally Engle Merry, eds. FIRE AND THE GENUS HOMO Corporate Lives: New Perspectives on the Social Life of the Corporate Form. Damani Partridge, Marina Welker, and Rebecca Hardin, eds. GUEST EDITORS: DENNIS M. SANDGATHE AND FRANCESCO BERNA The Origins of Agriculture: New Data, New Ideas. T. Douglas Price and Ofer Bar-Yosef, eds. Fire and the Genus Homo The Biological Anthropology of Living Human Populations: World Histories, Recognizing Fire in the Paleolithic Archaeological Record National Styles, and International Networks. Susan Lindee and

Volume 58 Experimental Approaches to Archaeological Fire Features and Their Ricardo Ventura Santos, eds. Behavioral Relevance Human Biology and the Origins of Homo. Susan Antón and Leslie C. Aiello, eds. Evidence of Burning from Bushfires in Southern and East Africa and Potentiality and Humanness: Revisiting the Anthropological Object in Its Relevance to Hominin Evolution Contemporary Biomedicine. Klaus Hoeyer and Karen-Sue Taussig, eds. Ethnoarchaeology of Paleolithic Fire: Methodological Considerations Alternative Pathways to Complexity: Evolutionary Trajectories in the Middle Aboriginal Use of Fire in a Landscape Context: Investigating Presence and Paleolithic and Middle Stone Age. Steven L. Kuhn and Erella Hovers, eds. Absence of Heat-Retainer Hearths in Western New South Wales, Australia Supplement 16 Crisis, Value, and Hope: Rethinking the Economy. Susana Narotzky and Researching the Nature of Fire at 1.5 Mya on the Site of FxJj20 AB, Koobi Niko Besnier, eds. Fora, Kenya, Using High-Resolution Spatial Analysis and FTIR Spectrometry The Anthropology of Christianity: Unity, Diversity, New Directions. Joel Robbins Spatial Analysis of Fire: Archaeological Approach to Recognizing Early Fire and Naomi Haynes, eds. Evidence of Hominin Use and Maintenance of Fire at Zhoukoudian Politics of the Urban Poor: Aesthetics, Ethics, Volatility, Precarity. Veena Das and How Did Hominins Adapt to Ice Age Europe without Fire? Shalini Randeria, eds. Technologies for the Control of Heat and Light in the Vézère Valley The Life and Death of the Secret. Lenore Manderson, Mark Davis, and Aurignacian Chip Colwell, eds. Control of Fire in the Paleolithic: Evaluating the Cooking Hypothesis Reintegrating Anthropology: From Inside Out. Agustin Fuentes and Fire for a Reason: Barbecue at Middle Pleistocene Qesem Cave, Israel Polly Wiessner, eds.

Pages S163–S370 Neanderthal Cooking and the Costs of Fire New Media, New Publics? Charles Hirschkind, Maria José A. de Abreu, and Savanna Chimpanzees at Fongoli, Senegal, Navigate a Fire Landscape Carlo Caduff, eds. Toward a Long Prehistory of Fire Current Anthropology is sponsored by The Identifying and Describing Pattern and Process in the Evolution of Wenner-Gren Foundation for Anthropological Hominin Use of Fire Research, a foundation endowed for scientific, educational, and charitable purposes. The Foundation, however, is not to be understood as endorsing, by virtue of its financial support, any of Sponsored by the Wenner-Gren Foundation for Anthropological Research the statements made, or views expressed, herein.

T H E U N I V E R S I T Y O F C H I C A G O P R E S S Wenner-Gren Symposium Series Editor: Leslie Aiello Wenner-Gren Symposium Series Managing Editor: Laurie Obbink Current Anthropology Editor: Mark Aldenderfer Current Anthropology Managing Editor: Lisa McKamy Book Reviews Editor: Holley Moyes Corresponding Editors: Claudia Briones (IIDyPCa-Universidad Nacional de Río Negro, Argentina; [email protected]), Michalis Kontopodis (Humboldt Universität zu Berlin, Germany; [email protected]), José Luis Lanata (Universidad Nacional de Río Negro San Carlos de Bariloche, Argentina; [email protected]), David Palmer (Hong Kong University, China; [email protected]), Anne de Sales (Centre National de la Recherche Scientifique, France; [email protected]), Zhang Yinong (Shanghai University, China; [email protected])

Please send all editorial correspondence to (877) 705-1878. Fax: (773) 753-0811 or toll-free (877) 705- Mark Aldenderfer 1879. E-mail: [email protected]. School of Social Sciences, Humanities, and Arts University of California, Merced Reasons of practicality or law make it necessary or desirable 5200 North Lake Road to circulate Current Anthropology without charge in certain Merced, CA 95343, U.S.A. portions of the world; it is hoped, however, that recipients of (fax: 209-228-4007; e-mail: [email protected]) this journal without charge will individually or collectively in various groups apply funds or time and energy to the world Individual subscription rates for 2017: $79 print + elec- good of humankind through the human sciences. Information tronic, $47 print-only, $46 e-only. Institutional subscription concerning applicable countries is available on request. rates are tiered according to an institution’s type and research output: $346 to $727 print + electronic and $301 to $632 q 2017 by The Wenner-Gren Foundation for Anthropological e-only. Institutional print-only is $397. Additional taxes or post- Research. All rights reserved. Current Anthropology (issn age for non-US subscriptions may apply. For additional rates, 0011-3204) is published bimonthly in February, April, June, including prices for full-run institutional access and single cop- August, October, and December by The University of Chicago ies, visit www.journals.uchicago.edu/journals/ca/about. Free Press, 1427 East 60th Street, Chicago, IL 60637-2954. or deeply discounted access is available in most developing na- Periodicals postage paid at Chicago, IL, and at additional tions through the Chicago Emerging Nations Initiative (www mailing offices. Postmaster: Send address changes to .journals.uchicago.edu/ceni). Current Anthropology, P.O. Box 37005, Chicago, IL 60637.

Please direct subscription inquiries, back-issue requests, and address changes to the University of Chicago Press, Jour- nals Division, P.O. Box 37005, Chicago, IL 60637. Telephone: (773) 753-3347 or toll-free in the United States and Canada Current Anthropology Volume 58 Supplement 16 August 2017

Fire and the Genus Homo

Leslie C. Aiello Fire and the Genus Homo Wenner-Gren Symposium Supplement 16 S163 Dennis M. Sandgathe and Francesco Berna Fire and the Genus Homo An Introduction to Supplement 16 S165 Paul Goldberg, Christopher E. Miller, and Susan M. Mentzer Recognizing Fire in the Paleolithic Archaeological Record S175 Vera Aldeias Experimental Approaches to Archaeological Fire Features and Their Behavioral Relevance S191 J. A. J. Gowlett, J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina Evidence of Burning from Bushfires in Southern and East Africa and Its Relevance to Hominin Evolution S206 Carolina Mallol and Auréade Henry Ethnoarchaeology of Paleolithic Fire: Methodological Considerations S217 Simon J. Holdaway, Benjamin Davies, and Patricia C. Fanning Aboriginal Use of Fire in a Landscape Context: Investigating Presence and Absence of Heat-Retainer Hearths in Western New South Wales, Australia S230 Sarah Hlubik, Francesco Berna, Craig Feibel, David Braun, and John W. K. Harris Researching the Nature of Fire at 1.5 Mya on the Site of FxJj20 AB, Koobi Fora, Kenya, Using High-Resolution Spatial Analysis and FTIR Spectrometry S243 Nira Alperson-Afil Spatial Analysis of Fire: Archaeological Approach to Recognizing Early Fire S258

http://www.journals.uchicago.edu/CA Xing Gao, Shuangquan Zhang, Yue Zhang, and Fuyou Chen Evidence of Hominin Use and Maintenance of Fire at Zhoukoudian S267 Harold L. Dibble, Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Shannon P. McPherron, and Dennis M. Sandgathe How Did Hominins Adapt to Ice Age Europe without Fire? S278 Randall White, Romain Mensan, Amy E. Clark, Elise Tartar, Laurent Marquer, Raphaëlle Bourrillon, Paul Goldberg, Laurent Chiotti, Catherine Cretin, William Rendu, Anne Pike-Tay, and Sarah Ranlett Technologies for the Control of Heat and Light in the Vézère Valley Aurignacian S288 Richard Wrangham Control of Fire in the Paleolithic: Evaluating the Cooking Hypothesis S303 Ran Barkai, Jordi Rosell, Ruth Blasco, and Avi Gopher Fire for a Reason: Barbecue at Middle Pleistocene Qesem Cave, Israel S314 Amanda G. Henry Neanderthal Cooking and the Costs of Fire S329 Jill D. Pruetz and Nicole M. Herzog Savanna Chimpanzees at Fongoli, Senegal, Navigate a Fire Landscape S337 Michael Chazan Toward a Long Prehistory of Fire S351 Dennis M. Sandgathe Identifying and Describing Pattern and Process in the Evolution of Hominin Use of Fire S360 Current Anthropology Volume 58, Supplement 16, August 2017 S163

Fire and the Genus Homo Wenner-Gren Symposium Supplement 16

Leslie C. Aiello

“Fire and the Genus Homo” was the 152nd symposium in the Homo, drawing on evidence from both the fossil record and Wenner-Gren series and the sixteenth open-access supplement human biology, and Kuhn and Hovers (2013) explored human of the Foundation’s journal, Current Anthropology. The sym- evolutionary trajectories in the Middle Paleolithic and the Mid- posium was organized by Dennis M. Sandgathe and Francesco dle Stone Age. There is also a forthcoming issue on the human Berna (Simon Fraser University) and was held October 16–22, colonization of Asia in the Late Pleistocene (Bae, Douka, and 2015, at Tivoli Palácio de Seteais in Sintra, Portugal (fig. 1). Petraglia, forthcoming). Wenner-Gren symposia are a long-standing tradition of the Sandgathe and Berna (2017) provide many reasons to focus Foundation, having been initiated in 1958 at the Foundation’s on fire at this particular point in time. There has been an im- European conference center, Burg Wartenstein castle, Austria. pressive upsurge in research into the evolution of human fire Many of the edited volumes resulting from those meetings be- usage over the past decade. In addition, some firmly embedded came landmark contributions to the discipline. However, over preconceptions and popular theoretical models (e.g., the cook- the years, as the field grew in both size and diversity and op- ing hypothesis) generate considerable debate and discussion. portunities for publication increased, edited volumes lost much For example, Sandgathe and Berna point out that although of their earlier cachet. In 2010 we introduced the open-access many researchers believe that by the Middle Paleolithic, if not supplementary issues of Current Anthropology specifically to earlier, hominins regularly made and used fire, there is no ro- raise the visibility of Wenner-Gren symposia. By the end of bust evidence to support this before the appearance of ana- 2016 our symposium supplementary issues had been accessed tomically modern humans (see also Dibble et al. 2017; but see a total of 900,000 times with an average of 70,000 accesses per Henry 2017). This leaves considerable controversy over the issue. We could not be happier with this success. significance of evidence for fire (or lack thereof ) in the earlier “Fire and the Genus Homo” follows in our tradition of sup- hominin sites in Africa, the Near East, and Europe. The debate porting symposia on big-issue topics. In their introduction, surrounding the significance of cooking to human adaptation Sandgathe and Berna, the guest editors, express surprise that and brain evolution by approximately 1.5 million years ago is Wenner-Gren has not sponsored prior symposia on fire. From part of this discussion (Wrangham 2017). Why is the evidence the Foundation’s point of view, this is not surprising. As Sand- for fire usage at this stage so sparse and ambiguous when it gathe and Berna (2017) say, until recently fire was both under- would appear to be fundamentally important to the evolution studied and considered by many to be without problem or major of Homo? interest. “Fire and the Genus Homo” addresses these issues and many The Foundation has, however, sponsored many meetings on more. It provides the most current and up-to-date summary human evolution and human adaptation. In the early years and discussion of the evidence for and debates surrounding the there were a number of meetings that focused primarily on the human use of fire. For this reason alone it will be a go-to ref- then-emerging East African fossil evidence for human evolu- erence for anyone interested in fire and its importance to hu- tion (e.g., Bishop and Clark 1967; Butzer and Isaac 1975; Coppen man evolution. Sandgathe and Berna (2017) conclude by call- et al. 1976; Washburn 1964). Subsequent symposia have em- ing for more collaborative work on the evolution of human use phasized both the evolution of human adaptation and the em- of fire and for new analytical methods and skills to confidently pirical basis for inference. For example, Gibson and Ingold interpret the archaeological record. There are still many un- (1993) addressed the relationship between tools, language, and answered questions to be addressed, but no one can doubt the cognition in human evolution, Enfield and Levinson (2006) ex- importance of fire to humans. Sometimes the most obvious amined the roots of human sociality, and Wynn and Coolidge aspects of life are those that are taken for granted and receive (2010) looked at the working memory in relation to the evo- the least scientific attention. It is exciting to see the growing lution of the brain, language, and symbolism. Antón and Aiello anthropological interest in fire and to speculate on how many (2012) addressed the evolution of adaptation in the genus of our current preconceptions will be overturned by future evi- dence. This is exciting research and certainly a big issue in the Leslie C. Aiello is President of the Wenner-Gren Foundation for modern discipline. Anthropological Research (470 Park Avenue South, 8th Floor North, Wenner-Gren symposia are partnerships between the Foun- New York, New York 10016, USA [[email protected]]). dation and the academic organizers, and we are always looking q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0001$10.00. DOI: 10.1086/692275 S164 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 1. Participants in the symposium “Fire and the Genus Homo.” Front row, from left: Laurie Obbink (Wenner-Gren), Sarah Hlubik, Meg Thibodeau, Vera Aldeias, Carolina Mallol, Ran Barkai, Xing Gao. Middle row: Nira Alperson-Afil, Leslie Aiello (Wenner-Gren), Simon Holdaway, Amanda Henry, Michael Chazan, Jill Pruetz, Paul Goldberg. Top row: John Gowlett, Richard Wrangham, Harold Dibble, Randall White, Dennis Sandgathe (organizer), Francesco Berna (organizer), Fatima Pinto (Wenner-Gren). A color version of this figure is available online.

for new and important ideas from all branches of anthropology raphy, paleoecology, and evolution. Prehistory Archeology and Ecology Series. Chicago: University of Chicago Press. for future symposia and eventual CA publication. Please contact Dibble, Harold L., Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Shan- us with your proposals. Information about the Wenner-Gren non P. McPherron, and Dennis M. Sandgathe. 2017. How did hominins adapt Foundation, the symposium program, and what constitutes a good to Ice Age Europe without fire? Current Anthropology 58(suppl. 16):S278– ’ S287. symposium topic can be found on the Foundation s website Enfield, Nicholas J., and Stephen C. Levinson, eds. 2006. Roots of human soci- (http://www.wennergren.org/programs/international-symposia). ality: culture, cognition, and interaction. Oxford: Berg. Gibson, Kathleen, and Tim Ingold, eds. 1993. Tools, language and cognition in human evolution. Cambridge: Cambridge University Press. Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current References Cited Anthropology 58(suppl. 16):S329–S336. Kuhn, Steven L., and Erella Hovers, eds. 2013. Alternative pathways to com- Antón, Susan C., and Leslie C. Aiello, eds. 2012. Human biology and the origins plexity: evolutionary trajectories in the Middle Paleolithic and Middle Stone of Homo. Current Anthropology 53(suppl. 6):S267–S496. Age. Current Anthropology 54(suppl. 8):S176–S182. Bae, Christopher J., Katerina Douka, and Michael D. Petraglia, eds. Forthcom- Sandgathe, Dennis M., and Francesco Berna. 2017. Fire and the genus Homo: ing. Human colonization of Asia in the Late Pleistocene. Current Anthro- an introduction to supplement 16. Current Anthropology 58(suppl. 16):S165– pology 58(suppl. 17). S174. Bishop, Walter William, and J. Desmond Clark, eds. 1967. Background to evo- Washburn, Sherwood L., ed. 1964. Classification and human evolution. Viking lution in Africa. Chicago: University of Chicago Press. Fund Publications in Anthropology, no. 37 (Wenner-Gren Foundation for Butzer, Karl W., and Glynn Isaac, eds. 1975. After the Australopithecines: Anthropological Research). Chicago: Aldine. stratigraphy, ecology and culture change in the Middle Pleistocene. World Wrangham, Richard. 2017. Control of fire in the Paleolithic: evaluating the Anthropology Series. The Hague: Mouton. cooking hypothesis. Current Anthropology 58(suppl. 16):S303–S313. Coppen, Yves, Francis Clark Howell, Glynn L. Isaac, and Richard E. F. Leakey, Wynn, Thomas, and Frederick L. Coolidge, eds. 2010. Working memory: be- eds. 1976. Earliest man and environments in the Lake Rudolf Basin: stratig- yond language and symbolism. Current Anthropology 51(suppl. 1):S1–S200. Current Anthropology Volume 58, Supplement 16, August 2017 S165

Fire and the Genus Homo An Introduction to Supplement 16

by Dennis M. Sandgathe and Francesco Berna

Employing fire as an adaptive aid represents one of the most important technological developments in the course of hominin evolution, and, not surprisingly, research into the prehistoric use of fire has a long history. Over the last decade or so there has been a notable increase in research. Some people have continued to focus on better under- standing of the timing of the beginning of fire use, but some have also been trying to understand something of its role in the evolution of the genus Homo. In the fall of 2015 a symposium was held in Portugal that brought together 17 re- searchers who have contributed significantly in recent years to this subject. These contributions include improving the type and quality of archaeological and ethnoarchaeological data, collecting and interpreting fire residue data from archaeological sites from various time periods and regions, and developing models of fire as an ecological resource and the role of cooking in hominin evolution. A result of the symposium was the recognition of the need to focus less on data from individual sites and more on the broader role of fire in hominin adaptations and to concentrate more on developing the analytical methods and skills to confidently interpret what we see in the archaeological record.

While it may be difficult to identify major events in the evo- and safer to consume and at the same time significantly in- lution of human physiology, we can certainly identify im- crease its caloric returns. Beyond significantly increasing our portant cultural ones, such as the initial use of stone tools, the adaptive range and potential, the use of fire very likely played development of artificial shelter, or the beginning of food a major role in subsequent physiological changes such as our production. Of course, researchers may debate which devel- decrease in gut size and increase in encephalization. The pa- opments are the most significant, but when it comes to major pers in this volume, originally presented at a workshop titled technological developments over the course of the evolution “Fire and the Genus Homo,” bring together a wide range of of the genus Homo, almost all would certainly include the use current perspectives and data on this important topic. of fire. It is difficult to overemphasize the significance of this development. In whatever manner fire was initially acquired, Early Research whatever it was initially used for, whatever its function at various times and places over the course of subsequent pre- The recognition of the importance of fire in hominin adapta- history, and however the hominin-fire relationship developed tions can be traced back to very early in the history of prehis- over time, its far-reaching potential is clear. It allowed us to toric archaeology and human evolution research (e.g., Darwin significantly modify our environment and its resources, to sur- 1871). Since then questions about early fire use have been evolv- vive in extremely cold environments, and to keep dangerous ing, but rather slowly. For much of the earlier history of re- animals at bay. With fire we can see better in the dark, thus search, interest was focused heavily on simply determining when effectively extending the period in each day for activities; we and where fire was first used (e.g., Black 1932; Hough 1926; can modify raw materials to expand the range of our material Oakley 1955, 1956). Moreover, much of this research was not culture; and we can cook our food in order to make it easier necessarily hampered by any concern about the authenticity of potential evidence for fire use at early sites (e.g., Black 1932; Stewart 1956), and there was also a strong presumption that Dennis M. Sandgathe is Lecturer in the Department of Archaeology once fire use was “discovered,” it subsequently became a uni- at Simon Fraser University (8888 University Drive, Burnaby, British versal and ubiquitous component of all hominin populations Columbia V54 1S6, Canada) and at the Museum of Archaeology and (e.g., Oakley 1955, 1956; Stewart 1956). Likewise in the early Anthropology of the University of Pennsylvania (3260 South Street, years there was little specific interest in how fire use came to be Philadelphia, Pennsylvania 19104, USA [[email protected]]). Francesco Berna is Assistant Professor in the Department of Archaeology of incorporated into the hominin technological repertoire: it was Simon Fraser University (8888 University Drive, Burnaby, British simply the result of a discovery. And there was also limited in- fi Columbia V54 1S6, Canada [[email protected]]). This paper was sub- terest in exploring what early re might have been used for, mitted 25 VII 16, accepted 3 II 17, and electronically published 27 other than general suggestions related to cooking food, provid- VII 17. ing heat, or scaring off dangerous animals (e.g., Oakley 1955). A q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0002$10.00. DOI: 10.1086/691424 S166 Current Anthropology Volume 58, Supplement 16, August 2017 significant exception to much of this early history is the work prehistory fire use had become common. In fact, the litera- of Kenneth Oakley. In the 1950s he published important ture would suggest that currently the majority of researchers articles on the (at that point very limited) evidence for Paleo- believe that by the Middle Paleolithic, if not earlier, hominins lithic use of fire. This included discussions of the nature of this were regularly using fire and had developed the technology to evidence and potential problems with some of it (including the create it at will; conversely, in contexts where evidence for fire potential for natural fire residues to be mistaken for anthro- was lacking, the absence was interpreted simply as being a pogenic ones; Oakley 1954, 1955, 1956, 1961). However, there product of taphonomic processes. However, there was no were no real discussions about the “development” of fire use— evidence of any hominin-fire interaction predating the emer- that is, as a process rather than an event—and there was no gence of the genus Homo. This continues to be the case today: discussion about the interaction between fire use and hominin regardless of the specific timing for when fire came to be physiological evolution. used, it is clearly a behavior associated exclusively with spe- In the early decades of Paleolithic research there were lim- cies of Homo and does not appear to predate the emergence ited field methods available for the accurate identification and of Homo ergaster. We should note, however, that the recog- interpretation of sediments that could represent potential ev- nition of a certain degree of passive interaction between chim- idence of hominin fire use. At a number of sites this led to panzees and fire (Pruetz and Herzog 2017; Pruetz and LaDuke the misidentification of the presence of fire based on sediments 2010) as well as some species of monkeys (Herzog et al. 2014) that were in fact unrelated to burning (Goldberg, Miller, and suggests the possibility of similar behaviors extending back Mentzer 2017). For example, at the site of Fontechevade in into the Pliocene. southwestern France patches of manganese dioxide were mis- taken for charcoal (Chase et al. 2009), and at Zhoukoudian, Research in Recent Years China (Goldberg et al. 2001; Weiner et al. 1998), lenses of or- ganic matter and layers of very fine silts from Middle Pleis- While there has long been a general interest in the question of tocene contexts were mistaken for charcoal and ash. A simi- Paleolithic fire use, the last couple of decades—and especially lar mistaken interpretation occurred at South Africa’sCaveof the last several years—have seen a particularly strong uptick Hearths (Oakley 1957), and more recently there is the exam- in research on this topic. Since 2000 there has been a three- to pleofSchöningeninnorthernGermany(Stahlschmidtetal. fourfold increase in the number of journal publications spe- 2015). There was also a serious lack of detail in the descrip- cifically on Paleolithic fire use. This upsurge is the product of tions of real or potential fire residues in early sites. For ex- a number of factors that include important new discoveries of ample, the term “hearth” was often used rather indiscrimi- fire residues associated with various Homo species in Africa, nately to refer to what were simply dark or reddish patches in southwest Asia, and Europe, the influence of the cooking hy- the sediments or a scattering of a few burned bones, charcoal, pothesis (Alperson-Afil and Goren-Inbar 2006; Alperson-Afil, or burned lithics, and such evidence was rarely quantified (e.g., Richter, and Goren-Inbar 2007; Berna et al. 2012; Goren- Bader and Klein 1965; Breuil 1922; de Lumley and Bard 1972; Inbar et al. 2004; Gowlett and Wrangham 2013; Karkanas Garrod 1951; Henri-Martin 1965; Klein 1965; Oakley 1954, et al. 2007; Preece et al. 2006; Schiegl et al. 1996; Wrangham 1956, 1961; Stewart 1956; Waechter 1951). This meant that 2009), and the development of new methods associated with other researchers could not properly assess such findings in the recovery and interpretation of archaeological deposits (e.g., terms of how accurately they reflected evidence for fire use in Goldberg, Miller, and Mentzer 2017; Mentzer 2014). The latter the archaeological record. So the recovery of minute traces of development has put us in a much better position to distinguish what was assumed to be ash or charcoal resulted, in many actual fire residues and features from deposits that can poten- cases, in those sites being put forward as examples of the fre- tially mimic them (Bellomo 1994; Berna and Goldberg 2008). quent use of fire. Over time, such lack of rigor resulted in an A small number of researchers have also begun to broaden overestimation of the frequency of fire residues in Paleolithic our understanding of the nature of natural fires, their resulting sites and, in turn, how common fire use likely had been through- residues, and their potential effect on archaeological remains out the Paleolithic (e.g., Roebroeks and Villa 2011; Sandgathe (Barbetti et al. 1980; Gowlett et al. 1981, 2017). These two de- et al. 2011b; Stahlschmidt et al. 2015). velopments have resulted in a more generally critical approach However, while the evidence for fire use from Lower and to claims for fire, especially in particularly old contexts (James Middle Pleistocene contexts was limited and subject to debate 1989; Roebroeks and Villa 2011; Sandgathe et al. 2011b). (Barbetti 1986; Bellomo 1993, 1994; Gowlett 2006; James New approaches are now being used that allow us to test 1989), unmistakable examples of hominin use of fire, includ- the effects of taphonomy on the presence or absence of ephem- ing well-preserved combustion features, had been recognized eral fire residues in site components. Direct fire residues like in Late Pleistocene contexts. This was especially the case with charcoal, ash, and heat-altered sediments are easily eroded or sites associated with Neanderthals in Western Eurasia (e.g., dissolved by a range of taphonomic processes, and while we Oakley 1956). Fire residues, however variously described, oc- recognize that the absence of these residues cannot be taken as curred frequently enough in these contexts that, taken at face evidence for a lack of fire use at a site, neither can we build value, they seemed to indicate that at least by this point in models or construct hypotheses about prehistoric fire use on Sandgathe and Berna Fire and the Genus Homo S167 contentions that fire could have been used in contexts where hominin interaction with fire. These are still our primary sources there is no positive evidence for it. However, some fire resi- of data. However, we are coming to recognize that the relation- dues such as fat-derived char (e.g., Goldberg et al. 2009; Mallol ship between hominins and fire was likely a long and complex et al. 2013) and burned lithics (and to a lesser extent burned one, possibly following many varied stages of development. bones) are far less affected by common taphonomic processes. Perhaps not at first, but as this relationship became more en- Several research projects have been investigating how infor- tangled, fire came to play ever more important roles in hom- mative these residues are about the apparent frequency of fire inin adaptations, and eventually the multifaceted use of fire use at Paleolithic sites (Aldeias et al. 2012; Dibble et al. 2009, became a necessity for the survival of modern humans: we be- 2017; Goldberg et al. 2012; Sandgathe et al. 2011a; Shimelmitz came obligate fire users. et al. 2014). Others have been looking at spatial patterning in burned lithics (e.g., Alperson-Afil, Richter, and Goren-Inbar 2007; Gowlett 2006), often with the implicit goal to identify The Organization of This Volume criteria by which the results of natural fire can be distin- This volume follows the general organization of the sympo- guished from results of anthropogenic fire (e.g., Gowlett et al. sium on which it is based. The topic “Fire and the Genus 2017; Hlubik et al. 2017). As a result of improvements in the Homo” includes a wide range of distinct but overlapping areas types and quality of data and an increased appreciation for the of research, and we have tried to organize the participants’ importance of biocultural evolution, there are some clear and contributions in this manner. significant shifts from previous decades in how most research- ers view the probable complexity of hominin interaction with and use of fire. This change has resulted in changes in the way Prehistoric Fire Use: The Data Paleolithic fireuseshouldbestudiedandinwhatconstitutes evidence for various fire-related behaviors. Regardless of the varied approaches that researchers take to understand the nature of Paleolithic interaction with fire, all The Symposium “Fire and the Genus Homo” interpretation ultimately rests on the nature and quality of our most basic types of data—residues and alterations di- When we first proposed to Leslie Aiello and Laurie Obbink at rectly resulting from fire—and researchers have been expand- the Wenner-Gren Foundation that they devote one of their ing the types of data that can be brought to bear on our ques- two annual symposia to early fire use, we were quite surprised tions. We rely on these data to provide the most basic contexts to find that over the Wenner-Gren symposium history that for our subsequent interpretations: Are sediments in a site ac- spans almost 60 years (and over 150 conferences!), this topic tually fire residues? Are these fire residues the result of nat- had never been covered before. This is unexpected consid- ural or anthropogenic fire? What type of fuel was used? What ering how much interest archaeologists and anthropologists temperature did the fire reach? But we are now trying to have always shown in the subject of early hominin fire use. expand our questions to include investigations into the struc- However, considering the increased emphasis on research on ture and function of fires, the latter being the most difficult this topic over the last several years, this did seem like a par- aspect to demonstrate. These questions lead us to consider even ticularly opportune time to finally bring together a significant more carefully what types of data should be recorded, what number of the scholars currently leading research into this types of samples should be taken, and what analyses should area of hominin behavior and adaptation. We only regret not be carried out as a normal course of excavation in sites with being able to include all those researchers who are currently potential evidence of early hominin interaction with fire. doing research on Paleolithic fire or who have made major Building on the work of Barbetti (1986), Goldberg, Miller, and contributions to this topic in the past. Mentzer (2017) formalize three basic questions that should It is becoming increasingly clear that we are now at the comprise the identification and interpretation of early Homo point where we can begin developing a more evolutionary ap- fire use: (1) Are the sediments or objects in question actually proach to the role of fire in hominin (or homininae?!) adap- burned? (2) If they are burned, what was the nature and con- tations, and this symposium was intended to instigate and fa- text of their deposition? (3) Were they burned by hominins? cilitate discussion and collaboration along those lines. During Currently there is a range of data, analytical techniques, and the discussions in this symposium it became clear that the methodologies that are necessary in trying to answer these field needs to move on from treating hominins establishing questions, and many of these should be considered as minimal “control” of fire as a onetime event at some distant time in the standards in the recovery and interpretation of potential evi- ancient past, from debating the timing of this, and from as- dence of early fire use. Goldberg, Miller, and Mentzer (2017) suming that from that point on, fire was a regular component and Aldeias (2017) summarize the types of data and analyses of all hominins’ adaptations. Obviously, we still need to es- as well as quantification that can differentiate actual burned tablish an understanding of the timing of the process of hom- residues from sediments that can mimic these. Ultimately, the inin development of fire use, and so it is still very important most effective approach is going to be the integration of high- that we identify examples in the archaeological record of early resolution spatial analysis of the archaeological record and S168 Current Anthropology Volume 58, Supplement 16, August 2017 microscopic and chemical-physical characterization of the ar- different types of combustion features and different fuel types; chaeological deposits (microcontextual approach). and under what conditions will combustion residues be pre- However, our understanding of the relationship between served? prehistoric fire and the resulting archaeological residues de- Goldberg, Miller, and Mentzer’s (2017) third question pends heavily on an understanding of the interaction between (whether or not the evidence of burning is a product of hom- fire and its immediate environment and the residues and al- inins) is especially important in many Paleolithic contexts, es- terations that result from this interaction. This understanding pecially the open-air sites, because fire residues may be a prod- relies very heavily on experimental work (e.g., Aldeias 2017; uct of natural fires with no relationship to hominin behavior. Aldeias et al. 2016; Bellomo 1993; Berna et al. 2007; Gowlett In many ecosystems, especially in warmer midlatitudes and for- et al. 2017; Mallol and Henry 2017; Mallol et al. 2007, 2013; ested regions, natural fires are very common, which means that March 1992; March et al. 2014; Théry-Parisot 2001; Théry- for many archaeological sites natural fi res are an equally plau- Parisot and Costamagno 2005). Having an understanding of sible source for fire residues and alterations of sediments, what residues and alterations of the immediate environment bones, and lithics (for a particular example, see the prelude in occur under different heat conditions (e.g., the size of the fire, Goldberg, Miller, and Mentzer 2017). It has become apparent the duration of the fire, what fuel was used, the nature of the that we need a much better understanding of the frequency, substrates upon which the fire was constructed) and under variability, effects on substrates, and resulting products of nat- what conditions these residues and alterations can survive in ural fire across a range of different environments so that we the archaeological record is essential for the further develop- can begin to distinguish these from the products of anthro- ment of methods that can address the three questions laid out pogenic fires—in essence, control for the natural fire “back- by Goldberg, Miller, and Mentzer (2017). ground noise” across the landscape. Based on observations and A second source of information about fire in the archae- experiments with natural and controlled-burn fires in eastern ological context comes from the ethnographic record. We have and southern Africa, Gowlett et al. (2017) call into question come to recognize the problems inherent with trying to apply some previous assumptions held about the differences between ethnographic data or models wholesale to prehistoric contexts, natural fires and anthropogenic ones. For example, they ob- especially when we are dealing with premodern hominin spe- served that in natural grass fires temperatures can reach up to cies. There is no a priori reason to expect exactly the same and beyond the limit typically ascribed to campfires. However, interactions with the environment and the exact same limits the nature of residues and alterations produced by a fire de- on the range of behaviors between modern humans and ex- pend heavily on the duration of the fire and the temperatures tinct species of Homo. Indeed ethnographic analogy should reached. There is significant variability in these variables across not be completely trusted as compelling evidence in prehistory different types of wildfires and even within individual wildfires, (Wylie 1985). However, we also recognize that ethnographic depending on the nature, concentration, and density of the data can be a powerful source of insight into the potential types vegetation. andrangesofevidenceleftbehindbybehaviorsandtechnol- ogies that hominins may have adopted under certain environ- Lower and Middle Pleistocene Use of Fire mental circumstances (David and Kramer 2001). The ethnoar- chaeology of fire use among living small-scale societies provides It is in African Lower Pleistocene contexts where we get the a wealth of comparative materials for the interpretation of ar- earliest potential evidence for hominin-fire interaction, and chaeological fire remains. Such studies can provide important for many researchers this indicates that Homo ergaster/Homo insights into how fire creation technologies, fire function, fire erectus could have been regularly using fire well before a mil- structures, the repeated use and duration of use of a single fire, lion years ago (e.g., Alperson-Afil 2017; Bellomo 1994; Brain and fires using different fuels are transformed when they en- and Sillen 1988; Clark and Harris 1985; Gowlett and Wrang- ter the archaeological record (Mallol and Henry 2017; Mallol ham 2013; Shimelmitz et al. 2014). However, the number of et al. 2007). In this volume, ethnoarchaeological investigations examples of Lower Pleistocene fire residues associated with of fire use among two different small-scale society groups— hominin sites is very small when compared with the totality a group of Hadza foragers in Tanzania and Evenk reindeer of coeval sites, and the evidence (patches of rubefied sediments herders and hunters of southeastern Siberia—included residue that may indicate hearth locations, isolated burned bones, and analysis of combustion remains (Mallol and Henry 2017). As isolated or small concentrations of burned lithics) is uncon- with other experimental approaches, these types of studies vincing to many (Chazan 2017; Roebroeks and Villa 2011). As provide data against which archaeological observations can be discussed above, many researchers recognize that given the fre- compared. For example, how do different types of fuels behave quencies of natural fires in Africa today, especially in Savannah in different combustion settings, and what sort of fire functions and Sahel ecosystems, there is a reasonable possibility that can they be linked to? Geoarchaeological analysis of ethno- such fire residues in archaeological sites are not the products graphic fire use provides major insights into taphonomy and of hominin behavior (e.g., Gowlett et al. 2017; Hlubik et al. formation processes: what is the nature of the residues left 2017; Roebroeks and Villa 2011; Sandgathe et al. 2011b). Over behind, if any; what sort of alterations occur to substrates with the hundreds of thousands of years that these sites lay on the Sandgathe and Berna Fire and the Genus Homo S169 landscape, natural fires must have passed over them many interaction or the role that fire was playing. To get at this we times. For archaeological sites that are not deeply buried, the need more information about, among other things, the ac- chances are not insignificant that the later fires could affect tual frequency of use of fire at these sites. Was fire being used artifacts, bones, and sediments (Buenger 2003). In the case of every time a group occupied a site? Or was its use intermittent, a burning tree or shrub, the result can be a concentration of and if so, how much time passed between uses, and why was it burned sediments, bones, or lithics that could mimic the re- just used at some sites and not others? It would be important mains of a hearth (Bellomo 1994; Gowlett et al. 2017). This to have some understanding of how regular fire use was, how means that simply measuring (through methods like magnetic integrated it had become in hominin adaptations, and what susceptibility and thermoluminescence) the temperatures that function or functions it served before we are able to construct altered sediments or heated flints or performing simple spatial models about the role fire played and its potential impact on analysis of these residues are not enough to distinguish nat- hominin biocultural adaptations at this point in prehistory. ural from anthropogenic fires. However, we might be able to get at this problem through the careful analysis of patterning Fire Use in Late Pleistocene Europe and Holocene Australia in the burning in these early archaeological sites. This is the approach that Hlubik et al. have been taking in their study of The current evidence from several Spanish sites in the Ata- burned basalt, bone, and soil aggregates at the site of FxJj20AB puerca Hills and Orce Basin indicates that some species of at Koobi Fora in Kenya. They attempt to identify patterning in Homo (some version of H. erectus?) moved into Europe some- the distribution of burned artifacts that cannot be explained as time prior to 1.25 Mya (Carbonell et al. 2008; Moyano et al. a result of the passage of natural fires through or over the site 2011). However, evidence for any interaction with fire in any and that indicate that the fire and the occupation of the site European sites older than around 400 kya is almost nonexistent co-occurred. This would certainly make it much more likely (the one exception, so far, being Cueva Negra in Spain, which that the hominins were interacting with and perhaps respon- has fire residues that date to around 800 kya; Walker et al. sible for the fire. 2016). While Lower and Middle Pleistocene occupations of Researchers have identified sites dated to the end of the Europe may have correlated strongly with warm climatic pe- Lower Pleistocene with potential evidence for hominin in- riods and warmer Mediterranean latitudes, it is still surprising teraction with fire. One of these is Wonderwerk Cave in South that these African-adapted hominins were able to colonize Africa, which has yielded ash and burned bone fragments deep European latitudes without the regular use of fire. However, inside the cave in deposits dated to a million years ago (Berna that fire was not a requisite technology for hominins to do this et al. 2012; Chazan 2017). It is difficult to propose natural fires was recognized many years ago (e.g., Perlès 1981). By the end as a source for these residues as there is no evidence that any of the Middle Pleistocene, the majority of Eurasian site com- of the sediments washed into the cave and it seems very un- ponents still have little or no evidence for fire, but some do, likely that natural fire could occur so deeply inside the cave. and by the start of the Late Pleistocene a significant number There are also sites of similar age outside of Africa. Gesher do have very good evidence that Neanderthals regularly used Benot Ya’akov in Israel, where a number of concentrations of fire (Roebroeks and Villa 2011). Because of this, for several de- burned lithics across the site strongly suggest the locations of cades now it appears that the general consensus is that re- hearths dated to approximately 800 kya (Alperson-Afil 2017; gardless of what was going on in previous time periods and Alperson-Afil, Richter, and Goren-Inbar 2007) and Cueva Ne- with other hominins, Neanderthals were creating and using gra in Spain where, like Wonderwerk Cave, fire residues occur fire at will and that it was an integral component of their adap- inside a cave (Walker at al. 2016). tation (e.g., Barkai et al. 2017; Daniau, d’Errico, and Goñi The record of fire in archaeological sites changes dramat- 2010; Gowlett 2016; Pettitt 1997; Rolland 2004). In fact, it is ically in the Middle Pleistocene, with a large increase in the difficult to imagine that even an apparently physiologically number of sites and site components with fire residues and cold-adapted species like Neanderthals could survive the ex- undeniable examples of hominin use of fire. This includes in- treme colds of Ice Age Eurasia without fire. However, Dibble tact hearths in several sites in Israel—Qesem Cave (Barkai et al. et al. (2017) present evidence from several French Middle 2017; Karkanas et al. 2007), Hayonim Cave (Goldberg and Paleolithic sites that appears to support the idea that while Bar-Yosef 2002), and Tabun Cave (Schimelmitz et al. 2014)— Neanderthals were certainly using fire regularly during some and hearth structures identified during recent excavations in periods, there are sites with evidence for major successive oc- layer 4 of Zoukoudian in China (Gao et al. 2017). Whereas we cupations spanning many thousands of years with no evidence have no idea how fire was being obtained, it is clear that the that they were using fire over that period. The data from these hominins occupying these sites were using fire and often used sites seem to show a pattern of fire use that correlates coun- it repeatedly, in the same locations within the sites (e.g., at terintuitively with climate: Neanderthals were using fire dur- Qesem Cave; Shahack-Gross et al. 2014). These sites dating to ing warm periods and apparently not during cold periods. 300–400 kya are the earliest evidence so far for fire beginning Dibble et al. suggest a number of potential explanations for to take on a more important role in Homo adaptations. It is this that bring into question whether by the Late Pleistocene all very dif ficult to get a realistic idea of the actual nature of this hominins in all regions had the technology to make fire at will. S170 Current Anthropology Volume 58, Supplement 16, August 2017

General perceptions among Paleolithic researchers appear man 2016). For over a decade Wrangham (e.g., Wrangham to be that with the arrival of modern humans in Europe and 2007, 2009; Wrangham et al. 1999; Wrangham and Conklin- throughout the subsequent Upper Paleolithic period, fire use Brittain 2003) has made a compelling argument that modern was ubiquitous and had taken on an even more complex role humans need to cook our food in order to obtain the net en- in human adaptations (e.g., Chazan 2017; White et al. 2017). ergy returns required to support such a large brain while also As presented by White et al., Upper Paleolithic sites in Eu- having such a small gut and small teeth. Furthermore, he ar- rope contain a range of different fire structures, some with rel- gues that this obligatory relationship between brain, gut, teeth, atively complex stone arrangements (see also Movius 1966) that and cooking extends back to H. ergaster/H. erectus around or make it clear that many of these are not just general-use shortly after 2 Mya. Wrangham’s cooking hypothesis has come “hearths” providing heat for warmth and cooking. While the to be one of the more influential ideas in paleoanthropology actual functions of these different structures has yet to be de- today. However, the cooking hypothesis is arguing that cook- termined in most cases, their evidence strongly suggests a much ing and fire use would have essentially been a daily activity wider range of applications of fire than anything observed in over the last 2 Mya, while the current archaeological evidence earlier periods. There is also a clear increase in the complexity for fire use over much of this time is sparse at best. Wrang- of the spatial arrangement of occupation sites, with fire struc- ham (2017) presents further arguments in support of the very tures being major components of this organization. early adoption of a cooked diet. He also addresses many of the However, the archaeological record is a product of taph- potential problems the hypothesis faces and goes a long way onomic processes as much as—and often more than—the di- toward reconciling some of the apparent inconsistencies be- rect result of patterns of behavior (Holdaway, Fanning, and tween the hypothesis and the archaeological data. Shiner 2005). This is an important point when it comes to Two other symposium contributions represent major data interpreting any aspects of the archaeological record but is points along the prehistoric development of cooking. Barkai especially pertinent when we are dealing with remains that are et al. (2017) discuss the significant fire residues at Qesem particularly susceptible to erosion and loss, as is the case with Cave, Israel, dated between 420 and 200 kya. They interpret direct fire residues like charcoal and ash. Holdaway, Davies, these residues as the remains of meat-roasting fires that were and Fanning (2017) present an important example of this with used regularly and repeatedly, and they argue that this be- Holocene age fire features in Australia. In the presented case, havior was an essential component of hominin adaptations in the preservation and visibility of any evidence of fire-related the Levant in this period. What species of archaic Homo is behavior is facilitated by the prehistoric use of stone structures represented at the sites is unclear, but presumably it is either as heat retainers. While the ephemeral residues like charcoal Homo heidelbergensis or early Neanderthals. and ash generally disappeared very quickly, these structures sur- Henry (2010) and Henry, Brooks, and Piperno (2011) have vived more or less intact. One might then examine the tem- made significant contributions to our understanding of cook- poral and spatial distribution of these features as direct sources ing and the role of plant foods in Neanderthal diets. Their of information about patterns of prehistoric use of fire. How- analysis of phytoliths and starch grains incorporated into Nean- ever, the analysis of Holdaway, Davies, and Fanning indicates derthal dental calculi strongly suggests that some plants were that the frequency of these hearth structures in the archaeo- being integrated into their diet. The morphology of some starch logical record is also heavily influenced by taphonomic pro- grains in the calculi also suggests that some plants were being cesses, particularly geomorphic development of landscapes. cooked before they were consumed (Henry 2015). However, This serves as one cautionary tale (and a particularly powerful much of the other evidence (stable isotope data and the faunal one considering the young age of these hearth structures) for record) suggests that the Neanderthal diet was heavily meat any attempts to infer behavioral patterns directly from the focused (e.g., Richards and Trinkaus 2009) and that Neander- presence or absence and frequency of any archaeological phe- thals were not cooking their food at all times and in all places nomena, not just fires. (e.g., Dibble et al. 2017; Sandgathe et al. 2011a, 2011b). Henry (2017) has taken a novel approach to understanding Nean- derthal fire use patterns (although the implications are not Cooking restricted to any one species of Homo). She suggests that un- While fire has undoubtedly served a wide range of functions like recent modern human foraging societies, Neanderthals over the course of prehistory, a century of ethnographic ob- may have viewed fire as just another potentially exploitable servations from around the world has shown that cooking is resource among many available in the landscape. Whether or universal among human groups. Thus it made sense to have a not they chose to exploit each resource depended on a cost- special section devoted to this. benefit approach, and this, of course, changes depending on People cook food for a variety of reasons, for example, the changing circumstances. At times, the cost of collecting fuel making rotting meat safe to eat (e.g., Dibble et al. 2017) or and starting and maintaining a fire may outweigh the potential denaturizing meat or plant food to make it easier to chew benefits. Under such circumstances they may have chosen not (e.g., Wrangham and Conklin-Brittain 2003; Zink and Lieber- to use it. This idea is complementary to, and may present an Sandgathe and Berna Fire and the Genus Homo S171 alternative explanation for, the observation in Dibble et al. of Homo speciation. An attractive component of this model is (2017) that at least in some regions Neanderthals were not how readily it can be operationalized—how directly research- using fire for very long periods of time. ers can develop hypotheses that are testable with archaeolog- ical data. Finally, Sandgathe (2017) discusses two major current is- The Long View of Hominin Fire Use sues in the development of broader, more process-oriented Ultimately, one of the major goals of the symposium was to views of the hominin fire use. The first issue is a practical one instigate discussions about and perhaps reevaluate the role of terminology. A number of terms have come into common fire played in the evolution of the genus Homo. Starting at the usage among early fire researchers (e.g., “control” and “ha- level of the analysis of data from individual archaeological sites, bitual”), but there is little common understanding of what we want to be able to start to construct models and hypoth- individual researchers mean by these terms, with the result eses about how this process might have gone and what stages that there are undoubtedly some mutual misunderstandings of increasing complexity might characterize it (e.g., Parker et al. and a certain degree of talking past each other. The second 2016). An important source of data for constructing such mod- and more important issue is the overarching tendency until els will come from observations of how nonhuman primates recently to see the development of fire use as an event (see interact with natural fires (e.g., Atwell, Kovarovic, and Kendall also Chazan 2017). In recent years another “event” has been 2015; Parker et al. 2016). Some work has already been done added to this discussion: when fire use becomes “habitual.” that indicates that some primates take advantage of natural The addition of a second, later event is a small step forward fires and can organize their behaviors around exploiting food among researchers in that it starts to hint at the existence of a by-products derived by the passage of a fire (e.g., vervet mon- longer process that likely includes several major stages. How- keys; Herzog et al. 2014). Pruetz and Herzog (2017) present ever, it still fails to recognize the greater potential complexity the results of their observations of chimpanzees in Fongoli, that could have, and likely did, underlie the development of Senegal, interacting with wildfires. This interaction is limited, fire use. This process was undoubtedly a very long, slow one, and the observed chimpanzees did not actively manipulate and there is no a priori reason to assume that this develop- natural fire, but they developed an understanding of its behavior ment was linear, or at least was always so. The prehistory of that allows them to predict its movement. Such an under- fire use may have included multiple examples of populations standing in turn lessens their fear of it and allows them to exploit not using fire after long periods of time when they did. And changes to the fire-altered landscape, such as accessing food there is no reason to think that the process of development of resources that were altered or exposed by the fire. With the fire use was the same in all regions. The big picture of the logical caveat that applying modern examples as direct analogs prehistory of hominin fire use was likely very messy. to the past is problematic, such research is obviously necessary if we want to develop models of early human interaction with Results of the Symposium fire that, presumably, is a required first step in the development of a more complex relationship with fire that would include Over the course of the discussions that characterized the active manipulation and maintenance. symposium a number of major issues with prehistoric fire use Chazan (2017) proposes a model of the prehistory of fire were identified by the discussants. These are issues that seemed use. He addresses one of the major problems hindering the to be commonly recognized by the majority if not all of the development of a broader evolutionary view of hominin fire symposium participants and, presumably, by other researchers use: our tendency to construct narratives to describe and ex- whom we were unable to invite. plain the past. The result of this tendency in the context of First, we as a community of researchers interested in the this topic is that we continually focus on trying to identify an same questions need to shift the emphasis away from dis- “origin of fire,” looking for an event that can be pinpointed in cussions and debates about individual sites and what the na- time, after which everything changes for all hominins. Cha- ture of their fire residues might mean. While most of the zan’s model proposes three broad stages that would charac- basic data are going to continue to come from work carried terize the process of increasing complexity in hominin-fire out at individual sites, we need to focus more on the broader interaction. The first stage involves opportunistic interaction research goal of understanding the role of fire in hominin with natural fire, and so its use was limited to where and when adaptations, how this evolved over time from an initial sim- it was available. This could be compared to or overlaps with ple interaction with fire to its becoming an integral compo- the chimpanzee behavior observed by Pruetz and Herzog (2017; nent of hominin adaptations, and the development of the see also Pruetz and LaDuke 2010). The second stage involves technology to make it at will. This can only come from ex- the development of fire maintenance behaviors. The third stage amining collectively the data from many sites. We also need sees the development of fire containment strategies and pre- to recognize the importance of integrating the rest of the ar- sumably includes the development of techniques for creating chaeological data: the fire data alone will not be enough for us fire. Chazan then positions these stages within the framework to understand its role in adaptations. S172 Current Anthropology Volume 58, Supplement 16, August 2017

Second, we need to concentrate more on developing the • A certain degree of standardization in the types of data analytical methods and skills to confidently interpret what we and methods of data collection is necessary for compara- see in the archaeological record. Major sticking points in the bility of different researchers’ data. How can we compare debate have been identifying in very early sites, with broadly what different researchers are finding in the archaeological held confidence, actual fire residues and fire residues that are record if they individually are collecting very different the product of hominin behavior. This is a basic requirement types of data or using very different units of measure? How if we are to be able to identify the nature of hominin use of can we determine the weight of interpretations if re- fire at different times and places. searchers are not collecting or addressing the lack of all These two broader goals require the acknowledgment of sev- the pertinent data? eral other important guiding principles and considerations: • We need more explicit definitions for the terminology we use so that we know what each researcher means • Context is of paramount importance in interpreting the when they discuss their data and how they have inter- archaeological record. This is particularly important for preted it, to eliminate unnecessary ambiguities. fire residues, and a number of the papers in this volume address this issue directly (Aldeias 2017; Goldberg, Miller, Through the discussions that took place at the symposium and Mentzer 2017; Holdaway, Davies, and Fanning 2017; and from reading the resulting papers, it is clear that, as with Mallol and Henry 2017). Context involves several levels all areas of research, there are some disagreements and po- of resolution as well. For example, what was the nature of tential confl icts between the approaches and data that dif- vegetation cover of the site at the time it was occupied? ferent researchers are employing in their research into early How many years are represented by the millimeter-scale fire use. However, there is also far more common ground, and gap between two hearths in the same location? What was what was most obvious coming out of the symposium was the the substrate beneath a hearth composed of? Are the shared goal by all researchers involved to better understand burned bones associated with a fire feature the result of the nature of the evolution of fire use in the genus Homo.We cooking meat, burning bone as fuel, or simply building a hope the symposium represents an important step in that fire on top of previously discarded bone? By what pro- process. cesses was the deposit with the fire evidence formed and altered? • Since it is logical to assume that no scholar can be de- Acknowledgments termined to have discovered the first use of fire in pre- history, this pursuit is a useless exercise in terms of what That this symposium happened at all and was such a big suc- it can tell us about fire and hominin adaptations. What cess is due to the hard work and support of a number of peo- is important is understanding the role fire played in hom- ple and institutions. We are very grateful to the Wenner-Gren inin adaptations through time and across space. In order Foundation for Anthropological Research for the opportunity to achieve this, the scientific community should focus on to cohost this symposium and to do so in such a beautiful frequency of use, intensity of use, and functions of fire and locale. We are also indebted to all the participants for both the the changes that took place in these aspects through time quality of their papers and the intellectual quality and collegial and by region. nature of their contributions to the discussions around the • On their own, individual site records are of limited value, symposium table; it was a truly enjoyable and enlightening especially single occupation sites. We need to study the experience. We thank Meg Thibodeau for serving as such an evidence for fire use at multiple sites and through mul- effective rapporteur. We owe special thanks to Leslie Aiello, tiple occupations to get meaningful temporal and spa- president of the Wenner-Gren Foundation, and Laurie Obbink, tial frequency data. Therefore, while some researchers conference program associate, for their amazing patience and do have access to multiple sites, realistically, we will only organizational magic. We thank all the other members of the make significant progress through broad collaboration. Wenner-Gren staff who worked behind the scenes to make • We need to tailor questions to investigate available and the symposium a success and to produce the resulting volume. fi future data. There are some important questions we And nally, while all errors in format and content in this in- would currently very much like to answer that we simply troduction are ours, we are indebted to a number of anony- lack the data to realistically address at this point in time. mous reviewers for improving it considerably. For example, how, when, and where was the technology for making fire developed? Did it have a single origin? Or did it occur independently with different modalities References Cited in different areas and at different times? Getting to the Aldeias, Vera. 2017. Experimental approaches to archaeological fire features – point where we can confidently address these questions and their behavioral relevance. Current Anthropology 58(suppl. 16):S191 S205. comes back to focusing on the development of better data- Aldeias, Vera, Harold L. Dibble, Dennis M. Sandgathe, Paul Goldberg, and recovery methodologies and incorporating new types of data. Shannon J. P. McPherron. 2016. How heat alters underlying deposits and Sandgathe and Berna Fire and the Genus Homo S173

implications for archaeological fire features: a controlled experiment. Jour- du Würmien II dans le Midi méditerranéen, vol. 1. Marseille: Université de nal of Archaeological Science 67:64–79. Provence, Laboratoire de paléontologie humaine et de préhistoire. Aldeias, Vera, Paul Goldberg, Dennis M. Sandgathe, Francesco Berna, Harold Dibble, Harold L., Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Shan- L. Dibble, Shannon P. McPherron, and R. Zeljko. 2012. Evidence for Ne- non P. McPherron, and Dennis M. Sandgathe. 2017. How did hominins andertal use of fire at Roc de Marsal (France). Journal of Archaeological adapt to Ice Age Europe without fire? Current Anthropology 58(suppl. 16): Science 39(7):2414–2423. doi.org/10.1016/j.jas.2012.01.039. S278–S287. Alperson-Afil, Nira. 2017. Spatial analysis of fire: archaeological approach to Dibble, Harold L., Francesco Berna, Paul Goldberg, Shannon P. McPherron, recognizing early fire. Current Anthropology 58(suppl. 16):S258–S266. Susan M. Mentzer, L. Niven, D. Richter, Dennis M. Sandgathe, I. Théry- Alperson-Afil, Nira, and Naama Goren-Inbar. 2006. Out of Africa and into Parisot, and Alain Turq. 2009. A preliminary report on Pech de l’Azé IV, Eurasia with controlled use of fire: evidence from Gesher Benot Ya’aqov, layer 8 (Middle Paleolithic, France). PaleoAnthropology 2009:182–219. Israel. Archaeology, Ethnology and Anthropology of Eurasia 28(1):63–78. Gao, Xing, Shuangquan Zhang, Yue Zhang, and Fuyou Chen. 2017. Evidence Alperson-Afil, Nira, D. Richter, and Naama Goren-Inbar. 2007. Phantom of hominin use and maintenance of fire at Zhoukoudian. Current Anthro- hearths and the use of fire at Gesher Benot Ya’akov, Israel. Palaeoan- pology 58(suppl. 16):S267–S277. thropology 2007:1215. Garrod, D. A. 1951. A transitional industry from the base of the Upper Atwell, L., K. Kovarovic, and J. R. Kendall. 2015. Fire in the Plio-Pleistocene: Palaeolithic in Palestine and Syria. Journal of the Anthropological Institute the functions of hominin fire use, and the mechanistic, developmental and of Great Britain and Ireland 81(1/2):121–130. evolutionary consequences. Journal of Anthropological Sciences 93:1–20. Goldberg, Paul, and Ofer Bar-Yosef. 2002. Site formation processes in Ke- Bader, O. N., and R. G. Klein. 1965. The Palaeolithic of the Urals and the bara and Hayonim caves and their significance in Levantine prehistoric peopling of the North. Arctic Anthropology 3(1):77–90. caves. In Neandertals and modern humans in Western Asia. Takeru Aka- Barbetti, M. 1986. Traces of fire in the archaeological record, before one zawa, Kenichi Aoki, and Ofer Bar-Yosef, eds. Pp. 107–125. New York: million years ago? Journal of Human Evolution 15(8):771–781. Springer US. Barbetti, M., J. D. Clark, F. M. Williams, and M. A. J. Williams. 1980. Palaeo- Goldberg, Paul, Harold L. Dibble, Francesco Berna, Dennis M. Sandgathe, magnetism and the search for very ancient fireplaces in Africa. Anthropol- Shannon J. P. McPherron, and Alain Turq. 2012. New evidence on Ne- ogie 18:299–304. andertal use of fire: examples from Roc de Marsal and Pech de l’Azé IV. Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a rea- Quaternary International 247(1):325–340. son: barbecue at Middle Pleistocene Qesem Cave, Israel. Current Anthro- Goldberg, Paul, Christopher E. Miller, and Susan M. Mentzer. 2017. Recog- pology 58(suppl. 16):S314–S328. nizing fire in the Paleolithic archaeological record. Current Anthropology 58 Bellomo, R. V. 1993. A methodological approach for identifying archaeo- (suppl. 16):S175–S190. logical evidence of fire resulting from human activities. Journal of Ar- Goldberg, Paul, Christopher E. Miller, S. Schiegl, B. Ligouis, Francesco Berna, chaeological Science 20(5):525–553. N. J. Conard, and L. Wadley. 2009. Bedding, hearths, and site maintenance ———. 1994. Methods of determining early hominid behavioral activities in the Middle Stone Age of Sibudu Cave, KwaZulu-Natal, South Africa. Ar- associated with the controlled use of fire at FxJj 20, Main, Koobi Fora, chaeological and Anthropological Sciences 1(2):95–122. Kenya. Journal of Human Evolution 27:173–195. Goldberg, Paul, S. Weiner, Ofer Bar-Yosef, Q. Xu, and J. Liu. 2001. Site for- Berna, Francesco, A. Behar, Ruth Shahack-Gross, J. Berg, J. Zorn, E. Boaretto, mation processes at Zhoukoudian, China. Journal of Human Evolution 41: A. Gilboa, et al. 2007. Sediments exposed to high temperatures: recon- 483–530. structing pyrotechnological practices in Late Bronze and Iron Age strata Goren-Inbar, Naama, Nira Alperson, M. E. Kislev, O. Simchoni, Y. Melamed, at Tel Dor (Israel). Journal of Archaeological Science 34:358–373. A. Ben-Nun, and E. Werker. 2004. Evidence of hominin control of fire at Berna, Francesco, and Paul Goldberg. 2008. Assessing Paleolithic pyrotech- Gesher Benot Yaaqov, Israel. Science 304(5671):725–727. nology and associated hominin behavior in Israel. Israel Journal of Earth Gowlett, J. A. J. 2006. The early settlement of northern Europe: fire history in Sciences 56:107–121. the context of climate change and the social brain. Comptes Rendus Palevol Berna, Francesco, Paul Goldberg, L. Kolska Horwitz, J. Brink, S. Holt, M. 5(1):299–310. Bamford, and Michael Chazan. 2012. Microstratigraphic evidence of in situ ———. 2016. The discovery of fire by humans: a long and convoluted pro- fire in the Acheulean strata of Wonderwerk Cave, Northern Cape province, cess. Philosophical Transactions of the Royal Society B 371(1696):20150164. South Africa. Proceedings of the National Academy of Sciences of the USA Gowlett, J. A. J., J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina. 2017. 109:7593–7594. Evidence of burning from bushfires in southern and east Africa and its rel- Black, Davidson. 1932. Evidences of the use of fire by Sinanthropus. Acta evance to hominin evolution. Current Anthropology 58(suppl. 16):S206– Geologica Sinica (English edition) 11(2):107–108. S216. Brain, C. K., and A. Sillen. 1988. Evidence from the Swartkrans Cave for the Gowlett, J. A. J., John W. K. Harris, D. Walton, and B. A. Wood. 1981. Early earliest use of fire. Nature 336:464–466. archaeological sites, hominid remains and traces of fire from Chesowanja, Breuil, L. A. H. 1922. Palaeolithic Man at Gibraltar: new and old facts. Journal Kenya. Nature 294:125–129. of the Royal Anthropological Institute of Great Britain and Ireland 52:46–54. Gowlett, J. A. J., and Richard W. Wrangham. 2013. Earliest fire in Africa: Buenger, B. A. 2003. The impact of wildland and prescribed fire on archaeo- towards the convergence of archaeological evidence and the cooking hy- logical resources. University of Kansas. Final Report. Hot Springs, SD: Wind pothesis. Azania: Archaeological Research in Africa 48(1):5–30. Cave National Park. Henri-Martin, G. 1965. La grotte de Fontéchevade. Bulletin de l’Association Carbonell, E., J. M. B. de Castro, J. M. Parés, A. Pérez-González, G. Cuenca- Française pour l’Etude du Quaternaire 2(3):211–216. Bescós, A. Ollé, M. Mosquera, et al. 2008. The first hominin of Europe. Henry, Amanda G. 2010. Plant foods and the dietary ecology of Neandertals Nature 452(7186):465–469. and modern humans. Washington, DC: George Washington University. Chase, P. G., A. Débenath, Harold L. Dibble, and Shannon P. McPherron. ———. 2015. Formation and taphonomic processes affecting starch granules. 2009. The cave of Fontéchevade: recent excavations and their paleoanthro- In Method and theory in paleoethnobotany.JohnM.Marston,JadeD’Alpoim pological implications. Cambridge: Cambridge University Press. Guedes, and Christina Warinner, eds. Pp. 35–50. Boulder: University of Col- Chazan, Michael. 2017. Toward a long prehistory of fire. Current Anthro- orado Press. pology 58(suppl. 16):S351–S359. ———. 2017. Neanderthal cooking and the costs of fire. Current Anthropology Clark, J. D., and John W. Harris. 1985. Fire and its roles in early hominid 58(suppl. 16):S329–S336. lifeways. African Archaeological Review 3(1):3–27. Henry, Amanda G., A. S. Brooks, and D. R. Piperno. 2011. Microfossils in Daniau, A. L., F. d'Errico, and M. F. S. Goñi. 2010. Testing the hypothesis of calculus demonstrate consumption of plants and cooked foods in Nean- fire use for ecosystem management by Neanderthal and Upper Palaeolithic derthal diets (Shanidar III, Iraq; Spy I and II, Belgium). Proceedings of the modern human populations. PLoS ONE 5(2):e9157. National Academy of Sciences of the USA 108(2):486–491. Darwin, Charles. 1871. The descent of man and selection in relation to sex. Herzog, Nicole M., C. H. Parker, E. R. Keefe, J. Coxworth, A. Barrett, and K. London: J. Murray. Hawkes. 2014. Fire and home range expansion: a behavioral response to David, N., and C. Kramer. 2001. Ethnoarchaeology in action. Cambridge: burning among savanna dwelling vervet monkeys (Chlorocebus aethiops). Cambridge University Press. American Journal of Physical Anthropology 154(4):554–560. de Lumley, H., and J. P. Bard. 1972. La grotte de l’Hortus (Valflaunès, Hérault): les Hlubik, Sarah, Francesco Berna, Craig Feibel, David Braun, and John W. K. chasseurs néandertaliens et leur milieu de vie. Élaboration d’une chronologie Harris. 2017. Researching the nature of fire at 1.5 Mya on the site of FxJj20 S174 Current Anthropology Volume 58, Supplement 16, August 2017

AB, Koobi Fora, Kenya, using high-resolution spatial analysis and FTIR ceptualization of “fire behavior” and the case for a chimpanzee model. spectrometry. Current Anthropology 58(suppl. 16):S243–S257. American Journal of Physical Anthropology 141(4):646–650. Holdaway, Simon J., Benjamin Davies, and Patricia C. Fanning. 2017. Aborig- Richards, M. P., and E. Trinkaus. 2009. Isotopic evidence for the diets of inal use of fire in a landscape context: investigating presence and absence of European Neanderthals and early modern humans. Proceedings of the Na- heat-retainer hearths in western New South Wales, Australia. Current An- tional Academy of Sciences of the USA 106(38):16034–16039. thropology 58(suppl. 16):S230–S242. Roebroeks, W., and P. Villa. 2011. On the earliest evidence for habitual use of Holdaway, Simon J., Patricia C. Fanning, and J. Shiner. 2005. Absence of ev- fire in Europe. Proceedings of the National Academy of Sciences of the USA idence or evidence of absence? understanding the chronology of indigenous 108(13):5209–5214. occupation of western New South Wales, Australia. Archaeology in Oceania Rolland, N. 2004. Was the emergence of home bases and domestic fire a 40(2):33–49. punctuated event? a review of the Middle Pleistocene record in Eurasia. Hough, W. 1926. Fire as an agent in human culture. Bulletin of the US Natural Asian Perspectives 43(2):248–280. Museum, No. 139. Washington, DC: Government Printing Office. Sandgathe, Dennis M. 2017. Identifying and describing pattern and process James, S. R. 1989. Hominid use of fire in the Lower and Middle Pleistocene. in the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): Current Anthropology 30(1):1–26. S360–S370. Karkanas, P., Ruth Shahack-Gross, Avner Ayalon, Mira Bar-Matthews, Ran Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. Barkai, Amos Frumkin, Avi Gopher, and Mary C. Stiner. 2007. Evidence for McPherron, Alain Turq, L. Niven, and J. Hodgkins. 2011a. On the role of fire habitual use of fire at the end of the Lower Paleolithic: site-formation pro- in Neandertal adaptations in Western Europe: evidence from Pech de l’Azé cesses at Qesem Cave, Israel. Journal of Human Evolution 53(2):197–212. IV and Roc de Marsal, France. PaleoAnthropology 2011:216–242. Klein, R. G. 1965. The Middle Palaeolithic of the Crimea. Arctic Anthropology ———. 2011b. Timing of the appearance of habitual fire use. Letter. Pro- 3(1):34–68. ceedings of the National Academy of Sciences of the USA 108(29):E298. Mallol, Carolina, and Auréade Henry. 2017. Ethnoarchaeology of Paleolithic Schiegl, S., Paul Goldberg, Ofer Bar-Yosef, and S. Weiner. 1996. Ash deposits fire: methodological considerations. Current Anthropology 58(suppl. 16): in Hayonim and Kebara caves, Israel: macroscopic, microscopic and min- S217–S229. eralogical observations, and their archaeological implications. Journal of Mallol, Carolina, C. M. Hernández, D. Cabanes, J. Machado, A. Sistiaga, L. Archaeological Science 23(5):763–781. Pérez, and B. Galván. 2013. Human actions performed on simple combus- Shahack-Gross, R., Francesco Berna, P. Karkanas, C. Lemorini, Avi Gopher, tion structures: an experimental approach to the study of Middle Palaeo- and Ran Barkai. 2014. Evidence for the repeated use of a central hearth at lithic fire. Quaternary International 315(0):3–15. Middle Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of Archaeo- Mallol, Carolina, F. W. Marlowe, B. M. Wood, and C. C. Porter. 2007. Earth, logical Science 44:12–21. wind, and fire: ethnoarchaeological signals of Hadza fires. Journal of Ar- Shimelmitz, R., S. L. Kuhn, A. J. Jelinek, A. Ronen, Amy E. Clark, and Mina chaeological Science 34(12):2035–2052. Weinstein-Evron. 2014. “Fire at will”: the emergence of habitual fire use March, R. J. 1992. L’utilisation du bois dans les foyers préhistoriques: une 350,000 years ago. Journal of Human Evolution 77:196–203. approche expérimentale. Bulletin de la Société Botanique de France. Actua- Stahlschmidt, M. C., C. E. Miller, B. Ligouis, U. Hambach, Paul Goldberg, lités Botaniques 139(2–4):245–253. Francesco Berna, D. Richter, B. Urban, J. Serangeli, and N. J. Conard. 2015. March, R. J., A. Lucquin, D. Joly, J. C. Ferreri, and M. Muhieddine. 2014. On the evidence for human use and control of fire at Schöningen. Journal Processes of formation and alteration of archaeological fire structures: of Human Evolution 89:181–201. complexity viewed in the light of experimental approaches. Journal of Stewart, O. C. 1956. Fire as the first great force employed by man. In Wenner- Archaeological Method and Theory 21:1–45. Gren International Symposium: man’s role in changing the face of the earth. Mentzer, Susan M. 2014. Microarchaeological approaches to the identification W. L. Thomas, ed. Pp. 10–13. Chicago: University of Chicago Press. and interpretation of combustion features in prehistoric archaeological sites. Théry-Parisot, I. 2001. Economie des combustibles au paléolithique: expéri- Journal of Archaeological Method and Theory 21:616–668. mentation, taphonomie, anthracologie, No. 20. Paris: CNRS Editions. Movius, H. L. 1966. The hearths of the Upper Perigordian and Aurignacian Théry-Parisot, I., and S. Costamagno. 2005. Propriétés combustibles des horizons at the Abri Pataud, Les Eyzies (Dordogne), and their possible sig- ossements: Données expérimentales et réflexions archéologiques sur leur nificance. American Anthropologist 68(2):296–325. emploi dans les sites paléolithiques. Gallia Préhistoire 47(1):235–254. Moyano, I. T., D. Barsky, D. Cauche, V. Celiberti, S. Grégoire, F. Lebegue, Waechter, J. D. A. 1951. Excavations at Gorham’s Cave, Gibraltar. Proceedings M. H. Moncel, and H. de Lumley. 2011. The archaic stone tool industry of the Prehistoric Society, n.s., 17(1):83–92. from Barranco León and Fuente Nueva 3, (Orce, Spain): evidence of the Walker, M. J., D. Anesin, D. E. Angelucci, A. Avilés-Fernández, Francesco earliest hominin presence in southern Europe. Quaternary International 243 Berna, A. T. Buitrago-López, Y. Fernández-Jalvo, et al. 2016. Combustion at (1):80–91. the late Early Pleistocene site of Cueva Negra del Estrecho del Río Quípar Oakley, Kenneth. 1954. Evidence of fire in South African cave deposits. Na- (Murcia, Spain). Antiquity 90(351):571–589. ture 174:261–262. Weiner, S., Q. Xu, Paul Goldberg, J. Liu, and Ofer Bar-Yosef. 1998. Evidence ———. 1955. Fire as a Palaeolithic tool and weapon. Proceedings of the for the use of fire at Zhoukoudian, China. Science 281(5374):251–253. Prehistoric Society 21:36–48. White, Randall, Romain Mensan, Amy E. Clark, Elise Tartar, Laurent Mar- ———. 1956. The earliest fire-makers. Antiquity 30(118):102–107. quer, Raphaëlle Bourrillon, Paul Goldberg, et al. Technologies for the con- ———. 1957. The earliest use of fire. In Proceedings of the Third Pan-African trol of heat and light in the Vézère Valley Aurignacian. Current Anthro- Congress on Prehistory, Livingstone, 1955. J. Desmond Clark, ed., assisted pology 58(suppl. 16):S288–S302. by Sonia Cole. London: Chatto & Windus. Wrangham, Richard W. 2007. The cooking enigma. In Evolution of the human ———. 1961. Possible origins of the use of fire. Man 61:244. diet: the known, the unknown, and the unknowable. Peter S. Ungar, ed. Parker, C. H., E. R. Keefe, Nicole M. Herzog, J. F. O’Connell, and K. Hawkes. Pp. 308–323. Oxford: Oxford University Press. 2016. The pyrophilic primate hypothesis. Evolutionary Anthropology: Issues, ———. 2009. Catching fire: how cooking made us human.NewYork:Basic. News, and Reviews 25(2):54–63. ———. 2017. The cooking hypothesis. Current Anthropology 58(suppl. 16): Perlès, C. 1981. Hearth and home in the Old Stone Age. Natural History 90 S303–S313. (10):38–41. Wrangham, Richard, and N. Conklin-Brittain. 2003. Cooking as a biological Pettitt, P. B. 1997. High resolution Neanderthals? interpreting middle Palaeo- trait. Comparative Biochemistry and Physiology Part A: Molecular and In- lithic intrasite spatial data. World Archaeology 29(2):208–224. tegrative Physiology 136(1):35–46. Preece, R. C., J. A. Gowlett, S. A. Parfitt, D. R. Bridgland, and S. G. Lewis. Wrangham, Richard W., J. H. Jones, G. Laden, D. Pilbeam, and N. Conklin- 2006. Humans in the Hoxnian: habitat, context and fire use at Beeches Pit, Brittain. 1999. The raw and the stolen: cooking and the ecology of human West Stow, Suffolk, UK. Journal of Quaternary Science 21(5):485–496. origins. Current Anthropology 40(5):567–594. Pruetz, Jill D., and Nicole M. Herzog. 2017. Savanna chimpanzees at Fongoli, Wylie, A. 1985. The reaction against analogy. Advances in Archaeological Senegal, navigate a fire landscape. Current Anthropology 58(suppl. 16): Method and Theory 8:63–111. S337–S350. Zink, K. D., and D. E. Lieberman. 2016. Impact of meat and Lower Palaeo- Pruetz, Jill D., and T. C. LaDuke. 2010. Brief communication: reaction to fire lithic food processing techniques on chewing in humans. Nature 531:500– by Savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: con- 503. Current Anthropology Volume 58, Supplement 16, August 2017 S175

Recognizing Fire in the Paleolithic Archaeological Record

by Paul Goldberg, Christopher E. Miller, and Susan M. Mentzer

Everyone agrees that fire has played an important part in the history of the genus Homo. However, because of the sometimes ephemeral and ambiguous nature of the evidence for fire in the Paleolithic record, establishing when and how hominins actively interacted with fire has been difficult. Over the past several decades, multiple techniques have been developed and employed in the search for the origins of human use of fire. Because fire is a natural phenomenon, the identification of burned remains at an archaeological site is generally not considered to be, on its own, convincing evidence for human use of fire. Rather, much of the difficulty of identifying early evidence for fire use has hinged on the question of how to establish a more direct link between burned materials and human activity. Here, we advocate for an approach to the investigation of the history of hominin use of fire that emphasizes an integration of multiple techniques. In particular, we argue that a contextualized study conducted at the microscopic scale—what we call a microcontextual approach—shows the most promise for establishing a behavioral connection between hominins and fire in the archae- ological record.

Prelude structure, with obvious doorways and subdivisions of rooms. But the house had clearly been abandoned for a long time fi Look at gure 1. On a recent trip to the high Andes of Peru, since shrubs and other vegetation now grew throughout the Susan Mentzer and Chris Miller stopped to stretch their legs building. during the long ascent to the Puna. They pulled over onto the “Hey,” Mentzer said excitedly. “Look at this! There was a side of the rut-worn road and took in the sparsely vegetated bushfire!” Sure enough, many of the shrubs outside the struc- “ ” “ ” landscape. Look over there, Susan said. A ruined house. ture had been burned, with ashes and charcoal collected in neat Kurt Rademaker, who was leading the expedition, spoke up: circles below the charred remains of branches. “ Yeah, there are lots of those up here. Some of them date to Inca “Wow,” Miller said. “Some of the shrubs inside the ruin are ” or even pre-Inca times. Mentzer and Miller got out of the truck burned too!” Maybe it was the low oxygen of the high Andes, crammed with students and equipment and headed over. The but their minds began to wander. Shooting each other a know- remnants of the stone walls still provided a clear outline of the ing glance, they started to think of what this ruin would look like to archaeologists in the future. “ ’ fi ” “ Paul Goldberg It s clearly a bush re, Miller said. You can see where it is Professor Emeritus in the Department of Archae- ” ology at Boston University (675 Commonwealth Avenue, Boston, started over there at the base of the hill. “ ’ Massachusetts 02215, USA [[email protected]]), at the Institute for Yeah, but it didn t burn evenly across the surface . . . the Archaeological Sciences at the University of Tübingen (Rümelinstrasse vegetation is patchy and so each shrub, when it burned, left 23, 72070 Tübingen, Germany), and at the School of Earth and En- a circular feature of charcoal and ash behind.” vironmental Sciences of the University of Wollongong (Wollongong, “So it’s natural.” New South Wales 2502, Australia). Christopher E. Miller is Professor “Yes, but some of the shrubs inside the house also burned at the Institute for Archaeological Sciences at the University of Tü- and left those circular patches. You know,” Mentzer said, “in bingen (Rümelinstrasse 23, 72070 Tübingen, Germany [christopher ten or twenty years, or if this site ever becomes buried, those [email protected]]) and at the Senckenberg Centre for Human patches would look a lot like hearths to archaeologists.” Evolution and Paleoenvironment (Rümelinstrasse 23, 72070 Tübingen, “Yeah,” Miller said, a bit despondently. Germany). Susan M. Mentzer is a Research Scientist at the Institute for “And these burned patches are natural, but they are associ- Archaeological Sciences at the University of Tübingen (Rümelinstrasse 23, 72070 Tübingen, Germany [[email protected]]) ated with an archaeological site. And since the burned patches and at the School of Anthropology at the University of Arizona (1009 are inside the structure, it would be easy to assume that they ” East South Campus Drive, Tucson, Arizona 85721, USA). This paper are hearths, Mentzer said. was submitted 25 VII 16, accepted 21 III 17, and electronically published “Yeah,” Miller said again. “But how could archaeologists dis- 6 VII 17. tinguish between a natural fire like this and a hearth?” q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0003$10.00. DOI: 10.1086/692729 S176 Current Anthropology Volume 58, Supplement 16, August 2017

vated material remains and human activities and behavior. In sites from most time periods, the evidence for anthropogenic fire can appear to be quite convincing: constructed hearths, kilns, and other installations can be readily compared with analogous features known from historic and ethnographic studies (Gur- Arieh et al. 2013). Similarly, the presence of ceramic sherds, glass, bricks, metal, and slag all clearly point to the control of fire for craft and industrial practices in the past. Archaeological evidence for the use of fire becomes spottier when dealing with hunter-gatherer or other nonsedentary so- cieties, particularly those from the Pleistocene. Although some Upper Paleolithic people lined and paved their hearths with clay and stone (Karkanas et al. 2004; Movius 1966), thus mak- ing them more readily identifiable by archaeologists, most ev- idence for fire in the Pleistocene takes the form of scatters and concentrations of charcoal, ash, burned bone, heated stone, and reddened substrates. Recognizing and interpreting these mate- rials as evidence for use of fire, let alone control or production of fire, can be difficult (see Sandgathe 2017 for definitions of use, control, and production of fire). Furthermore, this difficulty is only compounded the farther back in time we look when we search for the origins of hominin interactions with fire. The evidence for fire that we encounter in the archaeologi- cal record can be different from many other types of artifacts found at Paleolithic sites. Unlike handaxes, cave paintings, or butchered bones, materials such as charcoal, ash, and other burned remains can be produced by humans but also by nat- ural processes. Evidence for naturally occurring fires in the form Figure 1. Archaeological site affected by a recent brush fire in Peru. of preserved charcoal is known from the geological record dat- A, Shrubs growing within and around an abandoned stone struc- ing to the Devonian, more than 350 Mya (Scott 2000). Simi- ture have been differentially affected by brush fires. One concen- tration of burned shrubs (arrow) is located in the center of the room. larly, Bordes (1957) reports evidence for heated chert from de- B, Detail of the burned shrubs pictured in A. The burning reached posits dating to the Miocene. Fire is essentially a natural chemical an intensity such that round concentrations of charcoal and ashes reaction that humans have learned to anticipate, use, control, formed. A color version of this figure is available online. and produce. The difficulties of first identifying heated materials in the Pa- “Easy,” Mentzer replied. “If you can date the house archi- leolithic record and second determining whether humans were tecture stylistically, and then run a radiocarbon date on the the agents responsible have been recognized by archaeologists charcoal from the hearth, you could see that they wouldn’t since at least the mid-twentieth century (Oakley 1954). Later match, assuming that there is a large enough hiatus between researchers, in particular Barbetti (1986), formalized the prob- abandonment and the bushfire.” lem: “Demonstrating that fire was used at an archaeological site “OK,” Miller said after a long pause. “But what if this is a two-step process. One must first find the evidence and show weren’t a historical site, but a Paleolithic site?” that fire was present. It is then necessary to demonstrate that They stared at each other without saying a word and silently it was associated with human activity” (771). returned to the truck to continue the ascent into the moun- In the 30 years since Barbetti laid out the two-step process of tains.1 evaluating evidence for fire in the archaeological record, many new methods and techniques have been introduced that aid in identifying burned materials at Paleolithic sites. Similarly, Introduction many new sites have been excavated and novel studies con- Fire—as with any aspect of the past studied by archaeologists— ducted that help us evaluate the nature of the association of requires us to establish clear theoretical links between exca- burned materials and human activity. However, despite these advances, Barbetti’s two-step approach should still form the foundation for any archaeological study of fire. 1. This is a fictionalized account of a true event. The dialogue has been Therefore, in this paper, we have decided to revisit and ex- changed and some literary license has been taken, but the scientificcon- pand on Barbetti’s arguments, focusing on how we address cerns are real. the following three questions about materials encountered in Goldberg, Miller, and Mentzer Recognizing Paleolithic Fire S177 archaeological sites: (1) Are they burned? (2) Were they re- natural and anthropogenic sediments, it is widely used as a covered from the primary location of heating, and if not, what first-order approach to the study of combustion features and is the nature of their deposition? and (3) Were they burned by burned materials (Berna and Goldberg 2008; Mentzer 2013; humans? Miller, Goldberg, and Berna 2013). We will present a range of approaches to identifying heated A number of additional microanalytical techniques outlined materials from archaeological sites. But we will also show how in table 1 can be conducted on either loose sediment or ma- positive results—those that show that materials have been terials present in micromorphology samples. Ideally, a study heated—do not alone constitute sufficient evidence for human of evidence for fire applies the same techniques on both loose use of fire. Rather, we argue that these techniques need to be and micromorphological samples, thereby supporting the in- applied in conjunction with high-resolution analyses of site tegration of results obtained microscopically with observations formation processes in order to form a link between material made at the site scale. Defined narrowly (or optimistically), a evidence for heating and human behavior. microcontextual approach to the study of a combustion fea- ture might mean that every analysis conducted as part of the The Microcontextual Approach study should be conducted directly on materials visible in a micromorphological thin section or in its corresponding in- The microcontextual approach was put forward by Goldberg durated sediment block. and Berna (2010) from concepts developed by researchers such Micromorphology and a microcontextual approach can also as Boivin (2004), Goldberg and Macphail (2006), Matthews be used as a type of “methodological filter” for determin- (2005), and Matthews et al. (1997). It is essentially the in situ ing which additional techniques could be applied to identify analysis of microscopic components in thin section using a va- heating and to assess the depositional association of heated riety of microanalytical techniques. The microcontextual ap- materials with other traces of human activity. In the possible proach provides a framework for the integration of these data combustion feature illustrated in figure 2, observation of three by extending the context of an archaeological artifact—its types of potentially heated material and fuel residues as well as matrix, its provenience, and its associations with other artifacts anthropogenic inclusions (fig. 2B) lead to an analytical trajec- (Renfrew and Bahn 2007)—to the microscopic scale. In this tory that includes micro-FTIR (m-FTIR; Fourier transform in- framework, we treat individual particles of anthropogenic sed- frared spectroscopy), organic petrology, and the scanning elec- iments as artifacts and microscopic deposits as assemblages tron microscope (SEM; fig. 2E). Similarly, the microcontextual (fig. 2). This approach has proven successful in the analysis of approach can also help discourage the use of certain other combustion features and burned materials in sites of all ages, techniques that may be unnecessary or produce misleading re- ranging from the Paleolithic (Goldberg and Berna 2010) to the sults. In the same example, the observation that the suspected Iron Age (Mentzer, Romano, and Voyatzis 2015; Toffolo et al. ashes have been chemically altered (fig. 2C) precludes the 2012). measurement of stable isotopes of carbon and oxygen from the The most effective and efficient way to monitor and control calcareous ashes (Mentzer and Quaid 2013). A microcontextual for (micro)context is the use of soil micromorphology (Courty, study ideally does not focus on a single feature or only on po- Goldberg, and Macphail 1989; Goldberg and Macphail 2006; tential evidence for human use of fire. Rather, microcontextual Macphail and Cruise 2001), which is the meso- and micro- analyses should be part of a broader geoarchaeological study scopic analysis of intact blocks of sediments or soils that have that attempts to understand how potential evidence for use of been indurated, sliced, polished, and prepared into thin sec- fire fi ts within a holistic site formation model. tions (Courty, Goldberg, and Macphail 1989). Observations made on micromorphological samples can include the miner- Examples alogical and organic composition of sedimentary components; the size, shape and sorting of grains; the porosity and micro- Roebroeks and Villa (2011), in their assessment and ranking structure of deposits; the presence of bedding or other micro- of sites containing evidence for early use of fire, give particular scopic sedimentary structures; and the fabric—the internal or- weight to studies that identify combustion features. In par- ganization of all these attributes (Bullock et al. 1985; Courty, ticular, they privilege studies that employ micromorphology or Goldberg, and Macphail 1989; Stoops 2003). Furthermore, mi- use multiple analytical techniques over those that rely on only cromorphological analysis can reveal both physical and chem- one method or on one line of evidence. The microcontextual ical postdepositional alteration of deposits that may be related approach, by providing a methodological and theoretical link to either natural or human agents. Anthropogenic physical al- between multiple analytical techniques and micromorphology, teration might include trampling or reworking of combustion is therefore ideally suited for the study of Paleolithic fire. In residues by dumping, middening, or hearth rake out. Natural, practice, this has only been carried out to a limited extent. chemical alteration might entail dissolution of ashes or bone Many of the techniques for studying burned materials were or the secondary precipitation of carbonates or phosphates. Be- developed first using loose sediment samples, and only recently cause the use of micromorphology facilitates the analysis of have they been applied to micromorphological samples. For postdepositional alteration and enables the distinction between example, Shahack-Gross et al. (2008) first investigated the Figure 2. Application of the microcontextual approach to a suspected combustion feature. A, Middle Stone Age deposits in the site of Sibudu, South Africa, contain discrete features (arrow) that appear in the field to be composed of charred material overlain by ashes. These features were sampled for micromorphological analysis. For more detailed information about the micromorphology at this site, see Goldberg et al. (2009). B, Thin section prepared from a different feature but with similar characteristics (incident light scan). The area photographed in parts C and D is indicated with the red box. C, Feature viewed using a petrographic microscope (plane-polarized light; PPL) can be likened to a microscopic stratigraphic section containing what appears to be a basal layer of rubified sediment (Ru) overlain by charred plant fragments (Ch) and topped with ashes (Ash). The sequence also contains microscopic artifacts such as bone fragments (B) and lithic debitage (L). The depositional history of the three layers is clear at this scale of observation. The contacts are gradational over several millimeters with no indication that they are derived from discrete depositional events. This relationship is consistent with a formation model for intact hearths proposed by Meignen et al. (2001). Each of the materials visible in this feature can be analyzed using additional techniques described in table 1. The bone fragments and charred plants can be analyzed using organic petrology. The ash layer can be investigated at higher resolution using a petrographic microscope or SEM, which facilitates morpho- logical identifications of different components (e.g., ash rhombs, phytoliths). The molecular and elemental compositions of the bones Goldberg, Miller, and Mentzer Recognizing Paleolithic Fire S179 isotopic composition of ashes using experimental materials and cined. Similarly, carbon and oxygen isotopic analysis suggested archaeological samples from the site of Amud Cave (Israel). that at least some of the deposits at the site consisted of re- This approach was extended to micromorphological samples crystallized ashes or a mixture of ashes and geogenic calcite. several years later (Mentzer and Quade 2013). Moreover, until At Qesem, a combination of micromorphology, FTIR, and recently, many microanalytical instruments were available to isotopic analyses clearly demonstrated the presence of fire. How- archaeologists on a limited basis. Nevertheless, there are a few ever, it was the contextualization of these analyses at the mi- cases in which a microcontextual study was conducted on croscopic scale that allowed the authors to demonstrate that Paleolithic fire features (Berna and Goldberg 2008; Berna et al. the burned remains were probably related to human behavior. 2012; Goldberg and Berna 2010; Lowe et al. 2016; Mallol, Karkanas et al. (2007) point out that some authors (e.g., Har- Mentzer, and Miller, forthcoming; Shahack-Gross et al. 2014; rold and Otte 2001) argue that natural fires, such as those caused Stahlschmidt et al. 2015). Here, we draw on several examples by lightning strikes or spontaneous combustion of guano, could from three Lower Paleolithic sites to illustrate how the ap- account for the presence of burned materials in caves. In con- proach can yield positive, negative, and ambiguous results. At trast, Karkanas et al. (2007) argue that spontaneous combus- the end, we highlight an application where we believe the ap- tion is unlikely because these types of fires require significant proach is most promising: in the analysis of traces of fire that amounts of organic material that were generally absent from have been affected by postdepositional processes. the deposits. Additionally, guano fires, according to the authors, burn at relatively low temperatures and do not produce com- pletely combusted remains. At Qesem, much of the evidence Positive Results from Qesem Cave for fire consists of completely combusted wood (in the form of Microcontextual analyses have been essential in establishing ashes) and charred and calcined bones, which generally would some of the earliest clear evidence for repeated use of fire by be expected to occur with higher-temperature fires. The most humans. At Qesem Cave (Israel), which is associated with an convincing argument is that small fragments of burned bone Acheulo-Yabrudian occupation that dates between ca. 400 and are found in thin, discrete lenses composed of pure ash. The 200 kya (Barkai et al. 2003; Gopher et al. 2005; Karkanas et al. micromorphological study documented the repetitive occur- 2007), researchers conducted a microcontextual study that re- rence of these distinct “microcontexts” throughout the se- lied heavily on micromorphology but was augmented with quence at Qesem, thereby providing strong evidence for the other techniques, such as FTIR analysis on loose sediment repeated use of fire by the site’s inhabitants. samples and bones and isotopic analysis of calcareous deposits. A further study at Qesem (Shahack-Gross et al. 2014) fo- The analysis on loose sediment samples returned negative or cused on a 300 kya stone-lined feature and employed a broader ambiguous results. For example, FTIR analysis of loose sedi- range of microcontextual analyses. The ca. 4 m2 feature was ment did not reveal any evidence that the clay fraction of the lined by a ring of rocks and contained laminated deposits cave deposits had been heated to above 5007C. In addition, within it (Shahack-Gross et al. 2014). In addition to micro- no phytoliths or siliceous aggregates were found in the acid- morphology and FTIR analysis on loose sediment samples insoluble fraction of the loose samples, possibly reflecting un- from the laminated deposits, the authors also conducted favorable preservation conditions in the alkaline environment m-FTIR measurements directly on thin sections including bone of the cave. However, in thin section, Karkanas et al. (2007) fragments and soil aggregates. These measurements confirmed identified clear evidence for fire in the form of numerous ex- that the bone fragments found in association within the lami- amples of calcareous ashes. The authors reported that at a mi- nations of ashes were burned to above 5007C. The authors also croscopic scale, the ashes were found in discrete layers, generally demonstrated the presence of clay aggregates that had been 2 cm thick and in association with small fragments of burned heated to above 4007C. The identification of heated clay ag- bone and clay-rich soil aggregates that appeared reddened and gregates using m-FTIR in the ash laminations shows in partic- probably heated. Supporting the micromorphological evidence ular the strength of the microcontextual approach because pre- for fire at the site, FTIR analysis of bone showed that some vious FTIR analysis of loose sediments did not provide any fragments contained pyrolyzed collagen, and some were cal- clear evidence for heating of clays above 5007C at the site. and ashes can be documented using m-FTIR and SEM-EDS. D, Same view as C, cross polarized light. Analyses with a light microscope provide some information about the composition of certain materials, serving as a methodological filter for further investigations. Here, the ash layer is isotropic, which indicates that it is not composed of calcite, a primary mineral constituent of wood ash (Canti 2003; Shahack-Gross and Ayalon 2013). Additional analyses that address the composition of this layer are warranted. E, Observation of the “ash” layer under higher magnification and PPL reveals that the characteristic morphologies of ash components (Canti 2003) are absent. F, Analyses of the area in E using m-FTIR reveal the presence of two main phases. The spectrum in red contains peaks indicative of phosphate (Ph) and silicates (Si), further identified as apatite and quartz, respectively. The spectrum in blue contains peaks indicative of an amorphous silicate, such as opal. Resin peaks (R) are also present. Volumetric replacement of ashes with phosphatic minerals has been documented in other Paleolithic contexts (Karkanas and Goldberg 2010; Goldberg et al. 2007). In this case, the micromorphology and m-FTIR results neither support nor refute the hypothesis that the feature is an intact hearth, but further microcontextual analyses of other components may prove more fruitful in the future. Table 1. Geoscientific analytical techniques for the study of ancient fire

Used to identify presence/ Analytical technique, sample type Type of information/data absence of heating? Other information Key references Micromorphology: • Oriented blocks of sediment Visual observation of sedimentary Yes. Can be used to identify some Allows the analyst to observe the Canti 2003; Courty, Goldberg, and indurated with resin and processed components using a petrographic by-products of combustion based spatial relationship between different Macphail 1989; Goldberg and into petrographic thin sections microscope and different sources of on mineralogical composition and components of the feature. Fabric Macphail 2003; Meignen et al. light (plane-polarized, cross polarized, morphology (e.g., ashes). and structure can be used to make 2001; Mentzer, Romano, and darkfield, fluorescent, reflected, interpretations about depositional Voyatzis 2015; Wattez 1988 oblique incident) context. Other components provide information about postdepositional history. FTIR:a • Loose sediment samples Type and strength of molecular bonds; Yes. Certain materials (e.g., clay Provides temperature ranges of heating Berna et al. 2007; Forget et al. 2015; • Heated lithics and rocks output in the form of spectra minerals, bone) can be analyzed for bone and clay minerals; can also Goldberg and Berna 2010; • In situ measurements on intact to document molecular changes that be used to identify primary and Regev et al. 2010; Thompson, blocks and thin sections collected occur with heating. secondary minerals. Under ideal Islam, and Bonniere 2013; S180 for micromorphological analyses preservation conditions, pyrogenic Weiner 2010; Weiner and Bar- using m-FTIR calcite can be distinguished from Yosef 1990; Weiner and Goldberg geogenic calcite. 1990; Weiner, Goldberg, and Bar- Yosef 1993; Weiner et al. 1995, 1998; Xu et al. 2015 XRD:b • Loose sediment samples Spacing and arrangement of atoms in Yes. Certain materials (e.g., clay Can be used to identify primary and Domanski and Webb 1992; and artifacts (bone, lithics) a crystal lattice; output in the form minerals, bone) can be analyzed secondary minerals. Can be used Rogers and Daniels 2002; • In situ measurements on of diffraction patterns to document molecular changes that to study heat treatment of lithic Schmidt et al. 2012, 2013; micromorphological samples occur with heating. materials. Weymouth and Mandeville 1975 using micro-XRD SEM-EDS:c • Loose sediment samples Elemental abundance for major and No. When one mineral phase is present, Karkanas 2010; Karkanas et al. 2002; • In situ measurements on polished or trace elements (Na through U); the relative abundance of elements Schiegl et al. 1996 carbon-coated micromorphological output in the form of spectra can be used to estimate the chemical samples formula and identify the mineral. Used primarily for the study of diagenesis of burned materials. Phytolith analysis: • Phytoliths extracted from loose Qualitative description of phytolith Yes. Measurements of the refractive Identification of phytoliths to plant Albert, Berna, and Goldberg 2012; sediment samples morphology; quantitative informa- index of individual phytoliths can type based on morphology can be Albert and Cabanes 2007; • In situ observation of phytoliths tion from analysis of populations be used to determine heating. used to study the types of fuels in Albert et al. 1999, 2000; visible in micromorphological hearths. Cabanes et al. 2010; Elbaum et al. samples 2003; Madella et al. 2002; Piperno 2006 Table 1 (Continued)

Used to identify presence/ Analytical technique, sample type Type of information/data absence of heating? Other information Key references Organic petrology: • In situ measurements on polished, Identification of microscopic fragments Yes. Charred plant tissues have Provides information about fuels and Clark and Ligouis 2010; resin-indurated block sediment of plants based on morphology, reflectance values that are different their state of decomposition before Goldberg et al. 2009; samples (from micromorphological reflectance measurements of tissues from those of humified tissues. burning. Stahlschmidt et al. 2015; Suárez- samples) and gels, and quantitative fluorescence Ruiz 2012 • Other materials (e.g., bones) mounted in epoxy and polished Electron spin resonance: • Measurements on heated bones Measurement of the g-value of charred Yes. Heated bones have characteristic Can be used to reconstruct heating Michel, Falguères, and Dolo 1998; organic component of bone ESR spectra. temperature. Schurr and Hayes 2008 Luminescence measurements: • Thermoluminescence on heated Measurement of the amount of radiation Yes. Heated materials have a measurable Can be used to study heat treatment Brodard et al. 2012; Mercier et al. flints or sediment damage sustained by a material luminescence signal. of lithic materials. Both techniques 2007; Richter 2007 • Optically stimulated luminescence following a heating event provide ages for burning events. on heated sediment Magnetic measurements: • Magnetic field survey on Enrichment in ferromagnetic minerals Yes. Heated sediments have higher Magnetic susceptibility measurements Barbetti 1986; Bellomo 1993; buried features (magnetometry) are visible as magnetic anomalies. magnetic susceptibility than unheated can be complicated by soil forming Dalan and Banerjee 1998; • Magnetic susceptibility on loose Magnetic susceptibility is a measure sediments. Iron-bearing minerals in processes with equifinality producing Gose 2000 sediment samples of the abundance of magnetized grains heated rocks record the intensity and ambiguous results in some cases. • Palaeomagnetism on heated rocks within sediment. Measurements of direction of the magnetic field at the Components of magnetization remnant magnetism can indicate time of burning. can be used to determine heating whether rocks were heated in the past temperatures. Paleomagnetic S181 measurements can be used to determine whether burned rocks are in situ. Organic chemistry: • Gas chromatography mass Identification of alkanes from plants Yes. Charring reduces the relative May be useful for identifying residues Almendros, Martin, and Gonzalez- spectrometry on sediment and fatty acids from burned animal abundance of long-chain n-alkenes. from cooking or animal processing Vila 1988; Buonasera 2005; containing organic material tissues and conditions of heating. Applications Buonasera et al. 2015; are thus far limited to Holocene Eckmeier and Wiesenberg 2009; contexts. Mallol et al. 2013; March 2013; March, Ferreri, and Guez 1993; Sistiaga et al. 2011 Measurements of stable isotopes of carbon and oxygen: • Loose sediment samples Identification of calcite sourced from No. Can indicate low- or high-temperature Mentzer and Quade 2013; Shahack- • Quasi in situ measurements wood ashes and its degree of burning in well-preserved samples. Gross and Ayalon 2013; Shahack- on powders drilled from recrystallization in an alkaline Gross et al. 2008 micromorphological blocks environment a FTIR p Fourier transform infrared spectroscopy. b XRD p X-ray diffraction. c SEM/EDS p scanning electron microscopy/energy-dispersive X-ray spectroscopy. S182 Current Anthropology Volume 58, Supplement 16, August 2017

Combining field observations with the results of the micro- study showed that the lithological character of the sediment contextual study, the authors could convincingly argue that the within the features did not differ significantly from those from feature probably represented a large-scale hearth that was re- outside of the features. Finally, none of the mineral magnetic peatedly used and around which various activities were con- parameters recorded from the sediment samples from the fea- ducted. Taken together, the studies by Karkanas et al. (2007) tures corresponded with parameters measured from control and Shahack-Gross et al. (2014) suggest that fire was repeatedly samples of the calcareous lake deposits that were experimen- used by humans at Qesem. Furthermore, the authors could also tally heated. The results of this study strongly suggested that the show that the way in which the occupants engaged with fire reddened features formed as a result of natural oxidation of the changed over time. Although different techniques such as mi- iron sulfide (pyrite) and organic material in the lake deposits. cromorphology, FTIR, and isotopic analysis could reveal the This process was probably induced by the artificial lowering of presence of heat-altered remains at Qesem, it was the integra- the water table at Schöningen during mining operations in the tion of these data within a microcontextual research framework late twentieth century. that allowed the authors to make clear behavioral inferences The Schöningen microcontextual study did yield some ev- regarding hominin interactions with fire in the past (Barkai idence for fire at the site, but the authors argue that it was et al. 2017). probably not related to human behavior. Sand-size fragments of charcoal are present within the features. However, organic petrographic analysis of these fragments showed that the char- Negative Results from Schöningen and Zhoukoudian coal was not derived from wood but rather was composed of Other microcontextual studies have shown that some Paleo- rounded fragments of peat. The authors argued that these lithic sites that were reported and widely cited as containing fragments came from natural peat fires and were subsequently evidence for fire in fact do not. One such case is Schöningen redeposited in the lake. (Lower Saxony, Germany), an open-air Lower Paleolithic site Similarly, the site of Zhoukoudian, Locality 1 (China), has dating to ca. 300 kya that is renowned for its excellent organic long been cited as representing early evidence for fire since its preservation, including wooden hunting implements. Thieme initial excavations in the 1920s and 1930s: the presence of burned (1997, 1999) presented several lines of evidence for anthro- bones and reddish layers were reported from both Layers 10 pogenic fire at Schöningen, including four purportedly burned and 4 and were interpreted as hearths (Black 1931; Jia-Chia wooden implements as well as several features that he called 1980; Jia and Huang 1990; Teilhard 1934; Wu and Lin 1983). Feuerstellen, or hearths. These features consisted of localized Micromorphology and FTIR analysis of the sediments and fea- reddened areas of calcareous marl that were on average 1 m2 in tures in both layers (Goldberg et al. 2001; Weiner et al. 1998) size. Excavators noted that there were no obvious remains of suggested that they were not related to heating. Layer 10 con- charcoal or ash. In a preliminary micromorphological study of sists of finely laminated silt and clay interbedded with yellow the features, Schiegl and Thieme (2007) suggested that the and reddish brown organic fragments with localized fragments evidence for fire alteration of the substrate was ambiguous of limestone; charcoal was not observed in the field. The finely based on the presence of mollusk shells that had not been heat laminated nature of the Layer 10 deposits is consistent with low altered. energy water or ponded deposition by the Zhoukoudian River, Expanding on the initial study of the features by Schiegl and which at that time entered the cave. In the field, channeling and Thieme, Stahlschmidt et al. (2015) conducted a more detailed erosion of Layer 10 is clearly visible. Additionally, the deposits microcontextual analysis employing micromorphology, FTIR of Layer 10 completely lack the red-black-white structure of and m-FTIR, organic petrology, mineral magnetic parameters, a combustion feature or hearth (Goldberg et al. 2001; fig. 2). and thermoluminescence. Their analysis of the reddened areas Micromorphological analysis also demonstrated that the red- showed that these features consisted of a thick, consolidated dening of Layer 4 is a result of diagenesis evidenced by the pres- crust of iron oxide that impregnated the calcareous lake deposits ence of fine, silt-size grains of hematite. These grains probably of the substrate. In thin section, the reddening observed in the formed during a period of subaerial exposure after the roof of field appeared to correlate with sediment containing staining the site had collapsed. Moreover, FTIR and elemental analyses and matrix impregnations of hematite, which result from nat- revealed no evidence of heat alteration of the limestone (Wei- ural redox processes (Lindbo, Stolt, and Vepraskas 2010). In ner et al. 1998). However, new data from recent excavations direct association with the hematite were calcareous lake sed- presented by Gao et al. (2017) may reveal the presence of in- iment and calcareous fragments of mollusk shells, which sug- disputable fireplaces. gest that the substrate could not have been heated above 5007C. Similarly, the presence of kaolinite within the feature, which Ambiguous Results from Wonderwerk was identified using FTIR analysis, suggested that the substrate had never been heated above 4007C. A study of experimentally Whereas the microcontextual studies at Zhoukoudian and heated sediment collected from Schöningen suggested that the Schöningen clearly demonstrated absence of evidence for an- degree of reddening observed in the features would occur only thropogenic fire and the study at Qesem produced clear evi- at higher temperatures. Additionally, the thermoluminescence dence for repeated use of fire, other microcontextual studies Goldberg, Miller, and Mentzer Recognizing Paleolithic Fire S183 have produced results that are more difficult to interpret. At where less-than-ideal preservation conditions at Paleolithic sites Wonderwerk Cave (Northern Cape, South Africa), Berna et al. have greatly altered the evidence for human use of fire. Indeed, (2012) presented several lines of evidence for heated materials many of the analytical techniques that we champion in earlier from a single layer dated to ca. 1.0 Ma. The study integrated sections of this paper were developed for the study of com- micromorphology with m-FTIR, FTIR analysis of excavated bustion features affected by chemical diagenesis and physical bone fragments, and lithic analysis. The authors report potlid- reworking or a combination thereof at the Levantine Middle ded lithics, which they interpret as having fractured as a re- Paleolithic sites of Kebara, Hayonim, and Amud (Berna and sult of heating, probably above 5007C. In addition, they noted Goldberg 2008; Schiegl et al. 1994; Weiner, Goldberg, and Bar- that bones recovered during excavation exhibit colors that are Yosef 2002; Weiner et al. 1995, 2007). In each of these sites, consistent with bones that have been exposed to heat (Stiner research teams expanded on an initial desire to document the et al. 1995). FTIR analysis of several of these bones demon- presence of burned materials in primary or secondary position, strated that they were heated to above 4007C, and additionally first integrating micromorphological studies with mineralogy some altered clay adhering to a gray-colored bone suggested and chemical analyses and later microsampling the intact features heating of the clay to somewhere between 4007C and 7007C. for other types of microscopic artifacts, such as phytoliths. In this FTIR spectra collected on white bone fragments did not con- way, these sites served as laboratories for the development of tain peaks that would indicate the high temperatures that are analytical techniques that, thanks to recent improvements in typically associated with calcination (Thompson, Islam, and instrumentation, can now be directly applied to micromor- Bonniere 2013). In micromorphological samples, Berna and col- phological samples (see table 1). leagues (2012) identified pseudomorphic grains of oxalate crys- At the sites of Kebara and Hayonim, evidence for fire seems tals composed of calcite, which the authors identified as ashes readily apparent in the field (Meignen, Goldberg, and Bar- following (Canti 2003; Wattez 1988); however, Berna et al. Yosef 2007; Meignen et al. 2001), where the most obvious ex- (2012) and Goldberg, Berna, and Chazan (2015) note that cal- pression is structured combustion features consisting of a some- cified plant remains in the form of rhizoliths and other plant times rubified basal substrate overlain by a charcoal-rich zone, fragments are also present in the same deposits, suggesting that which is capped by calcareous or diagenetically altered ashes the formation history of these layers may be complex. Addi- consisting of apatite (dahllite) and other phosphate minerals tionally, m-FTIR analysis conducted directly on the thin sec- (Meignen et al. 2001) In profile, these features vary from tions identified bone that was also heated to between 4007C centimeter- to decimeter-diameter lenses (e.g., Hayonim Cave) and 5507C. to thin, centimeter-thick stringers within depressions. In plan The authors employed microcontextual analyses to identify view the features are generally circular (Berna and Goldberg the presence of heated materials at Wonderwerk Cave. To ad- 2008; Goldberg and Bar-Yosef 1998; Meignen, Goldberg, and dress the association of this evidence with hominin behavior, Bar-Yosef 2007), although at Amud Cave, only the cemented they considered both microscopic and site-scale formation pro- ashy portions of the tripartite sequence of combustion features cesses of the deposits in question. The authors first addressed seem to have survived, the rest of the sediments having been whether the burning could have occurred independently of homogenized by bioturbation (Berna and Goldberg 2008; Ma- human activity. They argued that because the burned materials della et al. 2002). At Kebara and Hayonim, similar vertical se- are distributed throughout a thick archaeological stratum, it is quences of basal rubification, burned organic matter or charcoal, unlikely that wildfires could have repeatedly extended deep into and ashes can be seen in thin section, which also reveals repeated the cave from the surrounding landscape. Second, they noted stacking of combusted layers that are not evident in the field. that whereas none of the burned materials were recovered in In the field, many of the features were identified as obvious primary position, the angularity of the bone and ash fragments combustion residues. Other features reveal an important ca- observed in thin section rules out the possibility that these veat in trying to evaluate the former presence of fire because of materials were transported into the cave by natural processes. diagenesis: the physical and chemical changes following de- Berna et al. (2012) point out that without a clear combustion position. In Kebara and Hayonim, some of the original com- feature, it is difficult to demonstrate hominin control of fire ponents, especially calcareous ashes, dissolved or were trans- (Roebroeks and Villa 2011). However, they argue that there is a formed into other minerals, such as phosphates (Berna and clear association of human activity and fire at Wonderwerk, Goldberg 2008; Schiegl et al. 1994, 1996; Weiner, Schiegl, and implying human knowledge, if not use, of fire in the cave. Bar-Yosef 1995; Weiner et al. 1995, 2007). Although these changes can be readily observed in the field where the original combustion structures still exist (e.g., Kebara Cave; Meignen, Difficult Cases: The Levantine Middle Paleolithic Goldberg, and Bar-Yosef 2007), other components, such as In the three types of scenarios described above, microcontex- bone, have been completely dissolved (Goldberg et al. 2007; tual studies have contributed to answering the three questions Schiegl et al. 1994; Weiner, Goldberg, and Bar-Yosef 1993, 2002; about burned materials that were raised at the beginning of this Weiner et al. 1995). Based on the typical sequence of rubified paper. We feel that in addition to linking humans to burned sediment overlain by charcoal and ashes that was documented materials, the microcontextual approach is instrumental in cases in better-preserved areas of the sites (Meignen et al. 2001), S184 Current Anthropology Volume 58, Supplement 16, August 2017 similar features containing phosphate minerals and amorphous samples in the field: it was not readily apparent what was re- silicates instead of ashes were interpreted as hearths affected by ally being measured. For example, apatite in a sediment sample diagenesis. could be sand- or silt-size fragments of bone or a secondary In addition to micromorphology, a number of other tech- precipitate. Thus, in order to have an idea of what is actually niques were employed to confirm the presence of anthropo- being analyzed in the sample, researchers have developed new genic fire within the sites. In Kebara Cave, researchers docu- microanalytical techniques such as m-FTIR and micro-XRD mented not only the presence of numerous hearth structures (m-XRD; Berna, forthcoming; Berthold and Mentzer, forth- but also the occurrence of wood phytoliths, which would not coming). Although these techniques were not available at the have accumulated within the cave naturally (Albert, Berna, and time, similar microcontextual analyses were conducted di- Goldberg 2012; Albert et al. 2000). Furthermore, wood phy- rectly on thin sections from Kebara and Hayonim using SEM/ toliths also occur in cave sediments that do not contain hearth energy-dispersive X-ray spectroscopy (SEM/EDS; Schiegl et al. structures that are visible in the field; these types of phytoliths 1996). Kebara Cave was also the site of one of the first stud- are absent from control samples of terra rossa collected outside ies to integrate phytolith analyses from loose samples with the site. In Amud Cave, phytoliths from calcareous ashes and micromorphology of combustion features (Albert, Berna, and surrounding sediments were studied using micromorphology Goldberg 2012), an approach that was recently extended to and stable isotope analysis. The phytoliths were predominantly include phytolith identifications in thin section (Wadley et al. from wood (including palm and fig), whereas grass phytoliths, 2011). which are fresh and composed of spikelets, were thought to Despite these methodological advances, true microcon- have been brought into the cave and accumulated within the textual studies from the Levantine “laboratory” sites are lim- anthropogenic ashy units (Madella et al. 2002). Thus, the ubiq- ited. However, a test case for the microcontextual approach uitous presence of wood phytoliths in ashes suggested that these was developed for the site of Kebara and presented by Berna siliceous materials could be an indication of fireplaces in cases and Goldberg (2008). They analyzed a large, well-structured when the more soluble calcitic component might have been hearth containing an organic-rich layer with bones overlain dissolved. by ashes with both micromorphology and m-FTIR, which One of the most important things to come out of the Le- provided temperature estimates of the different components: vantine cave studies was an improved understanding of the (a) combusted plant fuel, heated at ~3007C, and the included microscopic expression of burned materials in different depo- burned bones heated at ~3007C; ( b) the overlying !1 mm thick sitional contexts. The fabric and structure of burned mate- accumulation of phosphatized calcitic ashes with (c) inclusions rials, which are visible in thin section, can help one determine of calcined bone heated at ≥5507C, and (d) burned soil particles whether the burned materials (1) are in their original place; heated at ≥5007C. These results yield evidence of heating of (2) have been reworked locally by natural processes such as various components within the different parts of the hearth bioturbation (Berna and Goldberg 2008; Goldberg and Bar- structure and also provide additional FTIR data to corroborate Yosef 1998; Madella et al. 2002), colluviation, or runoff (Gold- previous studies of the diagenesis of the features (e.g., Schiegl berg et al. 2007); or (3) have been reworked by human activities, et al. 1996; Weiner et al. 2007). This work demonstrates how such as dumping or hearth rake out (Goldberg 2003; Kuhn et al. such an approach yields a more holistic picture of the condi- 2009; Meignen, Goldberg, and Bar-Yosef 2007; Miller 2015; tions of formation and preservation of hearths. Schiegl et al. 2003). At the site of Kebara, a large concentra- tion of ashes in the rear of the cave was interpreted as a midden only after micromorphological analyses ruled out other agents Approaches to Recognizing Heated of redeposition. Such evidence for intentional collection and Archaeological Materials movement of burned materials within the living environment not only provides an indirect marker of human intentionality Steps toward positive recognition of fire in the Paleolithic ar- in the control and maintenance of fire but also shows that this chaeological record take place in both the field and in the lab- behavior was repeated over long periods of time. oratory. The field is the first place where the context of possible Many of the techniques that, when combined, form the basis burned material can be established, including the mutual as- of a microcontextual analysis of a combustion feature, were sociations among objects and features as well as their connec- developed during the decades-long excavations at these sites. tion to the deposits that contain them. In Kebara and Hayonim Caves, mineralogical analyses were In the field, one can note possible indicators of in situ fire, classically carried out using FTIR in the field on bulk sediment including rubification of sediments, the presence of ashy sed- samples that were ground and made into a pellet using KBr as a iment, charcoal and char, and other burned objects, such as fire- binder; supplemental confirmations followed in the laboratory cracked rock and burned or calcined bone. In younger sites using X-ray diffraction (XRD; e.g., Weiner, Goldberg, and Bar- intact, structured features, such as pits (including stone-lined or Yosef 1993; Weiner et al. 1995, 1998). Within the past decade, plastered pits; e.g., Thoms 2008, 2009) can be primary com- however, it became clear that certain contextual evidence was bustion features. Mounded accumulations and middens can lost using bulk samples, even if they were collected as !1g also be loci of reworked combusted materials, like those ob- Goldberg, Miller, and Mentzer Recognizing Paleolithic Fire S185 served at sites such as Kebara and Üçağızlı (e.g., Goldberg 2003; mation they provide, and the capacities in which they can be Kuhn et al. 2009; Meignen, Goldberg, and Bar-Yosef 2007). informative about past use of fire are summarized in table 1. Fieldwork aimed at recovering data about fire should be Despite the plethora of techniques that can be successfully focused on observing the setting of the site and its lithostra- employed in the identification of materials subjected to heat tigraphy. For example, noting the shape, size, nature, and con- in the past, none of the techniques on their own can clearly text of suspected burned zones may be helpful for deciding demonstrate that the source of this heat was directly related whether a reddened zone is the result of heating or diagenesis to human behavior. Barbetti (1986) suggested that the mac- (Cushing et al. 1986; Pigati et al. 2014). roscopic association of burned areas with concentrations of In addition, objects and features should be recorded precisely unambiguously cultural material may provide strong but not in the field using, for example, a theodolite system (Dibble and necessarily conclusive evidence for anthropogenic heating. How- McPherron 1988; McPherron and Dibble 2002; McPherron, ever, as in the Peru example from the beginning of this paper Dibble, and Goldberg 2005). This or a similar type of strategy (fig. 1), natural fires can mimic anthropogenic ones, so that provides both accurate 3-D data on objects, features, and sam- the simple association of demonstrably burned materials with ples and also virtual real-time data in the field to assess the in- cultural artifacts can be ambiguous at best and misleading at tegrity of deposits, as well as the location and context of sam- worst. Since 1986, archaeological scientists have heeded Bar- ples collected for further analysis in the laboratory. betti’s call for the application of new techniques for identifying Laboratory analyses are needed to supply qualitative and heated remains in the archaeological record, but the problem of quantitative data that, when combined with field observations, demonstrating human agency in the thermal alteration of ar- can document—or at least suggest—the former presence of chaeological materials remains. More attention must be paid to fire. The materials Paleolithic archaeologists often cite as evi- the deposit or deposits in which burned materials are found. dence for past fire include ashes, charcoal, burned bone, heated We believe that the application of geoarchaeological meth- (often reddened) substrates, and fire-altered rocks; most of ods is necessary to establish clear links between human behav- these materials are initially identified in the field. For example, ior and fire in the Paleolithic archaeological record. As geo- a deposit that appears grayish and silty is described as ashy, archaeologists working at the microscopic scale, we argue here black flecks encountered during excavation are called charcoal, for a research strategy that moves beyond simply associating black and white bones are termed charred and calcined, and cultural artifacts with burned remains in the field. Rather, we reddened patches of sediment are fire-altered substrate. Al- advocate for the use of high-resolution approaches to the study though these types of observations provide the first line of of burned materials in Paleolithic archaeological sites. These evidence for the possible presence of fire at an archaeologi- approaches can be broadly grouped under the category of “mi- cal site, on their own, field observations are not sufficient to croarchaeology” (sensu Weiner 2010). But more importantly, demonstrate heating, let alone human use of fire. This is be- we stress the integration of different types of analyses and con- cause interpreting these field observations is confounded by textualization of burned materials within macroscopic and mi- equifinality: the characteristics that we most often use to iden- croscopic deposits. tify potential fire evidence in the field are not solely produced fi fi fi by re. For example, most eld identi cations of burned ma- Summary and Concluding Comments terials—following guidelines that we advocate (e.g., Meignen et al. 2001)—are based on color: charcoal is black, burned bone As seen in the preceding case studies, determining whether one is black or white, ash is gray, and heated substrates are red. has evidence for human use of fire at a Paleolithic site is more However, color is not always a useful indicator for evidence of complicated than simply establishing whether burned remains heating. Bone that has been subjected to heating can undergo are associated with cultural artifacts. First, one must determine color changes (Shipman, Foster, and Schoeninger 1984; Stiner that what is suspected to be burned—whether charcoal, ash, et al. 2001), but similar color changes, particularly blackening, burned bone, or heat-altered stone and substrate—is in fact can be mimicked by mineral staining (Shahack-Gross, Bar- burned. Multiple studies including those at Schöningen (Stahl- Yosef, and Weiner 1997). Additionally, black-colored organic schmidt et al. 2015) and Zhoukoudian (Goldberg et al. 2001; material may be carbonized; however, humified organic ma- Weiner et al. 1998) have repeatedly shown that simple field terial can also appear black (e.g., Stahlschmidt et al. 2015; Tay- identifications of heated remains are not reliable; laboratory lor et al. 1998). Similarly, reddened patches of sediment can be tests are necessary for firmly establishing whether an object caused by heating, but they can also form through natural pro- has been subjected to heat in the past. Thankfully, numerous cesses of oxidation (Canti and Linford 2000; Stahlschmidt et al. techniques have been developed over the past several decades 2015). Therefore, any field identification of potential evidence for examining how archaeological materials have been trans- for fire must be followed up by laboratory analyses that can formed through heating. either confirm or refute the field observations. Over the past On the other hand, when these techniques are used alone, several decades, numerous laboratory techniques and methods they cannot demonstrate that the agent of heating was human have been developed and applied to suspected burned materials rather than natural (see also Aldeias 2017). One must establish from archaeological sites. These techniques, the type of infor- whether the burned materials are located where originally S186 Current Anthropology Volume 58, Supplement 16, August 2017 burned or whether they have been redeposited or transported a microcontextual framework—can be either difficult to in- since combustion. Determining the depositional history of terpret or ambiguous. For sites like Kebara and Hayonim, the burned materials—along with any other type of material con- researchers were able to demonstrate that the repeated burning tained within an archaeological site—requires that one con- of hearths produced the sedimentary sequence seen at these duct a geoarchaeological study that develops a holistic site sites. However, because the calcareous ashes that originally com- formation model. A geoarchaeological perspective is necessary posed these hearths were susceptible to diagenesis, the researchers because context is more than the association of certain types needed to study the chemical environment of the cave and how of materials within a specific layer or site. Rather, context is the this influenced the preservation of ashes. A microcontextual entire history of how these objects came to be associated with approach here was essential for understanding the complex one another. By establishing the depositional history of burned interplay of depositional and postdepositional processes and remains and cultural materials, we can more readily assess the for establishing repeated use and control of fire by Neander- role that humans played in the formation of burned archae- thals in these sites. ological remains. The researchers at Wonderwerk (Berna et al. 2012) could To illustrate this point, we can look at two examples from the demonstrate at a microscopic scale that burned remains were case studies above. At Schöningen, Stahlschmidt et al. (2015) found in association with cultural artifacts. And while they reported the occurrence of small fragments of charcoal within could argue that the burned materials are probably anthro- layers containing cultural remains, namely, wooden spears and pogenic, they were circumspect in their interpretation of the butchered bones. If we assume that context is simply a matter results: an association of burned remains and artifacts may of association, then one could conclude that the fragments of suggest knowledge and possible use of fire, but it does not charcoal were produced by humans. However, when they ap- necessarily imply control or production. plied a microcontextual approach, Stahlschmidt et al. (2015) We can now revisit the ambiguous case from the Peruvian demonstrated that the fragments of charcoal were likely pro- Andes presented at the beginning of the paper. We believe we duced elsewhere as a result of peat fires and were subsequently were rightly concerned that if the remains of the bushfires transported from their original place of combustion and re- entered into the archaeological record, it may be difficult—if deposited within the lake sediments. These results strongly not impossible—to determine whether they were natural or suggest that the charcoal fragments were formed by natural cultural. Certainly, field observations would establish that cir- fires despite their association with cultural remains. In contrast, cular patches of charcoal and ash were associated with cultural at Qesem Cave, Karkanas et al. (2007) and Shahack-Gross remains. Even the location of one of the patches—within the et al. (2014) were able to demonstrate that various types of center of the structure—would imply, based solely on associ- remains—such as ashes, burned bone, and heated soil aggre- ation, that it was likely cultural and may even represent a cen- gates—were associated with one another within single lenses. tral, domestic hearth. We suggested that given the possibility of By establishing the association of these burned materials within stylistically dating the architecture of the house and radio- individual depositional contexts and by investigating the for- metrically dating the charcoal from the “hearth” one could mation processes of the cave, the authors could convincingly potentially establish a chronological association of the feature argue that such depositional contexts were not natural and with the architecture. This scenario could still prove prob- were therefore probably anthropogenic. lematic if there is an old-wood effect, or if the house had been Recent studies, particularly those conducted at Qesem (Kar- used for a long period of time. Even these concerns would kanas et al. 2007; Shahack-Gross et al. 2014) show how a micro- be moot in a similar situation from a Paleolithic site, where contextual approach can provide insights into fire-related stylistic dating of artifacts covers broad time periods and ra- behavior not readily gained from other types of studies. At diometric dating has large standard errors or is not directly Qesem, the researchers presented clear evidence for repeated, applicable. habitual use of fire throughout the sequence and for repeated However, a microcontextual analysis that would examine control of fire in the form of a multiphase hearth. These types the depositional context of the burned remains may help re- of observations go a long way in directly testing evolutionary solve the issue. For example, a micromorphological study may models of hominin interaction with fire. However, they are provide evidence for an occupational hiatus between the last possible only through a microscopic approach that emphasizes use of the structure (such as a floor or concentrated occupa- integration and contextualization of multiple analytical tech- tional debris) and the burned feature. This evidence could be a niques. thin layer formed from the collapse of the roof of the structure Despite the promise that the microcontextual approach has or could include a thin layer of windblown sand, which would shown, we should not see it as a panacea for making the iden- only accumulate once the building had been abandoned. Sim- tification and interpretation of fire use in the Paleolithic easy ilarly, a micromorphological study may identify increased bio- or simple. In fact, applications of the microcontextual approach logical activity directly below the burned feature or even for- to sites with potential evidence for early use of fire have shown mation of a soil, which would imply stasis in accumulation and how complicated the issue can be. As seen in several of the case probably abandonment of the structure. Although all of these studies presented above, the results—even when placed within scenarios are hypothetical, they illustrate how an interpreta- Goldberg, Miller, and Mentzer Recognizing Paleolithic Fire S187 tion of human interaction with fire requires a thorough and Almendros, G., F. Martin, and F. J. Gonzalez-Vila. 1988. Effects of fire on humic and lipid fractions in a Dystric Xerochrept in Spain. Geoderma 42: detailed understanding of the formation history of an entire 115–127. site that can only be obtained at the microscopic scale. Barbetti, M. 1986. Traces of fire in the archaeological record, before one Researchers have successfully applied the microcontextual million years ago. Journal of Human Evolution 8:771–781. Barkai, Ran, Avi Gopher, S. Lauritzen, and A. Frumkin. 2003. Uranium series approach to critical sites that are, or have been, often quoted dates from Qesem Cave, Israel, and the end of the Lower Palaeolithic. Nature to document the presence of fire in early human history, in- 423:977–979. cluding Zhoukoudian and Schöningen. Microcontextual anal- Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a reason: fi barbecue at Middle Pleistocene Qesem Cave, Israel. Current Anthropology 58 yses have also provided suggestive evidence of re at sites that (suppl. 16):S314–S328. were not readily thought to have them at first glance (e.g., Bellomo, R. 1993. A methodological approach for identifying archaeological Wonderwerk). In this light, it is clear that such multimethod evidence of fire resulting from human activities. Journal of Archaeological Science 20:525–554. approaches need to be expanded to other sites, particularly Berna, Francesco. Forthcoming. Micro-FTIR. In Encyclopedia of archaeological early ones when circumstances permit, that have been thought soil and sediment micromorphology. G. Stoops and C. Nicosia, eds. Hoboken, to have evidence of early use of fire (e.g., Swartkrans, Beeches NJ: Wiley-Blackwell. Berna, Francesco, A. Behar, R. Shahack-Gross, J. Berg, J. Zorn, E. Boaretto, Pit, Chesowanja, Olorgesailie) or other, yet-to-be-discovered A. Gilboa, et al. 2007. Sediments exposed to high temperatures: reconstruct- sites. ing pyrotechnological processes in Late Bronze and Iron Age strata at Tel Dor Attempts to study hominin interactions with fire in the past (Israel). Journal of Archaeological Science 34:358–373. Berna, Francesco, and Paul Goldberg. 2008. Assessing Paleolithic pyro- necessarily depend on the context of the objects, features, de- technology and associated hominin behavior in Israel. Israel Journal of posits, and the site in its landscape (see examples in Aldeias Earth Sciences 56:107–121. Berna, Francesco, Paul Goldberg, L. K. Horwitz, J. Brink, S. Holt, M. Bamford, 2017; James 1989; Sandgathe 2017). So much of the older fi — and Michael Chazan. 2012. Microstratigraphic evidence of in situ re in the literature and excavations and sadly, many current ones as Acheulean strata of Wonderwerk Cave, Northern Cape Province, South well—fail to document or appreciate deposits, stratigraphy, Africa. Proceedings of the National Academy of Sciences of the USA 109:7593– geoarchaeology, and site formation, so we have a rather eclectic 7594. fi Berthold, C., and Susan M. Mentzer. Forthcoming. Micro X-ray diffraction. In and fuzzy view of re from ancient and even more recent Encyclopedia of archaeological soil and sediment micromorphology. G. Stoops prehistoric sites. The message for the future seems clear: we and C. Nicosia, eds. Hoboken, NJ: Wiley-Blackwell. have to make a concerted effort to closely examine the context Black, D. 1931. Evidence of the use of fire by Sinanthropus. Bulletin of the Geological Society of China 11:107–108. of archaeological deposits at all scales using microcontextual Boivin, N. 2004. Landscape and cosmology in the South Indian Neolithic: new approaches (e.g., micromorphology, m-FTIR, micro-XRF, m-XRD). perspectives on the Deccan ashmounds. Cambridge Archaeological Journal Without such an approach, we can hope to achieve only in- 14:235–257. Bordes, F. 1957. Review of K. P. Oakley “Fire as a Paleolithic tool and weapon.” complete results and answers. L’Anthropologie 61:314–317. Brodard, A., P. Guibert, F. Lévêque, V. Mathé, L. Carozza, and A. Burens. 2012. Thermal characterization of ancient hearths from the cave of Les Acknowledgments Fraux (Dordogne, France) by thermoluminescence and magnetic suscep- tibility measurements. Quaternary Geochronology 10:353–358. A. Sistiaga Gutiérrez is gratefully acknowledged for a primer in Bullock, P., N. Fedoroff, A. Jongerius, G. Stoops, T. Tursina, and U. Babel. 1985. Handbook for soil thin section description. Wolverhampton: Waine organic matter, and comments from V. Aldeias and D. Sand- Research. gathe helped to clarify some of the text in its early stages. Buonasera, T. 2005. Fatty acid analysis of prehistoric burned rocks: a case Deutsche Forschungsgemeinschaft grant MI 1748/1-1 funded study from central California. Journal of Archaeological Science 32:957–965. analyses presented in figure 2. K. Rademaker facilitated our Buonasera, T. Y., A. H. Tremayne, C. M. Darwent, J. W. Eerkens, and O. K. Mason. 2015. Lipid biomarkers and compound specific d13C analysis indi- fortuitous discovery in the Andes. cate early development of a dual-economic system for the Arctic Small Tool tradition in northern Alaska. Journal of Archaeological Science 61:129–138. Cabanes, D., Carolina Mallol, I. Exposito, and J. Baena. 2010. Phytolith evi- References Cited dence for hearths and beds in the late Mousterian occupations of Esquilleu cave (Cantabria, Spain). Journal of Archaeological Science 37:2947–2957. Albert, R. M., Francesco Berna, and Paul Goldberg. 2012. Insights on Nean- Canti, M., and N. Linford. 2000. The effects of fire on archaeological soils and derthal fire use at Kebara Cave (Israel) through high resolution study of sediments: temperature and colour relationships. Proceedings of the Pre- prehistoric combustion features: evidence from phytoliths and thin sections. historic Society 66:385–395. Quaternary International 247:278–293. Canti, M. G. 2003. Aspects of the chemical and microscopic characteristics of Albert, R. M., and D. Cabanes. 2007. Fire in prehistory: an experimental plant ashes found in archaeological soils. Catena 54:339–361. approach to combustion processes and phytolith remains. Israel Journal of Clark, J. L., and B. Ligouis. 2010. Burned bone in the Howieson’s Poort and Earth Sciences 56:175–189. post-Howieson’s Poort Middle Stone Age deposits at Sibudu (South Africa): Albert, R. M., O. Lavi, L. Estroff, S. Weiner, A. Tsatskin, A. Ronen, and S. Lev- behavioral and taphonomic implications. Journal of Archaeological Science Yadun. 1999. Mode of occupation of Tabun Cave, Mt Carmel, Israel during 37:2650–2661. the Mousterian period: a study of the sediments and phytoliths. Journal of Courty, M.-A., Paul Goldberg, and R. I. Macphail. 1989. Soils and micromor- Archaeological Science 26:1249–1260. phology in archaeology. Cambridge: Cambridge University Press. Albert, R. M., S. Weiner, O. Bar-Yosef, and L. Meignen. 2000. Phytoliths in Cushing, J., A. M. Wenner, E. Noble, and M. Daily. 1986. A groundwater the Middle Palaeolithic deposits of Kebara Cave, Mt Carmel, Israel: study of hypothesis for the origin of “fire areas” on the northern Channel Islands, the plant materials used for fuel and other purposes. Journal of Archaeo- California. Quaternary Research 26:207–217. logical Science 27:931–947. Dalan, R. A., and S. K. Banerjee. 1998. Solving archaeological problems using Aldeias, Vera. 2017. Experimental approaches to archaeological fire features techniques of soil magnetism. Geoarchaeology 13:3–36. and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– Dibble, Harold L., and Shannon P. McPherron. 1988. On the computerization S205. of archaeological projects. Journal of Field Archaeology 15:431–440. S188 Current Anthropology Volume 58, Supplement 16, August 2017

Domanski, M., and J. A. Webb. 1992. Effect of heat treatment on siliceous Grotte XVI, Dordogne, France. Journal of Archaeological Science 29:721– rocks used in prehistoric lithic technology. Journal of Archaeological Science 732. 19:601–614. Karkanas, P., R. Shahack-Gross, A. Ayalon, M. Bar-Matthews, Ran Barkai, Eckmeier, E., and G. L. B. Wiesenberg. 2009. Short-chain n-alkanes (C16–20) A. Frumkin, Avi Gopher, and M. C. Stiner. 2007. Evidence for habitual use in ancient soil are useful molecular markers for prehistoric biomass burning. of fire at the end of the Lower Paleolithic: site-formation processes at Qesem Journal of Archaeological Science 36(7):1590–1596. Cave, Israel. Journal of Human Evolution 53:197–212. Elbaum,R.,S.Weiner,R.M.Albert,andM.Elbaum.2003.Detectionofburning Kuhn, S. L., M. C. Stiner, E. Güleç, I. Özer, H. Yilmaz, I. Baykara, A. AçIkkol, of plant materials in the archaeological record by changes in the refractive in- et al. 2009. The early Upper Paleolithic occupations at ÜçagIzlI Cave (Hatay, dices of siliceous phytoliths. Journal of Archaeological Science 30:217–226. Turkey). Journal of Human Evolution 56:87–113. Forget, M. C. L., L. Regev, D. E. Friesem, and R. Shahack-Gross. 2015. Physical Lindbo, D. L., M. H. Stolt, and M. J. Vepraskas. 2010. Redoximorphic features. and mineralogical properties of experimentally heated chaff-tempered mud In Interpretation of micromorphological features of soils and regoliths.G. bricks: implications for reconstruction of environmental factors influencing Stoops, V. Marcelino, and F. Mees, eds. Pp. 129–147. Amsterdam: Elsevier. the appearance of mud bricks in archaeological conflagration events. Journal Lowe, K. M., Susan M. Mentzer, L. A. Wallis, and J. Shulmeister. 2016. A multi- of Archaeological Science: Reports 2:80–93. proxy study of anthropogenic sedimentation and human occupation of Gleds- Gao, Xing, Shuangquan Zhang, Yue Zhang, and Fuyou Chen. 2017. Evidence of wood Shelter 1: exploring an interior sandstone rockshelter in Northern hominin use and maintenance of fire at Zhoukoudian. Current Anthropology Australia. Archaeological and Anthropological Sciences, doi:10.1007/s12520 58(suppl. 16):S267–S277. -016-0354-8. Goldberg, Paul. 2003. Some observations on Middle and Upper Palaeolithic Macphail, R. I., and G. M. Cruise. 2001. The soil micromorphologist as team ashy cave and rockshelter deposits in the Near East. In More than meets the player: a multianalytical approach to the study of European microstra- eye: studies on Upper Palaeolithic diversity in the Near East. A. N. Goring- tigraphy. In Earth science and archaeology. Paul Goldberg, V. Holliday, and Morris and A. Belfer-Cohen, eds. Pp. 19–32. Oxford: Oxbow. R. Ferring, eds. Pp. 241 –267. New York: Plenum. Goldberg, Paul, and O. Bar-Yosef. 1998. Site formation processes in Kebara Madella, M., M. K. Jones, Paul Goldberg, Y. Goren, and E. Hovers. 2002. Ex- and Hayonim Caves and their significance in Levantine prehistoric caves. ploitation of plant resources by Neanderthals in Amud Cave (Israel): the In Neandertals and modern humans in western Asia. T. Akazawa, K. Aoki, evidence from phytolith studies. Journal of Archaeological Science 29:703– and O. Bar-Yosef, eds. Pp. 107–125. New York: Plenum. 719. Goldberg, Paul, and Francesco Berna. 2010. Micromorphology and context. Mallol, Carolina, C. M. Hernández, D. Cabanes, A. Sistiaga, J. Machado, Á. Quaternary International 214:56–62. Rodríguez, L. Pérez, and B. Galván. 2013. The black layer of Middle Palaeo- Goldberg, Paul, Francesco Berna, and Michael Chazan. 2015. Deposition and lithic combustion structures: interpretation and archaeostratigraphic implica- diagenesis in the Earlier Stone Age of Wonderwerk Cave, Excavation 1, tions. Journal of Archaeological Science 40:2515–2537. South Africa. African Archaeological Review 32:613–643. Mallol, Carolina M., Susan M. Mentzer, and Christopher E. Miller. Forthcom- Goldberg, Paul, H. Laville, L. Meignen, and O. Bar-Yosef. 2007. Stratigraphy ing. Combustion features. In Encyclopedia of archaeological soil and sedi- and geoarchaeological history of Kebara Cave, Mount Carmel. In Kebara ment micromorphology. G. Stoops and C. Nicosia, eds. Hoboken, NJ: Wiley- Cave, Mt. Carmel, Israel: the Middle and Upper Paleolithic archaeology, pt. 1. Blackwell. O. Bar-Yosef and L. Meignen, eds. Pp. 49–89. American School of Prehistoric March, R. J. 2013. Searching for the functions of fire structures in Eynan Research Bulletin 49. Cambridge, MA: Peabody Museum of Archaeology (Mallaha) and their formation processes: a geochemical approach. In Natu- and Ethnology, Harvard University. fian foragers in the Levant. O. Bar-Yosef and F. Valla, eds. Pp. 227–283. Ann Goldberg, Paul, and R. I. Macphail. 2003. Strategies and techniques in collecting Arbor, MI: International Monographs in Prehistory. micromorphology samples. Geoarchaeology 18:571–578. March, R. J., J. C. Ferreri, and C. Guez. 1993. Étude des foyers préhistoriques ———. 2006. Practical and theoretical geoarchaeology. Oxford: Blackwell. des gisements magdaléniens du Bassin Parisien: l’approche expérimentale. Goldberg, Paul, Christopher E. Miller, S. Schiegl, Francesco Berna, B. Ligouis, Mémoires du Groupement Archéologique de Seine-et-Marne 1:87–95. N. J. Conard, and L. Wadley. 2009. Bedding, hearths, and site maintenance Matthews, W. 2005. Micromorphological and microstratigraphic traces of in the Middle Stone Age of Sibudu Cave, KwaZulu-Natal, South Africa. uses and concepts of space. In Inhabiting Çatalhöyük: reports from the 1995– Archaeological and Anthropological Sciences 1:95–122. 1999 seasons. I. Hodder, ed. Pp. 355–398. BIAA Monograph 40. Ankara: Goldberg, Paul, S. Weiner, O. Bar-Yosef, Q. Xu, and J. Liu. 2001. Site for- McDonald Institute Monographs. mation processes at Zhoukoudian, China. Journal of Human Evolution 41: Matthews, W., C. A. I. French, T. Lawrence, D. F. Cutler, and M. K. Jones. 1997. 483–530. Microstratigraphic traces of site formation processes and human activities. Gopher, Avi, Ran Barkai, R. Shimelmitz, M. Khalaily, C. Lemorini, I. Heshkovitz, World Archaeology 29:281–308. and M. Stiner. 2005. Qesem Cave: an Amudian site in Central Israel. Journal McPherron, Shannon P., and Harold Dibble. 2002. Using computers in archae- of the Israel Prehistoric Society 35:69–92. ology: a practical guide. New York: McGraw Hill. Gose, W. 2000. Paleomagnetic studies of burned rocks. Journal of Archaeo- McPherron, Shannon J. P., Harold L. Dibble, and Paul Goldberg. 2005. Z. Geo- logical Science 27:409–421. archaeology 20:243–262. Gur-Arieh, S., E. Mintz, E. Boaretto, and R. Shahack-Gross. 2013. An ethno- Meignen, L., O. Bar-Yosef, Paul Goldberg, and S. Weiner. 2001. Le feu au archaeological study of cooking installations in rural Uzbekistan: develop- Paléolithique moyen: recherches sur les structures de combustion et le ment of a new method for identification of fuel sources. Journal of Archaeo- statut des foyers: l’exemple du Proche-Orient. Paléorient 26:9–22. logical Science 40:4331–4347. Meignen, L., Paul Goldberg, and O. Bar-Yosef. 2007. The hearths at Kebara Harrold, F., and M. Otte. 2001. Time, space, and cultural process in the Eu- Cave and their role in site formation processes. In Kebara Cave, Mt. Carmel, ropean Middle-Upper Palaeolithic transition. In Questioning the answers: re- Israel: the Middle and Upper Paleolithic archaeology, pt. 1. O. Bar-Yosef and solving fundamental problems of the Early Upper Paleolithic. M. A. Hays and L. Meignen, eds. Pp. 91–122. American School of Prehistoric Research P. T. Thacker, eds. Pp. 3–12. British Archaeological Reports International Bulletin 49. Cambridge, MA: Peabody Museum of Archaeology and Eth- Series 1005. Oxford: Archaeopress. nology, Harvard University. James, S. R. 1989. Hominid use of fire in the Lower and Middle Pleistocene. Mentzer, Susan. 2013. Microarchaeological approaches to the identification Current Anthropology 30:1–26. and interpretation of combustion features in prehistoric archaeological sites. Jia, L., and W. Huang. 1990. The story of Peking Man. Beijing: Foreign Journal of Archaeological Method and Theory 21:616–668. Languages Press. Mentzer, Susan M., and J. Quade. 2013. Compositional and isotopic analytical Jia, L. P. 1980. Early man in China. Beijing: Foreign Language Press. methods in archaeological micromorphology. Geoarchaeology 28:87–97. Karkanas, P. 2010. Preservation of anthropogenic materials under different Mentzer, Susan M., D. G. Romano, and M. E. Voyatzis. 2015. Micromorpho- geochemical processes: a mineralogical approach. Quaternary International logical contributions to the study of ritual behavior at the ash altar to Zeus on 214:63–69. Mt. Lykaion, Greece. Archaeological and Anthropological Sciences, doi:10.1007 Karkanas, P., M. Koumouzelis, J. K. Kozlowski, V. Sitlivy, K. Sawbucks, Fran- /s12520-014-0219-y. cesco Berna, and S. Weiner. 2004. The earliest evidence for clay hearths: Au- Mercier, N., H. Valladas, L. Froget, J.-L. Joron, J.-L. Reyss, S. Weiner, Paul rignacian features in Closure Cave 1, southern Greece. Antiquity 78:513–525. Goldberg, L. Meignen, O. Bar-Yosef, and A. Belfer-Cohen. 2007. Hayonim Karkanas, P., J. P. Rigged, J. F. Semite, R. M. Albert, and S. Weiner. 2002. Ash Cave: a TL-based chronology for this Levantine Mousterian sequence. Jour- bones and guano: a study of the minerals and phytoliths in the sediments of nal of Archaeological Science 34:1064–1077. Goldberg, Miller, and Mentzer Recognizing Paleolithic Fire S189

Michel, V., C. Falguères, and J.-M. Dolo. 1998. ESR signal behavior study at Shahack-Gross, R., O. Bar-Yosef, and S. Weiner. 1997. Black-coloured bones g ∼ 2.002 of modern and fossil bones for heating palaeotemperature as- in Hayonim Cave, Israel: differentiating between burning and oxide staining. sessment. Radiation measurements 29:95–103. Journal of Archaeological Science 24:439–446. Miller, Christopher E. 2015. A tale of two Swabian caves: geoarchaeological Shahack-Gross, R., Francesco Berna, P. Karkanas, C. Lemorini, Avi Gopher, investigations at Hohle Fels and Geißenklösterle. Tübingen Publications in and Ran Barkai. 2014. Evidence for the repeated use of a central hearth at Prehistory. Tübingen: Kerns. Middle Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of Archaeo- Miller, Christopher E., Paul Goldberg, and Francesco Berna. 2013. Geoar- logical Science 44:12–21. chaeological investigations at Diepkloof Rock Shelter, Western Cape, South Shipman, P., G. Foster, and M. Schoeninger. 1984. Burnt bones and teeth: an Africa. Journal of Archaeological Science 40:3432–3452. experimental study of color, morphology, crystal structure and shrinkage. Movius, H. L. J. 1966. The hearths of the Upper Perigordian and Aurignacian Journal of Archaeological Science 11:307–325. horizons at the Abri Pataud, Les Eyzies (Dordogne), and their possible sig- Sistiaga, A., R. March, C. M. Hernández Gómez, and B. Galván Santos. 2011. nificance. American Anthropologist 68:296–325. Aproximación desde la química orgánica al estudio de los hogares del Oakley, K. P. 1954. Evidence of fire in South African cave deposits. Nature yacimiento del Paleolítico medio de El Salt (Alicante, España). Recerques 174:261–262. del Museu d’Alcoi 20:47–70. Pigati, J. S., J. P. McGeehin, G. L. Skipp, and D. R. Muhs. 2014. Evidence Stahlschmidt, M. C., Christopher E. Miller, B. Ligouis, U. Hambach, Paul of repeated wildfires prior to human occupation on San Nicolas Island, Goldberg, Francesco Berna, D. Richter, B. Urban, J. Serangeli, and N. J. California. Monographs of the Western North American Naturalist 7:35–47. Conard. 2015. On the evidence for human use and control of fire at Schö- Piperno, D. R. 2006. Phytoliths: a comprehensive guide for archaeologists and ningen. Journal of Human Evolution 89:181–201. paleoecologists. Lanham, MD: AltaMira. Stiner, M. C., S. L. Kuhn, T. A. Surovell, Paul Goldberg, L. Meignen, S. Weiner, Regev, L., K. M. Poduska, L. Addadi, S. Weiner, and E. Boaretto. 2010. Dis- and O. Bar-Yosef. 2001. Bone preservation in Hayonim Cave (Israel): a tinguishing between calcites formed by different mechanisms using infrared macroscopic and mineralogical study. Journal of Archaeological Science 28: spectrometry: archaeological applications. Journal of Archaeological Science 643–659. 37:3022–3029. Stiner, M. C., S. L. Kuhn, S. Weiner, and O. Bar-Yosef. 1995. Differential Renfrew, C., and P. Bahn. 2007. Archaeological essentials. New York: Thames & burning, recyrstallization, and fragmentation of archaeological bone. Jour- Hudson. nal of Archaeological Science 22:223–237. Richter, D. 2007. Advantages and limitations of thermoluminescence dating Stoops, G. 2003. Guidelines for analysis and description of soil and regolith of heated flint from Paleolithic sites. Geoarchaeology 22:671–683. thin sections. Madison, WI: Soil Science Society of America. Roebroeks, W., and P. Villa. 2011. On the earliest evidence for habitual use of Suárez-Ruiz, I. 2012. Organic petrology: an overview. In Petrology: new per- fire in Europe. Proceedings of the National Academy of Sciences of the USA spectives and applications. A. Al-Juboury, ed. Rijeka, Croatia: INTECH, doi:10 108:5209–5214. .5772/23431. https://www.intechopen.com/books/petrology-new-perspectives Rogers, K., and P. Daniels. 2002. An X-ray diffraction study of the effects of -and-applications/organic-petrology-an-overview. heat treatment on bone mineral microstructure. Biomaterials 23:2577–2585. Taylor, G. H., M. Teichmüller, D. A., Diessel, C. F. K., R. Littke, and P. Robert. Sandgathe, Dennis M. 2017. Identifying and describing pattern and process in 1998. Organic petrology. Berlin-Stuttgart: Gebrüder Bornträger. the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): Teilhard de Chardin, Pierre. 1934. Les fouilles préhistoriques de Péking. Revue S360–S370. des Questions Scientifiques 25:181–193. Schiegl, S., Paul Goldberg, O. Bar-Yosef, and S. Weiner. 1996. Ash deposits in Thieme, H. 1997. Lower Palaeolithic hunting spears from Germany. Nature Hayonim and Kebara Caves, Israel: macroscopic, microscopic and miner- 385:807–810. alogical observations, and their archaeological implications. Journal of ———. 1999. Altpaläolithische Holzgeräte aus Schöningen, Lkr. Helmstedt: Archaeological Science 23:763–781. Bedeutsame Funde zur Kulturentwicklung des frühen Menschen. Germania Schiegl, S., Paul Goldberg, H.-U. Pfretzschner, and N. J. Conard. 2003. Pa- 77:451–487. leolithic burnt bone horizons from the Swabian Jura: distinguishing be- Thompson, T. J. U., M. Islam, and M. Bonniere. 2013. A new statistical ap- tween in situ fi re places and dumping areas. Geoarchaeology 18:541–565. proach for determining the crystallinity of heat-altered bone mineral from Schiegl, S., S. Lev-Yadun, O. Bar-Yosef, A. El Goresy, and S. Weiner. 1994. FTIR spectra. Journal of Archaeological Science 40:416–422. Siliceous aggregates from prehistoric wood ash: a major component of Thoms, A. V. 2008. The fire stones carry: ethnographic records and archae- sediments in Kebara and Hayonim Caves (Israel). Israel Journal of Earth ological expectations for hot-rock cookery in western North America. Journal Sciences 43:267–278. of Anthropological Archaeology 27:443–460. Schiegl, S., and H. Thieme. 2007. Auf den Spuren von Feuer in Schöningen 13 ———. 2009. Rocks of ages: propagation of hot-rock cookery in western II-4. In Die Schöninger Speere-Menschen und Jagd vor 400,000 Jahren. North America. Journal of Archaeological Science 36:573–591. H. Thieme, ed. Pp. 167–171. Stuttgart: Theis. Toffolo, M., A. M. Maeir, J. R. Chadwick, and E. Boaretto. 2012. Character- Schmidt, P., S. Masse, G. Laurent, A. Slodczyk, E. Le Bourhis, C. Perrenoud, J. ization of contexts for radiocarbon dating: results from the early Iron Age Livage, and F. Fröhlich. 2012. Crystallographic and structural transformations at Tell es-Safi/Gath, Israel. Radiocarbon 54:371–390. of sedimentary chalcedony in flint upon heat treatment. Journal of Archae- Wadley, L., C. Sievers, M. Bamford, Paul Goldberg, Francesco Berna, and ological Science 39:135–144. Christopher Miller. 2011. Middle Stone Age bedding construction and Schmidt, P., G. Porraz, A. Slodczyk, L. Bellot-Gurlet, W. Archer, and settlement patterns at Sibudu, South Africa. Science 334:1388–1391. Christopher E. Miller. 2013. Heat treatment in the South African Middle Wattez, J. 1988. Contribution à la connaissance des foyers préhistoriques par Stone Age: temperature induced transformations of silcrete and their tech- l’étude des cendres. Bulletin de la Société Préhistorique Française 85:353– nological implications. Journal of Archaeological Science 40:3519–3531. 366. Schurr, M. R., and R. G. Hayes. 2008. Stable carbon- and nitrogen-isotope ratios Weiner, S. 2010. Microarchaeology: beyond the visible archaeological record. and electron spin resonance (ESR) g-values of charred bones: changes with New York: Cambridge University Press. heating and a critical evaluation of the utility of g-values for reconstructing Weiner, S., and O. Bar-Yosef. 1990. States of preservation of bones from pre- thermal history and original isotope ratios. Journal of Archaeological Science historic sites in the Near East: a survey. Journal of Archaeological Science 35:2017–2031. 17:187–196. Scott, A. C. 2000. The pre-Quaternary history of fire. Palaeogeography, Pa- Weiner, S., Francesco Berna, I. Cohen-Ofri, R. Shahack-Gross, R. M. Albert, laeoclimatology, Palaeoecology 164:281–329. P. Karkanas, L. Meignen, and O. Bar-Yosef. 2007. Mineral distributions in Shahack-Gross, R., and A. Ayalon. 2013. Stable carbon and oxygen isotopic Kebara Cave: diagenesis and its effect on the archaeological record. In Kebara compositions of wood ash: an experimental study with archaeological im- Cave, Mt. Carmel, Israel: the Middle and Upper Paleolithic archaeology. plications. Journal of Archaeological Science 40:570–578. O. Bar-Yosef and L. Meignen, eds. Pp. 131–146. American School of Pre- Shahack-Gross, R., A. Ayalon, Paul Goldberg, Y. Goren, B. Ofek, R. Rabino- historic Research Bulletin 49. Cambridge, MA: Peabody Museum of Archae- vich, and E. Hovers. 2008. Formation processes of cemented features in ology and Ethnology, Harvard University. karstic cave sites revealed using stable oxygen and carbon isotopic analyses: Weiner, S., and Paul Goldberg. 1990. On site Fourier transform-infrared a case study at Middle Paleolithic Amud Cave, Israel. Geoarchaeology 23: spectrometry at an archaeological excavation. Spectroscopy International 2(2): 43–62. 38–42. S190 Current Anthropology Volume 58, Supplement 16, August 2017

Weiner, S., Paul Goldberg, and O. Bar-Yosef. 1993. Bone preservation in preservation, and wood ash remnants. Israel Journal of Chemistry 35:143– Kebara Cave, Israel using on-site Fourier transform infrared spectrometry. 154. Journal of Archaeological Science 20:613–627. Weiner, S., Q. Xu, Paul Goldberg, J. Liu, and O. Bar-Yosef. 1998. Evidence for ———. 2002. Three-dimensional distribution of minerals in the sediments of the use of fire at Zhoukoudian, China. Science 281:251–253. Hayonim Cave, Israel: diagenetic processes and archaeological implications. Weymouth, J. W., and M. Mandeville. 1975. An X-ray diffraction study Journal of Archaeological Science 29:1289–1308. of heat-treated chert and its archaeological implications. Archaeometry 17: Weiner, S., S. Schiegl, and O. Bar-Yosef. 1995. Recognizing ash deposits in the 61–67. archaeological record: a mineralogical study at Kebara and Hayonim caves, Wu, R., and S. Lin. 1983. Peking Man. Scientific American 246:86–95. Israel. Acta Anthropologica Sinica 14:340–351. Xu, B., M. B. Toffolo, L. Regev, E. Boaretto, and K. M. Poduska. 2015. Structural Weiner, S., S. Schiegl, Paul Goldberg, and O. Bar-Yosef. 1995. Mineral as- differences in archaeologically relevant calcite. Analytical Methods 7:9304– semblages in Kebara and Hayonim Caves, Israel: excavation strategies, bone 9309. Current Anthropology Volume 58, Supplement 16, August 2017 S191

Experimental Approaches to Archaeological Fire Features and Their Behavioral Relevance

by Vera Aldeias

The uses and functions of fire in early human adaptations are commonly debated and at times very controversial topics. It is important to recognize under what circumstances and conditions specific fire-related traces can be pro- duced and preserved in the archaeological record. Currently, a growing body of data is emerging on the application of experimental research to the study of archaeological hearths and their residues. In this review, I draw together aspects of such available experimental data, particularly those pertaining to the sedimentary expression and components pro- duced during simple campfires. I highlight not only what one can find in ideal preservation conditions but also what type of indirect alteration proxies can be expected to survive in the archaeological record. I then discuss the implications of such data for analyzing anthropic fire features, their timing, and their meaning in terms of behavioral complexity in the use and manufacture of fire during the Paleolithic.

1. Introduction users (Aldeias et al. 2012; Dibble et al. 2017; Sandgathe et al. 2011a, 2011b). The lack of consensus on both the timing of Fire is a key behavioral and technological adaptation in human the first use of fire and its recurrent incorporation into hu- fi evolution, and today all modern societies routinely rely on re. man adaptations largely stems from the inherent difficulty in fi Early re use by the genus Homo may have been a crucial ad- identifying intentionally used, maintained, and manufactured aptation leading to behavioral traits, such as cooking, extend- hearths. There are several confounding aspects, such as natural ing activity time by providing light, warmth, protection against landscape fires affecting archaeological occupations (Bellomo predators, and as a driving mechanism for technological in- 1993; Buenger 2003; Gowlett 2017), geological processes pro- novations (Brown et al. 2009; Mazza et al. 2006). Fire use might ducing materials that may be mistaken as combustion residues also have been an important catalyst for the evolution of bio- (Stahlschmidt et al. 2015; Weiner et al. 1998), and issues re- logical and social traits (Gowlett 2006; Twomey 2013; Wrang- lating to preservation of the original fire components (Albert, ham 2006, 2017). It is therefore not surprising that decades of Bamford, and Cabanes 2006; Cabanes, Weiner, and Shahack- fi fi research have focused on the rst evidence for re in the ar- Gross 2011; Huisman et al. 2012; March 2013; Stiner et al. chaeological record (James 1989). A related aspect, and one of- 1995). We also lack criteria that would allow us to distinguish fi ten less investigated, is when re actually started to be system- the maintenance and manufacture of fire versus its harvesting atically incorporated in the adaptive tool kit of past humans. as a natural landscape resource. fi fi This differs from research dealing with rst appearance of re The archaeological visibility of fire residues greatly depends fi use and focuses on when humans became pro cient in mak- on our ability to identify and interpret combustion remains as fi ing and repeatedly employing re in their activities. Roebroeks pertaining to instances of intentional anthropic fire use. It is, and Villa (2011) suggest that early European hominins did therefore, of utmost importance to recognize under what cir- “ ” fi not habitually use re before 400 kya, whereas clear evidence cumstances and conditions specific fire-related traces can be for pyrotechnology exists in later Neanderthal contexts. Con- produced and preserved in the archaeological record. One way versely, other authors argue that although Neanderthals did to obtain such data is through archaeological experimentation. fi use re, this was not an essential part of their adaptation, and There is a growing body of data from experimental work fi European Neanderthals might not have been obligated re dealing with fire features and artifactual thermal alteration. In this paper I review some of the available experimental data in the framework of Paleolithic contexts, focusing mainly Vera Aldeias is Researcher in the Department of Human Evolution at on sedimentary components and signatures and, to a lesser ex- the Max Planck Institute for Evolutionary Anthropology (Deutscher Platz 6, 04107 Leipzig, Germany [[email protected]]) and Re- tent, on alterations of artifacts that result from association with fi fi searcher at the Interdisciplinary Center for Archaeology and Evolu- res. First I will focus on what the proxies for re are. Then I tion of Human Behavior at the Universidade do Algarve (Campus da will discuss experimental results for the types of fuel used, the Penha, 80085-139 Faro, Portugal). This paper was submitted 25 VII 16, effects of hearth location on artifact alteration, and elements accepted 20 I 17, and electronically published 16 V 17. indicating the intact or reworked nature of hearths (fig. 1). q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0004$10.00. DOI: 10.1086/691210 S192 Current Anthropology Volume 58, Supplement 16, August 2017

Finally, based on the available data, I discuss the archaeolog- are the by-products created by burning. The nature of these ical significance of burned artifacts and inferences for hearth by-products is intrinsically dependent on the type of fuel used functions. The overallgoalistopresenta comprehensive though and consists mainly of calcitic ashes, charcoals, siliceous com- not exhaustive review of the application of experimental re- ponents (phytoliths and siliceous aggregates), calcined bone search for the understanding of combustion features. fragments, or dung calcitic spherulites. A simple campfire can create diagnostic sedimentary signatures expressed as a suc- cession of discrete sedimentary lenses, typically with an up- 1.1. Fire Proxies and Their Contextual Arrangement permost ash-rich lens commonly resting on blackish (often Combustion residues can be broadly divided into direct (pri- charcoal-rich) deposits in wood-fueled fires. However, if these mary) and indirect (secondary) evidence. Direct fire residues somewhat fragile sediments are not protected or rapidly bur-

Figure 1. Schematic diagram illustrating the several phases of hearth construction, use, and preservation, with main factors influ- encing the archaeological expression of such evidence. The numbered circles refer to sections in this paper that deal with the available data from experimental archaeology. A color version of this figure is available online. Aldeias Fire Experiments S193 ied, they can then be easily displaced or chemically altered and properties of the plant matter that was used (e.g., physiological may become essentially invisible in the archaeological record. condition, anatomic section, density, moisture content, size; Another type of evidence for past fires relates to indirect Berna and Goldberg 2008; Braadbaart and Poole 2008; Braad- proxies. These are artifacts or sediments altered because of fire baart et al. 2012; Brochier 1983; Etiegni and Campbell 1991; temperatures. Depending on the temperature threshold, such March 1992; March et al. 2014; Théry-Parisot 2001; Théry- heating can result in important structural and mineralogical Parisot, Chabal, and Chrzavzez 2010; Weiner 2010). transformations (Aldeias et al. 2016a, 2016b; Chu et al. 2008; The by-products of combusted plant materials are mainly Elbaum et al. 2003; Maki, Homburg, and Brosowske 2006; charcoals and ash. Charcoals result from the carbonization and Schmidt 2013; Schmidt et al. 2013; Stiner et al. 1995; Toffolo incomplete charring of wood (Braadbaart and Poole 2008), and Boaretto 2014; Weiner et al. 2015). Therefore, thermally whereas wood ash comprises the residual inorganic fraction left altered artifacts (bones, lithics, shells, seeds, etc.) or deposits after organic material degradation. Ash is mainly composed (soils or sediment aggregates mixed with the original fuel, or of micritic (!4 µm) calcium carbonate resulting from the al- the heated substrate over which a fire is built) are relevant in- teration of cellular calcium oxalate crystals contained in plant direct proxies for past fires. Diagenesis and taphonomic pro- tissues (e.g., Brochier 1996; Canti 2003; Mentzer 2014; Pereira, cesses can equally affect burned artifacts and deposits, leading to Úbeda, and Martin 2012). Other residual components are si- their displacement and loss of primary position and contextual liceous in nature, namely phytoliths and silica aggregates, as association (fig. 1). What is important, however, is that several well as inorganic minerals from soil material originally con- heat-related transformations are irreversible even at a geological tained or attached to the used fuel source (Albert, Berna, and timescale. Therefore, diagnostic artifacts (e.g., burned lithics or Goldberg 2012; Canti 1998; Elbaum et al. 2003; Mentzer 2014; bones) can have a higher chance of surviving in the archaeo- Schiegl et al. 1994, 1996; Weiner 2010). This residual acid in- logical record when compared with the more “fragile” compo- soluble fraction was experimentally estimated to be on the nents such as wood ash. order of only 2% by volume or weight of wood ash (Schiegl et al. 1996). Moreover, the proportion of different constituents var- 2. The Evidence for Fuel ies: phytoliths, for instance, occur in greater concentrations in certain portions of a plant (e.g., leaves and bark) and in grasses The type of fuels being burned can result in distinct sedimen- or sedges than in wooden species (Albert and Cabanes 2007; tary accumulations and are an essential variable for combus- Piperno et al. 1999; Tsartsidou et al. 2007; Weiner 2010). tion parameters, such as the duration of the fire, management In thin section, ash accumulations have very distinct sedi- strategies, and associated average and maximum temperatures mentary characteristics with a calcitic crystallitic groundmass, (fig. 1; table 1). Fuel characteristics can also be relevant in- and it is sometimes possible to isolate individual or articulated dicators of paleoenvironmental aspects (e.g., the type of veg- pseudomorph crystals of calcium oxalate (Canti 2003; Mentzer etation cover surrounding a site) as well as higher-level in- 2014; Schiegl et al. 1996). While these pseudomorphs reflect ferences about human behavior in terms of selection criteria, original crystal shapes in the plant tissues, experimental re- gathering efforts and costs (Henry 2017; March 1992; Théry- search trying to link charred morphologies to the specific fuel Parisot 2002a, 2002b; Théry-Parisot, Chabal, and Chrzavzez types have been largely unsuccessful (Wattez 1996; Wattez 2010), and efficiency in relation to desired hearth functions and Courty 1987). Experiments by Simpson et al. (2003) sug- (Henry and Théry-Parisot 2014; March 1992; March and gest that some distinction between willow and birch woods Wunsch 2003; Simpson et al. 2003; Théry-Parisot 2002a, could be attained by microscopically examining the crystallitic 2002b; Théry-Parisot and Henry 2012; Villa, Bon, and Castel b-fabric of ash pseudomorphs and its clustering in relation to 2002). Substantial experimental work has been done in the embedded charcoal fragments at combustions below 4007C. domain of paleobotany and to a lesser extent with the sedi- Still, further research is needed to confirm such observations. mentary expressions of fuel-related variables. Here I will focus Relatively few experiments have targeted the burning of mainly on the latter. grasses or sedges. Miller and Sievers (2012) conducted a series of fires using these types of fuels. Their micromorphological results show the stratigraphically intermixed production of 2.1. Wood and Grasses as Fuel Source laminated fibrous charcoals and abundant phytoliths, several Wood, and to a smaller extent grasses or sedges, are commonly of which preserve a well-defined anatomical articulation. It is used as fuel sources because of their optimal pyrogenic prop- also noteworthy that by burning grasses and sedges, the over- erties. Their combustion by-products dominate the remains whelming quantity of combustion by-products were phyto- found in ancient hearths. Thermal degradation of the organic liths, with no observed calcitic ash rhombs (Miller and Sievers compounds starts at ~3007C in a chain reaction that is often 2012). There is little experimental data on the sedimentary complex (Braadbaart and Poole 2008; Pereira, Úbeda, and manifestation of mixed-plant fuel sources, namely, the com- Martin 2012). The composition of plant combustion residues bined combustion of wood and grasses within hearths. An depends on pyrogenic variables (e.g., temperature, fire duration, interesting archaeological case study, followed by experimen- oxygen availability, environmental factors, etc.) and intrinsic tal research, is described by Meignen, Goldberg, and Bar-Yosef Table 1. Fuel types and maximum temperatures recorded in experimental fires

Recorded temperature (7C) Fuel condition Fuel type Maximum Maximum mean Fire duration Fire shape/dimensions (moisture/state) Study Wood 900–1,000 NR Open flat fires NR Stiner et al. 1995 Wood 995 4 days Open flat fires NR Canti and Linford 2000 Wood 920 1 day Open flat fires NR Canti and Linford 2000 Wood 892 4 days Open flat fires NR Canti and Linford 2000 Wood 855 1 day Open flat fires NR Canti and Linford 2000 Wood 818 1 day Open flat fires NR Canti and Linford 2000 Wood 600 1–3 hours Open flat fires Bellomo 1993 Wood (sicklebush) 466, 760 3 hours 45 minutes, Open flat fires (ca. 40 cm Wheel-dried (12%–20% Bentsen 2013 5 hours diameter) moisture) Wood (sicklebush) 562, 730 4 hours 15 minutes, Open flat fires (ca. 40 cm Wheel-dried (12%–20% Bentsen 2013 5 hours diameter) moisture) Sedge (Cyperus involucratus) 1800 NR Flat horizontally laid sedges Sun-dried Miller and Sievers 2012 Grass (Themeda triandra and 225 Flat horizontally laid sedges Bellomo 1993 Imporada cylindrica) Grass (Setaria plicatilis) 157 NR Open flat fires (30 cm # Not fully dry Miller and Sievers 2012 30 cm) Bone (with some wood) 800–900 73–122 minutes Slightly lowered, cleared Mentzer 2009 surface (60 cm diameter) Bone (with some wood) 825–605 NR Several (fresh and dry) Théry-Parisot et al. 2005 Tree stump 250 NR Bellomo 1993 Peat 800 NR Air-dried for 1 month Braadbaart et al. 2012 Cow dung 800 NR Air-dried for 1 month Braadbaart et al. 2012 Cattle and sheep dung 630 Shahack-Gross et al. 2005 Note. It is not uncommon for only the maximum peak temperature to be reported, though such a peak can be very short lived. Increased research on the average amount of heat and its durability might be of interest in many questions dealing with archaeological hearths. NR p not reported. Aldeias Fire Experiments S195

(2014). This study was conducted in order to clarify the pro- terial with a loss of up to 98% of the original thickness of fuel venience of millimeter-sized reddish rounded soil aggregates during combustion. embedded in ash layers at the site of Hayonim Cave. The au- These values can give some idea of the minimum quantity thors observed similar clumps of terra rossa attached to the of material present in an archaeological site. However, while roots of grasses growing in the soils around the site, and these experimental data tends to show some degree of correlation would turn to similar reddish colorations when exposed to between amount of fuel and thickness of the final ash layer heat. Meignen, Goldberg, and Bar-Yosef (2014) concluded that (March 1992; March et al. 2014), processes such as compres- bushes or small branches were used as additional fuel sources, sion, fragmentation, and postdepositional dissolution can af- possibly attesting to expedient gathering practices near to the fect calculations of ash thickness and original fuel yield without site. a clear control for site-specific diagenetic rates in the archaeo- One important sedimentary expression of fires is the cal- logical record. culation of the expected volume of ash and charcoal produced by combustion events. Ash yield seems to be a function of two 2.2. Bone as Fuel Source main variables: the completeness of the combustion (which can be more efficiently achieved by using dry wood sources Bone can be another fuel resource. Evidence for bone-fueled and higher temperatures) and the relative amount of insoluble hearths has been proposed for several Paleolithic sites, mainly fraction per weight of the original fuel material. This was in- in the Upper Paleolithic (Beresford-Jones et al. 2010; Marquer ferred from several experimental studies. For instance, Schiegl et al. 2010; Schiegl et al. 2003; Théry-Parisot 2002a) and in et al. (1996) found that, although normally used in smaller Middle Paleolithic or Middle Stone Age contexts (Gabucio et al. quantities, the burning of bark, cones, and leaves actually 2014; see also Cain 2005; Dibble et al. 2009; Morin 2010; Yra- produces substantially more ash than the combustion of wood vedra and Uzquiano 2013). because of their higher content of siliceous aggregates and Because of their relatively high critical heat flux for ignition phytoliths. Ash yield can also vary according to fuel condition, (around 3807C; Laloy 1981), it seems that bones cannot be used namely, if the fuel source was fresh or naturally dried wood. as the sole fuel to start a fire (Théry-Parisot and Costamagno A twofold increase in ash production was obtained experi- 2005). In experimental studies, a mixture of wood and bone is mentally using naturally dried fuels as compared with green commonly used, particularly during the initial stages of com- wood (Albert and Cabanes 2007), and this difference is mainly bustion. The assessment of use of bone as fuel is based on a set due to the increased water content in green wood. Minor dif- of criteria, namely, the high percentage of calcined bone pieces, ferences in relative ash weight and volume were also reported a higher proportion of bone to charcoal ratio, and intense frag- in the case of different wood species (Schiegl et al. 1996). In mentation indexes with an abundance of bones in size ranges limited oxygen conditions and with decreased heating (e.g., smaller than 2 cm (Costamagno et al. 2005; Joly and March with temperatures 13007C, restricted ventilation, or with abrupt 2003; Marquer et al. 2010; Mentzer 2009). The size distribution interruption of burning phase due to rain), the combustion of of burned fragments seems to be more a function of intensity of organic compounds ends. This results in an increase in solid combustion as proposed by Stiner et al. (1995) than minor var- carbonaceous residues in the form of charcoal in relation to the iations due to initial bone size (Costamagno et al. 2005; Mentzer ash content (e.g., Braadbaart and Poole 2008; March 1992; 2009). March et al. 2014). However, as emphasized by Théry-Parisot, Experimental research shows differences in pyrotechnology Chabal, and Chrzavzez (2010), experimental results on the of bone fires when compared with simple wood-fueled fires, amount of charcoal produced and its underlying parameters and it has been proposed that the use of bone fuels can pro- varies greatly between different studies, making it difficult to vide certain advantages. Controlled experiments demonstrate correlate the expected amount of charcoal to more specific that the use of fresh, spongy bones will significantly increase combustion factors or wood selection criteria before burning. the duration of a combustion event (Costamagno et al. 2005; In actualistic fire experiments, the resulting ash layer tends Mentzer 2009; Théry-Parisot and Costamagno 2005; Théry- to be fairly thin, with reported maximum thickness rarely above Parisot et al. 2005), with a direct positive relationship between 2 cm (e.g., Bentsen 2012; Mallol et al. 2013a, 2013b; March the amount of bone used and the longevity of flames in a fire et al. 2014). Such experimental hearths typically reflect short- (Théry-Parisot 2002a). This increased efficiency is due to the duration events, with the combustion lasting for a couple of flammability of fatty components, indicating that complete hours and only rarely for durations of more than one day. It fresh bones rich in grease content (i.e., spongy bone sections) is indeed difficult to test experimentally the effects of hearths are substantially more effective combustibles than wood (Théry- that are continuously used over the course of several months— Parisot 2002a; Théry-Parisot and Costamagno 2005). As can be though ethnographic evidence shows that these can produce seen in table 1, there is not a substantial difference in maximum ash layers more than 20 cm thick (Mallol et al. 2007). In their ex- temperatures achieved with bone instead of wood fuels in ex- periments burning grasses and sedges, Miller and Sievers (2012) perimental fires, with high temperatures of 8007–9007C reached observed a dramatic decrease in the volume of the burned ma- in bone fires as in wood (see also Joly and March 2003). S196 Current Anthropology Volume 58, Supplement 16, August 2017

Conversely, fragmented, dry, and compact bones exhibit sues (which can include ash pseudomorphs), and residual in- quite different pyrogenic properties when they are used as the organic components (Brochier 1983; Matthews 2010; Shahack- main fuel source. Although such pieces can more easily be Gross 2011). Gur-Arieh et al. (2013) used a ratio of wood ash ignited because of their minor water content, experimentation pseudomorphs to dung spherulites to characterize the pre- indicates that these are quickly consumed, and fires fueled with dominant use of dung versus wood fuels. Experiments by fragmented dry bones tend to be of shorter duration and with Simpson et al. (2003) show the differences between the calcitic lower average temperature than those using complete fresh nature of wood ashes, the organic-rich isotropic composition of bones (Mentzer 2009; Théry-Parisot and Costamagno 2005). dung materials containing calcitic spherulites (in sheep dung), Similarly, Théry-Parisot and Costamagno (2005) demonstrate and the rubefied (reddened) material with phytoliths and di- that the use of compact bones, which are poor in fat content, atoms obtained from burning peat and turf. Braadbaart et al. will not produce combustion durations distinct from those (2012) note that peat and cow dung, in particular, result in a observed in purely wood-fueled fires. higher release of smoke than wood fires. These important though subtle differences in the condition and type of bones used before burning seem to indicate that 3. Hearth Location and Its Effect on Artifact the advantages of bone fuels are not straightforward. If we Alteration could demonstrate their exploitation by past humans, this would in turn indicate an important degree of knowledge about Thus far we have focused the discussion on fires and their bone pyrogenic properties. Unfortunately, however, combined direct combustion residues. However, because of preservation effects of combustion and taphonomy can confound the visi- of these fragile residues, it is common to rely on the concen- bility of such differences in the archaeological record (e.g., Cain tration and dispersion of noncombustible artifacts to infer the 2005; Joly and March 2003). Original size selection of complete presence of fire. In fact, archaeological interpretations of past fresh bones might not be detectable after burning. Further- fires are often solely based on indirect evidence for fire, that more, as pointed out by Costamagno et al. (2005), subsequent is, the thermally altered artifacts found in a site. It is, there- reuse of the same hearth will entail further fragmentation and fore, important to reconstruct when and how fires affect sur- calcination of bones from previous events, creating increased rounding artifacts and deposits. confounding evidence. Experiments dealing with heat-induced Archaeological visibility of past fire use is influenced by changes in bone cremation practices have also revealed that it hearth location (fig. 2), namely, if hearths were built in an oc- is challenging to infer bone condition (fleshed, green, or dry cupational site or outside it. The latter will entail little or no bones) before burning (Gonçalves et al. 2011). Finally, pres- archaeological visibility, particularly in the case of simple, non- ervation is an equally important aspect for assessing origi- structured combustion structures such as those commonly as- nal bone type selection. Stiner et al. (1995) used agitation and sociated with Paleolithic contexts. trampling experiments to demonstrate that burned bones are Several variables relate to the behavioral choice of hearth more likely to be reduced to powder by both compaction and location within a site, and these in turn affect the type of evi- trampling when compared with mildly heated or unburned dence produced. For instance, we should take into account the fragments. Calcined bones are the most susceptible to such preservation bias toward fires built inside the drip line of caves taphonomic damage (Nicholson 1992), and their macroscopic versus fires built outside, the latter being substantially more visibility in the archaeological record might, therefore, be con- exposed to taphonomical processes and increased archaeo- siderably biased. logical “invisibility.” The selection of location is also particu- larly pertinent for its association with notions of occupation duration and spatial organization of the habitat. Emergent 2.3. Animal Dung, Peat, and Turf as Fuel Sources organizational patterns in a given archaeological site—namely The use of other fossil (e.g., coal), animal (herbivore dung), activities developed around a fire—can be relevant for recon- and organo-mineral-based (e.g., peat and turf) fuel sources are structing space management and activity areas. In this sense, known mainly for later prehistoric and historic archaeological evidence for stacked hearths (i.e., distinct and vertically super- periods. Only rarely have instances of such fuel sources been imposed combustion events) has been proposed for several proposed for Paleolithic sites (e.g., Théry et al. 1996), and here Paleolithic sites, including early fire evidence at Qesem Cave I only briefly discuss experimental data comparing these types around 400 kya (Barkai et al. 2017; Karkanas et al. 2007; of fuel sources with wood. Shahack-Gross et al. 2014) and in later contexts (e.g., Aldeias Livestock dung fuels have been investigated in terms of their et al. 2012; Courty et al. 2012; Goldberg et al. 2012; Meignen, archaeological evidence and experimentally described (e.g., Goldberg, and Bar-Yosef 2007; Schiegl et al. 1996). Experi- Braadbaart et al. 2012; Canti 2003; Gur-Arieh et al. 2014; Mat- mental work on the superimposition of fire events by Mallol thews 2010; Shahack-Gross 2011; Shahack-Gross and Finkel- et al. (2013b) has noted the difficulty in discerning relighting stein 2008). The burning of dung produces microscopic sedi- events over short-term intervals even at a microscale ofanalysis. mentary components that can be readily identified by the Similarly, on the basis of macroscopic observations, Bentsen presence of calcitic dung spherulites, burned organic plant tis- (2012) reports on multiple superimposed fires not resulting in Aldeias Fire Experiments S197

Figure 2. From the behavioral choice of hearth location to the summarized effects in terms of archaeological visibility. A color version of this figure is available online. larger hearth areas when compared with single combustion onomic displacements (e.g., bioturbation) or behavioral man- events. Such experiments suggest that discernible stacked com- agement of combustion residues (e.g., heat treatment of lithics bustion features will be identified only in situations of either an or cleaning of previous residues), as will be further discussed intentional deposition of material between distinct combustion below. events or those related to a substantial amount of time elapsed between each fire event (Aldeias et al. 2012; Mallol et al. 2013b). 3.2. Vertical Proximity to Hearths A somewhat more complex issue is to reconstruct how deeply 3.1. Lateral Proximity to Hearths a fire affects underlying substrate and embedded artifacts. The degree to which a surface fire affects the surrounding Reports on maximum subsurface temperatures beneath a fire sediments and artifacts depends mainly on two variables: tem- vary greatly because of uncontrolled variables and different peratures reached during combustion and, most importantly, experimental conditions (Bennett 1999; Campbell et al. 1995; the distance to the fire feature itself. Temperatures within the March et al. 2014; Sievers and Wadley 2008; Wadley 2009; limits of a fire rise substantially during initial combustion with Werts and Jahren 2007). In our recent experimentation (Al- all types of fuels (see table 1), with a rapid spike in tempera- deias et al. 2016a), we used a set of controlled parameters to tures observed within the first few minutes of ignition. Our investigate subsurface heat transfer under a wide range of experiments (Aldeias et al. 2016a) show that temperatures drop conditions and variables that have archaeological applicability. dramatically outside the fireplace limits, and this is true for both This experimental work shows that while maximum temper- the surface immediately adjacent to it and laterally buried atures and rate of heat transfer depends on a range of factors deposits. Other experimental work has shown that, as expected, (substrate type, moisture content, porosity, fire temperature, substantial thermal alteration is visible in artifacts directly fire duration, etc.), significant thermal alteration is expected added into an active fire, while those positioned immediately only directly underneath a fire and not to its sides. The di- outside the fire’s limits remain largely unaffected (Mallol et al. ameter of this alteration is a direct function of hearth size; that 2013a, 2013b; Sergant, Crombé, and Perdaen 2006; Wadley is, larger fires will effectively alter equally larger subsurface 2009). Only small-sized artifacts, namely pot lids, were dis- deposits than smaller, restricted features. This study suggests persed farther away, being ejected distances of up to 3 m (Mallol that under a variety of experimental conditions, temperatures et al. 2013b; Sergant, Crombé, and Perdaen 2006). around 4007C and as high as 8007C can be reached at shallow These results suggest that the distribution of burned re- depths (2 cm below the surface) and that at 10 cm below a fire, mains are a function of fire area and can mimic the limits and maximum temperatures may range from 857C to 2507C (Al- extension of hearths. Therefore, dissociation between burned deias et al. 2016a). Subsurface temperatures continue to in- artifacts and direct evidence for a hearth in any archaeologi- crease well after a fire is extinct, and it is expected that longer cal context may point to a certain degree of artifact movement fire durations will further increase subsurface exposure to heat. and reworking. Causes of such dispersion can be either taph- These results are generally in accordance with previous studies S198 Current Anthropology Volume 58, Supplement 16, August 2017 reporting on subsurface temperatures (Bennett 1999; Bentsen cavations (Mallol et al. 2013a, 2013b), and special attention to 2013; Campbell et al. 1995; March et al. 2014; Sievers and the provenience of burned artifacts versus the location of the Wadley 2008; Wadley 2009; Werts and Jahren 2007). Although direct fire residues is required, particularly when using such temperatures consistently decrease with depth independently materials for chronometric dating or inferences about hearth of substrate material (Aldeias et al. 2016a; Canti and Linford function. 2000; March et al. 2014), it is interesting that Bellomo (1993) observed that burning of tree stumps might show increased 4. Evidence for In Situ versus Reworked Hearths subsurface temperatures linked to the smoldering root system. 4.1. The Effects of Hearth Shape In general, experimentally obtained subsurface temperatures suggest that artifact alteration can occur postdepositionally. There are a series of diagnostic sedimentary signatures created An important variable is the duration of the fire event. For by a simple horizontal fire, such as a hearth or a campfire. Un- instance, while in the experimental study by Stiner et al. (1995) disturbed fires tend to present a microstratigraphy with an calcination levels were never achieved in buried bones, similar uppermost ash-rich lens that may contain occasional charcoal experiments carried out by Bennett (1999) showed that longer fragments. These deposits can rest on a charcoal-rich layer (48-hour fire durations) did in fact produce black bone alter- that in turn overlies thermally altered substrates. This discrete ation at 10 cm of depth and calcined bones at shallower depths. stratigraphic arrangement in well-preserved hearths has been Bennet (1999) further hypothesized that some differentiations attested experimentally (e.g., Godino et al. 2011; Mallol et al. might be made in terms of surface alteration of bones indi- 2013a, 2013b; March, Ferreri, and Guez 1993; March et al. rectly exposed to heat (buried bones) as compared with those 2014; Miller et al. 2010; Wattez 1996). directly exposed to fire, with the former exhibiting more uni- Extensive experimental work by March and colleagues form surface color alterations associated with minimum frac- (March 1992, 1996; March, Ferreri, and Guez 1993; March turing and warping. Other studies have shown the effects of et al. 2014) has included research on different hearth shapes, fire duration in the degree of artifact alterations. For instance, namely simple, open horizontal fires and those constructed Wadley (2009) notes that red ochre might be artificially over inside “cuvette” structures (i.e., shallow basin-shaped depres- represented in archaeological contexts as the result of thermal sions). These experiments indicate that fires built inside a de- modification of yellow ochre and iron-rich rocks buried un- pression have a relatively high thermal efficiency when com- derneath a fire. In her fire experiments, no color alteration was pared with flat open hearths. Cuvette fires tend to show uniform observed in a short-lived fire event (lasting for 1.3 hours with temperatures throughout a larger surface area and, because temperatures 12507C), whereas with longer fire duration (tem- of incomplete and slower combustion, have a higher ratio of peratures 12507C for 19 hours) all of the buried materials be- charcoal production compared with open horizontal features came reddened, even those buried at 10 cm below the surface (March 1992; March et al. 2014). These differences seem to (Wadley 2009). Given that temperatures commonly reach above result from more controlled conditions of air circulation in- 1007C at up to 10 cm in depth (Aldeias et al. 2016a), experi- side shallow depressions versus surface fires. Accordingly, the mental research suggests that other artifacts—such as organic exposure to increased fluctuations in open ventilation con- materials like seeds, fruits, plant litter, grass, and rootlets—can ditions results in higher average temperatures in flat fires (see be postdepositionally altered. Similarly, experimental fires by also Canti and Linford 2000), which, in turn, demands higher Miller and Sievers (2012) resulted in the carbonization of a quantities of wood to keep a fire burning for a similar duration layer of sedges buried below 5 cm of sediments (see also Sievers event. Interestingly, March et al. (2014) also observed that the and Wadley 2008). Mallol et al. (2013a) note that in their substrates underlying cuvette fires may not show any signs of experimental fires the black layer sharply underlying ash oxidation. This evidence is interpreted as resulting from the remains is not innately linked to the fire event per se but in- rapid infilling of the depression with combustion residues, stead represents the secondary charring of the surface in which namely ashes and charcoals, acting as thermal isolators and the fireplace was built. Similarly, in our experiments we ob- topographically raising the center of combustion (March et al. served the charring of an organic-rich layer buried beneath 2014). Such a hypothesis is in agreement with data on the 2 cm of sand with the development of a thin underlying rube- thermal isolation effects of ashes reported by Canti and Lin- fied lens—a microstratigraphic superimposition that tends ford (2000) and further tested experimentally by Aldeias et al. to reflect the typical arrangement of a surface fire but that in (2016a). Finally, variations in the diameter of open fires can this case actually represents the postdepositional alteration of be broadly correlated with the amount of fuel used, as shown previously deposited sediments (Aldeias et al. 2016a). in the experimental work of Bentsen (2012). The presence and extent of postdepositional alterations due to overlying later fires are relevant because any archaeological materials embedded in secondarily altered deposits are tem- 4.2. Alteration of Underlying Substrate porally and consequently behaviorally unrelated to the fire fea- ture and its use. Accordingly, these materials and sediments A hearth built on top of sedimentary substrates will result in should be separated from the direct fire residues during ex- some degree of thermal alteration of the material directly un- Aldeias Fire Experiments S199 derlying. Thermally altered substrates are a good indication some degree of rubefi cation with the exception of heat ap- that a fire was present in this exact location. Still, the degree and plied to limestone sand and calcitic ashes (Aldeias et al. 2016a). aspect of thermal alteration varies greatly because it is intrin- In both of these cases, there were mineralogical and color sically dependent on the nature and content of the substrate changes associated with thermal alteration, but these did not itself and equally affected by pyrogenic properties, namely, fire assume the form of a rubefied substrate. In the cases where temperature and duration. Thus far, the majority of work re- reddening was observed, the main factors driving its exten- porting on fire-substrate alterations tends to concentrate on sion were sediment type and, to a lesser extent, mean tem- the direct replication of site-specific archaeological evidence perature and combustion duration. The average thickness of (e.g., Brodard et al. 2015; Mallol et al. 2013a;MillerandSievers the rubefied layer was ~6 cm, with a maximum thickness of 2012; Sherwood and Chapman 2005). A small number of ex- 8.5 cm obtained in wet quartz sands heated for a longer du- periments provide a wider range of factors that allow for in- ration of 19 hours at ~6007C. Similar to the results of pre- depth generalizations about the type of modifications expected vious researchers (Canti and Linford 2000; March et al. 2014), under certain conditions and, consequentially, what such al- when organic matter was present in the subsurface, there was terations may mean for inferring past pyrogenic conditions. a marked decrease of visible alteration compared with exam- One of the common thermally induced modifications is the ples from the same heating conditions and the same type of development of a rubefied (i.e., reddening) layer usually re- sediments but without an organic component; specifically, sed- lated to oxidation of iron-rich minerals. Experimental work by iments with organic matter showed only a 3 cm thick rubefied Canti and Linford (2000) produced varied results, with some substrate, when compared with a 6 cm thick rubefied layer using of the experimental fires showing a 2–3 cm thick rubefied base the same type of sediments but without organic matter (Aldeias whereas others showed no macroscopic alteration. These dif- et al. 2016a). ferences were not necessarily a direct function of temperature, In terms of the relationship between substrate alteration and because fires registering lower subsurface temperatures actu- hearth shape, experimentation has shown that a fire built on a ally became reddened while the hottest fires did not. These fires flat surface can produce an underlying semispherical config- were not built on top of the same type of sediments, with fires uration of altered sediments when seen in cross section (Al- on humic topsoil not presenting oxidized bases (Canti and deias et al. 2016a). This topography, readily visible in the case Linford 2000). Similar results were obtained by Bellomo (1993), of rubefication, is an outcome of the way heat transfer occurs who also did not observe any rubefication with the burning of in the subsurface and is not related to the original shape of the several tree stumps on sandy humic soils, though in this case it hearth. That flat surface fires can have a cuvette-shaped sub- is unclear whether this can also be related to the lower tem- strate is, therefore, shown through controlled experimentation, peratures reached (a mean maximum of 2507C was recorded). actualisticstudies(Bellomo1993),andheat-transfermodels(e.g., Bellomo further suggests that wood bark may act as a thermal Brodard et al. 2015; March et al. 2014). These secondarily altered insulator preventing heat transfer to the surrounding soil, a deposits should not be interpreted as intentional hearth con- hypothesis that rests largely untested in the case of natural tree struction in a depression. stump fires. In any case, these studies suggest that soil organic matter might prevent substantial rubefication. 4.3. The Role of Anthropic Actions Not all substrate thermal alteration is in the form of red- dened substrates. Other experimental fires show that substrate Besides forming combustion residues, humans can also act as organic matter has an important role on the formation of agents of erosion and reworking of anthropogenic sediments. black thermally altered zones underlying fire features. March Of particular interest is the identification of syndepositional et al. (2014) noted the formation of thick blackish zones un- (i.e., after formation and before burial) human interactions derneath fires on humic soils and volcanic or aeolian silts rich with fire residues, as such data can give us information on in organic matter. The formation of black basal layers was space use, maintenance activities, and, at times, fire functions. further investigated experimentally by Mallol et al. (2013a), For instance, in our experimental work with hearths used for suggesting that these deposits can represent the fire-altered roasting shellfish, Aldeias et al. (2016b) showed that spread- organic-rich surface and are not the result of direct combus- ing and dumping of fire residues outside the cooking area was tion residues (i.e., the common attribution of black layers to an essential step in the cooking procedures. Such actions re- charcoal-rich residues underlying an ash layer). sulted in the absence of a microstratigraphy associated with The nature and components of the substrate logically play intact hearths, which in this case would not be due to poor an important role in oxidation and associated reddening (i.e., preservation of the fire features but because removal of com- whether there are iron minerals to be oxidized in the first bustion residues was an intentional activity that directly related place). March et al. (2014) showed that oxidation dimensions to hearth use. are related to both the temperatures attained and the nature Asserting possible functions for human manipulation in- of the substrate, with reddish alteration visible between 3007C volves better characterization of the effects of actions such as and 5007C depending on soil types. In our experimental stud- trampling, scooping, sweeping, and dumping of fire residues. ies with a variety of sediment types, all sediments presented These have also been tested experimentally. Although having S200 Current Anthropology Volume 58, Supplement 16, August 2017 the lowest effects in terms of disturbing the original integrity and Cabanes 2007; Berna 2010; Berna, Matthews, and Weiner of the fire features, experiments with trampling produce the 2004; Berna et al. 2007; Braadbaart and Poole 2008; Cabanes, most diagnostic sedimentary signatures. Miller et al. (2010) Weiner, and Shahack-Gross 2011; Cohen-Ofri et al. 2006; noted that after short episodes of trampling (1 minute du- Karkanas 2010; Karkanas et al. 2000; Schiegl et al. 1996; ration), the original structure and microstratigraphic orga- Weiner 2010; Weiner, Goldberg, and Bar-Yosef 1993, 2002). nization of the hearths was still discernible in thin section, From such data, the degree of preservation and taphonomy of with the charcoal layer overlying altered substrate. Similar re- combustion-associated sediments can be indirectly assessed, sults were obtained by Mallol et al. (2013b) in trampling epi- and past geochemical conditions can be inferred. sodes of much longer duration (over the course of 21 days), In terms of the effects of bioturbation, a recent study has where the blurred limits of the original structures were visible specifically dealt with the interaction of some carnivores with macroscopically, though in thin section the uppermost ash hearth features, which can result in the reworking of the com- component was intrinsically mixed with angular charcoal frag- bustion structures (Camarós et al., forthcoming). Bears, in ments, subrounded sediment rip ups, and few rounded ash particular, were observed to interact and significantly disturb aggregates. The underlying microstratigraphy showed little fire residues used for meat cooking by rubbing themselves in modification from nontrampled substrates. Both studies sug- the ashes and charcoal, digging holes up to 50 cm in diameter, gest that trampling results only in alterations of the uppermost and displacing stones and bones away from the original hearth deposits (only a few centimeters thick) with increased sedi- location. As a result, the original experimental hearth sediments mentary compaction, microscopic granular structure, abundant and stone arrangement were no longer recognizable (Camarós in situ crushing of fragile artifacts (e.g., snapped and crushed et al., forthcoming). bones), and the incorporation of burned remains into the under- lying sediments (Mallol et al. 2013b; Miller et al. 2010). Increased sedimentary disturbance was obtained in exper- 5. Archaeological Significance iments involving sweeping and dumping of combustion resi- The archaeological significance of fires depends on being able dues. Sweeping produced the accumulation of a heterogeneous to attest their anthropogenic origin. We need of course to keep mix of deposits with a reworked ash layer embedding frag- in mind that fires are natural processes and a phenomenon ments of thermally altered substrate and a chaotic mixture of that has occurred throughout Earth’s geological history. Ulti- artifacts showing different degrees of burning (Mallol et al. mately, therefore, reconstructing instances of anthropogenic 2013b; Miller et al. 2010). This accumulation rests on top of (human-made) fire cannot rely solely on the nature of the substrates notaffectedbyheating.Scoopinganddumpingoffire artifacts per se (the aforementioned direct and indirect prox- residues performed by Miller et al. (2010) resulted in somewhat ies) but needs to take into consideration their overall contex- similar deposits, with an open chaotic structure containing tual association (Goldberg, Miller, and Mentzer 2017). “Con- burned artifacts and thermally altered sediment aggregates text” refers to the entirety of data pertaining to the internal embedded in a matrix of rounded calcitic ashes. Miller et al. distribution, orientation patterns, and external association of (2010) propose a possible distinction in terms of grain-size burned sedimentary components. In other words, it embodies distribution between sweeping and dumping, with coarser het- the internal structure of combusted residues and their asso- erogeneous deposits in dumped deposits and finer grain sort- ciation with adjacent archaeological data. Understanding the ing potentially associated with sweeping actions. In general, contextual framework is relevant for higher behavioral infer- these experimental results show that combustion accumula- ences beyond the assessment of whether a material is burned or tions resulting from sweeping and dumping do not preserve the not and is essential to tease apart whether (1) burned residues original hearth structure. Instead, the spreading of combusted are in fact related to anthropogenic activities, (2) the presence materials leads to a set of sedimentary characteristics associ- or absence of burned residues is due to human behavior, or ated with secondary ash dumps that allows for their identifi- (3) burned residues are instead associated with natural land- cation as disturbed, not in situ, accumulations of combustion scape burning episodes. It is this understanding of the broader material in the archaeological record. context of combustion remains that is indeed fundamental for well-grounded interpretations of past human pyrotechnology.

4.4. The Effects of Biogenic and Geogenic Processes 5.1. Indirect Proxies: Burned, Yes, but When? Experimentally testing the effects of biogenic and geogenic processes faces the issue of time. It is indeed difficult to test Archaeological sites are far from being pristine contexts even diagenetic processes that occur over hundreds and thousands under ideal situations of preservation. Therefore, as mentioned of years in archaeological fire features. Therefore, the over- above, the identification of traces for fire often relies on the de- whelming quantity of data we have on taphonomy and dia- tection and analysis of the burned artifacts that tend to with- genesis deals with individual combustion components and their stand the passage of time better than the more fragile com- stability under varied environmental conditions (e.g., Albert bustion residues. These are the so-called indirect proxies for Aldeias Fire Experiments S201

fires. Despite their unquestionable relevance, what the research to distinguish between discrete activities that might have oc- described above does show is that we must distinguish between curred above or around a fire. On the other hand, there is also a different types of indirect fire proxies. On one hand, there are conceptual issue. Several functions attributed to fires are dif- materials that are used, consumed, and burned at the same ficult to test for or are even untestable through archaeological time that a hearth is active. This heating is, therefore, contem- data simply because they leave no sedimentary signature. For poraneous with the fire event and potentially related to hearth instance, light and warmth are automatic outcomes of every fire function, the type of artifacts present, and the kinds of hu- regardless of whether their production was the intended ob- man activities that took place around the hearth. Moreover, jective in the first place. Tackling such functions directly might the study of these burned materials can provide a wealth of be a frustrating endeavor. However, it might be possible to information about the fire itself because different materials better discern patterns in the role of fire by comparing the ev- will respond to temperatures in distinct ways. idence between several sites and resorting to specifically tar- Again, however, not all burning results from direct contact geted experimentation to build testable research hypotheses. with a fire, as there might be a considerable elapsed interval of For example, the amount of light produced by one type of fuel time between artifact manufacture, discard, burial, and post- compared to another can be tested experimentally (e.g., does depositional exposure to heat. For example, a fire built on top bone fuel produce more luminosity than pine wood fires?). In of previous occupations has the potential to directly alter un- turn, the obtained results might drive expectations that can derlying deposits and embedded artifacts. In such cases, the be verified between different combustion features and distinct thermal alteration may significantly postdate the fire event, sites, attributing a degree of probability that, in this example, and these burned materials are neither temporally nor behav- light was a likely intended hearth function. iorally associated with the overlying hearth. This is particularly Such intersite comparative analyses (with comparable ex- pertinent in archaeological sites where a thin layer of sediments cavation and recording methodologies) are still fairly infre- often denotes several human occupations overlying each other quent in archaeological research, though substantial efforts for and represents sedimentary accumulations over hundreds or data collection and excavation standardization have been seen even thousands of years. in recent years. In order to interpret hearth function, our re- Thus, in order to distinguish between instances of contem- search questions need to be tailored to the type of data avail- poraneous burning from postdepositional alteration, we need able. Archaeological experimentation is key for assessing such to better understand how fires (both anthropic hearths and information on what to expect. As graphically illustrated in natural fires) affect the surrounding and especially the under- figure 3, putative hearth uses might be dependent on multiple lying deposits under a wide variety of conditions and substrate variables, and these need to be addressed from an interdisci- sediment types. Above all, such interpretations must be based plinary perspective—not just from the study of the burned ar- on context, for which detailed characterization of the sedimen- tifacts themselves but also the sediments embedding them and tary and microstratigraphic association of burned materials is their contextual arrangement. essential. For instance, we can expect the majority of heated artifacts embedded in ashes to be related to the active fires, 6. Final Comments and Future Directions whereas those retrieved from thermally altered substrates most fi fi probably represent older artifacts pertaining to distinct and The identi cation of anthropic re use is an important be- fi possibly markedly different types of human occupations. This havioral question both in terms of its rst appearance and the distinction is important for archaeological interpretations, stud- ies on spatial analyses, and dating of human occupations.

5.2. What Can We Say about Hearth Function? The array of behaviorally relevant functions of fires (light, warmth, cooking, protection against predators, etc.) is often stated and lies at the core of why fire technology is seen as a major technological advancement in human evolution. It is also, nonetheless, one of the most frustrating and elusive as- pects of ancient pyrotechnological research because pinpoint- ing the exact function of a particular Paleolithic combustion feature has proven to be extremely challenging. The reasons for this are varied. On one hand, there is the nature of the direct data we are dealing with; that is, the incompleteness of the ar- chaeological record even in instances where a hearth has been clearly identified. As seen throughout this review, the available Figure 3. Graphical illustration of the potentially complex inter- artifactual and sedimentary data often lack enough resolution playing of variables for hearth function determination. S202 Current Anthropology Volume 58, Supplement 16, August 2017 recurrence of its application in human adaptive strategies. De- techniques can be used to closely investigate whether a mate- tailed analyses on the content and context of fire residues can rial was burned, the degree of its alteration, and its micro- provide a wealth of information. Still, we face conceptual and contextual association. Moreover, an important development operational issues on how to identify ephemeral or patchy evi- is the close integration that archaeological-driven experimen- dence of burning, particularly in older contexts where archae- tation has had with other disciplines, for example, borrowing ological visibility and association might be an issue. Equally analytical techniques from the earth sciences, chemistry, and challenging is the archaeological identification of fire produc- biology. A growing line of research applies microcontextual tion, that is, being able to start a fire, versus simply its manage- approaches instead of averaging out the sedimentary debris ment. Experimental data described above show the sedimen- from fire features. Such innovative methodologies are promis- tary invisibility of several relevant aspects, namely relighting ing research avenues for the analyses of particular sedimen- episodes. This suggests that thus far, we cannot distinguish be- tary components that until recently had remained virtually tween fires that were maintained continuously lit from those invisible. revived in short intervals of time. Such potential lack of sedi- mentary evidence has consequences for questions relating to the capacity of past humans (e.g., western European Neander- Acknowledgments thals) to manufacture fire (Roebroeks and Villa 2011; Sandgathe This research was supported by the Max Planck Society, and I et al. 2011a, 2011b). Consequently, questions dealing with would like to acknowledge the support provided by Jean- the capability of humans to start a fire versus their choice to Jacques Hublin and Shannon McPherron. I would like to thank use or not use fire resources might go undetectable in the ar- Paul Goldberg and Harold L. Dibble for their comments on an chaeological record. earlier version of this manuscript. The commentaries of two How much fire was present in a given site and what those anonymous reviewers greatly helped to improve this manu- combustion events suggest in terms of human selection and script. A special thanks to Dennis Sandgathe, Francesco Berna, choices (e.g., fuels, hearth construction, and syndepositional and the Wenner-Gren Foundation for the invitation to and the maintenance activities) are both questions that profit from a organization of the symposium “Fire and the Genus Homo.” solid experimental background in order to interpret the com- plexity of the archaeological evidence. The majority of ar- chaeological experimentation tends to focus on what can hap- References Cited pen but not as much on identifying the circumstances under which it does—or did—happen. This distinction is crucial. Albert, R. M., M. K. Bamford, and D. Cabanes. 2006. Taphonomy of phytoliths fi and macroplants in different soils from Olduvai Gorge (Tanzania) and the Many of the data currently available on re experimentation application to Plio-Pleistocene palaeoanthropological samples. Quaternary deal with site-specific objectives, that is, the replication of what International 148(1):78–94. is seen in a particular archaeological layer. Such actualistic ex- Albert, R. M., F. Berna, and P. Goldberg. 2012. Insights on Neanderthal fire use at Kebara Cave (Israel) through high resolution study of prehistoric com- periments (also called realistic or replicative experiments) are bustion features: evidence from phytoliths and thin sections. Quaternary driven by the goal of seeing whether a certain type of evidence International 247(1):278–293. can be manufactured or produced or replicated. While these Albert, R. M., and D. Cabanes. 2007. Fire in prehistory: an experimental ap- proach to combustion processes and phytolith remains. Israel Journal of Earth are indeed important research questions, they tend to bear Sciences 56(2–4):175–189. little applicability for other archaeological settings and con- Aldeias, V., P. Goldberg, D. Sandgathe, F. Berna, H. L. Dibble, S. P. McPherron, texts. For instance, many experiments use sieved sediments A. Turq, and Z. Rezek. 2012. Evidence for Neandertal use of fire at Roc de Marsal (France). Journal of Archaeological Science 39(7):2414–2423. from a particular site to conduct their experimentation. While Aldeias, Vera, Harold L. Dibble, Dennis Sandgathe, Paul Goldberg, and Shan- there might be issues on how analogous such sediments truly non J. P. McPherron. 2016a. How heat alters underlying deposits and im- are to the actual stratigraphic layer, what is also important is plications for archaeological fire features: a controlled experiment. Journal of Archaeological Science 67:64–79. that the obtained results relate to that site alone, entailing lit- Aldeias, Vera, Shira Gur-Arieh, Raquel Maria, Patricia Monteiro, and Pedro tle applicability in other contexts. Moreover, many actualistic Cura. 2016b. Shell we cook it? an experimental approach to the microar- experiments produce confounding results, and we are left with chaeological record of shellfish roasting. Archaeological and Anthropological Sciences, doi:10.1007/s12520-016-0413-1. contradictory, nonreplicable data that are at most anecdotal Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a reason: evidence or are difficult to interpret. This is mainly due to the barbecue at Middle Pleistocene Qesem Cave, Israel. Current Anthropology lack of control on certain variables during experimentation. 58(suppl. 16):S314–S328. Bellomo, R. V. 1993. A methodological approach for identifying archaeo- Fortunately, in recent years, we have increasingly seen re- logical evidence of fire resulting from human activities. Journal of Archaeo- search applying more tightly controlled experimentation. By logical Science 20(5):525–553. limiting the number of variables, both in laboratory and field Bennett, J. L. 1999. Thermal alteration of buried bone. Journal of Archaeological Science 26(1):1–8. conditions, we have progressively gained a clearer picture of Bentsen, Silje Evjenth. 2012. Size matters: preliminary results from an experi- relevant factors in the formation and integrity of fire resi- mental approach to interpret Middle Stone Age hearths. Quaternary Inter- national 270:95–102. dues. For a long time, any blackened or reddened sediments ——— — — . 2013. Controlling the heat: an experimental approach to Middle or artifacts were commonly and often wrongly interpreted Stone Age pyrotechnology. South African Archaeological Bulletin 68(198): as hearths, whereas currently we understand that an array of 137–145. Aldeias Fire Experiments S203

Beresford-Jones, D. G., K. Johnson, A. G. Pullen, A. J. E. Pryor, J. Svoboda, archaeological applications. In Biosphere to lithosphere: new studies in and M. K. Jones. 2010. Burning wood or burning bone? a reconsideration of vertebrate taphonomy.T.O’Connor, ed. Pp. 51–62. London: Oxbow. flotation evidence from Upper Palaeolithic (Gravettian) sites in the Mora- Courty, M. A., E. Carbonell, J. Vallverdú Poch, and R. Banerjee. 2012. Micro- vian Corridor. Journal of Archaeological Science 37(11):2799–2811. stratigraphic and multi-analytical evidence for advanced Neanderthal pyro- Berna, F. 2010. Bone alteration and diagenesis. In Scientific methods and technology at Abric Romaní (Capellades, Spain). Quaternary International cultural heritage: an introduction to the application of materials science to 247(1):294–312. archaeometry and conservation science. G. Artioli, ed. Pp. 364–367. Oxford: Dibble, Harold L., Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Shan- Oxford University Press. non P. McPherron, and Dennis M. Sandgathe. 2017. How did hominins Berna, F., and P. Goldberg. 2008. Assessing Paleolithic pyrotechnology and adapt to Ice Age Europe without fire? Current Anthropology 58(suppl. 16): associated hominin behavior in Israel. Israel Journal of Earth Sciences 56: S278–S287. 107–121. Dibble, Harold L., Francesco Berna, Paul Goldberg, Shannon P. McPherron, Berna, F., A. Matthews, and S. Weiner. 2004. Solubilities of bone mineral from Susan Mentzer, Laura Niven, Daniel Richter, Dennis Sandgathe, Isabelle archaeological sites: the recrystallization window. Journal of Archaeological Théry-Parisot, and Alan Turq. 2009. A preliminary report on Pech de l’Azé Science 31(7):867–882. IV, Layer 8 (Middle Paleolithic, France). PaleoAnthropology 2009:182–219, Berna, Francesco, Adi Behar, Ruth Shahack-Gross, John Berg, Elisabetta doi:10.4207/PA.2009.ART30. Boaretto, Ayelet Gilboa, Ilan Sharon, et al. 2007. Sediments exposed to high Elbaum, Rivka, Steve Weiner, Rosa M. Albert, and Michael Elbaum. 2003. temperatures: reconstructing pyrotechnological processes in Late Bronze Detection of burning of plant materials in the archaeological record by and Iron Age strata at Tel Dor (Israel). Journal of Archaeological Science 34 changes in the refractive indices of siliceous phytoliths. Journal of Archae- (3):358–373. ological Science 30(2):217–226. Braadbaart, Freek, and Imogen Poole. 2008. Morphological, chemical and Etiegni, L., and A. G. Campbell. 1991. Physical and chemical characteristics of physical changes during charcoalification of wood and its relevance to ar- wood ash. Bioresource Technology 37(2):173–178. chaeological contexts. Journal of Archaeological Science 35(9):2434–2445. Gabucio, Maria Joana, Isabel Cáceres, Jordi Rosell, Palmira Saladié, and Josep Braadbaart, Freek, Imogen Poole, Hans D. J. Huisman, and Bertil van Os. Vallverdú. 2014. From small bone fragments to Neanderthal activity areas: 2012. Fuel, fire and heat: an experimental approach to highlight the potential the case of Level O of the Abric Romaní (Capellades, Barcelona, Spain). of studying ash and char remains from archaeological contexts. Journal of Quaternary International 330:36–51. Archaeological Science 39(4):836–847. Godino, Ivan Briz, Myrian Álvarez, Andrea Balbo, Débora Zurro, Marco Brochier, Jacques E. 1983. Combustion et parcage des herbivores domes- Madella, Ximena Villagrán, and Charles French. 2011. Towards high- tiques: le point de vue du sédimentologue. Bulletin de la Société Préhistorique resolution shell midden archaeology: experimental and ethnoarchaeology Française: Comptes Rendus des Séances Mensuelles Paris 80(5):143–145. in Tierra del Fuego (Argentina). Quaternary International 239:125–134. ———. 1996. Feuilles ou fumiers? observations sur le rôle des poussières Goldberg, P., H. Dibble, F. Berna, D. Sandgathe, S. J. P. McPherron, and A. sphérolitiques dans l’interprétation des dépôts archéologiques holocènes. Turq. 2012. New evidence on Neandertal use of fire: examples from Roc de Anthropozoologica 24:19–30. Marsal and Pech de l’Azé IV. Quaternary International 247:325–340. Brodard, Aurélie, Delphine Lacanette-Puyo, Pierre Guibert, François Lévêque, Goldberg, Paul, Christopher E. Miller, and Susan M. Mentzer. 2017. Recog- Albane Burens, and Laurent Carozza. 2015. A new process of reconstructing nizing fire in the Paleolithic archaeological record. Current Anthropology archaeological fires from their impact on sediment: a coupled experimental 58(suppl. 16):S175–S190. and numerical approach based on the case study of hearths from the cave of Gonçalves, D., T. J. U. Thompson, and E. Cunha. 2011. Implications of heat- Les Fraux (Dordogne, France). Archaeological and Anthropological Sciences, induced changes in bone on the interpretation of funerary behaviour and doi:10.1007/s12520-015-0250-7. practice. Journal of Archaeological Science 38(6):1308–1313. Brown, Kyle S., Curtis W. Marean, Andy I. R. Herries, Zenobia Jacobs, Chantal Gowlett, J. A. J. 2006. The early settlement of northern Europe: fire history in Tribolo, David Braun, David L. Roberts, Michael C. Meyer, and Jocelyn the context of climate change and the social brain. Comtes Rendu Palevol 5: Bernatchez. 2009. Fire as an engineering tool of early modern humans. 299–310. Science 325(5942):859–862. Gowlett, J. A. J., J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina. 2017. Buenger, Brent A. 2003. The impact of wildland and prescribed fire on ar- Evidence of burning from bushfires in southern and east Africa and its rel- chaeological resources. PhD dissertation, University of Kansas, Lawrence. evance to hominin evolution. Current Anthropology 58(suppl. 16):S206– Cabanes, D., S. Weiner, and R. Shahack-Gross. 2011. Stability of phytoliths in S216. the archaeological record: a dissolution study of modern and fossil phyto- Gur-Arieh, S., E. Mintz, E. Boaretto, and R. Shahack-Gross. 2013. An ethno- liths. Journal of Archaeological Science 38(9):2480–2490. archaeological study of cooking installations in rural Uzbekistan: develop- Cain, Chester R. 2005. Using burned animal bone to look at Middle Stone ment of a new method for identification of fuel sources. Journal of Archae- Age occupation and behavior. Journal of Archaeological Science 32(6):873– ological Science 40(12):4331–4347. 884. Gur-Arieh, Shira, Ruth Shahack-Gross, Aren M. Maeir, Gunnar Lehmann, Camarós, E., M. Cueto, L. Teira, Susanne C. Münzel, F. Plassard, P. Arias, and Louise A. Hitchcock, and Elisabetta Boaretto. 2014. The taphonomy and F. Rivals. Forthcoming. Bears in the scene: Pleistocene complex interactions preservation of wood and dung ashes found in archaeological cooking in- with implications concerning the study of Neanderthal behavior. Quater- stallations: case studies from Iron Age Israel. Journal of Archaeological Sci- nary International. http://dx.doi.org/10.1016/j.quaint.2015.11.027. ence 46:50–67. Campbell, G. S., J. D. Jungbauer Jr., K. L. Bristow, and R. D. Hungerford. 1995. Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current Soil temperature and water content beneath a surface fire. Soil Science 159 Anthropology 58(suppl. 16):S329–S336. (6):363–374. Henry, Auréade, and Isabelle Théry-Parisot. 2014. From Evenk campfires to Canti, M. 1998. Origin of calcium carbonate granules found in buried soils prehistoric hearths: charcoal analysis as a tool for identifying the use of rot- and Quaternary deposits. Boreas 27(4):275–288. ten wood as fuel. Journal of Archaeological Science 52:321–336. ———. 2003. Aspects of the chemical and microscopic characteristics of plant Huisman, D. J., F. Braadbaart, I. M. van Wijk, and B. J. H. van Os. 2012. ashes found in archaeological soils. Catena 54(3):339–361. Ashes to ashes, charcoal to dust: micromorphological evidence for ash- Canti, M. G., and N. Linford. 2000. The effects of fire on archaeological soils induced disintegration of charcoal in Early Neolithic (LBK) soil features in and sediments: temperature and colour relationships. Proceedings of the Elsloo (The Netherlands). Journal of Archaeological Science 39(4):994–1004. Prehistoric Society 66:385–395. James, Steven, R. W. Dennell, Allan S. Gilbert, Henry T. Lewis, J. A. J. Gowlett, Chu, Vikki, Lior Regev, Steve Weiner, and Elisabetta Boaretto. 2008. Differ- Thomas F. Lynch, W. C. McGrew, et al. 1989. Hominid use of fire in the entiating between anthropogenic calcite in plaster, ash and natural calcite Lower and Middle Pleistocene. Current Anthropology 30(1):1–26. using infrared spectroscopy: implications in archaeology. Journal of Archae- Joly, D., and R. March. 2003. Étude des ossements brûlés: nouvelles méthodes ological Science 35(4):905–911. pour la détermination des températures. In Le feu domestique et ses structures Cohen-Ofri, I., L. Weiner, E. Boaretto, G. Mintz, and S. Weiner. 2006. Modern au néolithique et aux âges des métaux: actes du colloque de Bourg-en-Bresse and fossil charcoal: aspects of structure and diagenesis. Journal of Archaeo- et Beaune, 7 et 8 octobre 2000. Montagnac: M. Mergoil. logical Science 33(3):428–439. Karkanas, P. 2010. Preservation of anthropogenic materials under different Costamagno, S., I. Théry-Parisot, J. P. Brugal, and R. Guilbert. 2005. Taph- geochemical processes: a mineralogical approach. Quaternary International onomic consequences of the use of bones as fuel: experimental data and 214:63–69. S204 Current Anthropology Volume 58, Supplement 16, August 2017

Karkanas, P., O. Bar-Yosef, P. Goldberg, and S. Weiner. 2000. Diagenesis in Sandrine Costamagno, and Auréade Henry, eds. Pp. 53–64. Oxford: Ar- prehistoric caves: the use of minerals that form in situ to assess the com- chaeopress. pleteness of the archaeological record. Journal of Archaeological Science 27 ———. 2014. Microarchaeological approaches to the identification and inter- (10):915–929. pretation of combustion features in prehistoric archaeological sites. Journal Karkanas, P., R. Shahack-Gross, A. Ayalon, M. Bar-Matthews, R. Barkai, A. of Archaeological Method and Theory 21(3):616–668. Frumkin, A. Gopher, and M. C. Stiner. 2007. Evidence for habitual use of Miller, Christopher E., N. Conard, P. Goldberg, and F. Berna. 2010. Dump- fire at the end of the Lower Paleolithic: site-formation processes at Qesem ing, sweeping and trampling: experimental micromorphological analysis Cave, Israel. Journal of Human Evolution 53(2):197–212. of anthropogenically modified combustion features. Palethnology 2:25–37. Laloy, J. 1981. Recherche d’une méthode pour l’exploitation des témoins de Miller, Christopher E., and Christine Sievers. 2012. An experimental micro- combustion. Paris: Centre de Recherches Préhistoriques, Cahiers. morphological investigation of bedding construction in the Middle Stone Maki, David, Jeffrey A. Homburg, and Scott D. Brosowske. 2006. Thermally Age of Sibudu, South Africa. Journal of Archaeological Science 39(10):3039– activated mineralogical transformations in archaeological hearths: inver- 3051. sion from maghemite gFe2O4 phase to haematite aFe2O4 form. Archae- Morin, Eugène. 2010. Taphonomic implications of the use of bone as fuel. ological Prospection 13(3):207–227. Palethnology 2:209–217. Mallol, C., C. M. Hernández, D. Cabanes, A. Sistiaga, J. Machado, T. Rodrí- Nicholson, Rebecca A. 1992. Bone survival: the effects of sedimentary abra- guez, L. Pérez, and B. Galván. 2013a. The black layer of Middle Palaeolithic sion and trampling on fresh and cooked bone. International Journal of combustion structures: interpretation and archaeostratigraphic implications. Osteoarchaeology 2(1):79–90. Journal of Archaeological Science 40(5):2515–2537. Pereira, Paulo, Xavier Úbeda, and Deborah A. Martin. 2012. Fire severity effects Mallol, C., F. W. Marlowe, B. M. Wood, and C. C. Porter. 2007. Earth, wind, on ash chemical composition and water-extractable elements. Geoderma and fire: ethnoarchaeological signals of Hadza fires. Journal of Archaeo- 191:105–114. logical Science 34(12):2035–2052. Piperno, D. R., D. M. Pearsall, R. A. Benfer Jr., L. Kealhofer, Z. Zhao, and Mallol, Carolina, Cristo M. Hernández, Dan Cabanes, Jorge Machado, Ainara Q. Jiang. 1999. Phytolith morphology. Science 283(5406):1265–1266. Sistiaga, Leopoldo Pérez, and Bertila Galván. 2013b. Human actions per- Roebroeks, W., and P. Villa. 2011. On the earliest evidence for habitual use of formed on simple combustion structures: an experimental approach to the fire in Europe. Proceedings of the National Academy of Sciences of the USA study of Middle Palaeolithic fire. Quaternary International 315:3–15. 108(13):5209–5214. March, R. J., J. C. Ferreri, and C. Guez. 1993. Étude des foyers préhisto- Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. Mc- riques des gisements Magdaléniens du Bassin parisien: l’approche expéri- Pherron, Alain Turq, L. Niven, and J. Hodgkins. 2011a. Timing of the ap- mentale. Mémoires du Groupement Archéologique de Seine-et-Marne (1):87– pearance of habitual fire use. Proceedings of the National Academy of Sci- 95. ences of the USA 108(29), doi:10.1073/pnas.1106759108. March, R. J., and G. Wunsch. 2003. Loupes et lentilles obscures: à propos de la ———. 2011b. On the role of fire in Neandertal adaptations in western Europe: fonction des structures de combustion. In Le feu domestique et ses struc- evidence from Pech de l’Azé and Roc de Marsal, France. PaleoAnthropology tures au néolithique et aux âges des métaux: actes du colloque de Bourg-en- 2011:216–242. Bresse et Beaune, 7 et 8 octobre 2000. Montagnac: M. Mergoil. Schiegl, S., P. Goldberg, O. Bar-Yosef, and S. Weiner. 1996. Ash deposits in March, Ramiro J. 1992. L’utilisation du bois dans les foyers préhistoriques: une Hayonim and Kebara caves, Israel: macroscopic, microscopic and minera- approche expérimentale. Bulletin de la Societé Botanique Française: Actua- logical observations, and their archaeological implications. Journal of Ar- lités Botaniques 139(2–4):245–253, doi:10.1080/01811789.1992.10827103. chaeological Science 23(5):763–781. ———. 2013. Searching for the functions of fi re structures in Eynan (Mallaha) Schiegl, S., P. Goldberg, H. U. Pfretzschner, and N. J. Conard. 2003. Paleolithic and their formation processes: a geochemical approach. In Natufian foragers burnt bone horizons from the Swabian Jura: distinguishing between in situ in the Levant: terminal Pleistocene social changes in western Asia. Ofer Bar- fireplaces and dumping areas. Geoarchaeology 18(5):541–565. Yosef and François R. Valla, eds. Ann Arbor, MI: International Monographs Schiegl, Solveig, Simcha Lev-Yadun, Ofer Bar-Yosef, Ahmed El Goresy, and in Prehistory. Stephen Weiner. 1994. Siliceous aggregates from prehistoric wood ash: a March, Ramiro Javier. 1996. L’Etude des structures de combustion préhisto- major component of sediments in Kebara and Hayonim caves (Israel). Israel riques: une approche interdisciplinaire. In XIII International Congress of Journal of Earth Sciences 43(3/4):267–278. Prehistoric and Protohistoric Sciences. O. Bar Yosef, L. Cavalli-Sforza, R. J. Schmidt, Patrick. 2013. Le traitement thermique des matières lithiques: que March, and D. R. Piperno, ed. Forli-Italia-8/14, pp. 251–275. se passe-t-il lors de la chauffe?. BAR International Series 2470. Oxford: March, Ramiro Javier, Alexandre Lucquin, Delphine Joly, Juan Carlos Ferreri, Archaeopress. and Mohamad Muhieddine. 2014. Processes of formation and alteration Schmidt, Patrick, Guillaume Porraz, Aneta Slodczyk, Ludovic Bellot-Gurlet, of archaeological fire structures: complexity viewed in the light of experi- William Archer, and Christopher E. Miller. 2013. Heat treatment in the mental approaches. Journal of Archaeological Method and Theory 21(1):1– South African Middle Stone Age: temperature induced transformations of 45. silcrete and their technological implications. Journal of Archaeological Sci- Marquer, L., T. Otto, R. Nespoulet, and L. Chiotti. 2010. A new approach to ence 40(9):3519–3531. study the fuel used in hearths by hunter-gatherers at the Upper Palaeolithic Sergant, J., P. Crombé, and Y. Perdaen. 2006. The “invisible” hearths: a con- site of Abri Pataud (Dordogne, France). Journal of Archaeological Science tribution to the discernment of Mesolithic non-structured surface hearths. 37(11):2735–2746. Journal of Archaeological Science 33(7):999–1007. Matthews, W. 2010. Geoarchaeology and taphonomy of plant remains and Shahack-Gross, R., F. Berna, P. Karkanas, C. Lemorini, A. Gopher, and R. microarchaeological residues in early urban environments in the Ancient Barkai. 2014. Evidence for the repeated use of a central hearth at Middle Near East. Quaternary International 214(1/2):98–113. Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of Archaeological Sci- Mazza, Paul Peter Anthony, Fabio Martini, Benedetto Sala, Maurizio Magi, ence 44:12–21. Maria Perla Colombini, Gianna Giachi, Francesco Landucci, Cristina Le- Shahack-Gross, R., and I. Finkelstein. 2008. Subsistence practices in an arid morini, Francesca Modugno, and Erika Ribechini. 2006. A new Palaeolithic environment: a geoarchaeological investigation in an Iron Age site, the Negev discovery: tar-hafted stone tools in a European Mid-Pleistocene bone-bearing Highlands, Israel. Journal of Archaeological Science 35(4):965–982. bed. Journal of Archaeological Science 33(9):1310–1318. Shahack-Gross, Ruth. 2011. Herbivorous livestock dung: formation, taphon- Meignen, L., P. Goldberg, and O. Bar-Yosef. 2007. The hearths at Kebara Cave omy, methods for identification, and archaeological significance. Journal of and their role in site formation processes. In Kebara Cave Mt. Carmel, Archaeological Science 38(2):205–218. Israel: the middle and upper Paleolithic archaeology, pt. 1. O. Bar-Yosef and Sherwood, Sarah C., and Jefferson Chapman. 2005. The identification and po- L. Meignen, eds. Pp. 91–122. Cambridge, MA: Peabody Museum of Ar- tential significance of Early Holocene prepared clay surfaces: examples from chaeology and Ethnology, Harvard University. Dust Cave and Icehouse Bottom. Southeastern Archaeology 24(1):70–82. Meignen, Liliane, Paul Goldberg, and Ofer Bar-Yosef. 2014. Together in the Sievers, C., and L. Wadley. 2008. Going underground: experimental carbon- field: interdisciplinary work in Kebara and Hayonim caves (Israel). Ar- ization of fruiting structures under hearths. Journal of Archaeological Sci- chaeological and Anthropological Sciences, doi:10.1007/s12520-014-0185-4. ence 35(11):2909–2917. Mentzer, Susan M. 2009. Bone as a fuel source: the effects of initial fragment Simpson, Ian A., Orri Vésteinsson, W. Paul Adderley, and Thomas H. Mc- size distribution. In Gestion des combustibles au Paleolithique et au Meso- Govern. 2003. Fuel resource utilisation in landscapes of settlement. Journal lithique: nouveaux outiles, nouvelles interpretations. Isabelle Théry-Parisot, of Archaeological Science 30(11):1401–1420. Aldeias Fire Experiments S205

Stahlschmidt, Mareike C., Christopher E. Miller, Bertrand Ligouis, Ulrich weaknesses evaluated based on a quantitative modern reference collection Hambach, Paul Goldberg, Francesco Berna, Daniel Richter, Brigitte Urban, from Greece. Journal of Archaeological Science 34(8):1262–1275. Jordi Serangeli, and Nicholas J. Conard. 2015. On the evidence for human Twomey, T. 2013. The cognitive implications of controlled fire use by early use and control of fire at Schöningen. Journal of Human Evolution 89:181– humans. Cambridge Archaeological Journal 23(1):113–128. 201. Villa, Paola, Francois Bon, and J. C. Castel. 2002. Fuel, fire and fireplaces in Stiner, M. C., S. L. Kuhn, S. Weiner, and O. Bar-Yosef. 1995. Differential the Palaeolithic of western Europe. Review of Archaeology 23(1):33–42. burning, recrystallization, and fragmentation of archaeological bone. Journal Wadley, L. 2009. Post-depositional heating may cause over-representation of of Archaeological Science 22(2):223–237. red-coloured ochre in stone age sites. South African Archaeological Bulletin Théry, I., J. Gril, J. L. Vernet, L. Meignen, and J. Maury. 1996. Coal used for 64(190):166–171. fuel at two prehistoric sites in southern France: Les Canalettes (Mousterian) Wattez, J., and M. A. Courty. 1987. Morphology of ash of some plant materials. and Les Usclades (Mesolithic). Journal of Archaeological Science 23(4):509– Soil Micromorphology. N. Fedoroff, L. M. Bresson, and M. A. Courty, eds. 512. Pp. 677–683. Plasir: Association française pour l’étude du sol. Théry-Parisot, I., L. Chabal, and J. Chrzavzez. 2010. Anthracology and ta- Wattez, Julia. 1996. Modes de formation des structures de combustión: ap- phonomy, from wood gathering to charcoal analysis: a review of the taph- proche méthodologique et implications archéologique. In The Lower and onomic processes modifying charcoal assemblages, in archaeological con- Middle Palaeolithic. Ofer Bar-Yosef, ed. Forlì: A.B.A.C.O. texts. Palaeogeography, Palaeoclimatology, Palaeoecology 291(1/2):142–153. Weiner, S. 2010. Microarchaeology beyond the visible archaeological record. Théry-Parisot, I., and A. Henry. 2012. Seasoned or green? radial cracks analysis Cambridge: Cambridge University Press. as a method for identifying the use of green wood as fuel in archaeological Weiner, S., P. Goldberg, and O. Bar-Yosef. 1993. Bone Preservation in Kebara charcoal. Journal of Archaeological Science 39(2):381–388. Cave, Israel using on-site Fourier transform infrared spectrometry. Journal Théry-Parisot, Isabelle. 2001. Economie des combustibles au paléolithique: of Archaeological Science 20(6):613–627. expérimentation, taphonomie, anthracologie. Paris: CNRS. ———. 2002. Three-dimensional distribution of minerals in the sediments of ———. 2002a. Fuel management (bone and wood) during the lower Auri- Hayonim Cave, Israel: diagenetic processes and archaeological implications. gnacian in the Pataud rock shelter (Lower Palaeolithic, Les Eyzies de Tayac, Journal of Archaeological Science 29(11):1289–1308. Dordogne, France): contribution of experimentation. Journal of Archaeo- Weiner, S., Q. Xu, P. Goldberg, J. Liu, and O. Bar-Yosef. 1998. Evidence for logical Science 29(12):1415–1421. the use of fire at Zhoukoudian, China. Science 281(5374):251–253. ———. 2002b. Gathering of firewood during the Palaeolithic. Bar Interna- Weiner, Steve, Vlad Brumfeld, Ofer Marder, and Omry Barzilai. 2015. Heating tional Series 1063:243–250. of flint debitage from Upper Palaeolithic contexts at Manot Cave, Israel: Théry-Parisot, Isabelle, and Sandrine Costamagno. 2005. Propriétés com- changes in atomic organization due to heating using infrared spectroscopy. bustibles des ossements. Gallia Préhistoire 47:235–254. Journal of Archaeological Science 54:45–53. Théry-Parisot, Isabelle, Sandrine Costamagno, Jean-Philip Brugal, Philippe Werts, S. P., and A. H. Jahren. 2007. Estimation of temperatures beneath ar- Fosse, and Raphaële Guilbert. 2005. The use of bone as fuel during the chaeological campfires using carbon stable isotope composition of soil or- Palaeolithic: experimental study of bone combustible properties. In The ganic matter. Journal of Archaeological Science 34(6):850–857. zooarchaeology of fats, oils, milk, and dairying. J. Mulville and A. Outram, Wrangham, Richard W. 2006. The cooking enigma. In Evolution of the hu- eds. Oxford: Oxbow. man diet: the known, the unknown, and the unknowable. Peter S. Ungar, ed. Toffolo, Michael B., and Elisabetta Boaretto. 2014. Nucleation of aragonite Pp. 308–323. Oxford: Oxford University Press. upon carbonation of calcium oxide and calcium hydroxide at ambient ———. 2017. Control of fire in the Paleolithic: evaluating the cooking hy- temperatures and pressures: a new indicator of fire-related human activi- pothesis. Current Anthropology 58(suppl. 16):S303–S313. ties. Journal of Archaeological Science 49:237–248. Yravedra, J., and P. Uzquiano. 2013. Burnt bone assemblages from El Esquilleu Tsartsidou, G., S. Lev-Yadun, R. M. Albert, A. Miller-Rosen, N. Efstratiou, Cave (Cantabria, Northern Spain): deliberate use for fuel or systematic dis- and S. Weiner. 2007. The phytolith archaeological record: strengths and posal of organic waste? Quaternary Science Reviews 68:175–190. S206 Current Anthropology Volume 58, Number S16, August 2017

Evidence of Burning from Bushfires in Southern and East Africa and Its Relevance to Hominin Evolution

by J. A. J. Gowlett, J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina

Online enhancements: supplemental figures

Early human fire use is of great scientific interest, but little comparative work has been undertaken across the eco- logical settings in which natural fire occurs or on the taphonomy of fire and circumstances in which natural and human-controlled fire could be confused. We present here results of experiments carried out with fire fronts from grass- and bushland in South and East Africa. Our work illustrates that in these circumstances hominins would have been able to walk with and exploit fires, and we emphasize that there can be different levels of fire use. The results also indicate that traditional assumptions about the discrimination of these are not reliable. Grass fires pass through the landscape rapidly in burns of less than 5 minutes duration, but areas of denser vegetation burn to much higher temperatures and for much longer. Trees are also caught in fires and may burn back into their roots, baking sediments. Animal bones on the surface can also become burned, so that presence of burned bone has to be used with care as an indicator of human activity. Duration of burning, repeated nature of burning, and copresence of features of human activity may give a better indication of human involvement.

Studies of early fire have come to the fore in recent years with 2012; Belcher, Collinson, and Scott 2013; Bond and Keeley important discoveries of new archaeological evidence at sites 2005; Bowman et al. 2011; Roos et al. 2014; Scott 2000, 2009; in Africa and other parts of the Old World and new hypothe- Scott et al. 2016). The subject of the human role remains con- ses about its evolutionary role (e.g., Berna et al. 2012; Gow- troversial because evidence of most burning disappears very lett 2010, 2016; Parker et al. 2016; Wrangham 2009, 2017) and rapidly, and there are also possibilities of confusion between significance in the biosphere (Archibald, Staver, and Levin evidence of controlled fire and evidence of wildfire. An under- standing of early human fire use can be obtained from a number of sources, including physiological and genetic sources (evi- J. A. J. Gowlett is Professor in the Department of Archaeology, Clas- dence based on metabolism, digestion, or genetics is inevitably sics, and Egyptology, School of Histories, Languages, and Cultures at largely indirect), but the archaeological evidence has the po- the University of Liverpool (Liverpool L69 3BX, United Kingdom [[email protected]]). J. S. Brink is Senior Specialist Scientist tential to be the most direct providing that there is an adequate and Head of the Department of Florisbad Quaternary Research at the framework for interpreting it (cf. Sandgathe 2017). fi National Museum (PO Box 266, Bloemfontein 9300, South Africa Exploration of the issues of early re use demands the [[email protected]]) and at the Centre for Environmental Man- availability of a set of “actualistic” comparative material, which agement of the University of the Free State (South Africa). Adam Caris remains largely lacking despite important pioneering efforts is Researcher in the Department of Archaeology, Classics, and Egyp- (e.g., Bellomo 1993, 1994; the first fire ethology appears to be tology, School of Histories, Languages, and Cultures at the Univer- owed to Al-Jahiz in the eighth century AD: see Stott 2012). The sity of Liverpool (Liverpool L69 3BX, United Kingdom [adamscaris scope of such work needs to be expanded now, partly because Sally Hoare @hotmail.com]). is Researcher in the Department of Ar- of the great variety of ecological circumstances in which fire chaeology, Classics, and Egyptology, School of Histories, Languages, occurs and partly because of poor understanding of when fire and Cultures at the University of Liverpool (Liverpool L69 3BX, United evidence is likely to be preserved or to decay. Here we present Kingdom [[email protected]]). S. M. Rucina is Senior Research Scientist and Head of the Department of Botany and the Department of the results of observations and experiments concerned with fi Palynology in the National Museums of Kenya (PO Box 40658-00100, African bush res that make available new information about Nairobi, Kenya [[email protected]]). This paper was sub- burning temperatures and durations and about effects on bone. mitted 25 VII 16, accepted 21 III 17, and electronically published 16 We emphasize that this is a field study, not a laboratory study. VI 17. Considerable efforts had to be made to control the fires and to q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0005$10.00. DOI: 10.1086/692249 Gowlett et al. Burning from Bushfires in Africa S207 ensure that they ran across the target areas where measure- Ideally one would follow a complete cycle of burning, decay of ment probes had been inserted. organic remains, and vegetation recovery, but this would in- volve great expense, and in any case some useful observations are the result of purely chance factors (e.g., encounter of a fire Previous Research on a given day). Evidence of patterns and consequences of burning have been The research addressed three particular questions. (1) To presented a number of times (e.g., Alperson-Afil and Goren- what extent do the temperatures observed in grass fires and Inbar 2010; Bliege Bird et al. 2012; Bond and Keeley 2005; Da- brush fires match those commonly accepted in archaeological vies, Holdaway, and Fanning 2016; Griffin 2002; Perlès 1977; observations and interpretations? (2) To what extent do tree Sergant, Crombe, and Perdaen 2006; Scott 2000, 2009; Vale fires occur, and to what extent are the traces left by these able 2002; see also Alperson-Afil 2017; Barkai et al. 2017; Dibble to mimic hearths? (3) To what extent do grass fires and bush- et al. 2017; Henry 2017; Holdaway, Davies, and Fanning 2017; fires affect surface bones? White et al. 2017). Bellomo’s work (1993, 1994) remains one of the most general treatments of methodology, based on a num- The Fires: Outline ber of experimental studies as well as literature and aimed to give documentation to the main kinds of natural fire as well as Most of our observations were made around Florisbad, about to the nature of hearths. 40 km north of Bloemfontein in South Africa (fig. 1), where a Bellomo’s research was geared to discriminating human fire reserve extends to the north of the museum (fig. 2). This re- from natural fire. Like many authors, Bellomo was inclined to search was stimulated by observation of an accidentally started emphasize temperature as a discriminator. He also suggested bushfire close to the museum in 2008 and also by similar en- that grass fires reached high temperatures only for a very short counters with fires at Kilombe in Kenya, possibly inadvertently period (cf. Clements 2010) and therefore do not create any started by charcoal burners. The range burning at Florisbad, considerable burning of materials. He set out criteria for rec- however, takes place as a means toward restoring native veg- ognizing hearths. He believed that they could be discriminated etation and improving grazing resources rather than for the from natural bole fires, regarding the latter as rare. In his ex- periments Bellomo found it difficult to generate tree stump fires, and he tended to minimize their role although they can occur at times even in rainforest environments (Tutin, White, and Mackangamissandzou 1996). Some assumptions drawn from Bellomo’s work are wide- spread, most especially that temperature is useful for discrimi- nation of hearth fire and wildfire. The presence of burned bone, although not much used by Bellomo, is frequently also taken as an indicator of human activity (e.g., Bosinksi 2006; Brain 2005; Brain and Sillen 1988). Again, further testing of the issues would be highly desirable, and what we report in this paper takes steps in that direction.

Pattern of Experimental Fire Research We report here on recent experiments and observations in South Africa and East Africa. They cover (1) experiments in burning set up as part of landscape management at Florisbad Quaternary Research Station, (2) following a bushfire and study of traces of previous bushfires at Soetdoring Nature Reserve near Florisbad, and (3) observation of very recent bushfires and charcoal burning at Kilombe farm in Kenya together with the recording of temperature variations with depth in exper- imental campfires. In this work it was possible to compare some burning events as they happened with the results of other fires that had oc- curred recently nearby. In this respect the pattern is similar to that in Robbins’s classic study of the decay of Turkana huts, where he was able to compare some as they were abandoned Figure 1. Map of Africa showing fire sites mentioned in the text. with others that were in the process of collapse (Robbins 1973). A color version of this figure is available online. S208 Current Anthropology Volume 58, Supplement 16, August 2017 Fires at Florisbad Quaternary Research Centre and Museum The museum is set within a region of grassland and small kar- roid shrubs (sweet veld) with occasional trees, especially closer to water (see Janecke and du Preez 2005). Locally, an area approaching 100 hectares had been farmed historically but has been museum land since 1980. Bush and scrub clearing by fire was organized in this area by the National Museum in 2012 and 2013. The aim of current conservation is to remedy the effects of overgrazing during past farm use and to restore grassland veld vegetation that can support an antelope popu- lation. The land is situated at about 1,250 m asl and descends gently to the north toward the Soutpan depression (Kuman, Inbar, and Clarke 1999; Scott and Brink 1992; fig. 2). In the course of the systematic burns at Florisbad, it was possible to measure burning temperatures near vegetation us- ing thermocouples (see n. 1). Additionally, we were able to place partial carcasses of two antelope for study of burning ef- fects, discussed below. The burns took place over a number of days in September 2012 and 2013 toward the end of winter and before the onset of any rains. Around the world records show that many lightning-induced fires occur at the start of rains following a dry season (e.g., Gisbourne 1931; Johnson 1992). In all cases here the vegetation was in much the state of dryness that would come into contact with the first thunderstorms a few weeks later. Examples of the burning are described and explained below. Figure 2. Museum reserve area at Florisbad. The grassland ex- tends from the museum at the southern end to the edge of the Soutpan at the north (satellite imagery courtesy of Google Earth). Temperatures of Burning in Open Grassland A color version of this figure is available online. Grass burns were conducted on a number of days when wind speeds were relatively low (burns were not permitted when research. In total we were able to observe fire in grassland, in wind speeds were higher than about 16 kph). The grassland low scrub, and in savanna (or riparian woodland), including at Florisbad is made up of a variety of species similar to those groups of trees of up to about 5–8 m in height. We were able to determine effects on vegetation and also on organic materials such as bone. In general, wind speeds were relatively low at a few miles per hour. In all cases fire fronts moved through relatively fast, with most immediate burning taking place within a space of a few minutes and of less than 30 minutes even in the case of trees (figs. 3–6 below). Temperatures, however, were exceedingly variable according to type and density of burned plant mate- rial, as demonstrated by measurements taken with thermo- couples attached to long heat-resistant leads (cf. Govender, Trollope, and Van Willgen 2006).1

1. Temperatures were measured with the use of long thermocouples. The tip lengths of ca. 100 cm were heat resistant up to 1,1007C. The re- maining lengths of ca. 12 m could withstand 3007C and were therefore buried at a shallow depth. Data were logged on a standard temperature recorder from a fire lane either at intervals or continuously recorded with Figure 3. Temperature measurements of two fire fronts in grass- the aid of a video camera. land (veld) at Florisbad. Gowlett et al. Burning from Bushfires in Africa S209

lustrates temperatures in an additional similar case of bush and grass burning. Another case illustrates the way in which the initial rapid burn can lead to continued burning and higher temperatures in localized areas of denser vegetation (fig. 6). One probe was attached to a small prickly bush, another to densely matted grass. The bush burned fiercely and rapidly without very high temperature, but temperatures of the burning grass rose to almost 6007C and were maintained at greater than 2007C for around 15 minutes. Most interesting, adjacent dung was ig- nited, but it did not burn rapidly; rather, the temperature was rising toward 507C as the others declined, and it had the effect of prolonging local burning to at least 30 minutes. In this case the longer burning may reflect input of oily material from the bush coupled with the dense matting of the grass. Figures 4 and 5 also show a slight bounce back of temperature, possibly be- Figure 4. Temperature measurements of two fire fronts in The- cause woody material took longer to ignite and burn than the meda triandra (red grass) stands. grasses. The principal finding is that although most grass fires had a predictably rapid burning plume structure, time duration, and at nearby Soetdoring (Janeke and du Preez 2005), sometimes temperature profile (cf. Clements 2010), in practice wide ranges occurring in dense stands, sometimes patchy. At Florisbad of temperature variation were observed. Grass fires were of the vegetation/grass height is generally 30–60 cm. The grasses were order of 3007C, but the plume spike rose to about 7007Cas ignited easily and burned in moving fronts. gases ignited (cf. Clements 2010). The temperature and dura- Figure 3 illustrates measurements from two adjacent probes tion of burning were notably raised when the fire caught hold in open grassland. In one of these temperature rapidly rose of shrubs. from ambient to more than 5007C, but in 2 minutes it had fallen 7– 7 back sharply toward ambient (i.e., 20 30 C). The other probe Fires at Soetdoring Nature Reserve records a lower peak (possibly because the highest temperature endured for just a few seconds), but temperature of 12007C Soetdoring lies about 5 km south of Florisbad. Observations was recorded for 4 minutes. Two measurements made on The- from three bushfires at Soetdoring provided further evidence meda triandra (red grass) the following day in similar condi- about burning of vegetation, especially trees, and also of bone. tions indicate a similar picture with peak temperature reached The game reserve flanks a reservoir extending behind the Kru- and falling right back within 3 minutes (fig. 4; see also fig. S1; gersdrift dam on the Modder River and lies 40 km north of figs. S1–S8 available online). Bloemfontein and about 5 km from the Florisbad museum

Temperatures of Burning: Small Shrubs and Grasses This example from the same suite of fires demonstrates tem- peratures and speed of burning in grassland in relation to a small shrub (GPS point 358: 28 45.159S/26 05.974E; Septem- ber 17, 2012): small karroid shrubs occur patchily in the area (Janecke and du Preez 2005) and may significantly affect in- tensity of burning. Four thermocouples were set on a bush at heights of 60 cm and 10 cm, on an adjacent small stump at a height of 25 cm, and in grassy brush at 5 cm. At 9:45 a.m. am- bient temperatures on these were 187–247C. Fire was set lo- cally at 10:03 a.m. As the front hit the area around 10:05 a.m., temperatures rose to 1517C, 2407C, 407C, and 1837C respec- tively. All temperatures then sharply declined except that of the grass brush, where the temperature rose briefly to 5007C. By 10:30 a.m., temperatures had fallen to 437C, 397C, 277C, and 257C, closing toward the original ambient temperatures. A piece of cattle dung 2 m from the bush was recorded burning at Figure 5. Temperature measurements on small bush and grasses 6057C at 10:16 a.m. and at 1,0457C at 10:37 a.m. Figure 5 il- at Florisbad. S210 Current Anthropology Volume 58, Supplement 16, August 2017

milis; Janecke and du Preez 2005). Larger trees are found to- ward the river, including especially sweet thorn (soetdoring), a species of Acacia. We were able to observe and investigate evidence from three bushfires: (1) a fire in progress on Sunday, September 23, 2012; (2) a fire of July/August 2012, about 6 weeks old at the time of observation; and (3) a fire of Monday, September 10, 2012, ob- served at the end of the burn and through the following week. The approximate extents of the burns are shown in figure 7. The cause of ignition of fire 3 was reported to be a barbecue fire. Fires 1 and 2 may also have been caused by human ac- tivities, such as the dropping of a cigarette. We discuss the fires out of chronological order so as to follow a train of argument.

fi Figure 6. Temperature measurements on prickly bush, dense Fire 1: Bush re Followed in Action matted grass, and dung. The dung would have ignited shortly after This fire was observed by chance at the reserve on Sunday the fire front passed, but a thermocouple was inserted only when the dung was noted smoking as the other fires subsided. September 23, 2012. It was immediately visible from about 2 km as it began to consume trees (fig. S2). The cause of ig- nition is not known, but there did not appear to be any visitors site. The landscape is to the eye a mosaic of grassland and sa- in that area of the park. We were able to follow the fire for more vanna, and in botanic terms it is part of the Grassland Biome of than 500 m, up to the point where it came up to a creek, which the sweet veld and similarly regarded as a mosaic dominated it eventually passed around. either by grasses or by dwarf karroid shrubs (such as Chryso- It was striking that the fire very rapidly consumed trees coma ciliata, Pentzia incana, Pentzia globosa, and Rosenia hu- (fig. S3), moving rapidly and loudly, but also that the burning

Figure 7. Soetdoring Nature Reserve. Approximate outlines of the fires in 2012 (base map courtesy of Google Earth). The fires were all in wooded grassland along the south bank of the Modder River. A color version of this figure is available online. Gowlett et al. Burning from Bushfires in Africa S211 was selective, with adjacent trees and even grass patches often more. In each case these had been lying in grass, and there is no unaffected. We were able to observe the way in which trunks doubt from the taphonomic indications discussed below that burned through at the height of 1–1.5 m and in which the the animals had already been dead for at least several months upper parts of trees were often entirely consumed so that and their soft parts were fully decomposed at the time that the nothing remained except fine ash within a few minutes. fire passed. In general, the fire burned out by the water’s edge, but in places it consumed reed beds, with dramatic flaming and smoke. Bones of medium-size bovid (hartebeest Alcelaphus buse- We were not able to use temperature probes but inferred that laphus). The skeleton was essentially complete, with the bones there was a great range of burning temperatures even very scattered across an axis of about 4 m and bone positions partly locally. representing past anatomical connections. Bones were vari- ably burned, with the skull and horn cores having suffered particularly, perhaps because they stood a few centimeters Fire 2: Trees and Associated Debris Left by Fire higher above the surface than the other bones. The skull is the This fire, the earliest of the three, was not observed by us: it most burned (fig. S7), and it is separated into two pieces. Nasal must have been similar in character to fire 3 but perhaps hot- passages were the focus of burning (the main part of the cra- ter, judging by the damage, and more extensive (fig. 7). An area nium had been lying on its base, with the frontal uppermost). was recorded about 700 m south of the believed fire source at In contrast, the hemimandibles were scarcely burned except (28 49.693S/26 03.088E). Here the fire had passed through a for the tip of the condyles on both sides. The scapulae had lain group of trees forming a sparse copse about 25 m across, and flat on the ground: one was charred brown on the dorsal sur- two areas of 15 # 9 m were plotted (fig. S3). face, the other just on the dorsal spine. Both sides of the pelvis The burning illustrated several features characteristic of the are quite strongly charred, especially around the pubic area, area. The trees had burned to a varied degree. One stump in- but also around the margins of the ilia (fig. S8). dicated a cut by flames at about 1.5 m (fig. S4). At least two or On the limb bones there was very little sign of burning ex- three trees had been consumed entirely, with only fine ash cept light charring around the distal right humerus, probably trails preserving the pattern of the fallen trunks and branches after the epiphysis had detached (the equivalent ephiphysis In several instances the trees had burned right back into was in position on the other side, with no obvious burning). their roots, creating fire-baked zones up to 2 m across (figs. S5, The ribs showed little sign of burning, except at vertebral ends, S6). These are of particular interest in the sense that they could where there was some charring. mimic hearths and create clasts of baked clay possibly simi- Charring of the vertebrae is very variable, prominent in lar to those found at Chesowanja (Gowlett et al. 1981). We some pieces, absent in others, but the place of burning varies excavated a section across one of these combustion zones and between equivalent vertebrae. Dorsal spines were most com- found that the baked zone extended to about 20 cm deep monly charred. (fig. S6), almost certainly indicating high temperature main- tained for considerable time (we discuss hearth temperatures Jackal (Canis mesomelas). The jackal had apparently be- further below in relation to experiments at Kilombe). In gen- come ensnared at the foot of an old wire fence, probably no eral, from inspection of quite a number of such burned out more than a few months previously, as some of its pelt re- stumps, we observe that they could be distinguishable from mained as a mass. Its skeleton is virtually complete; again, its hearths if they preserve the shape of spreading descending bones were variably burned. Some show little or no trace of roots (acriterion alsonotedbyBellomo 1993).Even wherethere burning, but others, including the flat plates of vertebral spines, is a wide-open depression, the roots can be observed at the are considerably charred. The skull/mandible has slight char- margins, sometimes with charred wood preserved (note that ring around the nasal aperture. The cervical vertebrae show Melson and Potts [2002] described other ways in which natural charring of spines, above or below, and one more than the baking of sediments can occur in East African environments). others. The thoracic vertebrae have the heaviest burning on Last, there were scattered bones on the surface by the trees the spines (six badly charred vs. five scarcely charred). The of a duiker-size bovid. The scapula and humerus were quite lumbar vertebrae have minor burning to the spines. The pelvis heavily burned, suggesting that the brush had burned quite has some charring to the margin of the ilium on one side. About intensely around them. half of the ribs are charred at one end, usually the proximal end. And the upper and lower limbs have very little visible damage, with just slight charring to the ridge of one scapula. The foot Fire 3: Bushfire and Effects on Animal Carcasses bones, metapodials still articulated, are uncharred. This fire, close to the main entrance of Soetdoring Park, we were able to observe while some areas were still smouldering. Small antelope. The small duiker was about 70% complete: It gave the best opportunities to study effects on the bones of one forelimb had vanished, and two feet were missing, as was carcasses. These were of animals that had died and decayed most of the thoracic vertebrae. The ribs were fragmented. Fire down to their skeletons before the fire, perhaps by a year or damage was as follows. The skull has light charring of the edges S212 Current Anthropology Volume 58, Supplement 16, August 2017 of nasal cavity. The mandibles show light charring of the tip of flesh and skin on. These parts were left in the path of the fire one condyle. On the cervical vertebrae, there is darkening at so that the fire could run over them, as if in a natural fire. the tip of one spine. One complete thoracic vertebrae is lightly Temperature probes were attached to the limbs of the bles- browned. Of the lumbar vertebrae, five articulating, there is bok and to the wildebeest limb. In an effort to maximize burn- plain charring of dorsal spines and very light charring of lateral ing, the specimens were set beside a fairly large bush (other- spine tips. The forelimb shows taphonomic damage to the wise the situation was as in the other grass fires described). plate of the scapula and charring of the chewed edge and on the Figure 8 shows the resultant temperature profile and burning ridge. There is none to the cup. The distal end of the humerus duration. is missing, and there are signs of light chewing; the sharp bone The two curves suggest that the temperature for one ther- edge is considerably charred. The proximal end of the radio- mocouple was maximized by being in the plume of burning, ulna is considerably charred, and the radius is fragmented and the other minimized perhaps by being on the top of the car- charred on one edge. The ilia of the pelvis had separated, and cass and not exposed to direct flame. They do, however, have there is light chewing of the tips and edges and light charring of similar profiles, showing the usual rapid heating with the fire the pubic area. There is light charring to parts of both femurs front followed by a cooling off of about 10–15 minutes. (also taphonomic damage from chewing), one tibia, and one When inspected, the blesbok skeleton, which had meat and calcaneum. There is no obvious burning damage to the foot skin attached, showed very few signs of burning. These were parts preserved. confined largely to lower parts of limbs. The left scapula spine was darkened; the lower part of the left front limb was dark- ened (distal shaft of metacarpal) or had became white or gray Experimental Burning of Antelope Carcasses and brittle (phalanges). Tibia and metatarsal of the left hind at Florisbad limb also became white and brittle. The right front limb of fl We set out to determine whether fresh antelope carcasses the black wildebeest, with its esh but no skin, showed only a would become burned in a comparable way in terms of vis- slight darkening to the proximal lateral part of the radio-ulna. fi ible damage to bones and whether the presence of meat would In summary, we found that in the passing re the meat at- affect the burning. To this end, two culled carcasses, a black tached to the blesbok and to the right limb of the wildebeest fi wildebeest and a blesbok, were acquired from the Caledon did not heat suf ciently to cook substantially and that it served Nature Reserve (a reserve managed by the Free State Provin- to protect the bones. cial Government). The rest of the skeleton of the wildebeest, which was de- fleshed and placed in a separate fire path, could have been more directly exposed to burning because of the defleshing, but in Black Wildebeest fact there were very few signs of burning other than slight darkening here and there. The blades of the ilia were slightly This was a young adult black wildebeest male with the third darkened and brittle. As with the blesbok, the tibiae were molar erupted but not yet in wear. The carcass was complete affected, with the left in particular being brittle and crumbly. but skinned and with the intestines removed. At Florisbad we defleshed the carcass, except for the right front limb, which was detached and kept separately.

Blesbok This was an adult female blesbok with a broken left meta- carpal that was incompletely healed. (The break must have occurred a few seasons ago, because there was much addi- tional bone growth around the break, which forms in an attempt to stabilize the bone. From the pelvis it is clear that she had had several calves.) The carcass was received intact, with skin still on but with the intestines removed. The carcass was kept as it was. We wanted to see the effect of the grass (veld) fire on (1) defleshed bone (the carcass of the black wildebeest without the right front limb), (2) bone with flesh on but without skin (the right front limb of the black wildebeest), and (3) bone with flesh and skin on (the carcass of the blesbok). Therefore, we put out the defleshed black wildebeest carcass, the right Figure 8. Temperature measurements of wildebeest and blesbok front limb with flesh still on, and the blesbok carcass with limb bones in grass fire and bushfire at Florisbad. Gowlett et al. Burning from Bushfires in Africa S213 Experimental Campfires at Kilombe, Kenya for discriminating between human-controlled and natural fire but also emphasize that various factors influence the degree of Experimental fires were carried out by Sally Hoare on clay baking of sediments. rich substrates in the area of the archaeological site of Ki- lombe, Kenya (Gowlett et al. 2015), and the aim was to pro- Discussion vide a further check on the common statement that campfires burn in the range of 3007C–8007C (e.g., Bellomo 1993, 1994 The fire experiments detailed here bring together a number of as previously cited; also Gowlett et al. 1981). On the first day, points that may sometimes be common knowledge in forest the ground was prepared by removing the first 5 cm of soil fire studies (e.g., papers in Scott et al. 2016) but that have not and then marking out an area of 150 cm by 150 cm. K-type been established or applied in strict archaeological terms. We thermocouples were used in order to take temperature read- have concentrated here on documentation of the visible ta- ings at regular intervals of 15 minutes during burning over a phonomy of fire remains rather than on behavioral interpre- period of 1 hour 30 minutes. This procedure was repeated for tations. The experiments were often carried out as ancillary a further campfire on the second day. The thermocouples (T) activities to range management or other research and clearly were arranged in two rows, one in the center of the fire and mark only a beginning to work of this kind. They were car- the second on the periphery of the fire and at the different ried out in the open veld in the context of fire hazards, and depths of 1–2 cm (T2), 3–4 cm (T3), and 6–7 cm (T4). A considerable efforts had to be made to ensure that the fires further thermocouple was used to record the temperature of passed over the target thermocouples and to keep the fires the actual flames during burning (T1); that is, it was lodged within control. Our efforts therefore concentrated on estab- above the fire. The fuel used was local wood, chiefly Acacia. lishing temperatures and rates of burning. As we took the The highest temperatures were recorded by thermocouples opportunities presented by chance factors, such as the on- T1 and T2, which were recording the temperatures of the going accidental fires at Soetdoring, we were not always able actual flames and at 1–2 cm depth, respectively. Maximum to have prior choice of the experimental materials. We do not values for T1 varied between 4807C and 6207C, and T2 varied know exactly how long bone distributions had lain on the between 3807C and 4027C. Temperatures recorded at depth surface or how much the collagen in the bones had decayed. were much lower and did not exceed maximum temperatures With temperatures and durations of burning established, how- of 2487Cat3–4 cm (thermocouple T3) and 1507Cat6–7cm ever, such factors can be much more easily studied in experi- (thermocouple T4). Thermocouples T5–T7, placed at the pe- mental situations that do not involve bushfires but that rather riphery of the campfire, did not record temperatures above make use of smaller fixed experimental fires or even ovens. 587C. Temperatures were similar for the second 90-minute Similar observations can be made about the burning of sed- fire the following day. iments. We believe that most campfires, like those at Kilombe, These temperatures clearly occupy a similar range to those do not generate a great deal of heat below and do not bake of the passing grass fires. The suggestion that temperatures of sediments to any depth. The passing bushfires also often have hearths and campfires can range from 3007C to 8007Cis little effect on the lowest vegetation and plant roots. On the likely to be accurate in relation to the flames but not neces- other hand, tree burns generate great heat and can bake sed- sarily in terms of the firing temperature of the underlying iments through 20 cm or more. Continuing controlled work soils and sediments, a point that may have implications for on the burning of sediments is clearly valuable for establish- the use of temperature measurements to identify human ac- ing a broader picture (Aldeias 2017). tivity in the archaeological record. Much lower temperatures At the level of animal behavior, it is plain that for all its than are generally attributed to human campfires were re- great importance fire does not create resources, rather it con- corded during the experimental campfires at Kilombe, spe- verts or attracts resources. The fires studied in South Africa cifically at depths of 3–4 cm, although at a depth of 1–2cm were set in a low resource area, and although there were signs temperatures just over 4007C were recorded. This may in part of effects on rare invertebrates, birds, and small animals such be due to the relatively short burning times of the particular as tortoises, all of which would be of interest to hunter- experiments. (The fire fronts described above also often passed gatherers, these were very infrequent in these particular land- without consuming much of the lowest level of grass and dry scapes. In another environment, hominins might find that plant debris, suggesting that there would be little if any effect the density of resources would give far higher incentives for on the underlying sediments.) In other camp fires we have following fire. As far back as the ninth century AD, Al-Jahiz seen considerable baking of sediments, which did not occur commented on the large numbers of animal species attracted here, suggesting that longer duration experiments and also to a fire (Stott 2012). In the fires observed by us, the com- variations in fire size are needed for establishing a range of bination of relatively low vegetation fuel load and low wind comparisons. There is an issue of preservation, as complete speeds allowed safe transit for humans around the fire areas, deposits rarely survive on early hominin sites, so traces of a as has also been observed for chimps in West African sa- temperature gradient could easily be lost. The figures again vanna (Pruetz and Herzog 2017; Pruetz and LaDuke 2010), suggest that temperature on its own is not a reliable indicator but the greater intensity of the burn at Soetdoring on Sep- S214 Current Anthropology Volume 58, Supplement 16, August 2017 tember 23 dramatically highlighted the power of fire and haps the century following use. Most environments are ero- particularly its ability to destroy trees very rapidly. sional: even in a depositional environment, out movement of Stump fires are of particular interest here as they were material may balance in movement of sediment. It is worth generated with difficulty in Bellomo’s (1993, 1994) work and restating a point made by Richard Preece that most European feature little in his scheme but could potentially lead to cre- open sites with hearths are preserved through coincidence with ation of hearth-like features. It is therefore useful that we tufaceous deposits (Preece et al. 2006, 2007); most others are observed evidence of stump fires numerous times in the Soet- preserved through rapid deposition of silts, as at Pincevent doring fires. In the African savannas many trees are fire re- (Julien 2003), or loess, as on the Russian plains (Holliday et al. sistant, and we have rarely observed evidence of stump fires 2007; Klein 1973). In Africa, there is a similar rarity of open- such as presented here, although evidence of tree stump fires air hearths even in relatively recent periods where human use has been observed at Chesowanja in Kenya, where a wooded of fire is well accepted (Florisbad is one example, preserved area was destroyed in the 1930s (Clark and Harris 1985) and in sands: Henderson 2001; Kuman, Inbar, and Clarke 1999). charred stumps were still visible in recent years. Animal bones, too, have a relatively short survival time on the We observed baked clay evidence, similar to that observed surface, in Africa ~20 years (see Gifford 1980). It is ques- in the Soetdoring fires, in Kenya when a bushfire took place tionable, then, whether any of the sites studied in our ex- in January 2011, progressing from the northeast to the south- periments would have passed into a long-term record. Studies west across large parts of Kilombe farm. Our group narrowly such as those of Davies, Holdaway, and Fanning (2016) and missed observing the fire but was able to search its consequen- Holdaway, Davies, and Fanning (2017) can cast further light tial evidence some months later in July 2011. One burned on the probabilities through modeling and large-scale field sur- stump recorded by us at Kilombe appears (from oral evidence veying of hearth distributions, which is difficult or impossible and inspection) to have been a dead tree with its core already for earlier periods. rotted away. Sometimes these are found at the heart of a termite mound that has built up around the tree, and in that Conclusions case baked clay would also be produced in a burn. Most of the trees burned at Soetdoring, however, were certainly alive. The field experiments around Florisbad and at Kilombe em- At Soetdoring there was a common scenario of a fire plume phasize a number of points on a landscape scale and in relation from grass and shrub fire that, as it passed, would attack the to detailed archaeological evidence. A first point is the great tree boles at a height of around 1.5 m. They would then burn visibility of bushfires and that they can often be followed on through at this point, with the upper part dropping on its base the ground with reasonable safety. They attract attention and (figs. S3, S4); both parts would then continue to burn, the can signify the possibility that food resources will be available. stump downward into its roots and the upper part outward Next, temperature is in no sense a reliable discriminator toward its extremities, which sometimes remained unburned. between grass fires and humanly managed fires. Both clearly The obvious archaeological relevance is that such fires burn range through 2007C–8007C on a regular basis. hot, certainly up to 8007C—the full range of hearth fires—and Baked clay is created by stump fires, by falling burning that they can also create baked areas. trunks, and by human activities. The natural instances as ob- Although this evidence seems a cautionary tale—that nat- served by us result from intense combustion of tree wood. ural tree fires may cause hearth-size baked features—it is These instances should have the character—for instance, in encouraging that they also corroborated an observation of magnetic studies—of a single high-temperature peak followed Bellomo (1993) that the shape of the tree root features is dis- by very gradual cooling. Any trace of repeated episodes of heat- tinctive and could be distinguishable from a hearth. As in- ing would be a very strong indicator of human activity. The dicated in figures S5 and S6, the spreading roots directed the stump fires also have a different physical conformation from shape of the combustion feature produced. hearth “bowls.” At Soetdoring, Adam Caris located one such tree burn Bone lying on the surface during a wildfire can clearly be- shortly after the fire. Ash filled the hollow in the baked clay. come charred, even in rapidly passing grass fires. Paradoxi- Not only did it have a temperature of around 4007C, this was cally, meaty bones appear far less likely to become charred. maintained for several days afterward. It suggests the possi- The coincidence of charring and butchery marks on bone, as bility that in the early days of interactions with fire, hominins at Swartkrans (Brain 2005; Brain and Sillen 1988; Pickering could easily have been drawn to any feature that naturally 2012), appears to argue far more strongly for human fire use retained a high temperature for a long period. than does charring of bone alone. Again, points made about The studies we report here can be related to early human repeatedness of human fire use in one locality are relevant behavior and the likelihood of its preservation in a number of (Alperson-Afil and Goren-Inbar 2010; Gowlett 2010): repeated ways. Chance factors ought to be emphasized both in the pri- fires may well be necessary for bones with combined evidence mary creation of a record and its later preservation. It seems of charring and butchery to result in any number. likely that in most circumstances the great majority of hearths In these experimental fires animal dung turned out to be will be eroded away through a period of time, generally per- one of the most important factors in the prolongation of burn- Gowlett et al. Burning from Bushfires in Africa S215 ing. It can frequently be seen smouldering after a fire front has Berna, F., P. Goldberg, L. K. Horwitz, J. Brink, D. Holt, M. Bamford, and M. Chazan. 2012. Microstratigraphic evidence of in situ fire in the Acheulean passed, and it does not seem speculative to suggest that hom- strata of Wonderwerk Cave, Northern Cape Province, South Africa. Pro- inins would have been aware of this. The gathering and car- ceedings of the National Academy of Sciences of the USA 109(20):E1215– rying of burning dung could have been one of the first steps E1220, doi/10.1073/pnas.1117620109. fi Berthold, P., H. G. Bauer, and V. Westhead. 2001. Bird migration: a general for hominins becoming engaged with re use. It is worth re- survey. Oxford: Oxford University Press. stating another point, that natural fire is rarely a good chef. Bliege Bird, R., B. F. Coding, P. G. Kauhanen, and D. W. Bird. 2012. Ab- On numbers of occasions, here and elsewhere, we have noted original hunting buffers climate-driven fire-size variability in Australia’s fi spinifex grasslands. Proceedings of the National Academy of Sciences of the that a re passes too fast to complete cooking, or conversely, USA 109:10287–10292, doi:10.1073/pnas.1204585109. it may burn eggs completely. There would be a strong incen- Bond, W. J., and J. E. Keeley. 2005. Fire as global “herbivore”: the ecology and tive for hominins following fires to intervene, even to a minor evolution of flammable ecosystems. Trends in Ecology and Evolution 20: 387–394, doi:10.1016/j.tree.2005.04.025. degree, to improve the quality of cooking either by putting Bosinski, G. 2006. Les premiers peuplements de l’Europe centrale et de l’Est. food resources back into the fire or pulling them out. The In Climats, cultures et sociétés aux temps préhistoriques: de l’apparition des signs of continued burning behind the fire front in dung and hominidés jusqu’au Néolithique. H. de Lumley, ed. Special issue, Comptes Rendus Palevol 5(1/2):311–317. stumps might be a strong pull to humans to engage in that Bowman, D. M. J. S., J. Balch, P. Artaxo, W. J. Bond, M. A. Cochrane, C. M. kind of activity. D’Antonio, R. Defries, et al. 2011. The human dimension of fire regimes on If foraging is the most natural route into fire using, as Earth. Journal of Biogeography 38:2223–2236, doi:10.1111/j.1365-2699.2011 .02595.x. might be suggested by the large numbers of bird and animal Brain, C. K. 2005. Essential attributes of any technologically competent ani- fire followers (Berthold, Bauer, and Westhead 2001; Gowlett mal. In From tools to symbols: from early hominids to modern humans.F. 2010), it would require for hominins a very detailed knowl- d’Errico and L. Backwell, eds. Pp. 38–51. Johannesburg: Witwatersrand University Press. edge of potential resources and the way in which these could Brain, C. K., and A. Sillen. 1988. Evidence from the Swartkrans cave for the be managed—the beginnings of “fire farming,” which can be earliest use of fire. Nature 336:464–466. Clark, J. D., and J. W. K. Harris. 1985. Fire and its roles in early hominid seen as a somewhat different problem from the origins of – fi lifeways. African Archaeological Review 3:3 27. hearth re that are the focus of many studies. Clements, C. B. 2010. Thermodynamic structure of a grass fire plume. In- ternational Journal of Wildland Fire 19:895–902. Davies, B., S. J. Holdaway, and P. C. Fanning. 2016. Modelling the palimpsest: Acknowledgments an exploratory agent-based model of surface archaeological deposit for- mation in a fluvial arid Australian landscape. Holocene 26:450–463, doi:10 We thank National Museums of South Africa and all the .1177/09596836.15609754. work crew at Florisbad (especially Jaco Smith, Isaac Thapo, Dibble, Harold L., Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Abel Dichakane, Jacob Maine, and Johannes Motshabi); the Shannon P. McPherron, and Dennis M. Sandgathe. 2017. How did hom- inins adapt to Ice Age Europe without fire? Current Anthropology 58 management of the Soetdoring Nature Reserve; National Mu- (suppl. 16):S278–S287. seums of Kenya; the National Commission for Science, Tech- Gifford, D. P. 1980. Ethnoarchaeological contributions to the taphonomy of nology, and Innovation; the British Academy, which gave sup- human sites. In Fossils in the making: vertebrate taphonomy and paleo- ecology . A. K. Behrensmeyer and A. P. Hill, eds. Pp. 93–106. Chicago: Uni- port through the Lucy to Language project at the start of this versity of Chicago Press. work and through the Mobility and Links project with Na- Gisbourne, H. T. 1931. A five-year record of lightning storms and forest fires. tional Museums of Kenya; and to Maura Butler. Monthly Weather Review 59(4):139–149. Govender, N., W. S. W. Trollope, and B. W. Van Wilgen. 2006. The effect of fire season, fire frequency, rainfall and management on fire intensity in References Cited savanna vegetationinSouthAfrica.Journal of Applied Ecology 43:748– 758. Aldeias, Vera. 2017. Experimental approaches to archaeological fire features Gowlett, J. 2010. Firing up the social brain. In Social brain and distributed and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– mind. R. Dunbar, C. Gamble, and J. Gowlett, eds. Pp. 345–370. London: S205. British Academy. Alperson-Afil, Nira, and N. Goren-Inbar. 2010. The Acheulian site of Gesher Gowlett, J. A. J. 2016. The discovery of fire by humans: a long and convoluted Benot Ya‘aqov, vol. 2 of Ancient flames and controlled use of fire. Vertebrate process. Philosophical Transactions of the Royal Society B 371:20150164, Paleobiology and Paleoanthropology. Dordrecht: Springer. doi:10.1098/rstb.20150164. Alperson-Afil, Nira. 2017. Spatial analysis of fire: archaeological approach to Gowlett, J. A. J., J. S. Brink, A. I. R. Herries, S. Hoare, I. Onjala, and S. M. recognizing early fire. Current Anthropology 58(suppl. 16):S258–S266. Rucina. 2015. At the heart of the African Acheulean: the physical, social Archibald, A., A. C. Staver, and S. A. Levin. 2012. Evolution of human-driven and cognitive landscapes of Kilombe. In Settlement, society and cognition fire regimes in Africa. Proceedings of the National Academy of Sciences of in human evolution: landscapes in mind. F. Coward, R. Hosfield, M. Pope, the USA 109:847–852, doi:10.1073/pnas.1118648109. and F. Wenban-Smith, eds. Pp. 75–93. Cambridge: Cambridge University Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a Press. reason: barbecue at Middle Pleistocene Qesem Cave, Israel. Current An- Gowlett, J. A. J., J. W. K. Harris, D. Walton, and B. A. Wood. 1981. Early thropology 58(suppl. 16):S314–S328. archaeological sites, hominid remains and traces of fire from Chesowanja, Belcher, C. M., M. E. Collinson, and A. C. Scott. 2013. A 450 million year Kenya. Nature 294:125–129. record of fire. In Fire phenomena in the earth system: an interdisciplinary Griffin, D. 2002. Prehistoric human impacts on fire regimes and vegetation in approach to fire science. C. M. Belcher, ed. Pp. 229–249. London: Wiley. the northern intermountain West. In Fire, native peoples and the natural Bellomo, R. V. 1993. A methodological approach for identifying archaeo- landscape. T. R. Vale, ed. Pp. 77–100. Washington, DC: Island. logical evidence of fire resulting from human activities. Journal of Ar- Henderson, Z. 2001. The integrity of the Middle Stone Age horizon at chaeological Science 20:525–555. Florisbad, South Africa. Navorsinge van die nasionale Museum Bloemfon- ———. 1994. Methods of determining early hominid behavioural activities tein 17(2):26–52. associated with the controlled use of fire at FxJj20 Main, Koobi Fora, Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current Kenya. Journal of Human Evolution 27:173–195. Anthropology 58(suppl. 16):S329–S336. S216 Current Anthropology Volume 58, Supplement 16, August 2017

Holdaway, Simon J., Benjamin Davies, and Patricia C. Fanning. 2017. Ab- Pruetz, Jill D., and Nicole M. Herzog. 2017. Savanna chimpanzees at Fongoli, original use of fire in a landscape context: investigating presence and ab- Senegal, navigate a fire landscape. Current Anthropology 58(suppl. 16): sence of heat-retainer hearths in western New South Wales, Australia. Cur- S337–S350. rent Anthropology 58(suppl. 16):S230–S242. Robbins, L. H. 1973. Turkana material culture viewed from an archaeological Holliday, V. T., J. F. Hoffecker, P. Goldberg, R. Macphail, S. L. Forman, M. perspective. World Archaeology 5:209–214. Anikovich, and A. Sinitsyn. 2007. Geoarchaeology of the Kostenki- Roos, C. I., D. M. J. S. Bowman, J. K. Balch, P. Artaxo, W. J. Bond, M. Borshchevo sites, Don River Valley, Russia. Geoarchaeology 22:181–228, Cochrane, C. M. D’Antonio, et al. 2014. Pyrogeography, historical ecology, doi:10.1002/gea.20163. and the human dimensions of fire regimes. Journal of Biogeography 41 Janecke, B. B., and P. J. du Preez. 2005. A synoptic view on the grassland (4):833–836, doi:10.1111/jbi.12285. vegetation of Soetdoring Nature Reserve, Free State Province. South African Sandgathe, Dennis M. 2017. Identifying and describing pattern and process in Journal of Botany 71:339–348. the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): Johnson, E. A. 1992. Fire and vegetation dynamics: studies from the North S360–S370. American boreal forest. Cambridge: Cambridge University Press. Scott, A. C. 2000. The pre-Quaternary history of fire. Palaeogeography, Palaeo- Julien, M. 2003. A Magdalenian base camp at Pincevent (France). In Perceived climatology and Palaeoecology 164:297–334. landscapes and built environments: the cultural geography of late Palaeo- ———. Forest fire in the fossil record. In Fire effects on soils and restoration lithic Eurasia. S. A. Vasilev, O. Soffer, and J. Kozlowski, eds. Pp. 105–111. strategies. A. Cerdà and P. Robichaud, eds. Pp. 1–37. Enfield, NH: Science BAR International Series 1122. Oxford: Archaeopress. Publishers. Klein, R. G. 1973. Ice-Age hunters of the Ukraine. Chicago: University of Scott, A. C., W. G. Chaloner, C. M. Belcher, and C. I. Roos, eds. 2016. The Chicago Press. interaction of fire and mankind. Thematic issue, Philosophical Transactions Kuman, K., M. Inbar, and R. J. Clarke. 1999. Palaeoenvironments and cultural of the Royal Society B 371:1696. sequence of the Florisbad Middle Stone Age hominid site, South Africa. Scott, L., and J. S. Brink. 1992. Quaternary palaeoenvironments of pans in Journal of Archaeological Science 26:1409–1425. central South Africa: palynological and palaeontological evidence. South Melson, W. G., and R. Potts. 2002. Origin of reddened and melted zones in African Geographer/Suid-Afrikaanse geograaf 19:22–34. Pleistocene sediments of the Olorgesailie Basin, southern Kenya Rift. Sergant, J., P. Crombe, and Y. Perdaen. 2006. The “invisible” hearths: a Journal of Archaeological Science 29:307–316. contribution to the discernment of Mesolithic non-structured surface hearths. Parker, C. H., E. R. Keefe, N. M. Herzog, J. F. O’Connell, and K. Hawkes. 2016. Journal of Archaeological Science 33:999–1007. The pyrophilic primate hypothesis. Evolutionary Anthropology 25:54–63. Stott, R. 2012. Darwin’s ghosts: in search of the first evolutionists. London: Perlès, C. 1977. La préhistoire du feu. Paris: Mason. Bloomsbury. Pickering, T. R. 2012. What’s new is old: comments on (more) archaeological Tutin, C. E. G., L. J. T White, and A. Mackangamissandzou. 1996. Lightning evidence of one-million-year-old fire from South Africa. South African Jour- strike burns large forest tree in the Lopé Reserve, Gabon. Global Ecology nal of Science 108, 5/6, doi:10.4102/sajs.v108i5/6.1250. and Biogeography Letters 5:36–41. Preece, R. C., J. A. J. Gowlett, S. A. Parfitt, D. R. Bridgland, and S. G. Lewis. Vale, T. R., ed. 2002. Fire, native peoples and the natural landscape. Wash- 2006. Humans in the Hoxnian: habitat, context and fire use at Beeches Pit, ington, DC: Island. West Stow, Suffolk, UK. Journal of Quaternary Science 21:485–496. White, Randall, Romain Mensan, Amy E. Clark, Elise Tartar, Laurent Marquer, Preece, R. C., S. A. Parfitt, D. R. Bridgland, S. G. Lewis, I. Candy, H. I. Raphaëlle Bourrillon, Paul Goldberg, Laurent Chiotti, Catherine Cretin, Griffiths, J. E. Whittaker, and C. Gleed-Owen. 2007. Terrestrial environ- William Rendu, Anne Pike-Tay, and Sarah Ranlett. 2017. Technologies for ments during MIS 11: evidence from the Palaeolithic site at West Stow, the control of heat and light in the Vézère Valley Aurignacian. Current Suffolk, UK. Quaternary Science Reviews 26:1236–1300. Anthropology 58(suppl. 16):S288–S302. Pruetz, J. D., and T. C. LaDuke. 2010. Reaction to fire by savanna chim- Wrangham, Richard. 2009. Catching fire: how cooking made us human. New panzees (Pan troglodytes verus) at Fongoli, Senegal: conceptualization of York: Basic. “fi re behavior” and the case for a chimpanzee model. American Journal of ———. 2017. Control of fire in the Paleolithic: evaluating the cooking hy- Physical Anthropology 141:646–650. pothesis. Current Anthropology 58(suppl. 16):S303–S313. Current Anthropology Volume 58, Supplement 16, August 2017 S217

Ethnoarchaeology of Paleolithic Fire Methodological Considerations

by Carolina Mallol and Auréade Henry

Most of the ethnoarchaeological literature on hearths is scattered within general works that target many different aspects of foraging or hunter-gatherer societies. Although these works are a good source of ideas and clues for the interpretation of macroscopically observable features of Paleolithic hearths, there is hardly any high-resolution ethnoarchaeological reference material with which to compare microstratigraphic evidence of archaeological fire. Our ethnoarchaeological research at this scale has focused on exploring differential preservation of open-air hearths and the potential to identify fire-related activities and different variables of fire technology (fuel, temperature, and function) using micromorphological and anthracological analysis. Although these studies have been useful sources of analogy, further case studies as well as ethnoarchaeological examples of superposed and imbricated hearths and reference material from enclosed settings such as caves and rock shelters are strongly called for. In this paper we summarize and discuss aspects of our previous work to highlight the strengths and weaknesses of the ethnoarchaeological approach for the study of Paleolithic fire and propose possible avenues for future research on the topic.

Fire has played a fundamental role in the biological and social disciplines (particularly geoarchaeology, archaeobotany, and evolution of humankind. Omnipresent in the domestic and zooarchaeology) have contributed valuable data to advance our ritual spheres since Paleolithic times, it has been influential for knowledge of Paleolithic fire and fire-related activities (Aldeias our diet and associated with the emergence of technological, 2017; Alperson-Afil 2017; Costamagno et al. 2009; Hlubik et al. sociological, and artistic expressions in different past societies 2017; Holdaway, Davies, and Fanning 2017; Mentzer 2014; (Perlès 1977). However, despite its potential to furnish valu- Théry-Parisot, Chabal, and Costamagno 2010). able behavioral information on our most distant past, anthro- Among these approaches, bioarchaeological analyses (in the pogenic fire has been an elusive topic in prehistoric research. sense of analysis of organic materials) of combustion remains— Most researchers have focused on trying to establish the pres- commonly charcoal, but also bone or phytoliths—provide ence or absence of fire in Paleolithic contexts in order to as- valuable taphonomic data in relation to fuel management and certain the timing of fire control and use, rarely examining the hearth functionality, thus allowing us to better understand en- evidence in order to better understand human lifeways. vironmental and economic aspects of past societies (Théry- Nevertheless, there is a growing number of studies on ar- Parisot, Costamagno, and Henry 2009). Similarly, geoarchae- chaeological combustion structures, combustion residues, and ological analyses of hearths or combustion structures, which are other elements of the archaeological record bearing traces of in fact sedimentary artifacts, contribute similar kinds of in- burning. Such archaeological remains convey not only socio- formation in addition to an understanding of the hearths’ economic, dietary, and paleoenvironmental information but formation processes (Goldberg, Miller, and Mentzer 2017; also clues to site formation and taphonomy. All of these Mallol, Mentzer, and Miller, forthcoming; Mentzer 2014). aspects are central to Paleolithic research. Accordingly, recent With increasing studies of anthropogenic fire in Paleolithic methodological approaches from different archaeological sub- archaeology, it has become relevant to incorporate this topic into the ethnoarchaeological research agenda. Some archaeo- logical subdisciplines, such as zooarchaeology, have incorpo- Carolina Mallol is Ramón y Cajal Researcher in the Departamento de rated ethnoarchaeological research into their interpretive Geografía e Historia of the Universidad de La Laguna (Campus de framework (Abe 2005; Costamagno and David 2009; Kent Guajara, 38071, La Laguna, Tenerife, Spain) and in the Instituto Uni- 1993; Monahan 1998; Sázelová et. al. 2015; Svoboda et. al. versitario de Bio-Orgánica Antonio González (Avenida astrofísico Fran- 2011; Waguespack 2002). Over the years, ethnoarchaeological cisco Sánchez, 2, 38206, La Laguna, Tenerife, Spain [[email protected]]). research has also incorporated a microscopic scale of obser- Auréade Henry is a Postdoctoral Researcher at Centre National de la Recherche Scientifique, Unité Mixte de Recherche (CNRS-UMR 7264 vation, as in the study of ethnographic artifacts as reference CEPAM, Université Côte d’Azur, CNRS, Nice, France [aureade.henry data sets for archaeological use-wear analyses (Beyries 1995; @cepam.cnrs.fr]). This paper was submitted 25 VII 16, accepted 3 II 17, González-Urquijo, Beyries, and Ibáñez 2015; Mansur 1983). and electronically published 16 V 17. Despite these efforts, the ethnoarchaeological fire record is q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0006$10.00. DOI: 10.1086/691422 S218 Current Anthropology Volume 58, Supplement 16, August 2017 understudied, and the reason might be that fire residues are published ethnographic/ethnoarchaeological works. Besides predominantly sedimentary in nature (ash and minute charred Binford’s work, the ethnographic/ethnoarchaeological litera- residues embedded in sediment) and the study of the sedi- ture containing information on fire is mostly scattered within mentary context in ethnoarchaeology is relatively recent, with general works about the daily life of particular societies, and only a few published case studies (see Friesem 2016 for a re- very few address technical issues of fuel selection and fire view on this topic). management. Nevertheless, cultural-ecological studies including In this paper we first briefly review and discuss the role of fire as part of both symbolic and economical practices commonly fire in ethnoarchaeological research applied to Paleolithic ar- provide examples of fuel type selection and management, fire chaeology and then summarize and discuss two examples of functionality and techniques used in cooking fires, duration, our previous research on fire in ethnoarchaeological contexts. spatial distribution, and the entity of group members in charge Our goal is to highlight the kinds of questions that can be of fire in a number of published studies on such topics (e.g., addressed by studying the ethnoarchaeological record of an- Brandisauskas 2007, 2010; Gur-Arieh et al. 2013; Lavrillier 2005; thropogenic fire and possible pitfalls. We also set forth a few Osgood 1970 [1936]; Picornell Gelabert, Asouti, and Allué methodological guidelines that, based on our experience in Martí 2011; Sillar 2000). geoarchaeology and anthracology applied to ethnoarchaeolog- Other factors—such as burning temperatures, hearth re- ical contexts, may aid in procuring useful ethnoarchaeological lighting, reworking of ashes, or charcoal dispersal and micro- data toward an understanding of archaeological combustion anatomy—are less present in ethnoarchaeological accounts structure formation processes. (with exceptions such as Ntinou 2002; Zapata Peña et al. 2003). Such factors, which are particularly significant in the interpre- fi Ethnoarchaeology and Paleolithic Archaeology: tation of Paleolithic re because they can be approached The Role of Fire through the archaeological record, often emerge from questions raised by the study of the sedimentary archaeological record or Why have the hearths of contemporary traditional societies the bioarchaeological record (i.e., charcoal and other organic not been studied in-depth to advance Paleolithic research? residues) at microstratigraphic/microscopic scales of observa- Ethnoarchaeological data have been an important source of tion, which are generally overlooked in ethnoarchaeological information in Paleolithic archaeology, from Binford’s ex- contexts. haustive documentation on the Nunamiut of northcentral As pointed out by Goldberg and Macphail (2008) and more Alaska (Binford 1978) to other referential studies on hunter- recently by Friesem (2016), geoarchaeology and bioarchaeology gatherer societies around the globe (e.g., Beyries and Pétrequin did not incorporate ethnoarchaeological investigations into their 2001; Beyries and Vaté 2007; Brooks and Yellen 1987; Fewster agendas until very recently. The applicationof geoarchaeology to and Zvelebil 2001; González-Ruibal, Hernando, and Politis ethnoarchaeological contexts, or geo-ethnoarchaeology (a term 2011; Kelly 1995; Leroi-Gourhan and Leroi-Gourhan 1989; coined by Brochier et al. 1992), is a relatively young field with Lim 1985; O’Connell 1987; Peterson 1971, 1973; Pétrequin great potential for advancing our understanding of formation 1988). Traditional ethnoarchaeological data concern material processes of archaeological sites and features. So far, there are and spatial aspects of human behavior analyzed with a holistic few exploratory geo-ethnoarchaeological studies, and these have approach in order to understand the link between material focused on the micromorphological and geochemical identifi- culture and social structure and beliefs (Binford 1980:5). cation of activity area sedimentary indicators (e.g., Anderson Concerning fire, Binford set forth hypotheses about hearth et al. 2014; Brochier et al. 1992; Shahack-Gross, Marshall, and function, such as in his interpretation of certain archaeological Weiner 2003; Shahack-Gross et al. 2004; Wattez 1992), post- combustion features as smudge pits based on ethnoarchaeo- depositional processes affecting household elements (Friesem logical observations (Binford 1967). He also used ethnoar- et al. 2011, 2014a, 2014b; Goldberg and Whitebread 1993), and chaeological data to approach areas of human activity and to the sedimentary manifestation of different kinds of open-air identify some of the factors of archaeological assemblage for- hearths (Mallol et al. 2007). mation (e.g., Binford 1978). For instance, ethnoarchaeology- Within the field of fuel analysis, a few studies have under- based models such as Binford’s “hearth-related assemblage,” taken investigations of ethnoarchaeological contexts related to “toss and drop zone” (Binford 1978), or collector/forager models fire by focusing on different types of fuel, mainly dung and (Binford 1980) have been widely used to interpret intrasite ar- wood. Only the most recent ethnoarchaeological studies on the chaeological spatial distribution patterns and to infer intersite use of dung among traditional societies include sampling pro- Paleolithic settlement dynamics, respectively. Actually, the study tocols that aim at producing reference data sets for identifying of settlement dynamics elements such as territorial mobility or the use of dung as a fuel within archaeological cooking struc- site type has relied heavily on ethnoarchaeological information tures (Gur-Arieh et al. 2013; Lancelotti and Madella 2012; Lan- (e.g., Binford 1980; Brooks and Yellen 1987; Cameron and celotti, Ruiz-Pérez, and García 2016). Tomka 1996; Grøn 2005; Kuznetsov 2007). Regarding wood fuel, even though the potential benefits of On the whole, it is generally difficult to obtain detailed incorporating ethnoarchaeology into charcoal analysis were information about fire-related practices based on the existing pointed out more than 20 years ago (Chabal 1994), “ethno- Mallol and Henry Ethnoarchaeology of Paleolithic Fire S219 anthracology” (Henry 2011; Henry, Théry-Parisot, and Vo- fissuring, and soil organic matter carbonization) in days-old to ronkova 2009) has only been applied to very few ethnographic months-old open-air hearths made by a group of Hadza charcoal assemblages (Henry and Théry-Parisot 2014a; Joly foragers (Tanzania) and documented ethnographically. To et al. 2009; Ntinou 2002; Vidal-Matutano 2013). The aims of this end, the ethnographers collected a series of undisturbed such studies have been to assess the paleoecological accuracy of micromorphology sediment blocks from abandoned open-air, charcoal analysis and/or to explore the extent to which different cooking, and sleeping hearths (fig. 1). Some of the hearths were methods are able to identify human practices. recent (days old), while others had been abandoned for 1 year. In sum, although there have been ethnoarchaeological ap- Regarding the duration of these fires, some of them were proaches to Paleolithic fire since the 1970s, these have rarely continuously used for 3–4 months, while others were brief fires focused on combustion residues, which are direct transmitters (less than an hour in duration) for roasting food items. of environmental and behavioral information. In recent years, The results of this study, detailed in Mallol et. al. (2007), with the prominence of interdisciplinary studies in archaeology, included field and micromorphological observations on the the anthropogenic combustion record in the ethnoarchaeolog- sedimentary substrate associated with the abandoned hearths. ical context is starting to be queried, and Paleolithic archaeology At a microscopic scale, all the samples, including those from can largely benefit from such a source of analogy. hearths that had been abandoned for a year, yielded com- bustion residues (wood ash, charcoal, and charred plant/ani- fi Research Avenues for the Ethnoarchaeological Study mal tissue) attesting to the presence of anthropogenic re as of Fire: Lessons from Previous Case Studies well as microstructural sedimentary features such as matrix disaggregation and browning or “masked birefringence.” Over- Given an interest in hearths and hearth-assemblage formation all, the micromorphological components and features observed processes, we aim at approaching some of the factors involved in are in agreement with those proposed by Wattez (1992) as the formation of archaeologically observable features formed representative of moderate to high intensity hearths involving throughout the depositional and postdepositional history of a temperatures from 3507Cto15007C. hearth. Such features may include thermally altered sediment, Certain micromorphological differences relating to function artifacts and bone fragments, 1–5 cm layers of ash or carbo- were observed among the different types of hearth. For instance, naceous matter, and concentrations of charcoal fragments of the sedimentary substrate of a brief fire made to roast an im- different sizes. In practice, this is no simple task. The ethno- pala yielded a few amorphous black impregnations and coatings archaeological context is complex and might lend itself to infi- (fig. 2A), and a tuber-roasting fire left behind microscopic nite query. Also, every case study is unique in the kinds of hu- charred plant tissue fragments (fig. 2B). Before this study, it was man actions performed around fire as well as in the nature and unknown whether or not specific activities related to anthro- condition of the local sedimentary substrate. Thus, it is likely pogenic fire might leave microscopic material evidence in the that specific questions and analytical parameters arise not only sediment. Unfortunately, these particular hearths were sampled in the beginning but also during the course of the investigation only days after they were made, and we do not know the pres- once the researchers are familiarized with the site. In the case of ervation potential of the observed features. There were also ethnoarchaeological fieldwork for the study of hearths, famil- micromorphological differences relating to taphonomic factors. iarization with the site includes gaining basic knowledge of the The hearths sampled 1 year later showed presence of ash only in sedimentary context, that is, the substrate of anthropogenic fire. a case where a layer of dry grass from a dismantled hut had We point this out because ethnographic research does not covered the hearth. They also showed signs of bioturbation usually focus on sedimentological descriptions. (channels and fresh rootlets dissecting the top of the combustion The motivation underlying ethnoarchaeological research de- structure; see fig. 2C), whereas those that were days or months sign may be (1) to test existing behavioral or taphonomic in- old did not (fig. 2D). Interestingly, bioturbation did not affect terpretations derived from archaeological or experimental data, the sedimentary fabric to the extent of precluding identification which will likely involve specific questions and analytical pa- of combustion features. rameters, or (2) to document the material expression of par- The study also showed that micromorphology is a powerful ticular human actions and their modification through time, tool to approach the genesis of sedimentary deposits through a which may lead to open-ended, exploratory studies. In the very peculiar finding: one of the samples showed micromor- following paragraphs, we provide two examples to illustrate phological features indicative of the presence of an abandoned both kinds of motivation and highlight some of the results and hut floor beneath one of the open-air hearths. The previous implications for each case. existence of a hut at the spot of that particular hearth had not been reported ethnographically. This finding shows that geo- ethnoarchaeology performed at a microstratigraphic scale of The Hadza Study on Open-Air Hearths observation adds a temporal dimension, thus contributing his- The Hadza study (Mallol et al. 2007) was designed as a test of the torical information. This contribution may help alleviate a short- visibility of previously established micromorphological features coming of ethnoarchaeology pointed out by Wobst (1978): “The associated with anthropogenic fire (such as browning-reddening, ethnographic record is insufficiently sensitive to deal with be- S220 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 1. Different types of Hadza hearths subjected to micromorphological analysis. A, Brief, 20-minute-long fire used to burn an impala and sampled 10 days later. B, Cooking hearth used recurrently for 4 months and sampled a year later. C, Abandoned sleeping hearth at the entrance of an abandoned hut. It was lit every night for 4 months, and the sample was collected a year after aban- donment. Note the layer of dry grass over it. D, Communal cooking hearth used continuously for 3 months and sampled 2 months later. E, Brief (15 minutes) tuber-roasting fire that was sampled the day after. A color version of this figure is available online. havioral variability. The living social context is complex and we particularly influential, as we saw that the ash layer of a can only attempt to perceive a limited range of its material ex- hearth that had been protected by a light grassy cover was still pressions.” By incorporating historical information, we broaden intact after a year in the open-air while those of exposed hearths the scope of possibly interrelated material expressions. had disappeared, or that the ash of a 10-day-old hearth dissi- Overall, our results provided an example of what different pated after a single rainfall event. The Hadza example also kinds of simple anthropogenic fires might leave behind in showed that even though the ash component is likely to erode ethnographic contexts after up to a year’s time. In this regard, away with time in cases of low sedimentary rates and deflation, the results suggest that rates of sedimentation and the effect the irreversible effects of fire in the top 2 cm of the sedimentary of postdepositional disturbance factors such as bioturbation substrate (charring of soil organic matter and clay alteration) or erosion are key factors in the preservation of open-air may remain intact and be readily identified through micro- simple hearths. Erosion by rain and deflation seemed to be morphology. Mallol and Henry Ethnoarchaeology of Paleolithic Fire S221

from the Hadza hearths, as we would not expect long-term preservation of ash in such simple open-air combustion features otherwise. The amount of ash reported for the in situ hearth at Nesher Ramla is quite small, suggesting either some period of deflation or erosion from rain before burial. Re- garding the charred substrate, the presence of blackened soil aggregates suggests a more organic-rich substrate than what we documented in the Hadza study, which was practically barren land with a light grass cover. In sum, the Hadza study showed the value of performing high-resolution, microstratigraphic investigations of the ethno- archaeological sedimentary record to distinguish between nat- ural and anthropogenic fire, to estimate burning intensities, to identify fuel types, and to assess preservation potentials. In hindsight, the study would have yielded more detailed results if we had included the following items in the research design: (1) the presence of a geoarchaeologist on site to enrich the eth- nographic record with data relevant for the study of site for- mation, and (2) the implementation of techniques comple- mentary to soil micromorphology. Current geoarchaeological combustion structure investigations are normally carried out Figure 2. Microphotographs from thin sections of different from a microcontextual perspective using interdisciplinary Hadza hearth samples. A, Detail of the topsoil in a brief impala- microstratigraphic methods (see examples in Mentzer 2014) burning hearth sampled after 10 days showing thin black coatings and involving techniques from inorganic and organic geochem- and cappings on some of the rock fragments. B,Detailofthetop- soil in a brief tuber-roasting hearth sampled after 1 day showing istry and geophysics, including, for example, organic petrol- fi the presence of charred plant tissue fragments (black particle at ogy for identi cation of microscopic charred particles, Fourier the center and smaller ones around it). C, Topsoil of a 1-year-old Transform Infrared Spectrometry for mineral identification, cooking hearth. Note the compact appearance and presence of gas chromatography mass spectrometry for lipid analysis, or rootlets. D, Topsoil of the impala-roasting hearth. Note its loose archaeomagnetism to approach firing temperatures and tem- appearance and lack of rootlets, channels, or other bioturbation features. A color version of this figure is available online. poral relationships between different hearths. Applying such techniques in the Hadza study might have strengthened the reliability of such techniques to provide data on thermal al- teration of the substrate and to identify combustion residues. All these observations can be useful in the interpretation of Finally, in order to further expand the results obtained in the Paleolithic fire evidence. Unfortunately, micromorphological Hadza study toward an understanding of Paleolithic com- studies of Paleolithic open-air combustion features are scarce. bustion structures, the results need to be tested against ex- The bulk of data was gathered by Wattez in the 1990s and is perimental taphonomic data on abandoned open-air hearths associated with Upper Paleolithic and Mesolithic open-air and also against micromorphological data from ethnoar- hearths, which are generally more complex, involving stone- chaeological examples of hearths in enclosed settings (caves or lined or stone-filled pit structures (e.g., Wattez 1992, 1994). rock shelters) as well as examples of superimposed and im- Therefore, there is not much data allowing us to make anal- bricated hearths, which are common in the Paleolithic fire ogies with the results of our study of simple Hadza hearths. record. No such data are currently available. One case is the Middle Paleolithic open-air site of Nesher Ramla, Israel, where an in situ simple combustion structure The Evenk Study of Specialized Open-Air Hearths and a wood ash midden were reported by Friesem, Zaidner, and Shahack-Gross (2013). The sediments from the com- The Evenk research program was first designed as a case study bustion structure showed a thin black layer composed of to test a model proposed for the European Paleolithic ac- blackened sediment aggregates overlain by wood ash residues cording to which fuel management is a complex system re- and calcined bone fragments (Friesem, Zaidner, and Shahack- sulting from a series of interacting environmental and cultural Gross 2013). parameters (Théry-Parisot 2001). An exploratory study was According to the authors, preservation of these features carried out among a group of Evenk reindeer herders and was possibly enhanced by their location in a topographic hunters from the Amur Region (South Eastern Siberia) in depression sheltered from the wind and receiving regular order to (1) observe firewood management practices from colluvial input, which would have buried the intact features. wood acquisition to the discard of fuel residues; (2) observe This interpretation is in agreement with our observations possible connections between the environment (available S222 Current Anthropology Volume 58, Supplement 16, August 2017 biomass), human activities, and fuel management; (3) assess evidence in archaeological context is infrequent, and we can which of these observations could be evidenced by the study of assume that identifying the presence of residual rotten wood in anthracological remains and sedimentary micromorphologi- an archaeological version of this hearth would require cal signatures; and (4) explore the possibility of tracing anal- investigations at a molecular scale. The same can be said about ogies between current nomadic lifeways in cold environments the substrate, which consisted exclusively of fresh plant litter and Paleolithic contexts (Henry 2011; Henry and Théry- without any apparent effects from the overlying combustion. Parisot 2014a; Henry, Théry-Parisot, and Voronkova 2009). This is explained by the way in which the fire was made: a small Fieldwork among the Evenks was carried out in two seasons, quantity of charcoal from sound wood was produced in a furnace one in late winter/early spring, and another in late summer/early elsewhere and redeposited on the ground at the hide-smoking autumn. Throughout the study, we observed that Evenk tra- spot, where it was then covered by rotten, crumbling wood. ditions are strongly influenced by seasonality and residential Interestingly, our microscopic observations suggest that sim- mobility. These factors have a decisive effect on the range and ilar hide-smoking hearths might have been previously made at configuration of activities performed at their camps (which in the same place, as indicated by the presence of rotten wood— archaeological terms would determine the site’sfunction).This an intrusive element brought by humans—contained within the observation also applies to combustion structure types, their plant litter substrate. As in the Hadza case, micromorphological function, and firewood management, with wood procurement analysis yielded information on past events. As we know from areas and modalities varying according to the time of year the Evenks we worked with, they normally return to their main (Henry 2017; Henry, Théry-Parisot, and Voronkova 2009). In campsites yearly, at least for periods of 4–5 years, and light their turn, combustion structure types and functions determined the fires at the same spots as in previous occupations. species and state of the selected wood. In Paleolithic research, In sum, we can predict that a possible archaeological ex- seasonal constraints have mainly been approached through pression of a hide-smoking combustion structure of the kind subsistence studies. By showing that a whole range of activity studied here, after many years of subaerial exposure, would areas—firewood management included—are strongly tied to exhibit a very weak sedimentary signature consisting of a diffuse seasonal and climatic conditions, this example invites us to ex- layer of charcoal resting on an unburned sedimentary substrate. pand our approach of seasonality, prehistoric settlement pat- Regarding our anthracological results, both hearths (smudge terns, and mobility. fire and hide-smoking hearth) were positively discriminated Furthermore, the study confirmed the positive correlation and allowed us to set forth a new method to diagnose the initial between the degree of fuel selectivity and of hearth speciali- soundness of firewood based on the identification of micro- zation implicitly suggested in earlier anthracology works morphological fungal alteration intensities observed on charcoal (Chabal 1982, 1991) as well as the importance not only of the fragments (Henry and Théry-Parisot 2014a; fig. 5). This method taxon but also of the state of the wood in the wood selection has proven to be effective for discussing fuel management strat- process (Théry-Parisot 2001). In turn, the implications for egies evidenced in Paleolithic and Mesolithic hunter-gatherer prehistoric anthracology are that the identification not only of sites (Henry and Boboeuf 2016; Henry and Teten’kin 2014; the taxon but also of the state of the wood used is crucial for the Vidal-Matutano, Henry, and Théry-Parisot 2017). characterization of prehistoric wood procurement modalities, These results have several implications for Paleolithic anth- combustion behavior, and the specialized/seasonal nature of racology because they evidence the nature of the link between the hearths (see also Henry and Théry-Parisot 2014b; Théry- environmental conditions, lifeways, and modalities of fire use. Parisot and Texier 2006). As in other research domains, comparisons with other ethno- Finally, the relationship between two kinds of fireplaces and graphic settings point at the great variability of practices and anthracological/micromorphological signatures were explored beliefs around fire. Nevertheless, regularities also exist between during the second field season, in which our Evenk hosts were different groups, revealing similar choices under comparable kind enough to allow us to collect charcoal from a smudge fire environments (Henry 2011). For example, the use of crumbling, and to excavate and sample their hide-smoking hearth for rotten wood for smoking hides is ubiquitous among northern charcoal and micromorphological analyses (fig. 3). We exca- hunter-gatherers living in forest environments from Siberia to vated half of it, collecting anthracological samples, and saved North America (Alix and Brewster 2004; Anikhovskij et al. 2012; the other half for micromorphological sampling, which re- Beyries 2002, 2008; Binford 1967; Brandisauskas 2007, 2010; quired dismantling of the hearth. Henry, Théry-Parisot, and Voronkova 2009; Lavrillier 2007; Micromorphological observation of three thin sections from Nelson 1986; Osgood 1970 [1936]). The prospect of being able the hide-smoking hearth, which produced no ash, showed that to discuss archaeological hearth functions thanks to new de- the combustion deposit consisted of a mix of charcoal and 1– velopments in ethnoanthracology is particularly exciting and 3 cm fragments of rotten wood without any visible signs of motivates new archaeoanthracological studies geared in this burning on a bed of relatively fresh plant litter (fig. 4). A few direction. One such study involves a hypothetical Middle rotten wood fragments were also observed embedded in the Paleolithic smoking hearth based on archaeoanthracological plant litter layer (fig. 4A). Rotten wood has a low preservation data (Vidal-Matutano 2016; Vidal-Matutano, Henry, and Théry- potential under aerobic conditions. Thus, encountering such Parisot 2017). Figure 3. Summary of the ethnoanthracological work with Evenki from Ulgen, Amur Region. Photos: Auréade Henry, ACI “Système Renne.” A color version of this figure is available online. S224 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 4. A, Scan of thin section from micromorphological sample collected from an Evenk hide-smoking hearth. Note the presence of rotten wood fragments (lighter colored) at the top but also embedded within a plant litter deposit. B, C, Microphotographs from the same thin section and two others showing a mixed presence of charcoal (dense, black particles) and rotten wood fragments (yellowish-brown particles) on a pristine bed of plant litter (lower half of image in A). D, Rotten wood fragment. Note the presence of fecal pellets (small granules filling gaps between healthy portions of tissue). E, Detail of rotten wood fragment showing deformed cell walls due to fungal attack. Photos taken in Plane Polarized Light; E taken in Blue Light (400–440 BL). A color version of this figure is available online.

The downside of the Evenk study is that the limited amount parable to Paleolithic counterparts. Among the things to of hearth samples is insufficient to fully evaluate the potential consider are planning for the implementation of interdisci- and limitations of our results. Also, as with the Hadza study, plinary methods to test the preservation state of the target applying complementary geoarchaeological techniques would features before intervention. In this section, we provide a few have allowed for richer, more robust data. In the future, many more specific and practical guidelines that may facilitate the more hearths need to be sampled in order to obtain a stronger research design. reference data set and establish a protocol that incorporates multiproxy geoarchaeological and bioarchaeological data. 1. Compile ethnographic data and assess their quality. What A final observation is that studying hearths that are still in is the potential of this data to address archaeologically use has many advantages, because complete operational chains relevant questions regarding anthropogenic fire? When a (chaînes opératoires) of specific activities such as meat curing or pertinent research context has been defined, it is equally hide smoking can be documented as well as their timing and important to assess the marginality of the recorded activity duration, fuel amounts and procurement distances, and other or process: is this type of behavior recurrent today among such variables involved in fire technology. However, one dis- traditional societies and at what scale? What other related advantage is that sampling can be difficult or even impossible social and environmental domains need to be described? because it involves destruction of a structure within or outside 2. In the case of abandoned contexts, which may be exposed the perimeter of ongoing domestic activity. Some of these struc- or buried, assess the degree of integrity of the record tures may have a strong cultural or symbolic meaning for the through surveys and test pits. For each proxy, control community. samples from nearby natural deposits or activity areas should also be collected to validate the diagnostic value of Methodological Guidelines the results. 3. Perform excavation and collect sediment samples for The two examples presented here convey a series of aspects to multiple microstratigraphic analytical techniques. Unlike take into consideration when designing an ethnoarchaeo- other ethnoarchaeological remains, such as exposed re- logical project to study contemporary simple hearths com- mains, combustion structures representing open hearths Mallol and Henry Ethnoarchaeology of Paleolithic Fire S225

Figure 5. Differences in the proportions of microscopic fungal degradation features on charcoal samples according to the initial soundness of the wood used in the fire. Photos: Auréade Henry, ACI “Système Renne.” The SEM photos were taken for A. Henry by Monique Repoux, Centre de Mise en Form des Materiaux, Sophia Antipolis, France. A color version of this figure is available online.

always require excavation because of their sedimentary ject. Keep in mind that different analytical tech- nature. Excavation and sampling may be performed fol- niques (e.g., soil micromorphology, organic and lowing a protocol for archaeological combustion features. inorganic geochemistry, archaeomagnetism, anal- We propose to follow these steps (fig. 6): ysis of phytoliths, spherulites, and pollen if appli- cable) require different specific sampling protocols. a. Determine the perimeter and plot it within a co- d. Collect control sediment samples (or hand spec- ordinate system. imens in the case of macrobotanical studies) from b. Divide the feature in two equal portions and excavate outside the combustion structure for each of the half of it, leaving the other half for microstratigraphic analytical techniques. sampling. Excavate following stratigraphic layers and place the sediment from different layers in dif- Discussion: Some Pros and Cons of ferent bags. If possible, collect all the sediment for Ethnoarchaeological Research to future studies on fuel loss, ash yield, etc. This sedi- Approach Paleolithic Fire ment will be subsequently sorted to recover different kinds of macroremains (charcoal, seeds, bone frag- Ethnoarchaeology may be viewed as a polemic field of research ments, microfauna, etc.). In cases of occurrence of because at first it seems rather impossible, if not dangerous and anthropogenic bone remains other than fuel or ar- unscientific, to use ethnographic data for the interpretation of tifacts, these should be plotted as material remains archaeological evidence. Societies are complex dynamic systems within the coordinate system, including their depth. fashioned by multiple interrelated factors across space and time. c. While collecting samples from the second half, Some authors are pessimistic about the potential of ethno- keep all the remaining sediment and piece-plot any archaeology unless it is carried out through an ontological ap- artifact, bone, or other visible anthropogenic ob- proach (González-Ruibal, Hernando, and Politis 2011). S226 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 6. A, Excavation of a recent cave entrance hearth. Note that the excavator is reserving half of the combustion structure for sampling. The location of two micromorphological samples in the other half has been marked with nails and thread. B, Archaeo- magnetic sampling of half of a 5-year-old combustion structure. A color version of this figure is available online.

In practice, this means that in the study of a particular aspect mation processes and the nature of our archaeological remains of human behavior, such as behavior around fire, researchers (including sedimentary residues). need to adopt a broad perspective and include other behavioral The challenge lies in achieving ethnoarchaeological research domains in order to approach more general aspects of societal design in a way that meets the standards of current archaeo- and/or environmental value. In other words, anthropogenic logical science. There are very few general works on ethno- fire, as any other element of human social behavior, must be archaeology (e.g., David and Kramer 2001) that include meth- considered within its social and environmental context. Thus, odological guidelines geared at multidisciplinary research and ideally, the ethnoarchaeological research team involved in a the incorporation of subdisciplines such as bioarchaeology and study of fire behavior should include experts from different geoarchaeology. Despite this, the existing literature reveals that fields documenting as many aspects of culture and environ- ethnoarchaeology may truly benefitmostfields of prehistoric ment as possible. research as long as two main targets are pursued: (1) obtaining Another polemic issue is the validity of ethnographic anal- contextualized ethnographic data through interdisciplinary ogy. Present-day observations should not be directly transposed approaches and (2) carrying out research motivated either by to the past. Caution should be taken when using contemporary archaeological facts and questions or by a need to obtain ref- societies to approach distant spatio-temporal contexts (e.g., erence material for the material residues of specific human comparing present-day subtropical agricultural societies with activities or lifestyles. Paleolithic hunter-gatherers of the Northern Hemisphere) or According to our personal experience in the field of the tracing analogies that may seem straightforward (e.g., Bronze ethnoarchaeology of fire, it has become obvious that ethno- Age vs. present-day traditional pastoral societies of the same archaeology is much more than a “real-size experiment.” Ethno- area). In all cases, it is important to bear in mind that “the true archaeology not only improves our reference data sets (and in role of ethnoarchaeology is not to provide the prehistorian this sense, it is complementary to experimentation) but also with analogical tidbits, but rather to be an important source for provides an enriching way of approaching the archaeological those wanting to build theoretical models for the relationships record through the possibility of apprehending the diversity of between people and things” (Skibo 2009:47). As shown by our environmental and societal variables and their effect on the own work, ethnoarchaeology also provides insight into site for- material record. Mallol and Henry Ethnoarchaeology of Paleolithic Fire S227

The Hadza and Evenk studies are good examples of how and enriched by geoarchaeological and bioarchaeological data ethnoarchaeological research may generate new knowledge from additional techniques. It is to be hoped that further re- that is complementary to archaeology and experimentation. In search of this kind will be carried out in the coming years and the Evenk case, while showing that the archaeological model that it will contribute data to aid in building robust reference was easily adaptable to any fuel management study, our ob- data sets for our understanding of Paleolithic hearth and hearth- servations also revealed complex patterns of human behavior assemblage contexts through ethnographic analogy. toward firewood, which could only be understood in regard to the social, economical, and seasonal background of the study. Acknowledgments In return, these results provided us with new ideas for inter- pretative pathways of prehistoric fuel selection and hearth We thank the Wenner-Gren Foundation and the organizers of function in relation to site function and seasonality. the “Fire and the Genus Homo” symposium for their kind in- The challenge was then to test the potential of archaeo- vitation to participate, and we thank our collaborators in the metric methods to document socioeconomic aspects through Hadza and Evenk fire projects (funded by NSF grant 0242455 to the study of their material expression: combustion structures. Frank Marlowe, the American School of Prehistoric Research at The positive results of the ethnoanthracological analysis, which Harvard University, and the French Research Council project have been subsequently validated by substantial experimental ACI TTT, “Adaptations biologiques et culturelles: Le Système replication (Henry and Théry-Parisot 2014a), opened up new Renne,” funded by the French Ministry of Research, to Sylvie methodological perspectives on fuel selection, hearth function, Beyries, Centre National de la Recherche Scientifique, Unité and even seasonality. Mixte de Recherche, 7264 Cultures et Environnements Préhis- In sum, ethnoarchaeology is complementary to but not re- toire, Antiquité, Moyen Âge). We would also like to give par- placeable by experimentation, because functional and mean- ticular thanks to Frank Marlowe, Claire Porter, Brian Wood, ingful anthropogenic deposits or sedimentary features are dif- Evgenja Zavadskaja, and especially the Nikiforovs and the ficult if not impossible to produce outside their traditional Struchkovs, who not only let us sample their hearths but opened context. On the other hand, experimentation can and should be their homes and their hearts to us. used to further investigate and validate ethnoarchaeological results before testing their applicability to archaeological con- texts. At a broader, systemic scale, ethnoarchaeology of fire is a References Cited powerful tool to evaluate the variability of human adaptations to Abe, Y. 2005. Hunting and butchery patterns of the Evenki in northern the natural environment through the establishment of causal Transbaikalia, Russia. PhD dissertation, Stony Brook University, State fi University of New York. relationships between culture and re technology. Aldeias, Vera. 2017. Experimental approaches to archaeological fire features and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– S205. Conclusions Alix, C., and K. Brewster. 2004. Not all driftwood is created equal: wood use and value along the Yukon and Kuskowim Rivers, Alaska. Alaska Journal of This holistic, multidisciplinary approach, which allows us to Anthropology 2(1):2–19. Alperson-Afil, Nira. 2017. Spatial analysis of fire: archaeological approach to take into account many different environmental and mate- fi – fl recognizing early re. Current Anthropology 58(suppl. 16):S258 S266. rial and nonmaterial parameters in uencing the nature of the Anderson, D. G., E. M. Ineshin, N. V. Kulagina, M. Lavento, and O. P. anthropogenic combustion remains, aims at contributing to Vinkovskaya. 2014. Landscape agency and Evenki-Iakut reindeer hus- the development of new analytical methods for the study of bandry along the Zhuia River, Eastern Siberia. Human Ecology 42(2):249– 266. prehistoric combustion structure formation processes, fuel Anikhovskij, S. A., E. A. Voronkova, A. P. Zabiyako, and R. A. Kobyzov. 2012. management systems, and hearth function. We have shown Evenki priamurja: olennaja tropa istorij i kultury. Blagoveschchensk: that ethnoarchaeological research, whether motivated by the Amurskij gosudartsvennij Universitet, Centr po cokhraneniju istoriko- kulturnovo nacledija Amurskoj Oblasti. need to test experimental or archaeological interpretations or Beyries, S. 1995. Expérimentation archéologique et savoir-faire traditionnel: by exploration, requires case-specific questions and analytical l’exemple de la découpe d’un cervidé. Techniques et Culture 22. http:// parameters. Our two case studies are examples of the great tc.revues.org/587 (accessed May 8, 2015). ———. 2002. Le travail du cuir chez les Tchouktches et les Athapaskans: potential in applying geoarchaeological and bioarchaeological implications ethno-archéologiques. In Le travail du cuir de la préhistoire à techniques and have allowed us to identify several important nos jours. S. Beyries and F. Audouin-Rouzeau, eds. Pp. 143–159. Actes des factors that should be taken into account in future research XXIIe rencontres internationales d’archéologie et d’histoire d’Antibes. Antibes: APDCA. design. These include the presence of specialists on site and ———. 2008. Modélisation du travail du cuir en ethnologie: proposition d’un application of multitechnique microstratigraphic studies. système ouvert à l’archéologie. Anthropozoologica 43(1):9–42. So far, our work has provided pilot micromorphological Beyries, S., and P. Pétrequin. 2001. Ethno-archaeology and its transfers. BAR International Series 983. Oxford: Archaeopress. and anthracological data that contribute to the study of Pa- Beyries, Sylvie, and Virginie Vaté. 2007. Les civilisations du renne d’hier et leolithic fire. However, it needs to be expanded significantly d’aujourd’hui: approches ethnohistoriques, archéologiques et anthropologiques. in different ways. We now have a few examples of open-air Antibes: APDCA. Binford, L. R. 1967. Smudge pits and hide smoking: the use of analogy in hearths that need to be complemented with examples of archaeological reasoning. American Antiquity 32(1):1–12. hearths from enclosed settings such as caves and rock shelters ———. 1978. Nunamiut ethnoarchaeology. New York: Academic Press. S228 Current Anthropology Volume 58, Supplement 16, August 2017

———. 1980. Willow smoke and dogs’ tails: hunter-gatherer settlement Manuel Marreiros, Juan F. Gibaja Bao, and Nuno Ferreira Bicho, eds. systems and archaeological site formation. American Antiquity 45(1):4–20. Pp. 27–40. New York: Springer. Brandisauskas, D. 2007. Symbolism and ecological uses of fire among Gur-Arieh, S., E. Mintz, E. Boaretto, and R. Shahack-Gross. 2013. An Orochen-Evenki. Sibirica 6(1):95–109. ethnoarchaeological study of cooking installations in rural Uzbekistan: ———. 2010. Hide tanning and its use in the taiga: the case of Orochen- development of a new method for identification of fuel sources. Journal of Evenki reindeer herders and hunters of Zabaikalye (East Siberia). Journal of Archaeological Science 40(12):4331–4347. Ethnology and Folkloristics 4(2):97–114. Henry, A. 2011. Paléoenvironnements et gestion du bois de feu au Méso- Brochier, J. E., P. Villa, M. Giacomarra, and A. Tagliacozzo. 1992. Shepherds lithique dans le sud-ouest de la France: anthracologie, ethno-archéologie et and sediments: geo-ethnoarchaeology of pastoral sites. Journal of Anthro- expérimentation. Doctoral dissertation, Université de Nice. pological Archaeology 11(1):47–102. Henry, A., and M. Boboeuf. 2016. Environnement ligneux et gestion du bois Brooks, A. S., and J. E. Yellen. 1987. The preservation of activity areas in the de feu au cours du Mésolithique au Clos de Poujol (Campagnac, Aveyron). archaeological record: ethnoarchaeological and archaeological work in Bulletin de la Société Préhistorique Française 113(1):5–30. northwest Ngamiland, Botswana. In Method and theory for activity area Henry, A., and Teten’kin A. 2014. Анализ образцов древесного угля с research. Susan Kent, ed. Pp. 63–106. New York: Columbia University местонахождения Коврижка III на Витиме (Байкало-Патомское Press. Нагорье, Иркутская Область) [Anthracological analysis of the wooden Cameron, C. M., and S. A. Tomka, eds. 1996. The abandonment of settlements pieces from Kovrizhka 3 site on Vitim Valley (Baikalo-Patom’s Escarp- and regions: ethnoarchaeological and archaeological approaches. New York: ment, Irkutskaja Oblast)]. Известия Лаборатории древних технологий Cambridge University Press. 2(11):9–19. Chabal, L. 1982. Méthodes de prélèvement des bois carbonisés protohisto- Henry, A., and I. Théry-Parisot. 2014a. From Evenk campfires to prehistoric riques pour l’étude des relations homme-végétation. Masters thesis, Uni- hearths: charcoal analysis as a tool for identifying the use of rotten wood as versité des Sciences et Techniques du Languedoc. fuel. Journal of Archaeological Science 52:321–336. ———. 1991. L’Homme et l’évolution de la végétation méditerranéenne des ———. 2014b. Fuel management during the Mesolithic: current research in âges des métaux à la période romaine: recherches anthracologiques théoriques, archaeobotany. In Des techniques aux territoires: nouveaux regards sur les appliquées principalement à des sites du Bas-Languedoc. Doctoral dissertation, cultures mésolithiques. Auréade Henry, Benjamin Marquebielle, Lorène Ches- Université de Montpellier 2. naux, and Sylvène Michel, eds. Palethnologie 6. http://blogs.univ-tlse2.fr ———. 1994. Apports récents de l’anthracologie à la connaissance des /palethnologie/2014-revue/. paysages passés: performances et limites. Histoire and Mesure 9(3):317– Henry, A., I. Théry-Parisot, and E. Voronkova. 2009. La gestion du bois de feu 338. en forêt boréale: problématique archéo-anthracologique et étude d’un cas Costamagno S., and F. David. 2009. Comparaison des pratiques bouchères et ethnographique (Région de l’Amour, Sibérie). In Gestion des combustibles culinaires de différents groupes sibériens vivant de la renniculture. Ar- au paléolithique et au mésolithique: Nouveaux outils, nouvelles interpréta- chaeofauna 18:9–25. tions. Isabelle Théry-Parisot, Sandrine Costamagno, and Auréade Henry, Costamagno, S., I. Théry-Parisot, J. C. Castel, and J.-P. Brugal. 2009. Com- eds. Pp. 17–37. BAR International Series 1914. Oxford: Archaeopress. bustible ou non? analyse multifactorielle et modèles explicatifs sur de Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current ossements brûles paléolithiques. In Gestion des combustibles au Paléolithique Anthropology 58(suppl. 16):S329–S336. et au mésolithique: noveaux outils, nouvelles interprétations. I. Théry-Parisot, Hlubik, Sarah, Francesco Berna, Craig Feibel, David Braun, and John W. K. S. Costamagno, and A. Henry, eds. Pp. 65–84. BAR International Series 1914. Harris. 2017. Researching the nature of fire at 1.5 Mya on the site of FxJj20 Oxford: Archaeopress. AB, Koobi Fora, Kenya, using high-resolution spatial analysis and FTIR David, N., and C. Kramer. 2001. Ethnoarchaeology in action. Cambridge: spectrometry. Current Anthropology 58(suppl. 16):S243–S257. Cambridge University Press. Holdaway, Simon J., Benjamin Davies, and Patricia C. Fanning. 2017. Ab- Fewster, K. J., and M. Zvelebil. 2001. Ethnoarchaeology and hunter-gatherers: original use of fire in a landscape context: investigating presence and ab- pictures at an exhibition. BAR International Series 955. Oxford: Archaeo- sence of heat-retainer hearths in western New South Wales, Australia. press. Current Anthropology 58(suppl. 16):S230–S242. Friesem, D., E. Boaretto, A. Eliyahu-Behar, and R. Shahack-Gross. 2011. Degra- Joly, D., R. March, D. Marguerie, and H. Yacobaccio. 2009. Gestion des dation of mud brick houses in an arid environment: a geoarchaeological model. combustibles dans la province de Jujuy (Puna, Argentine) depuis l’Holocène Journal of Archaeological Science 38(5):1135–1147. ancien: croisement des résultats ethnologiques et anthracologiques. In Gestion Friesem, D. E. 2016. Geo-ethnoarchaeology in action. Journal of Archaeo- des combustibles au paléolithique et au mésolithique: nouveaux outils, nou- logical Science 70:145–157. velles interprétations. I. Théry-Parisot, S. Costamagno and A. Henry, eds. Friesem, D. E., P. Karkanas, G. Tsartsidou, and R. Shahack-Gross. 2014a. Pp. 39–56. BAR International Series 1914. Oxford: Archaeopress. Sedimentary processes involved in mud bricks degradation in temperate Kelly, R. L. 1995. The foraging spectrum: diversity in hunter-gatherer lifeways. environment: an ethnoarchaeological case study from northern Greece. Washington, DC: Smithsonian Institution Press. Journal of Archaeological Science 41:556–567. Kent, Susan. 1993. Models of abandonment and material culture frequencies. Friesem, D. E., G. Tsartsidou, P. Karkanas, and R. Shahack-Gross. 2014b. In Abandonment of settlements and regions: ethnoarchaeological and ar- Where are the roofs? a geoethnoarchaeological study of mud structures and chaeological approaches. C. M. Cameron and S. A. Tomka, eds. Pp. 54–73. their collapse processes, focusing on the identification of roofs. Archaeo- Cambridge: Cambridge University Press. logical and Anthropological Sciences 6:73–92. Kuznetsov, O. V. 2007. Ethnoarchaeological approach to Late Palaeolithic Friesem, D. E., Y. Zaidner, and R. Shahack-Gross. 2013. Formation processes settlements and habitation structures (Transbaikal, Siberia). In Les and combustion features at the lower layers of the Middle Palaeolithic civilisations du renne d’hier et d’aujourd’hui: approches ethnohistoriques, open-air site of Nesher Ramla, Israel. Quaternary International 331:128–138. archéologiques et anthropologiques. S. Beyries and V. Vaté, eds. Pp. 111– Goldberg, P., and R. I. Macphail. 2008. Practical and theoretical geoarchae- 125. Antibes: APDCA. ology. Hoboken, NJ: Blackwell. Lancelotti, C., and M. Madella. 2012. The “invisible” product: developing Goldberg, P., and I. Whitebread. 1993. Micromorphological studies of a markers for identifying dung in archaeological contexts. Journal of Ar- Bedouin tent floor. In Formation processes in context. P. Goldberg, D. T. chaeological Science 39(4):953–963. Nash, and M. Petraglia, eds. Pp. 165–188. Monographs in World Ar- Lancelotti, C., J. Ruiz-Pérez, and J. J. García-Granero. 2016. Investigating fuel chaeology 17. Madison, WI: Prehistory Press. and fireplaces with a combination of phytoliths and multi-element analysis: Goldberg, Paul, Christopher E. Miller, and Susan M. Mentzer. 2017. Recog- an ethnographic experiment. Vegetation History and Archaeobotany 26 nizing fire in the Paleolithic archaeological record. Current Anthropology 58 (1):75–83. (suppl. 16):S175–S190. Lavrillier, A. 2005. Nomadisme et adaptations sédentaires chez les Évenks de González-Ruibal, A., A. Hernando, and G. Politis. 2011. Ontology of the self Sibérie postsoviétique: “jouer” pour vivre avec et sans chamanes. PhD and material culture: arrow-making among the Awá hunter-gatherers thesis, École Pratique des Hautes Études, Paris. (Brazil). Journal of Anthropological Archaeology 30(1):1–16. ———. 2007. Gestion duelle de l’espace à long terme chez les Evenks éleveurs González-Urquijo J., S. Beyries, and J. J. Ibáñez. 2015. Ethnoarchaeology and de rennes et chasseurs des monts Stanovoï: interférences ou cohérences des functional analysis. In Use-wear and residue analysis in archaeology. João zones sauvages et humanisées. In Les civilisations du renne d’hier et Mallol and Henry Ethnoarchaeology of Paleolithic Fire S229

d’aujourd’hui: approches ethnologiques, archéologiques et anthropologiques. Shahack-Gross, R., F. Marshall, and S. Weiner. 2003. Geo-ethnoarchaeology S. Beyries and V. Vaté, eds. Pp. 65–88. Antibes: APDCA. of pastoral sites: the identification of livestock enclosures in abandoned Leroi-Gourhan, Arlette, and André Leroi-Gourhan. 1989. Un voyage chez les Maasai settlements. Journal of Archaeological Science 30(4):439–459. Aïnous: Hokkaïdo 1938. Paris: Albin Michel. Sillar, B. 2000. Shaping culture: making pots and constructing households: an Lim, Imogene. 1985. Rock-shelter use today: an indicator of Usandawe pre- ethnoarchaeological study of pottery production, trade and use in the Andes. history. In Recent Advances in Indo-Pacific Prehistory. V. N. Misra and Oxford: J. & E. Hedges. Peter Bellwood, eds. Pp. 105–110. Leiden: Brill. Skibo, J. M. 2009. Archaeological theory and snake-oil peddling: the role of Mallol, C., F. W. Marlowe, B. M. Wood, and C. C. Porter. 2007. Earth, wind, ethnoarchaeology in archaeology. Ethnoarchaeology 1(1):27–57. and fire: ethnoarchaeological signals of Hadza fires. Journal of Archaeo- Svoboda, J., S. Sázelová, P. A. Kosintsev, V. Jankovská, and M. Holub. 2011. logical Science 34(12):2035–2052. Resources and spatial analysis at actual Nenets campsites: ethnoarchaeolog- Mallol, C., S. M. Mentzer, and C. E. Miller. Forthcoming. Combustion ical implications. Journal of Anthropological Archaeology 30(1):30–43. features. In Archaeological soil and sediment micromorphology. C. Nicosia Théry-Parisot, I. 2001. Economie des combustibles au paléolithique: expérimen- and G. Stoops, eds. Chichester: Wiley. tation, taphonomie, anthracologie. Dossier de documentation archéologique Mansur Franchomme, M. E. 1983. Traces d’utilisation et technologie lithique: 20. Paris: CNRS. exemple de la Patagonie. Doctoral dissertation, University of Bordeaux. Théry-Parisot, I., L. Chabal, and J. Chrzavzez. 2010. Anthracology and ta- Mentzer, S. M. 2014. Microarchaeological approaches to the identification and phonomy, from wood gathering to charcoal analysis: a review of the interpretation of combustion features in prehistoric archaeological sites. taphonomic processes modifying charcoal assemblages, in archaeological Journal of Archaeological Method and Theory 21(3):616–668. contexts. Palaeogeography, Palaeoclimatology, Palaeoecology 291(1/2):142– Monahan, C. M. 1998. The Hadza carcass transport debate revisited and its 153. archaeological implications. Journal of Archaeological Science 25(5):405–424. Théry-Parisot, I., L. Chabal, and S. Costamagno, eds. 2010. The taphonomy of Nelson, R. K. 1986. Hunters of the northern forest. 2nd edition. Chicago: burned organic residues and combustion features in archaeological con- University of Chicago Press. texts. Proceedings of the round-table Valbonne, May 27–29 2008, CEPAM. Ntinou, M. 2002. La paleovegetación en el norte de Grecia desde el Tardiglaciar Palethnologie 2. http://blogs.univ-tlse2.fr/palethnologie/en/2010-revue/. hasta el Atlántico: formaciones vegetales, recursos y usos. BAR International Théry-Parisot I., S. Costamagno, and A. Henry, eds. 2009. Gestion des Series 1038. Oxford: Archaeopress. combustibles au Paléolithique et au Mésolithique: nouveaux outils, nouvelles O’Connell, J. F. 1987. Alyawara site structure and its archaeological impli- interprétations. BAR International Series 1914. Oxford: Archaeopress. cations. American Antiquity 52(1):74–108. Théry-Parisot, I., and P. J. Texier. 2006. L’utilisation du bois mort dans le site Osgood, C. 1970 (1936). Contributions to the ethnography of the Kutchin. Yale moustérien de la Combette (Vaucluse): apport d’une approche morpho- University Publications in Anthropology 14. New Haven, CT: Yale Uni- métrique des charbons de bois à la définition des fonctions de site, au versity Press. Paléolithique. Bulletin de la Société Préhistorique Française 103(3):453–463. Perlès, C. 1977. Préhistoire du feu. Paris: Masson. Vidal-Matutano, P. 2013. Combustible vegetal y etnografía: estudio de un Peterson, N. 1971. Open sites and the ethnographic approach to the archaeology horno de pan en Ghuala (Argelia). Arkeogazte 2013 3:63–79. of hunter-gatherers. In Aboriginal man and environment in Australia.J. ———. 2016. Alrededor del fuego: paisaje, clima y gestión de los recursos Mulvaney and J. Golson, eds. Pp. 239–248. Canberra: Australian National leñosos en grupos cazadores-colectores durante el Paleolítico medio (Ali- University Press. cante, España). Doctoral dissertation, University of Valencia, Spain. ———. 1973. Camp site location amongst Australian hunter-gatherers: ar- Vidal-Matutano, P., A. Henry, and I. Théry-Parisot. 2017. Dead wood gath- chaeological and ethnographic evidence for a key determinant. Archaeology ering among Neanderthal groups: charcoal evidence from Abric del Pastor and Physical Anthropology in Oceania 8(3):173–193. and El Salt (Eastern Iberia). Journal of Archaeological Science 80:109–121. Pétrequin, A. M., and P. Pétrequin. 1988. Ethnoarcheologie de l’habitat en Waguespack, N. M. 2002. Caribou sharing and storage: refitting the Palangana grotte de Nouvelle-Guinee: une transposition de l’espace social et eco- site. Journal of Anthropological Archaeology 21(3):396–417. nomique. Archives Suisses d’Anthropologie Générale 1:61–82. Wattez, J. 1992. Dynamique de formation des structures de combustion de la Picornell Gelabert, L., E. Asouti, and E. A. Allué Martí. 2011. The ethnoar- fin du Paléolithique au Néolithique moyen. Doctoral dissertation, Université chaeology of firewood management in the Fang villages of Equatorial Guinea, de Paris I. central Africa: implications for the interpretation of wood fuel remains from ———. 1994. Micromorphologie des foyers d’Etiolles, de Pincevent et de archaeological sites. Journal of Anthropological Archaeology 30(3):375–384. Verberie: le milieu naturel et son exploitation. 2. Ressources et exploitations. Sázelová, S., J. Svoboda, Pavel A. Kosintsev, and M. Novák. 2015. Patterns of Documents d’Archéologie Française 43:120–127. change in a Nenets landscape: an ethnoarchaeological study of Yangana Pe, Wobst, H. M. 1978. The archaeo-ethnology of hunter-gatherers or the tyr- Polar Ural Mts. Russia. Human Ecology 43:283–294. anny of the ethnographic record in archaeology. American Antiquity 43 Shahack-Gross, R., F. Marshall, K. Ryan, and S. Weiner. 2004. Reconstruction (2):303–309. of spatial organization in abandoned Maasai settlements: implications for Zapata Peña, L., L. Peña-Chocarro, J. J. Ibáñez Estévez, and J. E. González site structure in the Pastoral Neolithic of East Africa. Journal of Archaeo- Urquijo. 2003. Ethnoarchaeology in the Moroccan Jebala (Western Rif): logical Science 31(10):1395–1411. wood and dung as fuel. Africa Praehistorica 15:163–175. S230 Current Anthropology Volume 58, Supplement 16, August 2017

Aboriginal Use of Fire in a Landscape Context Investigating Presence and Absence of Heat-Retainer Hearths in Western New South Wales, Australia

by Simon J. Holdaway, Benjamin Davies, and Patricia C. Fanning

A case study from western New South Wales, Australia, illustrates the age, preservation, and distribution of late Ho- locene heat-retainer hearths that are abundant in the semiarid archaeological record in the region. These hearths were constructed as underground ovens with stone heat retainers. They appear archaeologically as eroded concentrations of heat-fractured stone sometimes protecting charcoal deposits. We explore geomorphic processes influencing hearth temporal and spatial distributions using a neutral agent-based model. Parallels between model outcomes and the dis- tribution of hearths in space and time suggest that processes of sediment erosion and deposition are having complex effects on hearth survivorship and therefore on patterns of hearth frequency. We consider the various processes that explain why hearths were made in the past and how they manifest in the present. Despite the relatively recent age of the hearths when compared with evidence for fire use in the Paleolithic record, the presence and absence of these fire features reflect the outcome of a large number of processes interacting together, not all of them related to human behavior. We use the results of the case study to comment on current behavioral models for the presence and absence of fire use in the distant past.

Human use of fire in the semiarid regions of Australia is visi- Such a distribution is not unique to semiarid Australia, being ble archaeologically as abundant heat-retainer hearths that are reported in some other parts of the world where surface expo- found in eroded contexts along drainage lines. These hearths sure is favorable (e.g., Black and Thoms 2014; Brink and Dawe appear as concentrations of heat-fractured rocks clustered to- 2003; Milburn, Doan, and Huckabee 2009; Petraglia 2002; Schae- gether with tens to thousands of heat-retainer fragments. In fer et al. 2014; Sullivan et al. 2001; Thoms 2009). In common some hearths, these heat retainers protect charcoal deposits. with the Australian example, the distribution of these fire fea- However, as we discuss below, many of the hearths are erod- tures indicates an additional aspect of “on site” and “off site” ing, and once exposed, the charcoal is also susceptible to ero- fire use. Yet despite their ubiquity in different areas of the sion. The distribution of these hearths is of particular interest world, fire use in these settings has received relatively little at- because they do not conform to a pattern consistent with oc- tention. Here we outline the archaeological expression of this cupation sites and they are not the remains of broadcast fir- form of fire use in one part of semiarid Australia. In doing so, ing used to concentrate game or modify flora. Instead, these we emphasize how fire use can be understood as a set of in- hearths were constructed in almost all parts of drainage lines. terconnected relationships between people and environment in the way that human use of fire manifests, in the technology of fire use, and in the ways the outcomes of fire use are preserved Simon J. Holdaway is Professor in the School of Social Sciences at in the archaeological record. the University of Auckland (Private Bag 92019, Auckland 1142, New An older literature commented on changes in past Austra- Zealand [[email protected]]), in the Department of Archae- lian environments as a consequence of Aboriginal fire use (e.g., ology at the University of Queensland (Brisbane, St. Lucia, Queensland Flannery 1994; Kershaw et al. 2002; Singh, Kershaw, and Clark 4072, Australia), and in the Department of Environmental Sciences at 1981). More recently, changes in the frequency of radiometric Macquarie University (North Ryde, Sydney, New South Wales 2109, dates from archaeological fire features have been used as pri- Australia). Benjamin Davies is Postdoctoral Researcher in the Depart- mary evidence for directional changes in past human popu- ment of Anthropology of the School of Social Sciences at the Univer- lation and occupation intensity (Johnson and Brook 2011; Smith sity of Auckland (Private Bag 92019, Auckland 1142, New Zealand [[email protected]]). Patricia C. Fanning is Associate Profes- et al. 2008; Williams et al. 2015). However, patterns in large-scale fi sor in the Department of Environmental Sciences of Macquarie Uni- records of re are not always principally determined by human versity (North Ryde, Sydney, New South Wales 2109, Australia [patricia activities even when human action may have contributed to their [email protected]]). This paper was submitted 25 VII 16, accepted initial creation. For example, Mooney and colleagues (2011, 2012) 7 II 17, and electronically published 17 V 17. have shown that when large numbers of charcoal records are q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0007$10.00. DOI: 10.1086/691436 Holdaway, Davies, and Fanning Aboriginal Use of Fire in a Landscape Context S231 compared from across the continent, climate-modulated changes in vegetation explain more of the variation in fire prevalence than do human activities over timescales measured in centuries to millennia, a finding that parallels the results of global studies (Bowman et al. 2009, 2011). If patterning in archaeological fire features is to be used to understand past human behavior, then behavioral contributions to those patterns need to be contex- tualized within the larger suite of formational processes oper- ating on landscapes over time. The reasons that hearths constructed in different places are found, the form that these hearths took, and the archaeological preservation of these hearths can be understood by consider- ing fire use in relation to environmental and geomorphological history before behavioral inferences are drawn. While the his- tory of fire use in semiarid Australia is of interest in itself, the case study reported here also has wider relevance for studying human use of fire. The types of archaeological signatures fire use produces will vary contextually, and inference about fire use needs to be assessed accordingly. The Australian case study illustrates that where consideration of the influence of these contexts on the formation of patterning associated with fire use is absent, behavioral inferences drawn from the presence or absence of fire features in the archaeological record may be mis- leading. This particularly applies to studies that rely on the ab- sence of evidence for fire use (e.g., Roebroeks and Villa 2011; Sandgathe et al. 2011). How fire manifests archaeologically is partly determined by the contexts in which people found fire useful, and these con- texts in turn shape the archaeological record that preserves it. Fire functions are not self-evident in the sense that fire needs Figure 1. Western New South Wales study area. Source: Holdaway to be treated in relation to the ways people and objects inter- and Fanning (2014), pl. 1. acted with past environments before its presence or absence can be understood. This relates to the way fire is employed by hu- mans but also to how the material remains are preserved, it- fined to the larger watercourses. Both of these vegetation com- self the outcome of the interaction of the material remains of munities provided firewood. Extensive areas of stony (“gibber”) fire and a variety of environmental and geomorphological pro- and bare surfaces mantle the slopes and plains. Changes in cesses. Documenting such interactions from archaeological re- vegetation cover and geomorphic processes occurred with the mains raises significant issues because at one level, we can only introduction of European-style pastoralism in the nineteenth analyze what is preserved and what we can see. As is illustrated century (Fanning 1994, 1999). Topsoil from hillsides was de- in the case study below, this issue is not confined to very an- posited on valley floors as laminated sandy sediments. Streams cient archaeological records. incised into the valley fills, and renewed erosion on the valley floors followed, leaving hard-setting, saline subsoils exposed at the surface on which archaeological materials, mostly stone arti- Case Study: Hearths in Western New South Wales, facts but also heat-fractured stones from heat-retainer hearths, Australia now rest (Fanning 1999). Episodic flood events, resulting in erosion in some areas and The Semiarid Environment deposition in others, are a feature of the study area (Fanning, The Rutherfords Creek study area is located on the valley floor Holdaway, and Rhodes 2007). Mean annual rainfall is low, but margins of an ephemeral stream draining a catchment of 37.8 km2 variability is high, and the bulk of the rain falls during short, in western New South Wales (NSW) on the southeastern mar- intense rain depressions, especially in summer. Episodic events gin of the central Australian arid zone (fig. 1). The climate today such as this affect the sedimentary record. Where the contem- is semiarid, with mean annual rainfall less than 250 mm and porary creek lines are cut into alluvial sequences, unconfor- pan evaporation exceeding 2,000 mm (Holdaway, Fanning, and mities represent either substantial hiatuses in valley floor ag- Witter 2000). As a consequence, vegetation is sparse, with che- gradation or, more likely, periods dominated by valley floor nopod shrublands ubiquitous across the hillsides and trees con- erosion. For example, at Stud Creek (fig. 1), gaps of several S232 Current Anthropology Volume 58, Supplement 16, August 2017 thousand years occur in the depositional record of Holocene filling, burying, and preserving hearth remains. Once buried, and late Pleistocene deposits (Fanning and Holdaway 2001). hearths were probably difficult to identify even by the people The absence of buried paleosols in these deposits suggests the who made them. Their presence on the surface today is a con- unconformities represent more than just stable periods in the sequence of erosion processes that are both exposing them and evolution of the landscape when aggradation temporarily ceased. at the same time causing their destruction and, for some, re- Instead, it is likely that an older record of occupation was de- burial (Fanning et al. 2009). stroyed by erosion represented by the unconformities in the Heat-retainer hearths were identified during archaeological stratigraphic sequence (Fanning and Holdaway 2001:99). surveys of Rutherfords Creek as concentrations of heat-fractured Surface visibility, and therefore the quantity of the surface stone sometimes protecting an underlying deposit of charcoal, archaeological record, is highest where there is an absence of with 979 recorded in eroding valley floor deposits in the Ruth- vegetation and the surface is lagged, forming “scalds” where erfords Creek study area. Heat-fractured stone can be identi- topsoils (more correctly, the uppermost sedimentary unit) are fied by the presence of small nonconchoidal flakes shed during removed by a combination of wind and water erosion (Fan- heating (pot lids) and by cracking patterns as well as by irregular ning and Holdaway 2004). Stone artifacts and hearth stones fracture. Identification was easiest when the heat-fractured stones are exposed on scalds as the finer sediments are removed by were concentrated together but became progressively more dif- unconcentrated overland water flow. On slopes greater than two ficult as erosion dispersed the fragments. Identification was also degrees, this overland flow can also move small artifacts with a dependent on the nature of the land surface on which they rest. maximum clast dimension less than 20 mm. However, lateral They were, for instance, much harder to identify on surfaces movement is much less discernible among the larger stone covered with a gravel lag. To some degree these problems were artifacts and stone heat retainers (Fanning et al. 2009). In a overcome by using a fluxgate gradiometer to measure both mag- sense, therefore, sediment erosion has “excavated” the archae- netic susceptibility and thermoremanent magnetism as a result ological record, offering an opportunity for investigation of large of heating (Fanning, Holdaway, and Phillipps 2009; see also quantities of this record distributed across the landscape. Gose 2000). Because they are exposed by erosion, the condition of hearths differs depending on the extent of this erosion. Six categories Archaeological Survey are used to describe the degree of hearth preservation (fig. 2): From 2005 to 2008, archaeological and geomorphological sur- buried, partially exposed, intact, disturbed, scattered, and rem- veys were conducted along a 13 km length of the main channel nant. “Buried hearths” describes those that remain largely bur- of Rutherfords Creek on the eroding valley floor margin. Ar- ied with only the tips of the fire-cracked rock poking above the tifact deposits are visible on eroded patches (the “scalds” dis- surface (fig. 2A). “Partially exposed hearths” describes hearths cussed above) making up approximately 2 km2 of the valley where a portion of the cluster of hearth stones is exposed along floor. The detailed archaeological surveys were confined to a an erosion escarpment (fig. 2B) but the bulk of the hearth re- randomly selected sample of the approximately 2,267 mapped mains buried. “Intact hearths” are those where erosion has com- scalds, amounting to approximately 4.5% of the eroded valley pletely exposed the dense cluster of fire-cracked rock, but it has floor by area. not been dispersed (fig. 2C). The next three categories—“dis- Hearths were constructed by Aboriginal people in the past by turbed,”“scattered,” and “remnant”—refer to hearths dis- excavating a depression that formed the body of an oven into playing increasing amounts of disturbance of the heat retainers which stones and then food items could be placed for cook- (fig. 2D–2F). ing (Holdaway et al. 2002). Europeans who observed Aborig- All 2,267 scalds and gravel patches identified in Rutherfords inal people in the nineteenth and early twentieth centuries de- Creek were searched to determine the presence of heat-retainer scribed the construction and use of these hearths (e.g., Eyre hearths. In addition, the areas between the sampling units were 1845 and Parker 1905, cited in Allen 1972:280–281). In one systematically surveyed. Hearths found on the scalds and gravel example, Daniel George Brock, who accompanied the explorer patches totaled 737 with an additional 242 hearths located be- Charles Sturt (Peake-Jones 1988), recounts how hearths were tween the sampling units (fig. 3). The overall distribution of constructed by scooping out a hollow, adding oven stones, then hearths recorded per scald or gravel patch is clustered (Moran’s lighting a fire to heat them and burying plant and animal food I p 0.131, z p 9.954, P ! .001). However, an Anselin Moran’s in the scoop. The oven stones aided cooking by acting as heat I analysis indicates only two areas with substantial hearth retainers, helping to prolong the heat generated by the fire. clusters. The first is in the northeastern corner of the catch- Recent findings from the Mungo region show that the use of ment, an area with concentrated sheet wash erosion, and a sec- heat-retainer technology is present in Pleistocene deposits in ond area in the center of the catchment, where hearths are Australia (Fitzsimmons et al. 2015). clustered in scalds that are particularly eroded. In addition, As discussed below, heat-retainer technology dated to dif- there are isolated examples where there are an unusually high ferent periods is found elsewhere in the world and, in some number of hearths. Overall, while there are hearths visible in places, is still in use today (Petraglia 2002). In western NSW, all parts of the catchment valley floor margins, these are not subsequent abandonment of the hearth led to depression in- clustered in any one part of the valley. Figure 2. Hearth preservation categories: A p buried, B p partially exposed, C p intact, D p disturbed, E p scattered, F p remnant. Source: Fanning, Holdaway, and Phillipps (2009), fig. 3. S234 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 3. Hearths in Rutherfords Creek represented by black dots. Large gray circles represent Cluster 11 hearths. Inset shows Cluster 11 OSL sample locations (letters) and hearths (gray dots).

It is worth emphasizing that the total number of hearths iden- Hearth ages range from modern to ages in excess of 6000 cal tified reflects only those that were visible at the times when the BP, with the spatial patterning depending on their geomorphic survey was conducted. This total therefore reflects the outcome setting (Holdaway, Fanning, and Rhodes 2008). To use Cluster of processes of erosion and aggradation that initially exposed 11 as an example (fig. 3), the youngest hearths in this cluster hearths and in some cases destroyed them while at other times are found closest to the creek channel and to the north on the reburying them. Thus, the total number of observed hearths is true left bank. OSL sediment ages suggest that in these regions, a sample of a much larger number of hearths that still exist or ancient sediments are deeply buried. At OSL sample locations that once existed. The patterns that we see, both in the distri- L and M and N and O, Pleistocene-aged sediments occur but bution of hearths and in their ages, need to be interpreted with only at depths greater than 70 cm. Near-surface sediments are these processes in mind. An agent-based simulation study of more recent than 2000 BP. In contrast, sediments at location K hearth exposure (Davies, Holdaway, and Fanning 2015) was date to around 4000 BP at 35 cm depth. Here, hearths have developed and applied to illustrate these processes, as sum- ages that range from 2000 to 4000 BP together with some that marized below. are more recent. Thus, the distribution of hearth ages is likely to reflect surface preservation (Holdaway, Fanning, and Rhodes 2008). The northern hearths in Cluster 11 rest on sediments Hearth Dating that are recent in age, probably because at times in the past the Two hundred and fifty-six hearths of the 979 identified were creek has avulsed and removed older hearths through erosion. excavated to obtain dating samples, including hearthstones for Ancient sediments at this location are relatively deeply buried optically stimulated luminescence (OSL) dating (Rhodes, Fan- by more recent deposits. To the south of the creek, erosion is ning, and Holdaway 2010; Rhodes et al. 2009). Only about one less prevalent, leaving relatively ancient hearths intact (as de- third of the hearths excavated contained enough charcoal for termined by OSL measurements on hearthstones). radiocarbon dating. Hearths were excavated in clusters to deter- Hearth ages therefore partly reflect the age of the sediment mine whether groups of hearths shared similar or different ages. surfaces on which they rest. However, we also know that the Holdaway, Davies, and Fanning Aboriginal Use of Fire in a Landscape Context S235 hearths themselves are subject to change as indicated by the pared with the results obtained from the survey along Ruther- classification of forms discussed above. Of the hearths that we fords Creek, an agent-based model was constructed using the excavated, only a portion retained sufficient charcoal from which NetLogo modeling platform (Davies, Holdaway, and Fanning to obtain dating samples. These observations need to be con- 2015). The model consists of a grid world where each cell con- sidered when investigating the spatial and temporal patterns of tains an ordered list of dated sedimentary layers. In the model, the hearths that remain archaeologically. The current level of computational agents construct dated hearths at a constant rate hearth exposure in parts of Rutherfords Creek, such as Cluster 11, over random points on the grid during the period from 2000 to is likely to be a reflection of the increased erosion since European 200 BP. Any hearth with an age younger than or equal to the pastoral practices were introduced. However, to some degree, age of the most recent layer of sediment of the cell on which it semiarid regions of western NSW have always been subject to rests is considered visible, while any that are older are hidden erosion, as illustrated by the stratigraphic disconformities in as part of a subsurface deposit. At given time intervals, events sedimentary deposits dating back to the Pleistocene (Fanning occur that have the capacity to erode or deposit sediment in and Holdaway 2001). It is to be expected, therefore, that some a grid cell. If erosion occurs, the uppermost layer of sediment hearths will be destroyed through erosion while others will be is removed, and any hearths visible on the surface lose their modified through exposure and, in some cases, reburial. Ero- datable charcoal. If deposition occurs, a layer of sediment is sion affects the components of hearths in different ways, with added to the cell, and any hearths currently visible on the sur- charcoal deposits more susceptible to dispersal than stone heat face are hidden. Hearths that are hidden can become reexposed retainers. As we demonstrate below using agent-based mod- through subsequent erosion events. In the model, two variables eling, we need to understand the impact of erosional processes are initially explored: the probability of individual cells expe- in order to make inferences about the spatial and temporal riencing erosion versus deposition during a given event (mod- extent of hearth preservation. eled as Bernoulli process; Jaynes 2003:42) and the time interval between these events. Results generated from 1,000 random samples of 100 hearth Agent-Based Model of Hearth Visibility ages ordered from youngest to oldest are plotted in figure 4A As discussed above, patterning in the archaeological evidence with separate graphs for time intervals between events of 10 for fire is often explained in behavioral terms based on in- to 200 years and the probabilistic difference between erosion ferences about hearth function derived from the contents of and deposition of sediment varied between 0.1 and 0.9 (to en- hearths and at times from ethnographic analogy and with limited sure some hearths are always present). If no deposition or ero- consideration of the roles of visibility or formational processes. sion occurred, hearths would fall along a diagonal line from Frequencies of radiocarbon data obtained from hearths or other 200 years BP at bottom left to 2000 years BP at top right, de- features are used to demonstrate diachronic changes in popu- pending on their age. When sampled ages of surface hearths lation or occupation intensity in a given location (e.g., Gamble from all simulations are compared, curves all fall to the left of et al. 2004; Johnson and Brook 2011; Mulrooney 2013; Rick this line and so are weighted toward the present. Increasing the 1987; Smith et al. 2008; Williams 2013; cf. Attenbrow and His- frequency of events produces a record that is younger on av- cock 2015; Delgado, Aceituno, and Barrientos 2015). However, erage, while more mixed regimes feature a wider range of dates. the strength of any explanation invoking behavioral change As events become less frequent, the mean age of mixed-regime should be based on the ability to demonstrate a difference from surface hearths tends to increase, but the variability decreases as instances when human behavior is not assumed (Brantingham the number of exposing events is fewer. However, in all cases, 2003:490). It is therefore important to establish how different the upper quartile age of surface hearths falls within the last the pattern is when behavioral change is removed from the 400 years, showing that the modeled surface archaeological equation. Neutral models in which the formation of a given record is biased toward the present as a result of differential proxy is uniform are frequently used as mechanisms for devel- preservation. oping expectations (e.g., Contreras and Meadows 2014; Rhode Variants that have inverse erosion to deposition ratios (e.g., et al. 2014). If changes in human behavior were the primary 0.3 and 0.7) feature more or less identical distributions. This driver in the generation of fire-related patterning in the archae- is because under more erosional conditions, older hearths are ological record, then it is expected that such patterning would less likely to survive destruction, and thus the record is mostly show little similarity to that produced by a model that assumed younger, while under more depositional conditions, older hearths no variation in behavior (Brantingham 2003; Premo 2007). will be hidden by layers of sediment, with only the most recent Similarities between the record and outcomes from a model hearths being visible on the surface. Surfaces featuring similar with neutral assumptions can be used to evaluate the extent to distributions of hearth ages may have formed under highly di- which behavior needs to be considered as part of an explana- vergent geomorphological regimes. tory framework or whether the resolution of the patterning is As the length of intervals between geomorphic events in- sufficient to distinguish it from a neutral record (Lake 2015:14). creases, chronological gaps appear. Because erosion or depo- To explore the effects of differential erosion and deposition sition events affect all grid cells in the model, all hearths sit- on a uniform record of hearth manufacture that could be com- ting on the surface at those times (which includes all hearths Figure 4. Distributions of radiocarbon (black) and OSL (gray) data from simulations (A–C) compared with data recorded from Rutherfords Creek (D). A–C, Individual graphs showing samples of 100 dates for each proxy taken from 1,000 separate simulation runs, ordered youngest to oldest, while each graph set shows different surface stability settings (A p 0, B p 0.1, C p 0.5). D, Dots showing mean ages for 93 radiocarbon determinations and a random sample of 93 OSL determinations less than 2000 BP (bars p 1 SD). Holdaway, Davies, and Fanning Aboriginal Use of Fire in a Landscape Context S237 accumulated since the last event) will either be obscured by de- (fig. 4D). Plotting the radiocarbon and OSL ages together shows position or destroyed by erosion, leaving only hearths from variation in steepness similar to that predicted by the simula- previous intervals exposed on the surface to be joined by hearths tion indicating how erosion has a different effect on charcoal from the upcoming interval. Repeating this process produces from hearths compared to the hearth stones. The chronological interdigitating sets of surfaces containing hearths grouped by gaps that are visible in the radiocarbon sample are almost ab- alternating time periods, with older hearths within those groups sent in the OSL sample. becoming rarer through time. If ages were obtained from these surfaces at any given point in time, there would appear to be chronological gaps in the record, but these would be purely the Discussion result of geomorphic activity. Hearths in Western New South Wales Figure 4B and 4C shows the results when surface stability is introduced, simulated as the probability of a cell’s surface To some degree the term “hearth” is evocative. In Western remaining stable through an erosion or depositional event. In societies, we imagine people sitting around a fire as a center of this simulation, the erosion probability is set to 0.5 with the domestic activity (e.g., Wrangham et al. 1999). This is unlikely event interval set to 100 years, but the percentages of cells left to be the type of activity that occurred around the hearths we stable (s) is varied. As the proportion of stable cells increases, find in Rutherfords Creek. As heat-retainer ovens, they rep- the surfaces on which the hearths rest become less organized resent an aspect of food preparation but not one that acted as a by the sedimentary process, and the gaps begin to aggrade (s p focus for domestic activity. The ethnohistoric accounts cited 0.1; fig. 4B). As surface stability approaches 50% (s p 0.5; fig. 4C), above indicate that animals prepared and placed in an oven the gaps are completely extinguished, but a record of increas- might be left for several hours while the people responsible ing frequency toward the present remains. When stability reaches undertook tasks elsewhere. In contrast, fires used in domestic 100% (s p 1, not shown in fig. 4), the record undergoes no locations for heat and light tended to be small and were made as geomorphic change and thus displays the uniform record of needed (Gould 1971). The explorer Sturt, commenting about hearth generation. Aboriginal hearths on the central Darling River, stated, “Our The qualitative similarities between the modeled data and fires were always so much larger than those made by them- those obtained from the field are striking. The chronological selves, that, they fancied, perhaps we were going to roast them” gaps in the modeled data reflect the vulnerability of charcoal (Sturt 1834:150). A heat-retainer hearth might be in use for a to dispersal through erosion and to sediment deposition that considerable period, but it did not need to be tended. There- obscures the hearths themselves. However, hearth stones might fore, the location where many daily tasks were carried out be expected to show less dispersal than charcoal because they need not be the location where the hearth remains are found. are much larger than charcoal fragments (Fanning and Hold- While a great many hearths exist, and many more have likely away 2001), and thus OSL dates obtained from hearth stones been buried or destroyed, these do not cluster either spatially should show less evidence for the chronological gaps. We ex- or temporally into groups that reflect prolonged occupations. plored this in the hearth simulation model by taking another Aboriginal people were able to construct hearths throughout set of samples that included hearths that were visible but had the Rutherfords Creek valley, indicating that at different times, lost their charcoal and thus were not included in the radio- there was sufficient fuel and material for hearth stones readily carbon sample. We have shown that in the more depositional available. From our observations (Fanning, Holdaway, and Phil- model environments, the radiocarbon and OSL dates track well lipps 2009), stone was likely sourced from the immediate vi- together as fewer hearths are destroyed (Davies, Holdaway, and cinity of the hearth location. In a small number of cases, termite Fanning 2015). However, as conditions become more ero- mound clay was also used as a heat retainer. The locations sional, the curve of the radiocarbon data remains steeper than selected for hearth construction also contained suitable sedi- that obtained from the OSL dates. This is because hearths that ments for the excavation of shallow depressions with the tech- have lost their charcoal in erosional events can still be sampled nology that they had available. Aboriginal people were also able using the OSL method. Meanwhile, hearths that are obscured to access fire in the different locations where the hearths were by overlying sediments are still effectively invisible to both dat- constructed. Gould (1971) describes the technique he observed ing techniques. When conditions become completely erosional, for making fire during his time in the Western Desert in the the radiocarbon distribution returns to the exponential cur- 1960s involving friction when a wooden spear thrower or throw- vature also seen under the highly depositional settings, but the ing stick was rubbed across another piece of wood. He cites OSL distribution straightens out, reflecting the actual record other accounts where fire was produced by a stick drill and of hearth ages produced by the agents. Gaps that are clear in stone percussion, and he recounts how at times fire was car- the simulated radiocarbon chronologies under settings with no ried using a substantial stick or bark (see also Hallam 1975:44). surface stability are effectively absent from those in the simu- Whether or not these or other methods were employed in the lated OSL record. A similar pattern is apparent when the actual past, the spatially extensive record of hearths that we observed radiocarbon ages obtained from hearths in Rutherfords Creek in Rutherfords Creek attests to material availability as well as are compared to OSL ages obtained from the hearth stones the technical ability to make and transport fire. Finally, that so S238 Current Anthropology Volume 58, Supplement 16, August 2017 many hearths are distributed so widely also indicates that suit- the opposite is the case: this carpet of dots is better explained able food items for cooking were available and that the impetus by the high mobility of small numbers of people in a landscape to cook them struck in a range of different locations at different where the ecology and topography leads to the wide dispersal times. of the archaeological material remains (Holdaway and Fanning At some level, this wide distribution of hearths in time and 2014). space indicates mobility rather than repetitive place use. Our Taking all these factors into account, the distribution of work on portable stone artifacts from Rutherfords Creek as well hearths in western NSW makes sense. Hearths are distributed as other locations in western NSW similarly illustrates how throughout the valley because at different times, different Aboriginal people curated flakes, leaving behind concentrations places might be attractive for hearth construction. At these of artifacts that are in a sense the antithesis of occupation sites times, places had sufficient fuel and stones to allow for hearth (e.g., Douglass et al. 2008; Douglass et al. 2015; Holdaway, construction as well as suitable food stuffs to cook. However, Douglass, and Phillipps 2015; Holdaway and Douglass 2015). while hearths were places where food was cooked, they might The artifacts that remain for us to study are those that were not not be the places where other activities were concentrated. selected for transport because they were considered unsuitable Because of the unpredictability of resource availability, people (Holdaway and Douglass 2012). Thus, as Isaac (1983) long ago needed to be able to access different places at different times, illustrated by way of a cartoon of an archaeologist imagining something that we characterize as a low redundancy in place a group of hominins sitting around a fire, we have to be wary of use (Holdaway and Fanning 2014). The same processes that deriving social or behavioral explanations from what appear to explain the distribution of hearths also have an effect on their be self-evident features of the archaeological record. preservation and therefore the chronology that can be obtained At larger spatial scales, geomorphic processes affect the abun- from the datable materials found with the hearths. Patterns in dance of resources available for Aboriginal people to exploit the chronological distribution of hearth radiocarbon ages may (Holdaway et al. 2015). On the one hand, western NSW illus- be an outcome of differential dispersal of charcoal and the vis- trates a lack of topographic complexity, with small areas of ibility of the hearths that retain these charcoal deposits. Visi- relatively high topographic roughness bounded by large areas bility also has an effect on the hearth chronology obtained with where there is little or no change in relief. On the other hand, OSL, although in a different way from that using charcoal; there is a high degree of local landscape heterogeneity, a differential visibility and hearth erosion means that there are product of modern vegetation cover plus regolith variation. To more hearths with recent ages than those that are older using the degree that this heterogeneity can be used as a proxy for past both dating systems, but gaps in the chronological record are environmental resource abundance, it suggests that at a local most apparent when only the radiocarbon record is viewed. level, resources were likely to be highly variable. In western The simulation study results show how temporal hearth NSW, resource heterogeneity did not translate into the forma- patterns can emerge when hearths are produced consistently tion of regularly recurring resource patches largely as a conse- and the frequency and likelihood of sediment erosion and de- quence of low soil fertility combined with the rainfall variability position are varied. Erosion in Rutherfords Creek was at times (Holdaway, Douglass, and Fanning 2013). spatially extensive. Using hearths dated to the last 2,000 years Low fertility and intermittent rainfall produces a landscape from all observed clusters, the distribution of radiocarbon ages that varies both spatially but also temporally in ways that would shows a pattern of fluctuating hearth frequencies similar to that have been difficult to predict. For humans, this meant that a produced in the simulation when the level of surface stability is location rich in resources at one moment in time might become kept low and sedimentary events are infrequent (Davies, Hold- depleted at another, with little way of predicting when such away, and Fanning 2015). This suggests that something like the a change might occur. Individual resource patch locations in alternating effects of erosion and deposition as modeled may western NSW were neither continuously nor cyclically attrac- be having an effect on hearth distribution, an inference further tive for occupation (Holdaway, Douglass, and Fanning 2013). bolstered by additional patterns in the OSL data predicted by the If resources were episodic in their availability, both spatially model (fig. 4D). Modern-day rain events can result in erosion and temporally, it might be expected that many places would or burial of hearths (Fanning, Holdaway, and Rhodes 2007), see at least some use and therefore would see the creation of an and stratigraphic disconformities indicate that at times in the archaeological record. Periods after episodic, abundant rainfall past, considerable volumes of sediment were eroded from val- might lead to enhanced resource availability over large regions; ley floor fills (Fanning and Holdaway 2001). Correlations exist however, at these times more than adequate resources might be between the abundance of hearth ages and past continental scale available at more places than the available population could environmental shifts (Holdaway et al. 2010). These studies in- exploit at one time. In such a situation, no one place would be dicate the likely mechanisms responsible for the exposure, ero- significantly better than another. Using a “dots on map” ap- sion, reburial, and destruction of hearths in Rutherfords Creek. proach, the landscape would appear to be covered by a carpet of The hearths visible in the eroded valley floor sediments result occupation debris just as we see in the distribution of hearths from the intersection of all these processes. The interaction in Rutherfords Creek, as though all places were used at once. between Aboriginal people, the technological forms of fire con- However, as the results of the analyses discussed above show, trol they utilized, and the environment explains why they made Holdaway, Davies, and Fanning Aboriginal Use of Fire in a Landscape Context S239 hearths in the way that they did, but their visibility in the ar- gamut of broadcast fires used to modify the environment chaeological record is not likely to be a direct outcome of this (Scherjon et al. 2015). Instead, they reflect an alternative way in behavior. It is rather a consequence of the way the archaeo- which people in the past could use controlled fire (e.g., Milburn, logical record is formed and transformed through time. These Doan, and Huckabee 2009; Thoms 2008), apparent only if results suggest that Australian studies where the frequency of suitable locations in landscapes amenable to the use of this radiocarbon data obtained from hearths or other features are form of fire technology are investigated. Based on our studies used to support proposed changes in population or occupation of hearths and stone artifacts, the hearths in western NSW are intensity (e.g., Johnson and Brook 2011; Smith et al. 2008; consistent with people who practiced high levels of mobility. Williams 2013) require further consideration. That geomorphic We therefore need to tread carefully when assuming that fire process can determine the age and frequency of the archaeo- use will fall into the categories that we expect from our own logical record connected with fire even in very recent periods also experience or from simple archaeological divisions such as “on has implications for understanding the Paleolithic fire record. site” and “off site” that envisage movement to and from central places of occupation or indeed from limited readings of the ethnographic record involving only domestic fire features as- Implications for Understanding the Paleolithic Use of Fire sociated with these occupations. Absence of fire should not To determine when fire was used, we need to consider how be concluded when consideration is limited to such a small it was encountered, made, and controlled and how often and number of categories within a small range of spatial locations. where it was used, and all of these criteria are of course related This observation has relevance to interpretations of sites in to why fire was used, for what purposes, and in what contexts. Europe during the period 0.8–0.5 Ma that preserve large quan- Here the implications of the detailed case study presented tities of bone, none of which are charred. For some, this reflects above need to be considered. With some caveats, the case for the absence of fire, but as Gowlett and Wrangham (2013) note, the presence of human-controlled fire can be made if suitable the absence of evidence may alternatively indicate that fire was features are present, but the case for its absence is much harder used in places that almost never survive. They note, for ex- to demonstrate (Gowlett and Wrangham 2013). A great deal of ample, how fire at Beeches Pit in eastern Britain was used near a research has concentrated on determining the makeup of the water edge. Other similar locations exist in Africa (they note hearths themselves, involving ethnographic accounts (e.g., Florisbad), but in the main, evidence for fires in Europe comes Black and Thoms 2014; Mallol et al. 2007; Thoms 2008), ex- from caves. In these contexts, fires are more extensive during perimental studies (e.g., Brodard et al. 2015; Graesch et al. 2014; the Middle Paleolithic, although not continuously so in some Homsey 2009; March et al. 2014), and micromorphological sites (Sandgathe et al. 2011). What may have changed was the studies of likely hearth features (e.g., Aldeias et al. 2012; Berna locations where fires were created and therefore the probabil- and Goldberg 2007; Friesem, Zaidner, and Shahack-Goss 2014; ity of their survival in the archaeological record. Goldberg et al. 2012; Mentzer 2014). However, much less at- The Australian example also emphasizes the need to con- tention has been given to studying the distribution and visibility sider nonhuman, proxy-specific formation processes when con- of hearth features at a landscape scale. Based on the NSW study, sidering evidence of absence. The fire record in western NSW is we could modify Gowlett and Wrangham’s (2013:10) state- extensive, but it is probably patterned as a result of differential ment that “Hearths are the most valuable archaeological indi- preservation and visibility. The erosion system that is likely cator of fire use, but may be a small part of a general picture in responsible for preservation did not exist in all places and times which fire was also exploited on landscapes” to read “fire use during the Paleolithic or any other period and place (although in occupation sites may be only a small part of a general picture of course we would not exclude the possibility that such a sys- of hearth use in landscapes.” Those landscape studies that tem operated at particular times and places). However, the do exist often seek behavioral explanations for hearth distri- example underlines how we should not assume that the ar- bution, frequency, and age and typically only consider geomor- chaeological record is patterned by human behavior alone. To phic explanations to assess whether concentrations of heat- test whether human behavior is indeed responsible for archae- fractured rocks are anthropogenic in origin (e.g., Black and ological patterning, we need to employ approaches using a Thoms 2014; Schaefer et al. 2014; Sullivan et al. 2001). neutral model similar to that illustrated here. To better un- The hearths studied in western NSW were probably not derstand the mechanisms that lead to the presence and absence related to domestic activity as such. What we call heat-retainer of hearths and also their apparent increase through time in the hearths represent a cooking technology that, given their loca- Australian record, the use of models needs to be combined with tions, the nature of the environment, and the types of stone observations made from multiple locations so that the extent artifacts with which they are spatially associated, signals some- and nature of the processes involved in forming the archaeo- thing other than domestic camps. Thus, the hearths distributed logical record can be understood. This cannot be achieved by throughout the valley floor do not fall easily into the categories studying the record from one locality because absence at any of fire features that dominate the discussion of fire use within one period from such a place could be due to the interaction of Paleolithic archaeological sites (e.g., Berna and Goldberg 2007; multiple processes operating at different scales, many of them Sandgathe et al. 2011; Wadley 2012), nor do they fit within the natural rather than cultural. Indeed, as we demonstrate, “ab- S240 Current Anthropology Volume 58, Supplement 16, August 2017 sence” can even vary depending on what dating system is used. archaeological record of fire use. We await the result of such That a neutral model of hearth construction can produce pat- studies with interest. terns homologous to the empirical archaeological record from a late Holocene context should be of concern to those analyzing records from the deep past where the accumulated effect of Acknowledgments combined formation processes has the potential to be so much greater. We are grateful to Barkindji Aboriginal Traditional Owners of The early records of fire control seem to be very discontin- Country in our western NSW study areas for their permission fi uous through time, leading some to propose that fire use was to conduct research. The paper bene ted from discussions over either not required or intermittent (e.g., Roebroeks and Villa a number of years with Murray Butcher and Warlpa Thomp- 2011; Sandgathe et al. 2011). From a formation perspective, son. The Australian research was funded by an Australian Re- such an intermittent record may speak of issues connected with search Council Discovery Project Grant and Macquarie Uni- preservation and visibility more than it does with the nature versity and University of Auckland research grants to Simon J. of human behavior. Petraglia (2002) reviews the use of what Holdaway and Patricia C. Fanning. The simulation work was he terms “thermally altered stone,” here referred to as heat funded by a University of Auckland doctoral scholarship. Thanks retainers, providing case studies from mid-Atlantic Holocene to Dennis Sandgathe and Francesco Berna for the invitation “ ” sites in North America. All three of the sites he discusses are to attend the Fire and the Genus Homo symposium and to on terraces where plowing has revealed the thermally altered Laurie Obbink and the Wenner-Gren Foundation for support. rock. Petraglia (2002) comments that western European Mid- dle Paleolithic sites show limited evidence for heat retainers compared with those in the eastern United States, with much References Cited greater levels seen only from the Early Upper Paleolithic. He Aldeias, Vera, Paul Goldberg, Dennis Sandgathe, Francesco Berna, Harold also comments on the lack of thermally altered stones in the L. Dibble, Shannon P. McPherron, Alain Turq, and Zeljko Rezek. 2012. Evidence for Neandertal use of fire at Roc de Marsal (France). Journal of earlier Paleolithic sites in open-air locations. This could rep- Archaeological Science 39(7):2414–2423. resent an evolutionary change in the use of fire technology, but Allen, Harry. 1972. Where the crow flies backwards: man and land in the it could equally reflect changes in land-use strategies and en- Darling Basin. PhD thesis, Australian National University, Canberra. Attenbrow, Val, and Peter Hiscock. 2015. Dates and demography: are radio- suing changes in the availability of fuel and stone as well as food metric dates a robust proxy for long-term prehistoric demographic change? sources and shifts in the nature of contexts that preserve ma- Archaeology in Oceania 50:30–36. terial from the past. It is the combination of such changes that Berna, Francesco, and Paul Goldberg. 2007. Assessing Paleolithic pyrotechnol- ogy and associated hominin behavior in Israel. Israel Journal of Earth Sciences needs to be considered when assessing the absence of evidence 56(2):107–121. as well as the first appearance of fire use, for example, the ap- Binford, Lewis. R. 1983. Working at archaeology: the generation gap—reac- pearance of fire use after 400 Ka in Europe discussed above. As tionary arguments and theory building. In Working at archaeology. Lewis R. Binford, ed. Pp. 213–227. New York: Academic Press. the Australian example shows, all these aspects are intercon- Black, Stephen, and Alston Thoms. 2014. Hunter-gatherer earth ovens in nected, so we should not simply compare the presence and the archaeological record: fundamental concepts. American Antiquity 79(2): absence of fire features from different time periods and places 204–226. Bowman, David M. J. S., Jennifer K. Balch, Paulo Artaxo, William J. Bond, as though they are equivalent. Comparisons need to acknowl- Jean M. Carlson, Mark A. Cochrane, Carla M. D’Antonio, et al. 2009. Fire edge the potential complexity of hunter-gatherer behaviors no in the earth system. Science 324(5926):481–484. matter what the time period involved and recognize that what Bowman, David M. J. S., Jennifer Balch, Paulo Artaxo, William J. Bond, Mark A. Cochrane, Carla M. D’Antonio, Ruth DeFries, et al. 2011. The human di- we see archaeologically are windows into settlement systems mension of fire regimes on Earth. Journal of Biogeography 38(12):2223–2236. that were likely spatially extensive, as Binford (1983) long ago Brantingham, P. Jeffrey. 2003. A neutral model of stone raw material procure- observed. However, how settlement systems are manifest also ment. American Anthropologist 68(3):487–509. Brink, Jack W., and Bob Dawe. 2003. Hot rocks as scarce resources: the use, probably depends on the geomorphic processes, as the current re-use and abandonment of heating stones at Head-Smashed-In Buffalo Jump. study suggests. A great deal of attention in the literature is given Plains Anthropologist 48(186):85–104. to analyzing evidence that shows the presence of fire features. Brodard, Auréilie, Delphine Lacanette-Puyo, Pierre Guibert, François Lévêque, Albane Burens, and Laurent Carozza. 2015. A new process of reconstructing The results of this study suggest we should be giving equal archaeological fires from their impact on sediment: a coupled experimental attention to understanding, as Gowlett and Wrangham (2013) and numerical approach based on the case study of hearths from the cave have recently argued, what the absence of such features might of Les Fraux (Dordogne, France). Archaeological and Anthropological Sciences 8(4):673–687. mean. This involves considering a wide range of evidence that, Contreras, Daniel A., and John Meadows. 2014. Summed radiocarbon calibra- ironically, may seem unconnected (at least directly) to the human tions as a population proxy: a critical evaluation using a realistic simulation use of fire. It also means beginning with a model of formation approach. Journal of Archaeological Science 52:591–608. Davies, Ben, Simon J. Holdaway, and Patricia C. Fanning. 2015. Modelling the that does not assume the primacy of human behavior. While palimpsest: an exploratory agent-based model of surface archaeological this may seem contrary to the anthropological study of early deposit formation in a fluvial landscape. Holocene 26(3):450–463, doi:10.1177 human behavior, we will always run the risk of identifying false /0959683615609754. Delgado, Miguel, Francisco Javier Aceituno, and Gustavo Barrientos. 2015. negatives without understanding the wider systems in which 14C data and the early colonization of northwest South America: a critical fire operated and that are involved in the preservation of the assessment. Quaternary International 363:55–64. Holdaway, Davies, and Fanning Aboriginal Use of Fire in a Landscape Context S241

Douglass, Matthew, Simon J. Holdaway, Justin Shiner, and Patricia C. Fanning. intersections and transformations. David Frankel, Susan Lawrence, and 2015. Quartz and silcrete raw material use and selection in late Holocene Jennifer Webb, eds. Pp. 51–68. Melbourne: Routledge. assemblages from semi-arid Australia. Quaternary International 424:12–23, Holdaway, Simon J., Matthew J. Douglass, and Rebecca Phillipps. 2015. Flake doi:10.1016/j.quaint.2015.08.041 selection, assemblage variability and technological organization. In Works Douglass, Matthew J., Simon J. Holdaway, Patricia C. Fanning, and Justin I. in stone: contemporary perspectives on lithic analysis. Michael J. Shott, ed. Shiner. 2008. An assessment and archaeological application of cortex mea- Pp. 46–62. Salt Lake City: University of Utah Press. surement in lithic assemblages. American Antiquity 73(3):513–526. Holdaway, Simon J., and Patricia C. Fanning. 2014. A geoarchaeology of Ab- Fanning, Patricia C. 1994. Long-term contemporary erosion rates in an arid original landscapes in semiarid Australia. Collingwood, Victoria: CSIRO. rangelands environment in western New South Wales, Australia. Journal of Holdaway, Simon J., Patricia C. Fanning, Martin Jones, Justin Shiner, Dan C. Arid Environments 28(3):173–187. Witter, and Geoff Nicholls. 2002. Variability in the chronology of Late Ho- ———. 1999. Recent landscape history in arid western New South Wales, locene Aboriginal occupation on the arid margin of southeastern Australia. Australia: a model for regional change. Geomorphology 29(3/4):191–209. Journal of Archaeological Science 29(4):351–363. Fanning, Patricia, and Simon J. Holdaway. 2001. Temporal limits to the ar- Holdaway, Simon J., Patricia C. Fanning, and Edward J. Rhodes. 2008. Chal- chaeological record in arid western NSW, Australia: lessons from OSL and lenging intensification: human-environment interactions in the Holocene radiocarbon dating of hearths and sediments. In Australasian connections geoarchaeological record from western New South Wales, Australia. Holocene and new directions: Proceedings of the 7th Australasian Archaeometry Con- 18(3):403–412. ference. Martin Jones and Peter Sheppard, eds. Pp. 85–104. Research in Holdaway, Simon J., Patricia C. Fanning, Edward J. Rhodes, S. K. Marx, Bruce Anthropology and Linguistics 5. Auckland: Department of Anthropology, Floyd, and Matthew J. Douglass. 2010. Human response to palaeoenviron- University of Auckland. mental change and the question of temporal scale. Palaeogeography, Palaeo- ———. 2004. Artifact visibility at open sites in western New South Wales, climatology, Palaeoecology 292(1/2):192–200. Australia. Journal of Field Archaeology 29:255–271. Holdaway, Simon J., Patricia C. Fanning, and Dan Witter. 2000. Prehistoric Fanning, Patricia C., Simon J. Holdaway, and Rebecca Phillipps. 2009. Heat- Aboriginal occupation of the rangelands: interpreting the surface archaeo- retainer hearth identification as a component of archaeological survey in logical record of far western New South Wales, Australia. Rangeland Journal western NSW, Australia. In New directions in archaeological science. Andrew 22(1):44–57. Fairbairn and Sue O’Connor, eds. Pp. 13–23. Terra Australis 28. Canberra: Holdaway, Simon J., Geoffrey C. P. King, Matthew J. Douglass, and Patricia C. Australian National University E-Press. Fanning. 2015. Human-environment interactions at regional scales: the com- Fanning, Patricia C., Simon J. Holdaway, and Edward J. Rhodes. 2007. A geo- plex topography hypothesis applied to surface archaeological records in Aus- morphic framework for understanding the surface archaeological record in tralia and North America. Archaeology in Oceania 50:58–69. arid environments. Geodinamica Acta 20(4):275–286. Homsey, Lara K. 2009. The identification and prehistoric selection criteria Fanning, Patricia C., Simon J. Holdaway, Edward J. Rhodes, and Tessa G. of fire-cracked rock: an example from Dust Cave, Alabama. Southeastern Bryant. 2009. The surface archaeological record in arid Australia: geomor- Archaeology 28(1):101–116. phic controls on preservation, exposure, and visibility. Geoarchaeology 24(2): Isaac, Glynn L. 1983. Bones in contention: competing explanations for the 121–146. juxtaposition of Early Pleistocene artefacts and faunal remains. In Animals Fitzsimmons Kathryn E., Nicola Stern, Colin Murray-Wallace, William Truscott, and archaeology: hunters and their prey. Juliet Clutton-Brock and Caroline and Cornel Pop. 2015. The Mungo Mega-Lake Event, semi-arid Australia: non- Grigson, eds. Pp. 3–19. British Archaeological Reports International Series linear descent into the last Ice Age, implications for human behaviour. PLoS 163. Oxford: British Archaeological Reports. ONE 10(6):e0127008, doi:10.1371/journal.pone.012700. Jaynes, E. T. 2003. Probability theory: the logic of science. Cambridge: Cambridge Flannery, Tim. 1994. The future eaters: an ecological history of the Australasian University Press. lands and people. Port Melbourne: Reed. Johnson, Christopher N., and Barry W. Brook. 2011. Reconstructing the dy- Friesem, David. E., Yossi Zaidner, and Ruth Shahack-Gross. 2014. Formation namics of ancient human populations from radiocarbon dates: 10 000 years processes and combustion features at the lower layers of the Middle Palaeo- of population growth in Australia. Proceedings of the Royal Society B 278 lithic open-air site of Nesher Ramla, Israel. Quaternary International 331: (1725):3748–3754, doi:10.1098/rspb.2011.0343. 128–138. Kershaw, A. Peter, James S. Clark, A. Malcolm Gill, and Donna M. D’Costa. Gamble, Clive, William Davies, Paul Pettitt, and Martin Richards. 2004. Cli- 2002. A history of fire in Australia. In Flammable Australia: the fire regimes mate change and evolving human diversity in Europe during the last glacial. and biodiversity of a continent. Ross A. Bradstock, Jann E. Williams, and Philosophical Transactions of the Royal Society B 359(1442):243–254. Malcolm A. Gill, eds. Pp. 3–25. Cambridge: Cambridge University Press. Goldberg, Paul, Harold Dibble, Francesco Berna, Denis Sandgathe, Shannon J. Lake, Mark W. 2015. Explaining the past with ABM: on modelling philosophy. P. McPherron, and Alain Turq. 2012. New evidence on Neandertal use of fire: In Agent-Based modeling and simulation in archaeology. Gabriel Wurzer, examples from Roc de Marsal and Pech de l’Azé IV. Quaternary Interna- Kerstin Kowarik, and Hans Reschreiter, eds. Pp. 3–35. Advances in Geo- tional 247:325–340. graphic Information Science. New York: Springer. Gose, Wulf. 2000. Palaeomagnetic studies of burned rocks. Journal of Archae- Mallol, Carolina, Frank W. Marlowe, Brian M. Wood, and Claire C. Porter. 2007. ological Science 27:409–421. Earth, wind, and fire: ethnoarchaeological signals of Hadza fires. Journal of Gould, Richard A. 1971. Uses and effects of fire among the Western Desert Archaeological Science 34:2035–2052. aborigines of Australia. Mankind 8:14–24. March, Ramiro, Alexandra Lucquin, Delphine Joly, Juan Carlos Ferreri, and Gowlett, John A. J., and Richard W. Wrangham. 2013. Earliest fire in Africa: Mohamad Muhieddine. 2014. Processes of formation and alteration of ar- towards the convergence of archaeological evidence and the cooking hy- chaeological fire structures: complexity viewed in the light of experimental pothesis. Azania: Archaeological Research in Africa 48:5–30. approaches. Journal of Archaeological Method and Theory 21:1–45. Graesch, Anthony P., Tianna DiMare, Gregson Schachner, David M. Schaepe, Mentzer, Susan. 2014. Microarchaeological approaches to the identification and John Dallen. 2014. Thermally modified rock: the experimental study of and interpretation of combustion features in prehistoric archaeological “fire-cracked” byproducts of hot rock cooking. North American Archaeol- sites. Journal of Archaeological Method and Theory 21:616–668. ogist 35(2):167–200. Milburn, Douglas M., U. K. Doan, and Joanna Huckabee. 2009. Spatial and Hallam, Sylvia, 1975. Fire and hearth: a study of Aboriginal usage and Euro- temporal distributions of archaeological heated-rock cooking structures in the pean usurpation in South-Western Australia. Canberra: Australian Institute Transverse Mountain Range: proposed markers of land-use shifts since the of Aboriginal Studies. Early Holocene. Proceedings of the Society of California Archaeology 22:1–21. Holdaway, Simon J., and Matthew Douglass. 2012. A twenty-first century Mooney, S. D., S. P. Harrison, P. J. Bartlein, A. L. Daniau, J. Stevenson, K. C. archaeology of stone artifacts. Journal of Archaeological Method and Theory Brownlie, S. Buckman, et al. 2011. Late Quaternary fire regimes of Australasia. 19(1):101–131. Quaternary Science Reviews 30:28–46. ———. 2015. Use beyond manufacture: non-flint stone artifacts from Fowlers Mooney, Scott D., Sandy P. Harrison, Patrick J. Bartlein, and Janelle Stevenson. Gap, Australia. Lithic Technology 40(2):94–111. 2012. The prehistory of fire in Australasia. In Flammable Australia: fire Holdaway, Simon J., Matthew J. Douglass, and Patricia C. Fanning. 2013. A regimes, biodiversity and ecosystems in a changing world. Ross A. Bradstock, new ecological framework for understanding human-environment inter- A. Malcolm Gill, and Richard J. Williams, eds. Pp. 3–25. Collingwood: actions in arid Australia. In Archaeology in environment and technology CSIRO. S242 Current Anthropology Volume 58, Supplement 16, August 2017

Mulrooney, Mara A. 2013. An island-wide assessment of the chronology of in the western Colorado Desert, Alta California. California Archaeology 6: settlement and land use on Rapa Nui (Easter Island) based on radiocarbon 65–94. data. Journal of Archaeological Science 40(12):4377–4399. Scherjon, Fulco, Corrie Bakels, Katharine MacDonald, and Wil Roebroeks. Peake-Jones, K., ed. 1988. To the desert with Sturt: a diary of the 1844 expe- 2015. Burning the land: an ethnographic study of off-site fire use by current dition. Adelaide: South Australian Government Printer. and historically documented foragers and implications for the interpretation Petraglia, Michael D. 2002. The heated and the broken: thermally altered of past fire practices in the landscape. Current Anthropology 56:299–326. stone, human behavior, and archaeological site formation. North American Singh, G., A. P. Kershaw, and R. Clark. 1981. Quaternary vegetation and fire Archaeologist 23:241–269. history in Australia. In Fire and the Australian biota. A. M. Gill, R. H. Groves, Premo, Luke S. 2007. Exploratory agent-based models: towards an experi- and I. R. Noble, eds. Pp. 23–54. Canberra: Australian Academy of Science. mental ethnoarchaeology. In Digital discovery: exploring new frontiers in Smith, M. A., A. N. Williams, Chris S. M. Turney, and M. L. Cupper. 2008. human heritage: CAA 2006. Jeffrey T. Clark and E. M. Hagermeister, eds. Human-environment interactions in Australian drylands: exploratory time- Pp. 29–36. Budapest: Archaeolingua. series analysis of archaeological records. Holocene 18:389–401. Rhode, David, P. Jeffrey Brantingham, Charles Perreault, and David B. Madsen. Sturt, Charles N. 1834. Two expeditions into the interior of southern Australia, 2014. Mind the gaps: testing for hiatuses in regional radiocarbon date se- during the years 1828, 1829, 1830, and 1831, with observations on the soil, quences. Journal of Archaeological Science 52:567–577. climate, and general resources of the colony of New South Wales,vol.2.London: Rhodes, Edward J., Patricia C. Fanning, and Simon J. Holdaway. 2010. Devel- Smith, Elder, & Co. opments in optically stimulated luminescence age control for geoarchaeo- Sullivan, Alan P., Robert A. Cook, Matthew P. Purtill, and Patrick M. Uphus. logical sediments and hearths in western New South Wales, Australia. Qua- 2001. Economic and land-use implications of prehistoric fire-cracked-rock ternary Geochronology 5(2/3):348–352. piles, Northern Arizona. Journal of Field Archaeology 28:367–382. Rhodes, Edward J., Patricia C. Fanning, Simon J. Holdaway, and Cynthia Thoms, Alston V. 2008. The fire stones carry: ethnographic records and ar- Bolton. 2009. Ancient surfaces? dating archaeological surfaces in western chaeological expectations for hot-rock cookery in western North America. NSW using OSL. In New directions in archaeological science. A. Fairbairn and Journal of Anthropological Archaeology 27:443–460. S. O’Connor, eds. Pp. 189–200. Terra Australis 28. Canberra: Australian ———. 2009. Rocks of ages: propagation of hot-rock cookery in western National University E-Press. North America. Journal of Archaeological Science 36:573–591. Rick, John W. 1987 Dates as data: an examination of the Peruvian preceramic Wadley, Lyn. 2012. Some combustion features at Sibudu, South Africa, be- radiocarbon record. American Antiquity 52(1):55–73. tween 65,000 and 58,000 years ago. Quaternary International 247:341–349. Roebroeks, Wil, and Paola Villa. 2011. On the earliest evidence for habitual Williams, Alan N. 2013. A new population curve for prehistoric Australia. use of fire in Europe. Proceedings of the National Academy of Sciences of the Proceedings of the Royal Society B 280(1761), doi:10.1098/rspb.2013.0486. USA 108:5209–5214. Williams, Alan N., Sean Ulm, Chris S. M. Turney, David Rohde, and Gentry Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon McPherron, White. 2015. Holocene demographic changes and the emergence of com- Alain Turq, Laura Niven, and Jamie Hodgkins. 2011. On the role of fire in plex societies in prehistoric Australia. PLoS ONE 10(6):e0128661, doi:10.1371 Neandertal adaptations in western Europe: evidence from Pech de l’Azé IV /journal.pone.0128661. and Roc de Marsal, France. Paleoanthropology 2011:216–242. Wrangham, Richard W., James Holland Jones, Greg Laden, David Pilbeam, and Schaefer, Jerry, Don Laylander, Sherri Andrews, James Daniels, and Tony T. NancyLou Conklin-Brittain. 1999. The raw and the stolen: cooking and the Quach. 2014. Identifying, dating, and explaining fire-affected rock features ecology of human origins. Current Anthropology 40:567–594. Current Anthropology Volume 58, Supplement 16, August 2017 S243

Researching the Nature of Fire at 1.5 Mya on the Site of FxJj20 AB, Koobi Fora, Kenya, Using High-Resolution Spatial Analysis and FTIR Spectrometry

by Sarah Hlubik, Francesco Berna, Craig Feibel, David Braun, and John W. K. Harris

Online enhancements: supplemental figures

Some scholars explain the major anatomical characteristics that differentiate Homo erectus from its predecessor, Homo habilis, as the result of Homo erectus being adapted to use fire for cooking and other tasks. However, many scholars contend that the evidence of fire in Homo erectus sites is very scant and is not convincingly anthropogenic. This study presents a methodology to evaluate the evidence of fire associated with the 1.5-million-year-old Homo erectus site FxJj20 AB, Koobi Fora, Kenya. The evidence is in the form of thermally altered lithics, soil aggregates, and bone fragments identified using visual inspection and Fourier transform infrared spectroscopy (FTIR). We conducted high- resolution excavation focused on the recovery and high-resolution mapping of large and small finds (! 2 cm). ArcGIS spatial analysis and soil micromorphology were used to assess whether the evidence of fire at the site has a natural or anthropogenic origin. Preliminary results indicate that the spatial pattern of heated and unheated archaeological material is not inconsistent with prehistoric anthropogenic fire features found in archaeological sites of Europe and West Asia.

Several studies show fairly convincingly that modern humans fire use has deep roots in the Homo lineage (Wrangham 2017). depend physically and biologically on the controlled use of fire Morphological changes (larger body and brain, smaller teeth (Carmody and Wrangham 2009; Wrangham and Carmody and gut) in Homo erectus beginning around 2.0 Mya suggest 2010; Wrangham and Conklin-Brittain 2003; Wrangham et al. that major changes in feeding strategies such as cooking or 1999). However, the timing of the initial use of fire by hominins other food processing behaviors began at or before this time is contentious at best (James 1989; Sandgathe 2017; Sandgathe (Aiello and Wells 2002; Aiello and Wheeler 1995; Organ et al. et al. 2011). Some scholars hypothesize that the dependence on 2011). While an increase in meat eating (fat and protein) would have provided a greater number of valuable calories (e.g., Blu- menschine and Pobiner 2007; Milton 1999; Speth 1989), with- Sarah Hlubik is a graduate student in the Anthropology Depart- out hunting to ensure continuous access to fresh meat resources, ment at Rutgers University (Ruth Adams Building, 131 George Street, hominins would have been subject to toxins and pathogens New Brunswick, New Jersey 08901, USA [[email protected]]). that develop in spoiling meat (Ragir, Rosenberg, and Tierno Francesco Berna is Assistant Professor in the Department of Archae- 2000; Smith et al. 2015). Using fire to cook meat and vegetal ology at Simon Fraser University (8888 University Drive, Burnaby, foods would have reduced their potential toxicity and, most British Columbia V5A 1S6, Canada [[email protected]]). Craig importantly, increased the caloric return of these foods. This Feibel is Professor in the Geology Department at Rutgers University would reduce the amount of metabolic energy needed for di- (Ruth Adams Building, 131 George Street, New Brunswick, New Jersey gestion by diminishing chewing time and by breaking down David Braun 08901, USA [[email protected]]). is Associate Pro- complex starches and proteins into more efficiently digestible fessor in the Department of Anthropology at George Washington compounds (Carmody and Wrangham 2009; Wrangham and University (2110 G Street NW, Washington, DC 20052, USA [david [email protected]]). John W. K. Harris is Professor Emeritus in the Conklin-Brittain 2003; Wrangham et al. 1999). Evidence for Anthropology Department at Rutgers University (Ruth Adams Build- acquisition of meat resources dates to at least 2.6 Mya (Semaw ing, 131 George Street, New Brunswick, New Jersey 08901, USA et al. 2003) and potentially as early as 3.4 Mya (McPherron et al. [[email protected]]). This paper was submitted 25 VII 16, ac- 2010), while persistent carnivory is shown in the record begin- cepted 22 III 17, and electronically published 30 VI 17. ning about 2.0 Mya (Ferraro et al. 2013). The gap between in- q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0008$10.00. DOI: 10.1086/692530 S244 Current Anthropology Volume 58, Supplement 16, August 2017 cluding some meat in the diet and persistent carnivory indi- are not ideal settings to investigate evidence of controlled fire. cates that some barrier existed to the incorporation of meat in The Atapuerca sites, in northern Spain, contain dispersed char- the hominin diet before 2 Mya. It is possible that the acquisi- coal throughout their sequences that cannot be considered evi- tion of the use of fire in the behavioral repertoire of hominins dence for control of fire on a site (Roebroeks and Villa 2011). was responsible for this. To test this hypothesis, we must es- To determine whether fire was used by ancient hominins, fire tablish whether the use of fire, in some capacity, is present in evidence needs to be found in minimally disturbed anthropo- the Early Pleistocene archaeological record associated with H. genic contexts with a demonstrable association with hominin erectus or H. habilis. activities (Berna et al. 2012). In this paper we illustrate the nature of the 1.5 Mya evidence Researchers at Gesher Benot Ya‘akov (GBY) found a num- for fire at the site of FxJj20 AB, located in Koobi Fora, Kenya. ber of burned microartifacts (artifacts !2 cm in maximum The study includes high-resolution mapping and spatial anal- dimension) clustered together, while macroartifacts (artifacts ysis, soil micromorphology, and Fourier transform infrared 12 cm in maximum dimension) were found mostly unburned spectroscopy (FTIR) characterization of the archaeological and away from the burned microartifacts (Alperson-Afil 2017; record. We carefully examine the nature of fire evidence and its Alperson-Afil and Goren-Inbar 2010; Goren-Inbar et al. 2004). distribution and relation to hominin activities to determine The researchers proposed that this pattern mirrored the toss whether there is a statistically significant association between and drop zones proposed by Binford (1983). Macroscopic the spatial distribution of fire evidence and evidence of other components of the purported hearths at GBY no longer re- human activities (e.g., flint knapping or butchery). We rec- mained intact, but the pattern of burned and unburned ar- ognize the difficulty of claiming that ancient hominins con- tifacts was preserved and used to pinpoint their original lo- trolled fire at 1.5 Mya in the absence of numerous sites with cations (Alperson-Afil, Richter, and Goren-Inbar 2007). The similar patterns. Nevertheless, the work presented here rep- site provides strongly suggestive evidence for the controlled resents a fundamental data point and, more importantly, a use of fire at 780 kya even in the absence of commonly used methodological benchmark for the investigation of the nature proxies such as microscopic wood ash and/or burned sedi- of fire evidence in Early Pleistocene open-air sites of Africa and ment. Eurasia. Wonderwerk Cave contains evidence of fire in Acheulean contexts dating back to ca. 1 Mya (Berna et al. 2012). Micro- Background morphological investigation showed the presence of centimeter- thick intact paleosurfaces containing calcified and ashed plant Archaeological research in Europe and Southwestern Asia remains (Berna et al. 2012; Goldberg, Berna, and Chazan 2015; shows that control and maintenance of fire begins sometime in Thibodeau 2016). The microscopic evidence, combined with ev- the middle to late Middle Paleolithic (Roebroeks and Villa idence of bone burned at temperatures above 4007C and iron- 2011; Shimelmitz et al. 2014; Sorensen, Roebroeks, and van stone manuports with characteristic potlid fractures (resulting Gijn 2014). Qesem Cave, Israel, contains evidence of habitual from the exposure to high heat), suggests that hominins using use of fire (but see Sandgathe 2017) and hearth-like features Wonderwerk Cave were using fire in some capacity (Berna et al. associated with an Achuelo-Yabrudian industry at 300 kya 2012). (Barkai et al. 2017; Karkanas et al. 2007; Shahack-Gross et al. The Koobi Fora FxJj20 site complex spurred the debate over 2014; Shimelmitz et al. 2014). True hearth features appear in fire use by early hominins soon after the original excavations of the latter half of the Middle Pleistocene with the appearance of all three sites in 1972 and 1973. The sites FxJj20 East and Neanderthals (James 1989), while sites such as Campitello, FxJj20 Main were subsequently excavated in the 1970s and Italy (MIS 6; Mazza et al. 2006), have evidence of materials into the 1980s as more finds were discovered (Harris 1997). such as mastics that are not possible to produce without fire. During excavations at FxJj20 East and FxJj20 Main, discrete However, some evidence for potential use of fire at archaeo- concentrations of rubified sediment aggregates were found logical sites from the Early and Middle Pleistocene throughout throughout the site (Bellomo and Kean 1997; Harris 1978, 1997). the Old World extends much farther back in time (Alperson- It was proposed that these concentrations were the remains Afil and Goren-Inbar 2006, 2010; Berna et al. 2012; Clark and of ancient fires, and subsequent analysis using magnetometry, Harris 1985; Clark and Kurashina 1979; Goldberg, Miller, and thermoluminescence (TL; Bellomo 1994; Bellomo and Kean Mentzer 2017; Gowlett et al. 1981; James 1989; Preece et al. 1997), and phytoliths (Rowlett 2000) indicated that these may 2006; Roebroeks and Villa 2011; see table 1). At many of these indeed be related to fire. Spatial analysis of the materials re- sites, however, the anthropogenic nature of the fire is called covered during the excavation (Bellomo 1994) indicated that into question because of the contexts in which the residues are stone tool and bone concentrations were closely associated found (Isaac 1982; Roebroeks and Villa 2011; Stahlschmidt with the rubified sediment concentrations, further indicating et al. 2015). At Swartkrans, South Africa (1.0–1.5 Mya), fire that these patches may be the result of fire associated with residues are found in deposits filling karstic sinkholes (Brain hominin activities. 1993; Gibbon et al. 2014), while at Chesowanja, Kenya (~1.2 Mya), Earlier excavations at the FxJj20 sites did not have the ad- they are found in alluvial settings (Gowlett et al. 1981). These vantage of modern technology to aid in mapping and were not Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S245

Table 1. List of sites before 400 kya with evidence of fire

Location Period Evidence of fire Africa: Koobi Fora, FxJj20, Kenyaa 1.5–1.6 Mya Burned lithics, reddened sediment Chesowanja, GnJi1/6E, Kenyaa,b 1.42 Mya Reddened sediment Gadeb, Ethiopiaa .7–1.5 Mya Burned lithics Swartkrans, South Africac,d 1.0 Mya Burned bones Wonderwerk, South Africae 1.0 Mya Ash, charcoal, burned bone Southwest Asia: Gesher Benot Ya‘aqov, Israelf 780 kya Burned lithics, charcoal, burned seeds Qesem Cave, Israelg 400–200 kya Hearths, ash, burned bone Tabun, Israelh 350 kya Hearth, ash, phytoliths Hayonim, Israeli 250–100 kya Ash, charcoal, phytoliths Europe: Atapuerca, Spainj 1.2–.780 Mya Dispersed charcoal Isernia, Italy j 606 kya Burned bone, burned sediment Boxgrove, Englandj MIS 13 Charcoal Highlodge, Englandj MIS 13 Charcoal Beeches Pit, Englandj,k MIS 11 Burned lithics, burned bone, burned sediment, hearth Verteszöllös, Hungaryj MIS 9–11 Burned bone, hearth Biltzingleben, Germany j MIS 9–11 Charcoal, burned lithics, burned bone Terra Amata, France j 380–239 kya Charcoal, burned lithics, burned bone, hearth Orgnac, France j MIS 9–8 Burned bone, ash Petit Bost, Francej MIS 9/8 burned lithics a Clark and Harris 1985. b Gowlett et al. 1981. c Brain 1993. d Gibbon et al. 2014. e Berna et al. 2012. f Alperson-Afil and Goren-Inbar 2010. g Shahack-Gross et al. 2014. h Albert et al. 1999. i Shahack-Gross, Bar-Yosef, and Weiner 1997. j Roebroeks and Villa 2011. k Preece et al. 2006. undertaken with a methodology focused on the recovery of information systems enable the identification and visualiza- small finds. Artifacts and bone were plotted by hand and tion of density-related patterns (Anselin and Getis 1992; Henry smaller (!2 cm in maximum dimension) artifacts and bone 2011). were recovered in the screen. While items recovered in the Using methods such as soil micromorphology, FTIR, mag- screen are useful for many analyses, such as lithic technology netic susceptibility, paleomagnetism, and TL (Berna and Gold- and faunal lists, these materials are less suitable for high-resolution berg 2007; Goldberg, Miller, and Mentzer 2017; Shahack-Gross, spatial analysis as imprecise placement can skew analyses. Bar-Yosef, and Weiner 1997), fire itself is relatively easy to detect Since the original excavations at the FxJj20 sites, advance- in the archaeological record. Wildfire evidence is described as ments in recovery and analysis techniques have enhanced the diffuse and irregular (Gowlett et al. 2017), and recent paleo- ability to identify hominin fire use (Berna and Goldberg 2007; environmental studies show that with expansion of grasslands, Goldberg and Berna 2010; McPherron and Dibble 2002). Re- the East Africa landscapes became progressively fire prone at covery of all possible artifacts, piece plotted with the aid of the beginning of the Pleistocene (Magill, Ashley, and Freeman a total station, make it possible to map even the smallest 2013), so the probability that wildfire affected the remains of artifacts, ecofacts, samples, and feature boundaries with mil- prehistoric camps is higher in East Africa than in other geo- limeter accuracy (McPherron and Dibble 2002). Soil micro- graphic regions. Conversely, anthropogenic fire at base camps morphological techniques coupled with FTIR microscopy al- or in living spaces in the Early Pleistocene should be distin- low for the identification of microscopic fire evidence such as guishable as fairly localized evidence of fire closely associated wood ash, burned bone, burned soil, or microcharcoal (Berna with artifacts. Thus, to be able to distinguish between wildfire et al. 2012; Berna and Goldberg 2007; Goldberg and Berna and human-made fire, it is fundamental to produce high- 2010). The use of portable FTIR allows for the identification of resolution maps of heated and not heated macro- and micro- burned materials in the field or lab (Berna and Goldberg 2007; scopic artifacts and ecofacts. Evidence of exotic fuel brought to Weiner 2010). Spatial analysis using advanced geographical thesiteandfire-starting techniques (e.g., pyrite, marcasite, tin- S246 Current Anthropology Volume 58, Supplement 16, August 2017 der fragments) can be used to assess the anthropogenic origin of reddened sediment similar to those found at FxJj20 East of fire, especially if found in close association with evidence of and FxJj20 Main (J. W. K. Harris, FxJj20 site notes, 1973). Such human activities such as stone tools or butchered animal bone. concentrations were not found, but a preliminary spatial anal- All individual pieces of evidence need to be evaluated within ysis of the artifacts found at the site identified “Locus 1,” a the larger context of the site itself and in reference to other circular, low-density area surrounded by a higher-density evidence found. concentration in the northeast portion of the site (fig. 2). The artifacts making up a large concentration around Locus 1 are primarily smaller than 2 cm in maximum dimension (hence- Material and Methods forth “microartifacts”), while a larger concentration of larger artifacts (12 cm in maximum dimension) were recovered far- FxJj20 AB Site ther to the west of the largest concentration of microartifacts. All three FxJj20 sites are located on weakly lithified, fine- As all the surrounding squares had been excavated to the same grained, floodplain tuffaceous silts (Harris 1997). The site of elevation as the square containing Locus 1 and small artifacts as FxJj20 AB, discovered in 1973, is located about 150 m north- well as bone fragments surrounded Locus 1, we hypothesized east of the FxJj20 East and FxJj20 Main sites (fig. 1). The ar- that Locus 1 might be the location of a fire feature. On this tifacts and bone appear to be deposited on a gentle (~57) east to assumption, we have continued excavation and sampling of the west slope. A small 16 m2 test excavation was undertaken, and site to resolve this question. excavator notes describe recovering reddened clasts in the screen, though none were recorded in situ. The 1970s exca- Excavation Methods at FxJj20 AB Site vation resulted in a collection of lithics (n p 2,626) that are clearly the result of intentional flaking, often showing flake Between 2010 and 2015, 24 squares were excavated to the scars, but preserving no bulbs of percussion or platforms depth of the artifact-bearing layers, nine of which were fur- (Harris 1997). ther excavated to sterile layers. One was left unexcavated to In 2010, we resumed excavations at FxJj20 AB and expanded serve as a stratigraphic reference and allow for micromorpho- the site by 24 m2 to the east and north of the original excava- logy sample collection to document any microscopic strati- tion. We hoped to detect discrete, consolidated concentrations graphic features.

Figure 1. A, General location of the FxJj20 site complex, located on the Karari Ridge, in Koobi Fora, Northern Kenya. B, Location of the FxJj20 sites in relation to each other. FxJj20 AB is located approximately 150 m to the north and east of FxJj20 East and FxJj20 Main sites (image from GoogleEarth). C, Photo of the excavation of FxJj20 AB facing southwest across the site. A color version of this figure is available online. Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S247

Figure 2. Location of all artifacts and ecofacts found on FxJj20 AB. Potlids are pieces of stone debitage identified through lithic analysis as being consistent with experimentally derived potlids. Burned bone and burned sediment are identified through FTIR analysis. Locations of all materials from the new excavations are obtained through total station, while materials from original ex- cavation were plotted by hand. A color version of this figure is available online.

The excavation of FxJj20 AB is designed to document the history of local sedimentological processes. Several intact sed- three-dimensional locational data of as many artifacts as pos- iment blocks were collected throughout the site, following sible, particularly microartifacts. Dental tools and paintbrushes Stoops (2003) and Goldberg and Macphail (2006), to examine are the primary excavation tools; sediments are removed in the entire stratigraphic sequence and features observable in 5 cm levels. Spatial data are recovered using a total station the field. Intact sediment blocks were pedestalled in situ and and handheld data collector running EDM Mobile (McPherron wrapped with plaster of paris strips. The top and north-south and Dibble 2011). Obviously elongated artifacts are shot us- orientations were recorded on the blocks before removal. La- ing two shots on the long axis to determine artifact orienta- beled plastered blocks were cut out of the excavation and re- tion and help understand site formation processes (McPher- wrapped with more plaster of paris and packing tape to en- ron 2005). Sediment is screened through a 2 mm mesh screen sure sample integrity during transportation and shipment to to ensure recovery of small artifacts and ecofacts. Recovered the lab. In the lab, blocks were air-dried at 507C, embedded materials are classified and recorded (see supplemental ma- in styrene-diluted polyester resin under vacuum, cured, and terial online for details). With this methodology we collect cut into 2 # 3 inch and 1 # 2 inch blocks (chips). The chips three-dimensional coordinates for ~95% of the recovered were prepared into petrographic thin-section slides by Spectrum material, allowing high-resolution spatial analysis of the mate- Petrographics (Vancouver, WA). The thin-section slides were rials and enabling us to make predictions about activities on analyzed by petrographic microscopy with an Olympus BX41 the site. following the criteria in Stoops (2003) and by FTIR microscopy using a Thermo-Nicolet iN10 MX FTIR microscope (see be- low). The petrographic thin-section slides were examined for Soil Micromorphology composition and micromorphological features to identify re- Soil micromorphology analysis was performed to characterize mains of paleosurfaces, human activities, and natural processes the sediments at the microscopic scale and reconstruct the (i.e., bioturbation and pedogenesis). S248 Current Anthropology Volume 58, Supplement 16, August 2017

Heat Experiments press (Pike Technologies). The pellets were analyzed using a Thermo-Nicolet iS5 spectrometer collecting 64 scans in the Over the course of several seasons (2012–2015) we conducted 4,000 to 400 cm21 wavenumbers with a resolution of 4 cm21 a number of heat experiments to build a reference collection of wavenumbers. heated sediment, bone, and local lithic raw materials. We in- All materials (soil particles, bone fragments, lithic raw ma- vestigated various lithologies, including Gombe and Asille terial flakes, and pebbles) recovered from within Locus 1 (lo- basalt. Gombe basalt makes up approximately 50% of the ar- cation of a potential combustion feature) were analyzed by tifacts found at FxJj20 Main (Braun, Harris, and Maina 2009). FTIR to test for indications of heating. Materials from through- These heat experiments were conducted in the lab and in the out the site were tested by FTIR spectrometry to look for other field to document visible heat-related changes (e.g., discolor- heated items and identify potential natural or anthropogenic ation, fractures) and mineralogical transformations using FTIR combustion features. When burned material was identified, (see below). Experimental work in the geoarchaeology lab at lithics and bone fragments collected from its proximity were Simon Fraser University (SFU) included firing sediment samples also tested. Testing on random samples of bone and stone (taken from the center of FxJj20 AB), Gombe basalt, and Asille from throughout the site was also conducted. basalt in a muffle furnace (ThermoLyne 30400) at 4007C, A Thermo-Nicolet iN10 MX FTIR imaging microscope was 5507C, and 8007C for four hours. Field experiments placed used to analyze petrographic thin sections. Spectra of particles tested materials in camp fires fueled with wood locally avail- with diameter of 50 to 150 mm were collected in transmission able at Koobi Fora. Experimentally heated materials included and total reflectance modes with a Reflectocromat 15# objec- cobbles and flakes of basalt, ignimbrite, chert, and chalcedony, tive between 4,000 cm21 and 450 cm21 at 8 cm21 resolution. all of which would have been locally available to hominins in the past. An infrared thermometer was used to measure the temperature of the experimental camp fires; all fires reached Spatial Analysis temperatures above 5507C and lasted at least 6 hours. We vi- Spatial analysis was performed using ArcGIS on stone ar- sually inspected heated raw materials in the field at Koobi Fora tifacts, bone fragments, and rubified sediment aggregates re- Base Camp, at the National Museums of Kenya (NMK) in covered with three-dimensional coordinates. Materials re- Nairobi, Kenya, and at the SFU lab. FTIR testing of unheated covered in the screen (less than 5% of the material recovered and heated materials was performed with a portable FTIR during the 2010–2015 excavations) were excluded from the spectrometer (Thermo-Nicolet iS5) at the NMK in Nairobi spatial analysis. and at the SFU geoarchaeology lab. Laboratory experiments Vertical and horizontal spatial distribution of artifacts gives of the basalts and subsequent FTIR testing are still ongoing; insight into the stratigraphic makeup of the site. Analysis of the thus, the results presented here are considered to be prelimi- vertical distribution of the artifacts coupled with the soil mi- nary. cromorphological analysis may shed light on whether the ar- chaeological assemblage derived from a single occupation or a palimpsest of many occupations (Bailey 2007; Malinsky-Buller, FTIR Spectrometry and Microspectrometry Hovers, and Marder 2011). FTIR spectrometry has been used extensively to characterize the organic and inorganic composition of archaeological ma- Optimized hot spot analysis. ArcGIS-optimized hot spot terials, features, and deposits (Weiner 2010), and in particular analysis (OHSA) uses the Global Morans I statistic to calculate to identify heated materials in the archaeological record (Berna statistically significant clusters, or “hot spots,” within a speci- and Goldberg 2007; Berna et al. 2012; Shahack-Gross, Bar- fied geographical area. Unlike kernel-density estimates, which Yosef, and Weiner 1997; Shahack-Gross et al. 2014; Weiner will run with any sample size, OHSA requires a minimum 2010; Weiner et al. 2015). FTIR spectrometry makes it possible number of data points to work. When performed with the to recognize heat-induced mineralogical transformation in spatial analysis software ArcGIS (ESRI), OHSA aggregates bone, chert, calcite, and clay minerals (Berna 2010; Berna data within polygons. When no specific bounding polygon is et al. 2007, 2012; Berna and Goldberg 2007; Shahack-Gross, identified, the analysis uses a predefined fishnet grid over the Bar-Yosef, and Weiner 1997, 2014; Weiner et al. 2015). Micro- site, and empty polygons are discarded from the analysis. If destructive FTIR sampling (milligram particles) was conducted a bounding polygon is defined, empty polygons within the on archaeologically collected bulk and single-grain sediment boundary are included in the analysis. The analysis corrects samples, discolored aggregates, and bone and lithic fragments for a false discovery rate by assuming that objects close to each as well as unheated and experimentally heated local sediment other are likely similar and lowering the P value threshold and lithic raw materials. Initial FTIR analysis was conducted at for significance (e.g., P p .001 as opposed to P p .05). The the NMK, Nairobi, Kenya, at the close of the 2015 field season. analysis can be run with or without a weighting field. Weighting Further testing took place at the SFU lab. We ground samples data can be any type of data on which results are evaluated, with an agate mortar and pestle, mixed them with potassium such as population numbers in a city or number of cases of bromide (KBr), and pressed them into pellets using a hand disease. Without a weighting field, the analysis calculates spa- Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S249 tial similarity alone. With a weighting field, the analysis cal- of archaeologically recovered materials around a fire. Mod- culates spatial similarity in relation to the weighted field. eling his research after Stappert’s (1989) work at Pincevent, Forthisanalysis,OHSAswererunwithandwithoutweight- Henry (2012) proposed that around 120 cm from the fire ing. For all analyses, a 10 cm grid was overlaid on the site. For there would be a drop in numbers representing the “squat the weighted analysis, maximum dimension of artifacts was zone,” the place where an individual would sit or squat to used. Densities of larger materials should be identified as “hot work around the fire. He developed three hearthside ring spots,” while densities of smaller materials will be identified profiles to describe the activities on a site. In his analysis, he as “cold spots.” This method obviates the need to classify ma- looked at the frequencies of materials up to 2 m from the terials as micro- or macroartifacts and provides a more accu- center of the hearths found on the site (Henry 2012). He rate method of grouping materials according to size. The anal- defined three observable patterns: Type I, toss-zone domi- ysis will show whether artifacts in a spatial cluster are similarly nant, where the majority of artifacts are found behind the sized. individual working around the hearth; Type II, drop-zone The OHSA will define areas on the site where there is sig- dominant, where some materials are found in the toss zone nificant clustering and also determine whether materials on immediately behind the individual but most are found in the the site cluster according to size. We expect that if material is drop zone in front of the individual; and Type III, drop-zone clustering on the site, this will show up in the OHSA. If there dominant, where nearly all the material is found in front of are no clusters on the site, the analysis will return nonsignifi- the individual (Henry 2012). For Types I and II, Henry (2012) cant results. defines a “squat zone” where the individual working around the fire would be sitting and suggests that maintenance of the Toss and drop zones and ring distribution. Analyses of toss work area is responsible for these patterns. He suggests that and drop zones, as proposed by Binford (1983), have taken the Type III pattern indicates that an individual did not move several forms. They are used to identify the potential location from his or her workspace and did not try to maintain a clear of hearths where fire evidence such as ash and charcoal are workspace (Henry 2012). These analyses do not take size into absent (Alperson-Afil 2017; Alperson-Afil, Richter, and Goren- account, only frequency of artifacts from the center of the Inbar 2007; Sergant, Crombé, and Perdaen 2006; Vaquero and hearth. Comparisons between distributions from Tor Faraj, Pastó 2001), but they are also used to look at activities that Jordan (Henry 2012), Pincevent, France (Stappert 1989), and may have occurred around known hearths (Henry 2012; Stap- FxJj20 AB were made using a Kolmogorov-Smirnov test to de- pert 1989). While clusters of artifacts on a site can be useful to termine whether the material from FxJj20 AB fits the patterns define the potential location of a combustion feature, this in- found at other Middle and Upper Paleolithic sites. If the profile formation should be corroborated by the presence of burned of materials from FxJj20 AB matches one of the ring profiles material. Potential toss and drop zones are identified by look- proposed by Henry (2012), we can hypothesize that it derived ing at the spatial distribution of bone and lithic materials on a from behavior associated with a hearth. site coupled with an analysis of artifact size. As Binford (1983) directly noted around modern ethnographic campfires, smaller pieces of stone or bone (unintentionally produced waste) should Results be found in the drop zone, directly in front of the individual Excavation (Artifacts and Site Formation) working with these items. Larger pieces (intentionally dis- carded waste) should be found in the toss zone, which can be The number of artifacts recovered in situ from all excavated expected to be found up to several meters distance behind, to areas at FxJj20 AB totals 2,969 objects (table 2). The distri- the side, or in front of the individual (Binford 1983). Binford bution of recovered materials is given in figure 2. The vertical (1983) cites similar distributions patterns around hearths ob- spread of artifacts is between 8 cm in the lowest-density areas served by John Yellen in the Kalahari and Richard Gould in to 20 cm in the highest-density areas, with a maximum spread Australia. A comparable pattern is observable in the distribution of 40 cm at the eastern edge of the excavation (fig. S1; figs. S1– of burned archaeological materials at GBY (Alperson-Afil 2017; S9 available online). The distribution of all materials recovered Alperson-Afil and Goren-Inbar 2006, 2010; Alperson-Afil, Rich- from the site shows a high density of artifacts spreading from ter, and Goren-Inbar 2007; Goren-Inbar et al. 2004). If an the center to the northeastern portion of the site. A cross- activity area associated with fire is present, we expect small sectional view of the excavation shows that the archaeological materials, or debitage, to be clustered together (i.e., drop zone), material is found on a slight (~57) slope following the natural while larger materials will be found farther out and clustered topography of the area (see fig. S1). Large and small finds are separately from small materials. We would expect large lithic distributed throughout the excavation. There is no preferred tools, hammerstones, and flakes to cluster away from small orientation of the artifacts, indicating that no postdepositional debitage on the site. transport of materials occurred (see Hlubik et al. 2017 for de- Henry (2012), in a study of the Middle Paleolithic site of tails on orientation studies). Lithics from the excavation retain Tor Faraj, showed several potential configurations of the toss sharp edges and do not show any other signs of transport and and drop zones that can be distinguished by the distribution redeposition by water (fig. 3). Furthermore, there is a large S250 Current Anthropology Volume 58, Supplement 16, August 2017

Table 2. List of all finds from the excavation AB are consistent with those recorded by Schick (1986) from at FxJj20 AB knapping experiments before winnowing treatments (fig. S2; Hlubik et al. 2017). Excavated materials Count While some identifiable pieces of bone were found during Stone tools: the 1970s–1980s excavation, most taxonomic identifications Core 85 were made from tooth fragments that included representative Cobbles 16 species of the families Bovidae, Suidae, Equidae, Rhinocerotidae, Flakes 1,435 Whole flakes (n p 795) Hippopotamidae, Cercopithecidae, Hystricidae, Chelonia, Aves, Broken flakes (n p 640) and Thryonomyidae (Harris 1997). Much of the faunal evi- Flakes with potlid evidence (n p 3) dence found during the 2010–2015 excavations is extremely Angular fragments 1,372 fragmentary and not identifiable beyond “mammal,”“reptile,” Debitage (n p 1,328) “fi ” fi p or sh categories. Some are identi able to general element Potlids (n 4) “ ” Pebbles 52 (e.g., long bone ), but most are small fragments between 1 cm fi Hammerstones 9 and 5 cm in maximum dimension and are identi able only as Total 2,969 “bone.” Cortical surfaces are largely missing, making it diffi- Raw material: cult to determine whether butchery was occurring at the site. Basalt 2,807 Faunal analysis is ongoing to determine the total portion of Quartz 19 fi fi Ignimbrite 60 identi able fauna, whether any fragments re t, and whether fi Chert 58 more speci c elemental and taxonomic determinations can be Chalcedony 21 made. Aggregates of rubified sediment are found throughout Feldspar 4 the site, with the highest concentrations in the central and Total 2,969 southern portion of the site and much lower densities in the Fauna: fi Bone 1,242 northern portions ( g. S3). Tooth 102 It is important to note that there are several low-artifact- Total 4,313 density areas in the site, but one particular area (Locus 1) is located in the middle of several high-density clusters of objects. All squares adjacent to Locus 1 were excavated and had sig- proportion of small objects (40% have a maximum dimension nificant artifact densities. Approximately one quarter of Lo- less than 2 cm) that would be affected by significant hydraulic cus 1 (~0.7 m in diameter; see fig. 2) was excavated, and while action (Schick 1986). The numbers of artifacts found within some artifacts and bone were found, the density was much each size class from the new (2010–2015) excavations at FxJj20 lower than the surrounding areas. However, a number of ru-

Figure 3. Left, photographs of some artifacts from the FxJj20 excavations. The artifacts are fresh, not rolled, and show no signs of being transported in or by water. Right, photograph of an example of a potlid identified through lithic analysis; ventral side ( far right) is irregular with no signs of a bulb of percussion. A color version of this figure is available online. Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S251 bified sediment aggregates (n p 18) were recorded but not tuff. Interestingly, no calcium carbonate crystalitic b-fabric or collected from Locus 1. pedofeatures have been observed in any of the thin sections analyzed. Evidence of soil mixing appears to be limited to grass roots, arthropods, and earthworm action. There is no evidence Soil Micromorphology of major bioturbation from tree roots or by vertebrates such as Micromorphological analysis of the excavated area shows that rodents or reptiles. It is likely, given the fine-grained nature of the sediment is locally composed of very weakly pedogenized the sediments, the small size of the artifacts, and the lack of tuff with feldspathic, micaceous, and quartz silt to fine sand features produced by strong or prolonged pedogenesis, that the (fig. 4a). These sediment characteristics indicate that the ar- accumulation of sediment at the site was the result of seasonal chaeological material was deposited on a low-energy, rapidly or decadal flooding events from a nearby channel. The exact aggrading surface such as should be expected on an active flood- location of the channel has not yet been pinpointed (Harris plain. The porosity of the sedimentary column consists of a few 1997). Micromorphological analysis shows that the sediment of subcentimeter-size root channels and rare larger passage fea- Locus 1 has the same general composition and structure of the tures and other fabric pedofeatures. Limpid laminated yellow adjacent areas. It differs slightly because of the presence of rare clay coatings (fig. 4b) suggest postdepositional, low-energy, sub- rubified soil aggregates and a slightly higher content of fine surface translocation of clay formed from the weathering of the sand (fig. 4c). No evidence of wood ash or bone fragments has

Figure 4. Soil micromorphology at FxJj20 AB. a, Plane polarized light (left) and cross polarized light (right) scan of petrographic thin sections of intact block of local soil column, catalog no. 25015 (scale bar p 1 cm). Note the silty clay loam texture, channel micro- structure, and a few passage features (p.f.). b, Representative cross polarized light micrograph of the local ground mass (scale bar p 200 µm). Note the micaceous, feldspathic, and quartz composition of the poorly sorted silt and sand and the laminated clay coatings around pores and sand grains. c, Plane polarized light (left) and cross polarized light (right) scan of petrographic thin sections of intact block from inside Locus 1 (scale bar p 1 cm). Note sediment has similar composition and structure with respect to local sediment shown in figure 5b (sample catalog no. 25015) and contains rare rubified soil aggregates (r.a.). A color version of this figure is available online. S252 Current Anthropology Volume 58, Supplement 16, August 2017 been found in the blocks sampled so far from Locus 1 or other clear spectral differences between the unburned and burned areas. sediments with temperature-dependent pattern characteristics (fig. 5b). Over 600 pieces of bone, 170 pieces of rock, and 24 samples Experimental Work and Archaeological Evidence of Burning of discolored sediment were tested using FTIR. Of these, Upon being heated, Gombe and Asille basalt specimens did 50 pieces of bone and five samples (one in situ) of rubified not exhibit a color change, but large pieces exposed directly to sediment were found to show evidence of being heated. Of fire often spalled in convex subrounded flakes (aka potlids), thematerialsfoundwithinLocus1,onebonefragmenthad sometimes reducing an entire cobble through the process and FTIR absorptions characteristic of being burned above 5007C producing thousands of rock fragments. These potlids lack strik- and below 7007C (Berna 2010; fig. 5a). One sample of rubified ing platforms or bulbs of percussion (fig. 3) and are distin- sediment, found approximately 1 m east of Locus 1, was iden- guishable from flakes produced by knapping activities. Dur- tified using FTIR as being burned. Around this, six pieces of ing the firing experiments, the potlids projected out of the bone were identified as burned (fig. 2). An additional 43 bone fire, sometimes to a distance of several meters. Experimental specimens from other areas of the site were identified as potlids generally have an irregular surface texture that contrasts burned. with the smoother texture observed on debitage flakes derived from stone knapping. Smaller flaked pieces of basalt some- Spatial Analysis times sustained potlid fractures but more often emerged from the fire unchanged. Forty-four archaeological specimens of ba- Spatial analysis shows distinctive clusters of material on the salt were identified as potlids through comparison with exper- site, both horizontally and vertically. The results of individual imentally produced potlids and debitage. The 44 archaeolog- analyses are discussed below. ical potlids are found scattered throughout the site and are plotted in figure 2. This kind of spatial distribution is expected OHSA. Horizontally, the artifacts occur in one large dom- because potlids can be ejected several meters from the fire. The inant cluster in the northeastern portion of the site (fig. S5) incidence of numerous potlids on the site indicates the pres- with several smaller clusters to the southeast. The OHSA iden- ence of fire, but it is not expected to indicate the exact location tified the large cluster as significantly different from a random of the fire. distribution at a 99% confidence level (fig. S5). Overall the One piece of reddened chert was found roughly 3 cm to the clusters are composed of similarly sized materials (fig. S6). southeast of the edge of Locus 1. The chert from FxJj20 AB is Significant clusters of artifacts 12 cm are shown in red, while generally a yellow-brown color. This reddened chert retained significant clusters of artifacts !2 cm are shown in blue. Areas some of the original brown color along one edge, but over with no distinctive clustering of certain size-class specimens 90% of the piece is a deep red color. The reddening of the are indicated with cream-colored circles—these clusters con- piece is similar to the heat-induced discoloration observed in tain materials that fit into both large and small categories. other chert samples (which had similar coloration before heat- Specimens of bone tend to be found in clusters that overlap ing) and thus is considered to be burned (Purdy and Brooks with the concentrations of stone material, particularly the 1971). smaller artifacts (fig. 5). Clustering of artifacts and bone are Unheated and heated samples of the two principal local highest around Locus 1: smaller artifacts cluster primarily to raw materials, Gombe and Asille basalts, were analyzed by the west and southwest, bone clusters primarily to the south FTIR. Gombe and Asille basalts have overall similar FTIR and west as well but wraps around to the north, and larger spectra. Gombe basalt shows a unique hydroxyl absorption at artifacts cluster to the west of Locus 1. The small cluster of ca. 3,620 cm21, assigned to weathered olivine. Upon heating materials to the northeast of Locus 1 is primarily made up of Gombe basalt to 5007C, the FTIR absorption at 3,620 cm21 is artifacts that are both large and small. lost (fig. S4). Although it would be possible to use FTIR to One aggregate of rubified sediment was recovered within distinguish between unheated and heated Gombe Basalt, un- Locus 1, and more are found dispersed throughout the cen- fortunately the FTIR patterns of unheated and heated Asille tral and southern portions of the excavation site. A small basalts and heated Gombe basalt are statistically indistinguish- group of burned material (potlids, bone, sediment aggregates) able (fig. S4). Thus, we do not consider FTIR alone to be a re- is found to the west of Locus 1 and constitutes a statistically liable method for unequivocally distinguishing between burned significant cluster of burned materials, which could be thus and unburned Gombe and Asille basalts. considered the location of a contained fire. Experimental lab work showed that when sediment from the site was heated, it underwent a dramatic color change from a Toss and drop zones and ring distributions. The spatial light grayish tan to a deep red color (fig. 5b). This color change distribution around Locus 1 is compatible with that of com- was incremental according to temperature, with the material bustion features identified using the toss- and drop-zones burned at 4007C turning only slightly red and the material analysis (Binford 1980, 1983). Specifically, smaller materials burned at 8007C turning a deep red. FTIR analysis also shows are found close to Locus 1, along with high concentrations Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S253

Figure 5. a, FTIR spectrum of bone sample 12400 showing absorption at ca. 565, 603, 630, 960, 1,040, 1,090, 1,415, 1,455, and 3,570 cm21 characteristic of bone carbonate hydroxyapatite heated to ca. 5507C (Berna 2010). b, FTIR spectra of unheated (bottom pattern) and experimentally heated FxJj20 AB sediment to 4007C, 5507C, and 8007C. Note the weak absorptions at 915, 3,625, and 3,695 cm21 of kaolinite in the unheated sediment that disappear upon heating the sample to 4007C. Also note the shift of the major Si-O band from 1,026 to 1,089 cm21 upon the incremental heating of the sediment. A color version of this figure is available online. of bone, while larger materialsarefoundmoredistantfrom Using the center of the polygon describing the cluster of Locus1.Histogramsshowingthe distribution of artifacts burned material, we measured the distribution of artifacts in throughouttheentiresite(fig. S7) show the highest con- concentric rings from this polygon. A comparison of these centrations of artifacts are adjacent to Locus 1. The ring dis- ring distributions to Floor 3 (hearths 10 and 13) at Tor Faraj, tribution of the materials from around Locus 1 shows that the Jordan, and TII2 at Pincevent, France (figs. S8D,S9C), shows greatest numbers of materials are found between 80 and 120 cm that the FxJj20 AB Locus 2 assemblage is comparable to these from the center of Locus 1 (figs. S8C,S9B). Type III hearths are three archaeological hearths. Kolmogorov-Smirnov tests show not maintained in the same way as Type I and II hearths, but that the sample from FxJj20 AB is consistent with the distri- the ground immediately around the knapper is likely to have a butions found at Tor Faraj (Locus 1 and H10 and 13, P p .97; lower density than the surrounding area. Potential squat zones center of burned cluster and H10, P p .97; and center of for FxJj20 AB are labeled on figure S5. burned cluster and H13, P p .99) and at Pincevent (Locus 1, P p Another potential locus of fire activity is the area defined by .88; center of the burned material, P p .99). These Upper and the highest concentration of burned material (Locus 2, fig. 2). Middle Paleolithic hearths are classified as unimodal Type III, S254 Current Anthropology Volume 58, Supplement 16, August 2017 drop-zone dominant hearths, which bear resemblance to the number of heated bone fragments and sediment (0.014% of ring profile at FxJj20 AB. total material recovered) indicates that burning may have happened in discrete patches and was not widespread across fi Discussion the site, as a wild re would have been. More testing of materials found on the site will allow us to The meticulous recovery, high-resolution mapping, and anal- make a more definitive determination about the extent of fire ysis of the archaeological materials found at FxJj20 AB suggest on the site and the thermal history of the basalts found there. that the site has undergone weak postdepositional disturbance. Additional analysis of the sedimentological samples, both by None of the current analyses indicate deflation or artifact win- FTIR and magnetic techniques, may be able to further clarify nowing by hydraulic action. Soil micromorphology analysis the extent and nature of fire at FxJj20 AB. suggests the site is part of a fairly rapidly aggrading deposit, with At the moment we have identified two main locations where sediments brought onto the site by low-energy water. While fire may have been localized: Locus 1 and Locus 2 (figs. 2, S3). without refit studies we cannot rule out a palimpsest of over- Locus 1 is the low-artifact-density area located immediately lapping occupations, the vertical distribution is within the range east and north of the highest density of materials found on of refitting artifacts at FxJj50 (vertical dispersion at FxJj50 is 10– site and immediately southwest of a second high-density clus- 50 cm), a similar site in floodplain deposits located a few ter. We propose that Locus 1 may be the location of an ancient kilometers from FxJj20 AB (Bunn et al. 1980; Kroll 1994). The fire because of the lack of artifacts and the presence of burned soil micromorphological analysis revealed small root and in- bone found within this circular area and reddened chert adja- vertebrate action and minimal disturbance by larger bioturba- cent to it. Moreover, small materials dominate the high-density tion agents. The greatest spread in vertical distribution coincides cluster west of Locus 1. This is compatible with the phenomena with the high-density concentration of small (!2 cm) artifacts, of “drop zones” as described by Binford (1983). The highest which could be explained by small root action disturbing and density of these specimens is found within 120–140 cm of the displacing these artifacts. At FxJj50, Kroll determined, through center of Locus 1 (fig. S7). The minimal overlap between large refitting studies, that the material accumulated there was prob- and small materials found on the site is consistent with the ably the result of one or a few occupations on a single surface toss and drop profiles around hearths proposed by Binford before burial rather than multiple accumulations on several (1983) and comparable to a similar distribution described for surfaces overprinting each other and intermixing (Kroll 1994). Middle and Upper Paleolithic combustion features (Alperson- The small size of the artifacts at FxJj20 AB makes refitting Afil, Richter, and Goren-Inbar 2007; Henry et al. 1996; Sergant, studies difficult, but these are planned for the future to help Crombé, and Perdaen 2006; Vaquero and Pastó 2001). The dis- resolve the question of whether the site is a single accumulation tribution profile matches the unimodal Type III, drop-zone- or a palimpsest of a number of accumulations on a rapidly dominant profile identified by Henry (2012; fig. S8), and the aggrading surface. presence of burned materials within Locus 1 at the center of The clustering of materials at FxJj20 AB provides some this distribution further supports the hypothesis that the be- insights into hominin activities. The high proportion of flake havioral signal found on the site may be the result of fire- debitage and broken or damaged flakes suggests that much of control behavior by the hominins using the site. The lack of the flint-knapping activities occurred on the site. Comparison microscopic calcitic wood ash in Locus 1 deposit is explained to the experimental data reported in Schick (1986) indicates with ethnographic analogues and the local soil conditions that that the site is probably a site where flint-knapping activities do not favor calcium carbonate formation and preservation. took place (i.e., a primary production site). The mixture of Mallol et al. (2007) demonstrated that in several open-air Hadza lithic artifacts and bone fragments is not inconsistent with an combustion features, even in somelong-lastingones,theamount unspecialized camp, possibly a camp where individuals brought of wood ash that is incorporated in the subsoil is surprisingly resources together for processing. The distribution of materials low. The general absence of calcium carbonate components suggests the site may extend farther to the east, and future ex- and pedofeatures in the FxJj20 AB soil suggests that processes cavations will shed light on the total extent of the site. that result in calcium carbonate being leached away from the FTIR analysis shows that five rubified sediment aggregates top and subsoil horizons dominate the soil environment. and at least 50 pieces of bone were heated to or above 5007C. The second potential location of fire (Locus 2) corresponds Furthermore, while now there is no established morphological to the area defined by the greatest density of artifacts and definition of a potlid fracture, and the size and shape will likely burned materials. This area is where burned sediment from the vary with material type, analysis of the lithics at the site has site is found in conjunction with the majority of the burned identified a number of flakes that, based on comparison to bone. The position of Locus 2 to the west of Locus 1 (fig. S6) experimental examples, are very likely potlids resulting from suggests that Locus 2 corresponds to a combustion feature basalt objects being exposed to fire. Given the presence of subsequent to a Locus 1 combustion feature. The distribution potlids and FTIR-determined heated bone and sediment, it is of Locus 2 artifacts is consistent with ring profiles previously reasonable to consider the other rubified sediment aggregates described for other prehistoric hearth features. A comparison to be heated, but at temperatures below 5007C. The limited of a ring profile centered on a polygon describing the highest Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S255 density of burned materials (fig. S9) shows no significant dif- our work demonstrates that a recovery strategy designed to ference between the ring profile centered here and the hearth retrieve and map as much burned and unburned archaeolog- features described at Tor Faraj (Henry 2012) or Pincevent ical material as possible coupled with high-resolution spatial (Stappert 1989; fig. S9). Locus 2 is consistent with a fire loca- analysis has great potential to provide important information tion because of the high volume of burned materials found on the nature of the association of the archaeological record here. However, it is of some concern that the majority of with fire in primary contexts of great antiquity. materials found here are not burned, as would be expected for a localized fire. The mixed composition could be the result of material being moved around by animals or hominins or the Acknowledgments result of a rapidly aggrading palimpsest site. More study of the We thank the Canada Social Sciences and Humanities Research fi micromorphology of the site, re tting studies, and statistical Council (grant 430-2013-00546), the Wenner-Gren Founda- spatial analysis may shed light on postdeposition processes on tion for Anthropological Research (dissertation fieldwork grant the site. Gr.8984), and the National Science Foundation (doctoral dis- fi Future plans include the institution of a eld-based micro- sertation research improvement grant BCS-1443339). archaeology laboratory to facilitate experimental and archae- ological investigations for microscopic charcoal and phytolith remains to improve our understanding of the nature of fire References Cited residues found in Pleistocene archaeological contexts. None- Aiello, Leslie C., and Jonathan C. K. Wells. 2002. Energetics and the evolution theless, we can confirm the presence of burned items at FxJj20 of the genus Homo. Annual Review of Anthropology 31(1):323–338. AB and support the claims of Harris (1997) and Bellomo and Aiello, Leslie C., and Peter Wheeler. 1995. The expensive-tissue hypothesis: fi the brain and the digestive system in human and primate evolution. Cur- Kean (1997) for potential use of re at Koobi Fora and other rent Anthropology 36(2):199. Early Pleistocene localities in East Africa. Albert, Rosa M., Ofer Lavi, Lara Estroff, Steve Weiner, Alexander Tsatskin, Avraham Ronen, and Simcha Lev-Yadun. 1999. Mode of occupation of Tabun Cave, Mt Carmel, Israel during the Mousterian Period: a study of the Conclusion sediments and phytoliths. Journal of Archaeological Science 26(10):1249– 1260. The integration of techniques such as soil micromorphol- Alperson-Afil, Nira. 2017. Spatial analysis of fire: archaeological approach to fi recognizing early fire. Current Anthropology 58(suppl. 16):S258–S266. ogy, FTIR, and spatial analysis allowed the identi cation and Alperson-Afil, Nira, and Naama Goren-Inbar. 2006. Out of Africa and into precise contextualization of heated archaeological materials Eurasia with controlled use of fire: evidence from Gesher Benot Ya‘aqov, (lithics, bone, and sediment) in the 1.5 Mya old Early Stone Israel. Archaeology, Ethnology and Anthropology of Eurasia 2006(4):63–78. ———. 2010. Ancient flames and controlled use of fire, vol. 2 of The Acheulian Age context at FxJj20 AB, Koobi Fora, Kenya. We have come site of Gesher Benot Ya‘aqov. New York: Springer Science. to the following conclusions. (1) Spatial analysis reveals sta- Alperson-Afil, Nira, Daniel Richter, and Naama Goren-Inbar. 2007. Phantom tistically significant clusters of ecofacts and artifacts, indicating hearths and the use of fire at Gesher Benot Ya‘aqov, Israel. PaleoAnthro- pology 2007:1–15. that the archaeological material is in situ and is probably the Anselin, L., and A. Getis. 1992. Spatial statistical analysis and geographic in- result of various hominin activities during one or a few oc- formation systems. Annals of Regional Science 26(1):19–33. cupation phases over a short period of time. (2) We have found Bailey, Geoff. 2007. Time perspectives, palimpsests and the archaeology of fi time. Journal of Anthropological Archaeology 26(2):198–223. evidence of re associated with Early Stone Age archaeological Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a material in the form of heated basalt (potlids flakes), heated reason: barbecue at Middle Pleistocene Qesem Cave, Israel. Current An- chert, heated bone, and heated rubified sediment. To our thropology 58(suppl. 16):S314–S328. Bellomo, R. V. 1994. Methods of determining early hominid behavioral ac- knowledge this is, to date, the earliest securely documented tivities associated with controlled use of fire at FxJj 20 Main, Koobi Fora, evidence of fire in the archaeological context. (3) Spatial anal- Kenya. Journal of Human Evolution 27:173–195. ysis shows the presence of two potential fire loci. Both loci Bellomo, R. V., and W. F. Kean. 1997. Evidence of hominid-controlled fire at the FxJj20 site complex, Karari Escarpment. In Koobi Fora research project, contain a few heated items and are characterized by sur- vol. 5. G. L. Isaac and B. Issac, eds. Pp. 224–233 (app. 4A). Oxford: Clarendon. rounding artifact distributions with strong similarities to the Berna, Francesco. 2010. Bone alteration and diagenesis. In Scientific methods toss and drop zones and ring distribution patterns described and cultural heritage: an introduction to the application of materials science to archaeometry and conservation science. G. Artioli, ed. Pp. 364–367. for ethnographic and prehistoric hearths (Binford 1983; Henry Oxford: Oxford University Press. 2012; Stappert 1998). Berna, Francesco, Adi Behar, Ruth Shahack-Gross, John Berg, Elisabetta In summary, our results confirm the presence of fire in the Boaretto, Ayelet Gilboa, Ilan Sharon, et al. 2007. Sediments exposed to high temperatures: reconstructing pyrotechnological processes in Late Bronze Early Stone Age context of Koobi Fora FxJj20 previously and Iron Age strata at Tel Dor (Israel). Journal of Archaeological Science 34 documented by Harris (J. W. K. Harris, FxJj20 site notes, 1973, (March):358–373. 1997). In addition, our findings at FxJj20 AB support the hy- Berna, Francesco, and Paul Goldberg. 2007. Assessing Paleolithic pyrotechnology fi and associated hominin behavior in Israel. Israel Journal of Earth Sciences pothesis of a close association of re residues with hominin 56(2):107–121. activities. To test the hypothesis of an anthropogenic origin of Berna, Francesco, Paul Goldberg, Liora Kolska Horwitz, James Brink, Sharon the fire at FxJj20 AB, more intensive sampling for FTIR, soil Holt, Marion Bamford, and Michael Chazan. 2012. Microstratigraphic evi- dence of in situ fire in the Acheulean strata of Wonderwerk Cave, North- micromorphology, and other microscopic remains such as ern Cape Province, South Africa. Proceedings of the National Academy of microdebitage, microcharcoal, and phytoliths is needed. In fact, Sciences of the USA 109(20):E1215–E1220. S256 Current Anthropology Volume 58, Supplement 16, August 2017

Binford, L. R. 1980. Willow smoke and dogs’ tails: hunter-gatherer settlement ological sites, hominid remains, and traces of fire from Chesowanja, Kenya]. systems and archaeological site formation. American Antiquity 45(1):4–20. Nature 296:870. ———. 1983. In pursuit of the past: decoding the archaeological record. James, Steven R. 1989. Hominid use of fire in the Lower and Middle Pleis- Berkeley: University of California Press. tocene: a review of the evidence. Current Anthropology 30(1):1–26. Blumenschine, R. J., and B. Pobiner. 2007. Zooarchaeology and the ecology of Karkanas, Panagiotis, Ruth Shahack-Gross, Avner Ayalon, Mira Bar-Matthews, Oldowan hominin carnivory. In Evolution of the human diet: the known, Ran Barkai, Amos Frumkin, Avi Gopher, and Mary C. Stiner. 2007. Evidence the unknown, and the unknowable. P. Ungar, ed. Pp. 167–190. Oxford: for habitual use of fire at the end of the Lower Paleolithic: site-formation Oxford University Press. processes at Qesem Cave, Israel. Journal of Human Evolution 53:197–212. Brain, C. K. 1993. The occurrence of burnt bones at Swartkrans and their Kroll, Ellen. 1994. Behavioral implications of Plio-Pleistocene archaeological implications for the control of fire by early hominids. In Swartkrans: a cave’s site structure. Journal of Human Evolution 27:107–138. chronicle of early man. C. K. Brain, ed. Pp. 229–242. Pretoria: Transvaal Magill, Clayton R., Gail M. Ashley, and Katherine H. Freeman. 2013. Water, Museum. plants, and early human habitats in Eastern Africa. Proceedings of the Braun, David R., J. W. K. Harris, and D. N. Maina. 2009. Oldowan raw National Academy of Sciences of the USA 110(4):1175–1180. material procurement and use: evidence from the Koobi Fora Formation. Malinsky-Buller, Ariel, Erella Hovers, and Ofer Marder. 2011. Making time: Archaeometry 51(1):26–42. “living floors,”“palimpsests” and site formation processes: a perspective Bunn, Henry, J. W. K. Harris, Glynn Isaac, and Zefe Kaufulu. 1980. FxJj50: an from the open-air Lower Paleolithic site of Revadim Quarry, Israel. Journal Early Pleistocene site in Northern Kenya. World Archaeology 12(2):109– of Anthropological Archaeology 30(2):89–101. 136. Mallol, Carolina, Frank W. Marlowe, Brian M. Wood, and Claire C. Porter. Carmody, Rachel N., and Richard W. Wrangham. 2009. The energetic sig- 2007. Earth, wind, and fire: ethnoarchaeological signals of Hadza fires. nificance of cooking. Journal of Human Evolution 57:379–391. Journal of Archaeological Science 34:2035–2052. Clark, J. D., and J. W. K. Harris. 1985. Fire and its role in early hominid Mazza, Paul Peter Anthony, Fabio Martini, Benedetto Sala, Maurizio Magi, Maria lifeways. African Archaeological Review 3:3–27. Perla Colombini, Gianna Giachi, Francesco Landucci, Cristina Lemorini, Clark, J. D., and H. Kurashina. 1979. Hominid occupations of the East- Francesca Modugno, and Erika Ribechini. 2006. A new Palaeolithic discovery: Central Highland of Ethiopia in the Plio-Pleistocene. Nature 282(5734): tar-hafted stone tools in a European Mid-Pleistocene bone-bearing bed. 33–39. Journal of Archaeological Science 33(9):1310–1318. Ferraro, Joseph V., Thomas W. Plummer, Briana L. Pobiner, James S. Oliver, McPherron, Shannon P. 2005. Artifact orientations and site formation pro- Laura C. Bishop, David R. Braun, Peter W. Ditchfield, et al. 2013. Earliest cesses from total station proveniences. Journal of Archaeological Science 32 archaeological evidence of persistent hominin carnivory. PLoS ONE 8(4): (7):1003–1014. e62174. McPherron, Shannon P., Zeresenay Alemseged, Curtis W. Marean, Jonathan G. Gibbon, Ryan J., Travis Rayne Pickering, Morris B. Sutton, Jason L. Heaton, Wynn, Denné Reed, Denis Geraads, René Bobe, and Hamdallah A. Béarat. Kathleen Kuman, Ron J. Clarke, C. K. Brain, and Darryl E. Granger. 2014. 2010. Evidence for stone-tool-assisted consumption of animal tissues be- Cosmogenic nuclide burial dating of hominin-bearing Pleistocene cave fore 3.39 million years ago at Dikika, Ethiopia. Nature 466 (7308):857–860. deposits at Swartkrans, South Africa. Quaternary Geochronology 24:10–15. McPherron, Shannon P., and Harold L. Dibble. 2002. Using computers in Goldberg, Paul, and Francesco Berna. 2010. Micromorphology and context. archaeology: a practical guide. New York: McGraw Hill. Quaternary International 214(1/2):56–62. ———. 2011. EDM-Mobile. OldStoneAge.com, http://www.oldstoneage.com Goldberg, Paul, Francesco Berna, and Michael Chazan. 2015. Deposition and /software/edm-mobile.shtml. diagenesis in the earlier Stone Age of Wonderwerk Cave, Excavation 1, Milton, Katharine. 1999. A hypothesis to explain the role of meat eating in South Africa. African Archaeological Review 32(4):613–643. human evolution. Evolutionary Anthropology: Issues, News, and Reviews 8(1): Goldberg, Paul, and Richard I. Macphail. 2006. Practical and theoretical 11–21. geoarchaeology. Malden, MA: Blackwell. Organ, C., C. Nunn, Z. Machanda, and R. Wrangham. 2011. Phylogenetic rate Goldberg, Paul, Christopher E. Miller, and Susan M. Mentzer. 2017. Recog- shifts in feeding time during the evolution of Homo. Proceedings of the nizing fire in the Paleolithic archaeological record. Current Anthropology 58 National Academy of Sciences of the USA 108(35):14555–14559. (suppl. 16):S175–S190. Preece, R. C., J. A. J. Gowlett, S. A. Parfitt, D. R. Bridgland, and S. G. Lewis. Goren-Inbar, Naama, Nira Alperson-Afil, Mordechai E. Kislev, Orit Simchoni, 2006. Humans in the Hoxnian: habitat, context and fire use at Beeches Pit, Yoel Melamed, Adi Ben-Nun, and Ella Werker. 2004. Evidence of hominin West Stow, Suffolk, UK. Journal of Quaternary Science 21(5):485–496. control of fire at Gesher Benot Ya‘aqov, Israel. Science 304(5671):725–727. Purdy, B. A., and H. K. Brooks. 1971. Thermal alteration of silica minerals: an Gowlett, J. A. J., J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina. 2017. archeological approach. Science 173(3994):322–325. Evidence of burning from bushfires in southern and east Africa and its rel- Ragir, Sonia, Martin Rosenberg, and Philip Tierno. 2000. Gut morphology evance to hominin evolution. Current Anthropology 58(suppl. 16):S206– and the avoidance of carrion among chimpanzees, baboons, and early S216. hominids. Journal of Anthropological Research 56(4):477–512. Gowlett, J. A. J., J. W. K. Harris, D. Walton, and B. A. Wood. 1981. Early Roebroeks, Wil, and Paola Villa. 2011. On the earliest evidence for habitual archaeological sites, hominid remains, and traces of fire from Chesowanja, use of fire in Europe. Proceedings of the National Academy of Sciences of the Kenya. Comptes Rendus Palevol 5(1/2):299–310. USA 108(13):5209–5214. Harris, J. W. K. 1978. The Karari industry: its place in East African prehistory. Rowlett, Ralph M. 2000. Fire control by Homo erectus in East Africa and Asia. PhD dissertation, University of California, Berkeley. ACTA Anthropologica Sinica 19(suppl.):198–208. ———. 1997. FxJj 20. In Koobi Fora research project, vol. 5. G. L. Isaac and B. Sandgathe, Dennis M. 2017. Identifying and describing pattern and process in Isaac, eds. Pp. 147–169. Oxford: Clarendon. the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): Henry, D. O. 2011. Late Levantine Mousterian spatial patterns at landscape S360–S370. and intrasite scales in southern Jordan. Etudes et Recherches Archéologiques Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. McPherron, de l’Université de Liegè 126:1–18. Alain Turq, Laura Niven, and Jamie Hodgkins. 2011. Timing of the ap- ———. 2012. The palimpsest problem, hearth pattern analysis, and Middle pearance of habitual fire use. Proceedings of the National Academy of Sciences Paleolithic site structure. Quaternary International 247(1):246–266. of the USA 108(29):E298; author reply E299. Henry, D. O., S. A. Hall, H. J. Hietala, Y. E. Demidenko, V. I. Usik, A. M. Schick, K. 1986. Stone Age sites in the making. BAR International Series 319: Rosen, and P. A. Thomas. 1996. Middle Paleolithic behavioral organization: 313. 1993 excavation of Tor Faraj, southern Jordan. Journal of Field Archaeology Semaw, Sileshi, Michael J. Rogers, Jay Quade, Paul R. Renne, Robert F. Butler, 23:31–53. Manuel Dominguez-Rodrigo, Dietrich Stout, William S. Hart, Travis Picker- Hlubik, Sarah, David R. Braun, Francesco Berna, Craig S. Feibel, and John ing, and Scott W. Simpson. 2003. 2.6-million-year-old stone tools and asso- W. K. Harris. 2017. Site formation and integrity of FxJj20 AB, Koobi Fora, ciated bones from OGS-6 and OGS-7, Gona, Afar, Ethiopia. Journal of Hu- Kenya: implications for interpretation of Oldowan hominin fire use in the man Evolution 45(2):169–177. Early Pleistocene. PaleoAnthropology A:A1–A41. Sergant, Joris, Philippe Crombé, and Yves Perdaen. 2006. The “invisible” Isaac, G. 1982. Early hominids and fire at Chesowanja, Kenya [comment on hearths: a contribution to the discernment of Mesolithic non-structured J. A. J. Gowlett, J. W. K. Harris, D. Walton, and B. A. Wood, Early archae- surface hearths. Journal of Archaeological Science 33(7):999–1007. Hlubik et al. Nature of Fire at FxJj20 AB, Koobi Fora, Kenya S257

Shahack-Gross, R., F. Berna, P. Karkanas, C. Lemorini, A. Gopher, and R. Stoops, G. 2003. Guidelines for analysis and description of sand regolith thin Barkai. 2014. Evidence for the repeated use of a central hearth at Middle sections. Madison, WI: Soil Science Society of America. Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of Archaeological Thibodeau, Megan L. 2016. Identifying 1 mya fire in Wonderwerk Cave with Science 44(1):12–21. micromorphology and Fourier-Transform infrared microspectroscopy. Thesis, Shahack-Gross, Ruth, Ofer Bar-Yosef, and Steve Weiner. 1997. Black-coloured Simon Fraser University, Burnaby, British Columbia. bones in Hayonim Cave, Israel: differentiating between burning and oxide Vaquero, Manuel, and Ignasi Pastó. 2001. The definition of spatial units in staining. Journal of Archaeological Science 24(5):439–446. Middle Palaeolithic sites: the hearth-related assemblages. Journal of Ar- Shimelmitz, Ron, Steven L. Kuhn, Arthur J. Jelinek, Avraham Ronen, Amy E. chaeological Science 28(11):1209–1220. Clark, and Mina Weinstein-Evron. 2014. “Fire at will”: the emergence of Weiner, S. 2010. Infrared spectroscopy in archaeology. In Microarchaeology: habitual fire use 350,000 years ago. Journal of Human Evolution 77:196– beyond the visible archaeological record. P. 396. Cambridge: Cambridge 203. University Press. Smith, Alex R., Rachel N. Carmody, Rachel J. Dutton, and Richard W. Weiner, Steve, Vlad Brumfeld, Ofer Marder, and Omry Barzilai. 2015. Wrangham. 2015. The significance of cooking for early hominin scaveng- Heating of flint debitage from Upper Palaeolithic contexts at Manot Cave, ing. Journal of Human Evolution 84:62–70. Israel: changes in atomic organization due to heating using infrared spec- Sorensen, Andrew, Wil Roebroeks, and Annelou van Gijn. 2014. Fire pro- troscopy. Journal of Archaeological Science 54:45–53. duction in the deep past? the expedient strike-a-light model. Journal of Wrangham, R., and N. Conklin-Brittain. 2003. Cooking as a biological trait. Archaeological Science 42(February):476–486. Comparative Biochemistry and Physiology A 136(1):35–46. Speth, J. D. 1989. Early hominid hunting and scavenging: the role of meat as Wrangham, R. W., J. H. Jones, G. Laden, D. Pilbeam, and N. L. Conklin- an energy source. Journal of Human Evolution 18:329–343. Brittain. 1999. The raw and the stolen: cooking and the ecology of human Stahlschmidt, Mareike C., Christopher E. Miller, Bertrand Ligouis, Ulrich origins. Current Anthropology 40(5):567–594. Hambach, Paul Goldberg, Francesco Berna, Daniel Richter, Brigitte Urban, Wrangham, Richard. 2017. Control of fire in the Paleolithic: evaluating the Jordi Serangeli, and Nicholas J. Conard. 2015. On the evidence for human cooking hypothesis. Current Anthropology 58(suppl. 16):S303–S313. use and control of fire at Schöningen. Journal of Human Evolution 89:1–21. Wrangham, Richard, and Rachel Carmody. 2010. Human adaptation to the Stappert, D. 1989. The ring and sector method: intrasite spatial analysis of control of fire. Evolutionary Anthropology: Issues, News, and Reviews 199: Stone Age sites, with special reference to Pincevent. Palaeohistoria 31:1–57. 187–199. S258 Current Anthropology Volume 58, Number S16, August 2017

Spatial Analysis of Fire Archaeological Approach to Recognizing Early Fire

by Nira Alperson-Afil

The use of fire by early hominins is considered a significant technological and cultural revolution. Recently, the study of fire use has been affected by a troublesome trend that views chemical and microscopic techniques as the only ac- ceptable analyses of fire residues, thus ignoring basic archaeological observations and analyses. This paper discusses the diverse expressions of early fire, their variability, and their level of significance and suggests that the spatial analysis of burned residues is a reliable method for recognizing early fire. Evidence from the site of Gesher Benot Ya‘aqov (GBY) suggests that fire was routinely used by Acheulian hominins. In addition, new data on the spatial association between percussive activities and fire are presented. Such evidence of routine and habitual use of fire requires intensity and recurrence, documented in sites with long occupational sequences such as GBY. This requirement excludes habitual use of fire from the majority of early hominin habitations, documented by short-term occupations of open-air sites. To deny Early to Middle Pleistocene hominins of the habitual use of fire is to ignore the archaeological record of their evolution, behavior, and culture.

The use of fire by early hominins is considered a significant Identifying Prehistoric Fire technological and cultural revolution and has occupied archae- ological and anthropological studies for many decades. Recently, In the early years of prehistoric research, an archaeological as- the study of fire use has been affected by troublesome trends. sociation between stone tools and burned materials was con- fi One is the apparent competition of many recent studies for ex- sidered an adequate indication for the use of re (e.g., Garrod clusivity of the earliest evidence of the use of fire, the control of and Bate 1937). Later, issues of depositional integrity and the fi fi fire, or the habitual use of fire (e.g., Karkanas et al. 2007). An- probability of natural res were considered, adding re nement other is the view that chemical and microscopic techniques are to the association between human activities and burned resi- fi the only acceptable analyses of fire residues (e.g., Stahlschmidt dues. At present, even that relationship is deemed insuf cient, et al. 2015). These views are disturbing because they often ig- and some scholars consider the presence or absence of a hearth fi fi nore and disregard elementary archaeological observations and ( replace) the sole criterion for proving re use (e.g., Shahack- analyses. The range of available techniques (Goldberg, Mil- Gross et al. 2014), preferably with the support of a microcon- ler, and Mentzer 2017) contributed significantly to the study of textual approach (Goldberg, Miller, and Mentzer 2017). This early fire, yet they are not always feasible because they require is of prime importance, as hearths are often not preserved ar- favorable postdepositional conditions (e.g., Pop et al. 2016). chaeologically, and their absence is thus wrongly interpreted as This paper discusses the use of spatial analyses in the study lack of evidence. of fire use at the 0.78 Ma site of Gesher Benot Ya‘aqov (Israel; Hearths are features of all contemporary hunter-gatherer so- GBY). Clusters of burned flint microartifacts were identified cieties, and ethnographic data suggest that open hearths, which throughout the long cultural sequence at the site and inter- involve no construction, are more common than hearths com- preted as the remnants of hearths. In addition, new data on the prising dug pits, stone or wood linings, or a structure of some spatial association between the hearths and a variety of per- sort. In addition, when found, ethnographically or archaeolog- cussive activities are presented, providing additional support ically, hearths exhibit high variability of construction method, for the assertion that spatial analysis, though based “only” on size, and function (Galanidou 1997; Mallol and Henry 2017; archaeological data, is an efficient method for the study of early Mallol et al. 2007), and this variability increases when they are use of fire. exposed and excavated at archaeological sites. In his Diction- naire de la préhistoire, Leroi-Gourhan suggested that a hearth Nira Alperson-Afil is Senior Lecturer in the Department of Land of will exhibit discoloration (dark sediments) and that charcoal Israel Studies and Archaeology at Bar-Ilan University (Ramat Gan, should be preserved (Leroi-Gourhan 1988:405). Schiegl et al. fi 5290002, Israel [nira.alperson-afi[email protected]]). This paper was sub- (1996) suggest another de nition that similarly depends on the mitted 25 VII 16, accepted 5 III 17, and electronically published 13 state of preservation: “Good field evidence for the use of fire is VII 17. the presence of well-preserved hearths. Such hearths are usually q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0009$10.00. DOI: 10.1086/692721 Alperson-Afil Spatial Analysis of Fire S259 round or oval shaped and often have an upper layer composed study of the spatial consequences of former hominid activity of light coloured minerals, a lower layer rich in charcoal, and a patterns within and between features and structures and their substrate of reddened sediment” (Schiegl et al. 1996:763–764). articulation within sites, site systems and their environments” These descriptions, however, appear to suit the definition of (Clarke 1977:9). Two main approaches can be distinguished a “well-preserved hearth” better than that of a “hearth.” The among spatial analysis studies that focus on a single site (in- ethnogeoarchaeological study of Mallol et al. (2007) demon- trasite). The first involves studies that investigate the forma- strated that the preservation of combustion features is not a tion of archaeological spatial patterns in actualistic contexts— straightforward issue, particularly in open-air sites: “the an- ethnographic or experimental—aiming to provide models for thropogenic signature of open air combustion structures can be the formation of archaeologically observed spatial patterns (e.g., detected depending on the rates of sedimentation and the im- Bartram, Kroll, and Bunn 1991; Binford 1968, 1978; O’Connell, pact of postdepositional disturbance factors. . . . If the rates Hawkes, and Jones 1991). The second approach is using a va- of sedimentation are low, leading to erosion, the remains of an riety of methods for analysis of archaeological spatial patterns, ephemeral open air fire are likely to disappear” (Mallol et al. including the recognition and interpretation of similarities and 2007:2050). Similarly, discoloration of sediments around and differences in distribution among artifacts and features, map- beneath the hearth depends on a variety of factors (e.g., fuel ping of the excavated area, identifying clusters of artifacts, re- used, soil moisture, chemical variations in sediments) and re- fitting of lithics and bones, etc., and these methods are often quires favorable depositional and postdepositional conditions assisted by different spatial analysis software. in order to be preserved in the archaeological record (Bellomo The behavioral implications of spatial analysis studies are and Harris 1990; Canti and Linford 2000; Linford and Canti straightforward; that is, the remains of human activities are dis- 2001). Thus, as preservation depends on age and soil chemistry, tributed nonrandomly because human behavior is not random among other things, it is evident that the definition of a hearth across space. This derives from the behavior of our species for varies in terms of sedimentological setting, intensity, size, fuel which the segregation of activities is a means to organize and used, structure, and function. These variables will eventually manipulate cultural environments and to codify social relation- dictate the archaeological appearance of these features, that is, ships (e.g., Binford 1978, 1983, 2001; Galanidou 2000; Gargett whether hearths will exhibit a stone lining, whether ash and/or and Hayden 1991; O’Connell 1987; Yellen 1977). charcoal will be preserved, or whether discoloration of the sed- Clearly, before attempting to spatially analyze archaeological iments will occur. Consequently, as in the ethnographic record, remains, their spatial integrity should be established. A variety the archaeological occurrences of hearths are extremely vari- of postdepositional and taphonomic processes may have altered able and uneven, and hearths are independently defined for the original spatial configuration of the archaeological mate- each site. rial. In addition, any spatial analysis should follow clear strat- The only common feature of all hearths is the simple fact igraphic observations, reassuring that a distinct occupation is that people intentionally burn fuel in order to produce a fire. under analysis rather than palimpsests of several occupations Accordingly, the archaeological definition of a hearth should (e.g., cave sites). specify that it is an anthropogenic combustion area variable in structure, size, and depth that preserves the remains of burned Spatial Analysis and Fire materials. We should bear this in mind when reviewing inter- pretations of early evidence of fire use, as they often lack com- The fact that humans tend to carry out a vast range of activities prehension of this variability. The variability of evidence is fur- in close proximity to hearths is widely documented. The hearth ther complicated by certain lines of evidence that are bound to assembles the social group, and around it is the area in which be absent in sites in which postdepositional processes have social interactions, tool production, food processing, food con- concealed the evidence: “taphonomic problems related to the sumption, and ritual ceremonies are carried out (Binford 1983, preservation of ash . . . charcoal . . . and other indications of 1998; Brooks and Yellen 1987; Galanidou 1997, 2000; Spurling fire use . . . hinder the discovery and recognition of burnt re- andHayden1984;Yellen1977).Whilenumerousactivitiesleave mains” (Karkanas et al. 2007:198). These taphonomic complex- no tangible evidence for us to uncover (e.g., social interactions), ities constrain the attempts to identify early human use of fire other activities (e.g., tool making and food processing) con- and actually suggest that of the various possible methods, spa- tribute directly to the formation of the archaeological record. tial analysis of burned residues is the most accessible and effi- Hearths not only serve as spatial spots of accumulation but also cient, particularly when a durable component such as lithics is influence the patterns of distribution of certain size groups of concerned. the assemblage. Binford (1978, 1983) suggested that the forma- tion of certain spatial patterns during work around a hearth Spatial Analysis appears to be universal and that the distribution of waste often displays two concentric zones around the hearth: the drop zone A broad definition of spatial analysis was provided by Clarke in proximity to the hearth, where small fragments of bone/stone (1977), who perceived spatial archaeology as “the retrieval of are left in situ, and the toss zone, an area farther away from the information from archaeological spatial relationships and the hearth to which the larger debris is tossed. Thus, the area clos- S260 Current Anthropology Volume 58, Supplement 16, August 2017 est to the hearth is likely to display high quantities of small in sites of Champréveyres and Monruz in Switzerland, hearths are situ refuse. Notwithstanding, spatial analysis studies often con- characterized by various amounts of cobbles, stone slabs, and centrate on the larger refuse and features despite the fact that extremely abundant and well-preserved wood charcoal (Leesch “the data most likely to be informative . . . are very small refuse et al. 2005). Regardless of the remarkable preservation of these items, such as chipping debris, small bone fragments, and plant sites, the spatial distribution of burned flint microartifacts has macrofossils, which will often be found in primary context” proved to be the optimal indicator for the precise location of the (O’Connell 1987:104). Yet no conventional limit has been de- hearths (Leesch et al. 2005). fined as a critical size factor for determining what is considered small refuse, which can range from several millimeters (e.g., Dun- Spatial Analysis and Fire at Gesher Benot Ya‘aqov nell and Stein 1989; Fladmark 1982; Metcalfe and Heath 1990; Stein and Teltser 1989; Vance 1987) up to several centimeters The waterlogged site of GBY is located on the shores of the (DeBoer 1983; O’Connell 1987). paleo–Lake Hula in the northern Jordan Valley in the Dead Sea In conclusion, ethnographic observations suggest that the as- Rift. The Early to Middle Pleistocene sediments document an sociation between features (e.g., hearths) and the spatial dis- oscillating freshwater lake and represent some 100,000 years of tribution of artifacts can provide the contextual framework of hominin occupation (OIS 18–20) dating to 790,000 years ago. artifact concentrations. Consequently, in attempting to recon- Fourteen archaeological horizons indicate that Acheulian hom- struct the formation process of hearth-related spatial patterns, inins repeatedly occupied the lake margins, where they skillfully we can draw on the following inferences: a wide range of ac- produced stone tools, systematically butchered and exploited tivities is carried out in close proximity to hearths; hearths are animals, gathered plant food, and controlled fire. spatial spots of refuse accumulation, and small refuse is more The study of fire use at GBY followed the hearth-related spa- likely than large refuse to be left in situ; hearths are thus likely tial patterning discussed above. Our basic assumption was that to display dense concentrations of burned small refuse. Such clustersofdebris,specificallysmallburneddebris,areindicators hearth-related discard patterns have been reported from a va- of the locations of hearths, defined as “phantom hearths”— riety of archaeological settings, including open-air sites (e.g., features that lack other observable traits (e.g., structuring, dis- Gilead 1980; Gilead and Grigson 1984; Hietala 1983; Leesch coloration of sediments, ash, or charcoal). Leroi-Gourhan’sdef- et al. 2005; Sergant, Crombe, and Perdaen 2006) as well as rock- inition of structures latentes established the approach to such shelters and cave sites (e.g., Galanidou 1997; Vaquero and Pastó archaeological features, namely, that these can be discernible 2001), in all of which the hearths are readily identifiable fea- through observable patterns of artifacts’ spatial distributions tures. In addition, hearth-related spatial patternings were re- (Leroi-Gourhan and Brézillon 1972). These principles guided ported from different cultural settings of varying chronolo- the study of fire use at GBY, where large amounts of small gies. At the Middle Paleolithic occupation at Belvédère, clusters knapping waste products were recorded within the archaeo- of burned artifacts suggested the presence of hearths (Stapert logical layers (table 1). These include flint microartifacts (2– 1990). The Aurignacian of Abri Castanet (White et al. 2017) 20 mm in length) and macroartifacts (longer than 20 mm) re- presents spatial association between ivory beads and their covered during excavation and through postexcavation sorting production debris around the fire features. At the Magdalenian of the wet-sieved sediments. Experimental studies have dem-

Table 1. Area, volume, and counts of lithic artifacts in the levels of Layer II-6 from the topmost layer of the stratigraphic sequence (younger) to the lowermost (older)

Unburned flint Burned flint Basalt Limestone Level Areaa Volumeb 12cmc ≤2cmd 12cm ≤2cm 12cm ≤2cm 12cm ≤2cm L-1 23.79 4.28 1,192 53,081 17 754 1,044 2,745 42 1,506 L-2 25.62 3.07 469 73,064 5 563 915 3,889 23 2,154 L-3 17.92 2.50 494 91,050 16 877 656 2,597 33 1,570 L-4 16.64 2.16 635 105,022 16 1,406 1,033 7,279 45 4,727 L-4b 13.69 .82 151 7,182 5 443 579 1,008 33 8,778 L-5 13.39 1.20 186 31,862 12 1,211 227 2,540 25 1,996 L-6 12.62 1.38 325 12,282 9 251 363 493 35 331 L-7 12.60 1.38 598 23,455 19 677 412 955 69 828 Total 136.27 16.79 4,050 396,998 99 6,182 5,229 21,506 305 21,890 a Area (in square meters) represents the spatial extent of the excavated material. b Volume (in cubic meters) is the excavated area multiplied by the estimated mean of excavated thickness based on cross sections. c Items larger than 2 cm (i.e., macroartifacts: flakes and flake tools, cores and core tools, and bifacial tools). d Smaller items (i.e., microartifacts). Alperson-Afil Spatial Analysis of Fire S261 onstrated that direct contact of flint with fire of high tempera- lyzed, suggesting that percussive activities (e.g., nut processing) tures (3507C–5007C) results in distinctive thermal macrofrac- were carried out near the hearth and may have involved the use tures (e.g., potlid fractures), which are readily identified (Purdy of fire. 1982, 1975; Purdy and Brooks 1971; Sergant, Crombe, and Per- This paper presents new data on the possible association daen 2006). At GBY, the spatial distribution of flint artifacts was between percussive tools and fire at GBY. The study focuses examined using spatial analyses software (ESRI ArcMap 9.3), on the eight superimposed occupational levels of Layer II-6, available with the GIS package, in order to detect possible clus- a sedimentary sequence that represents some 10,000 years of ters of burned material. hominin occupation (Craig Feibel, personal communication). Such clusters of burned flint were recorded throughout the The levels of Layer II-6 were rapidly sealed, preserving the entire depositional sequence at GBY. They are relatively small original location of different artifacts (evidenced by the pres- and incorporate both burned and unburned flint microarti- ervation state of the lithics, the preservation of mollusk em- facts. Within the kernel of these clusters, the relative percent- bryos, the presence of conjoinable bones, and a lack of win- age of burned flint microartifacts is always higher than that of nowing; see, e.g., Ashkenazi et al. 2010, 2005; Feibel 2001; the unburned ones, and homogeneity analysis indicated that Goren-Inbar, Werker, and Feibel 2002; Rabinovich et al. 2012). the observed percentages of burned flint microartifacts in the center of these clusters are higher than what we would expect Percussive Tools if the distribution of the burning was uniform. In addition, the significance of the statistical tests carried out on the spatial dis- Percussive tools are an important component of the lithic as- tribution of burned flint microartifacts in each of the studied semblages of GBY (Goren-Inbar et al. 2015) and were most archaeological levels further substantiates the observed clus- tering. This extensive study was facilitated by the minimal effect of postdepositional processes on the spatial configuration of the archaeological material (Ashkenazi et al. 2005; Goren-Inbar, Werker, and Feibel 2002; Rabinovich et al. 2012), and its meth- odology and results were published in detail (Alperson-Afil 2008; Alperson-Afil and Goren-Inbar 2010; Alperson-Afil, Rich- ter, and Goren-Inbar 2007; Goren-Inbar et al. 2004), demon- strating with certainty that the hominins of GBY controlled fire and used it routinely.

Segregating Activities around the Fire at Gesher Benot Ya‘aqov Once the use and control of fire at GBY were established, the option to trace hearth-related spatial patterns became possi- ble. A single occupation level (Level 2), one of eight superim- posed occupational levels of Layer II-6, was selected, and its lithic, faunal, and botanical remains were spatially examined (Alperson-Afil et al. 2009). The analyses of Level 2 indicated that hominins carried out different activities in two distinct locations. Abundant knapping took place in the northwestern area, resulting in a dense concentration of unburned flint mi- croartifacts. Other significant aspects of this activity area in- clude the use of chopping tools and fish exploitation, the latter also recorded in the vicinity of the hearth, where greater var- iation was found in the activities carried out there. While flint knapping around the hearth was less intensive, basalt and lime- stone knapping was spatially restricted to the hearth, which also served as a focal point for biface modification and for ac- tivities involving the use of chopping tools, side scrapers, end scrapers, and awls. In addition, the differential preservation of fish and crabs, along with their spatial distribution, suggests that they were consumed near the hearth (Alperson-Afil et al. 2009). Another hearth-related activity was revealed when the Figure 1. Basalt percussors. a, #16326, Layer II-2/3. b, #14887, percussors and the pitted stones of Level 2 were spatially ana- Layer II-6, Level 3. S262 Current Anthropology Volume 58, Supplement 16, August 2017

flaked surfaces that sometimes bear signs of the heavy percus- sive force that caused their fragmentation. These anvils have a particular trapezoid geometry, with part of the upper surface occasionally sloping to form an acute angle with the lower surface (Goren-Inbar et al. 2015). Both thin anvils and percus- sors frequently exhibit pitting damage on their surfaces. Such damage, however, is recorded on a variety of other artifacts, all classified as pitted stones. Pitted stones (fig. 3) are thus either unflaked nodules or slabs or flaked artifacts that bear small pits (depressions) on their surfaces. These pits are often shallow, and their number and location on the surfaces varies. While some pits are organized in clusters of small pits on one of the artifact faces, others are less distinct and seem to resemble battering marks in restricted areas. In most cases the pits occur in the central part of the artifact and rarely on the marginal edges (Goren-Inbar et al. 2015)

Percussive Activities and Fire The eight sequential levels of Layer II-6 were spatially analyzed in order to explore possible associations between their hearth or hearths and their archaeological material. More specifically, the spatial position of the different types of basalt percussive tools was examined. The results of these analyses are presented in figure 4. Each level is illustrated in two stages; both present

Figure 2. Basalt thin anvils. a, #5255, Layer II-6, Level 1. b, #7695, Layer II-6, Level 6. likely used by the hominins for stone knapping, nut cracking, and a variety of food-processing activities (both vegetal and faunal). The percussive tools of GBY are modified mostly on ba- salt unmodified nodules or slabs and include percussors, pit- ted stones, and thin anvils. The basalt percussors (fig. 1) are rounded nodules, somewhat elongated and usually of cobble size, that present battering or small pitting damage on their surface. Some percussors are shattered as a result of their use in intensive knapping or percussive activity of hard materials. These artifacts show rough and uneven breakage surfaces that lack ventral face features and are classified as split percussors. Figure 3. Basalt-pitted stones. a, #2312 Layer II-6, Level 1. b, #14177 Thin anvils (fig. 2) are basalt slabs with two flat, parallel, un- Layer II-6, Level 2; giant core with a cluster of pits. Figure 4. Layer II-6, Levels 1–7. a, Complete field map superimposed on the kernel density map of burned flint microartifacts. b, Assemblage of basalt percussive tools superimposed on the kernel density map of burned flint microartifacts. S264 Current Anthropology Volume 58, Supplement 16, August 2017 the density map of the burned flint microartifacts of the ex- Where There’s Smoke, There’s Fire amined level and its association with (a) the map of the ar- Based on archaeological observations and analysis, fire was chaeological level as drafted in the field and (b) with the basalt used for the first time by hominins in Africa some 1.5 million percussive tools in their recorded excavated coordinates. years ago. Indications are available from cave sites in South Regardless of the percussive tools, examination of the field Africa (Swartkrans: Brain and Sillen 1988; Pickering 2012; Won- maps raises several interesting observations. First, there are derwerk: Beaumont 2011; Berna et al. 2012) and include burned clear differences between the occupational levels, probably re- sediments, bones, and stone artifacts. The use of fire in these flecting different site uses (e.g., elephant butchering in Level 1, two sites is acknowledged by most scholars despite the lack of biface production in Levels 4 and 4b). And yet despite these “hearth” features. Sites in East Africa (e.g., Koobi Fora: Isaac differences, fire is present in each of the occupational levels. and Harris 1978; Chesowanja: Gowlett et al. 1981) display clus- Interestingly, in many cases (particularly clear in Levels 3 and 6) ters of burned sediments in a hearth-like arrangement. How- the artifacts in the field map seem to be positioned in an arch- ever, as these are open-air sites that have been affected by ta- like manner, and their radial configuration looks as if it is sur- phonomic disturbances, the evidence is considered insufficient rounding the hearth or hearths (fig. 4). by many (but see Hlubik et al. 2017). In Eurasia, the earliest As for the distribution of percussive tools, the results show evidence is recorded from GBY, as discussed above. Elsewhere that in all occupational levels, the different classes of percussive in Eurasia, the use of fire is well attested from 0.4–0.3 Ma, and tools are found in association with each other; thus, where there accumulations of burned sediments, often in conjunction with is an anvil, there are percussors and pitted stones. This may in- burned stone artifacts and bones, occur regularly (Garrod and dicate that despite their morphological differences, they were Bate 1937; Gowlett 2006; Karkanas et al. 2007; Vertes and Do- used in similar or related tasks. When present, split percussors bosi 1990). At the 0.4 Ma site of Beeches Pit in England, the are spatially associated with percussors (excluding a single case evidence includes numerous burned flints, bones, shells, and in Level 5). Most important, the different classes of percussive charcoal, and their spatial patterning, including refitting series, tools are spatially associated with the hearths, suggesting that is in association with a hearth-like feature (Gowlett 2006; Gow- percussive activities, probably related to food processing, were lett et al. 2005). The difficulty is that Europe lacks conclusive carried out in the vicinity of the hearth (fig. 4). evidence for fire use during its early stages of occupation, ca. 1.0– 0.8 Ma (e.g., Parfitt et al. 2010). Some explain this gap as an in- Discussion dication that fire was used only sporadically before 0.4 Ma and that it was not an essential element in the subsistence of the tfirs Identifying Early Fire with Spatial Analysis Methods Europeans (Roebroeks and Villa 2011). Rightfully, Gowlett and The use of spatial analyses at GBY enabled the identification of Wrangham (2013) call for a more mature approach to early fire phantom hearths and their association with percussive activi- studies, one that does not ignore the findings of evolutionary, ties throughout the archaeological sequence. Our results illus- dietary, and social studies, which facilitate a comparative view trate that spatial analysis is an efficient method for identifying of early hominin adaptations and behavior. early use of fire, particularly at open-air sites where tapho- The concept of habitual use of fire is used to differentiate be- nomic conditions often limit the use of chemical and micro- tween an early, sporadic use and a later, controlled, and inten- scopic techniques. sive use. What, however, does sporadic use entail: limited pur- Such sites form the bulk of early hominin occupation areas, poses or short duration? Fire was discovered, domesticated, and and often their evidence for fire use is considered unreliable. used by early hominins. Habitual use, however, requires inten- Thus, review of the evidence reveals two major pitfalls—the sity and recurrence of the evidence, as documented in sites with misuse of hearths as exclusive indicators of fire use, and the long occupational sequences. This requirement excludes ha- question of fire use as sporadic versus habitual (Karkanas et al. bitual use of fire from the majority of early hominin habita- 2007; Roebroeks and Villa 2011; Shahack-Gross et al. 2014). tions, documented by short-term occupations of open-air sites. Hearths are variable features, and their preservation requires Clearly, along with the advantages that fire brought to homi- particular depositional conditions. Their identification requires nins, it also made them dependent on its benefits (in diet, tech- rejection of the involvement of natural agents (taphonomic nology, social life, and culture) and more and more vulnerable disturbances, natural fires) and the documentation of discrete to its loss. To deny Early to Middle Pleistocene hominins of the concentrations of burned residues. Such concentrations can habitual use of fire (e.g., Sandgathe 2017) is to ignore the ar- be identified using an array of methods: by the naked eye (e.g., chaeological record of their evolution, behavior, and culture. flint, bones, charcoal, ash) and by chemical and microscopic techniques (e.g., bones and sediments). This approach should References Cited be just as applicable to early prehistoric sites as it is to sites of modern humans, where burned flints are used to mark the lo- Alperson-Afil, Nira. 2008. Continual fire-making by hominins at Gesher Benot Ya‘aqov, Israel. Quaternary Science Reviews 27:1733–1739. cations of hearths regardless of chemical or microstratigraphic Alperson-Afil, Nira, and N. Goren-Inbar. 2010. The Acheulian site of Gesher Benot analyses (Sergant, Crombe, and Perdaen 2006). Ya‘aqov.Vol.2,Ancient flames and controlled use of fire.Dordrecht:Springer. Alperson-Afil Spatial Analysis of Fire S265

Alperson-Afil, Nira, D. Richter, and N. Goren-Inbar. 2007. Phantom hearths Gargett, R., and B. Hayden. 1991. Site structure, kinship, and sharing in Ab- and the use of fire at Gesher Benot Ya‘aqov, Israel. Paleoanthropology 1:1– original Australia. In The interpretation of archaeological spatial patterning. 15, http://www.paleoanthro.org/journal/content/PA20070001.pdf. E. M. Kroll and T. D. Price, eds. Pp. 11–32. New York: Plenum. Alperson-Afil, Nira, G. Sharon, M. Kislev, Y. Melamed, I. Zohar, S. Ashkenazi, Garrod, D. A. E., and D. M. A. Bate. 1937. The Stone Age of Mount Carmel. R. Rabinovich, et al. 2009. Spatial organization of hominin activities at Gesher London: Clarendon. Benot Ya‘aqov, Israel. Science 326:1677–1680. Gilead, I. 1980. A Middle Paleolithic open-air site near Tell Far’a, Western Ashkenazi, S., K. Klass, K. H. Mienis, B. Spiro, and R. Abel. 2010. Fossil em- Negev: preliminary report. Israel Exploration Journal 30:52–62. bryos and adult Viviparidae from the Early-Middle Pleistocene of Gesher Gilead, I., and C. Grigson. 1984. Fara II: a Middle Paleolithic open-air site in Benot Ya‘aqov, Israel: ecology, longevity and fecundity. Lethaia 43:116–127. the Northern Negev, Israel. Proceedings of the Prehistoric Society 50:71–97. Ashkenazi, S., U. Motro, N. Goren-Inbar, R. Biton, and R. Rabinovich. 2005. Goldberg, Paul, Christopher E. Miller, and Susan M. Mentzer. 2017. Recog- New morphometric parameters for assessment of body size in the fossil nizing fire in the Paleolithic archaeological record. Current Anthropology 58 freshwater crab assemblage from the Acheulian site of Gesher Benot Ya‘aqov, (suppl. 16):S175–S190. Israel. Journal of Archaeological Science 32:675–689. Goren-Inbar, N., Nira Alperson, M. E. Kislev, O. Simchoni, Y. Melamed, A. Bartram, L. E., E. M. Kroll, and H. T. Bunn. 1991. Variability in camp structure Ben-Nun, and E. Werker. 2004. Evidence of hominin control of fire at Gesher and bone food refuse patterning at Kua San hunter-gatherer camps. In The Benot Ya‘aqov, Israel. Science 304:725–727. interpretation of archaeological spatial patterning. E. M. Kroll and T. D. Price, Goren-Inbar, N., G. Sharon, Nira Alperson-Afil, and G. Herzlinger. 2015. A eds. Pp. 77–148. New York: Plenum. new type of anvil in the Acheulian of Gesher Benot Ya‘aqov, Israel. Phil- Bellomo, R. V., and John W. K. Harris. 1990. Preliminary report of actualistic osophical Transactions of the Royal Society B 370:20140353, doi:10.1098 studies of fire within Virunga National Park, Zaire: towards an understanding /rstb.2014.0353. of archaeological occurrences. In Evolution of environments and Hominidae Goren-Inbar, N., E. Werker, and Craig S. Feibel. 2002. The Acheulian site of in the African Western Rift Valley. N. T. Boaz, ed. Pp. 317–338. Martinsville: Gesher Benot Ya‘aqov. Vol. 2, The wood assemblage. Oxford: Oxbow. Virginia Museum of Natural History. Gowlett, J. A. J. 2006. The early settlement of northern Europe: fire history in Berna, Francesco, Paul Goldberg, Liora Kolska Horwitz, James Brink, Sharon the context of climate change and the social brain. Comptes Rendus Palevol Holt, Marion Bamford, and Michael Chazan. 2012. Microstratigraphic evi- 5:299–310. dence of in situ fire in the Acheulian strata of Wonderwerk Cave, Northern Gowlett, J. A. J., J. Hallos, S. Hounsell, V. Brant, and N. C. Debenham. 2005. Cape Province, South Africa. Proceedings of the National Academy of Sciences Beeches Pit: archaeology, assemblage dynamics and early fire history of a Mid- of the USA 109:1215–1220. dle Pleistocene site in East Anglia, UK. Journal of Eurasian Prehistory 3:3–38. Beumont, P. B. 2011. The edge: more on fire making by about 1.7 million years Gowlett, J. A. J., John W. K. Harris, D. Walton, and B. A. Wood. 1981. Early ago at Wonderwerk Cave in South Africa. Current Anthropology 52:585– archaeological sites, hominid remains and traces of fire from Chesowanja, 595. Kenya. Nature 294:125–129. Binford, L. R. 1968. Methodological considerations of the archaeological use of Gowlett, J. A. J., and Richard W. Wrangham. 2013. Earliest fire in Africa: to- ethnographic data. In Man the hunter. R. B. Lee and I. Devore, eds. Pp. 268– wards the convergence of archaeological evidence and the cooking hypoth- 273. Chicago: Aldine. esis. Azania: Archaeological Research in Africa 48:5–30. ———. 1978. Dimensional analysis of behavior and site structure: learning Isaac, G. L., and John W. K. Harris. 1978. Archaeology. In Koobi Fora Research from an Eskimo hunting stand. American Antiquity 43:330–361. Project, vol. 1. M. D. Leakey and R. E. F. Leakey, eds. Pp. 64–85. Oxford: ———. 1983. In pursuit of the past: decoding the archaeological record. Lon- Clarendon. don: Thames & Hudson. Hietala, H. 1983. Boker Tachtit: spatial distributions. In Prehistory and paleo- ———. 1998. Hearth and home: the spatial analysis of ethnographically doc- environments in the Central Negev, Israel, vol. 3. A. E. Marks, ed. Pp. 191– umented rock shelter occupations as a template for distinguishing between 216. Dallas, TX: Southern Methodist University. human and hominid use of sheltered space. In XIII UISPP Congress Pro- Hlubik, Sarah, Francesco Berna, Craig Feibel, David Braun, and John W. K. ceedings, Forli 8–14 September 1996. Carlo Peretto and Carlo Giunchi, eds. Harris. 2017. Researching the nature of fire at 1.5 Mya on the site of FxJj20 Pp. 229–239. Forli: ABACO Edizioni. AB, Koobi Fora, Kenya, using high-resolution spatial analysis and FTIR ———. 2001. Constructing frames of reference: an analytical method for ar- spectrometry. Current Anthropology 58(suppl. 16):S243–S257. chaeological theory building using hunter-gatherer and environmental data Karkanas, P., R. Shahack-Gross, A. Ayalon, M. Bar-Matthews, Ran Barkai, A. sets. Berkeley: University of California Press. Frumkin, Avi Gopher, and M. C. Stiner. 2007. Evidence for habitual use of Brain, C. K., and A. Sillen. 1988. Evidence from the Swartkrans Cave for the fire at the end of the Lower Paleolithic: site-formation processes at Qesem earliest use of fire. Nature 336:464–466. Cave, Israel. Journal of Human Evolution 53:197–212. Brooks, A. S., and J. E. Yellen. 1987. The preservation of activity areas in the Leesch, D., J. Bullinger, M.-I. Cattin, W. Muller, and N. Plumettaz. 2005. archaeological record: ethnoarchaeological and archaeological work in North- Hearths and hearth-related activities in Magdalenian open-air sites: the case west Ngamiland, Botswana. In Method and theory for activity area research: studies of Champréveyres and Monruz (Switzerland) and their relevance to an ethnoarchaeological approach. S. Kent, ed. Pp. 63–106. New York: Co- an understanding of Upper Paleolithic site structure. Paper given at “The lumbia University Press. Magdalenian in Central Europe: New Finds and Concepts,” Rzeszow, Po- Canti, M. G., and N. Linford. 2000. The effects of fire on archaeological soils land, September 20–23, 2005. and sediments: temperature and colour relationships. Proceedings of the Pre- Leroi-Gourhan, A. 1988. Dictionnaire de la préhistoire. Paris: Presses Univer- historic Society 66:385–395. sitaires de France. Clarke, D. L. 1977. Spatial information in archaeology. In Spatial archaeology. Leroi-Gourhan, A., and M. Brezillon. 1972. Fouilles de Pincevent: essai d’analyse D. L. Clarke, ed. Pp. 1–32. London: Academic Press. ethnographique d’un habitat Magdalenien. Paris: Editions du CNRS. DeBoer, W. R. 1983. The archaeological record as preserved death assemblage. Linford, N. T., and M. G. Canti. 2001. Geophysical evidence for fires in antiq- In Archaeological hammers and theories. J. A. Moore and A. S. Keene, eds. uity: preliminary results from an experimental study. Archaeological Pros- Pp. 19–36. New York: Academic Press. pection 8:211–225. Dunnell, R. C., and J. K. Stein. 1989. Theoretical issues in the interpretation of Mallol, Carolina, F. W. Marlowe, B. M. Wood, and C. C. Porter. 2007. Earth, microartifacts. Geoarchaeology 4:31–42. wind, and fire: ethnoarchaeological signals of Hadza fires. Journal of Archae- Feibel, Craig S. 2001. Archaeological sediments in lake margin environments. ological Science 34:2035–2052. In Sediments in archaeological contexts. J. K. Stein and W. R. Farrand, eds. Mallol, Carolina, and Auréade Henry. 2017. Ethnoarchaeology of Paleolithic Pp. 127–148. Salt Lake City: University of Utah Press. fire: methodological considerations. Current Anthropology 58(suppl. 16): Fladmark, K. R. 1982. Microdebitage analysis: initial considerations. Journal S217–S229. of Archaeological Science 9:205–220. Metcalfe, D., and K. M. Heath. 1990. Microrefuse and site structure: the hearths Galanidou, N. 1997. “Home is where the hearth is”: the spatial organization of and floors of the Heartbreak Hotel. American Antiquity 55:781–796. the Upper Palaeolithic rockshelter occupations at Klithi and Kastritsa in O’Connell, J. F. 1987. Alyawara site structure and its archaeological implica- Northwest Greece. BAR International Series 687. Oxford: BAR. tions. American Antiquity 52:74–108. ———. 2000. Patterns in caves: foragers, horticulturists, and the use of space. O’Connell, J. F., K. Hawkes, and N. B. Jones. 1991. Distribution of refuse- Journal of Anthropological Archaeology 19:243–275. producing activities at Hadza residential base camps: implications for anal- S266 Current Anthropology Volume 58, Supplement 16, August 2017

yses of archaeological site structure. In The interpretation of archaeological Sergant, J., P. Crombe, and Y. Perdaen. 2006. The “invisible” hearths: a con- spatial patterning. E. M. Kroll and T. D. Price, eds. Pp. 61–76. New York: tribution to the discernment of Mesolithic non-structured surface hearths. Plenum. Journal of Archaeological Science 33:999–1007. Parfitt, S. A., N. M. Ashton, S. G. Lewis, R. L. Abel, G. R. Coope, M. H. Field, Shahack-Gross, R., Francesco Berna, P. Karkanas, C. Lemorini, Avi Gopher, R. Gale, et al. 2010. Early Pleistocene human occupation at the edge of the and Ran Barkai. 2014. Evidence for the repeated use of a central hearth at boreal zone in northwest Europe. Nature 466:229–233. Middle Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of Archaeo- Pickering, T. R. 2012. What’s new is old: comments on (more) archaeological logical Science 44:12–21. evidence of one million-year-old fire from South Africa. South Africa Jour- Spurling, B., and B. Hayden. 1984. Ethnoarchaeology and intrasite spatial nal of Science 108:1–2. analysis: a case study from the Australian Western desert. In Intrasite spatial Pop, E., W. Kuijper, E. van Hees, G. Smith, A. García-Moreno, L. Kindler, analysis in archaeology. H. J. Hietala, ed. Pp. 224–241. Cambridge: Cam- S. Gaudzinski-Windheuser, and W. Roebroeks. 2016. Fires at Neumark- bridge University Press. Nord 2, Germany: an analysis of fire proxies from a Last Interglacial Middle Stahlschmidt, M. C., Christopher E. Miller, B. Ligouis, U. Hambach, P. Gold- Palaeolithic basin site. Journal of Field Archaeology 41:603–617. berg, Francesco Berna, D. Richter, B. Urban, J. Serangeli, and N. J. Conard. Purdy, B. A. 1975. Fractures for the archaeologist. In Lithic technology: making 2015. On the evidence for human use and control of fire at Schöningen. and using stone tools. E. Swanson, ed. Pp. 133–141. Paris: Mouton. Journal of Human Evolution 89:181–201. ———. 1982. Pyrotechnology: prehistoric application to chert materials in Stapert, D. 1990. Middle Palaeolithic dwellings: fact or fiction? some applica- North America. In Early pyrotechnology: the evolution of the first fire-using tions of the ring and sector method. Palaeohistoria 32:1–19. industries. T. A. Wertime and S. F. Wertime, eds. Pp. 31–44. Washington, Stein, J. K., and P. A. Teltser. 1989. Size distributions of artifact classes: com- DC: Smithsonian Institution Press. bining macro- and micro-fractions. Geoarchaeology 4:1–30. Purdy, B. A., and H. K. Brooks. 1971. Thermal alternation of silica minerals: Vance, E. D. 1987. Microdebitage and archaeological activity analysis. Archae- an archeological approach. Science 173:322–325. ology 40:58–59. Rabinovich, R., S. Gaudzinski-Windheuser, L. Kindler, and N. Goren-Inbar. Vaquero, M., and I. Pasto. 2001. The definition of spatial units in Middle Pa- 2012. The Acheulian site of Gesher Benot Ya‘aqov. Vol. 3, Mammal taphon- laeolithic sites: the hearth-related assemblages. Journal of Archaeological Sci- omy: the assemblages of Layers V-5 and V-6. Dordrecht: Springer. ence 28:1209–1220. Roebroeks, W., and P. Villa. 2011. On the earliest evidence for habitual use of Vertes, L., and V. T. Dobosi. 1990. Fireplaces of the settlement. In Vértesszőlős fire in Europe. Proceedings of the National Academy of Sciences of the USA site, man and culture. M. Kretzoi and V. T. Dobosi, eds. Pp. 519–521. Buda- 108:5209–5214. pest: Akadémiai Kiadó. Sandgathe, Dennis M. 2017. Identifying and describing pattern and process in White, Randall, Romain Mensan, Amy E. Clark, Elise Tartar, Laurent Marquer, the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): Raphaëlle Bourrillon, Paul Goldberg, Laurent Chiotti, Catherine Cretin, S360–S370. William Rendu, Anne Pike-Tay, and Sarah Ranlett. 2017. Technologies for Schiegl, Solveig, Paul Goldberg, Ofer Bar-Yosef, and Steve Weiner. 1996. Ash the control of heat and light in the Vézère Valley Aurignacian. Current An- deposits in Hayonim and Kebara caves, Israel: macroscopic, microscopic thropology 58(suppl. 16):S288–S302. and mineralogical observations, and their archaeological implications. Jour- Yellen, J. E. 1977. Archaeological approaches to the present: models for recon- nal of Archaeological Science 23:763–781. structing the past. New York: Academic Press. Current Anthropology Volume 58, Supplement 16, August 2017 S267

Evidence of Hominin Use and Maintenance of Fire at Zhoukoudian

by Xing Gao, Shuangquan Zhang, Yue Zhang, and Fuyou Chen

Evidence for “controlled use of fire” by Homo erectus pekinensis at Zhoukoudian Locality 1 was initially discovered in the early 1930s and was widely accepted as the earliest such record in human evolutionary history for more than half a century. However, since the mid-1980s, this evidence has been questioned. Some of the questions were based on new research results, including geochemical and taphonomic studies conducted in the 1990s. Others are hypothetical and to some extent stem from a theoretical shift in ideas about early hominin subsistence capabilities, including hunting big game and using fire. Limited access to Chinese archaeological collections and literature, limited geo- logical sampling, and postdepositional disturbance might all contribute to the results of these recent investigations. A thorough review of original field notes, excavation reports, and research papers leads to the conclusion that fossils, cultural materials, and traces of fireuseincertainhorizonsatthesiteareabundant,unambiguous,andmutuallysup- portive. New field investigations and laboratory analyses at Locality 1, ongoing since 2009, have yielded new evidence that indicates that Layer 4 contains clear-cut evidence for in situ use of fire. Future research may well reveal similar evidence for Layers 8–9 and 10, potentially resolving this ongoing debate comprehensively.

Zhoukoudian (ZKD, also Choukoutien or Chou-kou-tien in et al. 1985b). The detailed chronology of ZKD Locality 1 es- early literature) Locality 1, close to Beijing, has been well known tablished by early investigations was reviewed by Zhao et al. since the 1920s because of the discovery of Sinanthropus (1985a) and Goldberg et al. (2001). Shen et al. (2001) published (Homo erectus pekinensis) fossils, stone artifacts, and evidence older ages for key horizons at the site using mass spectro- of hominin-controlled use of fire (Black 1926, 1931; Pei 1934). metric U-series dating of intercalated calcite samples: 400 kya Deposits at the site have been divided into 13 stratigraphic for Layers 1–2, 500 kya for the upper part of Layer 5, and layers (fig. 1), with Layer 4 referred to as the “Upper Cultural 600 kya for the middle and lower parts of Layer 5. In 2009, Horizon” and Layer 8–9 the “Lower Cultural Horizon” (Jia Shen et al. used 26Al/10Be burial dating to determine the age of 1959; Pei and Zhang 1985). With respect to the time frame of the lower component of the deposits (Layers 7–10) yielding an hominin occupation at the site, fission-track dating was em- estimate of 770 kya (Shen et al. 2009). ployed in the early 1980s, which determined that the formation For more than half a century, burned items (including of Layer 10 probably took place 462 5 45 kya while Layer 4 stones, bones, and seeds; fig. 2), charcoal fragments, and “ash was deposited approximately 299 5 55 kya (Guo et al. 1980, accumulations” unearthed from the site had been proposed 1991). Thermoluminescence dating was also applied at the site, and generally accepted as the earliest evidence of “controlled generating an age of 417–592 kya for Layer 10 and 292–312 kya use of fire” by early hominins, evidence supplemented by the for Layer 4 (Pei et al. 1985). Uranium-series (U-series) dating results of chemical analyses that identified free carbon in the on animal teeth produced an age range of 230–256 kya for burned fossils and “ash” remnants (Black 1931; Jia and Huang Layers 1–3andanindefinite date of 1300 kya for Layer 8 (Zhao 1984; Oakley 1955; Pei and Zhang 1985; Shen et al. 2004; Wu 1999). Since the mid-1980s, some scholars have questioned the evidence of in situ burning at the site and thus the notion that H. erectus pekinensis had the capability to control fire (e.g., Binford and Ho 1985; Binford and Stone 1986; Boaz Xing Gao Shuangquan Zhang Yue Zhang is Professor and , , and et al. 2004; Goldberg et al. 2001; James 1989; Weiner et al. Fuyou Chen are Associate Professors in the Key Laboratory for 1998). Considering the new discoveries of evidence for hominin Vertebrate Evolution and Human Origins at the Institute of Verte- fi brate Paleontology and Paleoanthropology of the Chinese Academy of use of re from even earlier contexts in Israel and South Africa Sciences (142 Xizhimenwai Street, Beijing 100044, China [gaoxing@ivpp (Berna et al. 2012; Goren-Inbar et al. 2004), the possible signs of fi .ac.cn, [email protected], [email protected], chenfuyou re use at ZKD would no longer constitute the earliest. How- @ivpp.ac.cn]). This paper was submitted 25 VII 16, accepted 8 III 17, ever, ZKD Locality 1 is still worth further investigation as a case and electronically published 28 VI 17. study for Pleistocene hominin adaptation as well as with respect q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0010$10.00. DOI: 10.1086/692501 S268 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 1. Sketch of the stratigraphy of the Western Section of ZKD Location 1 (after Jia 1959).

to the variability and diversity of human behavior in different samples of the apparently burned materials to Paris for com- geographic regions. parative studies with similar items found in prehistoric sites in Europe. Chemical analysis of these materials was conducted at the Laboratory of Mineralogy of the Paris Museum. Free Early Research on the Use and Maintenance carbon was detected, indicating that the blackened antler and of Fire at Zhoukoudian mammal bones had indeed been burned. Later, similar anal- yses were carried out in the Department of Pharmacology of The ZKD site was treated as a fossil vertebrate locality when Peking Union Medical College on samples of possible ash and scientific investigations were initiated there in 1918. In 1929, associated charred bone fragments. Once again, the possibil- Pei Wenzhong (also Pei Wen-Chung and W. C. Pei), the ZKD ity of heavy metal contamination was ruled out, the presence field director at that time, began to pay attention to possible in the sediments of a significant amount of carbon was dem- evidence of fire use at the site: mainly charred or partly cal- onstrated, and a significant free carbon component was con- cined animal bones (Black 1931). In late 1930, P. Teilhard firmed (Black 1931). Based on such analyses and the associ- de Chardin, a principal investigator on the ZKD project, took ation of the burned materials with Homo erectus pekinensis Gao et al. Fire Evidence from Zhoukoudian S269

Figure 2. Burned stones, bones, and tree seeds collected in the 1930s. A color version of this figure is available online.

fossils and stone artifacts, Davison Black announced that “It is fire. (1) Homo erectus pekinensis knew how to use fire and thus clear beyond reasonable doubt that Sinanthropus knew had the ability to control and maintain it. (2) They did not the use of fire” (Black 1931:108). possess the capability to manufacture fire. Instead, they According to original field notes and other documentation, may have introduced naturally occurring fire into the cave. evidence of fire use detected in early excavations at ZKD in- (3) They were not adept at maintaining fire throughout the cluded burned bones, rocks, arboreal (Celtis) seeds, charcoal, period of their occupation of the site. Sometimes they had to and ash. Such remains were recovered from four layers in live without fire, which is indicated by the lack of continuous Locality 1, namely Layer 4–5, Quartz II, Layer 8–9, and the accumulations of ash and charcoal. (4) The use of fire was lower horizon of Layer 10. Burned items were often found to very important for their survival. They used fire to keep have been concentrated in certain spots; in particular, three dangerous predators out of their cave home, to cook food, to piles of ash were discovered and recorded in the Gezitang (Pi- illuminate the dark niches of the cave, and to keep warm geon Hall or Chamber of the Pigeons). Two were located con- during North China’s cold winters. (5) The discovery formably on top of a large limestone slab, and some lime- and recognition of evidence of fire use at ZKD is a milestone stone breccia was burned into lime, making the sediments a in the history of research on human evolution (Zhang 2004). multicolored mixture of dark red, brown, and white. In some The conclusion that in situ burning as a result of human horizons, fine sediments became layers of hard latosol (Pei activity occurred at the ZKD site was reinforced by later in- and Zhang 1985). vestigations. More burned materials and other pyrogenic traces Since the time of Black’s announcement, ZKD Locality 1 were unearthed, and more physical and chemical analyses were was widely accepted as preserving the earliest record of the conducted. In the 1980s, fission-track and thermolumines- “controlled use of fire” in human prehistory (Movius 1948; cence dating were carried out on sediment samples from the Oakley 1956; Zhang 1985). Hallam Movius even provided a site, yielding reasonable ages for key cultural horizons that vivid description of the way fire was used by H. erectus pe- are roughly consistent with the ages obtained by other dating kinensis: methods. This indicates that the sphenes in the Layer 4 and Layer 10 ash samples were heated above the annealing tem- Fire was a basic item in his daily life. He presumably cooked perature during H. erectus pekinensis’s occupation of the site. his meat over the open hearth in which he burned the wood Otherwise, ages obtained from these samples should be much of the Redbud (Cercis blackii), a type of shrub. Since fire older (Guo et al. 1980, 1991; Liu et al. 1985; Pei 1985). There- would have provided warmth in the then-existing cave, and fore, these analytical results signaled high-temperature burning since it would keep predatory animals away at night, it must events at the site, especially from the Layer 10 ash remains. have been an immense asset to him. (Movius 1948:402) In 2004, Shen et al. (2004) performed research on the pres- Zhang Senshui (S. S. Chang) summarized perceptions of ence of elemental carbon (EC) in various samples collected H. erectus pekinensis’s capabilities with respect to the use of from ZKD Locality 1. “Elemental carbon” is an abbreviated S270 Current Anthropology Volume 58, Supplement 16, August 2017 name for a C-rich, O-H-S-N-depleted substance produced in statement was subsequently used as a starting point for a se- the process of combustion, and it usually occurs in charcoal, ries of challenges to the notion that ZKD Locality 1 preserves soot, fusain, microcrystalline graphite, carbon black, and sim- evidence of human use and maintenance of fire. ilar substances and is considered one of the indicators of in In the following year, Binford and Stone (1986) published situ use of fire by humans. Analyzed samples were taken from another paper on ZKD. This article was the outcome of Bin- Layers 4, 7, and 10 and from natural, nonanthropogenic de- ford’s experiences during a short visit to the site and 4 days of posits outside the cave as well as from an ash lump collected examination of 1,523 pieces of animal bone from the cave. in the 1930s curated in the ZKD site museum. Shen et al. Binford identified carnivore chew marks, stone-tool cut marks, (2004) discovered that the concentration of EC in the samples manganese stains, traces of fire alteration, and discrepancies taken from Layer 10 is much higher than that from other in the taxa and skeletal elements represented. They concluded samples and concluded that the sampling loci might have that the dominant contributors to the ZKD bone accumu- been close to a hearth. lations were denning animals, especially hyenas, implying that Based on original field notes, excavation reports, and re- the hominins who occupied the cave were not big-game hunt- search papers, it can be concluded that the excavated materials ers but occasional scavengers. There was strong evidence for and traces relating to fire-use activities in certain horizons at the controlled use of fire by H. erectus pekinensis during the the ZKD site are abundant, comprehensive, and mutually sup- late phases of occupation at the site, indicated by a small num- portive. While evidence for the in situ use of fire at ZKD was ber of equid bones in the upper levels bearing evidence of being questioned by some scholars, other researchers voiced burning. However, they maintained that such evidence for the strong support for this evidence. They argued that evidence of early occupations was lacking because most of the blackened “controlled use of fire” at the site as a whole is clear, strong, bones were actually mineral stained, and the few burned dry and internally consistent and that any questions based on ei- bones from the earlier levels were likely the result of modifi- ther limited taphonomic observations or analyses of samples cation by natural fires and were, therefore, not related to the taken from a few localized spots would not be sufficient (Liu behavior of hominins. The multicolored layers identified as et al. 1998; Zhong et al. 2014). Some also agreed that further ash yielded no evidence that these strata were produced by research would be necessary to solve the issue and put an end fire, and they believed that until they were subjected to more to the controversy. sophisticated analysis, their identification as ash would remain in question (Binford and Stone 1986). The opportunity for such geochemical analyses finally ar- Historical Debates over the Evidence rived in 1996 and 1997 when Weiner et al. (1998) launched a fi for Fire Use at Zhoukoudian systematic eld project at ZKD that included section clean- ing, sampling, and geological observations on the Western Ideas about Homo erectus pekinensis’s capability to use fire Section of Locality 1 along with subsequent geological, chem- took a sudden turn in 1985 when Binford and Ho published ical, and taphonomic analyses and experimental studies. Their their paper on the taphonomy at ZKD. From a taphonomic principal interest was looking for evidence of in situ use of point of view, based on information derived from early field fire. During the cleaning of geological sections, they collected notes, photos, reports, and papers, they questioned the long- 42 large-animal and 278 microfaunal elements mainly from established notion that ZKD Locality 1 was the cave home of the upper part of Layer 10 (Weiner et al. 1998). Five of the H. erectus pekinensis, the location to which hominins brought macrofaunal bone fragments were uniformly black to gray in back game to share with the group and where fire was used and color in fresh fracture surfaces. Infrared spectra showed that maintained to cook food and provide light and heat within the the insoluble residues extracted from the blackened bones were cave. First, they asserted that the association of hominin fossils all characteristic of burned bone organic matrix. Seven micro- and stone tools was weak and that the so-called ash layers were faunal bones were uniformly black and probably burned. Many not hearth remnants and that no structured hearth arrange- of the microscopic pieces of bone were well rounded, possibly ments could be confirmed. Second, they concluded that many, due to transport or to carnivore digestion. In the “ash layer” in if not most, of the animal bones in ZKD Locality 1 were ac- the upper part of Layer 10, they did not detect siliceous ag- cumulated by hyenas, wolves, and other carnivores who oc- gregates or sufficient amounts of potassium and other prop- cupied the site rather than by hominins as the consequence of erties such as phytoliths as indicators of the presence of ash hominin hunting behavior. They also suggested that the ap- resulting from burning wood. As to the “hearths” reported in parent ash was largely owl or other raptor droppings or the the lower part of Layer 10, they found that the deposits were by-product of natural, nonanthropogenic fires. In the end, in fact composed of finely laminated silt, clay, and organic their conclusions were rather inconclusive: “What, then, was matter, and they pointed out that the fine lamination of such life like in the ‘cave home of Beijing man?’ We think we must sediment was created by still-water deposits. conclude that we do not know. Hominids were regularly in In addition, this research team found several stone artifacts the cave. They regularly used stone tools there and probably in the upper part of Layer 10 and observed a close association used fire there” (Binford and Ho 1985:429). Nevertheless, their between those artifacts and the burned bones. They also no- Gao et al. Fire Evidence from Zhoukoudian S271 ticed that the percentage of the faunal assemblages that were Recent Investigations and Research Results burned (2.5% of the microfaunal remains and 12% of the large-animal bones) was similar to those observed in younger Systematic research-oriented salvage excavations have been cave sites with clear evidence of the use of fire. Nevertheless, undertaken at the Western Section of ZKD Locality 1 since they noted that “as some of the sediments of Layer 10 were 2009. These new excavations are aimed at rescuing fossils, deposited under water, we cannot be sure that the bones, artifacts, and taphonomic information from blocks of sedi- including the large burned and unburned bones, as well as ments in imminent danger of erosion and collapse, stabilizing fi the artifacts are in their original discard location” (Weiner extant deposits and pro les, searching for new archaeological fi et al. 1998:253) and concluded that remains using up-to-date eld technologies, and conducting systematic geological sampling and multidisciplinary analyses on the basis of the absence of ash or ash remnants (siliceous to reconstruct the formation history of the site and to resolve aggregates) and of in situ hearth features that there is no various controversies, including the issue of hominin use of fi 2 direct evidence for in situ burning in Layers 4 and 10. Most re. The excavation area was initially about 20 m and is now 2 of the fine-grained sediments in the site were water laid, and less than 10 m in because of the slope of the South and West even if ash remains could be recognized, it would be difficult Sections and the sharp incline of the original cave wall to the to demonstrate where they were produced. The co-occurrence north. Thus far, the excavation has cut through Layers 3, of burned black bones and quartzite artifacts in the same 4, and 5 and has reached the upper part of Layer 6, during fl layers is only suggestive of a cultural association, and hence which nearly 5,000 stone artifacts (mainly akes, cores, chunks, of the use of fire by humans, but does not prove it. and scrapers made on vein quartz) were unearthed along with numerous vertebrate fossils (Zhang et al. 2016). Most of the This research not only “damps ancient Chinese fires” but also excavated materials derive from Layer 4, the Upper Cultural called into question whether Homo erectus was capable of using Horizon, which clearly indicates that the formation of this fire (Wuethrich 1998). unit is closely associated with hominin activities. Considerable In 2000, Boaz and his colleagues published a paper in progress has been made on the analysis of the lithic collec- fi fi fi which they concluded that hyenas were responsible for most tion and veri cation of re use as a result of these new eld of the faunal accumulation and modification at the site, in- investigations. cluding that of H. erectus, and that hominins did not occupy ZKD Locality 1. Instead, their remains were transported into New Discoveries and Observations in the Field the cave by hyenas. They suggested that further research was needed to determine the true nature and scope of the pres- Discoveries relevant to evidence of the use of fire by Homo ence of stone tools, cut marks on bone, and fire in the cave erectus pekinensis recently derived from Layer 4 include the (Boaz et al. 2000). following data classes and phenomena: In 2001, Goldberg and his colleagues published a detailed a) Burned areas or hearths. Three areas characterized by in- geological analysis of ZKD as another important outcome of tense rubification of sediments and concentrations of po- the 1996–1997 field project. They stated that sediments from tential ash remnants were encountered. Two loci yielded Layers 10 through 3 show extensive water deposition of fine, signals of high magnetic susceptibility, rubification, and silt-sized material derived from outside the cave. They sug- maximum heating temperatures (more information on such gested that the site was open to varying extents throughout analyses is given below) and thus can be identified as most of its depositional history and that the so-called ashes in hearths. One such feature is outlined by rocks (fig. 3a). Layer 4 represent subaerial water-laid silt deposits accumu- b) Burned bones. Many uniformly black mammal bone frag- lated after the collapse of the brecciated roof during the for- ments were recovered from this horizon, providing clear mation of Layer 6 (Goldberg et al. 2001). They concluded that information on provenience, taphonomy, and association no irrefutable evidence of in situ burning by humans in any with lithic artifacts (fig. 3b). horizon of the deposits at ZKD could be found. In 2004, Boaz c) Heated limestone and lime. A pile of limestone blocks in et al. (2004) reinforced their interpretation of H. erectus peki- the northeastern corner of the excavation pit was deter- nensis’s pyrotechnological skills. Based on their analysis of mined to have been severely heated, and some pieces were published and unpublished sources, they constructed a digi- already transformed into white lime. These rocks may have tized three-dimensional excavation grid of Locality 1 to assess been part of a constructed hearth; however, their enclos- the spatial relationships of the human fossils, artifacts, and ing matrix had already collapsed, thus no clear structure other materials. Their observations are largely a reiteration of remained. the research results reported by Goldberg et al. (2001), but d) Evidence of a cave or rock-shelter. A large pile of limestone Boaz et al. (2004:546) nevertheless concluded that “despite breccia of various sizes was encountered on top of Layer 4 recent geochemical and sedimentological studies that modify sediments in the northern part of the excavation pit. This earlier interpretations of the use of fire at the site, there are is clearly the result of the collapse of the cave roof, indi- signs that fire was used by hominids at Zhoukoudian.” cating that during the formation of Layer 4, at least part of S272 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 3. Hearth (a) and burned bones (b) unearthed recently from Layer 4. A color version of this figure is available online.

ZKD Locality 1 retained the integrity of a cave suitable for strata could be the result of primarily natural processes, some hominin occupation, and the collapse process at other may be closely related to the activities of human ancestors. parts of the cave could have started earlier. e) The role of water in the deposition of sediments. Bedding The Extraction of Siliceous Aggregates structures were detected in certain layers and loci in this and Experimental Studies horizon, but it is not a uniform phenomenon. Such de- posits occurred mostly in the southwestern corner of the The absence of siliceous aggregates and potassium in Layer 10 excavation area. Excavation clearly demonstrates that the is the principal reason that Weiner et al. (1998) argued that Layer 4 deposits dip toward the southwest. Therefore, that there was no clear evidence for in situ fire use at ZKD. New part of the cave must be lower than other locations and attempts have been recently made to extract siliceous aggre- thus more easily subject to the action of water. gates and other botanical microfossils from the upper part of f ) The Layer 4 deposit at ZKD Locality 1 is thick and com- the deposits at ZKD Locality 1 (Zhong et al. 2014). Four sam- plicated. This unit can be further divided into several dis- ples were extracted from potential ash accumulations in Layer 4 crete strata representing different taphonomic processes in- and Layer 6 (three from the former and one from the latter), cluding geological and anthropogenic agents. While some where new excavations exposed fresh deposits more suitable Gao et al. Fire Evidence from Zhoukoudian S273

Table 1. Elementary sorts and main compositions of the insoluble phase of geological samples and modern tree ash

Weight percentage (%) Samples Al Si K Fe Main composition of the insoluble phase

ZKD1 5.88 18.46 1.21 1.07 SiO2,C,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 ZKD2 6.04 22.21 1.27 3.53 SiO2,C,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4, KAlSi3O8 ZKD3 8.96 28.94 2.73 3.00 SiO2, C, KFe.28Al.72Si3O8, K(Al1.91Fe.09)(Si3Al)O10(OH), K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 ZKD4 6.61 31.34 2.94 1.89 SiO2, C, KFe.28Al.72Si3O8, K(Al1.91Fe.09)(Si3Al)O10(OH), K(Si3Al)O8, K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Gleditsia sp. 1.03 26.53 1.21 1.22 SiO2, C, KFe.28Al.72Si3O8, K(Si3Al)O8, KAlSi3O8,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Diospyros sp. 5.22 25.99 1.49 .76 SiO2,C,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4, K(Si3Al)O8, KFe.28Al.72Si3O8 Zanthoxylum sp. 3.28 17.25 1.61 .83 SiO2, C, KAlSi3O8,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Prunus sp. .59 23.13 4.99 1.07 SiO2, C, KFe.28Al.72Si3O8, K(Si3Al)O8 Ginkgo sp. 3.93 27.45 2.15 .88 SiO2, C, K(Si3Al)O8, KFe.28Al.72Si3O8,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Ziziphus sp. 3.56 22.46 1.07 .18 SiO2, C, KFe.28Al.72Si3O8, KAlSi3O8,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Cercis sp. 5.16 15.51 6.02 .05 SiO2, C, K(Si3Al)O8, KFe.28Al.72Si3O8 Morus sp. 2.28 26.58 2.32 1.07 SiO2, C, KFe.28Al.72Si3O8, KAlSi3O8,KA12(SiAl)4O10(OH)2 Celtis sp. .94 26.38 .33 .06 SiO2, C, KFe.28Al.72Si3O8 Styphnolobium sp. 1.06 29.37 .34 .31 SiO2, C, KFe.28Al.72Si3O8, KAlSi3O8,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Juglans sp. 2.54 21.98 1.54 .23 SiO2, C, K(Si3Al)O8 Populus sp. 2.29 16.20 1.00 .25 SiO2, C, KFe.28Al.72Si3O8, K(Si3Al)O8,K2(Al3.74Fe.26)(Si6Al2O20)(OH)4 Source. Zhong et al. 2014. for such analysis. The samples were soaked in 6N hydro- Fe, and K) contained in fresh ash resulting from burning the chloric acid (HCl) and 6N nitric acid (HNO3), respectively, wood of different trees (table 1). From this, we may reason- for 24 hours. After drying, a Hitachi S4800 field-emission ably conclude that the samples analyzed by Weiner et al. (1998) scanning electron microscope was used to analyze the ele- happened to have derived from the burning of arboreal spe- mental composition of the insoluble phases in these speci- cies containing low quantities of siliceous aggregates and el- mens, and a Bruker D8 advance X-ray diffractometer was emental potassium. Therefore, samples taken from different employed to analyze their material composition. loci in the section may produce highly variable results and The results revealed a large number of sintered agglom- lead to different conclusions. erates present in the insoluble residual phases of the samples, and the elements associated with siliceous aggregates (Al, Si, Research on Magnetic Susceptibility K, Fe) were evident in all samples (table 1). In all four sam- and Color Measurements ples, silicon content was highest, followed by aluminum. Po- tassium, which can only be preserved in siliceous aggregates, Another series of analyses carried out recently at ZKD Lo- and iron were detected in lower concentrations. Material cality 1 involved detailed measurements of magnetic suscep- composition analysis demonstrated that the substances in the tibility, color, and diffuse reflectance spectra of sediments in potential ash remnants existed mainly in the form of SiO2, the area newly exposed during recent excavations in Layer 4 EC, elemental potassium, and silicates. The test results show (Zhang et al. 2014). Magnetic analysis and color measure- clearly the presence of EC, elemental K, and siliceous aggre- ments have been proven to be effective methods for detecting gates in the insoluble phases of the ash from Layer 4 and the presence of heated areas in archaeological sites (e.g., Car- Layer 6 and that “these results should spur reconsideration rancho and Villalain 2011; Jordanova, Petrovsky, and Kova- of the conclusions drawn by Weiner and his colleagues and cheva 2001). In order to analyze the sedimentary features of provide additional supporting evidence of in situ controlled the excavation pits and determine the extent of the poten- use of fire by Homo erectus pekinensis at Zhoukoudian Lo- tially burned areas, systematic sampling was employed in cality 1” (Zhong et al. 2014:341). three zones identified as possible burned areas or “hearths” as Burning experiments were conducted at the site using well as apparently unburned areas in the same level. A total of wood sourced from 12 arboreal species including hackberry 405 samples were collected. Magnetic susceptibility was mea- (Celtis sp.) and Chinese redbud (Cercis chinensis). These spe- sured with an Agico MFK1-FA Kappabridge magnetic suscep- cies were found to have been in the environment of H. erectus tibility meter at frequencies of 976 Hz (vLF) and 15,616 Hz pekinensis, either recovered as burned fragments or identified (vHF). Redness was measured with a Konica Minolta CM- through pollen analysis on sediments extracted from the site. 700d spectrophotometer. Following magnetic susceptibility Through elemental composition and material analysis of the and redness measurements, some samples were then selected fresh ash, the presence of silicate aggregates was determined for DRS and high-temperature magnetic susceptibility mea- and, more importantly, an enormous discrepancy was dis- surements (fig. 4). The results indicate that the magnetic sus- covered in the levels of constituent chemical elements (Si, Al, ceptibility and redness of sediments taken from the hypothet- S274 Current Anthropology Volume 58, Supplement 16, August 2017

x x Figure 4. Contour maps for LF (A), redness (B), and FD (C) measurements. The numbers outside the frames set the excavation coordinate according to which the pits were named. The thick dashed lines (A) show the burned areas. Black circles (B) show the x fi locations of DRS samples. Diamonds (C) show locations of the 2T samples (after Zhang et al. 2014). A color version of this gure is available online. ical “hearths” were markedly higher than those from other early Middle Pleistocene, have been questioned because of the areas in the same level (up to 22 times greater for magnetic lack of solid supporting evidence (Sandgathe and Berna 2017). susceptibility and three times more for redness), and fine- How can we effectively evaluate evidence of in situ anthropo- grained magnetite and hematite grains contributed signifi- genic burning, and what criteria should be used and accepted cantly to the distinctly high values of magnetic susceptibility by researchers to achieve consensus on this matter? Certainly, and redness in these “hearth” sediments. High-temperature thepresenceoffire-cracked rocks, burned bones and artifacts, magnetic susceptibility measurements demonstrated that the ash, charred botanical materials, burned sediments or ground thermally altered sediments were heated to above 7007C. Those surfaces, and residues of combustion are important. However, changes in low-frequency magnetic susceptibility and redness in many cases it is intact fire features (fireplaces or “hearths”), cannot result from natural fires (Zhang et al. 2014). the context of these individual occurrences, and their associa- tion with other human remains (e.g., human fossils and arti- Discussion facts) that are emphasized. Weiner et al. (2000:221) proposed the following criteria and degrees of confidence for evaluating In the past, many Paleolithic sites, including ZKD, have been in situ burning at archaeological sites (table 2). reported to have yielded evidence for the use and mainte- In theory, the criteria proposed by these scholars are ob- nance of fire by early hominins. However, many such local- jective, clear cut, and feasible. However, in practice, they are ities, especially those of the Lower Pleistocene and even the often very difficult to apply to many Paleolithic cases because

Table 2. Criteria for evaluating in situ burning at archaeological sites

Objects and associations Degree of confidence Constructed fireplaces and hearths (e.g., depression, stone lined) preferably Clear-cut evidence for intentional fire use by humans in association with burned bones, lithics, sediments, etc. Unambiguous associations, e.g., in situ presence of wood ash in a cave where Evidence for fire use by humans trees are not normally found (i.e., interior portions) Ambiguous associations, e.g., burned bones associated with lithics in Suggestive evidence for fire use by humans a stratigraphic unit or layer Presence of burned materials dispersed in a depositional context without Evidence for fire at site but not proof of direct or purposeful direct association with anthropogenic products human use Source. Weiner et al. 2000:221. Gao et al. Fire Evidence from Zhoukoudian S275 human and other animals’ activities during the formation of related with the Quartz II layer in the Gezitang (Chamber or the archaeological record as well as geological factors of post- Hall of the Pigeons), where two concentrations of ash were depositional processes could all alter the traces left by hu- found deposited conformably on top of a large limestone slab mans who made use of the site. Thus, ideal archaeological sites (Pei and Zhang 1985). It is possible that hearth features, or at in which human behavioral evidence remains completely intact least some subset of them, were disturbed or even destroyed will be very difficult to find. In general, the older the archae- by natural postdepositional processes such as hyena denning ological deposit, the worse the evidence is preserved. However, and water movement. Early excavators at ZKD may have en- we can still use the criteria proposed by Weiner et al. (2000) to countered undisturbed features of this type in other parts of evaluate the nature and authenticity of the evidence for the the horizon, leading them to report the existence of evidence use and maintenance of fire from ZKD Locality 1, especially for hominin use and maintenance of fire in this stratum. We Layer 4. can only hope that future excavators at ZKD will be fortunate The presence of burned materials dispersed throughout the enough to find such clear-cut evidence testifying to the early deposit—including fire-cracked rocks, burned animal bones, development of nascent pyrotechnologies at the site. and charred botanical remains—has been clearly documented According to the field notes and research reports of earlier at ZKD Locality 1, especially in Layers 4, 8–9, and 10. There- ZKD excavation projects, Layer 8–9, the so-called Lower Cul- fore, while the site retains evidence of fire, no unambiguous tural Horizon, yielded much more evidence of human fire use proof of direct or purposeful human maintenance of fire has than did Layer 10, including several layers of ash and burned emerged. Although these items were found along with lithic objects (Jia 1959; Pei and Zhang 1985). However, little recent artifacts and human fossils in the same stratigraphic units, effort has been made to investigate this important horizon. these associations are regarded as ambiguous by some research- Even in the investigations by Goldberg et al. (2001), little at- ers. Thus, only suggestive evidence for the use and mainte- tention was paid to these strata. They provided a relatively nance of fire by early hominids at the site has been preserved. simple description of Layer 8–9: “The broken, dislodged and The recent field investigations in Layer 4 revealed that anthro- rotated nature of the aggregates and the interstitial infillings pogenic fireplaces and hearths illustrated by stone-outlined of well sorted silt, suggest disturbance of the original deposits features, siliceous aggregates, and extremely high redness val- by water, including inwashing of silt” (Goldberg et al. 2001: ues and magnetic susceptibility in association with burned 497–498). Considering the fact that abundant archaeological bones, stone artifacts, and latosols are clearly present in the materials and traces related to human use of fire have been deposits, thus providing clear-cut evidence for intentional fire documented in Layer 8–9 and that three Homo erectus crania use by hominins at the site, at least during the formation of were unearthed from this unit in 1937 (Jia and Huang 1990), that horizon. the Lower Cultural Horizon deserves much more research But what of the other Locality 1 horizons, especially Layer 8– attention because future excavation and analysis might pro- 9 and Layer 10, so long claimed to be rich in burned materials duce more convincing evidence of in situ use and mainte- and other pyrogenic traces, even hearth remnants? Goldberg nance of fire by early hominids similar to that obtained from et al. (2001) paid particular attention to Layer 10 during their Layer 4. field investigations. They argued that there was no clear evi- The debate over hominin use and maintenance of fire at dence of anthropogenic fire use in the Lower Unit of Layer 10 ZKD provides an interesting basis for examining the ways in and concluded that a major part of this unit is the product of which the scientific community has dealt with key academic deposition in standing or slowly flowing water. With respect issues and changes through time: some might relate to the to the Upper Unit of Layer 10, they expressed uncertainty. application of new technologies to extract new evidence, and They pointed out that the Upper Unit deposits contained both some might be subjective assertions inspired by theoretical quartzite pebbles presumed to have been brought to the site paradigm shifts. Binford and Ho’s (1985) claims (that burned by humans and allochthonous rock fragments that appear to items evident in ZKD Locality 1 were probably brought into have a fluviatile origin; therefore, the possibility of human ac- the cave by flowing water, that the so-called ash deposits were tivity could not be ruled out (Goldberg et al. 2001). actually made up of bat or bird dung or simply chemical- It is clear from these scholars’ description that Layer 10 is altered humus or the residues of nonanthropogenic wildfires, thick and has a long and very complicated depositional his- and that ZKD could not be considered the real “cave home” tory and that the site might have been open to varying extents of Peking Man but was instead the lair of denning carnivores) throughout most of the period of its formation. Therefore, were not based on any primary research at the site or on many agencies, both human and natural, may have contrib- excavated materials. They did briefly visit the site later and uted to the accumulation of sediments in this horizon. Shen analyzed collections of bone (Binford and Stone 1986), but a et al.’s (2004) analysis of the distribution of EC identified half-day observation of the profile and 4 days of examination much higher concentrations in Layer 10, strongly suggesting of 1,523 pieces of animal bone could hardly produce precise that fireplaces had existed here during this earlier period. and unbiased data (see comments by Behrensmeyer, Haynes, Another piece of evidence in support of this conclusion is the and Olsen in Binford and Stone 1986). Their argument was observation that the lower horizon of Layer 10 may be cor- essentially a consequence of the shift in ideas about the ca- S276 Current Anthropology Volume 58, Supplement 16, August 2017 pabilities and inclinations of early hominins to acquire meat, of course for more than half a century until it began to be that is, from perceptions emphasizing systematic hunting to seriously questioned in the mid-1980s. While some of the situational scavenging. This factor is very obvious in their will- questions have been based on new research results, some are ingness to accept evidence of fire use from Layer 4 but refute it at least partially hypothetical and speculative. In the latter from Layer 10. They took this stance not because they de- case, previous evidence and analytical results have been tected significant differences in the burned materials from these largely ignored. Geochemical investigations conducted at the two horizons but only because Layer 4 was deposited much site in the 1990s provided new perspectives and deepened later, thus the inferred capability of hominins to hunt large studies of this issue. However, the results have certain weak- game and use fire for that younger stage was more acceptable. nesses due, in part, to limitations in sample collection and the In doing so, earlier, potentially equally compelling, evidence resulting data. The field investigations and laboratory anal- was either denied or simply neglected. yses, ongoing since 2009, have unveiled new and reliable evi- Binford’s verdict on the nature of the ZKD site and the fate dence supporting the use and maintenance of fire by hom- of its evidence for hominin fire use had a profound effect on inids at ZKD. Current evidence suggests that Layer 4 of the other researchers who seemed to be following the same path. site contains clear-cut evidence for in situ use of fire, includ- This includes Boaz et al. (2000), who suggested that Homo ing hearth features. Future excavations and research con- erectus pekinensis did not occupy Locality 1 but that their re- ducted at Layers 8–9 and 10 may reveal similar evidence to mains were transported into the cave by hyenas, and they sub- definitively resolve this continuing controversy. sequently proposed “a model of transient hominid scavenging aided by the use of fire at the large hyenid den” (Boaz et al. 2004:519–520). It also includes Weiner et al. (1998) noticing a Acknowledgments close association among stone artifacts and burned bones We thank Guo Zhengtang, John W. Olsen, Zhong Maohua, from Layers 10 and 4 and burned bone patterns similar to Zhang Yan, Guan Ying, Li Feng, and other colleagues for col- those observed in caves with unequivocal evidence of fire use laboration and assistance. Thanks also go to Dennis Sandgathe by later humans but still claiming that it was “only suggestive and Francesco Berna for organizing the “Fire and the Genus of a cultural association, and hence of the use of fire by hu- Homo” workshop and helping with the draft manuscript of mans, but does not prove it” (Weiner et al. 1998:253). this article and to two anonymous reviewers for comments and These data can certainly be interpreted in other ways. For suggestions. The research was supported by the Key Research example, Wu Xinzhi (1999) used the burned bone values ob- Program of the Chinese Academy of Sciences (grant KZZD- tained by Weiner et al. (2.5% of the microfaunal and 12% of EW-15) and the Basic Research Program of the National Nat- the macrofaunal assemblages were burned) to refute their ural Science Foundation of China (grant 41672024). conclusion that those bones were transported into the cave by water runoff, arguing that water transport would have yielded much smaller proportions of macrofaunal bones compared References Cited with microfaunal elements and that stone artifacts would not Berna, Francesco, Paul Goldberg, Liora Kolska Horwitz, James Brink, Sharon have been easily moved under such conditions. Holt, Marion Bamford, and Michael Chazan. 2012. Microstratigraphic evidence of in situ fire in the Acheulean strata of Wonderwerk Cave, One especially interesting observation was made by Goldberg Northern Cape Province, South Africa. Proceedings of the National Academy et al. (2001). They suggested that the Upper Unit of Layer 10 of Sciences of the USA 109(20):E1215–E1220, doi/10.1073/pnas.1117620109. could be partially related to human activity because of the pre- Binford, L. R., and C. K. Ho. 1985. Taphonomy at a distance: Zhoukoudian, “the cave home of Beijing man?” Current Anthropology 26:413–442. sence of stone tools and blackened bones, and they believed that Binford, L. R., and N. M. Stone. 1986. Zhoukoudian: a close look. Current human occupation of the cave would not have been contem- Anthropology 27:453–475. porary with hyenas. This conclusion provided a kind of inspi- Black, D. 1926. Tertiary man in Asia: the Chou Kou Tien discovery. Nature 118:733–734. ration in addressing the question of whether ZKD was, in fact, ———. 1931. Evidence of the use of fire by Sinanthropus. Bulletin for Geo- the “cave home” of H. erectus pekinensis or the den of hyenas or logical Society of China 11:107–108. cave bears because both early hominids and large carnivores Boaz, N. T., R. L. Ciochon, Q. Xu, and J. Liu. 2000. Large mammalian carnivores as a taphonomic factor in the bone accumulation at Zhoukou- made use of the cave during certain periods throughout the dian. Acta Anthropologica Sinica 19(suppl.):224–234. long history of its formation and evolution. What we describe ———. 2004. Mapping and taphonomic analysis of the Homo erectus loci at here is a long and complex occupational story; the appearance Locality 1 Zhoukoudian, China. Journal of Human Evolution 46:519–549. Carrancho, A., and J. J. Villalain. 2011. Different mechanisms of magnetiza- on the scene of one character should not exclude others; they tion recorded in experimental fires: archaeomagnetic implications. Earth were all protagonists on the stage during different acts. Planet Science Letter 312:176–187. Goldberg, P., S. Weiner, O. Bar-Yosef, Q. Q. Xu, and J. Y. Liu. 2001. Site formation processes at Zhoukoudian, China. Journal of Human Evolution 41:483–530. Goren-Inbar, Naama, Nira Alperson, Mordechai E. Kislev, Orit Simchoni, Conclusions Yoel Melamed, Adi Ben-Nun, and Ella Werker. 2004. Evidence of hominin control of fire at Gesher Benot Ya‘aqov, Israel. Science 304:725–727. The evidence for Homo erectus pekinensis use and mainte- Guo, S. L., S. S. Liu, S. F. Sun, F. Zhang, S. H. Shou, X. H. Hao, W. Mang, F. nance of fire at ZKD Locality 1 had been accepted as a matter Shang, and J. F. Lui. 1991. Fission track dating of the 4th layer of the Peking Gao et al. Fire Evidence from Zhoukoudian S277

Man Site. Acta Anthropologica Sinica 10(1):73–77. [In Chinese with English Man Site at Zhoukoudian and the possibility of their application in the abstract.] development of evidence for the use of fire by humans. Chinese Science Guo, S. L., S. H. Liu, S. F. Sun, F. Zhang, S. H. Zhou, X. H. Liao, R. Y. Hu, W. Bulletin 49:612–616. Meng, P. F. Zhang, and J. F. Lin. 1980. Age determination of Peking Man Shen, G. J., X. Gao, B. Gao, and D. E. Granger. 2009. Age of Zhoukou- by fission track dating. Chinese Science Bulletin 25:535–536. dian Homo erectus determined with 26Al/10Be burial dating. Nature 458:198– James, S. R. 1989. Hominid use of fire in the Lower and Middle Pleistocene: a 200. review of the evidence. Current Anthropology 30:1–26. Shen, G. J., T. L. Ku, H. Cheng, R. I. Edwards, Z. X. Yuan, and Q. Wang. 2001. Jia, L. P. 1959. Report on excavations at the Sinanthropus site in 1958. Pa- High-precision U-series dating of Locality 1 at Zhoukoudian, China. Jour- laeovertebrata et Paleoanthropoligica 1(1):21–26. [In Chinese.] nal of Human Evolution 41:679–688. Jia, L. P., and W. W. Huang. 1984. Report on the excavation of Zhoukoudian. Weiner,S.,O.Bar-Yosef,P.Goldberg,Q.Q.Xu,andJ.Y.Liu.2000.Evidence Tianjin: Tianjin Science and Technology Press. [In Chinese.] for the use of fire at Zhoukoudian. Acta Anthropologica Sinica 19(suppl.):218– ———. 1990. The story of Peking Man: from archaeology to mystery. Hong 223. Kong: Oxford University Press. Weiner, S., Q. Xu, P. Goldberg, J. Liu, and O. Bar-Yosef. 1998. Evidence for Jordanova, N., E. Petrovsky, and M. Kovacheva. 2001. Factors determining the use of fire at Zhoukoudian, China. Science 281:251–253. magnetic enhancement of burnt clay from archaeological sites. Journal of Wu, X. Z. 1999. Investigating the possible use of fire at Zhoukoudian, China. Archaeological Science 28:1137–1148. Science 283:299. Liu, D. S., S. S. Zhang, X. Z. Wu, W. W. Huang, S. L. Guo, Z. C. Liu, C. L. Zhou, Q. Wuethrich, B. 1998. Geological analysis damps ancient Chinese fires. Science H. Huang, Q. Q. Xu, and J. Y. Liu. 1998. Comments on the paper published in 281:165–166. Science (1998, 281: 251–253) for the evidence of fire use in Locality 1 of Zhang S. Q., X. Gao., F. Y. Chen, Y. Li., Y. Zhang., X. L. Zhang, and J. S. Li. Zhoukoudian. Acta Anthropologica Sinica 17:317–329. [In Chinese.] 2016. A report of the 2009–2010 field excavations at the West Section of Liu, S. S., F. Zhang, R. Y. Hu, J. F. Liu, S. L. Guo, S. H. Zhou, W. Meng, P. F. Zhoukoudian Loc. 1. Acta Anthropologica Sinica 35(1):63–75. [In Chinese Zhang, S. F. Sun, and X. H. Hao. 1985. The application of fission track in with English abstract.] dating Peking Man’s age. In Multidisciplinary study of the Peking Man Site Zhang, S. S. 1985. The early Paleolithic in China. In Paleoanthropology and at Zhoukoudian. R. K. Wu, M. E. Ren, X. M. Zhu, Y. G. Yang, C. G. Hu, Paleolithic archaeology in the People’s Republic of China. Wu Rukang and Z. C. Kong, Y. Y. Xie, and S. S. Zhao, eds. Pp. 241–245. Beijing: Science J. W. Olsen, eds. Pp. 147–186. Orlando, FL: Academic Press. Press. [In Chinese.] ———. 2004. Zhoukoudian site annals. Beijing: Beijing Press. [In Chinese.] Movius, H. L. 1948. Lower Paleolithic culture of southern and eastern Asia. Zhang, Y., Z. T. Guo, C. L. Deng, S. Q. Zhang, H. H. Wu, C. X. Zhang, J. Y. Transactions of the American Philosophical Society 38(4):329–420. Ge, et al. 2014. The use of fire at Zhoukoudian: evidence from magnetic Oakley, K. P. 1956. The earliest fire makers. Antiquity 30:102–107. susceptibility and color measurements. Chinese Science Bulletin 59(10): Pei, J. X. 1985. Thermoluminescence dating for cave deposits of Zhoukoudian 1013–1020. and other cave sites. In Multidisciplinary study of the Peking Man Site at Zhao, S. S., J. X. Pei, S. L. Guo, S. S. Liu, F. Qian, S. H. Qiu, and X. G. Li. Zhoukoudian. R. K. Wu, M. E. Ren, X. M. Zhu, Z. G. Yang, C. G. Hu, Z. C. 1985a. Study of chronology of Peking Man Site. In Multidisciplinary study Kong, Y. Y. Xie, and S. S. Zhao, eds. Pp. 256–260. Beijing: Science Press. [In of the Peking Man Site at Zhoukoudian. R. K. Wu, M. E. Ren, X. M. Zhu, Chinese.] Z. G. Yang, C. G. Hu, Z. C. Kong, Y. Y. Xie, and S. S. Zhao, eds. Pp. 239– Pei, W. C. 1934. A preliminary report on the late Palaeolithic cave of Chou- 240. Beijing: Science Press. [In Chinese.] koutien. Bulletin of the Geological Society of China 13:327–358. Zhao, S. S., M. Xia, C. H. Zhang, M. L. Liu, S. X. Wang, Q. F. Wu, and Z. B. Pei, W. Z., and S. S. Zhang. 1985. A study on the lithic artifacts of Sinan- Ma. 1985b. Uranium-series dating of the Peking Man Site. In Multidisci- thropus. Beijing: Science Press. [In Chinese with English abstract.] plinary study of the Peking Man Site at Zhoukoudian. R. K. Wu, M. E. Ren, Sandgathe, Dennis M., and Francesco Berna. 2017. Fire and the genus X. M. Zhu, Z. G. Yang, C. G. Hu, Z. C. Kong, Y. Y. Xie, and S. S. Zhao, eds. Homo: an introduction to supplement 16. Current Anthropology 58(suppl. Pp. 246–250. Beijing: Science Press. [In Chinese.] 16):S165–S174. Zhong, M. H., C. L. Shi, X. Gao, X. Z. Wu, F. Y. Chen, S. Q. Zhang, X. K. Shen. C. D., W. X. Yi, Y. Yang, Y. M. Sun, C. Z. Jin, D. S. Liu, and B. X. Cai. Zhang, and J. W. Olsen. 2014. On the possible use of fire by Homo erectus 2004. Concentrations of “elemental carbon” in samples from the Peking at Zhoukoudian, China. Chinese Science Bulletin 59(3):335–343. S278 Current Anthropology Volume 58, Supplement 16, August 2017

How Did Hominins Adapt to Ice Age Europe without Fire? by Harold L. Dibble, Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Shannon P. McPherron, and Dennis M. Sandgathe

Online enhancements: supplemental figures

Analyses of archaeological material recovered from several Middle Paleolithic sites in southwest France have pro- vided strong corroborating data on Neanderthal use of fire. Both direct and indirect data show that Neanderthals in this region were frequently and/or intensively using fire during warmer periods, but such evidence declines sig- nificantly in occupations that took place during colder periods. One possible explanation for this pattern is that it reflects the inability of Western European Neanderthals to make fire, simply because natural sources of fire occur much more frequently during warmer climatic periods. Regardless of the explanation, the long periods of diminished evidence of fire shows that, unlike modern humans, these hominins were not obligate fire users, and this fact in itself raises important questions of how they adapted, physiologically and/or technologically, to the generally harsh glacial conditions of the middle latitude of Europe and to reduced energy returns typical of raw food. As a corollary, it also raises questions regarding their need for and use of fire during the warmer periods.

Throughout history, pyrotechnology, a uniquely human tech- dates and faunal remains, these periods of little or no use of nological innovation, has been seen as a major factor in allow- fire reflected a time of increasingly cold conditions. Given an ing humans to adapt to a wide range of environmental circum- abundance of evidence for fire use earlier in the sequences of stances. But as noted recently by Shimelmitz et al. (2014:196), both sites during the warmer periods of MIS 5, the lack of “only when fire use became a regular part of human behavioral evidence during the colder periods was especially surprising. adaptations could its benefits be fully realized and its evolu- Earlier, we suggested (Sandgathe et al. 2011a) that these Eu- tionary consequences fully expressed.” Recently a series of ropean hominins lacked the ability to start fires, relying in- papers was published (Aldeias et al. 2012; Goldberg et al. 2012; stead on natural fires. This suggestion was based on the fact that Sandgathe et al. 2011a, 2011b) looking at the evidence for most wildfires are caused by lightning strikes and that light- combustion features at two sites in southwest France, Roc de ning occurs much more frequently during warmer and wetter Marsal and Pech de l’Azé IV. At both sites two patterns were periods and more rarely during cold, dry periods. Regardless of apparent. First, there were long periods of time when, in spite the reason for the long hiatus, however, if fire were not being of extensive evidence for ongoing occupation, evidence for the used during long periods of occupation at these sites, this raises use or presence of fire was lacking. Second, as indicated by the important questions regarding the role of fire in the overall

Harold L. Dibble is Professor in the Department of Anthropology of the University of Pennsylvania (3340 Walnut Street, Philadelphia, ology at Boston University (675 Commonwealth, Boston, Massachu- Pennsylvania 19104, USA [[email protected]]), in the Institute setts 02215, USA [[email protected]]), at the School of Earth and for Human Origins of Arizona State University (PO Box 874101, Environmental Sciences of the University of Wollongong (Australia), Tempe, Arizona 85287, USA), and in the Department of Human and at the Institute for Archaeological Sciences of the Eberhard Karls Evolution of the Max Planck Institute for Evolutionary Anthropol- Universitat Tübingen (Tübingen, Germany). Shannon P. McPherron ogy (Leipzig, Germany). Aylar Abodolahzadeh is a Graduate Student is a Researcher in the Department of Human Evolution of the Max in the Department of Anthropology of the University of Pennsylvania Planck Institute for Evolutionary Anthropology (Leipzig, Germany (3340 Walnut Street, Philadelphia, Pennsylvania 19104, USA [aylar [[email protected]]). Dennis M. Sandgathe is Lecturer in the [email protected]]). Vera Aldeias is a Researcher in the De- Department of Archaeology of the Simon Fraser University (8888 partment of Human Evolution of the Max Planck Institute for Evo- University Drive, Burnaby, British Columbia V5A 1S6, Canada) and at lutionary Anthropology (Leipzig, Germany [[email protected] the Museum of Archaeology and Anthropology at the University of .de]) and at the Interdisciplinary Center for Archaeology and Evolution Pennsylvania (3260 South Street, Philadelphia, Pennsylvania 19104, of Human Behavior at the Universidade do Algarve (Faro, Portugal). USA [[email protected]]). This paper was submitted 25 VII 16, accepted 5 III Paul Goldberg is Professor Emeritus in the Department of Archae- 17, and electronically published 17 VII 17. q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0011$10.00. DOI: 10.1086/692628 Dibble et al. Ice Age Europe without Fire S279 adaptation of these hominins to the colder conditions of MIS 4 intrusive. Artifact densities in Layers 9 through 2 are very high and MIS 3. Some of these questions will be addressed later in (e.g., n p 8,100 lithics for Layer 9). Layer 1 is a disturbed this paper. First, however, it is important to review the evidence Holocene deposit. from these two sites and others that have been studied subse- Layers 9 through 5 comprise a single lithostratigraphic unit quently, taking into account some results of recent laboratory with significant interbedded darker anthropogenic compo- experiments on the transfer of heat from surface fires to un- nents including major concentrations of ash, charcoal, and derlying sediments (Aldeias 2017; Aldeias et al. 2016). burned objects. The faunal data (fig. A5) indicate temperate conditions with an abundance of forest species, suggesting a late Evidence for Presence and Absence of Fire MIS 5a date (Marquet in Sandgathe et al. 2008). The overlying at Pech de l’Azé IV and Roc de Marsal layers, Layers 4 through 2, show increasingly colder and drier conditions that would correspond to MIS 3 (see Castel et al. Pech de l’Azé IV (Pech IV) and Roc de Marsal are two Mous- 2016; Guérin et al. 2012, 2016; Sandgathe et al. 2008). terian cave sites in the Dordogne region of southwest France As at Pech IV, direct evidence for fire (see Aldeias 2017) is (fig. A1; figs. A1–A7 available online) recently excavated by the found not throughout the Roc de Marsal sequence but rather same multidisciplinary team from 2000 to 2010 (Dibble et al. only in the earliest layers. This evidence includes discrete 2009; McPherron, Soressi, and Dibble 2001; Sandgathe et al. charcoal and ash units, burned or calcined bone, burned 2008; Turq et al. 2009, 2011). Both sites span the period of time lithics, and rubefied sediments (Aldeias et al. 2012; Goldberg from mid to late MIS 5 into MIS 3. At both sites there is clear et al. 2012). In Layers 7 and 9 localized examples of “stacked” evidence for fire use in the basal layers and very limited evidence hearths are clearly visible in section view (fig. A6), which in- of it in the upper layers. dicates that individual hearths were repeatedly constructed in Pech IV contains several major geological strata of Pleistocene- more or less the same location throughout the duration of each age sediments (fig. A2). The lowermost, Layer 8, lies directly on of these stratigraphic components. However, not all of the lower bedrock (see Dibble et al. 2009) and is dated by thermolumi- layers exhibit direct and intensive evidence for fire and in fact, nescence (TL) and optically stimulated luminescence (OSL) to such evidence alternates: Layers 5, 7, and 9 are rich in such approximately 95 5 5 kya, or MIS 5c (Gibbard and van features, while Layers 6 and 8 have little or no direct evidence for Kolfschoten 2005; Jacobs et al. 2016; Richter et al. 2012; them. Abundant hearths in the lower levels of the site were also Winograd et al. 1997). In keeping with these dates, the as- mentioned in the unpublished notes of the previous excavator, sociated fauna also suggests a warm, humid climatic regime J. Lafille. (Dibble et al. 2009; Laquay 1981; Niven 2013). Evidence for To summarize (and as shown in fig. A7), the Middle Pa- fire, including clear organic- and ash-rich units, burned bone, leolithic occupations of Pech de l’Azé IV and Roc de Marsal burned lithics, and rubefied (reddened) sediments is abun- overlap considerably in time, with initial occupations during dant in this layer. a temperate period followed by occupations that occurred In the overlying layers, there is little direct evidence of fire. during a marked deterioration in climate. At both sites un- Following the heavily disturbed Layer 7, Layer 6 is also as- mistakable hearth features occur in their lower layers and sociated with a faunal assemblage (fig. A3) that reflects a rel- indicate that fire was certainly used at this time, that is, during atively temperate wooded environment and dates to approx- the time of relatively warmer conditions. But it is equally clear imately 77 kya (Jacobs et al. 2016). Although in his earlier from both sites that fire use was not a constant element in the excavation of the site, Bordes (1975) did observe some limited occupations. At Roc de Marsal, for example, the lower layers traces of fire residues in this layer, our own excavations did not. seem to alternate between those with clear fire residues and Following another partially disturbed layer (5B), Layer 5A those without. At both sites, however, the upper Mousterian represents the beginning of a colder period, with a significant layers (Layers 4 through 2 at Roc de Marsal and Layers 5 and 4 increase in reindeer, and this trend to colder conditions con- at Pech IV) contain no identifiable fire features, including no tinues through Layer 4, which dates to around the beginning of concentrations of charcoal or ash and limited quantities of MIS 4. The final Pleistocene layer, Layer 3, is dated by a burned bone or flint. At best, very rare and very small frag- number of methods (see also McPherron et al. 2012) to around ments (!0.5 cm) of charcoal were noted, and these were pri- 50 kya, or MIS 3. Virtually no direct evidence of fire was seen in marily confined to Layer 3 of Pech IV (where there are also Layers 4 and 5, although burned bones were observed in thin slightly more heated flints). section and some dispersed charcoal occurred in Layer 3. Another Middle Paleolithic site excavated by the same How Good Are These Data? team is Roc de Marsal, which is located about 20 km west of Pech IV (Turq et al. 2009). Here excavations recognized 13 strat- The primary question to be addressed first is whether or not igraphic layers (fig. A4). At the base, Layers 13 through 10 rep- the absence of direct evidence for fire is indeed evidence of resent locally mobilized sediments from weathering of the lime- fire’s absence. With regard to taphonomic factors, strong argu- stone bedrock. Layer 10 contains some limited archaeological ments can be made to show that postdepositional processes materials (n p 129 lithics), but these may be mostly or entirely were not significant factors in removing the direct evidence. S280 Current Anthropology Volume 58, Supplement 16, August 2017

First of all, at both sites well-preserved fire residues occur both figure 1 below shows the results of two heat experiments at two inside the caves as well as beyond what would have been the drip different temperatures (9507C and 6007C) at the ground surface, lines at the time of occupation. Therefore the degree of overhead with additional temperature readings at depths of 2, 6, 10, and cover is not a factor. Furthermore, there is no evidence in the 20 cm. Even at the cooler surface temperature of 6007C, objects form of edge damage on the lithic artifacts, preferred orien- 2 cm below the surface will heat to 3007C within an hour, and at tations of objects, winnowing of smaller objects, or micromor- higher temperatures and with longer duration, temperatures as phological studies of the sediments to indicate significant post- high as 2007C will be reached at even 10 cm below the surface. depositional disturbances or erosion in the upper layers of either Thus, even relatively small and/or brief fires will affect under- site. Thus, no site formation processes have been identified that lying objects. could have removed the fire residues. Instead of simply noting the presence of burned flints in Nonetheless, even if direct evidence for fire may be have the two sequences, the number of proximal and complete been removed, there should be indirect evidence due to the pieces showing signs of burning is expressed as a percentage effects that heat has on lithics, bones, and even the sediment, of all proximal and complete pieces (in both cases counts are and all of these should remain (Aldeias 2017). When flint is for pieces larger than 2.5 cm; see fig. 2). At both sites the exposed to heat, color changes can be visible at temperatures percentage of burned lithics and fauna decreases through the starting at 2507C, a distinctive luster begins to develop at tem- sequence, reaching minima of 1%–2% in the upper layers. As peratures of approximately 3507C, and crazing can occur start- shown in Sandgathe et al. (2011a), the numbers are a func- ing at 3207C (Julig et al. 1999:838 and citations within; Rott- tion of neither varying sample size nor artifact density. Even länder 1983). These effects are not limited to objects that are if the direct evidence for the combustion features were re- actually in the fire. A series of actualistic experiments (Sievers moved, it is highly unlikely that the same processes would and Wadley 2008; Stiner et al. 1995; Werts and Jahren 2007), as remove the objects that were either in direct association with well as more highly controlled ones (Aldeias 2017; Aldeias et al. the fire or embedded in sediments directly below the fire. 2016), has suggested that the heat from fires can transfer to At both sites, then, the percentage of burned lithic objects, underlying sediments, though exactly how far down and the rate which is not subject to preservation issues, generally agrees of transfer are dependent on several variables. For example, with the direct evidence of fire residues in that the use of fire

Figure 1. Results of controlled experiments on heat transfer from ground surface down into the substrate (in this case sand). The experiment was run with the surface heat source set at 9507C and again at 6007C. Temperatures were continually recorded at the ground surface and at 2, 6, 10, and 20 cm beneath the heat source (see Aldeias et al. 2016). Dibble et al. Ice Age Europe without Fire S281

traces in the upper layers. Furthermore, analysis of Bordes’s entire collection, which represents a much larger area than our own, shows an identical pattern of decreasing percentages of burned lithics through the sequence (see fig. 2). Moreover, these two excavations were concentrated directly in the middle of the major part of the deposits, as determined by topographic relief. In summary, both Pech IV and Roc de Marsal have ex- cellent preservation, and the correlation is high between the presence or absence of direct evidence for fire (i.e., ash, char- coal or burned bone, rubefied sediments) and the indirect evidence of burned artifacts (see Aldeias 2017). This rela- tionship is not surprising given the causal nature of one to the other, but it means that the presence of fire can be detected even though various taphonomic processes may have oblit- erated the more direct evidence. Therefore, in the absence of both direct and indirect evidence, the conclusion that fire was not present at particular times during the occupation of these Figure 2. Frequencies of burned lithics (as a percentage of all platform-bearing lithics) in each layer at both Pech de l’Azé IV two sites is much stronger than it would be by relying on the and Roc de Marsal. direct evidence alone. It would seem unavoidable to conclude, therefore, that, while fire was being used frequently and/or drops off significantly, if not completely, during the later intensely during the earlier occupations, its use drops to near occupations. It can be noted too that the same trends are ap- zero in the upper occupations. While evidence for the use of parent among the fauna and small finds. Two notable ex- fire does not disappear entirely, on the basis of the very low ceptions are Layers 6 and 8 at Roc de Marsal, where the per- frequency of burned flints, this evidence represents an insig- centage of heated flints remains high even though there are no nificant aspect of these later occupations. visible hearths in these layers. Although it might be suggested that the elevated percentage of heated flints might represent How Can We Explain These Patterns? some vertical movement of lithics, there is currently no mi- cromorphological or field evidence for this. Likewise, generally What has been shown is that at these two sites fire was absent there are some clear time-related patterns in these layers (9–5) over significant periods of time, potentially over thousands of that are not expected if substantial mixing had occurred. The years of repeated occupation and use of the sites. Is it possible far more likely explanation is that heat from fires in overlying to argue that how these sites were used changed over time? layers moved down through the sediments (i.e., the fires in The answer to this question would appear to be no. For both Layers 5 and 7 modified some of the lithics of Layers 6 and 8, Pech de l’Azé IV and Roc de Marsal, while understanding the respectively; see Aldeias 2017; Aldeias et al. 2016). duration and intensity of individual occupation events is It is also unlikely that during the later occupations fires difficult at best, it is clear that these were both consistently were constructed at other lateral and as yet unexcavated lo- used as occupation sites. The significant concentrations of cations at the sites. At Roc de Marsal, the majority of the site stone tools (made and used in all components of both sites) has now been excavated (when combining our own excava- and heavily butchered faunal remains throughout the entire tions with the previous excavations of Lafille). The morphol- sequences (Castel et al. 2016; Hodgkins et al. 2016; Niven ogy of the cave in relation to the remaining sediments makes it 2013) are evidence of this. It would be very difficult to argue very unlikely that evidence of fire remains to be detected there that at times, the sites served more ephemeral purposes (e.g., by some future excavation. Our own excavations extended that they were kill sites or secondary butchery sites) where the along the entire length of deposits from well in front of the drip use of fire may have been less likely. In addition, the alternating line to the rear of the cave, as well as laterally across the width pattern in the basal layers of Roc de Marsal—with fires present of the cave. If there had been any other fire residues (including in Layers 5, 7, and 9 and absent in Layers 6 and 8—provides burned flints), they would have been detected. This issue is a another argument against fires being associated with only cer- bit more problematic at Pech de l’Azé IV because our own tain kinds of site use (see figs. 3, 4). Both the burned and un- excavations were concentrated on the western section of the burned layers show virtually identical technological and typo- excavated area, that is, the side that is closest to the original logical characteristics. (and now collapsed) entrance of the cave (Turq et al. 2011). Although faunal data are incomplete for Roc de Marsal, it However, observation of the eastern section remaining from is clear that while the prey species varies throughout both site Bordes’s (1975) earlier excavation clearly indicates the same sequences, the kinds of anatomical elements represented are level of burning in the basal deposits and a similar lack of such virtually indistinguishable in layers with fire versus those S282 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 3. Percent frequencies of technological types in burned layers (layers with significant evidence for use of fire) compared to unburned layers (layers with little or no evidence for use of fire) at both Pech de l’Azé IV and Roc de Marsal. without (see fig. 5; Castel et al. 2016; Hodgkins et al. 2016; it would be difficult to argue that these Neanderthals simply Dibble et al. 2017, pt. 3). Again, this suggests that the pres- changed their tastes in food preparation for millennia. ence or absence of fire does not appear to reflect changes in Since one of the primary uses of fire ethnographically is for site use. Likewise, while there is no evidence either for or warmth, it is especially ironic that in these two sites the use of against the use of fire for cooking, the arguments of Wrang- fire drops off significantly during colder periods. This pattern ham (2009) concerning the benefits that could be derived from is repeated at several sites in the region where we have com- cooking suggest that if fire were available, it would tend to be parable data. At Combe-Capelle Bas, which was excavated and used for this purpose under all circumstances. In other words, analyzed using identical techniques, there was also no direct

Figure 4. Percent frequencies of major tool types in burned layers (layers with significant evidence for use of fire) compared to unburned layers (layers with little or no evidence for use of fire) at both Pech de l’Azé IV and Roc de Marsal. Dibble et al. Ice Age Europe without Fire S283

Figure 5. Relative frequencies of different anatomical components (skull portions, forelimb portions, and hind limb portions) compared to frequencies of burned lithics in each layer at both Pech de l’Azé IV (Pech IV; Niven 2013; L. Niven, personal com- munication) and Roc de Marsal (Hodgkins et al. 2016; J. Hodgkins, personal communication). evidence of fire residues, and the percentage of burned flints Laville, and Paquereau 1966). As can be seen in table 1, there is remained at !2% for the entire sequence (Dibble and Lenoir a general tendency, with some exceptions, for frequencies of 1995). This site, which contains Quina Mousterian assem- burned flints to increase in those layers that were identified, blages, has been dated to MIS 3 (Valladas et al. 2003) and is through associated fauna (see Chase 1986) and sedimentology thus contemporary with the nonfire occupations from Pech de (Guadelli and Laville 1988), as being associated with warmer l’Azé IV and Roc de Marsal. Similarly, the Quina layers at and wetter climatic periods. These data provide some of the Jonzac and La Quina (Bierwirth 1996; McPherron et al. 2006; comparable quantitative data called for by Sorensen (2017) but Niven et al. 2012) are associated with very high percentages of contradict his conclusions for this important site based on the reindeer, and the percentage of heated flints is less than 0.5%. number of observations (e.g., presence of purported fires in At Jonzac, the Quina layers have been TL dated to MIS 4 particular layers, heated bedrock, burned bones taken for AMS (Richter et al. 2013). In the overlying layers at these two sites, dating, TL samples taken for dating, etc.). More work is needed heated flints are never more than 4% of the assemblage, and to reconcile these differences, but it should be emphasized that they generally are much less. A similar pattern is noted at La comparable proxies on relative scales are preferable to presence Ferrassie, which exhibits Levallois industries similar to those at or absence data and that techniques such as micromorphology Roc de Marsal, little evidence of fire, and again occupations and FTIR (Fourier transform infrared spectroscopy) are pref- occurring during cold periods (Guérin 2015). erable for identifying fire features. This said, we also acknowl- One of the best sites in the region for testing a correlation edge that curation issues with the Combe Grenal collection of fire and climate is Combe Grenal, where some 64 layers may impact proxies based on counting heated flints (Dibble likely represent deposits from late MIS 6 through MIS 3. Data et al. 2009). from Bordes’s collection (the site is currently being reexcavated) We cannot exclude the possibility that all of these nonfire also indicates a correlation between climatic regime and fre- occupations were limited to summer months and that what quency of fire use. This is based on percentages of burned flints we are seeing at these sites is a switch in seasonality wherein recorded for 23 of the 64 layers identified by Bordes (Bordes, southwest France was only occupied during warm months. It S284 Current Anthropology Volume 58, Supplement 16, August 2017

Table 1. Climate and frequency of burned flints in sampled Obviously, arguments based on negative evidence are fraught layers of the Middle Paleolithic site Combe Grenal with difficulties, and admittedly, there are many potential ex- in southwest France planations—both behavioral and natural—for an absence of evidence for fire in these sites. The hypothesis that this absence Layer Climate % lithics burned N lithics is due to an inability to start fires will require quantitative data 1–4 Very cold, very dry No data No data from many other sites that are adequately dated and for which 5 Very cold, very dry .0 19 both indirect and direct evidence for fire is noted (Aldeias et al. 6 Very cold, very dry .0 170 2012; Goldberg et al. 2012; Sandgathe et al. 2011a). This is only a 7–8 Less cold, humid .0 430 9 Cold and dry 1.9 160 hypothesis advanced to explain the cyclical pattern apparent in fi fi 10 Cold and dry No data No data the data. The more important nding is that re is not as 11–13 Less cold, humid No data No data ubiquitous as it should be, given that virtually all recent hunter- 14 Very cold, very dry 1.9 1,227 gatherer populations, even prior to access to matches and – 15 16 Very cold, very dry No data No data lighters, used fire daily for a wide range of applications and at 17 Less cold, humid 4.5 1,099 18–19 Less cold, humid No data No data almost every location where people spend any time at all. Thus, 20 Mild, humid 21.7 1,395 the main conclusion to be drawn from the evidence at hand is 21–22 Mild, humid No data No data that use of fire by Western European Neanderthals is sporadic 23 Very cold, very dry 1.2 1,490 at best, and for any particular site—even those with relatively – 24 25 Very cold, very dry No data No data dense or intensive occupations—fire may be virtually absent 26–35 Cold and humid No data No data 36–37 Cold, very dry No data No data for long periods of time. What we are suggesting, therefore, is 38 More mild, humid 6.9 1,643 that during the European Late Pleistocene there is simply not 39 More mild, humid No data No data enough evidence of fire to assume that Neanderthals used fire, 40 Cold and very dry .8 491 especially for warmth and for cooking, to a degree equivalent 41–43 Temperate and humid .3 725 – to that seen among extant humans. At a minimum, this may 44 49 Cold and dry No data No data fi 50a Temperate and humid 11.8 678 mean that re had a very different role in Neanderthal adap- 50 Temperate and humid 14.7 2,878 tations in Western Europe than at, say, Kebara, Tabun (Schiegl 51 Temperate and humid 10.6 1,710 et al. 1996; Shimelmitz et al. 2014), or Qesem Cave in Israel 52 Temperate and humid 15.5 1,255 (Barkai et al. 2017) and that there may have been extended 53 Cold and humid 29.0 93 periods of time when fire was not used to a great extent. Thus, 54–55 Cold and humid No data No data 56–57 Cold and dry No data No data the available data we do have for this region of France indicates fi 58 Less cold, humid .0 402 that Neanderthals were not obligate re users (contra Sorensen 59 Cold and very dry No data No data 2017), and such evidence might constitute a signifi cantly dif- 60–61 Cold and very dry 1.2 1,294 ferent behavioral adaptation than those seen in more recent 62 Less cold, humid .0 83 modern human occupations (see Henry 2017). 63–64 Cold and dry No data No data Note. Climate descriptions from Guadelli and Laville (1988) and Chase (1986). Implications for Considering the Role of Fire in Western European Neanderthal Adaptations The primary focus of this paper is to emphasize the quality of would be difficult, however, to imagine that these sites were the data presented to show that fire was not present for long consistently used only in one season for the tens of thousands periods of time in this particular region. We are not arguing of years represented by those layers that have little evidence that Neanderthals, as a species, were fire incapable, always ate for fire. That said, even during the warm months, conditions raw food, and never used fire for heat. Rather we are simply were still far cooler than today’s and, again, modern hunter- presenting quantitative evidence (rather than simple presence gatherers in even substantially warmer environments still rely or absence) that at times and for long periods they did not on fire to warm themselves, especially at night. make extensive use of fire. Admittedly, this is based on only a For all of these reasons, it was suggested by Sandgathe et al. few Middle Paleolithic sites, and given the lack of comparable (2011a) that the presence or absence of fire was due to the in- data available in the literature (see Sandgathe 2017; Sand- ability to start fires and, therefore that fire would be more likely gathe et al. 2011b; Roebroeks and Villa 2011), it is still un- to be present when it was naturally available. If the primary known to what extent this pattern holds true for many other natural source of fires is lightning strikes, then the apparent times and places during the Pleistocene. But the evidence avail- pattern seen in these two sites—fires present during warm able from a number of sites is strong enough to make us question periods, absent during cold periods—would reflect the fact our assumption that once we find evidence of fire in the ar- that lightning strikes occur more frequently in warmer and chaeological record, it automatically means that it was an es- more humid climates and much less frequently in colder and sential part of human behavior and adaptation from that point drier ones. forward (cf. Barkai et al. 2017) and throughout the world. Dibble et al. Ice Age Europe without Fire S285

If this assumption is not valid, and the use of fire was much quent moves (Macdonald, Roebroeks, and Verpoorte 2009; more limited than previously believed, this does raise im- Verpoorte 2006). portant questions regarding European hominin behavior and Moreover, given that cooking raises the nutritional and adaptation in the Pleistocene. One of the more important energetic value of food (e.g., Carmody and Wrangham 2009; questions concerns the ability of hominins to survive during Wrangham 2009), then an inability to cook their food for the colder periods. Among modern human groups, it would extended periods would further increase the amount of cal- be impossible for foragers to inhabit more northerly latitudes ories required by Neanderthals to meet their daily energetic without fire unless they had very well-developed clothing and needs. Lower overall energy requirements could have given shelter technology. However, anatomically modern humans anatomically modern humans competitive advantages over are relative newcomers to higher latitudes, unlike Neanderthals Neanderthals in terms of reproductive success and demo- who, along with their direct ancestors, have a potential time graphic expansion (Froehle and Churchill 2009). There is depth in Europe of several hundred thousand years. While it is considerable evidence that Middle Paleolithic populations possible that Neanderthal populations migrated to some extent relied particularly heavily on sources of fat in their diet. This in response to major climatic changes and did not always in- is especially apparent in the breakage of long bones, pre- habit the most northerly European latitudes during colder sumably for the extraction of marrow, and the intense, in- periods (e.g., Roebroeks 2006; Steegmann, Cerny, and Holliday tentional fragmentation of cancellous bone, presumably to 2002), the presence of significant numbers of occupations in extract bone grease (Castel et al. 2016; Hodgkins et al. 2016). Europe during full glacial conditions indicates that Neander- In the face of increased caloric requirements for Neander- thals were adapted to such conditions. thals, and especially increased fat requirements, cooking meat The question is to what extent was theirs a physiological may have taken on an especially important role under certain adaptation versus a cultural or technological one? While we conditions. do see the advent of bone needles during the Gravettian of Besides increasing net caloric returns, cooking can also late MIS 3 and early MIS 2, evidence for tailored clothing reduce the danger presented by bacterial and parasitic loads prior to this is virtually nonexistent (Collard et al. 2016). It in meat. As the evidence from sites like Roc de Marsal and has been argued that simply draping or wrapping untailored Pech IV clearly indicates that meat was not always cooked, it animal hides around a person has very limited thermal ef- seems likely that consumption of raw meat was relatively fectiveness (e.g., Gilligan 2007; Wales 2012). Although there common during the Lower Paleolithic and during at least is ongoing discussion about how much of the difference in certain periods in the Middle Paleolithic. This might not morphology between Neanderthals and their African contem- present any major problems during colder climatic periods poraries is due to active selective pressures and how much is when meat acquired by hominins would tend to remain un- mainly due to random genetic drift (e.g., Weaver, Roseman, and spoiled longer and, during some cold periods in some regions, Stringer 2007), it has long been accepted that Neanderthals have was likely more easily acquired and in greater quantities due to significant cold-adapted features such as their short, squat, heavy the prevalence of large numbers of herd animals like reindeer, bodies with shorter, stockier limbs (e.g., Holliday 1997; Ruff et al. horse, and bison. During warmer climatic periods when the 1993; Steegmann, Cerny, and Holliday 2002; Trinkaus 1981). fauna was dominated by mainly smaller and less gregarious Assuming that an absence of fire means an absence of woodland species (like red deer, roe deer, and wild pigs), meat cooking, there are also implications for Neanderthal energet- was likely generally less available (Hodgkins et al. 2016). Meat ics. This is an area where Neanderthals differed significantly would also tend to spoil faster during warmer periods, and it is from anatomically modern humans and one with behavioral possible that simple meat preservation techniques, like smok- consequences that may have played a role in the replacement ing, could have been practiced. However, there is no evidence of the former by the latter. It is argued that due to their larger for such practices, and it may be that, if meat spoilage was a body mass and unique shape, Neanderthals would have had a consideration for Neanderthals, they were simply cooking it to higher basal metabolic rate than anatomically modern humans make it safe for consumption. Along with increased availability and therefore a proportionally larger total energy expenditure of natural fire, such a pattern of behavior would also help (e.g., Aiello and Wheeler 2003; Sorensen and Leonard 2001; explain the apparent association between increased fire use and Steegmann, Cerny, and Holliday 2002). Conservative estimates warmer climatic periods. suggest a 10% difference between Neanderthals and middle Upper Paleolithic humans (Churchill and Rhodes 2009; Froehle Conclusions and Churchill 2009; Macdonald, Roebroeks, and Verpoorte 2009). This estimate is based on the premise that Neanderthals Much of the current evidence suggests that as recently as the and anatomically modern humans derived the same caloric latter half of the Late Pleistocene, Neanderthals were not using benefits from the food consumed, and it means that if Nean- fire all the time, especially not during cold climatic periods. derthals and early anatomically modern humans had the same This raises important questions about how these hominins diet composition, Neanderthals would have been obliged to were able to adapt to high-latitude glacial conditions both in consume more. This in turn may have necessitated more fre- terms of their physiology and their technology. S286 Current Anthropology Volume 58, Supplement 16, August 2017

Roebroeks and Villa (2011) make an important point that Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a fi reason: barbecue at Middle Pleistocene Qesem Cave, Israel. Current An- there is no evidence supporting the idea that re was requisite thropology 58(suppl. 16):S314–S328. to the initial colonization of Europe. Although they conclude Bierwirth, S. L. 1996. Lithic analysis in southwestern France: Middle Paleolithic that the use of fire became “habitual” during the European assemblages from the site of La Quina, vol. 633. Oxford: British Archaeo- logical Reports. Middle Paleolithic, the data from Roc de Marsal and Pech IV Bordes, F. 1975. Le gisement de Pech de l’Azé IV: note préliminaire. Bulletin show that hominins occupied or used these sites for long periods de la Société Préhistorique Française 72:293–308. of time without using fire. Whether or not these results are Bordes, F., H. Laville, and M. M. Paquereau. 1966. Observations sur le Pléis- tocène supérieur du gisement de Combe-Grenal (Dordogne). Actes de la applicable to the entire European Middle Paleolithic is currently Société Linnéenne de Bordeaux 103(10):3–19. unknown, since quantified data (rather than simply presence or Carmody, R. N., and Richard W. Wrangham. 2009. The energetic significance absence), both direct and indirect, on the degree of fire use is of cooking. Journal of Human Evolution 57(4):379–391. Castel, J.-C., E. Discamps, M.-C. Soulier, Dennis M. Sandgathe, Harold L. largely unreported (see Sandgathe 2017). The same problem Dibble, Shannon P. McPherron, Paul Goldberg, and Alain Turq. 2016. Ne- applies to the Upper Paleolithic as well. For a long time, it has andertal subsistence strategies during the Quina Mousterian at Roc de been assumed that once the advantages of fire were discovered, Marsal (France). Quaternary International 433(B):140–156, http://doi.org /10.1016/j.quaint.2015.12.033 its use would spread quickly. But as the evidence is examined Chase, Philip G. 1986. Relationships between Mousterian lithic and faunal more closely, and as we try to evaluate the degree of fire use, it is assemblages at Combe Grenal. Current Anthropology 27(1):69–71. becoming fair to say that we simply do not know when fire use Churchill, Steven E., and Jill A. Rhodes. 2009. The evolution of the human capacity for “killing at a distance”: the human fossil evidence for the evo- became a regular, integral, and, eventually, necessary compo- lution of projectile weaponry. In The evolution of hominin diets. J. J. Hublin nent of hominin adaptations. and M. P. Richards, eds. Pp. 201–210. Vertebrate Paleobiology and Pale- oanthropology. Dordrecht: Springer. Collard, Mark, Lia Tarle, Dennis M. Sandgathe, and Alexander Allan. 2016. Faunal evidence for a difference in thermal effectiveness of early modern Acknowledgments human and Neanderthal clothing. Journal of Anthropological Archaeology 44:235–246. This paper is based on research funded by the National Science Dibble, Harold L., Francesco Berna, Paul Goldberg, Shannon P. McPherron, Foundation, the Leakey Foundation, the Conseil General de la Susan M. Mentzer, Laura Niven, Daniel Richter, I. Théry-Parisot, Dennis M. Sandgathe, and Alain Turq. 2009. A preliminary report on Pech de l’Azé Dordogne, the Max Planck Society, the Service Régional de – ’ I V, Layer 8 (Middle Paleolithic, France). PaleoAnthropology 2009:182 219. l Archéologie (Agquitaine), and the University of Pennsylva- Dibble, Harold L., and Michel Lenoir, eds. 1995. The Middle Paleolithic site of nia Research Foundation and University of Pennsylvania Mu- Combe-Capelle Bas (France). University Museum Monograph 91. Phila- seum. It was also made possible by the considerable support of delphia: University of Pennsylvania Press. Dibble, Harold L., Shannon P. McPherron, Paul Goldberg, and Dennis M. the National Museum of Prehistory at Les Eyzies, France, in Sandgathe, eds. 2017. Recent excavations at the Middle Palaeolithic site of particular the director, Cleyet-Merle, and curator, Stephen Mad- Pech de l’Azé IV. Cave and Karst Systems of the World. Cham, Switzerland: eleine, for helping us access collections held there. Thanks also Springer. Dibble, Harold L., Shannon P. McPherron, Dennis M. Sandgathe, Paul Gold- to our colleague Alain Turq for all of his help and support and berg, Alain Turq, and Michel Lenoir. 2009. Context, curation, and bias: an to Virginie Sinet-Mathiot for running our field lab. We would evaluation of the Middle Paleolithic collections of Combe-Grenal (France). – like to thank the Wenner-Gren Foundation for providing the Journal of Archaeological Science 36(11):2540 2550. Froehle, A., and S. E. Churchill. 2009. Energetic competition between Nean- funding for the symposium that led to this volume and give a dertals and anatomically modern humans. PaleoAnthropology 2009:96–116. very special thank you to Leslie Aiello, president of the Wenner- Gibbard, P., and T. van Kolfschoten. 2005. The Pleistocene and Holocene Gren Foundation, and Laurie Obbink, conference program as- Epochs. In A geologic time scale, 2004. F. Gradstein, J. Ogg, and A. Smith, eds. Pp. 441–452. Cambridge: Cambridge University Press. sociate, for organizing and hosting the symposium so effectively Gilligan, I. 2007. Neanderthal extinction and modern human behaviour: the and so graciously. We also acknowledge the constructive re- role of climate change and clothing. World Archaeology 39:499–514. views of two anonymous referees. Goldberg, Paul, Harold L. Dibble, Francesco Berna, Dennis M. Sandgathe, Shannon P. McPherron, and Alain Turq. 2012. New evidence on Neandertal use of fire: examples from Roc de Marsal and Pech de l’Azé IV. Quaternary International 247:325–340. References Cited Guadelli, J. L., and H. Laville. 1988. L’environnement climatique de la findu Moustérien à Combe-Grenal et à Camiac. In Paléolithique moyen récent et Aiello, Leslie C., and P. Wheeler. 2003. Neanderthal thermoregulation and the Paléolithique supérieur ancien en Europe. Ruptures et transitions: examen glacial climate. In Neanderthals and modern humans in the European critique des documents archéologiques, vol. 3. Mémoires du Musée de Pré- landscape of the last glaciation: archaeological results of the Stage 3 project. histoire d’Ile-de-France. APRAIF, éd. Pp. 43–48. Nemours: Presses du CNRS. T. H. van Andel and W. Davies, eds. Pp. 147–166. Cambridge: McDonald Guérin, Guillaume, Emmanuel Discamps, Christelle Lahaye, Norbert Mercier, Institute for Archaeological Research. Pierre Guibert, Alain Turq, Harold L. Dibble, Shannon P. McPherron, Dennis Aldeias, Vera. 2017. Experimental approaches to archaeological fire features M. Sandgathe, and Paul Goldberg. 2012. Multi-method (TL and OSL), multi- and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– material (quartz and flint) dating of the Mousterian site of Roc de Marsal S205. (Dordogne, France): correlating Neanderthal occupations with the climatic Aldeias, Vera, Harold L. Dibble, Dennis M. Sandgathe, Paul Goldberg, and variability of MIS 5–3. Journal of Archaeological Science 39(10):3071–3084. Shannon P. McPherron. 2016. How heat alters underlying deposits and im- Guérin, Guillaume, Marine Frouin, Sahra Talamo, Vera Aldeias, Laurent Bru- plications for archaeological fire features: a controlled experiment. Journal xelles, Laurent Chiotti, Harold L. Dibble, et al. 2015. A multi-method lu- of Archaeological Science 67:64–79. minescence dating of the Palaeolithic sequence of La Ferrassie based on new Aldeias, Vera, Paul Goldberg, Dennis M. Sandgathe, Francesco Berna, Harold excavations adjacent to the La Ferrassie 1 and 2 Skeletons. Journal of Ar- L. Dibble, Shannon P. McPherron, and R. Zeljko. 2012. Evidence for Ne- chaeological Science 58:147–166, doi:10.1016/j.jas.2015.01.019. andertal use of fire at Roc de Marsal (France). Journal of Archaeological Guérin, Guillaume, Marine Frouin, Joséphine Tuquoi, Kristina J. Thomsen, Science 39(7):2414–2423. Paul Goldberg, Vera Aldeias, Christelle Lahaye, et al. 2016. The comple- Dibble et al. Ice Age Europe without Fire S287

mentarity of luminescence dating methods illustrated on the Mousterian Se- Sandgathe, Dennis M. 2017. Identifying and describing pattern and process in quence of the Roc de Marsal: a series of reindeer-dominated, Quina Mousterian the evolution of hominin use of fire. Current Anthropology 58(suppl. 58): layers dated to MIS 3. Quaternary International 30.1:e14. doi:10.1016/j.quaint S360–S370. .2016.02.063. Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. McPherron, Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current Alain Turq, and Steve Schwortz. 2008. Roc de Marsal (Campagne-de-Bugue, Anthropology 58(suppl. 16):S329–S336. Dordoge) rapport d’operation pour l’année 2008.Reportonfile with the Ser- Hodgkins, Jamie, C. Marean, Alain Turq, Dennis M. Sandgathe, Shannon P. vice Régional de l’Archéologie et Conceil Général de la Dordogne, Dordogne, McPherron, and Harold L. Dibble. 2016. Climate-mediated shifts in Ne- France. andertal subsistence behaviors at Pech de l’Azé IV and Roc de Marsal Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. Mc- (Dordogne Valley, France). Journal of Human Evolution 96:1–18. Pherron, Alain Turq, Laura Niven, and Jamie Hodgkins. 2011a. On the role Holliday, T. W. 1997. Postcranial evidence of cold adaptation in European of fire in Neandertal adaptations in Western Europe: evidence from Pech Neandertals. American Journal of Physical Anthropology 104(2):245–258. de l’Azé IV and Roc de Marsal, France. PaleoAnthropology 2011:216–242. Jacobs, Z., N. Jankowski, Harold L. Dibble, Paul Goldberg, Shannon P. Mc- Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. Mc- Pherron, Dennis M. Sandgathe, and M. Soressi. 2016. The age of three Pherron, Alain Turq, Laura Niven, and Jamie Hodgkins. 2011b. Timing of Middle Palaeolithic sites: single-grain optically stimulated luminescence (OSL) the appearance of habitual fire use. Proceedings of the National Academy of chronologies for Pech de l’Azé I, II, and IV (France). Journal of Human Evo- Science 108(29):E298. lution 95:80–103. Schiegl, S., Paul Goldberg, Ofer Bar-Yosef, and S. Weiner. 1996. Ash deposits in Julig, P. J., D. G. F. Long, B. Schroeder, W. J. Rink, D. Richter, and H. P. Hayonim and Kebara caves, Israel: macroscopic, microscopic and mineral- Schwarcz. 1999. Geoarchaeology and new research at Jerf al-Ajla Cave, Syria. ogical observations, and their archaeological implications. Journal of Ar- Geoarchaeology 14:821–848. chaeological Science 23(5):763–781. Laquay, G. 1981. Recherches sur les faunes du Würm I en Périgord. Thèse de Shimelmitz, R., S. L. Kuhn, A. J. Jelinek, A. Ronen, A. E. Clark, and M. 3e cycle, Université de Bordeaux 1. Weinstein-Evron. 2014. “Fire at will”: the emergence of habitual fire use Macdonald, K., W. Roebroeks, and A. Verpoorte. 2009. An energetics per- 350,000 years ago. Journal of Human Evolution 77:196–203. spective on the Neandertal record. In The evolution of hominin diets. J.-J. Sievers, C., and L. Wadley. 2008. Going underground: experimental carbon- Hublin and M. P. Richards, eds. Pp. 211–220. Leipzig: Springer. ization of fruiting structures under hearths. Journal of Archaeological Sci- McPherron, Shannon P., J. J. Hublin, J. Jaubert, M. Soressi, S. Bailey, J.-G. ence 35(11):2909–2917. Bordes, E. Claud, et al. 2006. New excavations at the Paleolithic site of Sorensen, Andrew C. 2017. On the relationship between climate and Nean- Chez-Pinaud Jonzac (Charente-Maritime, France). Pp. 24–26. In Abstracts dertal fire use during the Last Glacial in south-west France. Quaternary for the Paleoanthropology Society Meeting, San Juan, Puerto Rico. International 436:114–128. McPherron, Shannon P., M. Soressi, and Harold L. Dibble. 2001. Deux nouveaux Sorensen, M. V., and W. R. Leonard. 2001. Neandertal energetics and foraging projets de recherché à Pech de l’Azé (Dordogne, France). Préhistoire du Sud- efficiency. Journal of Human Evolution 40(6):83–495. Ouest 8:11–30. Steegmann, A. J., F. Cerny, and T. Holliday. 2002. Neandertal cold adaptation: McPherron, Shannon P., Sahra Talamo, Paul Goldberg, Laura Niven, Dennis physiological and energetic factors. American Journal of Human Biology Sandgathe, Michael Richards, Daniel Richter, Alain Turq, and Harold L. 14:566–583. Dibble. 2012. Radiocarbon dates for the Late Middle Paleolithic at Pech de Stiner, M. C., S. L. Kuhn, S. Weiner, and Ofer Bar-Yosef. 1995. Differential l’Azé IV, France. Journal of Archaeological Science 39(2012):3436–3442. burning, recrystallization, and fragmentation of archaeological bone. Journal Niven, Laura. 2013. A diachronic evaluation of Neanderthal cervid exploita- of Archaeological Science 22:223–237. tion and site use at Pech de l’Azé IV, France. In Zooarchaeology and modern Trinkaus, E. 1981. Neanderthal limb proportions and cold adaptation. In As- human origins: human hunting behavior during the Later Pleistocene, ver- pects of human evolution. C. B. Stringer, ed. Pp. 187–219. London: Taylor & tebrate paleobiology and paleoanthropology. Jamie L. Clark and John D. Francis. Speth, eds. Pp. 151–161. Dordrecht: Springer Science and Business Media, Turq, Alain, Harold L. Dibble, Paul Goldberg, Shannon P. McPherron, and doi:10.1007/978–94–007–6766–9_9. Dennis M. Sandgathe. 2009. Roc de Marsal (Campagne-de-Bugue, Dordoge) Niven, Laura, T. E. Steele, W. Rendu, J. B. Mallye, Shannon P. McPherron, rapport d’operation pour l’année 2009. Report on file with the Service Ré- Marie Soressi, Jacques Jaubert, and Jean-Jacques Hublin. 2012. Neandertal gional de l’Archéologie et Conceil Général de la Dordogne, Dordogne, France. mobility and large-game hunting: the exploitation of reindeer during the Turq, Alain, Shannon P. McPherron, Harold L. Dibble, Paul Goldberg, Dennis Quina Mousterian at Chez-Pinaud Jonzac (Charente-Maritime, France). M. Sandgathe, H. Jones, K. Maddison, et al. 2011. Les fouilles recentes du Journal of Human Evolution 63(4):624–635. Pech de l’Azé IV (Dordogne). Gallia Préhistoire 53:1–58. Roebroeks, Wil. 2006. The human colonisation of Europe: where are we? Valladas, H., N. Mercier, J.-L. Joron, Shannon P. McPherron, Harold L. Dibble, Journal of Quaternary Science 21(5):425–435. and M. Lenoir. 2003. TL dates for the Middle Paleolithic site of Combe- Roebroeks, Wil, and Paola Villa. 2011. On the earliest evidence for habitual Capelle Bas, France. Journal of Archaeological Science 30:1443–1450. use of fire in Europe. Proceedings of the National Academy of Science 108 Verpoorte, Alexander. 2006. Neanderthal energetics and spatial behaviour. (13):5210–5014. Before Farming 2006(3):1–6. Richter, Daniel, Harold L. Dibble, Paul Goldberg, Shannon P. McPherron, Wales, N. 2012. Modeling Neanderthal clothing using ethnographic analogues. Laura Niven, Dennis M. Sandgathe, S. Talamo, and Alain Turq. 2012. The Journal of Human Evolution 63:781–795. Late Middle Palaeolithic in southwest France: new TL dates for the sequence Weaver, T. D., C. C. Roseman, and C. B. Stringer. 2007. Were Neandertal and of Pech de l’Azé IV (France). Quaternary International 294:160–167. modern human cranial differences produced by natural selection or genetic Richter, Daniel, Jean-Jacques Hublin, Jacques Jaubert, Shannon P. McPherron, drift? Journal of Human Evolution 53(2):135–145. Marie Soressi, and Jean-Pierre Texier. 2013. Thermoluminescence dates for Werts, S. P., and A. H. Jahren. 2007. Estimation of temperatures beneath the Middle Palaeolithic site of Chez-Pinaud Jonzac (France). Journal of archaeological campfires using carbon stable isotope composition of soil Archaeological Science 40(2):1176–1185, doi:10.1016/j.jas.2012.09.003. organic matter. Journal of Archaeological Science 34(6):850–857. Rottländer, R. C. A. 1983. Einführung in die naturwissenschaftlichen Methoden Winograd, I. J., J. M. Landwehr, K. R. Ludwig, T. B. Coplen, and A. C. Riggs. in der Archäologie. Tübingen: Archaeologica Venatoria. 1997. Duration and structure of the past four interglaciations. Quaternary Ruff, C. B., E. Trinkaus, A. Walker, and C. S. Larsen. 1993. Postcranial robusticity Research 48:141–154. in Homo. I. Temporal trends and mechanical interpretation. American Journal Wrangham, Richard. 2009. Catching fire: how cooking made us human. New of Physical Anthropology 91(1):21–53. York: Basic Books. S288 Current Anthropology Volume 58, Supplement 16, August 2017

Technologies for the Control of Heat and Light in the Vézère Valley Aurignacian

by Randall White, Romain Mensan, Amy E. Clark, Elise Tartar, Laurent Marquer, Raphaëlle Bourrillon, Paul Goldberg, Laurent Chiotti, Catherine Cretin, William Rendu, Anne Pike-Tay, and Sarah Ranlett

Online enhancements: supplemental figures

We can trace the beginnings of our knowledge of early Upper Paleolithic (Aurignacian) use of fire to the pioneering 1910–1911 excavations at Abri Blanchard undertaken by Louis Didon and Marcel Castanet. At Blanchard, the excavators recognized and described fire structures that correspond in many ways to features excavated more re- cently in Western and Central Europe. Here, we address the issue of heat and light management in the early Upper Paleolithic, demonstrating a pattern that builds on these early excavations but that is refined through our recent field operations. Topics to be discussed include (1) recently excavated fire structures that suggest complex fire man- agement and use, (2) the seemingly massive use of bone as fuel in most early Aurignacian sites, and (3) the anchoring of skin structures for purposes of heat retention with fireplaces behind animal-skin walls. Furthermore, new data on activities around fireplaces make it possible to infer social and organizational aspects of fire structures within Au- rignacian living spaces. The vast majority of early Aurignacian occupations, most of them now dated to between 33,000 and 32,000 BP (uncalibrated), occurred on a previously unoccupied bedrock platform into which the occupants dug their fire features.

The use of fire has long been recognized as a key innovation 2012; Karkanas et al. 2007; Roebroeks and Villa 2011; Stahl- in human evolution as a source of light and heat, a mecha- schmidt et al. 2015; Wiessner 2014; Wrangham and Caromed nism for cooking (Chazan 2017; Villa, Bon, and Castel 2002; 2010). Often, however, we are overly preoccupied with the Wrangham 2017), and a focal point for fireside activities and earliest occurrences of fire rather than its use in cultures where social bonding (Alperson-Afil 2008; Fernández Peris et al. we have long known controlled fire to be present.

Randall White is Professor at the Center for the Study of Human Origins in the Department of Anthropology at New York University Boston University (675 Commonwealth Avenue, Boston, Massachu- (25 Waverly Place, New York, New York 10003, USA) and Associate setts 02215, USA [[email protected]]) and at the Institute for Archae- Researcher at CREAP in the Maison des Sciences de l’Hommeetdela ological Sciences at the University of Tübingen (Rümelinstr. 23, 72070 Société of CNRS at the Université de Toulouse-Jean Jaurès (USR 3414, Tübingen, Germany) and the School of Earth and Environmental allées Antonio Machado, F-31058 Toulouse Cedex 9, France [randall Sciences at the University of Wollongong (Wollongong, New South [email protected]]). Romain Mensan is a Researcher in the Laboratoire Wales 2502, Australia). Laurent Chiotti is Researcher/Director in the TRACES, CNRS at the Université de Toulouse-Jean Jaurès (UMR 5608, Département de Préhistoire du Muséum National d’Histoire Naturelle F-31058 Toulouse Cedex 9, France [[email protected]]). Amy E. at UMR 7194 and the CNRS of the Musée de l’Abri Pataud (24620 Les Clark is a Postdoctoral Fellow in the School of Anthropology at the Eyzies-de-Tayac, France [[email protected]]). Catherine Cretin is Con- University of Arizona (1009 East South Campus Drive, Tucson, Arizona servator of Patrimony at the Musée National de Prehistoire in Les Eyzies 85721, USA [[email protected]]). Elise Tartar is a Researcher at (1 Rue du Musée, 24620 Les Eyzies-de-Tayac-Sireuil, France [catherine the Equipe Ethnologie Préhistorique of CNRS (UMR 7041, 21 allée de [email protected]]). William Rendu is a Researcher at PACEA of l’Université, 92923 Nanterre Cedex, France [[email protected] CNRS at the Université de Bordeaux (UMR 5199, 3074 Bordeaux Cedex, .fr]). Laurent Marquer is a Researcher at the Laboratoire GEODE of France [[email protected]]). Anne Pike-Tay is Professor in the Depart- CNRS at the Université de Toulouse-Jean Jaurès (UMR 5602, Toulouse ment of Anthropology at Vassar College (124 Raymond Avenue, Box 701, Cedex 9, France) and in the Department of Physical Geography and Poughkeepsie, New York 12604-0701, USA [[email protected]]). Sarah Ecosystems Science at Lund University (Box 117, 221 00 Lund, Sweden Ranlett is a Graduate Student in the Department of Anthropology at the [[email protected]]). Raphaëlle Bourrillon is a Researcher at University of Toronto (19 Russell Street, Toronto, Ontario M5S 2S2, CREAP (see above for CREAP info [[email protected]]). Paul Canada [[email protected]]). This paper was submitted 25 VII 16, Goldberg is Professor Emeritus in the Department of Archaeology at accepted 21 III 17, and electronically published 21 VII 17. q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0012$10.00. DOI: 10.1086/692708 White et al. Heat and Light in the Vézère Valley Aurignacian S289

Our research on classic sites in the Aurignacian shows that carved in the limestone to enclose the rock-shelter with skins not only was fire an essential part of life for human groups and retain heat. Many of our inferences surrounding these during this time but that it was manipulated and used in a topics make use of an abundance of experimental research standardized and consistent manner within an overarching over the past 20 years that has sought to investigate cultural system of heat and light capture and control. This includes and natural formation processes of combustion-related fea- altering the limestone bedrock to create concavities in which tures (see, e.g., Aldeias 2017; Costamagno et al. 2010; Lejay the fires sat and the creation of adjacent structures for the et al. 2016; Miller et al. 2010; Théry-Parisot et al. 2002). control and manipulation of different types of heat (e.g., the removal of hot coals from the primary fire feature to an ad- jacent location for a special use) and light (portable lamps). Classic Aurignacian Sites of the Vézère Valley In this paper, we will first present a series of early Aurignacian and Associated Fire Structures sites from the Vézère Valley, combining our recent interven- Abri Castanet and Abri Blanchard tions as well as observations from early excavations. A clear pattern will emerge: at site after site, the earliest Aurignacian When Marcel Castanet discovered a complex of fire features occupation sits directly on the bedrock, dates to between 32,000 at Abri Blanchard in May 1910, he and his employer, Louis and 33,000 years BP (uncalibrated), and exhibits complex fire Didon, were ill equipped to understand the importance of this structures within human-made depressions in the bedrock. The find, and given the state of knowledge before World War I, quality of information for each of these fire structures depends they had few points of comparison. Castanet drew a rough on when they were excavated, but a clear pattern can never- sketch in plan view (fig. 1), and after having recovered many theless be established. We do have a particularly clear window spectacular artifacts associated with these fire features, he on this pattern, however, that derives from our recent exca- simply backfilled the entire area containing the fire pits. vations at Abri Castanet. Between 2005 and 2010 we excavated Nearly 6 months later, in October 1910, Didon asked Cas- an interconnected set of fire structures that is testament to the tanet to reexpose the fire structures so that he could observe sophisticated management of fire in the Aurignacian. Finally, them himself. While there are allusions to photographs taken we will discuss a few key topics relevant to the management of by Didon, these appear not to have survived. The only record fire, heat, and light in the early Aurignacian. These include the is that plan view sketched by Castanet on May 21, 1910 (fig. 1), organization of activities around fire structures, use of bone or which shows one large elongate feature and three smaller wood as fuel, and the production of pierres à anneaux, rings satellite features all dug into the friable limestone bedrock.

Figure 1. Castanet’s May 21, 1910, plan view of the fire features at Abri Blanchard, annotated and translated. L. Didon archive (cf. Delluc and Delluc 1978). A color version of this figure is available online. S290 Current Anthropology Volume 58, Supplement 16, August 2017

In the Blanchard publication, Didon (1911) described the gnacian site, the Abri de la Souquette, just 50 m across the valley fire features as follows. from Abri Blanchard. Unfortunately, because of late medieval The four hearths were constituted of shallow pits dug into quarrying and massive pillaging at the beginning of the twen- the bedrock terrace. Three of the pits were circular, with an tieth century, only a small fragment of the site was available for 2 approximate diameter of 0.5 m and a depth of 0.2 m. The modern excavations by Roussot (1982) in the 1980s. Roussot fourth, rectangular with semicircular extremities, measured observed and excavated a few square meters of the Aurignacian fi 3.5 m long by 1.5 m wide, with a depth of ca. 0.4 m. They I level, situated directly on bedrock (see g. S2). were infilled with ash and calcined bone. Because of the We have been able to analyze and date this early Auri- thickness of this layer, the ashes seemed to date to the pre- gnacian assemblage, which is virtually identical in its contents vious day, and certain blocks of the material, a sort of breccia (typologically, technologically, artistically) to Blanchard (lower composed of ashes and conglomerated bones, were really level) and Castanet (lower level in the northern and southern fi impressive. I kept one of these blocks, and it was as if one sectors). Although no preserved re features were discovered in ’ needed simply to blow on it to revive the fire that had been the small area of Roussot s excavations, reindeer bone fragments fi 5 extinguished so many centuries ago. from this level yielded molecular ltration dates of 32,400 500 BP and 32,150 5 450 BP (dates uncalibrated; O’Hara et al. Didon describes these remarkable breccia-like blocks with 2015). This information from La Souquette reinforces the pat- greater detail in a letter to Breuil in February 1911, 4 months tern that in the Vézère Valley, the first Aurignacian I occupation after the pits had been reexposed by Castanet. occurred directly on bedrock and dates to around 32,000 BP Yesterday, I retrieved from a case some hearth fragments. I uncalibrated. use the term “hearth” in the usual sense of the word because these are far from being simple chunks of breccia. One block, Abri Cellier 30 # 12 cm, was composed of large fragments of carbonized bone and reindeer antler with ashes adhering to them. It In the summer of 1927, George Collie from Beloit College, seemed as if one simply needed to blow on them to reignite Wisconsin, undertook excavations at Abri Cellier (Collie 1928; the fire; it is very impressive. I retrieved them from the de- White and Knecht 1992) near Le Moustier in the Vézère Valley. pressions dug into bedrock (by the Aurignacians).1 Although there is some stratigraphic confusion (White 1992; White and Knecht 1992), our return to the site in 2014 did much Beginning in 1911, Denis Peyrony employed Marcel Cas- to clarify the situation (figs. 2, S3; White et al., forthcoming). As tanet to excavate the Abri Castanet, situated on the same bed- at Abri Castanet and Abri Blanchard, the initial occupation of rock platform just 50 m to the south of Abri Blanchard. At Abri the site by early Aurignacian groups was on an exposed bedrock Castanet, as at Abri Blanchard, the Aurignacian I layer (Pey- platform. It was only late in the excavation that Collie recog- rony’s layer A) sat directly on bedrock. Following the excavation nized bedrock at the site and thus did not clearly observe related of this layer, Peyrony (1935) described and illustrated precisely fire features. In cleaning the bedrock surface in 2014, we were the same kinds of fire features dug into the bedrock as those nonetheless able to observe fire-reddened depressions (cuvettes) found at Abri Blanchard (see fig. S1; figs. S1–S15 available on the surface of the bedrock and were even able to excavate online). He observed four features, one of them 1.3 m long. He highly calcined deposits within a residual fire pit feature pre- noted, as had Didon before him, that the most spectacular served in a deep hollow in the bedrock (fig. 3). Bone from this artifacts were concentrated around these fire features. At Blanch- structure yielded an ultrafiltration date of 33,600 5 550 BP ard, this Aurignacian I layer has now yielded to our team two (White et al., forthcoming). We were struck by the similarity of hydrozyproline dates on mammal bone of 33,420 5 350 and the Cellier features to those described at Blanchard and espe- 33,960 5 360 BP (uncalibrated; Bourrillon et al., forthcoming), cially to those described by Peyrony (1946) and excavated by us and at Castanet, 15 ultrafiltration dates on bone give an average at Abri Castanet. age of 32,400 BP for the same layer (White et al. 2012). We will return to Castanet later in this paper to describe our recent excavation of four such fire features within the previously un- Abri Pataud excavated southern portion of the site. In his excavations at Abri Pataud, Movius (1966) paid particular attention to fire features. Those in layer 14, situated directly on La Souquette bedrock, are indistinguishable from those described above in having been excavated into the bedrock platform (fig. 4). Blanchard and Castanet are situated at the base of the cliffs on Other sites in the Vézère have yielded the same pattern of the eastern slope of the small Castel-Merle valley. The opposing occupation by Aurignacians directly on bedrock with instal- western side of the valley has also yielded a rich early Auri-

2. Also, A. Roussot, unpublished administrative report, “Abri de la Sou- 1. Périgueux, February 2, 1911, Louis Didon to H. Breuil, Breuil quette, Commune de Sergeac Dordogne. Rapport de Fouilles. Récapitulatif, Archive, Muséum National d’Histoire Naturelle, Paris. 1980–1982.” White et al. Heat and Light in the Vézère Valley Aurignacian S291

Figure 2. Stratigraphic section on the eastern extremity of the Abri Cellier (from White et al., forthcoming). Layer 104 at the base of the sequence is an early Aurignacian level directly on bedrock. A color version of this figure is available online. lation of fire features directly on or in the fractured and first encounter to be either (1) several adjacent fire structures, spalled limestone surface (Delporte 1968). However, we have (2) a single combustion feature having been subjected to post- restricted our discussion here to those with which we have depositional alteration, (3) an ash dump, or (4) all of these at the personal experience. It is worth noting, however, that where same time. the early Aurignacian is followed by later Aurignacian levels In order to evaluate the relative merits of these different within the same stratigraphic sequence, such as at Abri Pataud hypotheses, we moved from excavation in .25 m2 units to a and Abri du Facteur, there is significant change through time .0625 m2 grid. We sought to sample and provenience all the in hearth structure (e.g., cobble lining) and arrangement (Del- fine-fraction bulk sediment in order to record, quantify, and porte 1968; Movius 1966). situate in three-dimensional space all of the products and residues of combustion. Our objective was to treat raw sedi- ment samples in the laboratory in order to detect, recover, and Returning to Castanet study microvestiges of combustion such as charcoal, burned bone, and phytoliths. Our own excavations at Abri Castanet between 2005 and 2010 In complement to the recovery of bulk sediments, we un- focused on what appeared at first to be a massive and diffuse dertook the extraction, embedding, and micromorphological fire feature (stratigraphic unit 109) that resolved into three study of oriented blocks of in situ sediment. Micromorphology separate features as we descended through the layer (the equiv- sought an understanding of the microstratigraphic context of alent of Peyrony’slayerA;fig. 5). All three of these features were each sample and a characterization of the taphonomic pro- excavated by the Aurignacians into the bedrock platform by cesses underlying the nature and distribution of macromi- removing a number of plaquettes of which the bedrock is con- crovestiges recovered from the raw sediments. stituted. These three features are named Structures 216, 217, As we descended through the fire feature, it became ap- and 218 (see fig. S4). parent that the diffuse black deposit (our stratigraphic unit 214) resolved into three observable structures, which we la- Excavation Strategy and Techniques beled 216, 217, and 218. We continued to employ the same fine-grained procedures but now carefully separated the bulk The meticulous, but routine excavation of the overlying, un- samples from adjacent structures 217 and 218. differentiated black stratum (our US 214), allowed us to iden- Structure 216 posed special problems because it contained tify a spatially extensive (ca. 2.6 m2) feature, which seemed at little sediment, and its heavily calcined bone infilling did not Figure 3. Abri Cellier, cleaned bedrock with small fire pit adjacent to two large fire-reddened depressions excavated by the Aurignacians into the underlying bedrock. A color version of this figure is available online. White et al. Heat and Light in the Vézère Valley Aurignacian S293

Figure 4. Left, Abri Pataud, early Aurignacian layer 14, hearths T and U in square B of trenches 3 and 4. Hearth T is completely excavated, leaving only a depression in the bedrock (from Movius 1966). Right, Abri Pataud, early Aurignacian layer 12, fire pits of differing dimensions, including a small feature filled with burned bone as at Castanet and Cellier (from Movius 1966). permit normal excavation because osseous tissue simply fell with no interstitial sediment except some heat disaggregated to dust under the slightest tool contact. The decision was made limestone bedrock. to remove the entire contents in bulk, again controlling the Structure 216 contained only 29 lithic pieces, two of which microstratigraphy with micromorphological samples. refit with a piece from Structure 218. All three are fire ex- ploded. Another piece refits with six others from outside the fi Magnetic Susceptibility re features per se. All seven pieces are products of the creation of a carinate scraper/core blank (Chiotti and Cretin 2011). The In 2010, we performed magnetic susceptibility analysis (Brodard constituents of both of the above refits are from the same block et al. 2016; Lecoanet, Léveque, and Ambrosi 2003; Lecoanet, of flint. Léveque, and Segura 1999) of the excavated area, including the The infilling is identical to that of Structure 217 with respect observed combustion features. One of the objectives of mapping to oxidized residues, but the magnetic signal is less intense, the magnetism of the substratum and immediately overlying leading to the suspicion that the contents of Structure 216 were deposits is to distinguish between zones where combustion extracted from Structure 217 while still hot. At its summit, this occurred in situ and those of secondary deposition. Because the fire feature had a large portion of a burned bovid horn core underlying Coniacian bedrock showed virtually no magnetic (visible in fig. S5). Observations at all scales converge to sug- signal, the surprisingly intense magnetic signal observed at Cas- gest that the nature and function of this feature are radically tanet had to emanate from the structures’ infilling. Our micro- different from Structures 217 and 218. In its absence of sedi- vestige analysis of the fire features’ contents enabled us to ment and dominance of burned/calcined bone, Structure 216 demonstrate that each of the three structures contained high resembles very closely the small fire features with “breccia” frequencies of iron oxides and kaolinite at both microscopic and described by Didon at Abri Blanchard, and it is very similar to macroscopic scales. During combustion these were transformed residual structure 1 from Abri Cellier. to magnetite and took on the prevailing planetary magnetic fi eld. The three structures appear to be contemporaneous and Structures 217 and 218 to have functioned more or less independently. Some of the conclusions of the magnetic analysis will be presented below. Structure 217 is a roughly circular structure with a diameter of ca. 0.6 m, bordered on the west by six vertically arranged Structure 216 plaquettes that close off the northwest side of the structure the margins of which are otherwise formed by the contours of This small fire feature, dug into the bedrock platform, was the depression dug into bedrock by the Aurignacians. Two roughly 30 cm in diameter with a maximum depth of 20 cm. profiles excavated into structure 217 show its infilling and It contained a dense deposit of burned and calcined bone topography (see fig. S7). It contained 64 lithic pieces, none Figure 5. Castanet southern sector with “footprints” of three distinct fire features after excavation. Grid in square meters divided into one-quarter square meters. A color version of this figure is available online. White et al. Heat and Light in the Vézère Valley Aurignacian S295 refitting with the other fire structures. All other refits were tools, fragments of colorants such as ochre and hematite, and internal to one of the fire structures. debris from the production of personal ornaments. Much of Magnetic susceptibility and micromorphology (see figs. S8– this constitutes the refuse from a diversity of activities that took S10) indicate that Structure 217 is relatively well preserved place around the fire features. As is the case for many Paleo- and that combustion took place there. In contrast, Structure 218 lithic and ethnographic sites (Binford 1998; Gamble 1991; Hahn is magnetically heterogeneous, suggesting that the contents are 1988; Miller et al 2010; Rigaud and Simek 1991; Stapert 1989; not at the site of original combustion. Possibilities include an ash Surovell and Waguespack 2007; Vaquero and Pastó 2001), the dump or an overflow of ashes and combusted material from fire features at Castanet served as a focal point or “tether” for Structure 217 that were subsequently subjected to trampling activities within the site. (see Schiegl et al. 2003). Curiously, 217 and 218 each had an It is noteworthy that carinate scrapers/bladelet cores were unburned cervid mandible on their upper surfaces. Structure produced around these fire features (Chiotti and Cretin 2011), 218 yielded 73 lithic pieces with a single, heavily burned piece and resulting microbladelets are abundant. However, the blades refitting with two pieces from Structure 216 also heavily burned. and flakes that served as blanks for these and other retouched Although we refer to 216, 217, and 218 as “structures” in the lithic tools were produced somewhere outside of the excavated plural, these features appear to have been components of one area. Not a single flint blade or flake core has been recovered large, simultaneously utilized fire complex. From the evidence from the 30 m2 excavated area of the southern sector of Abri presented above, we can infer that Structure 217 was used as Castanet. The act of blade/flake production, resulting in messy the primary fire location, and its contents were moved to and dangerous by-products, seems to have been excluded from Structures 216 and 218 for different purposes. Hot bone the living space immediately surrounding the Castanet fire embers seem to have been moved to Structure 216, probably as features. a way to control heat and flame for special tasks as yet unde- Two activities exemplify this pattern: osseous tool/weapon termined. Structure 218 was, at the very least, a dumping lo- manufacture and use, and personal ornament manufacture cation for the contents of Structure 217, but it may have served and use. It is particularly interesting to contrast the spatial more diverse purposes as well. The abundance of fire-altered location of formal finished artifacts and their production iron oxides within these features raises the possibility of heat debris. processing of this material for coloring and/or abrasive func- tions, an idea supported by thick deposits of paint (or at least ochre paste) on the bedrock adjacent to Structure 217 (fig. S11). Osseous Tool/Weapon Manufacture and Use In sum, this fire complex indicates a highly specialized and systematic use of fire that is spatially distinctive. An additional Observed patterns in the distribution of different formal cate- feature was found at Abri Castanet southern sector, residual gories of osseous tools, weapons, and debitage and the distri- structure 1, situated approximately 3 m from the 216, 217, and bution of burned and unburned osseous pieces almost cer- 218 complex (see fig. S6). tainly reflect functional differences among fire features and the The spatial orientation of these three fire structures was spatial organization of different activities associated with them. found replicated at Abri Cellier when we uncovered the bed- Fine-grained proveniencing permits a number of specific ob- rock (described above). Not only did we find a feature that servations. appeared to be a direct analogue to Structure 216, but adjacent Almost all formal osseous artifacts are situated outside the to this structure were two fire-reddened concavities in the perimeter of the three combustion features 216, 217, and 218 bedrock (fig. 6) that were the same size and orientation as (fig. 6), and none are burned. Split-based antler points and Structures 217 and 218. the tongued pieces related to their production (Tartar and Similar fire complexes were probably excavated by Didon White 2013) are concentrated to the south of these features. and Peyrony (or more accurately, Castanet) but were not ob- Smoothers (lissoirs) are concentrated at the northeastern mar- served as such. In particular, the large rectangular fire structure gins. found at Blanchard and described by Didon and Castanet (fig. 1) Bone retouchers, intermediate tools such as wedges and is suspiciously similar to our perception (in descending) of a chisels, as well as bone and antler waste and debitage prod- single spatially extensive feature before careful excavation ucts bearing technical stigmata show no special distribution revealed that it was made up of three interconnected but distinct (apart from four retouchers found adjacent to the 1995–1998 structures. fire feature 1). Some of these objects are found outside of the fire structures per se but within the perimeter (near its eastern Activities Concentrated around the Fire Features: limit) of the diffuse cloud of black sediment (US 109) en- Examples of Osseous Industry and Ornaments countered in descending through the layer (dotted outline on fig. 6) before the three fire features revealed themselves. The surface excavated by us at Castanet amounts to about 30 m2. Small burned pieces of antler (N p 55) showing no tech- A single level is present. This layer is dense with faunal remains, nical traces are abundant in both Structures 217 and 218 as debris from the final stages of production of lithic and osseous well as in the area just to their north, but they are entirely S296 Current Anthropology Volume 58, Supplement 16, August 2017

Figure 6. Spatial distribution of principal categories of osseous objects that make up the bone and antler industry in the southern sector of Abri Castanet. Grid in square meters divided into subsquares of one-quarter square meter. A color version of this figure is available online.

absent from Structure 216. The abundance of these nondietary with fragments being projected during high-impact direct per- faunal remains in Structures 217 and 218 might result from the cussion (Heckel and Wolf 2014). first steps of antler exploitation that were somehow linked to Farther along the bead production chain, unfinished bead these structures. If so, Structure 216 was not involved in such stages have a different distribution, with important concen- activities. trations to the north and west of the Structure 216, 217, 218 perimeters. Both raw ivory fragments and unfinished pro- duction stages show concentrations in square H12, while that Personal Ornament Manufacture and Use square does not contain a single finished bead (fig. 7). The distribution of objects linked to personal ornaments and Finished, unbroken beads show two hot spots, one to the their production is equally suggestive. Raw ivory fragments north of 217–218 and the other toward the southeastern ex- imply the on-site reduction of large chunks of subfossil mam- tremity of the excavated area (fig. 7). Broken, finished beads moth tusks. The creation of usable blanks by direct percussion (not shown) have a broader distribution, suggesting loss due and torsion of fragments of old tusks seems to have been the to breakage during use. objective. The morphology of the by-products of this percussion The spatial organization of the two examples presented varies from small flakes and splinters from the external tusk here, osseous technology and personal ornaments, indicate layers to angular fragments of complex morphology emanating that some spatial patterning between production debris and from direct percussion of the interior parts of already desiccated finished products can be observed around the fire features at tusks. Fresh tusks cannot be worked in this way. The spatial Abri Castanet. The presence of both production debris and distribution across the entire excavated area (fig. 7) is consistent finished products indicates that the fire features were a focal Figure 7. In all cases the grid is in square meters divided into subsquares of one-quarter square meter. Top left, Abri Castanet, southern sector, 1995–2010. Shading indicates frequency per one-quarter square meter of raw ivory fragments. Top right, Abri Castanet, southern sector, 1995–2010. Distribution of stages 1–4 of the basket-shaped bead production chain. Shading indicates frequency per one-quarter square meter of unfinished objects. Bottom, Abri Castanet, southern sector, 1995–2010. Spatial distribution of finished ivory basket-shaped beads including fragments. Shading indicates frequency per one-quarter square meter of finished beads. A color version of this figure is available online. S298 Current Anthropology Volume 58, Supplement 16, August 2017 point for a diversity of activities that were concentrated around its borders.

Use of Bone as Fuel Early Aurignacian sites in southwest France are famous for yielding very high frequencies of burned and calcined bone with relatively little wood charcoal being present (Costamagno et al. 2005; Marquer et al. 2010; Théry-Parisot 2001, 2002; Théry-Parisot et al. 2002). While this has often been inter- preted as reflecting an emphasis on bone as fuel in Pleistocene landscapes where tree cover is minimal, the link to low tree- cover environments is not always agreed on. Microcharcoal analysis (Marquer 2010; Marquer et al. 2010, 2015) of the Castanet fire feature contents provides new insights into the observed predominance of bone and scarcity of wood charcoal there. In fact, controlled sieving of the fire feature contents yielded an abundance of microscopic wood charcoal (microscopic charcoal results from combustion processes and secondary fragmentation of macroscopic charcoal), almost all of -it with a particle size of less than 63 mm(fig. 8). Therefore, while bone was certainly used as a fuel at Castanet, the abun- dance of wood charcoal at smaller particle sizes indicates that wood played a much more important role than was previously assumed. In sum, wood charcoal appears to be invisible to the usual array of recovery techniques because of taphonomic or cultural processes that remain to be determined (Marquer et al. 2010, 2012). The pattern found at Castanet does not necessarily extend to all early Aurignacian sites; although Marquer et al. (2010) found an abundance of microscopic charcoal at Abri Pataud, Figure 8. Numbers of fragments and particles of charcoal and burned bone still dominated all size fractions (Marquer et al. burned bones and their ratios relative to size classes of the re- 2010). Bone, however, is less susceptible to taphonomic pro- covered samples. cesses than is charcoal (Marquer et al. 2010), and based on rates of bone combustion established by experiments performed of the Dordogne, winter days are short (8 hours, 47 minutes of by Théry-Parisot (2002), an overly high dependence on bone as daylight at the winter solstice compared to 15 hours, 35 minutes fuel would result in the improbable situation of humans hunting at the summer solstice3), and south orientation would have to feed the fire. It cannot be assumed that bone was more provided maximum daily light. plentiful than wood in the environment; indeed, neither were A common occurrence in Vézère Valley early Aurignacian plentiful, and Aurignacians likely exploited any and all fuel sites is that of so-called pierres à anneaux, limestone blocks sources available to them. It is therefore most likely that both and slabs with a distinctive form of a carved ring. Castanet, wood and bone were utilized as fuel in ratios that varied by Blanchard, La Souquette, Cellier (fig. 9), and Pataud have all availability but that, in general, the use of wood in early Au- produced these. At Blanchard and Castanet, these were ob- rignacian sites is underestimated. served to be concentrated along the supposed original drip line of the shelter (Didon 1911; Peyrony 1935). In the last Heat and Light Capture and Retention year of excavations at Castanet, we too found anneaux in the rubble of the collapsed drip line of the shelter. Maximization of heat and light would have been beneficial to We have been able to demonstrate as well that these labor- Upper Paleolithic hunter-gatherers in the late Pleistocene lower intensive features were also placed on freestanding blocks on the midlatitudes of late Pleistocene Europe. It has long been rec- living surface of the site. Our experiments show that these rings ognized that in the Vézère Valley, the very placement of their withstand very little force, strain, and stress, and we also know occupations (see fig. S12) was an accommodation to these from our analysis of site formation that the ceiling of the shelter concerns (White 1983, 1985). While the advantage of solar was approximately 2 m above the occupational surface. energy capture on south-facing cliff faces is significant (Bouvier 1967; Legge 1972), it should not be forgotten that at the latitude 3. https://www.timeanddate.com/sun/france/bordeaux?monthp12. White et al. Heat and Light in the Vézère Valley Aurignacian S299

Figure 9. Abri Cellier, roof collapse block with anneau and engraved profile of mammoth. A color version of this figure is available online.

These data and observations incline us to support Denis The site has been the object of long-term, programmed Peyrony’s old hypothesis (Peyrony 1935) that the anneaux excavations by one of us (RM) since 2000.5 In the course of served to guide and suspend animal skins from the drip line this research, 28 combustion features have been excavated, of the shelter in order to retain heat, to reflect heat and light concentrated in two paleochannels that served to shelter the back into the shelter, and to serve as a windbreak vis-à-vis Aurignacian occupants. Around these fire features are dense westerly and northwesterly winds entering from the front of concentrations of archaeological material reflecting func- the shelter. Moreover, new seasonal data, based on cementum tionally complementary poles of activity. annuli analysis (Naji, Gourichon, and Rendu 2015; Rendu Two primary zones have been identified, one in each of the 2007, 2010) of reindeer and horse teeth, are categorical;4 Cas- perpendicularly oriented paleochannels. The first contains tanet was an exclusively cold season occupation (figs. 10, S13). several vast multifunctional structures that we interpret as The southern sector of Abri Castanet has also yielded a fat- belonging to a “domestic area.” The second is constituted of burning lamp, associated with Structure 218, which would several structures of a more specialized nature that we in- have complemented and rendered portable the light provided terpret as a “workshop area.” by the fires (see fig. S14). Such portable light technology (de The principal domestic area is organized around a con- Beaune and White 1993) would have allowed the Aurigna- centration of three fire structures. These are three pits, two cian occupants of Castanet to better adapt to long winter of which are characterized by intense traces of combustion nights and to the darkness of the enclosed living space and (12and16)andthethirdshowingbrownishimpregnations the outside surroundings. and an infilling of burned bones (see fig. S15). This organi- zation is strikingly similar to that of the southern sector of Régismont-le-Haut, a Point of Comparison Abri Castanet. The archaeological material associated reveals for the Vézère Valley Sites a range of activities such as bladelet production, tool re- Régismont-le-Haut (Poilhes, Hérault) is one of the rare open- air Aurignacian sites to have yielded well-preserved evidence 5. The lithic industry is rather original compared with other assem- of the organization and arrangement of living space separated blages available for comparison, making chronological attribution diffi- into distinguishable activity areas (Bon and Mensan 2007). cult. Attempts to date the site radiometrically (Bon 2002) do not allow attribution to a particular Aurignacian phase. It could be a regional facies 4. For example, A. Pike-Tay’s unpublished manuscript, 2000, “Dental of the early Aurignacian or a previously unobserved variant of a later growth mark analysis of Rangifer tarandus and Equus of Abri Castanet.” Aurignacian. S300 Current Anthropology Volume 58, Supplement 16, August 2017

pation of bare bedrock terraces in the period between 33,000 and 32,000 BP (uncalibrated). Early Aurignacians, moving directly onto the previously unoccupied bedrock, dug their fire pits directly into the substrate. This seems to have been a characteristic of the early Aurignacian only; cobble-lined or bordered fireplaces show up and become dominant in later stages of the Aurignacian. The fire features themselves also display a very consistent pattern. The fire complex that we excavated at Abri Castanet (consisting of Structures 216, 217, and 218) demonstrates a fire use that was intensive and easily adaptable to many spe- cialized functions. Given that Castanet was a winter occupied site, the fire in Structure 217 could have burned more or less continuously, with its contents subsequently moved to Struc- tures 216 and 218 for cleaning or for tasks that required the fl Figure 10. Cementum annuli analysis of season of death for 25 control of heat and ame. A nearly identical arrangement of reindeer teeth and one horse tooth from the southern sector of fire features found at Abri Cellier reinforces the view of a Abri Castanet based on a large comparative collection of known highly systematic management of fire in the early Aurignacian. age at death (Burke and Castanet 1995; Pike-Tay 1995, 1998). Evidence from Abri Castanet suggests that while these fires were fueled to a certain extent by bone, wood played a very sharpening, and the processing of mineral pigments as at important role, one that is probably significantly underrep- Castanet. resented because of taphonomic processes. Provisioning the At Régismont, all of the combustion features and sur- site with enough fuel to feed these intensive fire complexes rounding concentrations of material culture are attributed to would have come at a significant economic cost. a single, spatially continuous occupational surface. The re- The hypothesis of the anchoring of skin structures within dundancy in the spatial organization and association of dif- rock-shelters for purposes of heat retention, with fireplaces ferent groups of structures such as the one described above behind animal-skin walls, was made first in 1935 by Peyrony sustains the hypothesis of a single episode of occupation that (1935). Our recent research supports this hypothesis. It is took the form of a vast seasonal residential camp. The careful worth asking whether this is a seasonal phenomenon, as it is observation and documentation of the internal organization now clear that Abri Castanet was an entirely cold season of the site around combustion features makes Regismont-le- occupation. Haut a key point of comparison for the Aurignacian record of New robust data on activities around fireplaces at sites like the Vézère Valley (Anderson et al., forthcoming). That com- Castanet make it possible to infer social and organizational parison leads us to propose that the redundant fire feature aspects of fire structures within Aurignacian living spaces. organization that we observe in the Vézère was the expres- The spaces created within these Aurignacian sites, combining sion of a geographically widespread characteristic of Auri- complex fire structures, anneaux for enclosing the rock-shelters gnacian culture. with hides, and engraved and painted artwork on the walls and ceilings (White et al. 2012) would have been extremely impor- Conclusions tant sanctuaries for the protection from the cold and the ex- tension of daylight hours. Using both old archives and the new methods and data During the cold winter evenings, when daylight departed available to us, we are beginning to be able to address im- early, Aurignacian groups would have been able to continue to portant questions of heat and light management in the early make beads, produce and maintain their tools and weapons, Upper Paleolithic. In one region, the classic Vézère Valley of process pigments, prepare meals, and strengthen social bonds southwest France, we have seen the existence of complex and through fireside talk (Dunbar 2014; Wiessner 2014). This diverse fire features that require excavation and analysis with warm and protected space would have given these groups a forensic precision. The kinds of fire features and their ar- powerful advantage and may explain why Aurignacian peoples rangement across living surfaces repeat from one site to the seem to have thrived in the glacial conditions of late Pleisto- next and may well be part of a broader European pattern. We cene France. maintain that fire was manipulated and used in a standard- ized and consistent manner within an overarching system of Acknowledgments heat and light capture and control. A remarkable aspect of the early Aurignacian record in the The research described here has been supported since 1994 by Vézère Valley is that we can follow particular fire features generous grants from the US National Science Foundation, the across a time horizon marked by the availability for occu- Direction Régional des Affaires Culturelles d’Aquitaine (DRAC- White et al. Heat and Light in the Vézère Valley Aurignacian S301

Aquitaine), the L. S. B. Leakey Foundation, the Reed Founda- Burke, A., and J. Castanet. 1995. Histological observations of cement growth in horse teeth and their applications to archaeology. Journal of Archaeo- tion, the Rock Foundation, the Fine Foundation, the Center for logical Science 22:479–493. International Research in the Humanities and Social Sciences Chazan, Michael. 2017. Toward a long prehistory of fire. Current Anthro- (UMI 3199-CNRS-NYU), the Institute for Ice Age Studies, the pology 58(suppl. 16):S351–S359. Chiotti, L., and C. Cretin. 2011. Les mises en forme de Grattoirs Carénés/ Theodore Dubin Foundation, the Service Archéologique Dé- Nucléus de l’Aurignacien Ancien de l’ Abri Castanet (Sergeac, Dordogne). partemental de la Dordogne, New York University, the Fyssen Paléo 22:69–84. Foundation, and the Fulbright Foundation. Much of the re- Collie, G. 1928. The Aurignacians and their culture. Beloit, WI: Logan Mu- seum, Beloit College. search described here results from the Franco-American col- Costamagno, S., I. Théry-Parisot, J. P. Brugal, and R. Guibert. 2005. Tapho- laborative exchange titled “Aurignacian Genius: Art, Daily Life nomic consequences of the use of bone as fuel: experimental data and and Social Identity of the First Modern Humans of Europe” archaeological applications. In Biosphere to lithosphere: Proceedings of the 9th Conference of the International Council of Archaeozoology,T.O’Conner, (UMI 3199-CNRS-NYU and UMR 5608-TRACES), funded by ed. Pp. 51–61. Oxford: Oxbow. a 3-year grant from the Partner University Fund and the An- Costamagno S., I. Théry-Parisot, D. Kuntz, F. Bon, and R. Mensan. 2010. drew Mellon Foundation. The cementum analysis on the more Taphonomic impact of prolonged combustion on bones used as fuel. In The taphonomy of burned organic residues and combustion features in ar- recent sample was conducted within the CemeNTAA project chaeological contexts. I. Théry-Parisot, L. Chabal, and Costamagno S., eds. funded by the French National Agency for Research (ANR-14- Palethnologie 2:169–183. CE31-0011). The support of the Wenner-Gren Foundation for de Beaune, S., and R. White. 1993. Ice age lamps. Scientific American 268:108– “ ” 114. the Fire and the Genus Homo symposium is gratefully ac- Delluc, B., and G. Delluc. 1978. Les manifestations graphiques Aurignaciennes knowledged. We express our special thanks to the symposium sur support Rocheux des environs des Eyzies (Dordogne). Gallia Préhistoire organizers Dennis Sandgathe and Francesco Berna and to Leslie 21(1/2):213–438. Delporte, H. 1968. L’abri du facteur à Tursac (Dordogne). Paris: Presse du CNRS. Aiello and Laurie Obbink. The research described here was Didon, L. 1911. L’Abri Blanchard des Roches. Bulletin de la Société Historique made possible by the generosity and good will of site owners et Archéologique du Périgord 87:246–241, 321–345. Jean-Max Touron (Cellier) and Isabelle Castanet-Daumas (Cas- Dunbar, R. 2014. How conversations around campfires came to be. Proceedings of the National Academy of Sciences of the USA 111:14013–14014. tanet, Blanchard, La Souquette). Fernández Peris, J., V. Barciela González, R. Blasco, F. Cuartero, H. Fluck, P. Sañudo, and C. Verdasco. 2012. The earliest evidence of hearths in southern Europe: the case of Bolomor Cave (Valencia, Spain). Quaternary Interna- tional 247:267–277. References Cited Gamble, C. 1991. An introduction to the living spaces of mobile peoples. In Ethnoarchaeological approaches to mobile campsites: hunter-gatherer and Aldeias, Vera. 2017. Experimental approaches to archaeological fire features pastoralist case studies. C. Gamble and W. A. Boismier, eds. Pp. 1–23. Ann and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– Arbor, MI: International Monographs in Prehistory. S205. Hahn, J. 1988. Die Geissenklösterle-Höhle im Achtal bei Blaubeuren: Fund- Alperson-Afil, N. 2008. Continual fire-making by hominins at Gesher Benot horizontbildung und Besiedlung im Mittlelpaläolithikum und im Auri- Ya‘aqov, Israel. Quaternary Science Reviews 27:1733–1739. gnacien. Stuttgart: Theiss. Anderson, L., M. Lejay, J.-Ph. Brugal, S. Costamagno, C. Heckel, M. De Heckel, C., and S. Wolf. 2014. Ivory debitage by fracture in the Aurignacian: Araujo Igreja, J.-V. Pradeau, et al. Forthcoming. Insights into Aurignacian experimental and archaeological examples. Journal of Archaeological Sci- daily life and camp organization: the open-air site of Régismont-le-Haut. ence 42:1–14. Quaternary International. Karkanas, P., R. Shahack-Gross, A. Ayalon, M. Bar-Matthews, R. Barkai, A. Binford, Lewis R. 1998. Hearth and home: the spatial analysis of ethno- Frumkin, A. Gopher, and M. C. Stiner. 2007. Evidence for habitual use of graphically documented rock shelter occupations as a template for dis- fire at the end of the Lower Paleolithic: site-formation processes at Qesem tinguishing between human and hominid use of sheltered space. In Middle Cave, Israel. Journal of Human Evolution 53:197–212. Palaeolithic and Middle Stone Age settlement systems. N. J. Conard and F. Lecoanet, H., F. Léveque, and J. P. Ambrosi. 2003. Combination of magnetic Wendorf, eds. Pp. 229–239. Forli, Italy: ABACO. parameters: an efficient way to discriminate soil contamination sources Bon, F. 2002. L’Aurignacien entre mer et océan: réflexion sur l’unité des phases (south France). Environmental Pollution 122:229–234. anciennes de l’Aurignacien dans le sud de la France. Mémoire de la Société Lecoanet, H., F. Léveque, and S. Segura. 1999. Magnetic susceptibility in Préhistorique Française 29. Paris: Société préhistorique française. environmental applications: comparison of field probes. Physics of the Earth Bon, F., and R. Mensan (with the collaboration of Marina de Araujo Igreja, and Planetary Interior 115:191–204. Sandrine Costamagno, Philippe Gardère, Clément Ménard, Farid Sellami, Legge, A. J. 1972. Cave climates. In Papers in economic prehistory. E. Higgs, Carolyn Szmidt, and Isabelle Théry-Parisot). 2007. Le site de plein air de ed. Pp. 97–103. Cambridge: Cambridge University Press. Régismont-le-Haut: une halte Aurignacienne dans les Plaines du Languedoc. Lejay M., M. Alexis, K. Quénéa, F. Sellami, and F. Bon. 2016. Organic signatures In Qui est l’Aurignacien? Pp. 53–71. Aurignac: Musée-Forum. of fireplaces: experimental references for archaeological interpretations. Bourrillon, R., R. White, E. Tartar, L. Chiotti, R. Mensan, A. Clark, J.-C. Castel, Organic Geochemistry 99:67–77. et al. Forthcoming. A new discovery of Aurignacian art at Abri Blanchard Marquer, L. 2010. From microcharcoal to macrocharcoal: reconstruction of (Dordogne): implications for understanding Aurignacian graphic expression in the “wood charcoal” signature in Paleolithic archaeological contexts. In The Western and Central Europe. Quaternary International.doi:http://dx.doi.org taphonomy of burned organic residues and combustion features in archae- /10.1016/j.quaint.2016.09.063. ological contexts. I. Théry-Parisot, L. Chabal, and Costamagno S., eds. Bouvier, J.-M. 1967. Topographie et climatologie des Abris Paléolithiques Palethnologie 2:105–115. dans la Region des Eyzies. In Comptes Rendus de la VIIe Congres National Marquer, L., V. Lebreton, T. Otto, and E. Messager. 2015. Étude des macro- et de Speleologie, Bordeaux 28–30 mai 1966. Spelunca 4e séries. Mémoires des micro-charbons du Site Epigravettien de Mezhyrich (Ukraine): données no. 5:27–31. taphonomiques et anthracologiques. L’Anthropologie 119:487–504. Brodard, A., Delphine Lacanette-Puyo, Pierre Guibert, François Lévêque, Marquer, L., V. Lebreton, T. Otto, H. Valladas, P. Haesaerts, E. Messager, D. Albane Burens, and Larent Carozza. 2016. A new process of reconstructing Nuzhnyi, and S. Péan. 2012. Charcoal scarcity in Epigravettian settlements archaeological fires from their impact on sediment: a coupled experimental with mammoth bone dwellings: the taphonomic evidence from Mezhyrich and numerical approach based on the case study of hearths from the Cave of (Ukraine). Journal of Archaeological Science 39:109–120. Les Fraux (Dordogne, France). Archaeological and Anthropological Sciences Marquer, L., T. Otto, R. Nespoulet, and L. Chiotti. 2010. A new approach to 8(4):673–687. study the fuel used in hearths by hunter-gatherers at the Upper Palaeolithic S302 Current Anthropology Volume 58, Supplement 16, August 2017

site of Abri Pataud (Dordogne, France). Journal of Archaeological Science Stapert, Dick. 1989. The ring and sector method: intrasite spatial analysis of 37:2735–2746. Stone Age sites, with special reference to Pincevent. Palaeohistoria 31:1–57. Miller, C., N. Conard, P. Goldberg, and F. Berna. 2010. Dumping, sweeping Surovell, T. A., and N. Waguespack. 2007. Fulsome hearth-centered use of and trampling: experimental micromorphological analysis of anthropoge- space at Larger Gulch, Locality B. In Emerging frontiers in Colorado Paleo- nically modified combustion features. In The taphonomy of burned organic indian archaeology. R. S. Brunswig, and B. Pitblado, eds. Pp. 219–259. Boulder: residues and combustion features in archaeological contexts. I. Théry-Parisot, L. University of Colorado Press. Chabal, and Costamagno S., eds. Palethnologie 2:25–37. Tartar, E., and R. White. 2013. The manufacture of Aurignacian split-based Movius, H. L., Jr. 1966. The hearths of the Upper Perigordian and Auri- points: an experimental challenge. Journal of Archaeological Science 40: gnacian horizons at the Abri Pataud, Les Eyzies (Dordogne), and their 2723–2745. possible significance. American Anthropologist 68(2):296–325. Théry-Parisot I. 2001. Economie des combustibles au Paléolithique: expéri- Naji, S., L. Gourichon, and W. Rendu. 2015. La cémentochronologie. In mentation, taphonomie, anthracologie. Paris: CNRS Editions. Messages d’os: archéométrie du squelette animal et humain. M. Balasse, J.-P. ———. 2002. Fuel management (bone and wood) during the Lower Aurignacian Brugal, Y. Dauphin, E.-M. Geigl, C. Oberlin, and I. Reiche, eds. Pp. 217– in the Pataud Rock Shelter, Les Eyzies de Tayac, Dordogne, France: contri- 240. Paris: Sciences Archéologiques, Edition des Archives Contemporaines. bution of experimentation. Journal of Archaeological Science 29:1415–1421. O’Hara, J. F., R. White, Z. S. Garrett, T. Higham, and A. Roussot. 2015. The Théry-Parisot, I., S. Costamagno, J.-P. Brugal, P. Fosse, and R. Guilbert. 2002. Aurignacian site of the Abri de la Souquette (Commune de Sergeac, The use of bone as fuel during the Palaeolithic: experimental study of bone Dordogne): a history of archaeology. In Aurignacian genius: art, technology combustible properties. In The zooarchaeology of fats, oils, milk and dairy- and society of the first modern humans in Europe. R. White and R. Bourrillon, ing. J. Mulville and A. K. Outram, eds. Pp. 50–59. Oxford: Oxbow. eds. Palethnologie 7:98–117. Vaquero, M., and I. Pastó. 2001. The definition of spatial units in Middle Peyrony, D. 1935. Le Gisement de Castanet, Vallon de Castelmerle, Commune Palaeolithic sites: the hearth-related assemblages. Journal of Archaeological de Sergeac (Dordogne). Bulletin de la Société préhistorique française 32:418– Science 28:1209–1220. 443. Villa, P., F. Bon, and J.-C. Castel. 2002: Fuel, fire and fireplaces in the Palaeo- ———. 1946. Le Gisement Préhistorique de l’Abri Cellier, au Ruth, Com- lithic of Western Europe. Review of Archaeology 23(1):33–42. mune de Tursac (Dordogne). Gallia 4(1):294–301. White, R. 1983. Upper Paleolithic land use in the Périgord: a topographic ap- Pike-Tay, A. 1995. Variability and synchrony of seasonal indicators in dental proach to subsistence and settlement. British Archaeological Reports, Inter- cementum microstructure of the Kaminiriak caribou population. Archaeo- national Series 253. Oxford: BAR. fauna: International Journal of Archaeozoology 4:273–284. ———. 1985. The influence of solar aspect on Upper Paleolithic site location. ———. 1998. Dental growth mark analysis of a Rangifer tarandus sample New York University Journal of Anthropology 1:1–7. from Abri Castanet. In Rapport de Fouille Programmée Triannuelle 1995– ———. 1992. The history and research significance of the Logan Museum’s 1998, sec. A5. Bordeaux: Service Régional de l’Archéologie. French Paleolithic collections. In French Paleolithic collections in the Logan Rendu, W. 2007. Planification des activités de subsistance au sein du territoire Museum of Anthropology, Beloit College. R. White and L. Breitborde, eds. des derniers Moustériens cémentochronologie et approche archéozoo- Bulletin of the Logan Museum of Anthropology, n.s., 1(2):1–38. logique de gisements du Paléolithique moyen (Pech-de-l’Azé I, La Quina, White, R., R. Bourrillon, R. Mensan, A. Clark, L. Chiotti, T. Higham, S. Mauran) et Paléolithique supérieur ancien (Isturitz). Doctoral dissertation, Ranlett, E. Tartar, A. Morala, and M.-C. Soulier. Forthcoming. Newly Université Bordeaux 1. discovered Aurignacian engraved blocks from Abri Cellier: history, context ———. 2010. Hunting behavior and Neanderthal adaptability in the Late and dating. Quaternary International. doi: http://dx.doi.org/10.1016/j.quaint Pleistocene site of Pech-de-l’Azé I. Journal of Archaeological Science 37 .2017.02.001. (8):1798–1810. White, R., and H. Knecht. 1992. Abri Cellier (or la Ruth), Commune de Rigaud, J.-P., and J. F. Simek. 1991. Interpreting spatial patterns at the Grotte Tursac (Dordogne): results of the 1927 Beloit College excavations. In XV: a multiple-method approach. In The interpretation of archaeological spa- French Paleolithic collections in the Logan Museum of Anthropology, Beloit tial patterning. E. M. Kroll and T. D. Price, eds. Pp. 199–220. New York: College. R. White and L. Breitborde, eds. Bulletin of the Logan Museum of Plenum. Anthropology, n.s., 1(2):39–96. Roebroeks, W., and P. Villa. 2011. On the earliest evidence for habitual use of White, R., R. Mensan, R. Bourrillon, C. Cretin, T. Higham, A. E. Clark, M. fire in Europe. Proceedings of the National Academy of Sciences of the USA Sisk, et al. 2012. Context and dating of Aurignacian vulvar representations 108:5209–5214. from Abri Castanet, France. Proceedings of the National Academy of Sciences Roussot, A. 1982. Abri de la Souquette. Gallia Préhistoire 25:112–114. of the USA 1089:8450–8455. Schiegl, S., P. Goldberg, H. U. Pfretzschner, and N. J. Conard. 2003. Paleo- Wiessner, Polly W. 2014. Embers of society: firelight talk among the Ju/ lithic burnt bone horizons from the Swabian Jura: distinguishing between ’hoansi Bushmen. Proceedings of the National Academy of Sciences of the in situ fireplaces and dumping areas. Geoarchaeology 18:541–565. USA 111:14027–14035. Stahlschmidt, M. C., C. E. Miller, B. Ligouis, U. Hambach, P. Goldberg, F. Wrangham, R., and R. Caromed. 2010. Human adaptation to the control of Berna, D. Richter, B. Urban, J. Serangeli, and N. J. Conard. 2015. On the fire. Evolutionary Anthropology 19(5):187–199. evidence for human use and control of fire at Schöningen. Journal of Wrangham, Richard. 2017. Control of fire in the Paleolithic: evaluating the Human Evolution 89:1–21. cooking hypothesis. Current Anthropology 58(suppl. 16):S303–S313. Current Anthropology Volume 58, Supplement 16, August 2017 S303

Control of Fire in the Paleolithic Evaluating the Cooking Hypothesis

by Richard Wrangham

According to current evidence, Homo sapiens was unable to survive on a diet of raw wild foods. Because cooked diets have large physiological and behavioral consequences, a critical question for understanding human evolution is when the adaptive obligation to use fire developed. Archaeological evidence of fire use is scarce before ca. 400 ka, which suggests to some that the commitment to fire must have arisen in the mid-Pleistocene or later. However, weak jaws and small teeth make all proposals for a raw diet of early Pleistocene Homo problematic. Furthermore, the mid-Pleistocene anatomical changes seem too small to explain the substantial effect expected from the development of cooking. Here I explore these and other problems. At the present time no solution is satisfactory, but this does not mean the problem should be ignored.

In this paper I consider the current status of the cooking hy- melmitz et al. 2014). In Qesem, by contrast, the evidence of fire pothesis. I use “cooking” to mean the processing of food with use is impressive back to 420 ka (Barkai et al. 2017). heat. The cooking hypothesis posits that control of fire leads to One interpretation of such cases is that absence of evidence such a large increase in energy acquisition and reduces the phys- really is evidence of absence. The implication is that popula- ical challenges of eating food so greatly that the evolution of an tions of Homo occupied Europe during the Lower Paleolithic obligation to incorporate cooked food into the diet should be and later without the systematic control of fire, surviving on recognizable by evidence of novel digestive adaptations and raw food for hundreds of thousands of years (Roebroeks and increased energy use. It also claims that the only time in the Villa 2011; Shimelmitz et al. 2014). Alternatively, the archae- fossil record when the appropriate changes are seen is the early ological visibility of fire may vary too much to allow the history Lower Paleolithic (Parker et al. 2016; Wrangham 2006; Wrang- of its control to be confidently reconstructed. For example, Box- ham and Carmody 2010; Wrangham et al. 1999). grove was a damp shoreline site that might have been un- The evidence for Lower Paleolithic control of fire has been suitable for locating fire. Changes in the style of fire use over increasing in the last decade (Alperson-Afil 2008; Berna et al. time could have biased preservation. Middle Paleolithic fire 2012; Bosinski 2006; Walker et al. 2016). Nevertheless, nu- sites might have been larger, or more permanent, or sited more merous archaeological sites before the Upper Paleolithic chal- often in caves than in earlier times (Gowlett and Wrangham lenge the cooking hypothesis because they find no evidence for 2013). For these reasons the debate over the meaning of Homo the control of fire. European Lower Paleolithic sites such as sites that lack any evidence of fire is unresolved, and I will not Dmanisi, Atapuerca, La Caune d’Arago, and Boxgrove repre- discuss it here. sent in total a screening of thousands of unburned bones but I consider instead some conceptual and empirical challenges no burned bones (Gowlett and Wrangham 2013). Even in the arising from the assumption that cooked food did not become Middle Paleolithic many Neanderthal sites have no fire evi- obligatory until the mid-Pleistocene. The evolution of Homo dence (Sandgathe et al. 2011a, 2011b). Outside Europe, the few erectus is often discussed without mentioning the control of cases where preservation conditions allow long sequences also fire (e.g., Antón and Snodgrass 2012; Hardy 2009; Isler and offer puzzles. Tabun and Qesem caves lie near the Mediter- van Schaik 2012; Potts 2012). The implication is that either a ranean coast about 100 km apart. In Tabun, cave fire use was later time can be confidently found for the origin of obligatory “regular or habitual” by 350–320 ka, but burned flints were fire use, or fire control has only small effects on human adap- scarce or absent during at least 50 ka of prior occupations (Shi- tation, or both. But those implications are rarely considered carefully. I suggest that they are wrong. Here, therefore, I assess the difficulty of understanding a Richard Wrangham is Professor in the Department of Human Evo- date later than the origin of H. erectus for when cooking might fi lutionary Biology at Harvard University (11 Divinity Avenue, Cam- have become obligatory. In the rst section I review the current fi bridge, Massachusetts 02138, USA [[email protected]]). This status of the cooking hypothesis. I then consider some dif - paper was submitted 25 VII 16, accepted 5 III 17, and electronically culties that follow from the assumption that H. erectus could not published 25 V 17. cook. Finally, I consider problems for the cooking hypothesis. q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0013$10.00. DOI: 10.1086/692113 S304 Current Anthropology Volume 58, Supplement 16, August 2017 Current Status of the Cooking Hypothesis support this prediction for all three major macronutrients, that is, long-chain carbohydrates, proteins, and (plant) lipids (Car- Adaptation to Cooked Diets mody, Weintraub, and Wrangham 2011; Groopman, Carmody, The cooking hypothesis starts with the claim that unlike other and Wrangham 2015). On raw diets (sweet potato tuber, meat, animals, Homo sapiens has evolved an obligation to include or peanuts) mice typically lose weight, whereas on cooked diets cooked food in the diet such that we cannot live without it. The they maintain weight. idea has been intimated for at least half a century (Brace 1995; One reason is that digestibility is increased. With regard to Coon 1962; Symons 1998) and directly supported more re- starch, ileostomy experiments with five domesticated plants cently (Wrangham and Conklin-Brittain 2003). (oat, wheat, plantain, green banana, potato) eaten by humans Key evidence comes from research on raw foodists, that is, indicate that cooking raises starch digestibility in the small people who live for long periods on all-raw diets. Raw-foodist intestine by amounts that vary across foods from 28% to 109%. groups typically live in industrial societies on store-bought On the conventional assumption that because of fermentation foods. Even though raw foodists take little exercise compared by the microbiome, 50% of starch energy is recovered from the with hunter-gatherers and have fewer disease challenges, on colon, the median increase in net energy gained is 30.2% (Car- average they experience chronic energy shortage leading to low mody and Wrangham 2009). body mass index (BMI). In the only study of reproductive Schnorr et al. (2015) and Henry (2017) noted that the effects performance, incompetent or absent ovulation left more than of cooking in human evolution are best studied in wild plants 50% of women on an all-raw diet unable to reproduce (Koeb- roasted on open fires rather than in domesticated plants nick et al. 1999). cooked with contemporary methods. They assessed the ener- These physiological detriments are striking because the diet getic effect of brief roasting times (average 8.7 min) on four eaten by raw foodists is extremely high quality compared with species of tuber eaten by Hadza. Three of the species were any known hunter-gatherer diet (if their diet were eaten raw). normally cooked, and one was normally eaten raw. Schnorr Most of the raw foodists’ diet is rich in digestible energy and et al. (2015) found that such cooking led to much variation in low in structural fiber because it comes from domesticated spe- the in vitro starch digestibility both within and between spe- cies. Furthermore, raw foodists typically process the food ex- cies. In two species, cooking led to a 10% increase in glucose tensively by nonthermal means (such as by blending) and (in availability, whereas in the species typically eaten raw (Ipo- spite of their supposed adherence to raw) often use heat to moea transvaalensis), glucose absorption fell after roasting. lightly cook (up to around 1147F). In addition, raw foodists ex- Schnorr et al. (2015) concluded that starch gelatinization is not perience no important seasonal energy shortages (because they necessarily a route by which cooking increases net energy gain. buy from globally connected markets; Wrangham 2009). Instead, after finding that brief periods of roasting made tu- Low meat intake does not account for the problems faced by bers much easier to chew, they suggested that cooking could raw foodists. No relationship has been found for raw foodists increase net energy gain by making chewing more efficient. between BMI and the amount of meat eaten (Koebnick et al. Cooking also makes tubers easier to peel, which Henry (2017) 1999). Unlike meat-eating raw foodists, vegetarians eating suggested was its main benefit. Such experiments are prom- cooked diets thrive. They have high BMI and excellent ovarian ising, but as Schnorr et al. (2015) noted, they are at an early function (Barr 1999; Rosell, Appleby, and Key 2005). Cooking stage because they have considered only glucose, they have not is thus more important than meat for human welfare. taken account of variation in quality within food species, and On the basis of contemporary evidence, dogs (Canis fam- they rely on in vitro measures of digestibility. iliaris) are the only nonhuman species that might be adapted to Other energy benefits of cooking have been less well quan- cooked food given that their pancreatic amylases are adapted tified. They include a reduced cost of digestion (snakes: Bo- to a high-starch diet (Axelsson et al. 2013; Reiter, Jagoda, and back et al. 2007; rats: Carmody 2012), shorter and less vigorous Capellini 2016). However, it is unknown whether the relation- chewing times (humans: Zink, Lieberman, and Lukas 2014), ship between dogs and cooked food is obligatory. Possibly dogs increased chewing efficiency (Dominy et al. 2008), and reduced are merely able to take advantage of a cooked diet and can sur- investment in immune defenses when meat is eaten (Carmody vive without it if needed (as dingoes [Canis dingo]do;White- and Wrangham 2009; Carmody et al. 2016). In addition, cooked house 1977). So with the possible exception of some domesti- food is more quickly digested than raw food, which means that cated species, H. sapiens is unique among living species by being if extra food is available, the total rate of energy acquisition per evolutionarily restricted to diets that contain a substantial pro- day can be higher. portion of cooked food (Wrangham and Carmody 2010). In the case of plant lipids, a principal mechanism by which cooking increases digestibility concerns the common way in which they are stored in oilseeds. Oilseed lipids are stored in oil bodies surrounded by proteins called oleosins that present a How Does Cooked Food Benefit Consumer Physiology? hydrophobic interior surface and a hydrophilic exterior. Cook- Evidence of BMI reduction and reproductive problems among ing denatures the oleosins and thereby makes the lipids avail- raw foodists indicates that cooked food provides higher net able for faster digestion (Groopman, Carmody, and Wrangham energy gain than raw food. Feeding experiments with mice 2015). Wrangham The Cooking Hypothesis S305

No tests of the effects of cooking on energy gain have yet leaves the men thin, it raises the question of whether African been conducted with animal lipids, which are mostly stored as populations of H. sapiens might have been able to survive on a droplets inside adipocytes or other cells. The lipid droplets of diet that was sufficiently focused on equivalently fat-rich and vertebrates and insects are coated with proteins of the perilipin easily chewed raw wild foods, such as oilseeds, marrow, brains, family (Arrese and Soulages 2010; Brasaemle 2007). The effect or guts. of heat on denaturation of perilipins appears to be unknown. In African habitats, seasonal variation is a problem for both The physical effect of heating solid fats to liquid forms or thin oilseeds and marrow (Speth 2010). Oilseeds can be vanishingly layers could be important in promoting release of lipids from scarce, while the fat content of herbivore marrow declines adipocytes. Furthermore, the thinning of fats into oils could from more than 90% dry weight at the best time of year to 2%– facilitate more rapid digestion given that digestive lipases are 3% at the worst (late dry and early wet seasons; Dunham and active only at oil-water interfaces, where they depend for their Murray 1982; Lupo 1998). So oilseeds and marrow are unre- effectiveness on lipids presenting a high surface area (Lentle liable year-round sources of energy (Speth 2010). and Janssen 2011). How much such effects matter is unknown. However, brain fats are never depleted, remaining around There are numerous ethnographic reports of eating raw ani- 50%–60% dry weight all year (Carlson and Kingston 2007; mal fat (Ben-Dor, Gopher, and Barkai 2016). Variation in fat Stiner 1994). So conceivably a sufficient abundance of brains type could be important. It might be more beneficial to cook could give modern humans the basis for surviving as wild raw lipids that have high melting points. foodists during food-scarce seasons. How much brain would The uncertainty about effects on animal-sourced lipids as be required is, therefore, an interesting question. If the diet well as limited experimental study for all macronutrients mean needs to be ∼50% fat (i.e., for an animal-based diet; Speth that the energetic effect of cooking is not well quantified. 2010), and brains are roughly 50% fat, then a near-exclusive Progress in solving this problem will be slowed by numerous diet of brains might be predicted. Availability of other foods sources of variation, including the physical states of the diet, rich in fats or easily digested carbohydrates would lessen the such as whether it is eaten cold or hot, blended or whole, mixed reliance on brains. or pure, and fresh or fermented. However, even though the Intestines, especially of ungulates, could also represent an energy gain from eating food cooked is poorly known, there important item of diet partly because semidigested chyme would are clear indications that the amount is large enough to have provide an easily accessed source of energy. Furthermore, they major fitness consequences. The figure of 30.2% increase for are readily eaten by Hadza (Buck et al. 2016). starch may well be on the low side because it is derived solely In sum, there is no evidence of modern H. sapiens popula- from considering digestibility, excluding reduced costs of di- tions surviving on raw foods, but possible diets allowing earlier gestion, increased food safety, and reduced time digesting. H. sapiens to have lived without fire require easily chewed and Effects of cooking a starch-rich food appear broadly similar to fat-rich foods, of which brains and guts are the most plausible. those for cooking protein-rich food (meat) and lipid-rich food But these scenarios are speculative. Despite cooking their main (nuts; Groopman, Carmody, and Wrangham 2015). Thus, a meals, even well-adapted Arctic populations living on high-fat 30% gain in energy seems a reasonable starting point for con- diets suffer extreme food shortages (Balikci 1989; Hardy et al. sidering the effect of cooking. 2015). It is doubtful that they would survive if they were forced Much smaller increases in energy have large fitness effects in to eat all their food raw. the wild. When chimpanzees of the Kanyawara community in Kibale National Park, Uganda, were able to eat 5% more fruit Why Does Homo sapiens Find It So Difficult to Thrive in their diet than usual (thanks to improved fruit availability), on Raw Foods? they experienced a 4-month reduction in waiting time to con- ception (Emery Thompson and Wrangham 2008). Numerous The inferred inability of H. sapiens to survive on raw wild diets similar examples in large nonhuman primates indicate that even is explicable by a series of adaptations that have apparently modest increases in net energy gain have substantial positive promoted efficiency in processing easily digested foods at the effects on fitness. The estimated 30% increase in calorie gain is expense of being able to process relatively indigestible foods. therefore comparatively massive. It is also exceptional because The best-known adaptations are the diminution of the human compared with many changes in diet (e.g., addition of fat-rich mouth, jaw muscles, jaw, incisors, molars, stomach, cecum, and meat), it works both during periods of food scarcity and during colon compared with those of nonhuman primates (Martin periods of abundance. et al. 1985; Perry et al. 2015; Ungar 2012; Wrangham 2009). Overall, Aiello and Wheeler (1995) estimated that the human gut is 60% of the expected size for a primate. Note that Hladik, Could Homo sapiens Live without Cooked Food in the Wild? Chivers, and Pasquet (1999) presented data to claim that the No human populations are known to live without cooking, but human gut was the same size as expected in a nonhuman pri- might it be possible? Luke Glowacki (personal communication) mate of the same body size. However, their data were solely for observed that unmarried Nyangatom men in cattle camps in absorptive mucosa, that is, the small intestine. In humans the southwestern Ethiopia sometimes spend several weeks or a few small intestine is indeed the size expected by body mass, whereas months living solely on raw blood and milk. While this diet the big reductions are in the cecum and colon (Martin et al. S306 Current Anthropology Volume 58, Supplement 16, August 2017

1985). These reductions appear to be responsible for humans marked by an increase in total energy expenditure (e.g., Antón having a relatively small intake of dry weight of food per day and Snodgrass 2012). Importantly, there is no subsequent time compared with nonhuman primates (Barton 1992) and being in human evolution when a marked increase in energy use has relatively poor at digesting long-chain carbohydrates (Milton been suggested. Because cooking is strongly associated with an and Demment 1988). Presumably the reductions could not have increase in energy gain, it is therefore predicted to have been evolved until consumers had consistent (year-long) access to adopted by H. erectus (Wrangham 2006). Paleoanthropolo- appropriately calorie-dense, easily chewed, and easily digested gists routinely attribute increased energy use by H. erectus to foods. increased reliance on animal-sourced foods (e.g., Zink and Lie- Numerous parallel adaptations to cooked food can be ex- berman 2016). However, contemporary evidence indicates that pected in physiological traits, such as in digestive enzymes. A cooking has a much greater effect on energy gain than meat as leading candidate is the amylase system, which has been pu- shown by the robust performance of vegetarians eating cooked tatively associated with increased consumption of starch (Perry food compared with meat-eating raw foodists (see “Adapta- et al. 2007), specifically cooked starch (Hardy et al. 2015). Com- tion to Cooked Diets”). pared with chimpanzees and bonobos, humans have been found Changes in H. erectus anatomy are critical for reconstruct- to have a three or more times increase in copy number of the ing diet. Homo erectus digestive anatomy is known from teeth salivary amylase gene AMY1, which appears responsible for and jaws and inferred from rib cage and pelvis. Homo erectus levels of salivary amylase protein being at least three times as incisors and molars were markedly smaller than in Homo high as in Pan (Perry et al. 2007). No increase in salivary am- habilis, especially the third molar (Ungar 2012). Their jaws ylase copy number compared with Pan was found for Nean- were similar in absolute size to H. habilis (Antón and Snod- derthals or Denisovans (Perry et al. 2015). In theory this could grass 2012), which means that in relation to body mass, the jaw mean that adaptation to cooked diets occurred in the Homo of H. erectus was relatively small (Wood and Collard 1999; lineage after the splits from Neanderthals and Denisovans. body mass estimates: australopithecines 36–44 kg, H. habilis However, only one individual has been characterized for each 34 kg, H. erectus 57 kg). of the latter species, and it is not known whether duplication of Aiello and Wheeler (1995) proposed that H. erectus also amylase is associated with increased starch consumption (Car- experienced a major reduction in the size of the gut based on a penter et al. 2015). Even if it were, whether the amylase dif- reconstruction by Schmid (1983) of the rib cage of H. erectus as ference between Pan and H. sapiens is associated with cooking being barrel shaped compared with more bell shaped in aus- or only with an increase in the starch component of the diet is tralopithecines and living great apes: the flared shape is thought unknown (Perry et al. 2015). More information is therefore to allow a large intestinal capacity below the ribs. In addition, needed to make the amylase system informative about the the pelvis was seen as relatively narrow compared with earlier prehistory of cooking. hominins, suggesting a small intestinal floor supporting what A different approach to investigating physiological adapta- was therefore considered to be a relatively small gut compared tions to cooking uses gene expression. Carmody et al. (2016) with previous species. These points supported the idea that H. found that genes expressed in the livers of mice eating cooked erectus acquired a small gut at about the same time as getting food have been under positive selection in H. sapiens, Nean- smaller teeth, mouth, and jaws. derthals, and Denisovans. The cooking-related genes were as- However, subsequent analysis has changed the reconstruc- sociated with lipid-related metabolic processes on meat diets tion of the H. erectus pelvis because H. sapiens now appears to and with carbohydrate-metabolic processes on tuber diets. While be the only species of Homo with a reduced pelvis compared this research has not identified a specific cooking-related phe- with australopithecines (Holliday 2012). This means either notype, it suggests that human adaptation to a cooked diet oc- that the size of the pelvis is less informative about the volume curred before the evolution of Neanderthals and Denisovans, of the gut than Aiello and Wheeler (1995) suggested or that the that is, before 550–765 ka (Prüfer et al. 2014). However, similar Homo gut remained large until the evolution of H. sapiens. For cautions apply as to the amylase studies. the moment I assume the former conclusion, that is, the pelvis The question arises as to why Neanderthals would have had is not well correlated with gut size (see “There May Have Been “cooking genes” given evidence that they sometimes lived Important Variation in Gut Size within Post-Habiline Homo”). without using fire (Henry 2017). One possibility is that Nean- The idea that these changes to the jaws, teeth, and guts were derthals cooked sufficiently often to maintain their genetic adap- the result of a dietary improvement is universal, and the lead- tation. Another is that the genes were retained from a cooking- ing candidate diet has long been increased animal-sourced foods, dependent ancestor despite a low frequency of cooking. especially high-fat meat and marrow. However, the animal- food idea faces a major difficulty. Dietary adaptations must be relevant not only to preferred foods but also to fallback foods, that is, those eaten during periods of food scarcity. Such fall- What Diet Type Shaped the Digestive Anatomy back periods occur frequently, approximately every year, re- of Homo erectus? gardless of habitat: they are found even in rainforests for great The evidence of predictable high-energy gain from eating apes (Marshall and Wrangham 2007) and are well documented cooked food is helpful because the evolution of H. erectus was for African dry-country hunter-gatherers (Speth 2010). Wrangham The Cooking Hypothesis S307

So the difficulty is to understand what an importantly car- Humans benefitbyfinding honey more quickly, while honey- nivorous H. erectus would have eaten during the inevitable guides benefit by feeding on otherwise unattainable products periods when animal products were inadequate. Presumably (Spottiswoode, Begg, and Begg 2016). African hunter-gatherers H. erectus incorporated some plant matter in their diets just use smoke to quell the bees’ defensive response, suggesting an as recent hunter-gatherers do. But unlike hunter-gatherers, if ancient control of fire. However, this proposal has been weak- H. erectus did not control fire, they would have had to eat their ened by Kraft and Venkatamaran ’s (2015) finding that some plants raw. Because neither molars nor guts indicate an ability populations of honey collectors use plant volatiles rather than for H. erectus to utilize raw plants high in structural fiber, this smoke to control bees. So collecting Apis honey in the Paleo- makes no sense. Either a solution must be found to this prob- lithic might not have depended on controlling fire. lem or H. erectus had to cook. Two noncooking solutions are worth considering. First, the reduced digestive system could have been made Problems in Understanding Adaptations of Homo possible by a commitment to fat eating as implied by Ben-Dor, erectus If They Were Limited to Raw Food Gopher, and Barkai (2011). As discussed for H. sapiens, prey Scavenging on Raw Food Would Be Difficult Except brains would have been one of the few sources of fat in fallback for Marrow or Brains seasons for H. erectus. This hypothesis should eventually be testable by the fossil record. Eating of meat and marrow is evidenced back to 2.5 million Second, H. erectus might have physically processed plants years ago (de Heinzelin et al. 1999) and arguably to 3.3 million (probably underground or underwater storage organs [USOs]) years ago (McPherron et al. 2010; but see Domínguez-Rodrigo before consumption in ways that allowed teeth to be functional and Alcalá 2016). However, only by 1.8 million years ago do despite being small and that allowed fermentation to occur hominin sites indicate regular butchering of large animals fi without a large colon. Zink et al. (2014) and Zink and Lieber- (Potts 2012). This suggests that H. erectus was the rst species man (2016) proposed this scenario after showing that mechan- to rely extensively on access to animal foods and raises the ical tenderization decreased the toughness of tubers by 42%. question of how they escaped a high risk of disease from path- Against the proposed importance of physical processing, ogens (Ragir, Rosenberg, and Tierno 2000). One answer would however, mechanical tenderization has limited benefits to be a focus on eating marrow, which has a low bacterial load as a judge from the fact that raw foodists suffer energy deficiencies result of being protected inside bone (Smith et al. 2015). Brains despite using electrical blenders to produce smoothies. Fur- are probably similarly safe. thermore, no ethnographic or primate models seem to be Nevertheless, cut marks show that edible meat portions were known of physically processing a food for consumption raw. also commonly removed. Careful attention paid by the butch- Contemporary foragers sometimes use techniques such as ham- ers to identifying dangerously infected sections of meat would mering (to extract an edible seed), but the seed itself is not have helped reduce the dangers of eating raw wild meat, but smashed unless it is going to be cooked. These arguments cooking would still have been a safer strategy (Smith et al. mean that H. erectus was unlikely to have been able to live off 2015). raw plant foods even if they were mashed. Accordingly, even if H. erectus had an ape-sized gut, the puzzle remains of why Hunting Effort Would Be Mysterious (Time Would teeth, jaws, jaw muscles, and mouth size became reduced. Be Constrained If Food Is Raw) Although exploitation of meat and marrow can lead to high Did Homo erectus Have the Cognitive Ability to Cook? gains, nowadays those foods are not dependable as a source of Warneken and Rosati (2015) have shown that chimpanzees calories for tropical hunter-gatherers because the amount of have the cognitive ability to understand the transformative ef- animal food obtained on any given day is often inadequate fects of cooking and sufficient inhibition to carry food to a (Speth 2010). This suggests that the same would have been true cooker rather than eat it unripe. This indicates that H. erectus, in the Lower Paleolithic. Accordingly, individuals who invested with a presumably greater understanding of cause-effect rela- in trying to hunt or find carcasses would sometimes fail and tionships and more inhibitory ability, was plausibly able to would therefore need to have an alternative source of food that cook. day. The problem is easily solved in modern humans: when an entire group runs short of animal products, they are able to eat starchy plant foods that are consumable quickly because they Use of Honey have been cooked. Plant foods are generally less satisfying than The symbiotic relationship between humans and the greater animal foods, but their merit is that they are more easily ob- honeyguide bird (Indicator indicator) has been proposed to tained and are, therefore, relatively predictable. However, the depend on a long evolutionary history of controlling fire (Crit- value of plant foods depends on their being cooked not only tenden 2011; Marlowe et al. 2014; Wrangham 2011). Honey- because cooked plants provide more energy than raw plants guides are genetically adapted to leading humans toward Apis but also because they can be eaten much faster (perhaps in mellifera beehives, from which humans extract honeycomb. 10%–20% of the time needed for raw plants; Organ et al. 2011). S308 Current Anthropology Volume 58, Supplement 16, August 2017

Without access to cooked food, failed hunters would need ised on animal foods being more energetically efficient than many hours to chew and digest plant foods that would provide plant foods. In favor of this explanation, among carnivores a their daily energy requirements. This means that a significant diet richer in vertebrate animals is correlated with increased increase in the amount of meat eaten, as occurred most clearly relative brain size (Swanson et al. 2012). Against it, allometric with H. erectus, depended on having cooked food as a sub- variation in carnivore brain size is explained principally by stitute food resource on days when no animal-source foods between-family differences (Finarelli and Flynn 2009). Within were obtained (Wrangham 2009). families, the most striking variation in eating vertebrates is found in the Ursidae, from the vegetarian giant panda Ailu- ropoda melanoleuca to the vertebrate-eating polar bear Ursus Brain Size Increase in Homo erectus Would Be maritimus. If meat products are important for brain growth, Challenging to Understand polar bears should have relatively large brains. But although Aiello and Wheeler (1995) recognized two major rises in hu- larger bear species eat more vertebrates, the slope of brain man brain size (cf. Rightmire 2004). The first was around volume on body mass is the same in bears as in basal Carnivora. 2 million years ago, including H. habilis/rudolfensis and H. er- So bears have larger brains than other carnivores without any gaster, which they attributed to increased meat eating. The sec- evidence that a diet incorporating more meat influences this ond was in the latter half of the Middle Pleistocene. They sug- relationship (Finarelli and Flynn 2009). gested that cooked food could have been an important factor If cooked food explains the rise in brain size in H. erectus, in the second rise. what explains the later, mid-Pleistocene grade shift in brain If the mid-Pleistocene increase in cranial capacity was due size (Rightmire 2004)? One possibility is that developments in to hominin exploitation of cooked food, the fact that cooked hunting ability led to increased procurement of fat-rich prime food increases energy gain by 30% or more should be recog- adults compared with weakened, fat-depleted prey (Stiner, Bar- nizable in other changes as well. But in the mid-Pleistocene kai, and Gopher 2009). The interesting problem that remains there are no other significant signals of improved dietary qual- is how sufficient dietary fat could be maintained during sea- ity or any marked decline in tooth size that would be expected sons of food scarcity. to accompany an increasingly tender diet produced by cooking. This makes the idea of cooked food being introduced at that Running Would Not Be Favored without the Use of Fire time problematic. Fonseca-Azevedo and Herculano-Houzel (2012) drew at- Endurance running is a unique human ability compared with tention to a second difficulty by postulating that the positive other primates. Bramble and Lieberman (2004) argue that it is influence of cooked food on brain size was not achieved until made possible by various anatomical adaptations in H. sapiens the mid-Pleistocene (i.e., with Homo heidelbergensis). Based on that occur also in H. erectus but not earlier, including features extrapolation from living primates, they calculated the meta- that promote stabilization of head and trunk and energy stor- bolic cost of servicing bodies and brains of hominin species. age and shock absorption in the foot. Their proposals have Noting that a combination of large body and large brain is a been criticized on the basis that australopithecines may have difficult challenge because of the high maintenance costs of had the same adaptations, allowing them to be equally effective both, they found that if H. erectus ate only raw food, they endurance runners (Pontzer 2012). would be required to chew for as many hours per day as a Physiological adaptations for endurance running include gorilla, that is, up to 8 hours or more. Their calculation seems increased ability to lose heat compared with apes, such as longer to be an underestimate because it assumes that digestive ef- legs in relation to body mass (Pontzer 2012). Loss of body hair fectiveness was as high in H. erectus as in gorillas, which is would make a particularly important contribution to the abil- unlikely given the smaller teeth (and possibly smaller gut) of ity to lose heat. In support of an early loss of body hair (and the H. erectus. The demand of exceptionally high foraging time associated evolution of pubic hair), the human pubic louse (which would necessarily be associated with high resting time (Pthirus pubis) diverged an estimated 1.84–5.61 Ma from its to allow digestion to occur) appears to be impossible for a closest living relative, the gorilla louse (Pthirus gorillae; Reed species that supposedly had high travel distances (cf. Organ et al. 2007). However, a critical function of body hair is to et al. 2011). retain heat during sleep, which means that sufficient reduction Fonseca-Azevedo and Herculano-Houzel (2012) concluded of body hair to allow endurance running would seem to de- that H. erectus and subsequent species needed to eat cooked pend on a system of heat maintenance during sleep other than food in order to obtain enough calories per day to satisfy the an insulating layer of hair. combination of larger bodies and larger brains without spend- Clothes are one possibility, but parasite evidence indicates ing all day chewing. clothes were adopted relatively late. Thus, clothing appears to The major alternative is that Homo (starting presumably have been responsible for a functional split between head lice with H. habilis) fueled their larger brains thanks to a diet that and body lice (both Pediculus humanus), which are estimated included more raw meat and fat products of animals than to have been genetically separated since 83–170 ka (Perry 2014), before (Leonard, Snodgrass, and Robertson 2007). This is prem- long after the evidence for endurance running. Wrangham The Cooking Hypothesis S309

Other than clothing, the obvious way for a species with re- Wiessner indicated that predation risk on humans is sharply duced body hair to keep warm at night is to use fire (Wrang- higher when sleeping without fire than when fire is present ham 2009). This suggests that endurance running could not (Wrangham and Carmody 2010). have occurred without the loss of body hair made possible by In short, the pattern of sleep by large herbivores is not an the control of fire. argument against the claim that H. erectus would have been In sum, endurance running presumably depended on the very vulnerable if they slept on the ground unprotected by fire loss of an insulating layer of hair and, therefore, on non- or some alternative system. The proposal that H. erectus did insulated humans being able to warm themselves at night by not control fire, therefore, demands a novel explanation of their a fire. The evidence for Lower Paleolithic endurance running defenses at night, such as the evolution of an ungulate-like therefore suggests that fire was controlled by then. pattern of sleep, a surprisingly good ability to sprint, or the building of effective fences. Sleeping on the Ground Would Not Be Favored without the Use of Fire Problems for the Cooking Hypothesis The adaptations of H. erectus to terrestrial locomotion include reduced climbing ability. Having an essentially modern frame Why Were Population Densities So Low? with long legs, H. erectus cannot be expected to have been able Cooking presumably increased both the range of foods that to climb into trees every night to make a bed of leaves and twigs could be eaten and the total energy gained from those foods. in the way that most great apes do. They therefore presumably Yet although Homo achieved a wide geographical distribution slept on the ground (Coolidge and Wynn 2006). Gorillas reg- around the time of H. erectus, it appears to have been a rela- ularly sleep on the ground, as do chimpanzees in some popu- tively unsuccessful genus in terms of its population densities lations, but only where predators are not a serious risk (Koops and total numbers, at least intermittently. Later, Neanderthals et al. 2007). Terrestrial sleeping for humans in a predator-rich were in small, widely dispersed groups, and H. sapiens expe- savanna, by contrast, can be expected to be very dangerous. rienced a severe bottleneck around 70,000 years ago. How do Even today people in lion-rich environments are most vul- we reconcile the benefits of cooked food with the poor eco- nerable to predation shortly after dark (Packer et al. 2011). logical success of Homo before agriculture? Accordingly, the reduction of climbing adaptations that occurred One possibility is that by adapting to forego the ability to eat with H. erectus appears to signal the simultaneous evolution of fiber-rich foods (such as raw leaves, stems, and USOs), Homo a method of achieving safety at night. Control of fire is the boxed themselves into requiring even higher-quality foods obvious possibility because it is the principal method used by than before. Such foods would be animal products and low- modern humans sleeping in the kinds of habitats occupied by toxin, low-lignin plants containing high concentrations of sugar H. erectus (Wrangham and Carmody 2010). or starch. The acquisition of adaptations to take maximal ad- Shipman (2009) objected to this proposal by noting that vantage of cooked foods may thus have been a Faustian pact in antelope sleep on the ground without fire. However, antelope which the benefit of high-quality foods was set against the loss sleep less than humans (Richard Estes, personal communica- of ability to digest foods on which other hominoids (such as tion). Smaller species tend to hide at night, under bushes, for chimpanzees and orangutans) can readily survive. example. Larger species avoid cover and prefer to be in the open. Although no details are known about sleeping patterns of Af- rican ungulates, they certainly do not have long periods of re- Why Was Homo habilis Intermediate? laxed sleep. Even domesticated ungulates sleep briefly: accord- ing to Elgar et al. (1988, 1990), total sleep time per 24 hours is There is increasing evidence that the evolution of H. erectus less for artiodactyls (mean 5.3 h) and perissodactyls (4.8 h) was not a single “Adamic” event (Hublin 2015) but a “fuzzy thanin10otherordersofmammals,includingprimates(10.3h). transition” occurring in a complex series of shifts (Antón, Potts, In relation to body mass, domesticated ungulates also have the and Aiello 2014; Antón and Snodgrass 2012). This means that shortest cycle of REM sleep and slow-wave sleep of nine orders whatever the dietary change responsible for H. erectus, its effect of mammals, whereas primates have the longest. Short REM took time to be felt, and geographical variation suggests that the cycles in ungulates fit with the observation by Richard Estes process happened erratically in space. Thus, the conclusion that (personal communication) that wildebeest (Connochaetes tau- H. erectus was the first obligatory cook leads to the expectation rinus) sleep deeply for only a few minutes at a time. In general, that species before H. erectus would not show indications of be- species in riskier environments have less REM sleep (Lesku ing adapted to cooked food. In line with this prediction, H. ha- et al. 2006). bilis has a similar postcanine crown size to Australopithecus Thus, ungulates sleep little compared with humans, en- africanus and a similarly robust jaw in relation to body mass abling them to be relatively vigilant, and they sprint faster. (Eng et al. 2013; Wood and Collard 1999). Yet their mortality from predation is clearly much higher than However, there are at least two problems that the cooking in humans. Among Kalahari foragers, data collected by Polly hypothesis must deal with. First, based on one specimen (OH13), S310 Current Anthropology Volume 58, Supplement 16, August 2017

Eng et al. (2013) modeled H. habilis as producing only a low There May Have Been Important Variation in Gut Size maximum bite force in line with later Homo and different from within Post-Habiline Homo the higher bite forces of Australopithecus (and contemporary The idea that the inferred small gut size and observed small great apes). This result came from H. habilis having a relatively molar size of H. erectus are only explicable by a cooked diet is small second molar. Eng el al. (2013) therefore suggested that challenged by the observation that there is more difference in H. habilis might be adapted to foods that had been mechani- gut size among species of Homo than previously appreciated. cally processed to reduce their toughness or other physical This raises the possibility that the intestinal volume (including challenges. In support, dental microwear studies indicate that the colon) could have been so much larger in earlier Homo Australopithecus ate tougher foods than H. habilis (Ungar and than in H. sapiens that H. erectus was capable of surviving on a Scott 2009; Villmoare et al. 2015). Thus, a potential solution to raw diet. How much larger can only be guessed, but the idea H. habilis having a craniofacial structure that is to some extent raises the possibility that the colon was sufficiently large for intermediate between Australopithecus and subsequent Homo foods to be retained and well fermented. Ben-Dor, Gopher, is that a period of mechanical processing preceded cooking. and Barkai (2016) have similarly suggested that the digestive While this makes sense given that stone tools that could cut anatomy of Neanderthals differed significantly from that of and pound food were available long before Homo erectus,it H. sapiens by being relatively large. If this meant that they also raises the possibility that the reduced craniofacial ro- were able to live on raw diets, the fact that some Neanderthal busticity of H. erectus reflects a continuation of the same pro- sites were apparently occupied without fire would be more cess (Zink and Lieberman 2016). According to this idea, me- easily explicable (Sandgathe et al. 2011a, 2011b; Henry 2017). chanical, nonthermal processing became even more important An improved understanding of the relationship between trunk and effective in H. erectus than in H. habilis and accounts for morphology and digestive anatomy would be useful in estab- the small teeth (a reduction of ca. 25% in size), shorter face, lishing the potential for different species of Homo to exploit and more lightly built jaw. different diets. The second challenge has a similar implication of being able to explain H. erectus biology on the basis of adaptations begun in H. habilis. Before “Homo-ization” of the jaw and teeth, brain Biological Evidence for Late Adaptation to Fire size in H. habilis had risen from the australopithecine level Several recent results challenge the prediction that fire was – – of 385 571 cc to 510 750 cc (Antón, Potts, and Aiello 2014; controlled by species as early as H. erectus. Chisholm et al. Spoor et al. 2015). Whether this was due to increased animal (2016) concluded that the evolution of tuberculosis in Homo foods in the diet, nonthermal processing, or some other change, was a consequence of living with fire and that it likely hap- the increase in diet quality that is indicated by the rising brain pened between 6 ka and 70 ka. While analysis of dental calculus size in H. habilis could have been continued and caused the shows evidence of smoke inhalation between 300 and 400 ka changes seen in H. erectus. at Qesem (Hardy et al. 2015), at Atapuerca at 1.2 Ma it indi- A solution offered by the cooking hypothesis is that before cates consumption of raw foods and a lack of exposure to fire the obligatory adaptation to cooking indicated for H. erectus, (Hardy et al. 2016). Hubbard et al. (2016) found an allele con- fi H. habilis used re intermittently. As a result, they were able to ferring protection against smoke toxins in H. sapiens that was regularly eat relatively tender food (as indicated by microwear; absent in four Neanderthals and Denisovans, suggesting ad- fi Ungar 2012) and gain suf cient extra energy to promote an aptation to fire by at least 550 ka (Prüfer et al. 2014). increase in brain size. However, because they could not guar- antee having access to cooked food, they retained dental and Closing Remarks digestive adaptations that allowed them to effectively chew plant foods when animal foods were scarce. The cooking hypothesis is sometimes treated as if the fossil Despite these problems for the idea of a single shift en- evidence of an improved and more easily chewed diet at the compassing diverse features of Homo simultaneously, it is of- origin of Homo erectus is its only significant source of support ten concluded that habilines had a faster life history than (e.g., Armelagos 2014: “The link to changes in morphology H. erectus (e.g., “non-erectus early Homo was smaller and de- that are claimed to be related to cooking are confounded by the veloped more quickly than H. erectus ”; Antón and Snodgrass evidence for the control of fire” [1333]). Given the contrary 2012:S487). A slower life history could imply earlier weaning archaeological evidence, such a perspective can easily relegate and more dependence of juveniles on adults for food (Thomp- a dependence on cooking and the control of fire to a later time son and Nelson 2011). Early weaning is clearly hard to rec- in human evolution. The implication is that compared with oncile with a raw diet unless it predictably included such elu- other changes responsible for H. erectus, the evolution of cook- sive foods as brain and fat-rich marrow. Unfortunately, although ing and the control of fire have had relatively little influence these arguments are intriguing, they are premature given that on behavior and adaptation. This kind of thinking presum- it remains unclear how different the life history of H. erectus ably explains why some recent papers considering the origin of was from its antecedents (Schwartz 2012). H. erectus and the reasons for its higher-quality diet, larger Wrangham The Cooking Hypothesis S311 body, and larger brain have omitted any consideration of the Axelsson, Erik, Abhirami Ratnakumar, Maja-Louise Arendt, Khurram Maq- bool, Matthew T. Webster, Michele Perloski, Olof Liberg, Jon M. Arnemo, cooking hypothesis (Antón and Snodgrass 2012; Potts 2012). Åke Hedhammar, and Kerstin Lindblad-Toh. 2013. The genomic signature In fact, however, the evidence of increased dietary quality in of dog domestication reveals adaptation to a starch-rich diet. Nature 495 the early Lower Paleolithic is only one of several sources of (7441):360–364. Balikci, Asen. 1989. The Netsilik Eskimo. Prospect Heights, IL: Waveland. support for the cooking hypothesis, as discussed above. Fur- Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a rea- thermore, the emergence of H. erectus is not adequately ac- son: barbecue at Middle Pleistocene Qesem Cave, Israel. Current Anthro- counted for by an increased frequency of meat eating. Thus, pology 58(suppl. 16):S314–S328. Barr, Susan I. 1999. Vegetarianism and menstrual cycle disturbances: is there while the cooking hypothesis may be wrong, it cannot fairly be an association? American Journal of Clinical Nutrition 70:549S–554S. dismissed by ignoring it. Barton, R. A. 1992. Allometry of food intake in free-ranging anthropoid primates. Folia primatologica 58:56–59. Accordingly, two questions must be answered before the ’ fi Ben-Dor, Miki, Avi Gopher, and Ran Barkai. 2016. Neandertals large lower time for the control of re is assigned to the mid-Pleistocene. thorax may represent adaptation to high protein diet. American Journal of First, how could H. erectus use increased energy, reduce its Physical Anthropology 160(3):367–378. chewing efficiency, and sleep safely on the ground without fire? Ben-Dor, Miki, Avi Gopher, Israel Hershkovitz, and Ran Barkai. 2011. Man the fat hunter: the demise of Homo erectus and the emergence of a new Second, how could a cooked diet have been introduced to a hominin lineage in the Middle Pleistocene (ca. 400 kyr) Levant. PLoS ONE raw-foodist, mid-Pleistocene Homo without having major ef- 6(12):e28689. fects on its evolutionary biology? Satisfactory answers to these Berna, F., P. Goldberg, L. K. Horwitz, J. Brink, S. Holt, M. Bamford, and M. Chazan. 2012. Microstratigraphic evidence of in situ fire in the Acheu- questions will do much to resolve the tension between ar- lean strata of Wonderwerk Cave, Northern Cape Province, South Africa. chaeological and biological evidence. Proceedings of the National Academy of Sciences of the USA 109:E1215– The results should be rewarding. The control of fire and the E1220. Boback, Scott M., Christian L. Cox, Brian D. Ott, Rachel Carmody, Richard emergence of cooking had numerous effects on human biology W. Wrangham, and Stephen M. Secor. 2007. Cooking reduces the cost of and behavior, including cognition and cooperation (Attwell, meat digestion. Comparative Biochemistry and Physiology 148:651–656. Kovarovic, and Kendal 2015; Burton 2009; Dunbar and Gow- Bosinski, G. 2006. Les premiers peuplements de l’Europe centrale et de l’est. Comptes Rendus Palevol 5(1/2):311–317. lett 2014; Gintis, van Schaik, and Boehm 2015; Wiessner 2014; Brace, C. Loring. 1995. The stages of human evolution. 5th ed. Englewood Wrangham 2009). A better understanding of when the process Cliffs, NJ: Prentice-Hall. started will have wide-ranging implications for human biolog- Bramble, Dennis M., and Daniel E. Lieberman. 2004. Endurance running and the evolution of Homo. Nature 432:345–352. ical and social evolution. Brasaemle, Dawn L. 2007. The perilipin family of structural lipid droplet proteins: stabilization of lipid droplets and control of lipolysis. Journal of Lipid Research 48:2547–2559. Acknowledgments Buck, Laura T., J. Collette Berbesque, Brian M. Wood, and Chris B. Stringer. 2016. Tropical forager gastrophagy and its implications for extinct hominin I thank Francesco Berna, Dennis Sandgathe, Leslie Aiello, diets. Journal of Archaeological Science: Reports 5:672–679. Burton, F. D. 2009. Fire: the spark that ignited human evolution. Albuquerque: Laurie Obbink, and fellow participants for a fascinating Wenner- University of New Mexico Press. Gren conference, “Fire and the Genus Homo,” October 16–22, Carlson, Bryce A., and John D. Kingston. 2007. Docosahexaenoic acid, the 2015, in Sintra, Portugal. Rachel Carmody and Nancy-Lou aquatic diet, and hominin encephalization: difficulties in establishing evo- lutionary links. American Journal of Human Biology 19:132–141. Conklin-Brittain have been invaluable collaborators for a de- Carmody, Rachel N. 2012. Energetic consequences of thermal and non- cade and more. Dawn Brasaemle and Caroline Pond offered thermal food processing. PhD dissertation, Harvard University, Cambridge, helpful ideas about lipids. Manuel Dominguez-Rodrigo and two MA. Carmody, Rachel N., Michael Dannemann, Adrian W. Briggs, Birgit Nickel, anonymous reviewers kindly made constructive comments. Emily E. Groopman, Richard W. Wrangham, and Janet Kelso. 2016. Ge- netic evidence of human adaptation to a cooked diet. Genome Biology and Evolution 8(4):1091–1103. References Cited Carmody, Rachel N., Gil S. Weintraub, and Richard W. Wrangham. 2011. Energetic consequences of thermal and non-thermal food processing. Pro- Aiello, Leslie, and Peter Wheeler. 1995. The expensive-tissue hypothesis: the ceedings of the National Academy of Sciences of the USA 108(48):19199– brain and the digestive system in human and primate evolution. Current 19203. Anthropology 36:199–221. Carmody, Rachel N., and Richard W. Wrangham. 2009. The energetic sig- Alperson-Afil, N. 2008. Continual fire-making by hominins at Gesher Benot nificance of cooking. Journal of Human Evolution 57:379–391. Ya‘aqov, Israel. Quaternary Science Reviews 27:1733–1739. Carpenter, Danielle, Sugandha Dhar, Laura M. Mitchell, Beiyuan Fu, Jess Antón, Susan C., Richard Potts, and Leslie C. Aiello. 2014. Evolution of early Tyson, Nzar A. A. Shwan, Fengtang Yang, Mark G. Thomas, and John A. L. Homo: an integrated biological perspective. Science 345:1236828. Armour. 2015. Obesity, starch digestion and amylase: association between Antón, Susan C., and J. Josh Snodgrass. 2012. Origins and evolution of genus copy number variants at human salivary (AMY1) and pancreatic (AMY2) Homo: new perspectives. Current Anthropology 53(suppl. 6):S479–S496. amylase genes. Human Molecular Genetics 24(12):3472–3480. Armelagos, George J. 2014. Brain evolution, the determinates of food choice, Chisholm, Rebecca H., James M. Trauer, Darren Curnoe, and Mark M. and the omnivore’s dilemma. Critical Reviews in Food Science and Nutri- Tanaka. 2016. Controlled fire use in early humans might have triggered the tion 54(10):1330–1341. evolutionary emergence of tuberculosis. Proceedings of the National Acad- Arrese, Estela L., and Jose L. Soulages. 2010. Insect fat body: energy, metab- emy of Science of the USA 113(32):9051–9056. olism, and regulation. Annual Review of Entomology 55:207–225. Coolidge, Frederick L., and Thomas Wynn. 2006. The effects of the tree-to- Attwell, Laura, Kris Kovarovic, and Jeremy R. Kendal. 2015. Fire in the Plio- ground sleep transition in the evolution of cognition in early Homo. Before Pleistocene: the functions of hominin fire use, and the mechanistic, de- Farming 4:1–18. velopmental and evolutionary consequences. Journal of Anthropological Coon, Carleton S. 1962. The history of man: from the first human to primitive Sciences 93:1–20. culture and beyond. 2nd ed. London: Jonathan Cape. S312 Current Anthropology Volume 58, Supplement 16, August 2017

Crittenden, A. N. 2011. The importance of honey consumption in human results of a questionnaire survey. Annals of Nutrition and Metabolism 43:69– evolution. Food and Foodways 19:257–273. 79. de Heinzelin, Jean, J. Desmond Clark, Tim White, William Hart, Paul Renne, Koops, Kathelijne, Tatyana Humle, Elisabeth H. M. Stereck, and Tetsuro Giday WoldeGabriel, Yonas Beyene, and Elisabeth Vrba. 1999. Environment Matsuzawa. 2007. Ground-nesting by the chimpanzees of the Nimba Moun- and behavior of 2.5-million-year-old Bouri hominids. Science 284:625–629. tains, Guinea: environmentally or socially determined? American Journal of Domínguez-Rodrigo, Manuel, and Luis Alcalá. 2016. 3.3-million-year-old Primatology 69:407–419. stone tools and butchery traces? more evidence needed. Paleoanthropology Kraft, Thomas S., and Vivek V. Venkataraman. 2015. Could plant extracts 2016:46253. have enabled hominins to acquire honey before the control of fire? Journal Dominy, Nathaniel J., Erin R. Vogel, Justin D. Yeakel, Paul Constantino, and of Human Evolution 85:65–74. Peter W. Lucas. 2008. Mechanical properties of plant underground storage Lentle, Roger G., and Patrick W. M. Janssen. 2011. The physical processes of organs and implications for dietary models of early hominins. Evolutionary digestion. New York: Springer. Biology 35:159–175. Leonard, W. R., J. J. Snodgrass, and M. L. Robertson. 2007. Effects of brain Dunbar, R. I. M., and J. A. J. Gowlett. 2014. Fireside chat: the impact of fire on evolution on human nutrition and metabolism. Annual Review of Nutrition hominin socioecology. In Lucy to language: the benchmark papers.R.I.M. 27:311–327. Dunbar, C. Gamble, and J. A. J. Gowlett, eds. Pp. 277–296. Oxford: Oxford Lesku, John A., Timothy C. Roth II, Charles J. Amlaner, and Steven L. Lima. University Press. 2006. A phylogenetic analysis of sleep architecture in mammals: the inte- Dunham, K. M., and Martyn G. Murray. 1982. The fat reserves of impala, gration of anatomy, physiology, and ecology. American Naturalist 168(4): Aepyceros melampus. African Journal of Ecology 20:81–87. 441–453. Elgar, Mark, Mark D. Pagel, and Paul H. Harvey. 1988. Sleep in mammals. Lupo, Karen D. 1998. Experimentally derived extraction rates for marrow: Animal Behaviour 36:1407–1419. implications for body part exploitation strategies of Plio-Pleistocene hom- ———. 1990. Sources of variation in mammalian sleep. Animal Behaviour inid scavengers. Journal of Archaeological Science 25:657–675. 40:991–995. Marlowe, F. W., J. C. Berbesque, B. Wood, A. Crittenden, C. Porter, and A. Emery Thompson, Melissa, and Richard W. Wrangham. 2008. Diet and re- Mabulla. 2014. Honey, Hadza, hunter-gatherers, and human evolution. Jour- productive function in wild female chimpanzees (Pan troglodytes schwein- nal of Human Evolution 71:119–128. furthii). American Journal of Physical Anthropology 135(2):171–181. Marshall, Andrew J., and Richard W. Wrangham. 2007. Evolutionary con- Eng, Carolyn M., Daniel E. Lieberman, Katherine D. Zink, and Michael A. sequences of fallback foods. International Journal of Primatology 28:1219– Peters. 2013. Bite force and occlusal stress production in hominin evolu- 1235. tion. American Journal of Physical Anthropology 151:544–557. Martin, Robert D., David J. Chivers, Anne M. MacLarnon, and C. Marcel Finarelli, John A., and John J. Flynn. 2009. Brain-size evolution and sociality Hladik. 1985. Gastrointestinal allometry in primates and other mammals. in Carnivora. Proceedings of the National Academy of Sciences of the USA In Size and scaling in primate biology. W. L. Jungers, ed. Pp. 61–89. New 106(23):9345–9349. York: Plenum. Fonseca-Azevedo, Karina, and Suzana Herculano-Houzel. 2012. Metabolic McPherron, Shannon P., Zeresenay Alemseged, Curtis W. Marean, Jonathan constraint imposes tradeoff between body size and number of brain neu- G. Wynn, Denné Reed, Denis Geraads, René Bobe, and Hamdallah A. rons in human evolution. Proceedings of the National Academy of Sciences Béarat. 2010. Evidence for stone-tool-assisted consumption of animal tis- of the USA 109(45):18571–18576. sues before 3.39 million years ago at Dikika, Ethiopia. Nature 466:857–860. Gintis, Herbert, Carel van Schaik, and Christopher Boehm. 2015. The evo- Milton, Katharine, and Montague W. Demment. 1988. Chimpanzees fed high lutionary origins of human political systems. Current Anthropology 56(3): and low fiber diets and comparison with human data. Journal of Nutrition 327–353. 118:1082–1088. Gowlett, John A. J., and R. W. Wrangham. 2013. Earliest fire in Africa: to- Organ, Chris L., Charles L. Nunn, Zarin Machanda, and Richard W. Wrang- wards the convergence of archaeological evidence and the cooking hy- ham. 2011. Phylogenetic rate shifts in chewing time during the evolution of pothesis. Azania: Archaeological Research in Africa 48:5–30. Homo. Proceedings of the National Academy of Sciences of the USA 108: Groopman, Emily, R. N. Carmody, and R. W. Wrangham. 2015. Cooking 14555–14559. increases net energy gain from a lipid-rich food. American Journal of Packer, Craig, A. Swanson, D. Ikanda, and H. Kushnir. 2011. Fear of darkness, Physical Anthropology 156(1):11–18. the full moon and the nocturnal ecology of African lions. PLoS ONE 6(7): Hardy, Karen, Jennie Brand-Miller, Katherine D. Brown, Mark G. Thomas, e22285. and Les Copeland. 2015. The importance of dietary carbohydrate in human Parker, Christopher H., Earl R. Keefe, Nicole M. Herzog, James F. O’Connell, evolution. Quarterly Review of Biology 90(3):251–268. and Kristen Hawkes. 2016. The pyrophilic primate hypothesis. Evolution- Hardy, Karen, Anita Radini, Stephen Buckley, Ruth Blasco, Les Copeland, ary Anthropology 25:54–63. Francesc Burjachs, Josep Girbal, Riker Yll, Eudald Carbonell, and Jose María Perry, G. H., N. J. Dominy, K. G. Claw, A. S. Lee, H. Fiegler, R. Redon, J. Bermúdez de Castro. 2016. Diet and environment 1.2 million years ago Werner, et al. 2007. Diet and the evolution of human amylase gene copy revealed through analysis of dental calculus from Europe’s oldest hominin number variation. Nature Genetics 39:1256–1260. at Sima del Elefante, Spain. Science of Nature, doi:10.1007/s00114-016 Perry, G. H., L. Kistler, M. A. Kelaita, and A. J. Sams. 2015. Insights into -1420-x. hominin phenotypic and dietary evolution from ancient DNA sequence Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current data. Journal of Human Evolution 79:55–63. Anthropology 58(suppl. 16):S329–S336. Perry, George H. 2014. Parasites and human evolution. Evolutionary An- Hladik, C. Marcel, David J. Chivers, and P. Pasquet. 1999. On diet and gut size thropology 23:218–228. in non-human primates and humans: is there a relationship to brain size? Pontzer, Herman. 2012. Ecological energetics in early Homo. Current An- Current Anthropology 40:695–697. thropology 53(suppl. 6):S346–S358. Holliday, Trenton W. 2012. Body size, body shape, and the circumscription of Potts, Richard. 2012. Environmental and behavioral evidence pertaining to the genus Homo. Current Anthropology 53(suppl. 6):S330–S345. the evolution of early Homo . Current Anthropology 53(suppl. 6):S299–S317. Hrdy, Sarah Blaffer. 2009. Mothers and others: the evolutionary origins of Prüfer, K., F. Racimo, N. Patterson, F. Jay, S. Sankararaman, S. Sawyer, A. mutual understanding. Cambridge, MA: Harvard University Press. Heinze, et al. 2014. The complete genome sequence of a Neanderthal from Hubbard, Troy D., Iain A. Murray, William H. Bisson, Alexis P. Sullivan, the Altai Mountains. Nature 505:43–49. Aswathy Sebastian, George H. Perry, Nina G. Jablonski, and Gary H. Per- Ragir, S., M. Rosenberg, and P. Tierno. 2000. Gut morphology and the dew. 2016. Divergent Ah receptor ligand selectivity during hominin evo- avoidance of carrion among chimpanzees, baboons, and early hominids. lution. Molecular Biology and Evolution, doi:10.1093/molbeh/msw143. Journal of Anthropological Research 56(4):477–512. Hublin, Jean-Jacques. 2015. Paleoanthropology: how old is the oldest human? Reed, David L., Jessica E. Light, Julie M. Allen, and Jeremy J. Kirchman. 2007. Current Biology 25:R448–R469. Pair of lice lost or parasites regained: the evolutionary history of anthropoid Isler, Karin, and Carel P. van Schaik. 2012. How our ancestors broke through primate lice. BMC Biology 5(7):1–11. the gray ceiling: comparative evidence for cooperative breeding in early Reiter, T., E. Jagoda, and T. D. Capellini. 2016. Dietary variation and evo- Homo. Current Anthropology 53(suppl. 6):S453–S465. lution of gene copy number among dog breeds. PLoS ONE 11(2):e0148899. Koebnick, Corinna, C. Strassner, I. Hoffmann, and C. Leitzmann. 1999. Con- Rightmire, G. Philip. 2004. Brain size and encephalization in early to mid- sequences of a longterm raw food diet on body weight and menstruation: Pleistocene Homo. American Journal of Physical Anthropology 124:109–123. Wrangham The Cooking Hypothesis S313

Roebroeks, Wil, and Paola Villa. 2011. On the earliest evidence for habitual Thompson, Jennifer L., and Andrew J. Nelson. 2011. Middle childhood and use of fire in Europe. Proceedings of the National Academy of Sciences of the modern human origins. Human Nature 22(3):249–280. USA 108(13):5209–5214. Ungar, Peter S. 2012. Dental evidence for the reconstruction of diet in African Rosell, Magdalena, Paul Appleby, and Tim Key. 2005. Height, age at men- early Homo. Current Anthropology 53(suppl. 6):S318–S329. arche, body weight and body mass index in life-long vegetarians. Public Ungar, Peter S., and Robert S. Scott. 2009. Dental evidence for diets of early Health Nutrition 8(7):870–875. Homo.InThe first humans: origins of the genus Homo. F. E. Grine, J. G. Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. McPherron, Fleagle, and R. E. Leakey, eds. Pp. 121–134. New York: Springer. Alain Turq, Laura Niven, and Jamie Hodgkins. 2011a. On the role of fire in Villmoare, Brian, William H. Kimbel, Chalachew Seyoum, Christopher J. Neandertal adaptations in Western Europe: evidence from Pech de l’Azé IV Campisano, Erin DiMaggio, John Rowan, David R. Braun, J. Ramon Ar- and Roc de Marsal, France. Paleoanthropology 2011:216–242. rowsmith, and Kaye E. Reed. 2015. Early Homo at 2.8 Ma from Ledi- ———. 2011b. Timing of the appearance of habitual fire use. Proceedings of Geraru, Afar, Ethiopia. Science 347(6228):1352–1355. the National Academy of Sciences of the USA 108(29):E298. Walker, M. J., D. Anesin, D. Angelucci, A. Avilés-Fernández, F. Berna, A. T. Schmid, P. 1983. Eine Rekonstruktion des Skelettes von A.L. 288-1 (Hadar) Buitrago-López, J. S. Carrión, et al. 2016. A view from a cave: Cueva Negra und deren Konsequenzes. Folia Primatologica 40:283–306. del Estrecho del Río Quípar (Caravaca de la Cruz, Murcia, Southeastern Schnorr, Stephanie L., Alyssa N. Crittenden, Koen Venema, Frank W. Mar- Spain): reflections on fire, technological diversity, environmental exploita- lowe, and Amanda G. Henry. 2015. Assessing digestibility of Hadza tubers tion, and palaeoanthropological approaches. Human Evolution 31(1/2):1–67. using a dynamic in-vitro model. American Journal of Physical Anthropology Warneken, Felix, and Alex Rosati. 2015. Cognitive capacities for cooking in 158:371–385. chimpanzees. Proceedings of the Royal Society B 282:1809. Schwartz, Gary T. 2012. Growth, development, and life history throughout the Whitehouse, S. J. O. 1977. The diet of the dingo in Western Australia. Aus- evolution of Homo. Current Anthropology 53(suppl. 6):S395–S408. tralian Wildlife Research 4(2):145–150. Shimelmitz, Ron, Steven L. Kuhn, Arthur J. Jelinek, Avraham Ronen, Amy E. Wiessner, Polly W. 2014. Embers of society: firelight talk among the Ju/ Clark, and Mina Weinstein-Evron. 2014. “Fire at will”: the emergence of ’hoansi Bushmen. Proceedings of the National Academy of Sciences of the habitual fire use 350,000 years ago. Journal of Human Evolution 77(suppl. 1): USA 111(39):14027–14035. 196–203. Wood, Bernard A., and Mark Collard. 1999. The human genus. Science 284:65–71. Shipman, Pat. 2009. Cooking debate goes off the boil. Nature 459:1059–1060. Wrangham, R. W. 2011. Honey and fire in human evolution. In Casting the Smith, Alex R., Rachel N. Carmody, Rachel J. Dutton, and Richard W. net wide. J. Sept and D. Pilbeam, eds. Pp. 149–167. Cambridge, MA: Pea- Wrangham. 2015. The significance of cooking for early hominin scaveng- body Museum. ing. Journal of Human Evolution 84:62–70. Wrangham, Richard W. 2009. Catching fire: how cooking made us human. Speth, John. 2010. The paleoanthropology and archaeology of big-game New York: Basic. hunting: protein, fat or politics? New York: Springer. ———. 2006. The cooking enigma. In Early hominin diets: the known, the Spoor, Fred, Philipp Gunz, Simon Neubauer, Stefanie Stelzer, Nadia Scott, unknown, and the unknowable. P. Ungar, ed. Pp. 308–323. New York: Amandus Kwekason, and M. Christopher Dean. 2015. Reconstructed Homo Oxford University Press. habilis type OH 7 suggests deep-rooted species diversity in early Homo. Na- Wrangham, Richard W., and Rachel Carmody. 2010. Human adaptation to ture 519:83–86. the control of fire. Evolutionary Anthropology 19:187–199. Spottiswoode, Claire N., Keith S. Begg, and Colleen M. Begg. 2016. Reciprocal Wrangham, Richard W., and Nancy L. Conklin-Brittain. 2003. The biological signaling in honeyguide-human mutualism. Science 353:387–389. significance of cooking in human evolution. Comparative Biochemistry and Stiner, Mary C. 1994. Honor among thieves. Princeton, NJ: Princeton Uni- Physiology A 136:35–46. versity Press. Wrangham, Richard W., James Holland Jones, Greg Laden, David Pilbeam, Stiner, Mary C., Ran Barkai, and Avi Gopher. 2009. Cooperative hunting and and Nancy Lou Conklin-Brittain. 1999. The raw and the stolen: cooking meat sharing 400–200 kya at Qesem Cave, Israel. Proceedings of the Na- and the ecology of human origins. Current Anthropology 40:567–594. tional Academy of Sciences of the USA 106(32):13207–13212. Zink, K., and Daniel E. Lieberman. 2016. Impact of meat and Lower Swanson, Eli M., Kay E. Holekamp, Barbara L. Lundrigan, Bradley M. Arsznov, Palaeolithic food processing techniques on chewing in humans. Nature 531 and Sharleen T. Sakai. 2012. Multiple determinants of whole and regional (7595):500–503. brain volume among terrestrial carnivorans. PLoS ONE 7(6):e38447. Zink, Katherine D., Daniel E. Lieberman, and Peter W. Lucas. 2014. Food Symons, Michael. 1998. A history of cooks and cooking. Urbana: University of material properties and early hominin processing techniques. Journal of Illinois Press. Human Evolution 77:155–166. S314 Current Anthropology Volume 58, Number S16, August 2017

Fire for a Reason Barbecue at Middle Pleistocene Qesem Cave, Israel

by Ran Barkai, Jordi Rosell, Ruth Blasco, and Avi Gopher

Qesem Cave is a Middle Pleistocene site in Israel occupied between 420 and 200 ka. Excavations have revealed a wealth of innovative behaviors most likely practiced by a new hominin lineage. These include early evidence for the habitual and continuous use of fire, the repeated use of a central hearth, systematic flint and bone recycling, early blade pro- duction technologies, social hunting strategies and meat-sharing practices, and more. Fire was used throughout the 200,000 years of human occupation of the cave primarily for meat roasting and cooking. Roasting and cooking, we argue, had an important role in providing the necessary caloric intake of the cave’s inhabitants. We see fire as an essential element of the new post-Acheulian human adaptation in the Levant. The ample recurring evidence for fo- cused and repeated use of fire for dietary purposes suggests that fire production, control, use, and maintenance were habitually practiced by the cave’s inhabitants and that fire-induced calories became central for their survival. We present an integrative view regarding the use of fire at Qesem Cave and discuss the role of fire within the framework of the significant cultural and biological transformations that took shape in the post-Acheulian Levant during the Middle Pleistocene.

Introduction Paleolithic period in the Levant triggered human communities to make use of their extensive cultural and social capabilities as fi The use of re at Middle Pleistocene Qesem Cave (Qesem) has well as their profound familiarity with their environments and been discussed, and the repeated use of a central hearth at 300 ka their survival skills by using a new mode of adaptation. The use has been demonstrated in previous publications (Falguères et al. of fire for roasting meat and for cooking in general was a central 2015; Karakanas et al. 2007; Shahack-Gross et al. 2014). In this element of this new adaptation, and the evidence from Qesem is fi paper, our aim is to view the human use of re at Qesem in the consistent with this hypothesis. We start with a general introduc- context of cultural and biological transformations that took tion about the Acheulo-Yabrudian Cultural Complex (AYCC) shape in the Levant at ca. 400 ka and to specify the major uses of as reflected by the plethora of information gathered from the fi re at Qesem in particular and in the late Lower Paleolithic archaeological deposits of Qesem. Then we focus on direct and period in the Levant in general. We claim that the common and indirect evidence for the use of fire at the cave (ash, hearths, fi continuous use of re for roasting meat at 400 ka at Qesem was burned bones and lithics, stone tools used in butchering, and fi a Rubicon crossed for the rst time, and it characterizes human charcoal contained in dental calculus). Finally, we present our existence from that time to this very day. We contend that the hypothesis regarding the earliest, continuous, and common use fi combination of speci c circumstances during the late Lower of fire for roasting meat and cooking and its implications to human diet, culture, and adaptation.

Ran Barkai is Associate Professor at the Institute of Archaeology of The Acheulo-Yabrudian Cultural Complex and the Tel Aviv University (POB 39040, 69978 Tel Aviv, Israel [barkaran State of the Art of Qesem Cave Studies @post.tau.ac.il]). Jordi Rosell is Researcher in the Àrea de Prehistòria of the University Rovira i Virgili (URV, Av. Catalunya 35, 43002 Tar- The AYCC is a Middle Pleistocene, late Lower Paleolithic cul- ragona, Spain) and the Institut Català de Paleoecologia Humana I tural entity of the Levant. The AYCC consists of three distinct Evolució Social (IPHES, Zona Educacional 4, Campus Sescelades URV, lithic industries: the Acheulo-Yabruidan (a flake industry with fi Edi ci W3, 43007 Tarragona, Spain [[email protected]]). a notable presence of handaxes), the Yabrudian (a flake indus- Ruth Blasco is Researcher at the Centro Nacional de Investigación sobre try dominated by Quina and demi-Quina scrapers), and the la Evolución Humana (Paseo Sierra de Atapuerca 3, 09002 Burgos, Spain) and Institute of Archaeology, Tel Aviv University (POB 39040, 69978 Amudian (a blade-dominated industry; see Bar-Yosef 1994; Tel Aviv, Israel [[email protected]]). Avi Gopher is Professor at Copeland 2000; Rust 1950). Stratigraphically, the AYCC repeat- the Institute of Archaeology of Tel Aviv University (POB 39040, 69978 edly postdates the Lower Paleolithic Acheulian and predates the Tel Aviv, Israel [[email protected]]). This paper was submitted Middle Paleolithic Mousterian. The absolute chronology of the 25 VII 16, accepted 20 I 17, and electronically published 17 V 17. AYCC covers a range of over 200 kyr between ca. 420 and q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0014$10.00. DOI: 10.1086/691211 Barkai et al. Fire for a Reason S315

200 ka (Barkai et al. 2003; Gopher et al. 2010). New thermo- The turning point from “early fire” to the habitual use of fire luminescence and electron spin resonance dates from Qesem is a matter of dispute. Roebroeks and Villa (2011) describe and Misliya Caves accord well with this range (Falguères et al. habitual use of fire as a systematically repeated use of fire in 2015; Mercier et al. 2013; Valladas et al. 2013). AYCC sites are specific sites and/or regions. Based on an increased number of known from the central and southern Levant in both caves and archaeological sites associated with evidence for fire being ex- open air settings but mostly in caves or rock shelters. tensively used in domestic contexts, they suggested that the Blade production in the Amudian industry is of special in- earliest habitual use of fire occurred in Middle Pleistocene terest and is one of the major innovations of the AYCC (e.g., (ca. 400–300 ka) Europe and southwest Asia. On top of that, the Bar-Yosef and Kuhn 1999). Middle Pleistocene blade pro- probable use of fi re in the Acheulian was never demonstrated. duction is a unique “ahead of its time” technological innova- We argue that starting at ca. 400 ka, hearths were used, first and tion that is part of a set of cultural and behavioral transfor- foremost, for roasting meat, and as a result, burned bones are mations that took shape in the Levant at ca. 400 ka and that found in abundance in post-Acheulian sites (Fernández et al. should be viewed as a local innovation that persisted for 200 kyr. 2012; Karkanas et al. 2007). Thus we concur with the statement The Amudian industry is characterized by the systematic pro- that during the Lower Paleolithic, Acheulian meat was most duction of blades and tools made of blades. Alongside blade probably consumed raw (Bar-Yosef 2006). As for the AYCC, production, a significant flake component also appears, includ- the evidence from Qesem presented below indicates an un- ing side scrapers in various frequencies (Copeland 2000; Jelinek equivocal change regarding the presence and use of fire in ar- 1982, 1990). Qesem has shown that the Amudian represents a chaeological sites from this point in time onward. major industry of the AYCC, equivalent in scale to the other known industries (e.g., Gopher et al. 2005; Shimelmitz, Gopher, Qesem Cave and Barkai 2011, 2015). In the Yabrudian industry, the innovative appearance of the Qesem is located on the western slopes of the Samarian hills Quina chaîne opératoire for the shaping of Quina scrapers (see some 12 km east of the Mediterranean at 90 m asl. With nearby Bourguignon 2001) is of note. These distinctive scrapers are large springs and being located at the ecotone between the well known from Middle Paleolithic Mousterian of Europe. swampy basins of the coastal plain to the west and the moun- However, the Yabrudian is much older than the European man- tainous ridges of Samaria, this location provides a rich Medi- ifestations of the Quina phenomenon. In the Levant, Quina terranean biome. To date, the excavation exposed some 70 m2 scrapers appear in large numbers in the AYCC Yabrudian (by (130 m3) using 50 # 50 cm spatial units, full recovery through the thousands at Qesem and Tabun caves) but cease to appear a 2.4 mm net, and screen washing through 0.5–0.8 mm for in post-AYCC Middle Paleolithic Mousterian sites. The fact microfauna-bearing layers. that Quina scrapers are not known from earlier Acheulian con- Qesem is a sediment-filled karstic chamber that became avail- texts makes their presence in the Yabrudian unique and quite able to humans following slope erosion (Frumkin et al. 2009). enigmatic. The stratigraphic sequence was divided into two major parts: As for the makers of the AYCC, this remains an open the lower consists of a 16.5 m accumulation of sediments with question. The Galilee Man skull from Zuttiyeh Cave (Freidline clastic content, gravel, and clays, and the upper consists of 4.5 m et al. 2012; Keith 1927; Zeitoun 2001) and the dental remains of cemented sediment with a large ash component. The lower from Qesem hint at a new, post-Acheulian, post-erectus ho- part was deposited in a closed karstic chamber, while the upper minin lineage in the Levant at ca. 400 ka (Ben Dor et al. 2011; part was deposited when the cave was more open, as indicated Hershkovitz et al. 2011). This statement is supported by new by the presence of calcified rootlets (Karkanas et al. 2007). studies of Middle Pleistocene skeletal remains, mainly dental Qesem is a Middle Pleistocene (MIS 11-7) site dated by remains from the Levant and beyond (Le Cabec et al. 2013; Liu various methods (over 100 dates) to between 420 and 200 ka et al. 2013; Rink et al. 2013), as well as genetic evidence sug- (Barkai et al. 2003; Falguères et al. 2016; Gopher et al. 2010; gesting a Middle Pleistocene (before 300 ka) date for the ances- Mercier et al. 2013). The whole stratigraphic sequence is as- tors of modern humans and/or Neanderthals (e.g., Endicott, signed to the AYCC (Barkai and Gopher 2011, 2013). Ho, and Stringer 2010; Mendez et al. 2013). Ongoing research at the site provides ample evidence of in- As for the use of fire, the evidence indicates that human novative behaviors. This pertains, for example, to serial blade use of fire during the AYCC in the Levant was common and production (Shimelmitz, Gopher, and Barkai 2011), the acqui- relatively widespread (see a recent review in Shimelmitz et al. sition of raw material for selected tool types from underground 2014). Earlier use of fire in the Levant was reported at Gesher sources (Verri et al. 2004), the production of Quina scrapers Benot Ya‘aqov only (Alperson-Afil 2017; Alperson-Afil and using Quina debitage and retouch technologies (Lemorini et al. Goren-Inbar 2010; Goren-Inbar et al. 2004). Such “early fire” 2016), intensive and varied recycling of flint and bone (Parush, occurrences (Gowlett 2010), or the early interactions between Gopher, and Barkai 2016; Rosell et al. 2015), group hunting humans and fire, are evident during the Early and Middle of prime-aged animals (mainly fallow deer; Blasco et al. 2014, Pleistocene in Africa and Asia (e.g., Berna et al. 2012; Chazan 2016a), specialized butchering techniques and unique meat- 2017; Gao et al. 2017; Gowlett et al. 2017; Hlubik et al. 2017). sharing habits (Stiner et al. 2009, 2011), intensive use of bone S316 Current Anthropology Volume 58, Supplement 16, August 2017 retouchers, the habitual use of fire, hearth-centered and spatially jectories for the production of designated types of specific patterned activities, and more—all well established at Qesem sharp flakes and blades (Assaf et al. 2015; Barkai, Lemorini, and and present throughout the cave’s sequence. Qesem contains Gopher 2010; Parush et al. 2015; Parush, Gopher, and Barkai two of the three AYCC industries: the blade-dominated Amu- 2016) for targeted purposes (Lemorini et al. 2015). The scale of dian and the scraper-dominated Yabrudian. Viewing these two recycling may reach almost 10% of the debitage at Qesem (the as separate entities is at present challenged by data indicating highest densities may reach 180 recycled items per 1 m3;see their coexistence at Qesem. Below we will present a summary Gopher et al. 2016), and if recycled patinated items are added, of the data on the lithics and human remains. The fauna will the frequency rises significantly. Recycling is a well-established be presented separately because it directly relates to under- behavioral feature characterizing the AYCC (as well as earlier standing the use of fire at the cave. Acheulian technologies; see Agam, Marder, and Barkai 2015; Barkai and Gopher 2013; Shimelmitz 2015) and is most prob- ably oriented toward the intensification of cutting activities. The Lithic Assemblage As for tool typology, an “Upper Paleolithic” tool group made While Amudian-bearing levels are dominant throughout the on blades is conspicuous (Gopher et al. 2005). Large, wide, and cave’s stratigraphic sequence, the Yabrudian is less conspicuous thick blades were selected for tool shaping, and laminar-shaped and is stratigraphically and spatially restricted. Raw-material items are dominated by a variety of retouched and backed items studies were oriented to both characterizing flint types used and with few end scrapers. Yabrudian assemblages show an abun- locating their sources in the landscape. Preliminary results dance of Quina and demi-Quina scrapers making up to 50% of show the use of a large variety of flint types (over 80), some of the shaped items, while blades are less abundant. which were selected for specific types of tools or technologi- Asmallgroup(n p 16) of shaped stone balls (spheroids/poly- cal requirements. A concentration of over 10 geological sources hedrons) was found concentrated in specific stratigraphic and yielding flint types used at the cave was found within 5 km of spatial Amudian contexts (Barkai and Gopher 2016). Whether the site. Several sources are, however, located some 15 km away these were used for specific activities in specific contexts it is from the site (Wilson et al. 2016). Earlier raw-material studies too early to say. Biface production continued in the AYCC and provided evidence for flint procurement by both surface col- was quite conspicuous in the Acheulo-Yabrudian at Tabun. lection and quarrying from specific, designated primary sub- Although bifaces are indeed marginal at Qesem and appear surface sources (Verri et al. 2004, 2005). Quarried flint was as single items in both Amudian and Yabrudian assemblages directed toward the production of specific tool types (Boaretto (Barkai and Gopher 2013), they seem to have been given special et al. 2009). It is of note that the AYCC of Tabun cave has items attention. Bifaces appear in different stages—including rough- made of flint quarried from deep sources as well (Verri et al. outs. One of the outstanding roughouts is quite large (22 cm 2005). long, 15 cm wide, 10 cm thick, and 3,285 g; see Barkai et al. The Amudian blade industries provide very large samples. 2013). Blade production at Qesem shows a full chaîne opératoire,in- Archaeological evidence for knowledge transmission is cluding well-selected flat flint nodules, core shaping, and blade difficult to attain. We have suggested a change in knowledge- production, use, and discard. Amudian blades were mostly used transmission mechanisms between the Acheulian and the AYCC for cutting, butchering, and defleshing activities of soft tissues in the Levant in relation to new adaptive strategies and inno- and for a short time (Lemorini et al. 2006). Moreover, homo- vations in the lithic sphere, to hunting techniques and butch- geneity in blade production technology and in blade character- ering practices, to the habitual use of fire (firewood collec- istics was discerned throughout the Levant, reflecting a shared tion, making and maintaining fire), and especially to meat AYCC template regarding the properties of the target blades (and maybe other foods) roasting and cooking (Barkai and Go- (Shimelmitz, Barkai, and Gopher 2016). Amudian blade pro- pher 2013). These new AYCC behaviors necessitated knowledge- duction reflects a systematic, intensive, thoughtful, and straight- transmission mechanisms differing to a degree from those prac- forward technology—a conscious technological choice of skilled ticed in the Acheulian and supported by a new social milieu flint knappers constantly used for over 200 kyr (Shimelmitz, based on a possible new sociocultural discourse (see Assaf, Bar- Barkai, and Gopher 2016; Shimelmitz, Gopher and Barkai 2011). kai, and Gopher 2016; Barkai and Gopher 2013; Ben-Dor et al. Early blade industries were reported from Africa as well (John- 2011). A study of lithic knowledge transmission carried out at son and McBrearty 2010; Wilkins and Chazan 2012). The inter- Qesem recently relates to the technotypological characteristics relations between these industries and the Amudian is intrigu- of a lithic assemblage from the southern parts of the cave ing in light of a multiple origins (convergent) hypothesis for (earlier than 300 ka). The study of knapping trajectories dem- early Middle Pleistocene blade technology (Barkai and Gopher onstrated distinct features in this assemblage when compared 2013; Wilkins and Chazan 2012). with lithic assemblages from other areas of the cave. These Recycling flint is a clear component of both Amudian and features reflect various levels of knapping skills most probably Yabrudian assemblages at Qesem and in the AYCC as a whole characterizing both skilled knappers and knappers in the pro- (e.g., Shimelmitz 2015). Detailed studies of recycled items and cess of learning. This may permit a preliminary assessment of recycling products indicate technologically well-established tra- knowledge transmission relating to flint knapping that has taken Barkai et al. Fire for a Reason S317 place in a designated area (Assaf, Barkai, and Gopher 2016; Assaf for roasting and the new lithic technologies may be listed here et al. 2015). as two of the important new cultural elements related to this transformative biological and socioeconomic landscape. Qesem provides ample evidence on the environment, on Human Remains human behavioral and cultural adaptation, and on the biology To date, Qesem has yielded dental human remains only; of the hominins that inhabited it. Sealed by sediment accu- 13 teeth in total from different parts of the stratigraphic col- mulation at ca. 200 ka, shortly after desertion, and exposed only umn. These included deciduous and permanent teeth of a num- 15 years ago, Qesem preserves an outstanding potential not ber of individuals, many of whom were under the age of 20. The only for studying the communities that inhabited it but also for basic morphometric study indicates that the Qesem teeth are understanding the AYCC as a whole and its place within the clearly not of Homo erectus (sensu lato). It has highlighted that sequence between two relatively intensively studied entities— while some of the traits are more Neanderthal-like, they are the Lower Paleolithic Acheulian and the Middle Paleolithic generally similar to the Late Pleistocene local populations Mousterian. While these two may be regarded as pan-Eurasian of Skhul and Qafzeh caves dated to ca. 100 ka, (Hershkovitz phenomena, the AYCC is a local and distinct entity. Although et al. 2011, 2016). A 3-D scan of some of the teeth and various many of the AYCC lithic innovations do not continue into the subsequent analyses resulted in similar conclusions, although Mousterian and are replaced by Levallois-dominated indus- Neanderthal affinities were more emphasized in some of these tries, some of the most significant cultural and behavioral in- teeth (Fornai et al. 2016; Weber et al. 2016). While it is quite novative patterns of the AYCC do continue into later periods conceivable that the Acheulian Cultural Complex of the Le- (e.g., the habitual use of fire, hunting focused mainly on large vant was created by H. erectus (sensu lato; see Barkai and and medium-sized ungulates, meat roasting, and more). The Gopher 2013 for details), the dental evidence from Qesem, AYCC, mainly in its early stage, demonstrates a revolution, so augmented by the AYCC skull from Zuttiyeh Cave, indicates a to speak, reflecting a society open to innovative elements. New new, post-Acheulian, post-erectus hominin lineage starting at lithic technologies appeared and were maintained for a period ca. 400 ka. of 200 kyr. Known concepts such as flint recycling have been We thus see merit in our previous suggestions that the modified, changed, and intensified. We view the emergence Qesem finds and the AYCC as a whole were produced by a new of AYCC blade and Quina technologies in the Levant as an local human lineage. Obviously the question of why a biolog- original innovative behavior. The “African connection” of the ical change occurred in the Levant around 400 ka is of major Qesem assemblages (lithic and faunal) is practically nonexis- interest. Based on a bioenergetic model conjoined with the cul- tent, and it has no African counterparts. We suggest that sys- tural transformations demonstrated at Qesem, we offer an ex- tematic blade production and Quina-scraper production are planation accounting for the demise of H. erectus and the ap- local innovations aimed at manipulating medium-sized game. pearance of a new, locally evolved, post-erectus human lineage A study of Mousterian Quina scrapers from France (Claud et al. in the Levant around 400 ka (Ben-Dor et al. 2011). The model 2012) shows their use as butchering tools, and preliminary func- suggests that the disappearance of elephants from the human tional observations on the Qesem scrapers show similar results, diet in the Levant around this time triggered a selection process although hide working and bone working were observed as in favor of those who were better adapted to hunting larger well (Lemorini et al. 2016; Zupancich et al. 2016a, 2016b). numbers of smaller, faster animals with high fat content—that Amudian blades and Yabrudian scrapers, augmented by a va- is, those who were lighter and more agile. The ingredients of riety of recycling products, may thus be viewed as components this model include well-known data such as the fact that the of a new meat-cutting set that was developed in the Levant elephant is a unique and ideal food package exploited by Lower around 400 ka, replacing the long-lived Acheulian meat-cutting Paleolithic groups in the Levant for hundreds of thousands tool kit based on flakes and handaxes. This may have accom- of years. It is of note that no elephants are found in Levantine panied new hunting and meat-sharing practices following the post-Acheulian sites, which suggests this significant part of loss of calories previously obtained from elephants. This par- Acheulian life and diet had ceased in post-Acheulian times. Why ticular combination of blades and Quina scrapers (as well as elephants disappeared from the human diet in post-Acheulian recycled products) reflects a specific adaptation that has no times in the Levant remains an open question at the moment. counterparts in Africa or Europe. However, the fact that elephants are absent from all sites af- ter the Lower Paleolithic implies that rather than a culturally Fire Use at Qesem based avoidance or taboo, the most plausible explanation is that elephants were, for some reason, no longer available. Ad- The use of fire is apparent throughout the sequence at Qesem, ditionally, modern humans have known and generally accepted both directly by the large amounts of wood ash and the pres- ceilings on protein and vegetal food consumption, and fat is ence of hearths (Karkanas et al. 2007; Shahack-Gross et al. thus a compulsory component in the human diet for sufficient 2014) and indirectly by the large amounts of burnt flint and Daily Energy Expenditure (elephants are a notable package of burned bones, the organization of activities around the hearth, fat; see details in Ben-Dor et al. 2011). The habitual use of fire and the presence of charcoal fragments in human dental cal- S318 Current Anthropology Volume 58, Supplement 16, August 2017 culus. Thus, it is our contention that fire was used habitually, size, being superimposed, and bearing dense faunal and lithic commonly, and repeatedly as early as 400 ka, and one of its remains as well as evidence for spatial differentiation of activ- most important functions was most likely cooking (see Speth ities around it—provides a glance into fire-related behavior of 2012). This section presents this evidence in some detail. Middle Pleistocene humans. The sedimentary sequence post- dating this hearth at Qesem is characterized by a high content of ash. It is clear that the Qesem hominins possessed fire in the Direct Evidence for Fire Use sense of a maintainable technology. They built campfires inside The direct evidence from Qesem clearly indicates extensive, the cave, and a variety of activities were conducted in the vi- repeated use of fire between 420 and 200 ka. Micromorpho- cinity of these fireplaces. The ash-rich contents of the upper logical and isotopic evidence indicates recrystallization of wood strata as well as the abundant burned remains in the lower ash. Large quantities of burnt bone, characterized by a com- strata indicate many fire-building events, supporting the in- bination of microscopic and macroscopic criteria, and mod- terpretation of habitual use of fire in the cave. erately heated soil lumps are closely associated with the wood ash remains (Karakanas et al. 2007). All the ash structures are related to wood burning and complete combustion. Calcined The Faunal Assemblage bones are relatively common as identified, both mineralogically (using the Fourier Transform Infared Spectrometry [FTIR]) The Qesem faunal record is extremely rich in all of the layers and microscopically, by changes in color, birefringence, chem- and is dominated by fallow deer but includes red deer, horse, ical alteration, and interference colors. Such a mineralogical aurochs, wild pig, and wild ass. Although rare, small ungulates transformation occurs at either a very high temperature for a such as goat and roe deer and small prey such as birds are also short duration (above 6507C) or from prolonged combustion at present (Blasco et al. 2014, 2016a; Sánchez-Marco et al. 2016; temperatures as low as 5007C. This temperature range is com- Stiner, Gopher, and Barkai 2009 2011). Among the small prey, monly reached in campfires, and in the case of Qesem, calcined tortoises show a slightly higher level of representation (Blasco bones are mainly found in the area where the fireplace is located et al. 2016b). The faunal assemblages are characterized by an (see below). extremely rare presence of carnivores (Blasco et al. 2014, 2016a; A major component of the Qesem deposits consists of re- Stiner, Gopher, and Barkai 2009, 2011). A significant amount crystallized wood ashes. The upper ca. 4.5 m of the sequence of anthropogenic bone damage has been detected throughout consist mainly of anthropogenic sediments characterized by the stratigraphic sequence, including its earliest levels. The completely combusted, mostly reworked wood ashes associ- taphonomic characteristics of faunal remains indicate that all ated with large amounts of burnt bone, lithic artifacts, and assemblages were generated solely by humans occupying the moderately heated soil lumps. The strong cementation of the cave and were primarily damaged by their food-processing ac- deposits is explained by calcite that precipitated from dripping tivities. The ungulate mortality profile is dominated by adult-aged waters and the recrystallization of the ash. The isotopic anal- individuals, and in the case of fallow deer, the relative abun- ysis supports the presence of both preserved and recrystallized dance of infantile and young individuals suggests the devel- wood ash in the sediments (Karakanas et al. 2007). opment of seasonal hunting episodes (Blasco et al. 2014, 2016a; A central hearth was identified during fieldwork and con- Stiner, Gopher, and Barkai 2009, 2011). firmed later by mineralogical and microscopic criteria. Mi- Different types of butchery cut marks have been identified in cromorphological evidence shows two superimposed use cycles the form of incisions, sawing marks, scraping marks, and chop each composed of shorter episodes, possibly the earliest super- marks (ratios between !2% and 12% according to Blasco et al. imposed hearth securely identified to date (Shahack-Gross et al. 2014, 2016a, and Stiner, Gopher, and Barkai 2009). The un- 2014). The hearth covers ca. 4 m2, making it a uniquely large gulate specimens show cut marks distributed over most of the hearth in comparison with any contemporaneous hearth iden- skeletal elements, especially on limb bones (fig. 1). The loca- tified thus far. The hearth is located in the center of the cave and tions of the cut marks indicate that both long-bone epiphyses is associated with butchered animal remains and a dense flint and shaft fragments bear cuts, although there is a clear pre- assemblage. The central location of the hearth within the cave dominance of damage on limb-shaft fragments. Tortoise bones and the activities associated with it may reflect a pattern of the from the earliest levels show a relatively high rate of cut marks organization of space by the cave inhabitants. An array of in- (13.2%) on the shell and limbs (Blasco et al. 2016b). dependent lines of evidence was used to analyze this large, The faunal assemblage also includes damage caused during central, repeatedly used combustion feature. The evidence for bone breakage to access marrow (fig. 2). The studied samples this feature being a hearth is primarily the two microlaminated have preserved diagnostic elements of intentional bone break- gray-white layers composed of in situ wood ash that includes age of both long and flat bones, although, as in the case of cut charred and calcined bones, microcharcoal, burnt flint, and marks, the limb shafts show the highest proportions of damage. burnt microscopic clay aggregates. The micro-FTIR data in- The bone surface modification resulting from the anthropo- dicate a temperature range that exceeded 5007C (Shahack- genic breakage includes percussion pits, notches, impact flakes Gross et al. 2014). This central hearth—having a uniquely large (cortical flakes and scars included), counterblows, and peeling. Barkai et al. Fire for a Reason S319

Figure 1. Examples of burned bones bearing cut marks from the hearth faunal assemblage (A–E) and the earliest levels of the cave (F–H). A, Long-bone fragment of a small-sized animal showing “sandwich” coloration. B, Long-bone fragment with double coloration on the same surface. C, Proximal epiphyses of metapodial bones. D, Different burning tonalities on the same bone surface. E, Costal- bone fragment of tortoise showing burning (degree 3) on the dorsal surface while ventral surface remains unburned. Long-bone fragments of small-sized animals with degree 2 (F), degree 5 (G), and double coloration (degrees 0 and 3; H). A color version of this figure is available online.

Nine bone fragments from the hearth unit and 15 from earlier caused by the use of bone as a retoucher (Blasco et al. 2013a; levels show percussion marks related to the shaping of stone Rosell et al. 2015). tools. All of these items correspond to the long-bone shafts of Burning damage affects more than 30% of the bone frag- small, medium, and large animals, showing damage typically ments in all assemblages, including the earliest ones (figs. 1, 3). Figure 2. Experimental marrow removal through hammer-stone percussion (top) and fossil examples of damage generated during bone breakage to extract marrow in form of percussion notches on limb bones from the hearth assemblage (A), area around the hearth (B), and area under the shelf (C) at Qesem Cave (bottom). A color version of this figure is available online. Barkai et al. Fire for a Reason S321

Figure 3. Experimental examples of roasting meat (left) and experimental example of calcined bones (degree 5) after being thrown to the hearth for cleaning (right). Note the calcined bones become more fragile and susceptible to breakage, a circumstance leading to their fragmentation and destruction in some cases. A color version of this figure is available online.

In the case of the central hearth, this figure rises to 63.95% of 2013), the use of bone as fuel (Morin 2010; Théry-Parisot 2002), the specimens (Blasco et al. 2014). For the study of this specific the preparation of bones to facilitate their breakage (Caceres modification, macroscopic criteria based on color changes have et al. 2002), or the preparation of bone marrow for removal been used. According to several authors, two main variables (Oliver 1993; Speth et al. 2012). Burned bones could also be the seem to influence coloration, namely, the heat intensity and the unintended (accidental) consequence of postdepositional dam- exposure time (e.g., Brain 1981; Buikstra and Swegle 1989; Gil- age such as secondary burning when fireplaces are set up on christ and Mytum 1986; Johnson 1989; Mayne 1997; Nicholson bones buried close to the surface (e.g., Aldeias 2017; Bennett 1993; Shahack-Gross, Bar-Yosef and Weiner 1997; Shipman, 1999; Stiner, Kuhn, and Surovell 2001). Foster, and Schoeninger 1984; Spennemann and Colley 1989; The task becomes complicated when several of these pro- Stiner et al. 1995). The different responses of the organic and cesses occur during the formation of the same sedimentary inorganic components of the bones to the rise in temperature package. The superposition of different processes could mask lead to color changes and different chromatic stages (mainly the initial hominin activities, such as the roasting of meat before brown, black, gray, and white; e.g., David 1990; Grayson et al. defleshing. As in the case of tortoises, we have tried to detect 1988; Lyman 1994). For Qesem, the degree of alteration due to differential burning patterns on ungulate bones, that is, double burning has been classified into six categories of intensity ac- colorations on the cortical surface and no alteration on the cording to the bones’ color, structure, and homogeneity, with medullary surface, as it is expected that if bone is exposed when degree 0 representing unburned and degree 5 representing cal- meat is on the fire for roasting, the medullary surface will re- cined bones. “Sandwich” coloration has also been detected at main unchanged and the exposed areas (or those covered by Qesem, characterized by different coloration in the intracorti- only a thin meat layer) will be affected more intensely by the cal tissue with respect to that observed on the outer sides. This heat, thus acquiring a higher degree of coloration (Gabucio occurs when the skeletal element is subjected to partial thermal et al. 2014; Gifford-Gonzalez 1989; Rosell et al. 2012). Despite action or incomplete combustion while maintaining its fat con- complexities, a pattern based on the absence of burning on tent (Cerdá, García-Prósper, and Serra 2005). the medullar surface of ungulate limb bones was observed at Thermal alteration has been observed on every type of skel- Qesem. This pattern appears on bones that do show heat alter- etal element, albeit with a definite predominance on the long- ation (uniform or double coloration) on their cortical surfaces bone shafts of medium- and small-sized ungulates. Degree 3 (7.32% of specimens bearing differential burning patterns from represents the most abundant damage, followed by degree 2, the hearth archaeological context). In addition, some of these while degrees 1 and 5 are the least represented (Blasco et al. bones (2.1%) show a higher degree of burning on those parts that 2014, 2016a; Stiner, Gopher, and Barkai 2011). Some specimens would normally be exposed to fire when bone is placed on the bear “sandwich” coloration predominantly on the long bones. embers to roast the meat adhering to it (e.g., metaphysis). This The presence of burning on bone fragments might reflect fact has been interpreted as the result of episodes related to several phenomena. The challenge is certainly to identify roast- roasting. ing (e.g., Alhaique 1997; Solari et al. 2015; Speth 2006), as other The roasting “signal” could be viewed as weak if we assume intentional processes could lead to burning, such as the removal that only one fire-related activity occurred in the cave. How- of waste for cleaning purposes (e.g., Yravedra and Uzquiano ever, the development of a single activity is unlikely to be iden- S322 Current Anthropology Volume 58, Supplement 16, August 2017 tified for the following reasons: (1) the presence of burned as a model to analyze and elucidate this behavioral pattern bones over the entire excavation surface and volume, (2) al- (Blasco et al. 2014, 2016a). The succession of cycles of com- terations affecting all bone surfaces (both medullar and corti- bustion at the same location in the cave suggests a repeated cal), (3) structural changes of the bone tissues resulting from behavior and a patterned use of space during recurrent human intense exposure to fire (e.g., fissures, cracking, shrinkage), and occupations. This resulted in a significant quantity of faunal (4) significant concentration of burned material in the area of and lithic remains as well as evidence of the spatial differen- the fireplace (which includes the highest degrees of thermal tiation of activities around the hearth. The hearth and the area alteration [calcined bones degrees 4, 5]). In relation to this, south of it cover an area of approximately 15 m2 excavated to a Mentzer (2009) proposes that the bone fragments become maximum depth of 60 cm. The faunal assemblage comprises completely calcined when they are directly exposed to flame. 37,304 specimens (2,995 specimens, or 8.03%, have been iden- This might occur when bones are thrown into the hearth for tified to the species level) of which 15,464 come directly from cleaning (Yravedra and Uzquiano 2013), when fireplaces are set the hearth and 21,840 from surrounding zones. The faunal up on unburied or semiburied bones (Bennett 1999), or when assemblage includes 15 taxa and a minimum number of indi- bones are used as fuel (Costamagno et al. 2009). As noted in viduals (MNI) of 81 (fallow deer MNI p 41, red deer MNI p 8, Blasco et al. (2016b:204), these possibilities are not mutually horse MNI p 6, aurochs MNI p 5, wild pig MNI p 3, wild exclusive, as many processes may have occurred concurrently ass MNI p 3, rhinoceros MNI p 2, goat MNI p1, roe deer or sequentially, leading to the overlapping of bone alterations MNI p 2, large bird MNI p 3, Carnivore MNI p 1, Cervidae and even to the destruction of elements or signatures after a MNI p 2, and the tortoise MNI p 4; Blasco et al. 2014, 2016a). specific primary human activity. For example, some burned Perhaps one of the best examples of spatial differentiation bones show no homogeneous degree 2 coloration (brown), is the plot based on the size of bone fragments in the assem- which coincides with alterations described by Bennett (1999) blage and the degree of thermal alteration of the bones. Density of specimens burned after burial, indicating nonnutritive epi- maps were generated using geographic information systems sodes that occurred following the processing sequence. The set (ArcGIS 10) and include proximity calculations such as kernel of thermal alterations at Qesem certainly shows the complexity estimation or the nearest neighbor algorithm (see details in of its taphonomic history, where more than one fire-related Blasco et al. 2016a). Burned bones, mainly those showing a process, roasting included, seem to have occurred. higher degree of damage (degrees 4 and 5), are clustered in the The case of Testudo sp. is particularly significant because main combustion area. In contrast, the area around the hearth 52.9% of the burned tortoise bones from the central hearth makes up less than 1% of the total number of specimens re- showed thermal alteration on different surfaces, and the highest trieved with this degree of damage (Blasco et al. 2016a). This degree of burning was observed on the dorsal surfaces. This apparent organization of the remains is most obvious if we con- pattern was especially pronounced in the earliest levels of the sider the length of the bone fragments. Although the smallest cave, as 83.33% of the double-colored bones showed a higher specimens (!20 mm, the most abundant in the assemblage) degree of burning on the dorsal than on the ventral surface, are distributed over the entire occupied surface, the highest which tends to appear unburned (Blasco et al. 2016b). Although concentration is observed in the hearth area. Yet a most sig- a certain degree of variability is to be expected, this pattern fits nificant observation is the distribution of large bone fragments well with the idea of roasting tortoises by placing the whole (140 mm) in the outer area (Blasco et al. 2016a). This spatial animal upside down directly on the embers and allowing it to distribution seems to fit roughly with the model of cultural cook in the shell. These taphonomic patterns can be used as formation of hearth-related assemblages observed by Binford examples of new human-fauna relationships during the AYCC, (1978, 1983) in Nunamiut camps. The drop area is character- which include cooking as a regular component of the human’s ized by small bone splinters and lithic fragments resulting from behavioral repertoire. different domestic activities, such as bone breakage for marrow extraction or the processes of core reduction and stone tool shaping. The toss zone, in contrast, consists of larger fragments The Organization of Human Activities around the Central that have been intentionally tossed away to areas farther re- Hearth: The Faunal and Lithic Evidence moved from the activity areas. On this basis, a tentative stan- Spatially, the central hearth is an evident focus of intensive dardized pattern can be observed along the sedimentary for- activities and is very dense in both faunal and lithic finds. We mation of the hearth. Both the spatial distribution around the focus on the hearth itself and the area adjacent to it to the south, hearth and the subsistence strategies can be considered as fac- which seems to be related to the hearth (Blasco et al. 2016a; tors linked to the emergence of reference places of a residential Shahack-Gross et al. 2014). A preliminary spatial analysis of character—that is, places that would fit with the concept of the faunal remains in this area has recently been attempted (Blasco home base discussed by Rolland (2004). et al. 2016a), and the lithic spatial distribution is under study. Results of spatial density data based on a study of lithics at The abundant thermally altered items lead us to infer a well- 18 assemblages throughout the stratigraphic column and in implemented use of fire at Qesem through its whole strati- different parts of the cave included the fireplace area and the graphic sequence. The central hearth area has been used here area to its south (see Gopher et al. 2016). When lithic densities Barkai et al. Fire for a Reason S323 of the hearth area and the area to the south of it are compared of fire for roasting/cooking by the Qesem humans. Char- with the studied assemblages throughout the cave or, more coal microparticles up to 70 mm in diameter enter the mouth specifically, with four assemblages of similar stratigraphic po- during oral breathing, and while these fragments could also sition and roughly contemporary, some interesting results are result from ingestion of char adhering to roasted food, their evident. The lithic assemblage of the hearth area consists of size suggests they result from accidental inhalation (Hardy et al. 18,837 items and shows the highest density of all the assem- 2016). These findings, combined with the sedimentary and mi- blages of the cave (6,144 lithic items per m3 for the hearth it- cromorphological evidence from the cave, are indicative of self; see Gopher et al. 2016), indicating intensive lithic pro- fire and suggest a smoky atmosphere inside the cave. While duction, use, and discard in this area. This is reflected in the placing the hearth in a central place well inside the cave may relatively high density of cores and core trimming elements be linked to the increased intensity of roasting/cooking, flint (CTEs) as well (61 and 121 per m3, respectively). The area to the knapping, animal processing, and most, presumably, social in- south of the hearth is somewhat lower in density (3,106 items teractions, the need for smoke management became necessary. per m3) but shows high densities of cores and CTEs (37 and A smoky atmosphere can be an irritant and can at times cause 63 per m3, respectively). A conspicuous aspect of the hearth serious health problems. Smoke can cause coughing and eye area and the area south of it is the high density of cutting tools, irritation in addition to potentially more serious lung prob- including blades and naturally backed knives (NBKs) made on lems. For a successful use of fire to develop in an internal (cave) blades (both showing a density of 77 per m3) as well as NBKs location and fulfill its potential functional and social purposes made on flakes that are prominently dense in the area (98 per in a way that permitted the population to thrive, the health risks m3). Another outstanding aspect is the fact that the highest must be resolved. density of recycling is evident in the hearth area, including both “ ” the recycled parents and the recycling products (45 and Discussion and Conclusions 142 per m3, respectively), and south of the hearth (32 and 134 per m3, respectively). Many of the products of recycling The Late Lower Paleolithic period in the Levant, and the AYCC have shown meat-cutting use wear (Lemorini et al. 2015), and in particular, was a period of transformation in human biology the relatively high density of blades and NBKs (Lemorini et al. and culture. The habitual, common, and repeated use of fire 2006) in this area suggests that they could be interpreted as a for roasting meat (and possibly cooking in general) should be set of meat-cutting tools (Barkai, Lemorini, and Gopher 2010) conceived as an integral part of this set of transformations densely concentrated at the meat-roasting area. Interestingly, rather than an isolated phenomenon. We believe that a better the hearth shows a medium-low density of shaped items (tools), acquaintance with the environmental, behavioral, and adaptive while the area south of the hearth shows a very low density of contexts of the use of fire at ca. 400 ka in the Levant will pro- shaped items. This may indicate a frequent use of unshaped mote comprehension of this new aspect and its significance in items (mostly characterized by sharp edges) in the hearth area human evolution. and south of it. We may add that while blades and other cutting We claim that in the AYCC starting sometime at 400 ka, the tools (including recycling products) are conspicuous in the use of fire was clearly aimed at meat roasting and possibly hearth assemblage itself and the area south of it, assemblages cooking in general, and this innovation continued in the Le- of similar stratigraphy to the west and northwest of the hearth vant after the AYCC (see a statement on cooking as a Eurasian are poorer in blades and NBKs, poorer in recycled items and Middle Pleistocene innovation in Dennell 2009:476–477). A recycling products, and richer in shaped tools, including a con- detailed discussion of the sociocultural aspects of fireplaces spicuously high density of scrapers (27–43 scrapers per m3 as central foci of human activity is beyond the scope of this compared to 6 and 5 scrapers per m3 in the hearth and the area paper (see Wiessner 2014), but the fact that a central hearth to the south of it). This may indicate the use of blades and of was exposed is of major importance in this respect. The use of recycling products in some areas around the hearth while fire at Qesem is first and foremost related to the dietary scraper-related activities have taken place in other distinctive practices of the cave’s occupants. The diet of the Qesem inhab- and separated (though nearby and contemporaneous) areas of itants had a major meat component as reflected in the large the cave. number of animal bones and the dietary reconstruction we suggested (Ben-Dor et al. 2011). While floral remains were not preserved, we succeeded in extracting ingested floral remains Inhaled Charcoal in Dental Calculus from human dental calculus, including starch granules and Potentially inhaled and ingested materials were extracted from chemical compounds, providing a direct link to ingested plant dental calculus of the Qesem humans, including, among other food containing essential nutrients, including polyunsaturated things,microcharcoal,starchmolecules,andPinuspollen.These fatty acids and carbohydrates (Hardy et al. 2016). Following finds offer an insight into human diet and the environment the sources on which we based our “fat hunter” model, a hu- around the cave, and the microcharcoal highlights the need man diet based on meat and fat must have been complemented for smoke management in the enclosed environment of the by carbohydrates and thus included three major ingredients: cave—a challenge introduced after adopting the habitual use animal proteins, animal fat, and vegetal food. We have no S324 Current Anthropology Volume 58, Supplement 16, August 2017 doubt concerning vegetal food consumption in the Paleolithic ing and hunting capabilities took center stage. After hunting, (e.g., Melamed et al. 2016), and the above-quoted study sup- specific butchering practices characterized the AYCC at Qesem, ports this view, providing direct evidence for the consumption and these had to be culturally transmitted as well (Blasco et al. of vegetal food. A point in order here are results of a study of 2013b). Last but not least, the habitual use of fire in the AYCC masticatory wear patterns on human teeth carried out recently brought about a new set of knowledge and capabilities that that provided evidence suggesting that the Qesem people pos- had to do with firewood collection, production and mainte- sessed a strong masticatory system and fed on a wide range of nance of fire, ventilation of the cave in order to reduce pol- food types (Sarig et al. 2016). lution, and more, so the cost of making and keeping fire must A major element in our argument has to do with the absence also be taken into account (Henry 2017). Of course, roasting of elephants from AYCC sites. Elephants were part of the diet and cooking meat (and maybe other foods) had to be cultur- of Acheulian humans, and elephant bones, alongside other ally transmitted, too. It is our contention, therefore, that the taxa, were found in Acheulian sites in the Levant throughout multitude of new AYCC adaptations necessitated elaborate the 1-million-year span of the Acheulian Cultural Complex in knowledge-transmission mechanisms quite possibly different the region. Elephant bones are present in early, middle, and from those practiced by the Acheulians and supported by a late Acheulian sites (Ubeidya, Evron, Gesher Benot Ya‘aqov, new social milieu. Revadim, and Holon, respectively), and thus we may say that We suggest that the disappearance of the elephants at ca. Acheulian human presence in the Levant is constantly asso- 400 ka triggered the hunting of an increased number of smaller ciated with elephants and their use as a source of food. It is and faster animals to maintain protein supply and an adequate of course possible that elephants were not consumed at every fat content in the diet, and this was the evolutionary drive be- Acheulian site and that in certain times and at certain places, hind the emergence of the lighter, more agile, and cognitively Acheulian humans in the Levant survived with no elephant capable new humans in the Levant. Homo erectus in the Levant meat and fat on the menu. However, this was the exception must have been perfectly adapted to a diet in which the con- rather than the rule. Sites of the AYCC as well as later sites in sumption of elephant meat played a significant role. The need the Levant lack elephant remains. No elephant bones were to hunt larger numbers of selected medium-sized individuals found at any post-Acheulian site in the Levant, and they were with high fat content might have encouraged new social rela- not part of the post-Acheulian human diet, which was based tions based on new meat-sharing practices. The habitual use on medium-sized, prime-aged animals both in the late Lower of fire for roasting meat and cooking might be connected to the Paleolithic AYCC and the later Middle Paleolithic Mousterian need to extract more calories from every food item (e.g., Car- (e.g., Speth 2012). mody, Weintraub, and Wrangham 2011; Carmody and Wrang- The significant changes in human behavior during the AYCC ham 2009; Groopman, Carmody, and Wrangham 2015; Wrang- necessitated new knowledge-transmission mechanisms in or- ham 2017), and the new lithic technologies might have been der to cope with the many new aspects of behavior adopted. In aimed at a better manipulation of smaller game. The use of addition to the production of flakes and bifaces, AYCC hu- mans had to learn how to produce blades and Quina scrapers following strict standards. Moreover, knowledge and skills re- garding the identification of flint sources and quarrying tech- niques and procedures had to be transmitted as well as the concept and practice of flint recycling. The focus on hunting prime-aged fallow deer (highest fat content) necessitated pre- cise identification of specific deer to be targeted according to the color of the fur and the brightness of the skin. It is not without reason that the Saami of northern Norway, for example, use more than 600 words for describing reindeer according to their age, sex, color, coat, antlers, etc. (Clottes 2013). Tracking and hunting selected fallow deer must have been a practice based on specific knowledge and experience. Because we believe that Acheulian hominins hunted game including elephants and medium-sized animals, it goes without saying that parts of the tracking and hunting procedures of the AYCC were already practiced in the preceding Acheulian. However, because ele- phants contain large quantities of fat year round (Ben-Dor et al. fl 2011), fat-content-related choices had been marginal. When Figure 4. Experimental demi-Quina int scraper replica used to remove meat from bones throughout both longitudinal and trans- elephants were no longer consumed in the AYCC and later, it versal motions. An experiment conducted in the framework of a made a significant difference which deer was being hunted in use-wear study of the Qesem cave scrapers. Courtesy of Andrea order to supply not only meat but also fat, and thus new track- Zupanchich. A color version of this figure is available online. Barkai et al. Fire for a Reason S325

Figure 5. Different moments during a butchery activity using replicas of small recycled flint items. a,Defleshing hide of a wild boar. b, Disarticulating bone of a wild boar. c, Removing meat from a cow bone. d, e, Processing a sheep metapodium to obtain tendons. Experiment conducted in the framework of a use-wear study of the Qesem cave recycled items. Courtesy of Flavia Venditti. A color version of this figure is available online.

Quina scrapers and small cutting tools made of recycled items mode of adaptation in the Levant. This human trait must have might have been related to newly introduced processing meth- been highly successful, as it characterizes human behavior and ods (figs. 4, 5). Thus, the circumstances for the appearance of adaptation to this very day. the AYCC in the Levant and its characteristics provide a con- text for the earliest manifestations of the habitual use of fire for roasting and cooking meat. The AYCC provided a context open Acknowledgments for innovations and transformations, one of which was the adoption of fire. Such a major change in human behavior and We would like to thank Dennis Sandgathe and Francesco Berna, lifeways must have been negotiated and carefully tested before the organizers of the symposium, for their warm hospitality being habitually adopted, and this is the case for many of the and kindness during the symposium and for the opportunity other transformations that took shape during AYCC times. We to publish our work. We thank Leslie Aiello and Laurie Ob- regard Acheulian human adaptation as highly successful and bink for a perfectly organized and planned symposium carried as having enabled Lower Paleolithic communities in the Levant out in good spirit and with dedication. Special thanks go to the to thrive for over 1 million years without using fire commonly workshop participants for many hours of discussions, laughter, and continuously and without roasting meat. Modifying this and even some disagreements that made the time enjoyable and long-lived mode of adaptation and introducing a set of inno- fruitful. The Qesem Cave excavation project is supported by the vations such as the one characterizing the Levantine AYCC Israel Science Foundation, the CARE Archaeological Founda- must have been for very good reasons. At around 400 ka in the tion, the Leakey Foundation, the Wenner-Gren Foundation, Levant, human groups developed a new (post-Acheulian) ad- the Dan David Foundation, and the German Research Foun- aptation mode that enabled them to thrive for another 200 kyr, dation. J. Rosell and R. Blasco develop their work within the until the next set of transformations took place, as reflected by Spanish Ministerio de Economía y Competitividad/Fondo Eu- the appearance of Middle Paleolithic Mousterian lifeways. We ropeo de Desarrollo Regional projects CGL2015-65387-C3-1-P thus see the introduction, assimilation, and adoption of fire as and CGL2015-68604-P, the Generalitat de Catalunya-Agència a common human trait used for cooking meat as one of the de Gestió d’Adjuts Universitaris I de Recerca projects 2014 SGR innovative aspects that characterized the new post-Acheulian 900 and 2014/100573, and the SENECA Foundation project S326 Current Anthropology Volume 58, Supplement 16, August 2017

19434/PI/14. We thank Andrea Zupanchich and Flavia Vend- at both edges of the Mediterranean Sea. PLoS ONE 8(10):e76780, doi:10.1371 /journal.pone.0076780. itti for providing us illustrative material from their experimen- Blasco, Ruth, Jordi Rosell, Manuel Domínguez-Rodrigo, Segi Lozano, Ignasi tal work. Pastó, David Riba, and Eudald Carbonell. 2013b. Learning by heart: cultural patterns in the faunal processing sequence during the Middle Pleistocene. PLoS ONE 8(2):e55863. References Cited Blasco, Ruth, Jordi Rosell, Avi Gopher, and Ran Barkai. 2014. Subsistence economy and social life around the hearth: a zooarchaeological perspective Agam, Aviad, Ofer Marder, and Ran Barkai. 2015. Small flake production and from Middle Pleistocene (300 ka) Qesem Cave, Israel. Journal of Anthro- lithic recycling at late Acheulian Revadim, Israel. Quaternary International pological Archaeology 35:248–268. 361:46–60. Blasco, Ruth, Jordi Rosell, Avi Gopher, Pablo Sañudo, and Ran Barkai. 2016a. Aldeias, Vera. 2017. Experimental approaches to archaeological fire features What happens around a fire: faunal processing sequences and spatial distri- and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– bution at Qesem Cave (300 ka), Israel. Quaternary International 398:190. –209 S205. Blasco, Ruth, Jordi Rosell, Krister T. Smith, Lutz Christian Maul, Pablo Sañudo, Alhaique, Francesca. 1997. Do patterns of bone breakage differ between cooked Ran Barkai, and Avi Gopher. 2016b. Tortoises as a dietary supplement: a view and uncooked bones? an experimental approach. Anthropozoologica 25/ from the Middle Pleistocene site of Qesem Cave, Israel. Quaternary Science 26:49–56 Reviews 133:165–182. Alperson-Afil, Nira. 2017. Spatial analysis of fire: archaeological approach to Boaretto, Elisabetta, Ran Barkai, Avi Gopher, Francenco Berna, and Steve recognizing early fire. Current Anthropology 58(suppl. 16):S258–S266. Weiner. 2009. Specialized flint procurement strategies for hand axes, scrapers Alperson-Afil, Nira, and Naama Goren-Inbar. 2010. Ancient flames and con- and blades in the late Lower Paleolithic: a 10Be study at Qesem Cave, Israel. trolled use of fire, vol. 2 of The Acheulian site of Gesher Benot Ya‘aqov. New Journal of Human Evolution 24(1):1–12. York: Springer Science. Bourguignon, Laurence. 2001. Apports de l’expérimentation et de l’analyse Assaf, Ella, Ran Barkai, and Avi Gopher. 2016. Knowledge transmission and techno-morpho-fonctionnelle à la reconnaissance des processus d’aménage- apprentice flint-knappers in the Acheulo-Yabrudian: a case study from Qesem ment de la retouche Quina. In Préhistoire et approche expérimentale. Laurence Cave, Israel. Quaternary International 398:70–85. Bourguignon, llluminada Ortega, and Marie-Chantal Frère-Sautot, eds. Assaf, Ella, Yoni Parush, Avi Gopher, and Ran Barkai. 2015. Intra-site re- Pp. 35–66. Préhistoires no. 5. Montagnac: Monique Mergoil Editions. cycling variability at Qesem Cave, Israel: new evidence from an Amudian Brain, Charles K. 1981. The hunters or the hunted? an introduction to African and Yabrudian assemblages. Quaternary International 361:88–102. cave taphonomy. Chicago: University of Chicago Press. Barkai, Ran, and Avi Gopher. 2011. Innovative human behavior between Buikstra, Jane E., and Mark Swegle. 1989. Bone modification due to burning: Acheulian and Mousterian: a view from Qesem Cave, Israel. In The Lower and experimental evidence. In Bone modification. Robson Bonnichsen and Mar- Middle Paleolithic in the Middle East and neighboring regions. Jean-Marie Le cella H. Sorg, eds. Pp. 247–258. Orono: Center for the Study of the First Tensorer, Reto Jagher, and Marcel Otte, eds. Pp. 121–130. ERAUL, 126. Liège: Americans, Institute for Quaternary Studies, University of Maine. Centre de recherches archéologiques de l’Universitéde Liège. Cáceres, Isabel, Pilar Bravo, Montse Esteban, Isabel Exposito, and Palmira ———. 2013. Cultural and biological transformations in the Middle Pleis- Saladie. 2002. Fresh and heated bones breakage: an experimental approach. tocene Levant: a view from Qesem Cave, Israel. In Dynamics of Learning in In Current topics on taphonomy and fossilization. M. de Rienzi et al., eds. Neanderthals and Modern Humans, vol. 1. Pp. 115–137. Tokyo: Springer. Pp. 471–481. Valencia: Ayuntamiento de Valencia. ———. 2016. On anachronism: the curious presence of spheroids and poly- Carmody, Rachel N., Gil S. Weintraub, and Richard W. Wrangham. 2011. En- hedrons at Acheulo-Yabrudian Qesem Cave, Israel. Quaternary Interna- ergetic consequences of thermal and nonthermal food processing. Proceed- tional 398:118–128. ings of the National Academy of Sciences of the USA 108(48):19199–19203. Barkai, Ran, Avi Gopher, Stein-Erik Lauritzen, and Amos Frumkin. 2003. Carmody, Rachel N., and Richard Wrangham. 2009. The energetic signifi- Uranium series dates from Qesem Cave, Israel, and the end of the Lower cance of cooking. Journal of Human Evolution 57:379–391. Paleolithic. Nature 423:977–979. Cerdá, Manuel Polo, Elisa García-Prósper, and Alfred Sanchis Serra. 2005. Barkai, Ran, Avi Gopher, Natalia Solodenko, and Cristina Lemorini. 2013. An Estudio bioarqueológico de las cremaciones del monumento funerario ro- Amudian oddity: a giant biface from late Lower Palaeolithic Qesem cave. mano del “solar de la morería” de Sagunto. ARSE: Boletín anual del Centro Tel Aviv 40(2):176–186. Arqueológico Saguntino 39:229–268. Barkai, Ran, Cristina Lemorini, and Avi Gopher. 2010. Palaeolithic cutlery Chazan, Michael. 2017. Toward a long prehistory of fire. Current Anthro- 400 000–200 000 years ago: tiny meat-cutting tools from Qesem Cave, pology 58(suppl. 16):S351–S359. Israel. Antiquity 84:325. Claud, Émilie, Marie Soressi, Jacques Jaubert, and Jean-Jacques Hublin. 2012. Bar-Yosef, Ofer. 1994. The Lower Paleolithic of the Near East. Journal of World Étude tracéologique de l’outillage moustérien de type Quina du bonebed de Prehistory 8(3):211–265. Chez-Pinaud à Jonzac (Charente-Maritime): nouveaux éléments en faveur ———. 2006. The known and unknown about the Acheulian. In Acheulian d’un site de boucherie et de traitement des peaux. Gallia Préhistoire 54:3–32. tool-making from quarry to discard. Naama Goren-Inbar and Gonen Sharon, Clottes, Jean. 2013. Why did they draw in those caves? Time and Mind 6(1):7–14. eds. Pp. 479–494. Approaches to Anthropological Archaeology. London: Copeland, Lorraine. 2000. Yabrudian and related industries: the state of re- Equinox. search in 1996. In Toward modern humans: Yabrudian and Micoquian, 400– Bar-Yosef, Ofer, and Steven L. Kuhn. 1999. The big deal about blades: laminar 50 k years ago. Avraham Ronen and Mina Weinstein-Evron, eds. Pp. 97–118. technologies and human evolution. American Anthropologist 101(2):322–338. BAR International Series 850. Oxford: Archaeopress. Ben-Dor, Miki, Avi Gopher, Israel Hershkovitz, and Ran Barkai. 2011. Man the Costamagno, Sandrine, Isabelle Théry-Parisot, Jean-Christophe Castel, and fat hunter: the demise of Homo erectus and the emergence of a new hominin Jean-Philippe Brugal. 2009. Combustible ou non? analyse multifactorielle et lineage in the Middle Pleistocene (ca. 400 ka) Levant. PLoS ONE 6(12): modèles explicatifs sur des ossements brûlés paléolithiques. In Gestion des e28689, doi:10.1371/journal.pone.0028689. combustibles au Paléolithique et au Mésolithique: nouveaux outils, nouvelles Bennett, Joanne L. 1999. Thermal alteration of buried bone. Journal of Archae- interprétations. Isabelle Théry-Parisot, Sandrine Costamagno, and Auréade ological Science 26:1–8. Henry, eds. Pp. 65–84. BAR International Series 1914. Oxford: Archaeopress. Berna, Francesco, Paul Goldberg, Liora K. Horwitz, James Brink, Sharon Holt, David, Bruno. 1990. How was this bone burnt? In Problem solving in taphon- Marion Bamford, and Michael Chazan. 2012. Microstratigraphic evidence omy: archaeological and palaeontological studies from Europe, Africa and of in situ fire in the Acheulean strata of Wonderwerk Cave, Northern Cape Oceania. Su Solomon, Iain Davidson, and Di Watson, eds. Pp. 65–79. St. Lu- Province, South Africa. Proceedings of the National Academy of Sciences of cia: Anthropology Museum, University of Queensland. the USA 109(20):E1215–E1220. Dennell, Robin. 2009. The Paleolithic settlement of Asia. Cambridge: Cambridge Binford, Lewis. R. 1978. Nunamiut ethnoarchaeology. New York: Academic University Press. Press. Endicott, Phillip, Simon Y. W. Ho, and Chris Stringer. 2010. Using genetic evi- ———. 1983. In pursuit of the past: decoding the archaeological record. Lon- dence to evaluate four palaeoanthropological hypotheses for the timing of Ne- don: Thames & Hudson. anderthal and modern human origins. Journal of Human Evolution 59:87–95. Blasco, Ruth, Jordi Rosell, Felipe Cuartero, Josep Fernández-Peris, Avi Gopher, Falguères, Christophe, Maïlys Richard, Olivier Tombret, Qingfeng Shao, Jean- and Ran Barkai. 2013a. Using bones to shape stones: MIS 9 bone retouchers Jacques Bahain, Avi Gopher, and Ran Barkai. 2016. New ESR/U-series Barkai et al. Fire for a Reason S327

dates in Yabrudian and Amudian layers at Qesem cave, Israel. Quaternary Hlubik, Sarah, Francesco Berna, Craig Feibel, David Braun, and John W. K. International 398:6–12. Harris. 2017. Researching the nature of fire at 1.5 Mya on the site of FxJj20 Fernández Peris, Josep, Virginia Barciela, Ruth Blasco, Felipe Cuartero, Han- AB, Koobi Fora, Kenya, using high-resolution spatial analysis and FTIR nah Fluck, Pablo Sañudo, and Carlos Verdasco. 2012. The earliest evidence spectrometry. Current Anthropology 58(suppl. 16):S243–S257. of hearths in southern Europe: the case of Bolomor Cave (Valencia, Spain). Jelinek, Arthur J. 1982. The Middle Paleolithic in the Southern Levant, with Quaternary International 247:267–277. comments on the appearance of modern Homo sapiens.InThe transition Fornai, Cinzia, Stefano Benazzi, Avi Gopher, Ran Barkai, Rachel Sarig, Fred from Lower to Middle Paleolithic and the origin of modern man. Avraham L. Bookstein, Israel Hershkovitz, and Gerhard W. Weber. 2016. The Qesem Ronen, ed. Pp. 57–101. BAR International Series 151. Oxford: British Ar- Cave hominin material. 2. A morphometric analysis of dm2-QC2 decidu- chaeological Reports. ous lower second molar. Quaternary International 398:175–189. ———. 1990. The Amudian in the context of the Mugharan tradition at the Freidline, Sarah E., Philipp Gunz, Ivor Janković, Katerina Harvati, Max Planck Tabun Cave (Mount Carmel), Israel. In The emergence of modern humans. Hublin. 2012. A comprehensive morphometric analysis of the frontal and Paul Mellars, ed. Pp. 81–90. Ithaca, NY: Cornell University Press. zygomatic bone of the Zuttiyeh fossil from Israel. Journal of Human Evo- Johnson, Cara Roure, and Sally McBrearty. 2010. 500,000 year old blades from lution 62:225–241. the Kapthurin Formation, Kenya. Journal of Human Evolution 58:193–200. Frumkin, Amos, Panagiotis Karkanas, Miryam Bar-Matthews, Ran Barkai, Johnson, Eileen. 1989. Human modified bones from early southern plains sites. Avi Gopher, Ruth Shahack-Gross, and Anton Vaks. 2009. Gravitational In Bone modification. Robson Bonnichsen and Marcella H. Sorg, eds. Pp. 431– deformations and filling of aging caves: the example of Qesem Cave karst 471. Orono: Center for the Study of the First Americans, Institute for Qua- system, Israel. Geomorphology 106:154–164. ternary Studies, University of Maine. Gabucio, Joana, Isabel Cáceres, Jordi Rosell, Palmira Saladié, and Josep Vall- Karkanas, Panagiotis, Ruth Shahack-Gross, Avner Ayalon, Miryam Bar- verdú. 2014. From small bone fragments to Neanderthal activity areas: the Matthews, Ran Barkai, Amos Frumkin, Avi Gopher, and Mary C. Stiner. case of Level O of the Abric Romaní (Capellades, Barcelona, Spain). Quater- 2007. Evidence for habitual use of fire at the end of the Lower Paleolithic: nary International 330:36–51. site-formation processes at Qesem Cave, Israel. Journal of Human Evolu- Gao, Xing, Shuangquan Zhang, Yue Zhang, and Fuyou Chen. 2017. Evidence tion 53:197–212. of hominin use and maintenance of fire at Zhoukoudian. Current Anthro- Keith, Arthur. 1927. A report on the Galilee skull. In Researches in prehistoric pology 58(suppl. 16):S267–S277. Galilee, 1925–1926. Francis Turville-Pétre, ed. Pp. 593–623. London: Council Gifford-Gonzalez, Diana. 1989. Ethnographic analogues for interpreting mod- of the British School of Archaeology in Jerusalem. ified bones: some cases from East Africa. In Bone modification. Robson Bon- Le Cabec, Adeline, Philipp Gunz, Kornelius Kupczik, José Braga, and Jean- nichsen and Marcella H. Sorg, eds. Pp. 61–71. Orono: Center for the Study of Jacques Hublin. 2013. Anterior tooth root morphology and size in Nean- the First Americans, Institute for Quaternary Studies, University of Maine. derthals: taxonomic and functional implications. Journal of Human Evolution Gilchrist, Roberta, and Harold Mytum. 1986. Experimental archaeology and 64:169–193. burnt animal bone from archaeological sites. Circaea 4:29–38. Lemorini, Cristina, Laurance Bourguignon, Andrea Zupancich, Avi Gopher, Gopher, Avi, Avner Ayalon, Miryam Bar-Matthews, Ran Barkai, Amos Frumkin, and Ran Barkai. 2016. A scraper’s life history: morpho-techno-functional Panagiotis Karkanas, and Ruth Shahack-Gross. 2010. The chronology of the and use-wear analysis of Quina and demi-Quina scrapers from Qesem Cave, late Lower Paleolithic in the Levant based on U-Th ages of speleothems from Israel. Quaternary International 398:86–93. Qesem cave, Israel. Quaternary Geochronology 5(6):644–656. Lemorini, Cristina, Avi Gopher, Ron Shimelmitz, Mary Stiner, and Ran Barkai. Gopher, Avi, Ran Barkai, Ron Shimelmitz, Hamudi Khalaily, Cristina Lemorini, 2006. Use-wear analysis of an Amudian laminar assemblage from Acheuleo- Israel Hershkovitz, and Mary Stiner. 2005. Qesem Cave: an Amudian site in Yabrudian Qesem Cave, Israel. Journal of Archaeological Science 33:921–934. central Israel. Journal of the Israel Prehistoric Society 35:69–92. Lemorini, Cristina, Flavia Venditti, Ella Assaf, Yoni Parush, Ran Barkai, and Avi Gopher, Avi, Yoni Parush, Ella Assaf, and Ran Barkai. 2016. Spatial aspects Gopher. 2015. The function of recycled lithic items at late Lower Paleolithic as seen from a density analysis of lithics at Middle Pleistocene Qesem Qesem Cave, Israel: an overview of the use-wear data. Quaternary Interna- Cave: preliminary results and observations. Quaternary International 398: tional 361:103–112. 103–117. Liu, Wu, Lynne A. Schepartz, Song Xing, Sari Miller-Antonio, Xiujie Wu, Erik Goren-Inbar, Naama, Nira Alperson-Afil, Mordechai E. Kislev, Orit Simchoni, Trinkaus, and María Martinón-Torres. 2013. Late Middle Pleistocene ho- Yoel Melamed, Adi Ben-Nun, and Ella Werker. 2004. Evidence of hominin minin teeth from Panxian Dadong, South China. Journal of Human Evolution control of fire at Gesher Benot Ya‘aqov, Israel. Science 304(5671):725–727. 64:337–355. Gowlett, J. A. J., J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina. 2017. Lyman, L. R. 1994. Vertebratetaphonomy. Cambridge: Cambridge University Press. Evidence of burning from bushfires in southern and east Africa and its Mayne, Pamela M. 1997. Fire modification of bone: a review of the literature. relevance to hominin evolution. Current Anthropology 58(suppl. 16):S206– In Forensic taphonomy: the postmortem fate of human remains. William D. S216. Haglund and Marcella H. Sorg, eds. Pp. 275–293. Boca Raton, FL: CRC. Gowlett, John A. 2010. Firing up the social brain. In Social brain and distributed Melamed, Yoel, Mordechai E. Kislev, Eli Geffen, Simcha Lev-Yadun, and mind. Robin Dunbar, Clive Gamble, and John A. J. Gowlett, eds. Pp. 345– Naama Goren-Inbar. 2016. The plant component of an Acheulian diet at 370. London: British Academy. Gesher Benot Ya‘aqov, Israel. Proceedings of the National Academy of Sci- Grayson, Donald K., Paul W. Parmalee, Lee R. Lyman, and Jim I. Mead. 1988. ences of the USA 113(51):14674–14679. Danger Cave, Last Supper Cave, and Hanging Rock Shelter: the faunas. Mendez, Fernando L., Thomas Krahn, Bonnie Schrack, Astrid-Maria Krahn, Anthropological Papers of the American Museum of Natural History, vol. 66, Krishna R. Veeramah, August E. Woerner, Forka L. M. Fomine, et al. 2013. pt. 1. New York: American Museum of Natural History. An African American paternal lineage adds an extremely ancient root to Groopman, Emily E., Rachel N. Carmody, and Richard W. Wrangham. 2015. the human Y chromosome phylogenetic tree. American Journal of Human Cooking increases net energy gain from a lipid-rich food. American Journal Genetics 92(3):454–459. of Physical Anthropology 156(1):11–18. Mentzer, Suzan. 2009. Bone as a fuel source: the affects of initial fragment size Hardy, Karen, Anita Radini, Stephen Buckley, Rachel Sarig, Les Copeland, distribution. In Gestion des combustibles au Paléolithique et au Mésolithique: Avi Gopher, and Ran Barkai. 2016. Dental calculus reveals inhaled environ- nouveaux outils, nouvelles interprétations. I. Théry-Parisot, S. Costamagno, mental contamination and ingestion of essential plant-based nutrients at Lower and A. Henry, eds. Pp. 53– 64. BAR International Series 1914. Oxford: Ar- Palaeolithic Qesem Cave Israel. Quaternary International 398:129–135. chaeopress. Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current Mercier, Norbert, Hélène Valladas, Christophe Falguères, Qingfeng Shao, Avi Anthropology 58(suppl. 16):S329–S336. Gopher, Ran Barkai, Jean-Jacques Bahain, Laurence Vialettes, Jean-Louis Hershkovitz, Israel, Patricia Smith, Rachel Sarig, Rolf Quam, Laura Rodríguez, Joron, and Jean-Louis Reyss. 2013. New datings of Amudian layers at Qesem Rebeca García, Juan-Luis Arsuaga, Ran Barkai, and Avi Gopher. 2011. Middle cave (Israel): results of TL applied to burnt flints and ESR/U-series to teeth. Pleistocene dental remains from Qesem Cave, Israel. American Journal of Journal of Archaeological Science 40:3011–3020. Physical Anthropology 144:575–592. Morin, Eugine. 2010. Taphonomic implications of the use of bone as fuel. Hershkovitz, Israel, Gerhard W. Weber, Cinzia Fornai, Avi Gopher, Ran Barkai, Palethnologie 2:209–217. Viviane Slon, Rolf Quam, Gabet Yankel, and Rachel Sarig. 2016. New Middle Nicholson, Rebecca A. 1993. A morphological investigation of burnt animal Pleistocene dental remains from Qesem cave (Israel). Quaternary Interna- bone and an evaluation of its utility in archaeology. Journal of Archaeological tional 398:148–158. Science 20:411–428. S328 Current Anthropology Volume 58, Supplement 16, August 2017

Oliver, James S. 1993. Carcass processing by the Hadza: bone breakage from sition: evolution and stability in the Middle Paleolithic and Middle Stone butchery to consumption. In From bones to behavior: ethnoarchaelogical Age. E. Hovers and S. L. Kuhn, eds. Pp. 171–188. Interdisciplinary Con- and experimental contributions to the interpretation of faunal remains. tributions to Archaeology. New York: Springer. J. Hudson, ed. Pp. 200–227. Center for Archaeological Investigations, Oc- ———. 2012. Middle Palaeolithic subsistence in the Near East: zooarchaeo- casional Paper 21. Carbondale: Southern Illinois University, Center for Ar- logical perspectives past, present and future. Before Farming 2(1):1–45. chaeological Investigations. Speth, John D., L. Meignen, O. Bar-Yosef, and P. Goldberg. 2012. Spatial Parush, Yoni, Ella Assaf, Avi Gopher, and Ran Barkai. 2015. Looking for sharp organization of Middle Paleolithic occupation X in Kebara Cave (Israel): edges: modes of flint recycling at Middle Pleistocene Qesem Cave, Israel. concentrations of animal bones. Quaternary International 247:85–102 Quaternary International 361:61–87. Stiner, Mary C., Avi Gopher, and Ran Barkai. 2009. Cooperative hunting and Parush, Yoni, Avi Gopher, and Ran Barkai. 2016. Amudian versus Yabrudian meat sharing 400–200 kya at Qesem Cave, Israel. Proceedings of the Na- under the rock shelf: a study of two lithic assemblages from Qesem Cave, tional Academy of Sciences of the USA 106(32):13207–13212. Israel. Quaternary International 398:13–36. ———. 2011. Hearth-side socioeconomics, hunting and paleoecology during Rink, William J., Norbert Mercier, Dušan Mihalović, Mike W. Morley, Jeroen the late Lower Paleolithic at Qesem Cave, Israel. Journal of Human Evo- W. Thompson, and Mirjana Roksandic. 2013. New radiometric ages from lution 60:213–233. the BH-1 hominin from Balanica (Serbia): implications for understanding Stiner, Mary C., S. L. Kuhn, and T. A. Surovell. 2001. Bone preservation in the role of the Balkans in Middle Pleistocene human evolution. PLoS ONE Hayonim Cave (Israel): a macroscopic and mineralogical study. Journal of 8(2):e54608, doi:10.1371. Archaeological Science 28:643–659 Roebroeks, Wil, and Paola Villa. 2011. On the earliest evidence for habitual Stiner, Mary C., Steven L. Kuhn, Stephen Weiner, and Ofer Bar-Yosef. 1995. use of fire in Europe. Proceedings of the National Academy of Sciences of the Differential burning, recrystallization, and fragmentation of archaeological USA 108(13):5209–5214. bone. Journal of Archaeological Science 22(2):223–237. Rolland, Nicolas, 2004. Was the emergence of home bases and domestic fire a Théry-Parisot, Isabelle. 2002. Fuel management (bone and wood) during the punctuated event? a review of the Middle Pleistocene record in Eurasia. Lower Aurignacian in the Pataud Rock Shelter (Lower Palaeolithic, Les Eyzies Asian Perspectives 43(2):248–280. de Tayac, Dordogne, France): contribution of experimentation. Journal of Rosell, Jordi, Ruth Blasco, Avi Gopher, and Ran Barkai. 2015. Recycling bones Archaeological Science 29:1415–1421. in the Middle Pleistocene: some reflections from Gran Dolina TD10–1 Valladas, Hélène, Norbert Mercier, Israel Hershkovitz, Yossi Zaidner, Alexander (Spain), Bolomor Cave (Spain) and Qesem Cave (Israel). Quaternary In- Tsatskin, Reuven Yeshurun, Laurence Vialettes, Jean-Louis Joron, Jean-Louis ternational 361:297–312. Reyss, and Mina Weinstein-Evron. 2013. Dating the Lower to Middle Pa- Rosell, Jordi, Isabel Cáceres, Ruth Blasco, Maria Lluc Bennàsar, Pilar Bravo, leolithic transition in the Levant: a view from Misliya Cave, Mount Carmel, Gerard Campeny, Montse Esteban-Nadal, et al. 2012. A zooarchaeological Israel. Journal of Human Evolution 65(5):585–593. contribution to establish occupational patterns at Level J of Abric Romaní Verri, Giovanni, Ran Barkai, Cristian Bordeanu, Avi Gopher, Michael Hass, (Barcelona, Spain). Quaternary International 247:69–84. Aaron Kaufman, Peter W. Kubik, et al. 2004. Flint mining in prehistory record Rust, Alfred. 1950. Die Hohlenfunde von Jabrud (Syrien). Neumünster. by in situ produced cosmogenic 10Be. Proceedings of the National Academy of Sánchez-Marco, Antonio, Ruth Blasco, Jordi Rosell, Avi Gopher, and Ran Sciences of the USA 101(21):7880–7884. Barkai. 2016. Birds as indicators of high biodiversity zones around the Middle Verri, Giovanni, Ran Barkai, Avi Gopher, Michael Hass, Peter W. Kubik, Pleistocene Qesem Cave, Israel. Quaternary International 421:23–31. Michael Paul, Avraham Ronen, Steve Weiner, and Elisabetta Boaretto. 2005. Sarig, Rachel, Avi Gopher, Ran Barkai, Jordi Rosell, Ruth Blasco, Gerhard W. Flint procurement strategies in the late Lower Palaeolithic record by in situ Weber, Cinzia Fornai, Tatiana Sella-Tunis, and Israel Hershkovitz. 2016. produced cosmogenic 10Be in Tabun and Qesem Caves (Israel). Journal of How did the Qesem Cave people use their teeth? analysis of dental wear Archaeological Science 32:207–213. patterns. Quaternary International 398:136–147. Weber, Gerhard W., Cinzia Fornai, Avi Gopher, Ran Barkai, Rachel Sarig, and Shahack-Gross, Ruth, Ofer Bar-Yosef, and Steve Weiner. 1997. Black-coloured Israel Hershkovitz. 2016. The Qesem Cave hominin material. 1. A morpho- bones in Hayonim Cave, Israel: differentiating between burning and oxide metric analysis of the mandibular premolars and molar. Quaternary Inter- staining. Journal of Archaeological Science 24(5):439–446. national 398:159–174. Shahack-Gross, Ruth, Francesco Berna, Panagiotis Karkanas, Cristina Lemorini, Wiessner, Polly W. 2014. Embers of society: firelight talk among the Ju/’hoansi Avi Gopher, and Ran Barkai. 2014. Evidence for the repeated use of a central bushmen. Proceedings of the National Academy of Sciences of the USA 111 hearth at Middle Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of (39):14027–14035. Archaeological Science 44:12–21. Wilkins, Jane, and Michael Chazan. 2012. Blade production ~500 thousand Shimelmitz, R. 2015. The recycling of flint throughout the Lower and Middle years ago at Kathu Pan 1, South Africa: support for a multiple origins hy- PaleolithicsequenceofTabunCave,Israel.Quaternary International 361:34–46. pothesis for early Middle Pleistocene blade technology. Journal of Archaeo- Shimelmitz, Ron, Ran Barkai, and Avi Gopher. 2016. Regional variability in logical Science 39(6):1883–1990. late Lower Paleolithic Amudian blade technology: analyzing new data from Wilson, Lucy, Aviad Agam, Ran Barkai, and Avi Gopher. 2016. Preliminary Qesem, Tabun and Yabrud I. Quaternary International 398:37–60. evaluation of raw material choices in the Amudian versus the Yabrudian at Shimelmitz, Ron, Avi Gopher, and Ran Barkai. 2011. Systematic blade pro- Qesem Cave. Quaternary International 398:61–69. duction at late Lower Paleolithic (400–200 kyr) Qesem Cave, Israel. Journal Wrangham, Richard. 2017. Control of fire in the Paleolithic: evaluating the of Human Evolution 61:458–479. cooking hypothesis. Current Anthropology 58(suppl. 16):S303–S313. Shimelmitz, Ron, Steven L. Kuhn, Arthur J. Jelinek, Avraham Ronen, Amy E. Yravedra, José, and Paloma Uzquiano. 2013. Burnt bone assemblages from El Clark, and Mina Weinstein-Evron. 2014. “Fire at will”: the emergence of Esquilleu cave (Cantabria, northern Spain): deliberate use for fuel or sys- habitual fire use 350,000 years ago. Journal of Human Evolution 77:196–203. tematic disposal of organic waste? Quaternary Science Reviews 68:175–190. Shipman, Pat, Giraud Foster, and Margaret Schoeninger. 1984. Burnt bones Zeitoun, Valery. 2001. The taxonomical position of the skull of Zuttiyeh. and teeth: an experimental study of color, morphology, crystal structure Comptes Rendus de l’Académie des Sciences, série IIa: Sciences de la Terre and shrinkage. Journal of Archaeological Science 11(4):307–325. et des Planètes 332:521–525. Solari, Ana, Daniel E. Olivera, Inés Gordillo, Pedro Bosch, Geolar Fetter, Victor Zupancich, Andrea, Cristina Lemorini, Ran Barkai, and Avi Gopher. 2016a. Hugo Lara, and Omar Novelo. 2015. Cooked bones? method and practice On scraper handling: preliminary results from a use-wear and experimental for identifying bones treated at low temperature. International Journal of approaches on the Lower Palaeolithic site of Qesem Cave, Israel. Quater- Osteoarchaeology 25:426–440. nary International 398:94–102. Spennemann, Dirk H. R., and Sarah M. Colley. 1989. Fire in a pit: the effects Zupancich, Andrea, Stela Nunziante-Cesaro, Ruth Blasco, Jordi Rosell, Ema- of burning on faunal remains. Archaeozoologia 3(1/2):51–64. nuela Cristiani, Flavia Venditti, Cristina Lemorini, Ran Barkai, and Avi Speth, John D. 2006. Housekeeping, Neandertal-style: hearth placement and Gopher. 2016b. Early evidence of stone tool use in bone working activities midden formation in Kebara cave (Israel). In Transitions before the tran- at Qesem Cave, Israel. Scienti fic Reports 6:37686. Current Anthropology Volume 58, Supplement 16, August 2017 S329

Neanderthal Cooking and the Costs of Fire

by Amanda G. Henry

While it is clear that Neanderthals used fire for cooking their foods in some times and places, the record of their use of fire is somewhat patchy. We should not assume that Neanderthals had the same relationship with fire that we do; as a technological/cultural behavior, fire may be better understood as a tool that was used only when the costs of manufacture and maintenance were outweighed by the benefits.

Use of fire, particularly to transform food, is widely recognized processes that erased the record of fire, or is fire use perhaps as a major step in human evolution, allowing us increased ac- more variable in human history than we have previously as- cess to refractive foods (Stahl 1984), lowered cost of digestion sumed? (Carmody and Wrangham 2009), and decreased exposure to In this paper I address the question of Neanderthal use of fire, food-borne bacteria (Smith et al. 2015). Just when this process in particular for cooking their food. The fossil and archaeo- was first adopted by human ancestors is highly debated, with logical record of Neanderthals is the most complete among our the earliest evidence for fire being very unevenly distributed in hominin relatives, and there is clear evidence at many sites that time and space. Burned materials are found in association with Neanderthals used fire and cooked their food. Despite this hominin activity at a variety of archaeological sites, dating from wealth of data, many questions about Neanderthal use of fire 1.5 Ma at Koobi Fora (Hlubik et al. 2017), 1.4 Ma at Cheso- remain unanswered, and some have even suggested that fun- wanja (Gowlett et al. 1981), and 1 Ma at Wonderwerk Cave damental differences between Neanderthal and Early Modern (Berna et al. 2012). However, as Pruetz and Herzog (2017) and Human (EMH) use of fire may have contributed to the dis- Gowlett et al. (2017) indicate, naturally occurring fire is nearly appearance of the former. By exploring the factors that may ubiquitous, and it is extremely challenging to show that the have influenced how, when, and why Neanderthals used fire, hominins who created the archaeological scatters at these sites we may begin to build a more nuanced model of the influence were coeval with, or even directly interacted with, the fire. To of fire in human evolution. use the terminology developed by Sandgathe (2017), hominins fi may have been habituated to re, but discerning their use, Evidence for and against Neanderthal Use of Fire maintenance, or manufacture of fire is extremely difficult in the early record. In fact, regular use of fire does not appear to be Several European and Levantine Middle Paleolithic sites have widespread among northern-latitude hominins until much later ample evidence for the presence of fire, in the form of discrete (Roebroeks and Villa 2011; Stahlschmidt et al. 2015). Some have hearths of charcoal, ash, and fire-altered sediments, charred even argued that as late as the Middle Paleolithic, Neanderthals and calcined bone, and heated stone. These sites are found in may have been able to maintain fires but were not able to man- the Levant at Tabun, Kebara, and Hayonim (Goldberg 2003); ufacture them (Sandgathe et al. 2011). The benefits of cooking in France at St. Cesaire (Morin 2004), La Quina (Chase 1999), have been well documented (Wrangham 2009), and this has Pech de l’Azé IV, and Roc de Marsal (Goldberg et al. 2012); been argued to be a driving factor in promoting human use, and in Spain at Abric Romani (Vallverdú et al. 2012) and El maintenance, and manufacture of fire, such that once hominins Salt (Gómez de la Rua et al. 2010). Evidence for the use of fire first used fire, it quickly became an obligatory part of their includes the discovery of birch tar hafting, which would have niche. The disconnect between the very sparse archaeological required intentional heating (Mazza et al. 2006), found on record of fire and the potentially strong benefits of using fire several Middle Pleistocene flakes from Italy. The study of char- begs the question, Is the absence of fire evidence a result of coal and other fire remains, though initially used to document the local tree species, has become increasingly used as a means to document Neanderthal and EMH fire behavior, in large part Amanda G. Henry is Associate Professor in the Faculty of Archae- due to the extensive experimental works of Isabelle Théry- ology at Leiden University (Van Steenis Building, Einsteinweg 2, 2333 CC Leiden, The Netherlands [[email protected]]) and Parisot. She has shown that it is possible to differentiate among Associated Researcher at the Max Planck Institute for Evolutionary the burning of green, dry, and rotten woods, and that each of Anthropology (Deutscher Platz 6, 04103 Leipzig, Germany [amanda these fuels burns for a different amount of time with different [email protected]]). This paper was submitted 25 VII 16, accepted heating properties (e.g., convection, conduction, and illumi- 5 III 17, and electronically published 17 V 17. nation; Théry-Parisot 2001). Applying these methods to ar- q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0015$10.00. DOI: 10.1086/692095 S330 Current Anthropology Volume 58, Supplement 16, August 2017 chaeological sites, she has demonstrated that Paleolithic hom- cooking, burned as a way of disposing refuse, or simply heated inins (Neanderthals and EMH) generally burned dry stand- after burial when a fire occurred in the sediments above them. ing wood and did not appear to choose particular tree species. While some have developed metrics to help discern between However, she and others have also shown that in certain cases, cooked and otherwise fire-damaged bones (e.g., Costamagno Neanderthals burned coal or bone in addition to wood because et al. 2008; Speth and Tchernov 2001), these methods have been of their preferred burning properties (Dibble et al. 2009; Théry- applied only to a very small number of Neanderthal sites. At Parisot and Costamagno 2005; Théry-Parisot and Meignen Kebara, it seems the charred bones were the result of cooking 2000). Coal does not produce flames, but it is better than wood (Speth and Tchernov 2001), while at Pech de l’AzéIV,theevi- at conduction, so it is good for heating opaque, solid objects dence is inconclusive and could indicate cooking, refuse dis- (e.g., food placed on the fire to cook or flint and other raw posal, or both (Dibble et al. 2009). materials to improve their flaking characteristics; Théry-Parisot Direct evidence for cooking of plant foods is extremely rare and Meignen 2000). Mixes of bone and wood achieve longer in the Neanderthal record. Charred seed remains have only burning times and better conduction of heat than wood alone been reported from a handful of sites, including Kebara (Lev, (Théry-Parisot and Costamagno 2005). These results strongly Kislev, and Bar-Yosef 2005) and Douara Cave (Matsutani 1987) indicate that Neanderthals not only knew how to maintain a in the Levant, Franchthi Cave in Greece (Hansen 1991), and fire but had sufficiently experimented with different fuels to be Gorham’s and Vanguard Caves in Gibraltar (Gale and Car- able to choose among them to create fires with particular ruthers 2000). Most of these assemblages are very small, but burning properties. the Kebara assemblage consists of more than 4,000 charred In contrast to the above data, there is also an increasing em- seeds, mostly from legumes (Fabaceae) and nuts (Pistacia and phasis on the evidence that suggests that Neanderthals perhaps Quercus). A range of other plants, including members of the were able to use and maintain fire but were not able to man- Boraginaceae family, were also common at several sites. Though ufacture it. Sandgathe et al. (2011) point to the fact that, once these have been interpreted as food, even charred seeds may fire appears in the archaeological record, it is seen only at a not be an accurate record of cooking. Miller (1984, 1996) has small percentage of sites at any one particular time period strongly argued that the primary mechanism by which whole and often only within a small percentage of occupation layers seeds become charred is by the burning of animal dung as fuel, within each site. Based on the record at two French sites, they rather than direct cooking of plant foods, since there is little reject taphonomic processes, sampling biases, changes in func- benefit to be had by cooking whole seeds on an open-flame fire. tion of the site, or seasonality of use as causes for the apparent Even if Neanderthals did not collect and burn dung, it is pos- disappearance of direct and indirect evidence of fire. They argue sible that the seeds were accidentally burned rather than in- instead that the absence of fire reflects that the use of fire was tentionally cooked. Shallowly buried seeds may be inadver- “not an essential part of [Neanderthal] behavior” (Sandgathe tently charred when a fire is burned above them (Aldeias et al. et al. 2011:217). They emphasize that modern or recent his- 2016), calling into question the relationship between the fire torical foragers should not be taken as direct analogues for and the seeds. New sources of information about cooking and Neanderthals; instead, we must recognize that Neanderthals the consumption of cooked plant foods are clearly needed. had a deep time history in Europe that may have provided In the last 10 years, dental calculus has been increasingly ample opportunity to physiologically adapt to colder temper- explored as a source of new information about diet and health. atures and to acquire enough calories from uncooked food. This biological material is created through the proliferation of They conclude that Neanderthals used and probably maintained oral bacteria on the teeth and the subsequent mineralization of fire when it was convenient and available on the landscape— this biofilm due to the supersaturation of calcium phosphate in for example, in warmer periods when fuel was abundant and the saliva (Jin and Yip 2002; Lieverse 1999). The mineralized natural fires from lightning strikes were frequent—but that surface is recolonized by bacteria, and calculus builds up in suc- Neanderthals did not have the ability to manufacture fire. cessive layers throughout the lifetime of the individual, though the trigger of the mineralization process and the rate of for- Neanderthal Use of Fire for Cooking mation are generally unknown and likely vary among indi- viduals. As calculus forms, many of the organic residues found In the same manner that Neanderthal use of fire has a variable in the mouth, including human and bacterial proteins and record, evidence for cooking is inconstant and debated. The DNA, food lipids, and plant microremains, are trapped in the cooking of animal foods such as meat and fat is mostly evi- mineral matrix and can be preserved over archaeological time denced by heat-damaged bones. Blackened bones are frequently spans (Buckley et al. 2014; Henry and Piperno 2008; Warinner, found in sites, but discerning whether they have actually been Speller, and Collins 2015). There are two types of plant micro- charred or just stained requires a level of analysis that was not remains that are particularly useful in reconstructing diets— commonly done during the excavation of many Middle Paleo- phytoliths and starch grains. Phytoliths are particles of amor- lithic sites (Shahack-Gross, Bar-Yosef, and Weiner 1997). Even phous silica that form within and between the cells of a plant for bones that have been conclusively charred or calcined, there (Piperno 2006) and are generally regarded as a form of me- can be some question about whether they were damaged during chanical defense against herbivory (Massey and Hartley 2006). Henry Neanderthal Cooking and the Costs of Fire S331

Starch grains are structures that a plant produces from long- dence—gelatinized starches are particularly vulnerable to re- chain carbohydrates as a means of energy storage (BeMiller moval from the archaeological record. Other calculus inclu- and Whistler 2009). Both microremain types can have taxon- sions, including smoke residues, charcoal, and altered fats found specific morphology that can be used to identify the plant spe- in several individuals from El Sidrón and Qesem Cave, also cies and/or plant part in which they were formed. Furthermore, suggest cooking by these individuals (Hardy et al. 2012, 2016). plant microremains can often record changes due to process- There is much work to be done to assess the taphonomic ing, such as grinding and heating. Exposure to extremely high biases that may affect the record of plant microremains and temperatures can change the refractive index of phytoliths (El- other residues in dental calculus. Based on studies matching baum et al. 2003), while starches can undergo a variety of di- the diets of modern groups with their calculus microremains agnostic changes, including gelatinization, cracking, and loss of (Leonard et al. 2015; Power et al. 2015), we know that our re- organization, which indicate heating in the presence of water sults likely underrepresent the true number of plant foods con- (Babot 2003; Henry, Hudson, and Piperno 2009; Messner and sumed by Neanderthals. Furthermore, there are many taphonomic Schindler 2010). factors that potentially remove starches from the archaeological To date, over 70 Neanderthal dental calculus specimens from record (Henry 2015), and we have seen that gelatinized starches more than 40 individuals have been sampled for plant micro- are very quickly removed from buried stone tools (Debono remains and other residues, providing us a new glimpse into Spiteri et al. 2014) and thus might also be strongly underrep- their dietary behavior (Hardy et al. 2012, 2016; Henry 2010; resented in the calculus record. Residue analysis using mass Henry, Brooks, and Piperno 2011; Salazar-García et al. 2013). spectroscopy–based methods is still in its infancy, as no analy- These samples include material from the Levant (Shanidar, sisonpopulationswithknowndietshasbeenperformed.How- Qesem Cave), the Mediterranean (Sima de las Palomas, Ka- ever, given the increasingly large number of Neanderthal calcu- lamakia), and more “classic” areas in western Europe (Spy, La lus samples studied to date, it is likely that the apparent pattern Quina, La Ferrassie). Some of the earliest work demonstrated of variability in Neanderthal use of cooked plant foods is a that Neanderthals from a variety of environments consumed real one. plant foods, including several of the resources, like grass seeds and tubers, that became important in later agricultural socie- ties (Henry 2010; Henry, Brooks, and Piperno 2014). In rare Applying a Behavioral Ecology Framework occasions, gelatinized starch grains were found, indicating that to the Discussion of Fire the plants had been cooked prior to consumption (Henry, Brooks, and Piperno 2011). In the cases where the partially When combined with the data that suggest inconsistencies in gelatinized starches could be identified to taxon, they seem to Neanderthal control of fire, the pattern of cooking variability be all from grass seeds or other plants with hard, starchy en- raises the question of why Neanderthals would have fire and dosperm in their seeds, lending support to the interpretation cook with it in some places and times and not in others. Some that the charred macrobotanical remains found in other sites have argued that Neanderthals did not possess the capacity to do indeed represent cooking of seeds. However, this interpre- manufacture fire (Dibble et al. 2017; Sandgathe et al. 2011), tation must be tempered by some of the unresolved issues re- while others have invoked a more taphonomic explanation lating to differential survival of starches from different plants. (Gowlett and Wrangham 2013). However, in most studies so Furthermore, it is interesting to note that many of the Ne- far, researchers have considered only the possibility of an “either/ anderthal samples did not preserve any plant microremains. or” situation—either Neanderthals had the ability to manu- This pattern, with some specimens showing heavy use of plants facture fire and therefore did so in all times and all places, or while others seem to preserve none, has been found in every they did not. I propose that the use of fire needs to be examined subsequent study of calculus, across the entire geographic range using an explicitly economic framework to understand the ben- of sampled specimens. In fact, there has been to date no coher- efits but also the potential costs of manufacturing and main- ent pattern of plant use across the sampled specimens. Neither taining a fire. The relative costs to benefits in different envi- age nor geographic region influences the number of recovered ronments could explain the variability in Neanderthal fire use. plant microremains, nor the number of different types of plants Neanderthals may have had the ability to manufacture fire but represented (Henry, Brooks, and Piperno 2014). A study cur- in some cases may have chosen not to do so, and it is in ex- rently under way (Power et al., forthcoming) aims to explore ploring their choices that we gain a better understanding of whether a finer-grained geographical analysis that includes their behaviors. information about average temperature and tree cover at a site Like other technologies, the use of fire for cooking is some- might better explain the number of plants that Neanderthals thing that can be explained by using models derived from hu- consumed. Particularly relevant to the discussion of cooking, man behavioral ecology (Bird and O’Connell 2006). In these however, is the observation that gelatinized starches are ex- models, it is possible to predict an individual’s response to its tremely rare and appear only in a small handful of specimens. current environment. The basic premise is that an individual While this may indicate even lower frequencies of cooking, attempts to achieve some goal that will increase its fitness, but again, we must be careful with the problem of absence of evi- in order to do so, it has to choose between a variety of potential S332 Current Anthropology Volume 58, Supplement 16, August 2017 behaviors. Each of these behaviors has costs and benefits that deadwood to regenerate (Pryor et al. 2016). Clearly, securing are measured in a certain currency. In some cases, the indi- preferred fuel would have required some sort of behavioral vidual is prevented from using one of its potential behaviors. shift. In the case of the Gravettian groups, the authors proposed When using behavioral ecology models to examine real-world intentional management strategies, such as geographic mobil- data, the researcher makes a prediction about the exact goals, ity and the deliberate killing of trees in advance, but for groups choices, costs and benefits, currencies, and constraints that in- less dependent on fire than modern humans, it is possible that fluence the organism in order to predict the most optimal be- other adaptive strategies may have been used. havior. When the hypothesis matches the actual behavior, the Many studies have documented how increasing the cost of researcher chose the correct variables; when the actual behavior fuel can change behavior. Among people living today, fuel col- differs from the predicted, then either the predicted variables lection requires a significant time and energy investment, with were incorrectly chosen or the model was in some way inap- some households in rural Mexico spending nearly 4 person- propriate. This framework makes it possible to test whether hours per trip to collect fuel (Manning and Taylor 2014). Small certain environmental restrictions or behavioral limitations decreases in forest cover are connected with large increases in strongly influenced human use of fire, or we need instead to the amount of time collecting fuel, with one study linking a 1% invoke another explanation (e.g., taphonomic bias, cultural pres- increase in deforestation to a 0.3% decrease in fuel consumption sure sensu Wiessner [2014]) for the presence or absence of fire and a 0.6% increase in fuel collection (Kumar and Hotchkiss in the archaeological record. 1988). This may sound small, but it represents an extra 1.13 hours Specifically, in the case of using fire for cooking, most of the per day spent on fuel collection (Kumar and Hotchkiss 1988). benefits come in the form of caloric savings. Several experi- Recent Kenyan agripastoralists chose to prepare more quickly ments have documented that some tubers, meat, and even oil- cooked foods in times of fuel scarcity (van Wijngaarden 1984). rich seeds become more easily digestible once cooked (Boback Furthermore, while these agripastoralists often collected fuel in et al. 2007; Carmody and Wrangham 2009; Groopman, Car- conjunction with other tasks (such as herding cattle), they rec- mody, and Wrangham 2015). The risk of food-borne illnesses ognized the costs of carrying the wood, preferentially leaving cut drops once foods are cooked (Smith et al. 2015). Some plant green wood to dry in public areas, thus potentially risking its foods are only accessible to digestion after cooking, due to the loss, rather than carrying the heavy fresh wood back to the farm reduction in toxins and other antifeedants (Stahl 1984). All of to cure (van Wijngaarden 1984). Patterns of fuel use are less well these benefits can be measured as an increase in calories avail- studied in foraging populations than among developing socie- able to the consumer through a reduction in the caloric costs of ties. Women and children are the primary fuel gatherers in most the food, a reduction in costs related to illness, or an increase in societies (Murdock and Provost 1973), but there is a clear en- the number of potentially calorie-rich food sources, all of which vironmental variation in the value of and attitude toward fuel. would then allow more calories to be spent on other tasks Lee (1979:148) noted that for the !Kung San, “in the dry Kalahari (Aiello and Wheeler 1995). firewood is rarely a problem. Dead wood is plentiful and even Despite these rather significant benefits, there may be oc- living wood is dry enough to burn instantly if put into a fire.” In casions or environments where the costs of creating a fire out- contrast, the Mbendjele foragers in Congo-Brazzaville complain weigh the benefits of having one. The most obvious cost is that about having to cook beans, because they take too long and of collecting fuel (Ofek 2001). Dry standing wood is the pre- require too much wood (K. Janmaat, personal communication); ferred fuel for cooking, because rotten wood burns poorly and despite the abundance of trees, good fuel is limited in a rain- fresh wood requires significant time investment (6–36 months) forest. to properly dry (Théry-Parisot 2001), which limits its use by In the archaeological record, it has also been possible to mobile foraging groups. Dry standing wood is common in recognize behavioral shifts resulting from the costs of fuel. A many forested areas, and Théry-Parisot and Meignen (2000) series of elegant studies have indicated that, as local wood have calculated that there is enough deadwood within less than resources were depleted by smelting, Bronze Age inhabitants a kilometer radius of a site to burn four fires 24 hours a day for of the Levant shifted to using more tin in their bronze because 6 months. However, these calculations were done for the site of of its significantly lower melting temperature, which reduced Les Canalettes in France, which would have been surrounded their overall fuel needs (Kaufman 2013; Kaufman and Scott by thick forest during the period the authors were analyzing. In 2014). Even earlier in history, in the semiarid savanna of south- other environments, wood may not be so abundant (Elton west Texas and Coahuila, Mexico, Archaic period inhabitants 2008). Some of the earliest analyses of human choice of fuel relied on lechuguilla and sotol, two extremely refractive starchy have argued that preferred dry standing wood can quickly plant resources that required intensive processing, mostly roast- become depleted, even in areas of relatively low population ing in an earth oven for long periods of time. To create enough density (Shackleton and Prins 1992). A recent paper on fuel use food for 1 day for a small group of four to five individuals in the Pavlov Hills region of the Czech Republic proposed that required a minimum of 250 kg of wood (Dering 1999). This the Gravettian hunter-gatherers in the region would have quickly depleted local fuel sources, and Dering argued that fuel quickly eliminated the naturally occurring deadwood, and it and food scarcity, not water availability, led to increasing mo- may have taken several generations (40–120 years) for the bility among these foragers. Charcoal analyses at Mousterian Henry Neanderthal Cooking and the Costs of Fire S333 sites have already shown that Neanderthals were aware of the and potentially carcinogenic effects of the Mailliard reaction, qualities of their fires and probably also were aware of the which occurs when food is browned or burned over a fire availability of various fuel sources. Théry-Parisot and col- (Ledl and Schleicher 1990; Mottram, Wedzicha, and Dodson leagues (Théry-Parisot 2002; Théry-Parisot and Meignen 2000) 2002), to the actual risks of injury or death from burning and have argued that the introduction of coal and bone to wood the long-term health risks of smoke inhalation (Bruce, Perez- fires significantly reduces the amount of wood that is needed to Padilla, and Albalak 2000; Smith et al. 2000). Recent genetic burn for the same amount of time, suggesting a need to reduce studies have shown that modern humans, and not Neander- wood consumption perhaps due to a decreased availability of thals, Denisovans, or living African apes, have a fixed derived wood. Furthermore, the charcoal fragments at a Spanish Ne- variant of a gene (AHR) that reduces the deleterious effect of anderthal MIS 5–4 site showed signs of radial cracking and polycyclic aromatic hydrocarbons, a group of chemicals re- fungal infestation of wood (Vidal-Matutano et al. 2015). The leased in smoke and found on charred foods (Hubbard et al. former indicates the burning of green wood and the latter of 2016). This suggests that there was strong selective pressure rotten wood, neither of which is a preferred fuel source (Théry- among modern humans to avoid some of the negative health Parisot 2001), suggesting that perhaps the local environment effects of fire. Another study found no evidence in Neanderthals had become depleted in standing dry deadwood. These lines of for selection among a large number of detoxification genes to- evidence suggest that there may have been ample reasons for ward alleles that were more protective against the ill effects of Neanderthals to consider the amount of fuel they consumed. smoke (Aarts et al. 2016). It is unlikely that Neanderthals had an Though there would have been wood in sheltered places around understanding of microbial horizontal gene transfer or of link- Neanderthal sites even during cold, dry periods, it is possible ing cancer risks to browned food. However, these invisible costs that the time and energetic costs of collecting this limited re- of cooking may have had a noticeable effect on longevity and source may not have been offset by the benefits of fire. disease state, leading to a decrease in fitnessofgroupswhoreg- Though fuel collection is the major cost associated with fire, ularly cooked. there are other potential costs to consider. These include the Finally, and perhaps most importantly, the costs and benefits time spent curating a fire to cook the food, the potential loss of of cooking vary among environments, so that the value of cook- food to conspecific competitors, the removal of potentially ben- ing would not be the same across the entire geographic and eficial bacteria, and the health risks associated with fire itself. temporal spread of the Neanderthals. As mentioned above, the Time allocation is a major aspect of energy allocation. For ex- amount of available fuel depended heavily on the degree of ample, sitting around a fire while cooking food is a relatively tree cover and on the local distribution of other resources, such low-energy activity, and it is time that cannot be spent on as coal. For example, the cost to access standing deadwood in other, perhaps more fitness-enhancing, behaviors such as col- a steppe environment may have been very high, while other lecting additional food items. The cooked food also needs to be fuels might not have been an option if the foragers were mo- guarded from theft. Wrangham and colleagues (Wrangham bile (it requires a significant time investment to create and dry et al. 1999) noted that cooked food is of increased value because it dung cakes or to cure fresh-cut wood). Furthermore, the kinds is more digestible; also, it has been collected and gathered to a of foods available in different environments have different re- central place and is therefore at risk to be taken by “scroung- sponses to cooking. Within a single food type, plant underground ers.” The potential loss of food items should be considered storage organs, there is large variation among taxa in their nu- when weighing the benefits of cooking. Not only are cooked tritional qualities and their changes in digestibility when cooked. items vulnerable, but the fire itself is a shared good that is at Onestudyoftheglucosedigestibilityoftubersconsumedbythe risk of freeloaders (Twomey 2013) who might use the benefits Hadza foragers showed that of four species, one had improved of the fire but not contribute to its maintenance. Furthermore, digestibility when cooked, two had negligible changes, and one the process of cooking itself may produce harmful physical was actually less digestible after cooking (Schnorr et al. 2015). reactions. Cooking destroys bacteria, which can reduce food- The Hadza can and do consume these plants when raw, and borne illnesses. However, we are increasingly aware that hu- when they cook them, they do so for short (ca. 5 min) periods mans can acquire adaptive genes from exogenous bacteria and (Marlowe 2010). It seems that the Hadza cook the tubers to assimilate them within functional members of the complex make them easier to peel, but not to predigest the starch inside system of gut microbiota (GM; Smillie et al. 2011). One potent them. This is in contrast to the tubers available in rainforest example comes from a group of Japanese people who have environments, which often contain harmful toxins that must gained the ability to digest a type of refractive seaweed by in- be denatured by cooking (and sometimes more elaborate pro- corporating genes from naturally occurring bacteria into their cessing like leaching in lye) prior to consumption (Tanno 1981). GM (Hehemann et al. 2010). Studies of non-Western rural and Clearly, ascribing the value of cooking to one single factor can forager groups further indicate that the adoption of genes from lead to misunderstandings of its use in human history. environmental bacteria into the GM confers digestive benefits Fire is so embedded in the way present-day humans live that (De Filippo et al. 2010; Rampelli et al. 2015; Schnorr et al. 2014). it is hard for us to consider life without its benefits, including Finally, the process of creating a fire and cooking carries in- cooking, light, heat, and protection. However, given the im- trinsic health risks. These range from the nutrient-reducing mense time periods and significant environmental shifts in S334 Current Anthropology Volume 58, Supplement 16, August 2017 which Neanderthals lived, we must realize that using our pres- Cape Province, South Africa. Proceedings of the National Academy of Sci- ences of the USA 109(20):E1215–E1220, doi:10.1073/pnas.1117620109. ent, or even the recent historical past, as a reference may not be Bird, Douglas W., and James F. O’Connell. 2006. Behavioral ecology and appropriate. When the data suggest interpretations for Nean- archaeology. Journal of Archaeological Research 14(2):143–188. derthal interactions with fire that do not mimic our own, we Boback, Scott M., Christian L. Cox, Brian D. Ott, Rachel Carmody, Richard W. Wrangham, and Stephen M. Secor. 2007. Cooking and grinding reduces the must be open to the possibility that they had a different mode of cost of meat digestion. Comparative Biochemistry and Physiology A 148:651– living. Instead, models developed for studies of all living species, 656. such as those derived from behavioral ecology, allow us to Browne, Constance L., and Lucy Wilson. 2011. Resource selection of lithic raw materials in the Middle Palaeolithic in southern France. Journal of Human better understand Neanderthal behavior in terms of costs and Evolution 61(5):597–608, doi:10.1016/j.jhevol.2011.08.004. benefits rather than social or cognitive abilities. Behavioral ecol- Bruce, Nigel, Rogelio Perez-Padilla, and Rachel Albalak. 2000. Indoor air pol- ogy models have been successfully used to explore Neanderthal lution in developing countries: a major environmental and public health challenge. Bulletin of the World Health Organization 78(9):1078–1092, doi: decision-making in other areas (e.g., that the choice of lithic 10.1590/S0042–96862000000900004. raw material depends in no small part on the distance to the Buckley, Stephen, Donatella Usai, Tina Jakob, Anita Radini, and Karen Hardy. quarry and the difficulty of terrain to reach it [Browne and 2014. Dental calculus reveals unique insights into food items, cooking and plant processing in prehistoric central Sudan. PLoS ONE 9(7):e100808, doi: Wilson 2011; Wilson 2007a, 2007b]). A new comprehensive 10.1371/journal.pone.0100808. view of the costs of fire must be undertaken. While many of the Carmody, Rachel N., and Richard W. Wrangham. 2009. The energetic sig- potential costs cannot be directly compared to the potential nificance of cooking. Journal of Human Evolution 57:379–391. Chase, Philip G. 1999. Bison in the context of complex utilization of faunal gains of cooking, we can at least begin by exploring the caloric resources: a preliminary report on the Mousterian zooarchaeology of La costs of fuel collection in different environments and exploring Quina (Charente, France). In Le bison: gibier et moyen de subsistance des how much time and energy it requires to access different fuels hommes du Paléolithique aux Paléoindiens des Grandes Plaines. Jean-Philip Brugal, ed. Pp. 159–184. Antibes: APDCA. (e.g., dry standing wood, green wood, etc.). A study that in- Costamagno, Sandrine, Isabelle Théry-Parisot, Jean-Christophe Castel, and corporates the various caloric costs of fuel collection in dif- Jean-Philip Brugal. 2008. Combustible ou non? analyse multifactorielle et ferent environments and compares this to the relative caloric modèles explicatifs sur les ossements brûlés Paléolithiques. In Gestion des fi combustibles au Paléolithique et au Mésolithique: nouveaux outils, nouvelles bene ts of cooking local foods could illuminate why, or why not, interprétations. UISPP, XV Congress (Lisbon, 4–9 Septembre 2006). Isabelle Neanderthals chose to cook their dinners. Théry-Parisot, Sandrine Costamagno, and Auréade Henry, eds. Pp. 69–84. Oxford: Archaeopress. Debono Spiteri, Cynthianne, Antje Hutschenreuther, Jörg Watzke, and Amanda G. Henry. 2014. Starch taphonomy on stone tools: considering anthropo- genic alterations, climate, and soil chemistry. In Proceedings of the 23rd Acknowledgments meeting of the Paleoanthropology Society. Calgary, Alberta. De Filippo, Carlotta, Duccio Cavalieri, Monica Di Paola, Matteo Ramazzotti, This work was funded in part by the Max Planck Society and Jean Baptiste Poullet, Sebastien Massart, Silvia Collini, Giuseppe Pieraccini, by the European Research Council (ERC) under the European and Paolo Lionetti. 2010. Impact of diet in shaping gut microbiota revealed ’ by a comparative study in children from Europe and rural Africa. Proceedings Union s Horizon 2020 research and innovation program, un- – “ ” of the National Academy of Sciences of the USA 107(33):14691 14696, doi: der grant agreement STG-677576 ( HARVEST ). Many thanks 10.1073/pnas.1005963107. to the members of the Plant Foods Group at the MPI-EVA, Dering, Phil. 1999. Earth-oven plant processing in Archaic Period economies: especially Robert Power and Thomas Büdel, some of whose an example from a semi-arid savannah in south-central North America. American Antiquity 64(4):659–674, doi:10.2307/2694211. efforts made up part of this paper. Dibble, Harold L., Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, Shan- non P. McPherron, and Dennis M. Sandgathe. 2017. How did hominins adapt to Ice Age Europe without fire? Current Anthropology 58(suppl. 16): S278–S287. References Cited Dibble, Harold L., Francesco Berna, Paul Goldberg, Shannon P. McPherron, Susan M. Mentzer, Laura Niven, Daniel Richter, Dennis M. Sandgathe, Isabelle Aarts, Jac M. M. J. G., Gerrit M. Alink, Fulco Scherjon, Katharine MacDonald, Théry-Parisot, and Alan Turq. 2009. A preliminary report on Pech de l’Azé IV, Alison C. Smith, Harm Nijveen, and Wil Roebroeks. 2016. Fire usage and layer 8 (Middle Paleolithic, France). PaleoAnthropology 2009:182–219. ancient hominin detoxification genes: protective ancestral variants domi- Elbaum, Rivka, Steve Weiner, Rosa M. Albert, and Michael Elbaum. 2003. De- nate while additional derived risk variants appear in modern humans. PLoS tection of burning of plant materials in the archaeological record by changes ONE 11(9):e0161102, doi:10.1371/journal.pone.0161102. in the refractive indices of siliceous phytoliths. Journal of Archaeological Sci- Aiello, Leslie C., and Peter Wheeler. 1995. The expensive-tissue hypothesis: ence 30:217–226. the brain and the digestive system in human and primate evolution. Cur- Elton, Sarah. 2008. The environmental context of human evolutionary history rent Anthropology 36(2):199–221. in Eurasia and Africa. Journal of Anatomy 212(4):377–393, doi:10.1111/j Aldeias, Vera, Harold L. Dibble, Dennis M. Sandgathe, Paul Goldberg, and .1469–7580.2008.00872x. Shannon P. McPherron. 2016. A controlled experiment: how heat alters Gale, Rowena, and Wendy Carruthers. 2000. Charcoal and charred seed re- underlying deposits and implications for archaeological fire features. Journal mains from Middle Palaeolithic levels at Gorham’s and Vanguard Caves. In of Archaeological Science Reports 67:64–79. Neanderthals on the edge. Chris B. Stringer, R. N. E. Barton, and J. C. Fin- Babot, Maria del Pilar. 2003. Starch grain damage as an indicator of food layson, eds. Pp. 207–210. Oxford: Oxbow. processing. In Phytolith and starch research in the Australian-Pacific-Asian Goldberg, Paul. 2003. Some observations on Middle and Upper Palaeolithic regions: the state of the art. D. M. Hart and L. A Wallis, eds. Pp. 69–81. ashy cave and rockshelter deposits in the Near East. In More than meets the Canberra, Australia: Pandanus. eye: studies on Upper Palaeolithic diversity in the Near East. A. N. Goring- BeMiller, J. N, and R. L Whistler. 2009. Starch: chemistry and technology. Morris and Anna Belfer-Cohen, eds. Pp. 19–32. Oxford: Oxbow. Burlington, MA: Academic Press. Goldberg, Paul, Harold Dibble, Francesco Berna, Dennis Sandgathe, Shannon Berna, Francesco, Paul Goldberg, Liora Kolska Horwitz, James Brink, Sharon J. P. McPherron, and Alain Turq. 2012. New evidence on Neandertal use of Holt, Marion Bamford, and Michael Chazan. 2012. Microstratigraphic evi- fire: examples from Roc de Marsal and Pech de l’Azé IV. Quaternary In- dence of in situ fire in the Acheulean atrata of Wonderwerk Cave, Northern ternational 247:325–340. Henry Neanderthal Cooking and the Costs of Fire S335

Gómez de la Rua, D., C. Mallol, B. Galván, and C. M. Hernández Gómez. areas of Nepal. Washington, DC: International Food Policy Research In- 2010. Una visión geoarqueológica general del yacimiento Musteriense de El stitute. Salt (Alcoy, Alicante) a partir de la micromorfología de suelos. Recerques Ledl, Franz, and Erwin Schleicher. 1990. New aspects of the Maillard Reaction del Museu d’Alcoi 19:19–32. in foods and in the human body. Angewandte Chemie International Edition Gowlett, J. A. J., J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina. 2017. in English 29(6):565–594, doi:10.1002/anie.199005653. Evidence of burning from bushfires in southern and east Africa and its Lee, Richard Borshay. 1979. The !Kung San: men, women, and work in a relevance to hominin evolution. Current Anthropology 58(suppl. 16):S206– foraging society. Cambridge: Cambridge University Press. S216. Leonard, Chelsea, Layne Vashro, James F. O’Connell, and Amanda G. Henry. Gowlett, John A. J., Jack Harris, D. Walton, and Bernard A. Wood. 1981. 2015. Plant microremains in dental calculus as a record of plant con- Early archaeological sites, hominid remains and traces of fire from Che- sumption: a test with Twe forager-horticulturalists. Journal of Archaeo- sowanja, Kenya. Nature 294:125–129, doi:10.1038/294125a0. logical Science: Reports 2(June):449–457, doi:10.1016/j.jasrep.2015.03.009. Gowlett, John A. J., and Richard W. Wrangham. 2013. Earliest fire in Africa: Lev, Efraim, Mordechai E. Kislev, and Ofer Bar-Yosef. 2005. Mousterian towards the convergence of archaeological evidence and the cooking hy- vegetal food in Kebara Cave, Mt. Carmel. Journal of Archaeological Science pothesis. Azania: Archaeological Research in Africa 48(1):5–30, doi:10.1080 32:475–484. /0067270X.2012.756754. Lieverse, Angela R. 1999. Diet and the aetiology of dental calculus. Interna- Groopman, Emily E., Rachel N. Carmody, and Richard W. Wrangham. 2015. tional Journal of Osteoarchaeology 9:219–232. Cooking increases net energy gain from a lipid-rich food. American Journal Manning, Dale T., and J. Edward Taylor. 2014. Migration and fuel use in rural of Physical Anthropology 156(1):11–18, doi:10.1002/ajpa.22622. Mexico. Ecological Economics 102(June):126–136, doi:10.1016/j.ecolecon Hansen, Julie Marie. 1991. The palaeoethnobotany of Franchthi Cave: exca- .2014.03.012. vations at Franchthi Cave, Greece 7. Bloomington: Indiana University Press. Marlowe, Frank W. 2010. The Hadza: hunter-gatherers of Tanzania. Berkeley: Hardy, Karen, Stephen Buckley, Matthew J. Collins, Almudena Estalrrich, Don University of California Press. Brothwell, Les Copeland, Antonio García-Tabernero, et al. 2012. Neander- Massey, Fergus P., and Sue E. Hartley. 2006. Experimental demonstration of thal medics? evidence for food, cooking, and medicinal plants entrapped in the antiherbivore effects of silica in grasses: impacts on foliage digestibility dental calculus. Naturwissenschaften 99(8):617–626. and vole growth rates. Proceedings of the Royal Society B 273:2299–2304. Hardy, Karen, Anita Radini, Stephen Buckley, Rachel Sarig, Les Copeland, Avi Matsutani, A. 1987. Plant remains from the 1984 excavations at Douara Cave. Gopher, and Ran Barkai. 2016. Dental calculus reveals potential respiratory In Paleolithic site of Douara Cave, bulletin no. 21. T. Akazawa and Y. Sa- irritants and ingestion of essential plant-based nutrients at Lower Palaeo- kaguchi, eds. Pp. 117–122. Tokyo: Tokyo University. lithic Qesem Cave Israel. State of the art of the multidisciplinary research at Mazza, Paul, Peter Anthony, Fabio Martini, Benedetto Sala, Maurizio Magi, the Middle Pleistocene Qesem Cave, Israel, 2015. Special issue, Quaternary Maria Perla Colombini, Gianna Giachi, et al. 2006. A new Palaeolithic dis- International 398(April):129–135, doi:10.1016/j.quaint.2015.04.033. covery: tar-hafted stone tools in a European Mid-Pleistocene bone-bearing Hehemann, Jan-Hendrik, Gaëlle Correc, Tristan Barbeyron, William Helbert, bed. Journal of Archaeological Science 33(9):1310–1318, doi:10.1016/j.jas Mirjam Czjzek, and Gurvan Michel. 2010. Transfer of carbohydrate-active .2006.01.006. enzymes from marine bacteria to Japanese gut microbiota. Nature 464(7290): Messner, T. C., and B. Schindler. 2010. Plant processing strategies and their 908–912, doi:10.1038/nature08937. affect upon starch grain survival when rendering Peltandra virginica (L.) Henry, Amanda G. 2010. Plant foods and the dietary ecology of Neandertals Kunth, Araceae edible. Journal of Archaeological Science 37(2):328–336. and modern humans. PhD dissertation, George Washington University, Miller, Naomi F. 1984. The use of dung as fuel: an ethnographic example and Washington, DC. an archaeological application. Paléorient 10(2):71–79. ———. 2015. Formation and taphonomic processes affecting starch granules. ———. 1996. Seed eaters of the ancient Near East: human or herbivore? In Method and theory in paleoethnobotany. John Marston, Jade d’Alpoim Current Anthropology 37(3):521–528. Guedes, and Christina Warinner, eds. Pp. 35–50. Boulder: University Press Morin, Eugène. 2004. Late Pleistocene population interaction in western Eu- of Colorado. rope and modern human origins: new insights based on the faunal remains Henry, Amanda G., Alison S. Brooks, and Dolores R. Piperno. 2011. Micro- from Saint-Césaire, southwestern France. PhD dissertation, Université de fossils in calculus demonstrate consumption of plants and cooked foods in Bordeaux. Neanderthal diets (Shanidar III, Iraq; Spy I and II, Belgium). Proceedings of Mottram, Donald S., Bronislaw L. Wedzicha, and Andrew T. Dodson. 2002. the National Academy of Sciences of the USA 108(2):486–491. Food chemistry: acrylamide is formed in the Maillard Reaction. Nature 419 ———. 2014. Plant foods and the dietary ecology of Neanderthals and early (6906):448–449, doi:10.1038/419448a. modern humans. Journal of Human Evolution 69(April):44–54, doi:10.1016 Murdock, George P., and Caterina Provost. 1973. Factors in the division of /j.jhevol.2013.12.014. labor by sex: a cross-cultural analysis. Ethnology 12(2):203–225, doi:10.2307 Henry, Amanda G., Holly F. Hudson, and Dolores R. Piperno. 2009. Changes /3773347. in starch grain morphologies from cooking. Journal of Archaeological Sci- Ofek, Haim. 2001. Second nature: economic origins of human evolution. Cam- ence 36(3):915–922. bridge: Cambridge University Press. Henry, Amanda G., and Dolores R. Piperno. 2008. Using plant microfossils Piperno, Dolores R. 2006. Phytoliths: a comprehensive guide for archaeologists from dental calculus to recover human diet: a case study from Tell Al-Raqā’i, and paleoecologists. Lanham, MD: AltaMira. Syria. Journal of Archaeological Science 35(7):1943–1950. Power, Robert C., Domingo C. Salazar-García, Mauro Rubini, Andrea Darlas, Hlubik, Sarah, Francesco Berna, Craig Feibel, David Braun, and John W. K. Katerina Harvati, Michael Walker, Jean-Jacques Hublin, and Amanda G. Harris. 2017. Researching the nature of fire at 1.5 Mya on the site of FxJj20 Henry. Forthcoming. Dental calculus indicates widespread plant use within AB, Koobi Fora, Kenya, using high-resolution spatial analysis and FTIR the recurrent Neanderthal dietary niche. Journal of Human Evolution. spectrometry. Current Anthropology 58(suppl. 16):S243–S257. Power, Robert C., Domingo C. Salazar-García, Roman M. Wittig, Martin Hubbard, Troy D., Iain A. Murray, William H. Bisson, Alexis P. Sullivan, As- Freiberg, and Amanda G. Henry. 2015. Dental calculus evidence of Taï wathy Sebastian, George H. Perry, Nina G. Jablonski, and Gary H. Perdew. forest chimpanzee plant consumption and life history transitions. Scientific 2016. Divergent Ah receptor ligand selectivity during hominin evolution. Reports 5(October):15161, doi:10.1038/srep15161. Molecular Biology and Evolution, August, msw143, doi:10.1093/molbev/msw143. Pruetz, Jill D., and Nicole M. Herzog. 2017. Savanna chimpanzees at Fongoli, Jin, Ye, and Hak-Kong Yip. 2002. Supragingival calculus: formation and con- Senegal, navigate a fire landscape. Current Anthropology 58(suppl. 16): trol. Critical Reviews in Oral Biology and Medicine 13(5):426–441. S337–S350. Kaufman, Brett. 2013. Copper alloys from the ’Enot Shuni cemetery and the Pryor, A. J. E., A. Pullen, D. G. Beresford-Jones, J. A. Svoboda, and C. S. origins of bronze metallurgy in the EB IV–MB II Levant. Archaeometry 55 Gamble. 2016. Refl ections on Gravettian firewood procurement near the (4):663–690, doi:10.1111/j.1475–4754.2012.00699x. Pavlov Hills, Czech Republic. Journal of Anthropological Archaeology 43 Kaufman, Brett, and D. A. Scott. 2014. Fuel efficiency of ancient copper alloys: (September):1–12, doi:10.1016/j.jaa.2016.05.003. theoretical melting thermodynamics of copper, tin and arsenical copper and Rampelli, Simone, Stephanie L. Schnorr, Clarissa Consolandi, Silvia Turroni, timber conservation in the Bronze Age Levant. Archaeometry 57(6):1009– Marco Severgnini, Clelia Peano, Patrizia Brigidi, Alyssa N. Crittenden, 1024, doi:10.1111/arcm.12127. Amanda G. Henry, and Marco Candela. 2015. Metagenome sequencing of Kumar, Shubh K., and David Hotchkiss. 1988. Consequences of deforestation the Hadza hunter-gatherer gut microbiota. Current Biology 25(13):1682– for women’s time allocation, agricultural production, and nutrition in hill 1693, doi:10.1016/j.cub.2015.04.055. S336 Current Anthropology Volume 58, Supplement 16, August 2017

Roebroeks, Wil, and Paola Villa. 2011. On the earliest evidence for habitual Journal of Human Evolution 89(December):181–201, doi:10.1016/j.jhevol use of fire in Europe. Proceedings of the National Academy of Sciences of the .2015.04.004. USA 108(13):5209–5214, doi:10.1073/pnas.1018116108. Tanno, Tadashi. 1981. Plant utilization of the Mbuti pygmies—with special Salazar-García, Domingo C., Robert C. Power, Alfred Sanchis Serra, Valentín reference to their material culture and use of wild vegetable foods. African Villaverde, Michael J. Walker, and Amanda G. Henry. 2013. Neanderthal Study Monographs 1:1–53. diets in central and southeastern Mediterranean Iberia. Quaternary Inter- Théry-Parisot, Isabelle. 2001. Économie des combustibles au Paléolithique: ex- national 318:3–18, doi:10.1016/j.quaint.2013.06.007. périmentation, taphonomie, anthracologie. Dossier de Documentation Archéo- Sandgathe, Dennis M. 2017. Identifying and describing pattern and process in logique 20. Paris: CNRS Éditions. the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): ———. 2002. Fuel management (bone and wood) during the Lower Auri- S360–S370. gnacian in the Pataud rock shelter (Lower Palaeolithic, Les Eyzies de Tayac, Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. Dordogne, France): contribution of experimentation. Journal of Archaeo- McPherron, Alain Turq, Laura Niven, and Jamie Hodgkins. 2011. On the logical Science 29(12):1415–1421, doi:10.1006/jasc.2001.0781. role of fire in Neandertal adaptations in western Europe: evidence from Théry-Parisot, Isabelle, and Sandrine Costamagno. 2005. Proprétés combus- Pech de l’Azé IV and Roc de Marsal, France. PaleoAnthropology 2011:216– tibles des ossements. Gallia Préhistoire 47:235–254. 242. Théry-Parisot, Isabelle, and Liliane Meignen. 2000. Économie des combustibles Schnorr, Stephanie L., Marco Candela, Simone Rampelli, Manuela Centanni, (bois et lignite) dans l’abri Moustérien des Canalettes [l’expérimentation à la Clarissa Consolandi, Giulia Basaglia, Silvia Turroni, et al. 2014. Gut mi- simulation des besoins énergétiques]. Gallia Préhistoire 42(1):45–55, doi:10 crobiome of the Hadza hunter-gatherers. Nature Communications 5(April), .3406/galip.2000.2169. doi:10.1038/ncomms4654. Twomey, Terrence. 2013. How domesticating fire facilitated the evolution of Schnorr, Stephanie L., Alyssa N. Crittenden, Koen Venema, Frank W. human cooperation. Biology and Philosophy 29(1):89–99, doi:10.1007/s10539 Marlowe, and Amanda G. Henry. 2015. Assessing digestibility of Hadza –013-9402–2. tubers using a dynamic in-vitro model. American Journal of Physical An- Vallverdú, Josep, Susana Alonso, Amèlia Bargalló, Raúl Bartrolí, Gerard Cam- thropology 158(3):371–385, doi:10.1002/ajpa.22805. peny, Ángel Carrancho, Isabel Expósito, et al. 2012. Combustion structures Shackleton, C. M., and F. Prins. 1992. Charcoal analysis and the “principle of of archaeological level O and Mousterian activity areas with use of fire at the least effort”—a conceptual model. Journal of Archaeological Science 19 Abric Romaní rockshelter (NE Iberian Peninsula). Quaternary International (6):631–637, doi:10.1016/0305–4403(92)90033-Y. 247:313–324. Shahack-Gross, Ruth, Ofer Bar-Yosef, and Steve Weiner. 1997. Black-coloured van Wijngaarden, Jessica. 1984. The patterns of fuel gathering and use in a bones in Hayonim Cave, Israel: differentiating between burning and oxide rural area of Machakos District, Kenya. Wageningen, Netherlands: Wagen- staining. Journal of Archaeological Science 24(5):439–446, doi:10.1006/jasc ingen Agricultural University. .1996.0128. Vidal-Matutano, Paloma, Cristo M. Hernández, Bertila Galván, and Carolina Smillie, Chris S., Mark B. Smith, Jonathan Friedman, Otto X. Cordero, Lawrence Mallol. 2015. Neanderthal firewood management: evidence from strati- A. David, and Eric J. Alm. 2011. Ecology drives a global network of gene graphic unit IV of Abric Del Pastor (eastern Iberia). Quaternary Science exchange connecting the human microbiome. Nature 480(7376):241–244, Reviews 111(March):81–93, doi:10.1016/j.quascirev.2015.01.010. doi:10.1038/nature10571. Warinner, Christina, Camilla Speller, and Matthew J. Collins. 2015. A new era Smith, Alex R., Rachel N. Carmody, Rachel J. Dutton, and Richard W. in palaeomicrobiology: prospects for ancient dental calculus as a long-term Wrangham. 2015. The significance of cooking for early hominin scaveng- record of the human oral microbiome. Philosophical Transactions of the ing. Journal of Human Evolution 84:62–70, doi:10.1016/j.jhevol.2015.03 Royal Society B 370(1660):20130376, doi:10.1098/rstb.2013.0376. .013. Wiessner, Polly W. 2014. Embers of society: firelight talk among the Ju/ Smith, Kirk R., Jonathan M. Samet, Isabelle Romieu, and Nigel Bruce. 2000. ’hoansi bushmen. Proceedings of the National Academy of Sciences of the Indoor air pollution in developing countries and acute lower respiratory USA 111(39):14027–14035, doi:10.1073/pnas.1404212111. infections in children. Thorax 55(6):518–532, doi:10.1136/thorax.55.6.518. Wilson, Lucy. 2007a. Terrain difficulty as a factor in raw material procure- Speth, John D., and Eitan Tchernov. 2001. Neandertals hunting and meat- ment in the Middle Palaeolithic of France. Journal of Field Archaeology 32 processing in the Near East. In Meat-eating and human evolution.CraigB. (3):315–324, doi:10.1179/009346907791071539. Stanford and Henry T. Bunn, eds. Pp. 52–72. Oxford: Oxford University Press. ———. 2007b. Understanding prehistoric lithic raw material selection: ap- Stahl, Ann Brower. 1984. Hominid dietary selection before fire. Current An- plication of a gravity model. Journal of Archaeological Method and Theory thropology 25(2):151–168. 14(4):388–411, doi:10.1007/s10816–007-9042–4. Stahlschmidt, Mareike C., Christopher E. Miller, Bertrand Ligouis, Ulrich Wrangham, Richard W. 2009. Catching fire: how cooking made us human. Hambach, Paul Goldberg, Francesco Berna, Daniel Richter, Brigitte Urban, New York: Basic. Jordi Serangeli, and Nicholas J. Conard. 2015. On the evidence for human Wrangham, Richard W., James Holland Jones, Greg Laden, David R. Pilbeam, use and control of fire at Schöningen. In Excavations at Schöningen: new and Nancy Lou Conklin-Brittain. 1999. The raw and the stolen: cooking insights into Middle Pleistocene lifeways in northern Europe. Special issue, and the ecology of human origins. Current Anthropology 40(5):567–594. Current Anthropology Volume 58, Supplement 16, August 2017 S337

Savanna Chimpanzees at Fongoli, Senegal, Navigate a Fire Landscape

by Jill D. Pruetz and Nicole M. Herzog

Savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal, appear to be able to predict the “behavior” of wildfires of various intensities. Although most wildfires are avoided, even the most intense fires are met with relative calm and seemingly calculated movement by apes in this arid, hot, and open environment. In addition to reviewing instances of such behavior collected during the course of the Fongoli study, we also report chimpanzees’ use of burned landscapes during the dry season, when more than 75% of these apes’ home range may be burned annually. In burned areas, chimpanzees spent more time foraging and traveling than in unburned areas. Chimpanzees’ behavior in a fire con- text can help inform paleoanthropological hypotheses regarding early members of our own lineage and can provide insight into the ability of early hominins to conceptualize the behavior of fire and thus set the stage for our lineage’s use of fire.

Chimpanzees are frequently used as referential models to in- anthropologists as to how living apes adjust to a hot, open, form our knowledge of human evolution because observations and dry habitat, data on which we are lacking relative to our of living apes, when coupled with archaeological, ethnograph- knowledge of how monkeys adapt and adjust to such an en- ical, and paleoanthropological evidence, allow us to hypothe- vironment (Pruetz and Bertolani 2009). This approach, when size about the behavior of extinct hominins (defined here as bi- bolstered by a focus on general patterns of behavior dictated pedal apes). Data on living apes can be especially informative by specific ecological change (Vaesen 2014), can lead to pre- regarding subjects such as the origins of fire use when the dictions about probable ancestral responses to landscape fires paleoanthropological and archaeological record is expectedly and so provide the basis for hypothesizing possible archaeo- sparse. Using cladistic analyses of behavior, Pruetz and LaDuke logical or paleontological consequences of increased fire fre- (2010) argue that understanding the capabilities of living apes quency. can help anthropologists identify traits that are derived and Given the regularity of seasonal fires in southeastern Sen- those that probably also characterized early hominins, that is, egal, the Fongoli community of chimpanzees is an exception- primitive traits for our lineage. Specifically, chimpanzees using ally well-suited group from which to build hypotheses regard- more open habitats such as the mosaic savanna-woodland envi- ing the earliest hominin interactions with fire. We previously ronment in southeastern Senegal face environmental pressures reported the responses of these chimpanzees to wildfires in similar to those of early hominins who experienced increas- this savanna-woodland environment and put this behavior into ing aridification and savanna expansions at or near 2.8 Mya the context of a referential model (Pruetz and LaDuke 2010 (Cerling et al. 2011; deMenocal 2004, 2011, 1995; Wolde- sensu Moore 1996). Ideally, using a relational form of a ref- Gabriel et al. 2009; Wynn 2004). This environmental shift in- erential model (Moore 1996) in which savanna chimpanzee fluenced the structure of mammalian populations, including behavior is compared with chimpanzees living in forested en- hominins, in eastern Africa (Bobe and Behrensmeyer 2004; vironments helps control for the tendency to directly analogize Vrba 1995). Adaptation to this emerging mosaic habitat has chimpanzees as representative of early hominins. However, all been indicated as a driving factor in hominin evolution (Blu- other habituated chimpanzees live in forested habitats where menschine 1986; Potts 1998; Reed 1997). As such, knowledge fire is a rare and potentially catastrophic occurrence, such as of their behavioral responses to the local ecology can inform has been seen with Asian great apes and large-scale peat fires (Nellemann et al. 2007). To date, little opportunity to examine great ape interaction with fire in contexts such as those per- Jill D. Pruetz is Professor in the Department of Anthropology at haps experienced by early hominins has been available. Iowa State University (324 Curtiss Hall, Ames, Iowa 50011, USA [[email protected]]). Nicole M. Herzog is Postdoctoral Researcher Chimpanzees at the Fongoli study site adjust their move- fi in the Department of Anthropology at the University of Utah (201 ments to the intensity and movement of wild res during the fi Stewart, Salt Lake City, Utah 84112, USA [nicole.herzog@anthro late dry season, and because wild res are regular and exten- .utah.edu]). This paper was submitted 25 VII 16, accepted 5 III 17, sive at Fongoli, we suggested that chimpanzees here exhibit and electronically published 2 VI 17. behavior in the face of wildfires that reflects the earliest cog- q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0016$10.00. DOI: 10.1086/692112 S338 Current Anthropology Volume 58, Supplement 16, August 2017 nitive capabilities required of hominins for the use of fire Conceptualizing Fire Behavior: A Precondition (Pruetz and LaDuke 2010). Molecular and morphological evi- for Fire Control dence suggest that chimpanzees shared a last common ancestor with the human lineage 4–8 Mya (see reviews in Kumar et al. Fire is a potentially dangerous resource whose behavior is 2005; Langergraber et al. 2012), and making the parsimonious complex and variable (Trollope, de Ronde, and Geldenhuys assumption that this is a primitive hominid (here defined as the 2004). We noted that the ability to conceptualize a fire is a clade including the Pan and Homo lineages) trait, such a ca- prerequisite to fire use and defined this as “an understanding pability would support the assertion that use of fire was a rel- of the behavior of fire under varying conditions that would atively early event in the hominin lineage. If so, better under- allow one to predict its movement, thus permitting activity standing savanna-chimpanzee interaction with fire and burned in close proximity to the fire” (table 1; Pruetz and LaDuke landscapes can help inform hypotheses regarding the context 2010:4). This cognitive step has been largely overlooked in of early hominins’ exposure to and reaction to fire. reconstructions of hominin-fire interactions (but see Burton In addition to providing updated information on chim- 2009; Frazer 1930; Goudsblom 1986). As Sandgathe (2010) panzee encounters with wildfires, we present data on their use notes, however, the use of fire was probably characterized by of burned landscape during the dry season in order to test the increasing levels of sophistication, but a linear progress did hypothesis that Fongoli chimpanzees use such areas differ- not necessarily occur in all populations. Without understand- ently from nonburned areas in this savanna-woodland envi- ing the behavior of nonhumans in proximity to fire, predict- ronment. The frequent burning at Fongoli is probably atypical ing the earliest stages of fire use in hominins becomes even in terms of the frequency with which early hominins would more speculative, only partially explaining the derived traits have encountered wildfire. Nonetheless, it gives us insight into associated with fire use characterizing our lineage. the behavioral adjustments that an ape with a cranial capacity Little research has been done on the topic of immediate only approaching the earliest hominins in size can make in animal responses to fire. Those studies that do address this of- response to potentially life-threatening wildfire and can help ten simply note when subjects flee (Burton 2009) and, where inform hypotheses as to how and why remaining in close available, typically focus on animals’ survivorship following proximity to such fires could occur. If wild chimpanzees living wildfire (Berenstain 1986: Macaca fascicularus; Esque et al. today possess the cognitive sophistication of accurately pre- 2003: Gopherus agassizii; Jaffe and Isbell 2009: Cercopithecus dicting wildfire behavior, we can support putative evidence aethiops; Koprowski et al. 2006: Tamiasciurus hudsonicus;Mikami that hominins of the genus Homo were more than capable of et al. 2009: Chlamydera nuchalis;Pringleetal.2015:Lygodactylus safely navigating wildfire and understanding its behavior and keniensis, Acacia ant genera; Vernes 2000: Bettongia tropica;Woo- that perhaps even pre-Homo species possessed this capability. ley et al. 2008: Loxodonta africana). Fewer data on animals’ reac-

Table 1. Review of proposed deep-time steps/stages in fire control

Source and step/stage Associated behaviors/adaptations Age Stages of fire domestication (Burton 2009): Approach Approach fire, interact with food items altered by burning 6 Mya Associate Develop a cognitive relationship regarding the benefits of fire and 4–6 Mya seek it out; tending and spreading naturally occurring fires Manufacture Ability to create flames outside of naturally occurring fire events !250 kya Three cognitive stages in the control of fire (Pruetz and LaDuke 2010): Conceptualization of fire Understanding fire behavior under varying conditions; ability Before the human-chimpanzee to predict fire’s movement, thereby enabling close contact split, 16–10 Mya Fire control Ability to contain fire, provide or deprive the fire of fuel, and NA possibly the ability to extinguish fire Fire creation Ability to start a fire NA Pyrophilic shifts in hominin evolution (Parker et al. 2016): Passive pyrophilia Resistant to the effects of landscape fire Before the human-chimpanzee split, 16–10 Mya Active pyrophilia Reliance on fire-mediated foraging improvements After the human-chimpanzee split, !6–10 Mya Obligate pyrophilia Dependence on fire and cooked foods for survival and reproduction Before/coincident with emer- gence of Homo erectus ~2 Mya Pruetz and Herzog Chimpanzees Navigate a Fire Landscape S339 tion to, or interaction with, direct exposure to fire exist, with the innovation of fire use had taken place. Specifically, we pro- some exceptions (table 2). While the data are at present limited vide new data on chimpanzee reaction to fire and these apes’ (see Herzog et al 2014; table 1 for summary), nonhuman pri- use of burned landscapes and discuss these results as they relate mates are perhaps the best organisms to examine in order to to early hominins’ interaction with fire as well as fire use by the refine the differences between humans and other species. Spe- genus Homo. cifically, primates can serve as referential models of how our ear- liest ancestors may have responded to fire, especially given our Methods order’s emphasis on learning and cognition as well as the po- Study Subjects tential to identify homology in our taxon. We propose that chim- panzees at Fongoli can provide insight into a number of ques- The Fongoli chimpanzee community size is, on average, 32 in- tions associated with fire use in hominins. First, do chimpanzees dividuals (2005–2014; Pruetz et al. 2015). The community has exhibit behavior like monkeys (Armelagos 2010; Berenstain 1986; consistently contained more adult males (9–12) than females Harrison 1983, 1984; Herzog et al. 2014, 2016; Jaffe and Isbell (7–9), but a large number of immature individuals also char- 2009), which are attracted to burned or burning areas? Second, acterize the group (11–15). Adult males were used as focal sub- what would the impetus for attraction to fire be in early homi- jects per the Fongoli Savanna Chimpanzee Project’sprotocol. nins? Proximity to fire is, of course, a prerequisite for the ulti- One focal subject was chosen each day, based on the previous mate use of it. Minimally, conceptualization of fire behavior month’s order, so that focal sampling was not completely ran- would allow hominins to feed and range in areas prone to fire, dom but depended on when males were observed the previous thus avoiding costs associated with the disruption of activity in month, and observers attempted to follow this order. Male landscapes subject to burning. Not insignificant are the dan- absence from the larger social group or loss of a focal subject gers associated with remaining in the vicinity of wildfires, where meant that focal subject order varied monthly to some degree. narcotic gases can compromise animals’ central nervous systems Females are not targeted as focal subjects because of the slight and cardiovascular systems, rapidly incapacitating an individ- but real risk of adult females being targeted to obtain infants ual (Purser 1986). Notably, within three minutes, rapidly growing for the pet trade (1 case in 15 years; Pruetz and Kante 2010). flaming fires can reach heat and noxious gas levels threatening Females are fully habituated in the presence of adult males but to humans (Purser 1986). Here, we present data on apes with op- are nervous around observers when encountered alone or in portunities much like those the earliest hominins faced before small parties.

Table 2. Immediate reactions of nonhumans to close-proximity fire

Species Reaction Context Source Orangutan (Pongo pygmaeus) Imitated humans’ fire making Rehabilitant apes Russon 2000; Russon and Galdikas 1993 Orangutan (P. pygmaeus) Set bags on fire with lit cigarettes and Captive ape Osborn 1912, in Mitchell used them to set more bags on fire 1999 Bonobo (Pan paniscus) Taught to make (with lighter) and use Language-trained captive ape Segerdahl, Fields, and fire for cooking Savage-Rumbaugh 2005 Chimpanzee (Pan troglodytes) Ignore; display at and through campfire Captive apes experimentally Charmove 1996 provided with campfire Chimpanzee (P. troglodytes) Ability to light and extinguish cigarettes Captives in zoo given cigarettes Brink 1957 Chimpanzee (P. troglodytes verus) Foraged burned seeds Reintroduction of ex-captive apes Brewer 1978 Chimpanzee (P. troglodytes verus) Avoid, predict movement, navigate Naturally occurring seasonal Pruetz and LaDuke 2010 wildfires Rhesus monkey (Macaca mulatta) Foraged burned coconuts in ashes of fire Free-ranging, introduced popula- S. Gouzoules, in tion; details of fire not provided Armelagos 2010 Vervet monkey (Chlorocebus Ignored Controlled burn Herzog et al. 2014 aethiops) Reed frogs (Hyperolius nitidulus) Avoided by moving into protective cover Experimental exposure to sounds Grafe, Dobler, and of fire Linsenmair 2002 Raccoon (Procyon lotor) Avoided (≥400 m) Controlled burn Sunquist 1967 Meadow vole (Microtus Avoided above- and belowground; Controlled burn Geluso, Schroder, pennsylvanicus) one death and Bragg 1986 Rook (Corvus frugilegus), Eurasian Anting behavior with burning embers Captive birds Goodman 1960 Jay (Garrulus glandiarus) or lit matches applied to feathers S340 Current Anthropology Volume 58, Supplement 16, August 2017

Study Site Data Collection The Fongoli study site is located in a woodland-savanna hab- Immediate encounters with fire. Previous data on chim- itat in southeastern Senegal (12740N, 12713W). The Sudano- panzees’ encounters with wildfire (n p 2) from the 2005– Guinean vegetation at Fongoli is characterized by a grass un- 2006 study period (Pruetz and LaDuke 2010) are included. derstory that covers 190% of the chimpanzees approximately New data collected by Pruetz from late 2006 through early 90 km2 home range. Grasses, such as Pennisetum purpureum, 2014 follow a similar data collection format, but in two in- can grow more than 2 m in height, so the late rainy season stances more detailed data on chimpanzees’ interactions with and early dry season are characterized by poor visibility and fire were collected using instantaneous sampling (see below). more difficult travel, of which almost 100% is done terrestri- Because wildfires occur in the dry season (October–April), ally at Fongoli (Jill D. Pruetz et al., unpublished data). Grass- chimpanzees are subject to fire 7 months each year. Data pre- land (tall and short) and woodland (including bamboo wood- sented here come from a total of 15 months, 8 months in the land) habitat types account for most of the chimpanzees’ home first half or early dry season (October 2010; November 2010; range. Closed habitats (gallery forest and ecotone forest) ac- December 2008, 2010, 2011, 2012; January 2012, 2014) and count for less than 5% of the range. Anthropogenic distur- 7 months in the latter half or late dry season (February 2008, bance such as fields, villages, artisanal gold mines, and dirt 2013; March 2006, 2008, 2010, 2012; April 2006). roads make up approximately 5% of the Fongoli chimpanzees’ When we observed chimpanzees’ encounters with wildfires, range. we collected data on their reactions to fire. In several cases, we The Fongoli site is characterized by significant heat stresses collected data on temperature and wind speed using a hand- for apes. There is one long dry season (November–April) and held Kestrel data logger. In two cases, we conducted system- a short wet season (June–September), with May and October atic observations using 5-minute instantaneous sampling, where being transitional months when some rain may fall. In 2014 we recorded individuals closest to fire each 5 minutes and noted and 2015, however, less than 50 mm of rain fell in June, but their behavior as well as proximity (in meters) to the fire. In May 2014 was characterized by 150 mm of rain. Rainfall be- most cases, however, we could not collect such systematic data, fore 2010 averaged less than 1,000 mm annually, but between as observers also ensured that they maintained a safe distance 2010 and 2015 rainfall has become more erratic, with some wet from the fire and did not approach wildfires as closely as some years of 11,000 mm and some very dry years. This pattern chimpanzees. Observers also mapped onto the chimpanzees’ matches the drying (Galat-Luong and Galat 2005) and warm- behavior so that they could safely navigate the wildfires and, ing trend noted in the region (Hillyer, Armstrong, and Kor- given the tendency for smoke to obscure visibility, staying stjens 2015) and has implications for further environmental with chimpanzees was considered a more reliable solution to pressures on primates here. Maximum temperatures at Fon- avoiding the fire than setting out alone and trying to move goli exceed 407C during the late dry season (Pruetz 2007), and around it. chimpanzees are reduced to between two and four permanent water sources available during the peak of the dry season. In Use of burned landscapes. We collected data on chim- addition to almost daily drinking and therefore cyclic ranging panzees’ use of burned landscapes from 2011 to 2015 during among the few available water sources, chimpanzees’ activity the dry seasons. Data collection in 2014 and 2015 was min- also reflects the pressures of heat stress. They rest significantly imal following a travel ban due to risks from the Ebola epi- during the day at this time and have been observed to move demic, which began in the neighboring country of Guinea. and feed at night (Pruetz and Bertolani 2009). Dry season fires Data collection on burned landscape use consisted of record- therefore present an additional pressure that could be viewed ing whether the focal subject used unburned, burned (focal as significantly negative if the fires interfere with travel, feed- subject was within a burned area of at least 20 m in diameter), ing and foraging, or even resting behavior. and partly burned (focal subject was within an area of un- Frequent fires characterize grasslands, including woodlands burned and burned vegetation or the area burned was less dominated by a grass understory (Veldman et al. 2015). At than 20 m in diameter) areas. While 20 m may seem sub- Fongoli, wildfires begin annually in October, and we estimate jective, it was used to attempt to distinguish between pur- that at least 75% of the chimpanzee range is burned on av- poseful use of a burned area as opposed to encountering a erage each year. The vegetation is fire adapted (Tappan et al. burned area by chance in a largely unburned landscape. We 2004), and most fires here are probably due to human activity. recorded data every 5 minutes (instantaneous sampling) dur- Fongoli chimpanzees, like most chimpanzees in Senegal, live ing focal male follows, and an attempt was made to follow the outside of national park boundaries, where humans set fires focal subject from the time he exited his night nest until he to clear land for cultivation, to aid in hunting, and to aid in made a new one at the end of the day. navigating the landscape at the end of the dry season when We collected a total of 305 hours of focal data on 12 adult dead and dying grass makes walking very difficult. Fires are male subjects with a mean of 25 hours of data collected per prohibited but still occur, and late dry-season fires are espe- male (range 6–49 hours per subject). We collected opportu- cially destructive to the vegetation. nistic data (n p 10.25 hours; mean 0.85 hours per individual) Pruetz and Herzog Chimpanzees Navigate a Fire Landscape S341 on chimpanzees in other age-sex classes (7 adult females, 2 ad- subsequently pooled in analyses. Because sample sizes for in- olescent males, 2 juvenile females, 1 juvenile male) when focal dividuals other than adult males were small, we also examined subjects were out of sight in order to increase sample sizes. In the effect of these data on results via separate analyses. these cases, we recorded the behavior of the nearest visible Chimpanzees spent slightly more time in burned (38%) chimpanzee during the interim when focal male subjects were versus unburned (35%) or partially burned (27%) areas, but out of sight. These individuals were not targeted for follows, these data are not corrected for the availability of the different and a new male subject was chosen based on the previous burning conditions. General activity (feeding, rest, travel, so- month’s order if the original subject was not seen again for cial behavior) differed according to the three burned condi- 20 minutes. We conducted analyses with and without these tions (unburned, fully burned, partially burned) both when all individuals in order to examine whether results might be individuals were included (x2 p 194.613, df p 6, P p .000) skewed because of low sample size per individual (range 0.9– or when only adult male subjects were considered (x2 p 1.75 hours). 219.163, df p 6, P p .000). Resting behavior occurred less Variables recorded included (1) focal subject identity, (2) date, often, but travel behavior occurred more often in fully burned (3) time of day, (4) activity (feed/forage, travel, rest, social areas (fig. 4). Individuals spent less time in social behavior or [agonism, affiliation], nontravel movement, vigilance [follow- rest in burned or partially burned areas (40% of activity in ing Nishida et al. 1999]), (5) habitat type (gallery or ecotone each condition), whereas these behaviors were characteristic forest, woodland, grassland, bamboo woodland, field/road; of unburned areas (58% of activity; fig. 4). Feeding was done see Bogart and Pruetz 2011 for detailed definitions of habitat most often in partially burned areas (47%) compared to burned types at Fongoli), (6) fire condition (burned, partially burned, (40%) and unburned (33%) areas (fig. 4). unburned), and (7) substrate used (arboreal or terrestrial).

Discussion Results Current evidence for hominin fire use is confined to archae- Immediate Encounters with Fire ological data (e.g., hearths; for a review see Gowlett and Wrang- We recorded chimpanzees’ direct encounter with wildfires a ham 2013) and paleoanthropological data (e.g., cranial and gut total of 18 times over the course of the 15-month study (ta- morphology associated with ingestion of cooked foods; Car- bles 3–5), seven of which occurred during the late dry-season mody and Wrangham 2009; Fonseca-Azevedo and Herculano- months of February–April. When chimpanzees encountered Houzel 2012; Wrangham 2009; Zink, Lieberman, and Lucas fires, they frequently traveled through burned areas, often 2014). As Burton (2009) points out, however, hominins ex- minutes after a fire had passed through (figs. 1, 2). Overall, in hibiting such evidence of possible fire use and association must encounters with wildfire, chimpanzees most often appeared have experienced the forces driving these adaptations for some to ignore fire (42% of scores given, n p 10; fig. 3). However, time before their visibility in morphology. That hominins such they also tended to avoid or leave (move out of sight of the as Homo erectus show adaptations potentially related to a shift fire) a burning area as well (29%, n p 7; fig. 3). Other atten- toward cooked foods indicates a much earlier onset for both tion to fire included monitoring its movement (21%, n p 5) active and passive uses of fire (Burton 2009; Parker et al. 2016). and navigating (here defined as moving in close proximity or Therefore, archaeological and paleontological data alone may less then 50 m to wildfire) through or around the fire (17%, be insufficient in identifying the steps leading up to visible n p 4; fig. 3). evidence of fire manufacture. We and others (table 1) suggest Fires are potentially most deadly in the late dry season, several cognitive and behavioral steps as prerequisites to the when even fire-adapted vegetation succumbs to burning, and ultimate control of fire. Using this logic, we propose that the fuel is especially dry. Chimpanzees at Fongoli ignored fires ear- evolutionary effects of fire use and association should appear lier in the dry season; late dry-season fires (fig. 3) were moni- in earlier stages of hominin evolution. tored more often. Chimpanzees were also more likely to avoid Systematic observations of the costs and benefits that land- or leave a burning area in the late dry season, but they were scape fires provide for primates living in fire-prone regions observed navigating burning areas in the early dry season, al- should lead to predictions about likely ancestral responses to though sample sizes are low (fig. 3). landscape fires, including scenarios of how hominins might intentionally associate with burned landscapes. Savanna chim- panzees, in particular, serve as an exemplary model for two Use of Burned Landscapes reasons: (1) their frequent exposure to fires may be similar to According to box plot analyses (SYSTAT, ver. 13), three adult that of hominins also inhabiting an open mosaic savanna- males were characterized by outlying data points in the un- woodland habitat, and (2) they appear strategically to use burned condition, which could possibly skew results. Pearson burned landscapes and exhibit cognitive abilities necessary for x2 analyses run with and without these males did not signif- interacting with wildfires, which tentatively provides support icantly affect the outcomes, so data for all male subjects were for the early fire-use theory. Table 3. Summary of encounters between chimpanzees and wildfires over 15 dry-season months at Fongoli, Senegal

Individuals Dry-season observed in Encounter Date period Time of day vicinity of firea Detailed description of chimpanzee behavior in context of fire Behavior summary 1 March 14, 2006 Late 1210–1700 5 am, 2 im Pruetz and LaDuke 2010 Monitor and eventually leave 2 April 1, 2006 Late 1149–1325 2 am, 2 af, 7 im Pruetz and LaDuke 2010 Move ahead and monitor; eventually leave 3 February 27, 2008 Late 1928 7 am, 6 af, 11 im Seemingly ignore fire heard by observer WNW of nest site, Apparently ignore fire distance unknown. 4 March 12/13, 2008 Late 1912; 0626 9 am, 3 af, 3 im On evening of March 12, fire burned in streambed (!50 m) below Apparently ignore fire or chimpanzee nest site. On March 13, it appeared that fire had at least do not abandon burned through nest site, including some grass and fallen leaves, nest site although previous burn probably reduced intensity of this one (understory grasses and shrubs had been burned in January or February of this year).

S342 5 January 4, 2009 Early 1615 2 af, 1 im Recent fire has burned through the area, with fuel still smoldering. Ignore smoldering fire Chimpanzees ignore it. 6 March 23, 2010 Late 0920–1015 1 af, 1 im After fire is heard WNW (vegetation popping and smoke visible), Avoids fire then leaves female with infant moves in opposite direction. Wind p area 1.3 km/hour, temperature p 33.37C. She initially moves away from fire but cuts back across streambed to N then E. Area E and S of streambed is dry, tall elephant grass. She abandons fig feeding and leaves area at 1015 hours. No wind, temperature 34.47C. 7 October 19, 2010 Early 1814 1 af, 1 am, 1 im Fire near charcoal kiln. Party travels through area, feeding and Ignore moving. No discernible reaction to fire. 8 October 20, 2010 Early 1750–1820 6 am, 1 af, 1 im Observer hears fire at 1750 hours and party passes to N of fire, Avoid, use burned area as about 200–300 m from smoke. Party crosses burned area at corridor 1815 hours, using it as a travel corridor through the tall grass. 9 October 27, 2010 Early 1620–1655 4 am, 2 af, 2 im Observer hears fire at 1620 hours to SE. Party feeding and moving. Vigilant, probably leave Chimpanzees feeding in baobab tree seem to watch the fire, area which is 100–200 m away. Party leaves hurriedly at 1645 hours, either in response to the fire or perhaps a person approaching (though nothing besides their hurried behavior indicates a person is near). Adult male (alpha male DV) stops and scans the area frequently. Observer can smell smoke. 10 November 16/17, 2010 Early 1850 (November 16); 7 am, 4 af, 10 im Observer is about 630 m from nest site at Maragoundi (dry) Ignore 0600 (November 17) streambed and can hear fire, which we also heard at nest site. Fire never reaches nest site. November 17: can see fire never reached nest site. Not heard this morning. Table 3 (Continued)

Individuals Dry-season observed in Encounter Date period Time of day vicinity of firea Detailed description of chimpanzee behavior in context of fire Behavior summary 11 November 18, 2010 Early 1745–1930 7 am, 3 af, 8 im Can hear fire E of party at 1745 hours. Sounds closer at 1820 hours Ignore and upon leaving nested party (1930 hours). Observer finds fire about 400 m from their nest site. Fire is moving in that direc- tion. Chimpanzees remained in this nest site. 12 November 20/21, 2010 Early 1220, 1705–2007 2 am, 1 af, 1 im November 20: fire heard in distance to S at 1220, 1705 and 1903 Ignore (November 20); (party 1); 4 am, hours. Finds party that fissioned at 1903 more than 1 km 1300 (November 21) 3 af, 9 im away. Observer encounters fire about 900 m from nest site of (party 2) party 1 at 2007 hours. Fires burn throughout the general area. November 21: 0625 hours, nest site of party 2 apparently burned the previous day or during the night, but party remained at this nest site. At 1305, fire is heard closer, to the SW, but is never seen. Wind speed 1.7 km/hour, temperature 34.97C. 13 November 27, 2010 Early 1805 3 am, 4 af, 6 im Fire heard SSW at 1805 hours. Fire burned up to the dry Ignore streambed where chimpanzees nested this night, burning up the short-grass grassland and woodland adjacent to the riverbed. Chimpanzees apparently ignored the fire, as they remained in their nests (observer spent night out with party). 14 March 8, 2012 Late 1800 3 am, 2 af, 4 im Party appears to have left Kerouani (dry) streambed because of Leave area large wildfire approaching near the end of the day. 15 March 9, 2012 Late 1300–1700 3 am See table 5. Pass through area to feed after short period of vigilance; ultimately leave area S343 16 March 31, 2013 Late 1615 3 am, 1 af, 3 im Fire !50 m SW of chimpanzees, who ignore it. One adult male Ignore fire sleeps in cave at ravine’s edge as fire burns above. Average wind speed 2.2 km/hour, temperature 41.57C. 17 December 31, 2013 Early 1655–1730 12 am, 4 af, 7 im Party passes within 100 m of fire on near side of Oundoundo (dry) Navigate in close proximity; streambed and then within 50 m of flames, with one individual ignore fire within 25 m. Two adult females stop to eat bamboo pith. Small fingers of fire within 10 m as party travels through streambed area. At 1619, fire is heard at bottom of ravine, roaring through dry vegetation for about 10 minutes. Party sits at top of slope, waits. 18 January 6, 2014 Early 1300–1430 6 am, 1 af, 1 im See table 4. Rest, groom, and feed while waiting for fire to diminish/move; continue travel and feed Note. Cases in which the observer detected fire via sound or smoke but no distance estimation could be given and where its presence was also ignored by chimpanzees are not included here. a am p adult male; af p adult female; im p immature. S344 Current Anthropology Volume 58, Supplement 16, August 2017

Table 4. Detailed description of an extended close encounter with wildfire by Fongoli chimpanzees, January 6, 2014

Time Individuals closest to firea Reaction(s) by individuals closest to fire 1300 JM, TM, MM, KL, DV, Multiple individuals climb rock face to emerge on plateau, within 15 m of fire. They move to sit BI, BO, CY, and others !20 m from fire. Fire is low, and wind blows away from us. Habitat is woodland, interspersed with bamboo. The vegetation here is a mix of short and tall grasses, trees, and bamboo, interspersed with rock substrate (weathered laterite surface) covered by dead bamboo leaves. Smoke is scattered, along with “fingers” of flame, downwind of chimpanzees. Flames are usually !60 cm in height and do not reach over 1.5 to 2.0 m in height where taller grass fuels flames. 1305 JM, MM JM grooms MM within 20 m of flames. They sit on ground, in the shade. 1310 JM, MM Males still groom in same location, but flames are 20–25 m away. 1315 JM, MM Males still groom as flames reach to within 15 m of them. 1320 JM, MM Males still groom, as flames reach to within 20 m of them. 1325 JM, MM Males rest on ground here with flames within 10–15 m of them. 1330 JM, MM Males rest on ground here with flames !10 m from them. 1335 JM, MM Males rest on ground here with flames 10 m from them. Flames are very small, but as taller flames erupt !10 m away, JM sits up and then moves down slope, and MM whimpers (this could be because of the fire, JM’s exit, or DV’s ultimate arrival). KL runs out to MM for grooming as DV moves in the opposite direction to JM. KL cries (probably because of DV’smovement). 1340–1345 MM, KL Males groom on ground here in shade !10 m from flames. 1350 MM, KL Males groom here !7 m from flames. 1355–1400 MM, KL Males groom here !6 m from flames. Flames are mainly small ones, !30 cm in height. 1405 MM, KL Males groom here, but fire is 8 m from them, and flames are dying. Fire has stopped moving toward the chimpanzees. 1410 MM, KL MM rests about 15 m from fire, but KL has moved another 4 m away to feed on the slope. Observer moves closer to chimpanzees as fire approaches to within 10 m from her. 1415–1430 TM and others Party moves E as estrous female TM ascends slope to plateau and skirts the edge of the burned plateau, crossing it in some places. It seems as if party was waiting for fire to pass, and they now forage in immediately burned area, where ash is still warm, feeding on bamboo pith. a Adult males p JM, MM, KL, BO, BI, DV (alpha male); adult female p TM; juvenile male p CY (son of TM).

Preliminary data on Fongoli chimpanzees’ use of burned in burned areas, which might be predicted if Fongoli chim- landscapes in their savanna-woodland environment can pro- panzees use these areas to navigate rather than opting to travel vide some insight into the hypothetical precondition rela- through areas of unburned tall grass. Ease of access during the tionships between pre-fire-using hominins and their envi- early dry season, especially, characterized burned zones. Ad- ronment. At Fongoli, chimpanzees traveled and fed in burned ditionally, travel occurred soon after a fire had passed, and areas as well as unburned areas. Travel occurred most often chimpanzees’ behavior of resting and socializing in the vi-

Table 5. Detailed description of extended close encounter with wildfire by three adult male Fongoli chimpanzees, March 9, 2014

Time Reactions by individualsa closest to fire 1300–1520 Wildfire is 500 m E, although smoke blocks out the sun periodically. Males feed here until 1500, when they rest for about 25 minutes. 1525–1545 BN leads travel toward wildfire, which is in the vicinity of Hamdalaye ravine. Males eat along the way. 1550–1555 Can hear the roar of the fire/flames at Hamdalaye ravine as we near it. 1600 Males skirt the edge of the fire, passing !100 m from it. Males travel through unburned tall grass, arriving at ravine edge at 1607 hours. 1607–1614 BO eats figs about 35 m from fire’s edge. Flames are visible. BN and JM sit at ravine edge and watch fire. BO “hoos” periodically. Males descend ravine within 5 minutes of arriving. 1615 Males have crossed ravine to the unburned side and !150 m from fire. Grass is still green at the bottom of the dry streambed where they crossed but dry and brittle on slopes. Males feed on figs. 1620 Males eat figs !125 m from fire. 1625 Males eat figs !150 m from fire. 1630 Males eat figs, but fire is 1150 m away. 1635 Males descend ravine again in area where neither side is yet burned, moving toward the opposite side. Fire is 200 m away. We pass small fires burning in ravine as we cross it. 1640–1650 Males begin travel away from ravine and are again in the presence of a large fire, 200–300 m W. 1655 Males stop to termite fish more than 300 m from fire. 1700–1715 Males continue heading N, away from fire and eventually fusing with other individuals at their main seasonal water source more than a kilometer away. a Adult males BO, BN, JM. Pruetz and Herzog Chimpanzees Navigate a Fire Landscape S345

Figure 1. Fongoli chimpanzees travel through a recently burned area. Note smoke in background, on right. A color version of this figure is available online.

cinity of wildfires indicated they can accurately predict the observations presented here of the Fongoli chimpanzee com- leading edges of fire and assess other aspects of fire behavior munity prove most informative in aiding hypotheses regard- based on the available fuel in an area, which can move rapidly ing how hominins may have taken advantage of exposure to and vary greatly within a small space according to micro- fire such that an innovation ultimately led to use. This ques- habitat variation (Trollope, de Ronde, and Geldenhuys 2004). tion corresponds with our first stage (table 1; Pruetz and LaDuke In order to do so, they must take into account the following 2010), possessing the cognitive sophistication necessary to al- variables: resultant flame height and distance traveled, fuel low close contact with fire.Althoughthefrequencywithwhich and wind speed, and topography and climate. Understanding chimpanzees at Fongoli, Senegal, must deal with fire is probably this complex set of interactions perhaps involves forming a much higher than that of any early hominin, the fact that these mental prediction of the fire’s behavior (Pruetz and LaDuke relatively small-brained apes seem easily to incorporate navi- 2010). We surmise that Fongoli chimpanzees are adept at pre- gation around fires and through recently burned areas supports dicting fire behavior because they often exhibit the appropri- the hypothesis that early members of our own lineage would ate responses to different fire types. They appear unconcerned have had similar abilities. Rather than fleeing from fire, they regarding their frequent exposure to smoldering fires or early would have understood it as a force shaping their local ecol- flaming fires, but they are more likely to avoid fully developed ogy, incorporating its effects into their daily activities, such as or postflashover fires (sensu Purser 1986). continuing to feed and range within areas in proximity to wild- If the many questions associated with hominins’ use of fire fires as long as resources remained. Following from the asser- are simplified into (1) how, (2) why, and (3) when, using liv- tion that living apes possess the cognitive sophistication nec- ing apes as referential models can inform each of these. The essary to allow close contact with fire resources, we argue from Figure 2. Adult female Fongoli chimpanzee moves through recently burned area. A color version of this figure is available online.

Figure 3. Chimpanzees’ reactions to fire according to dry-season period (late vs. early). Pruetz and Herzog Chimpanzees Navigate a Fire Landscape S347

Figure 4. Fongoli chimpanzee activity (%) in burned and unburned conditions. homology to support the hypothesis that fire use was an early unpublished data). Saba liana are found at various levels of rather than a later event. Much like the difficulty associated the canopy. with identifying and interpreting transitional or newly derived However, certain food resources may be positively altered traits in the fossil record, early hominin attraction to and their on a longer temporal scale. For example, fire can alter the phy- very first use of fire will be difficult to ever identify archaeo- togeography of certain plants, creating long-term positive al- logically or paleoanthropologically (see Gowlett 2016; Gowlett terations in distribution and abundance. During the month of and Wrangham 2013; Herzog 2015; Parker et al. 2016). How- October, chimpanzees at Fongoli feed extensively on Vigna vine ever, the capabilities of living apes indicate that even the earliest legumes, which are a type of THV. While burning destroys hominins would have been cognitively sophisticated enough these plants in the short term, it may increase encounters the to understand some aspects of fire behavior. next year by promoting greater plant propagation (Rebetzke Regarding the question of why the earliest hominins may and Lawn 2006). have been attracted to fire-modified landscapes, evidence sug- Other foods important to chimpanzee dry-season diets in- gests at least three possible motivations: (1) changes in the dis- clude honey (Jill D. Pruetz, unpublished manuscript), which tribution of and access to food, (2) improvements in travel, and is almost always harvested from a tree hollow and frequently (3) decreased threat of predation. Dietary improvements, such from the baobab tree, and the Macrotermes termite (Bogart as improved encounter rates with preferred prey (Herzog et al. and Pruetz 2011), which is protected from wildfire by its large 2016), have been shown for other primates. However, stable termite-mound refugia. While isotopic data suggest predom- isotope studies, which provide the most appropriate assess- inately C3 diets for hominins, they also demonstrate that in ment for comparing similarities between early hominin and response to increasing aridity and climatic variability, some chimpanzee diets, show an almost exclusive C3 plant diet for lineages probably targeted foods abundant in grassland set- both before 4 Mya (Sponheimer et al. 2006, 2013), suggesting tings, such as invertebrates, tubers, and the fruits and pods of that fire many not alter the distribution of many key chim- arid adapted plants (Sponheimer et al. 2005a, 2005b). Fongoli panzee foods in the ways it does for other primates. Typical chimpanzees live in a similarly arid environment and con-

C3 foods, such as tree fruits, are less likely to be destroyed by sume many of the same foods. For example, while still largely fi wild re than C4 grasses and other terrestrial herbaceous veg- frugivorous, Fongoli chimpanzees spend more time obtaining etation (THV), save in areas where tall elephant grass pro- invertebrates than any other chimpanzee population studied, vides fuel of sufficient height to burn low tree branches. Dur- with total foraging time dedicated to invertebrate feeding at ing much of the dry season at Fongoli, chimpanzees focus on times outweighing that of fruit (Bogart and Pruetz 2011). baobab (Adansonia digitata) fruit (Lindshield 2014), which Observations among this population could shed light on the is rarely affected by fires. The majority of important chim- ways in which burning prompts increased insectivory (Burton panzee dry-season foods (defined according to time spent 2009; Herzog et al. 2015), a shift nominated as fundamental in feeding on these species) are tree fruits. One important fruit shaping our lineage (Bogart and Pruetz 2011; Burton 2009; that is affected by fire is the liana, Saba senegalensis,which McGrew 2001, 2014). can be severely compromised by late dry-season fires, with Savanna chimpanzees spend almost all of their total travel fruit being destroyed even if liana survive (Jill D. Pruetz, time moving terrestrially (Jill D. Pruetz et al., unpublished man- S348 Current Anthropology Volume 58, Supplement 16, August 2017 uscript). However, for chimpanzees, terrestrial locomotion is terms of understanding later stages of fire use—see Holdaway, more costly than arboreal locomotion (Pontzer and Wrangham Davies, and Fanning 2017 and Mallol and Henry 2017—or 2004). Therefore, among this population fire may have a two- using nonhuman primates to understand earlier stages or the fold effect on travel. First, the removal of an understory that is precondition for fire use). Several authors (Burton 2009; thick and difficult to move through may facilitate decreased Gowlett 2016; Gowlett and Wrangham 2009; Herzog 2015; travel time between patches. This phenomenon was observed Parker 2014; Parker et al. 2016; Pruetz and LaDuke 2010) among savanna-dwelling vervets, which not only expanded justify using primate models, especially apes, to understand into burned areas traveling almost exclusively terrestrially but the behavior of early hominins in respect to fire. They posit also moved faster in these areas than in unburned alternatives that ancestral fire use probably occurred earlier than definite (Herzog et al. 2014). Second, these newly bare landscapes may archaeological evidence suggests. While anecdotal informa- simply impose fewer energetic locomotor costs than unburned tion shows that nonhuman primates do not exhibit fear of fire alternatives. Chimpanzees in this study spent a higher propor- (summarized in Burton 2009), data are based on experimental tion of time traveling in burned than either unburned or par- contexts or human-produced campfires (table 2). Apes living tially burned areas. in fire-prone areas, therefore, can present a more natural con- Finally, burned areas may be attractive because of enhanced text for potential scenarios in which early hominins encoun- predator detection in fire-cleared grasslands (Herzog 2015). In tered natural fires and provide insight into the dynamics be- burned areas, vegetation that may have otherwise served as tween a relatively small-brained hominin (i.e., preceding the cover for ambush predators is gone. Without sufficient cover, genus Homo) and fire landscapes. If chimpanzees can be used predators may abandon efforts at hunting in these areas, and as a referential model for small-brained hominins living in research has suggested that predators avoid burned habitats similar wooded savanna environments, we are one step fur- (Ogen-Odoi and Dilworth 1984; Eby et al. 2013; Green et al. ther in better understanding the role that fire use had in our 2014). At Fongoli, where approximately 85%–90% of the un- own genus. derstory is grass, late rainy season and early dry-season travel is difficult. Chimpanzees have been observed taking advantage of recently burned areas as travel corridors, and it is possible Acknowledgments that these areas also provide them greater safety in terms of Research at Fongoli has been supported by Iowa State Uni- predator detection and reduced ambush sites for predators. versity, the National Geographic Society, the Leakey Founda- The final question, when, will be the most difficult to an- tion, Primate Conservation Inc., the American Society of Pri- swer. Recent archaeological and paleoanthropological recon- matologists, the Great Ape Trust of Iowa, the National Science structions are notably at odds regarding the emergence of this Foundation, and the Wenner-Gren Foundation for Anthropol- important milestone in human evolutionary history (see Gow- ogical Research. Permission to conduct research at Fongoli is lett 2016). Regarding the early control versus late control the- granted by the Department of Water, Soils and Forestry in Sen- ories of fire use (see Herzog 2015 for review), emerging data egal and by the Republic of Senegal. Numerous project manag- on nonhuman primates can at least provide an indication of ers and research assistants have contributed to data collection what is and is not unique regarding our lineage’s relationship at Fongoli, including Dondo Kante, Michel Sadiakho, Mboule with fire. As Herzog et al. (2014) point out in one of the only Camara, and Jacques Keita. other studies specifically designed to assess primates’ reactions to fires rather than post hoc or opportunistic inquiry, the fact that cercopithecines such as vervets (Chlorocebus aethiops) References Cited showed calmness in the face of wildfire indicates that such a prerequisite for eventual use could be a primitive trait for cat- Armelagos, George. 2010. Creatures of the flame: light and heat in human evolution. Evolutionary Anthropology 19:158–160. arrhines at least. As more data emerge on the topic, a concep- Berenstain, Leo. 1986. Responses of long-tailed macaques to drought and fire tual model (Herzog 2015) could add to the existing primate in eastern Borneo: a preliminary report. Biotropica 18:257–262. data, which has thus far been referential in nature. Attention Blumenschine, Robert J. 1986. Early hominid scavenging opportunities: im- plications of carcass availability in the Serengeti and Ngorongoro ecosystems. to the particular behavioral ecology of candidate nonhuman Oxford: BAR. primates to include would consider features of the diet and Bobe, Rene, and Anna K. Behrensmeyer. 2004. The expansion of grassland factors related to predator pressure, for example, that would ecosystems in Africa in relation to mammalian evolution and the origin of the genus Homo. Palaeogeography, Palaeoclimatology, Palaeoecology 207: be most informative when attempting to construct hypotheses 399–420. regarding the emergence of fire use in the hominin lineage. Bogart, Stephanie, and Jill Pruetz. 2011. Insectivory of savanna chimpanzees The use of fire is a uniquely derived trait that has been at Fongoli, Senegal. American Journal of Physical Anthropology 145:11–20. Brewer, Stella. 1978. The chimps of Mt. Asserik. New York: Knopf. used to distinguish humans from all other animals alive today Brink, A. S. 1957. Spontaneous fire-controlling reactions of two chimpanzee (Goudsblom 1986). However, considering scenarios that led smoking addicts. South African Journal of Science 53:241–247. to the use of fire by our lineage helps to shed light on the how Burton, Frances. 2009. Fire: the spark that ignited human evolution. Albu- fi querque: University of New Mexico Press. and why of such events, which are notoriously dif cult to iden- Carmody, Rachel N., and Richard Wrangham. 2009. The energetic signifi- tify without the use of behavioral models (either ethnographic, in cance of cooking. Journal of Human Evolution 57:379–391. Pruetz and Herzog Chimpanzees Navigate a Fire Landscape S349

Charmove, Arnold S. 1996. Enrichment: unpredictable ropes and fire. Shape gorillas suggest earlier divergence times in great ape and human evolution. of Enrichment 5:1–3. Proceedings of the National Academy of Sciences of the USA 109:15716– deMenocal, Peter B. 1995. Plio-Pleistocene African climate. Science 270 15721. (5233):53–59, doi:10.1126/science.270.5233.53. Lindshield, Stacy. 2014. Multilevel analysis of the foraging decisions of west- ———. 2004. African climate change and faunal evolution during the Pliocene- ern chimpanzees (Pan troglodytes verus) and resource scarcity in a savanna Pleistocene. Earth and Planetary Science Letters 220:3–24. environment at Fongoli, Senegal. PhD dissertation, Iowa State University, ———. 2011. Climate and human evolution. Science 331:540–542. Ames. Eby, S. A. Mosser, A. Swanson, C. Packer, and M. Ritchie. 2013. The impact of Mallol, Carolina, and Auréade Henry. 2017. Ethnoarchaeology of Paleolithic burning on lion Panthera leo habitat choice in an African savanna. Current fire: methodological considerations. Current Anthropology 58(suppl. 16): Zoology 59(3):335–339. S217–S229. Esque, T. C., C. R. Schwaibe, L. A. Defalco, R. B. Duncan, and T. J. Hughes. McGrew, William C. 2001. The other faunivory: primate insectivory and early 2003. Effects of desert wildfires on desert tortoise (Gopherus agassizii) and human diet. In Meat-eating and human evolution. Craig B. Stanford and other small vertebrates. Southwestern Naturalist 48(1):103–111. Henry T. Bunn, eds. Pp. 160–178. Oxford: Oxford University Press. Fonseca-Azevedo, Karina, and Suzana Herculano-Houzel. 2012. Metabolic ———. 2014. The “other faunivory” revisited: insectivory in human and non- constraint imposes tradeoff between body size and number of brain neu- human primates and the evolution of human diet. Journal of Human Evo- rons in human evolution. Proceedings of the National Academy of Sciences lution 71:4–11. of the USA 109:18571–18576. Mikami, O. K., Y. Katsuno, D. M. Yamashita, R. Noske, and K. Eguchi. 2010. Frazer, J.-G. 1930. Myths of the origin of fire. London: Macmillan. Bowers of the Great Bowerbird (Chlamydera nuchalis) remained unburned Galat-Luong, Anh, and Gerard Galat. 2005. Conservation and survival adap- after fire: is this an adaptation to fire? Journal of Ethnology 28:15–20. tations of Temminck’s red colobus (Procolobus badius temmincki) in Sen- Mitchell, Robert W. 1999. Scientific and popular conceptions of the psy- egal. International Journal of Primatology 26(3):585–603. chology of great apes from the 1700s to the 1970s: déjà vu all over again. Geluso, K. N., G. D. Schroder, and T. B. Bragg. 1986. Fire-avoidance behavior Primate Report 3–118. of meadow voles (Microtus pennsylvanicus). American Midland Naturalist Moore, Jim. 1996. Savanna chimpanzees, referential models and the last com- 116:202–205. mon ancestor. In Great ape societies. William C. McGrew, Linda F. Mar- Goodman, Jeanne Moore. 1960. Aves incendiaria. Wilson Bulletin 72:400–401. chant, and Toshisada Nishida, eds. Pp. 275–292. Cambridge: Cambridge Uni- Goudsblom, Johan. 1986. The human monopoly on the use of fire: its origins versity Press. and conditions. Human Evolution 1:517–523. Nellemann, Christian, Lera Miles, Bjørn P. Kaltenborn, Melanie Virtue, and Gowlett, John A. J. 2016. The discovery of fire by humans: a long and con- Hugo Ahlenius. 2007. The last stand of the orangutan: state of emergency; voluted process. Philosophical Transactions of the Royal Society B 371(1696), illegal logging, fire and palm oil in Indonesia’s national parks. Arendal, doi:10.1098/rstb.2015.0164. Norway: GRID-Arendal and UN Environment. Gowlett, John A. J., and Richard Wrangham. 2013. Earliest fire in Africa: Nishida, Toshisada, Takayoshi Kano, Jane Goodall, William McGrew, and towards the convergence of archaeological evidence and the cooking hy- Michio Nakamura. 1999. Ethogram and ethnography of Mahale chimpan- pothesis. Azania: Archaeological Research in Africa 48:5–50. zees. Anthropological Science 107(2):141–188. Grafe, T. Ulmar, Stephanie Dobler, and K. Eduard Linsenmair. 2002. Frogs Ogen-Odoi, Asaph A., and T. G. Dilworth. 1984. Effects of grassland burning flee from the sound of fire. Proceedings of the Royal Society B 269(1495): on the savanna hare-predator relationships in Uganda. African Journal of 999–1003. Ecology 22(2):101–106. Green, David S., Gary J. Roloff, Brian R. Heath, and Kay E. Holekamp. 2014. Parker, Christopher H. 2014. On the evolution of human fire use. PhD dis- Temporal dynamics of the responses by African mammals to prescribed sertation, University of Utah, Salt Lake City. fire. Journal of Wildlife Management 79:235–242. Parker, Christopher H., Earl R. Keefe, Nicole M. Herzog, James F. O’Connell, Harrison, Michael. 1983. Patterns of range use by the green monkey, Cer- and Kristen Hawkes. 2016. The pyrophilic primate hypothesis. Evolution- copithecus sabaeus, at Mt. Assirik, Senegal. Folia Primatologica 41:157–179. ary Anthropology 25(2):54–63. ———. 1984. Optimal foraging strategies in the diet of the green monkey, Pontzer, Herman, and Richard W. Wrangham. 2004. Climbing and the daily Cercopithecus sabaeus, at Mt. Assirik, Senegal. International Journal of Pri- energy cost of locomotion in wild chimpanzees: implications for hominoid matology 5:435–471. locomotor evolution. Journal of Human Evolution 46(3):315–333. Herzog, Nicole M. 2015. Primate behavioral responses to burning as a model Potts, Rick. 1998. Environmental hypotheses of hominin evolution. American for hominin fire use. PhD dissertation, University of Utah, Salt Lake City. Journal of Physical Anthropology 107:93–136. Herzog, Nicole M., Earl R. Keefe, Christopher H. Parker, and Kristen Hawkes. Pringle, Robert M., Duncan M. Kimuyu, Ryan L. Sensenig, Todd M. Palmer, 2016. What’s burning got to do with it? primate foraging opportunities in fire- Corinna Riginos, Kari E. Veblen, and Truman P. Young. 2015. Synergistic modified landscapes. American Journal of Physical Anthropology 159:432–441. effects of fire and elephants on arboreal animals in an African savannah. Herzog, Nicole M., Christopher H. Parker, Earl R. Keefe, James Coxworth, Journal of Animal Ecology, doi:10.1111/1365-2656.12404. Alan Barrett, and Kristen Hawkes. 2014. Fire and home range expansion: a Pruetz, Jill. 2007. Evidence of cave use by savanna chimpanzees (Pan trog- behavioral response to burning among savanna dwelling vervet monkeys lodytes verus) at Fongoli, Senegal. Primates 48:316–319. (Chlorocebus aethiops). American Journal of Physical Anthropology 154(4): Pruetz, Jill, and Paco Bertolani. 2009. Chimpanzee (Pan troglodytes verus) 554–560. behavioral responses to stresses associated with living in a savanna-mosaic Hillyer, Allison P., Roy Armstrong, and Amanda H. Korstjens. 2015. Dry environment: implications for hominin adaptations to open habitats. Pa- season drinking from terrestrial man-made watering holes in arboreal wild leoanthropology 2009:252–262. Temminck’s red colobus, The Gambia. Primate Biology 2(1):21–24. Pruetz, Jill D., Paco Bertolani, K. B. Onttl, S. Lindshield, M. Shelley, and E. G. Holdaway, Simon J., Benjamin Davies, and Patricia C. Fanning. 2017. Ab- Wessling. 2015. New evidence on the tool-assisted hunting exhibited by original use of fire in a landscape context: investigating presence and ab- chimpanzees (Pan troglodytes verus in a savannah habitat at Fongoli, Sén- sence of heat-retainer hearths in western New South Wales, Australia. Cur- égal. Royal Society Open Science 2(4), 140507. rent Anthropology 58(suppl. 16):S230–S242. Pruetz, Jill. D., and D. Kante. 2010. Successful return of a wild infant chim- Jaffe, Karen, and Lynne Isbell. 2009. After the fire: benefits of reduced ground panzee (Pan troglodytes verus) to its natal group after capture by poahcers. cover for vervet monkeys (Cercopithecus aethiops). American Journal of African Primates 7(1):35. Primatology 71:252–260. Pruetz, Jill, and Thomas LaDuke. 2010. Brief communication: reaction to fire Koprowski, J. L., K. M. Leonard, C. A. Zugmeyer, and J. L. Jolley. 2006. Direct by savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: con- effects of fi re on endangered Mount Graham red squirrels. Southwestern ceptualization of “fire behavior” and the case for a chimpanzee model. Naturalist 51:59–63. American Journal of Physical Anthropology 141:646–650. Kumar, S., A. Filipski, V. Swarna, A. Walker, and S. B. Hedges. 2005. Placing Purser, D. A. 1986. The effects of fire products on escape capability in pri- confidence limits on the molecular age of the human-chimpanzee diver- mates and human fire victims. International Association of Fire Safety Sci- gence. Proceedings of the National Academy of Sciences of the USA 102: ence 1:1101–1110. 18842–18847. Rebetzke, G. J., and R. J. Lawn. 2006. Root and shoot attributes of indigenous Langergraber, Kevin, K. Prüfer, C. Rowney, Christophe Boesch, C. Crockford, perennial accessions of the wild mungbean (Vigna radiata ssp. sublobata). K. Fawcett, E. Inoue, et al. 2012. Generation times in wild chimpanzees and Crop and Pasture Science 57(7):791–799. S350 Current Anthropology Volume 58, Supplement 16, August 2017

Reed, Kay. 1997. Early hominid evolution and ecological change through the Trollope, W. S. W., C. de Ronde, and C. J. Geldenhuys. 2004. Fire behaviour. African Plio-Pleistocene. Journal of Human Evolution 32:289–322. In Wildland fire management handbook for Sub-Sahara Africa. Johann G. Russon, Ann. 2000. Orangutans: wizards of the rain forest. Toronto: Firefly. Goldammer, and Cornelis de Ronde, eds. Pp. 27–59. Mainz: Global Fire Russon, Ann, and Birute Galdikas. 1993. Imitation in free-ranging rehabilitant Monitoring Center. orangutans. Journal of Comparative Psychology 44:23–33. Vaesen, Krist. 2014. Chimpocentrism and reconstructions of human evolu- Sandgathe, Dennis M. 2017. Identifying and describing pattern and process tion (a timely reminder). Studies in History and Philosophy of Science C 45: in the evolution of hominin use of fire. Current Anthropology 58(suppl. 16): 12–21. S360–S370. Veldman, Joseph W., Elise Buisson, Giselda Durigan, G. Wilson Fernandes, Segerdahl, Par, William Fields, and Sue Savage-Rumbaugh. 2005. Kanzi’sprimal Soizig Le Stradic, Gregory Mahy, Daniel Negreiros, et al. 2015. Toward an language: the cultural initiation of primates into language.NewYork:Palgrave old-growth concept for grasslands, savannas, and woodlands. Frontiers in Macmillan. Ecology and the Environment 13(3):154–162. Sponheimer M., Z. Alemseged, T. E. Cerling, F. E. Grine, W. H. Kimbel, M. G. Vernes, K. 2000. Immediate effects of fire on survivorship of the northern Leakey, J. A. Lee-Thorp, F. K. Manthi, K. E. Reed, B. A. Wood, and J. G. bettong (Bettongia tropica): an endangered Australian marsupial. Biological Wynn. 2013. Isotopic evidence of early hominin diets. Proceedings of the Conservation 93(3):305–309. National Academy of Sciences of the USA 110(26):10513–10518. Vrba, Elisabeth S. 1995. Paleoclimate and evolution, with emphasis on human Sponheimer, Matthew, Darryl de Ruiter, Julia Lee-Thorp, and Andreas Späth. origins. New Haven, CT: Yale University Press. 2005a. Sr/Ca and early hominin diets revisited: new data from modern and WoldeGabriel, Giday, Stanley H. Ambrose, Doris Barboni, Raymonde Bonne- fossil tooth enamel. Journal of Human Evolution 48:147–156, doi:10.1016/j fille, Laurent Bremond, Brian Currie, David DeGusta, et al. 2009. The geo- .jhevol.2004.09.003. logical, isotopic, botanical, invertebrate, and lower vertebrate surroundings Sponheimer, Matthew, Julia Lee-Thorp, Darryl de Ruiter, Daryl Codron, Jacqui of Ardipithecus ramidus. Science 326(5949):65e1–65e5. Codron, Alexander T. Baugh, and Francis Thackeray. 2005b. Hominins, Woolley, Leigh-Ann, Joshua J. Millspaugh, Rami J. Woods, Samantha J. Van sedges, and termites: new carbon isotope data from the Sterkfontein Valley Rensburg, Robin L. Mackey, Bruce Page, and Rob Slotow. 2008. Population and Kruger National Park. Journal of Human Evolution 48(3):301–312. and individual elephant response to a catastrophic fire in Pilanesberg Na- Sponheimer, Matthew, James E. Loudon, Michaela E. Howells, Jill D. Pruetz, tional Park. PLoS ONE 28:1–10. Daryl Codron, Jacqui Codron, Darryl de Ruiter, and Julia Lee-Thorp. 2006. Wrangham, Richard. 2009. Catching fire: how cooking made us human. New Do savanna chimpanzees consume C4 resources? Journal of Human Evo- York: Basic. lution 51(2):128–133. Wynn, Jonathan G. 2004. Influence of Plio-Pleistocene aridification on human Sunquist, Melvin E. 1967. Effects of fire on raccoon behavior. Journal of evolution: evidence from paleosols of the Turkana Basin, Kenya. American Mammalogy 48(4):673–674. Journal of Physical Anthropology 123(2):106–118. Tappan, Gray G., Moussa Sall, Eric C. Wood, and Matthew Cushing. 2004. Eco- Zink, Katherine D., Daniel E. Lieberman, and Peter W. Lucas. 2014. Food regions and land cover trends in Senegal. Journal of Arid Environments 59 material properties and early hominin processing techniques. Journal of Hu- (3):427–462. man Evolution 77:155. –166 Current Anthropology Volume 58, Supplement 16, August 2017 S351

Toward a Long Prehistory of Fire

by Michael Chazan

This article explores a conception of the origins of fire as a process of shifting human interactions with fire, a process that, in a sense, still continues today. This is a counterpoint to the dominant narrative that envisions a point of “dis- covery” or “invention” for fire. Following a discussion about what fire is and how it articulates with human society, I propose a potential scenario for the prehistory of fire, consisting of three major stages of development. From this per- spective, obligate cooking developed gradually in the course of human evolution, with full obligate cooking emerging subsequent to modern humans rather than synchronous with the appearance of Homo erectus as envisioned by the cooking hypothesis.

The study of human evolution is an undertaking that, like all fire. I will begin by discussing what fire is and how it articulates great scientific challenges, requires a combination of empirical with human society and then will present a potential scenario rigor, luck, creativity, and imagination. In On the Origin of for the prehistory of fire. Species Darwin wrote eloquently of the incompleteness of the From the perspective of evolutionary biology scenarios “are fossil record, and this is certainly where a healthy dose of luck, [phylogenetic trees] fleshed out with narratives of adaptation, and creativity, comes in handy. In the study of human evolu- competition, ecology etc.” (C. Patterson 1981:209). Tattersall tion there is a particular problem posed by the structuring of and Eldredge have described a scenario “as an extremely in- time periods that span hundreds of thousands of years. While tricate type of proposition” at a remove from testable hard data this excessively longue durée characterizes all of paleontology, (1977:205). Most archaeological research proceeds based on and in fact from this perspective human evolution is very brief, the construction of scenarios that are at least in part inductive there is good reason to think that there are challenges that are and at least potentially testable based on the archaeological unique to the study of human ancestors. The Paleolithic lies at record (Hodder 1999). The cooking hypothesis developed by the hinge between paleontology and history. In studying our Richard Wrangham also operates at the level of a scenario but own ancestors we tend to rely on narrative structures that, al- it is important to recognize that the basis for the construc- though never completely absent, are somewhat more weakly tion of the cooking hypothesis is fundamentally different from expressed in other aspects of evolutionary research. Landau an archaeological approach. Wrangham and his collaborators (1993) has demonstrated the relationship between scenarios work from the observation “that present-day humans cannot of human evolution and tropes drawn from folktales, while extract sufficient energy from uncooked wild diets” (Carmody Stoczkowski (1994, 2002) has elucidated the common struc- et al. 2016:1091). From this observation, the logical inference tural characteristics of these narratives. The demands of con- is that obligate cooking must have a point of origin in homi- temporary media (web-based, television, and print) also play a nin phylogeny. This point of origin is then mapped onto the major role in tunneling discussion toward a narrative with a increase in hominin brain and body size ca. 2 million years single point of origin. ago, leading to the proposition that obligate cooking began This article explores a conception of the origins of fire as a with Homo erectus and was a characteristic of subsequent taxa process of shifting human interactions with fire, a process that, on the hominin lineage. The power of the approach taken by in a sense, still continues today. This is a counterpoint to the the cooking hypothesis is that it is at least partially testable dominant narrative that envisions a point of “discovery” or “in- based on experimental studies on the physiological and molec- vention” for fire. One way of phrasing this perspective is that ular correlates of consumption of cooked food (see literature there is not an “origin” of fire but, rather, a long prehistory of cited in Carmody et al. 2016; Wrangham 2017). However, this approach also has a number of shortcomings. First, while ob- ligate cooking necessarily must have a point of phylogenetic Michael Chazan is Professor in the Department of Anthropology at origin, the same is not true for cooking that might become in- the University of Toronto (19 Russell Street, Toronto, Ontario M5S 2S2, Canada [[email protected]]) and at the Evolutionary tegrated into hominin adaptation through a process rather Studies Institute at the University of the Witwatersrand (1 Jan Smuts than as the result of a single event. Second, while many aspects Avenue, Braamfontein 2000, Johannesburg, South Africa). This pa- of human obligate cooking can be experimentally tested, there per was submitted 25 VII 16, accepted 5 III 17, and electronically is no current method for testing whether H. erectus required published 17 V 17. regular cooked food. In fact, the placement of the onset of q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0017$10.00. DOI: 10.1086/691988 S352 Current Anthropology Volume 58, Supplement 16, August 2017 obligate cooking at 2 million years ago is not directly testable a seasonal natural resource. As Wrangham (2009) emphasizes, without recourse to the archaeological record (including re- for modern humans, fire has become a necessity for life. In the covery of residues from fossils; see Hardy et al. 2017). context of human society, fire also becomes a technology, is deeply socially embedded, and can take on a spiritual dimen- Defining Fire sion. As is often the case with significant aspects of human culture, these various dimensions of fire—adaptation, technol- Fire is “a self-sustaining, high-temperature oxidation reaction ogy, society, and ideology—are often intertwined. To illustrate which releases heat and light; and which usually needs a small the functioning of fire in human society, a case study from the input of heat [ignition] to get it going” (Rossotti 1993:5). Fire archaeology of complex societies is useful. The choice of an involves the combustion of fuel, which in the archaeological example from complex societies rather than the ethnographic record is limited to organic compounds including wood, grass, record for hunter-gatherers is deliberate. The point is to em- fat, and bone. Combustion products such as ash and charcoal phasize that the relationship between humans and fire is an leave the material traces that, along with the burned substrate, ongoing process that does not apply only to hunter-gatherers. can be observed in well-preserved archaeological contexts. Ex- The Egyptians of the Old Kingdom, the period of pyramid posure to fire also causes fracture in siliceous rock such as construction, developed a distinctive technology for baking chert, including crazing and pot lids (L. W. Patterson 1995; bread (Chazan and Lehner 1990). In this technological system Purdy 1975). Burning can also alter both the magnetic prop- a thick clay pot served as a small oven. What makes Old King- erties and the thermoluminescent signal of siliceous rock (Bor- dom baking particularly interesting is that scenes on the walls radaile, Stewart, and Ross 1998; Richter, Alperson-Afil, and of Old Kingdom tombs, particularly the Tomb of Ti at Saq- Goren-Inbar 2011). Burned bone can be identified on the basis qara (fig. 1), provide a detailed account of how this technology of surface features such as color and of application of FTIR worked, and excavations of bakeries near the Giza pyramids (Fourier transform infrared spectroscopy) (see Berna et al. 2007, add an archaeological perspective (Lehner 1996; Mahmoud 2012; Stiner et al. 1995). and Eissa 2015). It is interesting that the fire itself is not depicted There is a firm empirical basis for considering natural fire in the scene from the Tomb of Ti, yet its effects are clearly to have been part of the ecology of early hominins. Geological evident in the actions of the figures crouched next to the oven, evidence of fire dates back to the Devonian (Scott 2000). Byers who protect their faces with their hands, and in the hieroglyph et al. (2014) provide evidence of a fire scar on a Triassic tree, that indicates that it is hot. Also absent from the scene is a de- suggesting that fire played an important role in plant ecology piction of how the fire is made; however, it is safe to assume the by the Triassic. There are also geological examples of fire as- use of some form of bow drill—a method used in Old Kingdom sociated with fossil fauna. For example, Brown, Collinson, and ground stone technology. Scott (2013) examine the association of charcoal with Creta- The scene found in the Tomb of Ti provides an unusually ceous hadrosaurs, ankylosaurs, and other fauna at Dinosaur complete perspective on the dynamics of an ancient techno- Provincial Park in Alberta, Canada. logical process. In Old Kingdom Egypt the production of bedja There is a wide body of literature on wildfires that aims bowls, which are thick walled with an organic-rich fabric to to understand the factors underlying wildfires and changes in resist thermal shock, and their use as ovens is a very particular wildfire due to global warming, anthropogenic activity, and for- technique. In this context fire is integrated into a chaîne op- est management. Average fire conditions, known as fire re- ératoire of baking. gimes, include fire density (number of fires within a given area), The technology of Old Kingdom baking is deeply integrated fire return interval, and fire intensity and severity (duration with the social structure of a complex society engaged in the and radiative power) (Chuvieco, Giglio, and Justice 2008). The construction of monumental architecture. Baking in bread seasonality of fire is linked to both covarying and stochastic molds allows for the centralized processing of grains into ed- weather patterns including temperature, wind speed, relative ible food, which removes food preparation from the household humidity, and radiation—as well as frequency of lightning and thus frees labor for monumental construction. The baking strikes (Johnson and Balice 2006). Analysis of fire records from of bread in molds also has the advantage of producing food as a giant sequoia groves in California links the frequency of fire to commodity. By controlling the size of the interior of the bread long-term climate change (Swetnam 1993). Despite the lim- mold, it is possible to create loaves that adhere to a tight unit itations posed by the human role in modern fire regimes, it is of measure and that can be transported and distributed. In the clear that from the perspective of early hominin ecology, fire Giza bakeries there are two sizes of bread mold, and the in- can be considered to be a seasonal resource (for further data, see terior volume of the vessels was highly standardized through Archibald et al. 2010; Johnson and Balice 2006; Swetnam 1993; control of the rim diameter and depth of the vessel. It is not an van Wagtendonk and Cayan 2008; van Wilgen and Scholes exaggeration to say that the construction of the pyramids at 1997). Giza was dependent on the technological system that allowed a fi Fire in Human Society re burning in a small, open oven to transform grain into stan- dardized loaves of bread. It is also true that the way the flame in In the context of human society, fire takes on multiple aspects the bakery oven was integrated into a technological system is that go beyond the physics of combustion or the dynamics of the result of the social structure of Old Kingdom Egypt. Chazan Long Prehistory of Fire S353

Figure 1. Process of bread baking as depicted in the Tomb of Ti, Saqqara. a, Grinding and sifting grain; b, mixing and pouring dough in large vats; c, heating bedja bowls (note how the crouched figure shields their face from the heat); d, pouring dough into a row of bedja bowls (note that bowls are then covered with inverted bowls); e, checking bread with a stick; f, removing bread from bowls after baking (after Epron 1939).

The engine of the developments on the Giza Plateau was far fire during the Paleolithic. It is important to emphasize that this from basic subsistence. The entire context, including the as- is a generalized scenario and that it is meant to highlight the sociated bakeries, makes little sense without considering the major tendencies in the development of fire. The intention is Egyptian beliefs about the divine nature of the king and the not to negate the likelihood that there were localized devel- king’s body. There is, however, no indication that the fire in opments that depart from these major tendencies. This sce- Old Kingdom bakeries was regarded as sacred. The Egyptians nario builds on the idea that the adaptive, technological, social, did have ideas about fire as an “effective weapon against . . . and ideational aspects of the use of fire are closely intertwined. nightmares and other nocturnal agents of chaos” (Szpakowska It is built on the currently available archaeological data and 2003:121); however, there is no mention of such powers in the thus is subject to testing and refutation. As noted above, this bakery scene in the Tomb of Ti. type of inductive approach is limited by the incompleteness of In sum, while fire is a chemical reaction and a natural re- the archaeological record, and there is always the risk that the source, it is also in the human context a technology, an aspect absence of evidence will be incorrectly interpreted as the ab- of society, a necessity, and an element of belief. Consideration sence of a behavior. However, the focus here is looking at pat- of Old Kingdom baking allows us to recognize that the way fire terning in the archaeological record over time rather than at is used to meet nutritional needs is highly contingent on both particular observations in isolation, so that the absence of ev- technology and social systems. While it might indeed be nec- idence is significant in relation to the subsequent presence of essary that humans eat cooked food, the way this cooking evidence. takes place is highly contingent. In the context of Old Kingdom Egypt, the energetic requirements were not simply the require- Stage 1: Opportunistic interaction with fire dates back 1 mil- ments for survival and reproduction but also were demands lion years but likely appears much earlier, perhaps even to the to release a tremendous amount of labor from subsistence ac- origins of Homo erectus. Fire at this stage was a seasonal re- tivities. source, and maintenance of fire was limited.

A Scenario for the Initial Development of the Field observations suggest that chimpanzees are able to fi Human Use of Fire: A Long Prehistory control their fear response to re and thus show a degree of conceptual understanding of fire (Pruetz and LaDuke 2010). The available archaeological data are consistent with a scenario Evidence for interaction between hominins and fire before of three stages in the development of human interaction with ca. 400,000 years ago is rare (for reviews of evidence, see Bar- S354 Current Anthropology Volume 58, Supplement 16, August 2017 betti 1986; Bellomo 1994; James 1989; Perlès 1977; Roebroeks activities were spatially distinct, then the likelihood of finding and Villa 2011; Rolland 2004). traces of fire, except in extraordinary contexts such as Won- The site of Gesher Benot Ya’aqov (ca. 700,000 years BP) derwerk Cave, becomes extremely low. Such a decoupling of has produced evidence of low-intensity traces of fire in small lithic activity areas and the use of fire might offer a potential patches within a dense scatter of lithics and fauna (Alperson- explanation for the absence of traces of fire on most early Eu- Afil, Richter, and Goren-Inbar 2007; Alperson-Afil et al. 2009; ropean archaeological localities. Goren-Inbar et al. 2004). If the pattern detected at Gesher Be- The focus on a long prehistory of fire proposed here shifts not Ya’aqov can be replicated, that would suggest that early in- the focus from simply identifying the earliest use of fire to- teractions with fire were short lived—as any long-term main- ward understanding the frequency and distribution of the use tenance would have produced a more pronounced signature, of fire. Sites lacking evidence for fire are thus as important as although the pattern of recurrent fireplaces in the same area re- those with such traces preserved. From this perspective, the quires consideration. There is no evidence from Gesher Benot contrast between the low-density Acheulean occupation of Ya’aqov of the technologies associated with making fire, so Wonderwerk Cave (located on the eastern slopes of the Ku- this scenario is unlikely (see discussion below). If Gesher Benot ruman Hills) and the remarkably high density of occupation Ya’aqov provides early evidence of collection of fire, this would in the sites of the Kathu Complex on the western fringes of suggest that in its early stages this behavior is not associated with the Kuruman Hills is extremely significant (for the sites of the practices for the long-term maintenance of fire. Kathu Complex, see Chazan et al. 2012; Matmon et al. 2015; At Wonderwerk Cave, South Africa, detailed micromorpho- Porat et al. 2009; Walker, Lukich, and Chazan 2014; Wilkins logical analysis combined with in situ micro-FTIR provides un- and Chazan 2012). The Acheulean and Fauresmith sites of the ambiguous evidence of burning in Stratum 10, including wood Kathu Complex comprise tens of millions of stone tools in ash and bone, dated to ca. 1 million years ago (Berna et al. 2012). stratigraphic contexts that rarely preserve bone (with the im- The location of the evidence of burning ca. 30 meters from the portant exception of Kathu Pan 1) and do not preserve micro- cave entrance makes it most likely that human activity is the ex- stratigraphic evidence relevant to fire. However, in the massive planation for fire in this context. It is significant that at Won- collections of artifacts from these sites, burned artifacts are not derwerk the traces of fire are associated with a low density of evident based on macroscopic features such as pot lids and stone tools and faunal remains and that there are no clear com- crazing, which makes it difficult to argue for the frequent use bustion features (although this issue is the focus of renewed of fire at these localities. Wonderwerk Stratum 10 is signifi- fieldwork at the site). The evidence from Wonderwerk makes cant not simply for the evidence of fire in and of itself but also it very difficult to sustain the argument that early hominins did within the broader context of fire use during the Acheulean. not collect or maintain fire. However, the scarcity of evidence The presence of fire in a low-density cave site contempora- of fire in the early archaeological record and the nature of the neous with the absence of fire on sites with massive accumu- archaeological context at Wonderwerk Cave are consistent with lations of lithics suggests that during this time period, aspects the possibility that early hominins did not maintain fire for long of hominin behavior related to the use of fire and those re- periods and that the way fire was used did not result in high ar- lated to the production and discard of large quantities of arti- chaeological visibility. Other claims for early use of fire include facts might have been somewhat spatially differentiated. Thus, Cueva Negra, Spain (Walker et al. 2013), but beyond the pres- a limitation of the archaeological record is that we are usually ence of burning in association with lithics and fauna, the ex- only able to identify the use of fire when evidence of burning is posure at this site is too limited to provide evidence of how fire spatially associated with archaeological sites defined by the was used. presence of detectable frequency of stone tools. The possibility At present there is no archaeological basis for associating of spatial decoupling of the use of fire and large-scale tool dis- the dispersal of H. erectus with aspects of the maintenance or card must be kept in mind when assessing the absence of evi- creation of fire by hominins (Hardy et al. 2017; Roebroeks and dence for the use of fire on most early sites. However, it never- Villa 2011). We are left with the difficulty of reconciling the theless remains difficult to reconcile the spatial differentiation apparent evidence that suggests fire played a minor role in the of the use of fire and the discard of tools with the pervasive role early hominin record and the revolutionary impact of fire on for fire in cooking envisioned by the cooking hypothesis. early hominin adaptations proposed by Wrangham (2009). It is now essential to determine how the seasonal exploitation of collected fire might have offered an adaptive advantage, which, Stage 2: Enhanced maintenance of fire was associated with the while more modest than the sweeping changes envisioned by development of base camp sites. The maintenance of fire re- Wrangham, was nonetheless significant. quires such occupation and there is strong evidence for the One of the greatest hazards in the study of early human in- development of this behavior between 400,000 and 200,000 teractions with fire is the possibility that during some periods years ago. the activities related to the use of fire would have been spatially distinct from the activities that result in the accumulation of Evidence from recent excavations at Schöningen, Beeches large assemblages of stone tools and faunal remains. If these Pit, and Qesem Cave suggests that by 400,000 years ago hom- Chazan Long Prehistory of Fire S355 inins had begun to collect and maintain fire (Karkanas et al. society that Isaac (1984) attributed to the earliest toolmaking 2007; Preece et al. 2006; Stiner, Gopher, and Barkai 2010; hominins. Thieme 2005; although for Schöningen, see also Stahlschmidt It is important to emphasize that the knowledge and skill et al. 2015). The most compelling evidence comes from Qesem in the manipulation of fire during this stage was extremely Cave, which is the only site with in situ micromorphological sophisticated, as is indicated by the emerging indications that evidence of burning (Barkai et al. 2017; Karkanas et al. 2007). fire was used to transform materials, including lithics and The lower component of Qesem Cave is primarily geogenic and resin (Brown et al. 2009; Koller et al. 2001; Pawlik and Thissen preserves the remains of solitary hearths. The upper sequence 2011). Although these techniques would enhance the role of follows a major rockfall and is primarily anthropogenic, con- fire, including for cooking, in hominin adaptation the reliance sisting of a 4.5 meter thick ash-rich deposit. The lithic industry on fire as a seasonal resource would fall below the bar of year- is blade dominated and attributed to the Amudian industry. round obligate cooking envisioned by the cooking hypothesis. Faunal analysis supports hunting of large game with hearth- based processing and sharing of meat (Stiner, Gopher, and Bar- Stage 3: Enclosure of fire is a late development not associ- kai 2010). ated with seasonal collection and maintenance (Sandgathe The evidence for the use of fire on archaeological sites in- et al. 2011). There appears to be an association between the tensifies around 200,000 years ago at the onset of the Middle enclosure of fire with habitual use of techniques to create Stone Age/Middle Paleolithic with sites with deep deposits of fire. This stage in the scenario proposed here would provide ash and charcoal and large quantities of burned bone and stone the basis for an adaptation dependent on year-round obli- tools becoming a major component of the archaeological rec- gate cooking. ord (see, e.g., Meignen et al. 2009). The source of fire in these fi contexts has not been the subject of signi cant discussion, but Perhaps one of the oddest aspects of the archaeological rec- fi I would set out as a proposition that the use of re continued ord from the Middle Stone Age/Middle Paleolithic is the ab- to be a seasonal aspect of adaptation based on the collection of sence of constructed hearths, although in some cases, stones fi fi re rather than evidence for a technology of re production. appear to have been pushed aside to clear an area (see Fer- If it is true that these adaptations included the maintenance nández Peris et al. 2010), and there is evidence of a stone-lined fi but not the creation of re, there are a number of implications. hearth at Qesem (see Shahack-Gross et al. 2014). As enclosure fi fi The rst of these is that evidence for re should not be found is seemingly cognitively undemanding, there is a need to ex- on all sites but should be limited to those sites occupied during plain why this behavior did not develop for over 300,000 years fi seasons of availability of natural sources of re (and perhaps of intensive hominin interaction with fire through collection not annually). For example, in publications on the site of Ho- and maintenance. lon, Israel, which is dated to ca. 200,000 years ago and contem- Similarly, there is no reason to question that early homi- porary with the later stages of the Qesem Cave sequence, Hor- nins had the dexterity or mental capacity needed to create fire. fi witz and I have pointed out the absence of any evidence of re The abilities for early hominins to operate on the basis of com- at this site, in marked contrast to Qesem (Chazan and Horwitz plex concepts and with high degrees of skill are well established 2006; Horwitz and Chazan 2016). based on studies of stone tool technology. However, there is no fi If re could only be collected but was becoming an increas- inherent reason to assume that the dexterity used in lithic pro- ingly critical aspect of hominin adaptation, this would also place duction would be transferred to the domain of producing fire fi a high degree of importance on the maintenance of re. As a simply because making fire would offer adaptive advantages. result there would be a degree of tethering of groups to places Modern hunter-gatherers relied on friction to create embers fi where re was maintained in order to prolong the availability that could then be coaxed into a flame. In his ethnography of of this resource. The buildup of hearths at sites like Kebara the !Kung San, Richard Lee provides a good description of the Cave appears to be consistent with this kind of activity, as does use of a fire drill to make fire, a skill he reports all !Kung men the demonstration based on phytolith analysis that consider- and women still possessed at the time of his fieldwork. How- able quantities of wood and wood bark were brought onto sites ever, he emphasizes that this is a difficult technique and it as fuel (Albert et al. 1999, 2000, 2003). An adaptation based on was widely abandoned in favor of a flint and steel kit that could fi the maintenance of seasonally collected re would encourage be purchased in trading stores (see also Friede 1978). It is the development of base camps and likely a division of la- worthwhile quoting Lee’s description of the use of the fire drill: bor such that some members of a group would remain on-site while others were engaged in hunting and gathering across the Two different kinds of wood are used: a hard wood such as landscape. Isaac’s home base/food-sharing model built on the Catophractes alexandri for the drill, and a softer wood such idea that there was a relationship between technology and as Ricinodendron rautanenii (mongongo) for the base. The social organization (Isaac 1978). As initially proposed by Rol- operator cuts a notch near the tip of the base stick held flat land (2004), the emergence of hearths as a central component on the ground, with a knife blade to receive the coal, and of hominin occupation sites might be the context for the emer- places the tip of the drill stick in the notch. He twirls the gence of the home base/food-sharing model of hunter-gatherer drill stick rapidly between his hands with a firm downward S356 Current Anthropology Volume 58, Supplement 16, August 2017

pressure, taking care to keep the drill tip from slipping out early hominin adaptation. At Wonderwerk there is a need to of the notch. . . . Drilling fire looks deceptively easy in the explore the possibility of creating fire by striking sparks off hands of a skilled operator; yet real muscular strength and ironstone. In the Fauresmith of Wonderwerk there is evidence control are required to get the fire started. Even experts in for incised lines on specularite, but it seems difficult to asso- the task are sweating with exertion after a minute’s drilling ciate these shallow features with efforts to create fire (Watts, (Lee 1979:148). Wilkins, and Chazan 2016). The emergence of rotative technologies and technologies Similar techniques have been documented using longitudinal involving grooving during the latter part of the Middle Stone friction in a trough to generate a spark (Perlès 1977; Rossotti Age and during the Upper Paleolithic make it plausible that 1993). during these periods hominins regularly created fire. It is in- Since the survival of wooden artifacts in early hominin con- teresting that in the Upper Paleolithic a range of methods also texts is rare and, to date, no drilling sticks have been identified, emerged for containing fire, including built hearths and por- we have no direct access to evidence of the initial creation of table lamps (on lamps, see de Beaune 1987; examples of built fire. However, we might be able to infer the presence of this hearths include Leesch, Cattin, and Müller 2004, fig. 244; Leroi- technology if we assume that specific pervasive techniques (as Gourhan and Brézillon 1973, figs. 122 and 131; Movius 1977, opposed to generalized cognitive capacity) will be transferred pls. 27 and 53; for a discussion of containers, see Gamble 2007). across raw materials. Thus, for example, during the Lower Pa- One possibility is that the adoption of built hearths was due leolithic percussion was applied to bone, apparently a transfer to new functional demands (see March, Muhieddine, and Ca- of a technique from lithics to bone (Gaudzinksi et al. 2005). If not 2009), but it is also possible that the shift toward enclos- drilling or grooving to create fire was an essential aspect of sur- ing fire, and the linked shift toward creating fire, are related to vival, it is reasonable to expect at least the occasional transfer of a change in the conceptualization of fire. This is an abstract these motions to other materials. construct, but it draws on recent theoretical writing on the na- Use wear on stone tools can provide a useful test for whether ture of artifacts that stresses that even chemical compounds— rotative action was part of the behavioral repertoire. Use wear such as water—can take on the status of artifacts (Bloom 2007; analysis of Middle Paleolithic assemblages focuses on damage Sperber 2007). It might be that what took place was a shift in to the tips of tools due to use as weapons, and I have been the nature of the concept of fire, away from being purely a unable to find explicit mention of use of tools as drills in the natural kind and toward hybridity. Paul Bloom makes a key Middle Paleolithic. For example, in his analysis of use wear distinction: “natural kinds [are] understood in terms of inter- on points from Kebara Cave, Units IX–XII, Shea (1988) does nal essences; artifacts are thought of in terms of consider- not report evidence of drilling. However, Plisson and Beyries ations such as creator’s intent, characteristic function, and the (1998) report that 25% of the tools studied from Stratum X and social and cultural context of the artifact’s creation and use” 6% of the tools from Stratum XI were used for perçage. From (Bloom 2007:154). However, Bloom argues that in the process the illustration provided, it is not clear whether this corre- of concept formation and learning language children accept sponds to a rotative action (see Tringham et al. 1974, fig. 20) that some categories are understood as both artifact types and or whether this is an act of applying pressure without rota- natural kinds. Could we conceive of the shift toward the pro- tion. It is interesting that the assemblage from Umm el Tlel did duction of fire as not only a change in technology, adaptation, not provide any evidence for the use of tools in this fashion and social structure but also a change in the conceptualiza- (Plisson and Beyries 1998). tion of the boundary between the natural and the artifact? Of Currently there is little compelling evidence in the archae- course, the adoption of methods to regularly create fire would ological record before the later part of the Middle Stone Age also have important implications not only for adaptation but and the early Upper Paleolithic for artifacts made using groov- also for social organization, notably in supporting increased ing or drilling. In the absence of such artifacts it would seem mobility. It is interesting that the Upper Paleolithic is linked to unlikely that hominin groups over this long time period con- the emergence of a specialized hunting tool kit that suggests sistently included individuals highly skilled in these techniques. emergence of hunting as a distinct sphere of activity in contrast There is an argument in the literature that fire could be created to domestic activity (Chazan 2010; Tartar et al. 2005). by striking flints against each other, resulting in a spark that – could then cause ignition of tinder (Perlès 1977:33 34). Al- The Prehistory of Fire and Hominin Speciation though there are many reports of using pyrite for this pur- pose (Sorensen, Roebroeks, and Van Gijn 2014; Stapert and Correlating the scenario for the prehistory of fire presented here Johansen 1999), and of course striking flint against metal was with hominin speciation events is challenging. In the cooking the basis for early firearms, it remains to be proven that the hypothesis (Wrangham 2009) the use of fire is connected to the sparks created by striking flints create enough energy to cause first appearance of Homo erectus. The scenario presented here ignition. Creating sparks through percussion might be signif- does not alter the possibility that the appearance of H. erectus icant for localized contexts where appropriate raw material is was linked with a shift in hominin engagement with fire, al- available but is not compelling as a general argument for fire in though as noted above, the implications would not have been as Chazan Long Prehistory of Fire S357 far-reaching as envisioned by the cooking hypothesis. Stage 2 speed that cannot be measured. Fire is extraordinary, but early is presented here as a gradual process that develops during humans were familiar with the extraordinary and were adept the period between 400,000 and 200,000 years ago. This would at manipulating forces that cannot be seen and that modern postdate the appearance of Homo heidelbergensis, if such a science still struggles to understand. The use of fire is one as- taxon has validity, but predate the first appearance of both pect of the ability of early humans to control forces beyond Neanderthals and modern humans (for discussion of H. heidel- comprehension—and from this perspective, the origins of fire bergensis, see Bräuer 2008; Stringer 2002). Stage 3 is also pre- is just one component of the ongoing process of human evo- sented as a gradual process that is possibly precocious in Africa, lution. where it occurred subsequent to the first appearance of modern humans. In Europe the appearance of stage 3 appears to cor- relate well with the Middle to Upper Paleolithic transition and Acknowledgments the arrival of modern humans. I would like to thank Dennis Sandgathe and Francesco Berna The important point here is precisely the lack of a direct for the opportunity to participate in the Wenner-Gren sym- correlation between speciation events in the hominin lineage posium and this publication. I particularly thank Francesco fi fi and shifts in the relationship with re, which ts with a con- Berna for years of collaboration in which he has opened my ception of a give-and-take between innovation in hominin eyes to evidence of fire that I had overlooked and provided a technology and aspects of biological evolution including spe- welcome empirical challenge to my theoretical ideas. I would ciation and dispersal. While mutations might have emerged also like to thank Liora Kolska Horwitz, Paul Goldberg, and all that led to shifts in hominin behavior and/or cognition, it is also the other members of the Wonderwerk Cave research project plausible that shifts in technology might have altered the se- and the Kathu Complex research project for sharing their lective advantage of traits already existing in populations, lead- diverse methodological skills and insights and also to Mark ing over time to changes in allele frequencies and ultimately to Lehner, who many years ago gave me the opportunity to work fi speciation. The concept of a long prehistory of re moves away on the Giza Plateau. This paper builds on research carried out from viewing technology as an endowment made possible by with funding from the Canadian Social Sciences and Human- favorable genetic mutations toward a more complex concep- ities Foundation and the Paleontological and Archaeological tion that looks at the interplay between genetics and culture in Scientific Trust (PAST) for which I am very grateful. driving hominin evolutions.

Conclusion References Cited Albert, R. M., O. Bar-Yosef, L. Meignen, and S. Weiner. 2000. Phytoliths in The cooking hypothesis is a bold intervention in the study of the Middle Paleolithic deposits of Kebara Cave, Mt. Carmel, Israel: study of human evolution. It is built on the idea of the origin of fire as a the plant materials used for fuel and other purposes. Journal of Archaeo- logical Science 27:931–947. single unitary event. An alternative scenario is presented here, ———. 2003. Quantitative phytolith study of hearths from the Natufian and in which the use of fire, and by extension, reliance on cooked Middle Palaeolithic levels of Hayonim Cave (Galilee, Israel). Journal of Ar- meat, was a process that extended over almost 2 million years. chaeological Science 30:461–480. fi Albert, R. M., A. Tsatskin, A. Ronen, O. Lavi, L. Estroff, S. Lev-Yadun, and The developing interactions with re played an integral role S. Weiner. 1999. Mode of occupation of Tabun Cave, Mt. Carmel, Israel in shifting selective pressures. Changes in the interaction with during the Mousterian Period: a study of the sediments and phytoliths. fi Journal of Archaeological Science 26:1249–1260. re were both a cause and a result of speciation and dispersal fi fi Alperson-A l, Nira, D. Richter, and Naama Goren-Inbar. 2007. Phantom events. I have emphasized the multiple aspects of re, partic- hearths and the use of fire at Gesher Benot Ya’akov, Israel. PaleoAnthro- ularly the relationships between society and technology. The pology 2007:1–15. adoption of a long prehistory of fire also provides a new per- Alperson-Afil, Nira, G. Sharon, M. Kislev, Y. Melamemd, I. Zohar, S. Ashke- nazi, R. Rabinovich, et al. 2009. Spatial organization of hominin activities at spective on criteria for acceptable evidence of early human use Gesher Benot Ya’aqov, Israel. Science 326:1677–1680. of fire, away from the particular case and toward synthesis of Archibald, S., R. J. Scholes, D. P. Roy, G. Roberts, and L. Boschetti. 2010. multiple lines of evidence. Southern African fire regimes as revealed by remote sensing. International Journal of Wildland Fire 19:861–878. In concluding, I would like to return to thinking about how Barbetti, M. 1986. Traces of fire in the archaeological record, before one million early humans would have experienced fire. Fire remains a mys- years ago. Journal of Human Evolution 15:771–781. terious force, a mystery that is often drawn into religious ritual. Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a rea- son: barbecue at Middle Pleistocene Qesem Cave, Israel. Current Anthro- Fire is a chemical reaction and escapes our normal experience pology 58(suppl. 16):S314–S328. of the tactile and real. While fire is extraordinary, it is also true Bellomo, R. V. 1994. Methods of determining early hominid behavioral ac- that early humans were in regular interaction with the invisi- tivities associated with the controlled use of fire at FxJj 20 Main, Koobi Fora, Kenya. Journal of Human Evolution 27:173–195. ble. Hunting would require an understanding of the behavior Berna, Francesco, A. Behar, Ruth Shahack-Gross, J. Berg, J. Zorn, E. Boaretto, of animals; collecting would have required similar knowledge A. Gilboa, et al. 2007. Sediments exposed to high temperatures: recon- of the processes through which plants grow and multiply. The structing pyrotechnological processes in Late Bronze and Iron Age Strata at Tel Dor (Israel). Journal of Archaeological Science 34:358–373. simple act of making a stone tool involves the propagation of Berna, Francesco, Paul Goldberg, Liora Kolska Horwitz, J. Brink, S. Holt, fracture, an act that irreversibly ruptures chemical bonds at a M. Bamford, and Michael Chazan. 2012. Microstratigraphic evidence of S358 Current Anthropology Volume 58, Supplement 16, August 2017

in situ fire in the Acheulean strata of Wonderwerk Cave, Northern Cape Johnson, S. D., and R. G. Balice. 2006. Seasons within the wildfire season: province, South Africa. Proceedings of the National Academy of Science 109: marking weather-related fire occurrence regimes. Fire Ecology 2(2):60–78. E1215–E1220. Karkanas, P., Ruth Shahack-Gross, A. Ayalon, M. Bar-Matthews, Ran Barkai, Bloom, Paul. 2007. Water as an artifact kind. In Creations of the mind: the- A. Frumkin, Avi Gopher, and M. C. Stiner. 2007. Evidence for habitual use of ories of artifacts and their representations. E. Margolis and S. Laurence, eds. fire at the end of the Lower Paleolithic: site formation processes at Qesem Pp. 150–156. Oxford: Oxford University Press. Cave, Israel. Journal of Human Evolution 53(2):197–212. Borradaile, Graham J., J. D. Stewart, and W. A. Ross. 1998. Characterizing Koller, J., U. Baumer, and D. Mania. 2001. High-tech in the Middle Palaeo- stone tools by rock-magnetic methods. Geoarchaeology 13(1):73–91. lithic: Neandertal-manufactured pitch identified. European Journal of Ar- Bräuer, G. 2008. The origin of modern anatomy: by speciation or intraspecific chaeology 4(3):385–397. evolution? Evolutionary Anthropology: Issues, News, and Reviews 17(1):22–37. Landau, M. 1993. Narratives of human evolution. New Haven, CT: Yale Uni- Brown, K., C. W. Marean, A. I. R. Herries, Z. Jacobs, C. Tribolo, D. Braun, D. L. versity Press. Roberts, M. C. Meyer, and J. Bernatchez. 2009. Fire as an engineering tool of Lee, Richard B. 1979. The !Kung San: men, women, and work in a foraging early modern humans. Science 325:859–862. society. Cambridge: Cambridge University Press. Brown, S. A., M. E. Collinson, and Andrew C. Scott. 2013. Did fire play a role in Leesch, D., M.-I. Cattin, and W. Müller. 2004. Témoins d’implantations formation of dinosaur-rich deposits? an example from the Late Cretaceous of magdaléniennes et aziliennes sur la rive nord du lac de Neuchâtel: Hauterive- Canada. Palaeobiodiversity and Palaeoenvironments 93(3):317–326. Champréveyres et Neuchâtel-Monruz. Hauterive: Service et Musée Cantonal Byers, B. A., S. R. Ash, D. Chaney, and L. DeSoto. 2014. First known fire scar d’Archéologie de Neuchâtel. on a fossil tree trunk provides evidence of Late Triassic wildfire. Palaeo- Lehner, Mark. 1996. Pyramid Age bakery reconstructed: experimental ar- geography, Palaeoclimatology, Palaeoecology 411:180–187. chaeology offers clues to ancient baking technology. AERAgram: Newsletter Carmody, R. N., M. Dannemann, A. W. Briggs, B. Nickel, E. E. Groopman, of the Ancient Egypt Research Associates 1(1):6–7. R. W. Wrangham, and J. Kelso. 2016. Genetic evidence of human adapta- Leroi-Gourhan, A., and M. Brézillon. 1973. Fouilles de Pincevent: essai d’analyse tion to a cooked diet. Genome Biology and Evolution 8(4):1091–1103. ethnographique d’un habitat magdalénien. (La section 36.) Paris: Centre Chazan, Michael. 2010. Technological perspectives on the Upper Paleolithic. National de la Recherche Scientifique. Evolutionary Anthropology: Issues, News, and Reviews 19:57–65. Mahmoud, H., and R. Eissa. 2015. Bakeries at the Heit al-Ghurab site: an in- Chazan, Michael, and Liora Kolska Horwitz. 2006. Finding the message in troduction. In Settlement and cemetery at Giza. F. Sadarangani and A. Wit- complexity: the association of lithics and fauna on Lower Paleolithic mul- sall, eds. Pp. 15–32. Boston: AERA. tiple carcass sites. Journal of Anthropological Archaeology 25:436–447. March, R. J., M. Muhieddine, and E. Canot. 2009. Simulation 3D des struc- Chazan, Michael, and Mark Lehner. 1990. An ancient analogy: pot baked bread tures de combustion préhistoriques. Archéovision (Actes du Colloque Vir- in ancient Egypt and Mesopotamia. Paléorient 16(2):21–35. tual Retrospect) 4:19–29. Chazan, Michael, J. Wilkins, D. Morris, and Francesco Berna. 2012. Bestwood Matmon, A., A. J. Hidy, S. Vainer, O. Crouvi, D. Fink, Y. Erel, L. K. Horwitz, 1: a newly discovered Earlier Stone Age living surface near Kathu, Northern Michael Chazan, and ASTER Team. 2015. New chronology for the southern Cape province, South Africa. Antiquity 86(331):Antiquity Gallery. http:// Kalahari Group sediments with implications for sediment-cycle dynamics www.antiquity.ac.uk/projgall/chazan331/. and early hominin occupation. Quaternary Research 84(1):118–132. Chuvieco, E., L. Giglio, and C. Justice. 2008. Global characterization of fire ac- Meignen, L., Paul Goldberg, R. M. Albert, and O. Bar-Yosef. 2009. Structures de tivity toward defining fire regimes from earth observation data. Global Change combustion, choix des combustibles et degré de mobilité des groupes dans le Biology 14:1488–1502. Paléolithique moyen du Proche-orient (grottes de Kébara et d’hayonim, is- de Beaune, S. 1987. Lampes et godets au Paleolithique. Gallia Préhistoire, sup- raël). In Fuel management during the Palaeolithic and Mesolithic periods: new plement 23. tools, new interpretations. Isabelle Théry-Parisot, Sandrine Costamagno, and Epron, L. 1939. Le Tombeau de Ti. Memoires Publies par les membres de Auréade Henry, eds. Pp. 101–118. Oxford: BAR International Series 1914, l’Institut Francais d’Archeologie Orientale du Caire 65. Cairo: l’Institut Archaeopress. Francais d’Archeologie Orientale du Caire. Movius, H. 1977. Excavations at the Abri Pataud, Les Eyzies (Dordogne): stra- Fernández Peris, J., V. Barciela González, Ruth Blasco, F. Cuartero, H. Fluck, tigraphy. Cambridge: Peabody Museum. P. Sañudo, and C. Verdasco. 2010. The earliest evidence of hearths in south- Patterson, C. 1981. Significance of fossils in determining evolutionary rela- ern Europe: the case of Bolomor Cave (Valencia, Spain). Quaternary Inter- tionships. Annual Review of Ecology and Systematics 12:195–223. national 247:267–277. Patterson, L. W. 1995. Thermal damage of chert. Lithic Technology 20:72–80. Friede, H. M. 1978. Methods of traditional fire-making in pre-industrial South Pawlik, A., and J. N. Thissen. 2011. Hafted armatures and multi-component Africa. African Studies 37(2):227–234. tool design at the Micoquian site of Inden-Altdorf, Germany. Journal of Ar- Gamble, C. 2007. Origins and revolutions: human identity in earliest prehis- chaeological Science 39:1699–1708. tory. Cambridge: Cambridge University Press. Perlès, C. 1977. Préhistoire du feu. Paris: Masson. Gaudzinski, S., E. Turner, A. P. Anzidei, E. Àlvarez-Fernández, J. Arroyo- Plisson, H., and S. Beyries. 1998. Pointes ou outils triangulaires? données Cabrales. J. Cinq-Mars, V. T. Dobosi, A. Hannus, E. Johnson, S. C. Münzel, fonctionnelles dans le Moustérien levantin. Paleorient 24:5–24. and A. Scheer. 2005. The use of Proboscidean remains in every-day Palaeo- Porat, N., Michael Chazan, R. Grun, M. Aubert, V. Eisenmann, and Liora lithic life. Quaternary International 126:179–194. Kolska Horwitz. 2009. New radiometric ages for the Fauresmith industry Goren-Inbar, Naama, Nira Alperson, M. Kislev, O. Simchoni, Y. Melamed, from Kathu Pan, Southern Africa: implications for the Earlier to Middle A. Ben-Nun, and E. Werker. 2004. Evidence of hominin control of fire at Stone Age transition. Journal of Archaeological Science 37:269–283. GBY, Israel. Science 304(5671):725–727. Preece, R. C., J. A. J. Gowlett, S. A. Parfitt, D. R. Bridgland, and S. G. Lewis. 2006. Hardy, K., A. Radini, S. Buckley, R. Blasco, L. Copeland, F. Burjachs, J. Girbal, Humans in the Hoxnian: habitat, context and fire use at Beeches Pit, West R. Yll, E. Carbonell, and J. M. B. de Castro. 2017. Diet and environment Stow, Suffolk, UK. Journal of Quaternary Science 21(5):485–496. 1.2 million years ago revealed through analysis of dental calculus from Eu- Pruetz, Jill D., and T. C. LaDuke. 2010. Brief communication: reaction to fire rope’s oldest hominin at Sima del Elefante, Spain. Science of Nature 104(1– by savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: con- 2):2. ceptualization of “fire behavior” and the case for a chimpanzee model. Hodder, I. 1999. The archaeological process: an introduction. Oxford: Blackwell. American Journal of Physical Anthropology 141(4):646–650. Horwitz, Liora, and Michael Chazan. 2016. Taking a Lèvy walk: early hominin Purdy, B. A. 1975. Fractures for the archaeologist. In Lithic technology: making mobility in the Lower Paleolithic of the Southern Levant. In Fresh fields and and using stone tools. E. H. Swanson, ed. Pp. 133–141. The Hague: Mouton. pastures new: papers presented in honor of Andrew M. T. Moore. K. Lillios Richter, D., Nira Alperson-Afil, and Naama Goren-Inbar. 2011. Employing and Michael Chazan, eds. Pp. 165–186. Leiden: Sidestone. TL methods for the verification of macroscopically determined heat alteration Isaac, Glynn L. 1978. Food sharing and human evolution: archaeological evi- of flint artefacts from Palaeolithic contexts. Archaeometry 53(4):842–857. dence from the Plio-Pleistocene of East Africa. Journal of Anthropological Roebroeks, Wil, and P. Villa. 2011. On the earliest evidence for habitual use of Research 34:311–325. fire in Europe. Proceedings of the National Academy of Sciences 108(13):5209– ———. 1984. The archaeology of human origins: studies of the Lower Pleis- 5214. tocene of East Africa, 1971–1981. Advances in Old World Archaeology 3:1–87. Rolland, N. 2004. Was the emergence of home bases and domestic fire a James, Steven R. 1989. Hominid use of fire in the Lower and Middle Pleis- punctuated event? a review of the Middle Pleistocene record in Eurasia. Asian tocene: a review of the evidence. Current Anthropology 30(1):1–26. Perspectives 43(2):248–280. Chazan Long Prehistory of Fire S359

Rossotti, H. 1993. Fire. Oxford: Oxford University Press. Tartar, E., N. Teyssandier, F. Bon. and D. Liolios. 2005. Equipement de chasse, Sandgathe, Dennis M., Harold L. Dibble, Paul Goldberg, Shannon P. Mc- équipement domestique: une distinction efficace? réflexion sur la notion Pherron, A. Turq, L. Niven, and J. Hodgkins. 2011. Timing of the appearance d’invetissement technique dans les industries aurignaciennes. In Normes of habitual fire use. Proceedings of the National Academy of Sciences 108(29): techniques et pratiques sociales de la simplicite´ de outillages pré-et proto- E298–E298. historiques. L. Astruc, F. Bon, V. Léa, P.-Y. Milcent, and S. Philbert, eds. Scott, Andrew C. 2000. The pre-Quaternary history of fire. Palaeogeography, Pp. 107–117. Antibes: Editions APDCA. Palaeoclimatology, Palaeoecology 164(1):281–329. Tattersall, I., and N. Eldredge. 1977. Fact, theory, and fantasy in human pale- Shahack-Gross, Ruth, Francesco Berna, P. Karkanas, C. Lemorini, Avi Gopher, ontology. American Scientist 65(2):204. and Ran Barkai. 2014. Evidence for the repeated use of a central hearth at Thieme, H. 2005. The Lower Palaeolithic art of hunting: the case of Schöningen Middle Pleistocene (300 ky ago) Qesem Cave, Israel. Journal of Archaeological 13 II-4, Lower Saxony, Germany. In Hominid individual in context: archae- Science 44:12–21. ological investigations of Lower and Middle Palaeolithic landscapes, locales, Shea, J. 1988. Spear points from the Middle Paleolithic of the Levant. Journal and artefacts. C. Gamble and M. Porr, eds. Pp. 115–132. London: Routledge. of Field Archaeology 15:441–450. Tringham, R., G. Cooper, G. Odell, B. Voytek, and A. Whitman. 1974. Ex- Sorensen, Andrew, Wil Roebroeks, and Annelou Van Gijn. 2014. Fire pro- perimentation in the formation of edge damage: a new approach to lithic duction in the deep past? the expedient strike-a-light model. Journal of Ar- analysis. Journal of Field Archaeology 1:171–196. chaeological Science 42:476–486. Walker, M. J., M. López-Martínez, J. S. Carrión-García, T. Rodríguez-Estrella, Sperber, D. 2007. Seedless grapes: nature and culture. In Creations of the mind: M. San-Nicolás Del-Toro, J.-L. Schwenninger, A. López-Jiménez, et al. 2013. theories of artifacts and their representations. E. Margolis and S. Laurence, Cueva Negra del Estrecho del Río Quípar (Murcia, Spain): a late Early eds. Pp. 124–137. Oxford: Oxford University Press. Pleistocene hominin site with an “Acheulo-Levalloiso-Mousteroid” Palaeo- Stahlschmidt, M. C., C. E. Miller, B. Ligouis, U. Hambach, Paul Goldberg, lithic assemblage. Quaternary International 294:135–150. Francesco Berna, D. Richter, B. Urban, J. Serangeli, and N. J. Conard. 2015. Walker, S. J. H., V. Lukich, and Michael Chazan. 2014. Kathu townlands: a On the evidence for human use and control of fire at Schöningen. Journal high density earlier Stone Age locality in the interior of South Africa. PloS of Human Evolution 89:181–201. One 9(7):e103436. Stapert, D., and L. Johansen. 1999. Making fire in the Stone Age: flint and van Wagtendonk, Jan W., and D. R. Cayan. 2008. Temporal and spatial dis- pyrite. Geologie en Mijnbouw 78:147–164. tribution of lightning strikes in California in relation to large-scale weather Stiner, M. C., Avi Gopher, and Ran Barkai. 2010. Hearth-side socioeconomics, patterns. Fire Ecology 4(1):34–56. hunting and paleoecology during the late Lower Paleolithic at Qesem Cave, van Wilgen, Brian W., and R. J. Scholes. 1997. The vegetation and fire regimes Israel. Journal of Human Evolution 60:1–21. of southern-hemisphere Africa. In Fire in southern African savannas: eco- Stiner, M. C., S. L. Kuhn, S. Weiner, and O. Bar-Yosef. 1995. Differential logical and atmospheric perspectives. B. W. van Wilgen, M. O. Andreae, J. G. burning, recrystallization, and fragmentation of archaeological bone. Jour- Goldammer, and J. A. Lindesay, eds. Pp. 27–46. Johannesburg: University nal of Archaeological Science 22(2):223–237. of Witwatersrand Press. Stoczkowski, W. 1994. Anthropologie naïve, anthropologie savante: de l’origine Watts, Ian, Michael Chazan, and Jayne Wilkins. 2016. Early evidence for bril- de l’homme, de l’imagination et des idées reçues. Paris: CNRS. liant ritualized display: specularite use in the Northern Cape (South Africa) ———. 2002. Explaining human origins: myth, imagination and conjecture. between ~500 and ~300 ka. Current Anthropology 57:287–310. Cambridge: Cambridge University Press. Wilkins, Jayne, and Michael Chazan. 2012. Blade production ~500 thousand Stringer, C. 2002. Modern human origins: progress and prospects. Philosophical years ago at Kathu Pan 1, South Africa: support for a multiple origins hy- Transactions of the Royal Society B: Biological Sciences 357(1420):563–579. pothesis for early Middle Pleistocene blade technology. Journal of Archae- Swetnam, T. W. 1993. Fire history and climate change in giant Sequoia groves. ological Science 39:1883–1900. Science 262:885–889. Wrangham, Richard. 2009. Catching fire: how cooking made us human. New Szpakowska, K. 2003. Playing with fire: initial observations on the religious York: Basic. uses of clay cobras from Amarna. Journal of the American Research Center ———. 2017. Control of fire in the Paleolithic: evaluating the cooking hypoth- in Egypt 40:113–122. esis. Current Anthropology 58(suppl. 16):S303–S313. S360 Current Anthropology Volume 58, Supplement 16, August 2017

Identifying and Describing Pattern and Process in the Evolution of Hominin Use of Fire

by Dennis M. Sandgathe

Although research relating to Paleolithic fire use has a long history, it has seen a particular resurgence in the last decade. This has been fueled in part by improved analytical techniques, improved standards of data collection and reporting, and the discovery of new sites with important fire residues in Africa, the Middle East, and Europe. A major component of this new research has been to identify when “controlled use” and “habitual use” of fire developed among Pleistocene hominins. However, an important starting point of this discussion is defining what is meant by “controlled use” and “habitual use,” as these terms have come to be used in undefined, inconsistent ways in the literature. We also need to lay out clearly how these behaviors might be recognized in the archaeological record and come to some understanding of what the potential implications of the development of these technologies and their geographic and climatic contexts are for the course of hominin evolution.

Research into the early use of fire in prehistory has tended to trolled” use of fire (e.g., Berna et al. 2012), and this is surely an focus on two major questions: when did hominins first begin important question, a major hurdle in attempts to reach some using fire, and when did fire use become an integral compo- understanding of the development of pyrotechnology continues nent of hominin adaptations. Both of these questions tended to to be a limited appreciation of the difficulty in distinguishing be dealt with in a rather simplistic manner in previous decades. residues of hominin use of fire from naturally occurring fire, This simplicity is not surprising for the early years of interest especially in Lower Pleistocene contexts (Barbetti 1986; see also in prehistoric fire use, as researchers were only beginning to the prelude in Goldberg, Miller, and Mentzer 2017). This is develop some understanding of the nature of the available data somewhat less of an issue with research in Middle Pleistocene and the potential implications for fire use in early hominin contexts, where we do have some clear examples of hominin use adaptations. In recent years, we are starting to come to terms of fire. In these cases, the goal appears to be more one of iden- with how potentially long and complex the process of hominin tifying examples of long-term, successive use of fire interpreted development of pyrotechnology might have been. However, to be evidence of “habitual” fire use. So far, claims of habitual fire the current discussion often seems to continue to ignore the use are mainly only from single sites as opposed to examples of probable complexity of this development and how problematic regional patterns of use (although see Roebroeks and Villa 2011). the available evidence continues to be (with significant ex- After the very earliest use, regular or successive use could be a ceptions, e.g., Chazan 2017; Parker et al. 2016). This issue is reasonable expectation as the next major step in the development reflected in the continued suggestion in some of the literature of hominin use of fire. However, what we mean by the terms that there will be a single point in prehistory at which fire use “controlled” and “habitual” and what their implications for hom- was adopted by hominins and that from that point on it was inin evolution and adaptation are need to be discussed and used by all hominins everywhere (e.g., Barkai et al. 2017; better defined (e.g., Alperson-Afil 2017; Barkai et al. 2017). This Daniau, d’Errico, and Sanchez Goni 2011:1). It is becoming requires the use of more explicit terminology and the develop- readily apparent that this scenario is a significant oversimpli- ment of a theoretical framework that better reflects the rela- fication of how the process probably occurred. While research- tionship between the evidence recovered from the archaeological ers do continue to look for the very earliest evidence for “con- record and our interpretations of it.

Dennis M. Sandgathe is Lecturer in the Department of Archaeology Issues in Current Research on the Development at Simon Fraser University (8888 University Drive, Burnaby, British of Pyrotechnology Columbia V5A 1S6, Canada [[email protected]]) and at the Museum of Archaeology and Anthropology at the University of Pennsylvania Much has already been written on the issue of identifying the fi (3260 South Street, Philadelphia, Pennsylvania 19104, USA). This earliest evidence of hominin use of re (e.g., Barbetti 1986; paper was submitted 25 VII 16, accepted 20 I 17, and electronically Bellomo 1993; Goudsblom 1986; Gowlett et al. 1981; Gowlett published 17 V 17. and Wrangham 2013; James 1989; Pickering et al. 2008; Roe- q 2017 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2017/58S16-0018$10.00. DOI: 10.1086/691459 Sandgathe Describing the Evolution of Hominin Use of Fire S361 broeks and Villa 2011), but that is not the focus of this paper. and Goudsblom 1986 for discussions of the idea of stages of However, this continues to be a point of major debate, and the development of fire use). One exception is Pruetz and LaDuke reasons for this underlie, to a large extent, debates about the (2010:4), who propose three cognitive and, presumably, chro- development of hominin fire use in general (see Sandgathe and nologically successive stages that hominids would have to go Berna 2017 for some discussion of this). What is of concern in through over the course of the development of fire use: this paper is an attempt to achieve some understanding of what researchers interested in the development of fire use really mean 1. Conceptualization of fire: an understanding of the be- when they use specific terms to describe the hominin behavior havior of fire, which would allow activity in close prox- they see reflected in the archaeological record and how this imity to it; behavior fits into broader ideas about the nature of the whole 2. The ability to control a fire: the knowledge and ability to process of development of fire use. How are researchers envi- contain, feed, and extinguish fire; sioning the long-term process of the development of pyrotech- 3. The ability to start a fire: the knowledge and technology nology, and what terms do they use to describe this? The term necessary to create fire at will. most commonly used in the literature in reference to the pre- historic development of fire use is “controlled use.” In the recent A second exception is Smith (Monica L. Smith, personal com- literature another term has begun to become common: “ha- munication, 2015), who proposed four stages, with very similar bitual use.” Both terms might be appropriate and useful in cer- components, “that reflect increasing deliberation and control”: tain circumstances, but neither has been well defined so that (1) habituation, (2) use, (3) curation, and (4) manufacture (see other researchers understand what is meant or intended by also Chazan 2017 for a similar discussion but slightly different their use. This makes it difficult for individuals to follow how criteria). other researchers are imagining the process of the development The second issue with the use of the term “control” is that of fire use went, which makes it difficult to take into consider- while it is almost never explicitly defined or explained, for ation and build on others’ work. many researchers the apparent implication seems to be that control of fire meant hominins had the ability to create fire at will (Alperson-Afil 2008, 2012; Attwell, Kovarovic, and Kendal 2015; Brown 2009). “Control of Fire”/“Controlled Use of Fire”? These distinctions are important considering the problems Much of the research on early fire use has been concerned we often have in even identifying genuine anthropogenic fire with identifying the appearance of the “controlled use of fire.” residues in early sites (e.g., Shöningen: Stahlschmidt et al. 2015; While the use of the term “control” (in specific reference to hom- and Zhoukoudien: Goldberg et al. 2001). One of the frequent inin fire use in the literature) probably appears quite early in problems in developing an understanding of the development of Paleolithic archaeology, it becomes almost ubiquitous by the pyrotechnology is achieving a reasonable degree of confidence late 1970s and early 1980s (e.g., Bellomo 1994; Clark and Har- that genuine fire residues identified at a site are actually an- ris 1985; Goudsblom 1986; Gowlett et al. 1981; James [and com- thropogenic and not the result of natural fires (this degree of ments] 1989; Rowlett 2000; it would not be difficult to find 30 confidence is, unfortunately, still a subjective thing and will vary or 40 more references) and is the current term of choice in between sites and researchers). For cave sites it has long been almost all the literature when referring to the early appearance recognized that this is not nearly as big a problem (Berna et al. of fire. 2012; Roebroeks and Villa 2011), although there are obviously There are two important issues with the use of the term situations in which natural fire residues can occur in cave de- “control.” The first issue is that for many researchers there is posits (spontaneous combustion of organic deposits or sedi- no discussion of, or indication of, an expectation of potential ments washing into a cave). For open-air sites, however, this stages in the development of fire use between when hominins must be seen as an important issue (see the prelude in Goldberg, were not using fire at all and their establishing control of it. Miller, and Mentzer 2017). In some regions and during some This might be because for these researchers “control” means climatic periods, natural fires of various types (grass fires, brush the most basic handling of fire, and so any potential evidence fires, forest fires) are essentially ubiquitous, especially in the of early fire use is considered evidence for control of it (e.g., context of geological or Paleolithic timescales. This is particu- Alperson-Afil 2012, 2017; Alperson-Afil, Richter, and Goren- larly the case for consistently warmer and drier climatic regions Inbar 2007; Bellomo 1993, 1994; Clark and Harris 1985; Goren- such as Africa and southwest Asia and the huge span of time Inbar et al. 2004). (Some researchers use both “use” and “control” represented by the Lower Pleistocene. The probability seems and seem explicitly to equate the terms, e.g., Bellomo [1993].) vanishingly small that the location of any open-air Early Stone There are some important exceptions. A few researchers have Age–Lower Paleolithic site would not have natural fires pass made a distinction between “using” fire and “controlling” fire over it at least once (and probably many times) in the period of (Goren-Inbar et al. 2004; James 1989) or “opportunistic” (Shi- time since its deposition. If the site is not too deeply buried, melmitz et al. 2014) versus “controlled” (James 1989) use or artifacts and bones can be altered by the heat of a passing natural “fortuitous use” versus control of fire (see also Bentsen 2014 fire, and charcoal and ash from natural fires can be introduced S362 Current Anthropology Volume 58, Supplement 16, August 2017 into the site sequence (see Aldeias 2017 and Aldeias et al. 2016 2. Fire residues identified at a site are demonstrably the result of for a discussion of heat transfer into sediment substrates, and hominins using fire, but we have no way to know how they see Gowlett et al. 2017 for a study of the effects of natural fires acquired it (James’s [1989] and Goren-Inbar et al.’ s [2004] on exposed objects). “using fire” and Smith’s [personal communication, 2015] Claims for early hominin use of fire are often based on ar- “use” of fire). guments that the fire residues in question either bear no rea- 3. Hominins were using fire that they collected from a nat- sonable similarity to residues generally associated with natural ural source (Shimelmitz et al.’s [2014] “opportunistic use” fires (e.g., Bellomo 1993, 1994; Berna et al. 2012; Gowlett et al. or James’s [1989] “fortuitous use”). 2005; Isaac 1982; James 1989; and Pickering et al. 2008) or 4. Hominins were using fire that they created with fire- the site deposits could not reasonably have been postdeposi- making technology. tionally altered by natural fires (e.g., Alperson-Afil, Richter, and Goren-Inbar 2007; Goren-Inbar et al. 2004; Pickering et al. These reflect increasing levels of understanding of hominin 2008). This might be a reasonable approach in some cases, but behavior that, to achieve, would necessarily require increas- currently we still do not know enough about natural fires and ingly better quality and types of data and increased confi- their potential range of resulting residues—either types of dence in our understanding of those data. I would suggest residues, patterns of their dispersion, or how the heat of such that for the majority of (perhaps all) claims of hominin use of fires may alter sediments and objects they come into contact fire associated with Middle Paleolithic–Middle Stone Age con- with—to realistically make such arguments in many (perhaps texts and earlier, we are, at best, at the second level: fire residues most?) cases. Some experimental work has been done on this, identified at a site are demonstrably the result of hominins using but most of this has been very limited and has not gone far fire, but we have no way to know how they acquired it. However, enough in bracketing the potential range of natural fire types, getting at some of these interpretations is going to be particu- their characteristics, their residues, and their effects on sub- larly difficult, for example, distinguishing hominin use of fire strates (e.g., Aldeias 2017; Aldeias et al. 2012; Bellomo 1993; that they created from hominin use of fire collected from a Canti and Linford 2000; Gowlett et al. 2017; March 1992). We natural source. do not know, for example, how frequently and in what cir- cumstances a tree or bush burning down into its root system will result in patches of blackened or rubified sediments that “Habitual” Use of Fire at Individual Sites look similar to the remains of an actual hearth. Or, as a further example, we do not have a good understanding at all of the Much of the recent work on early fire has been focussed more relationship between variation in natural fire frequencies and on identifying Middle Pleistocene examples of repeated and resulting charcoal distributions and concentrations in regional continuous fire use at individual sites (e.g., Aldeias et al. 2012; sediment records (e.g., Peters and Higuera 2007). With respect Alperson-Afil 2008; Blasco et al. 2015; Karkanas et al. 2007; to the issue of distinguishing residues of anthropogenic fire Shimelmitz et al. 2014). To date, the Lower Pleistocene record from natural fire, our interpretations have consistently been of hominin fire use is restricted entirely to Africa and southwest getting ahead of our understanding of the available data. Asia and is best characterized in these regions as sketchy, to Therefore, in many cases we are still debating the origin say the least. In light of this, the first appearances of examples of fire residues at archaeological sites, and so it is obviously of repeated fire use within a site rightly take on major signif- problematic to start assuming that hominins created the fire. icance in the history of development of pyrotechnology. Even in cases where it seems very clear that the fires were the The oldest of these is the open-air site of Gesher Benot result of hominin behavior, there still remains the possibil- Ya‘akov (Israel), dated to approximately 800 kya, which ap- ity that they acquired the fire from natural sources and did pears to have a few superimposed layers with fire residues not create it themselves. This possibility seems to be consis- (Alperson-Afil 2017). However, the earliest unquestionable ex- tently overlooked, underappreciated, or simply dismissed out amples of hominin use of fire and long-term, continuous fire of hand. use occur in cave sites in Israel dating from the latter half of It seems logical to attempt to make some distinction between the Middle Pleistocene. Between 350 and 200 kya we have the the different potential interpretations of the archaeological re- notable examples of Hayonim Cave, Qesem Cave, and Tabun cord. Depending on the nature of the available evidence and our Cave, where the sequences have recorded what appear to re- confidence in it, we may arrive at one of the following general flect regular and successive use of fire over much of this pe- interpretations: riod. This record includes impressive examples of sequences of stacked hearths (on the order of dozens at Hayonim Cave; 1. There are genuine fire residues associated with an archae- e.g., Bar Yosef et al. 2005; Goldberg and Bar-Yosef 1998; Shiegl ological site, but (at least currently) we have no way to et al. 1996) and notably high percentages of burned lithics in determine confidently whether they are associated with successive layers spanning tens of thousands of years in other hominin use of fire or are simply the result of natural fire cases (e.g., Tabun Cave; Shimelmitz et al. 2014). So far these and have no actual association with hominin behavior. sites appear to be the earliest evidence for fire use potentially Sandgathe Describing the Evolution of Hominin Use of Fire S363 being a “regularly” repeated behavior and perhaps an integral hominin use of fire starting by 400 kya (Roebroeks and Villa component of a local population’sadaptation. 2011). It seems probable that in this case the term “habitual” This long-term successive use of fire at individual sites has has a different intended meaning. been described as the first evidence for “habitual” use (Bentsen But even if we were able to demonstrate that fire was be- 2014; Karkanas et al. 2007; Roebroeks and Villa 2011; Shahack- ing used at every visit to a particular site over very long pe- Gross et al. 2014; Shimelmitz et al. 2014). The term “habitual” riods or being used frequently across a region, this still does has come into regular use in recent years. Shahack-Gross et al. not necessarily imply that these hominins were creating the (2014) define “habitual” (as they use it) as “systematically re- fire themselves. Examples of increased frequency or more reg- peated use of fire in specific sites and/or regions” (12). Shi- ular use of fire at a site or in a region may just be reflecting melmitz et al. (2014), in their observation of the fire record at regular access to natural fire sources in latitudes and climatic Tabun, make a distinction between “occasional and opportu- periods where the frequencies of natural fire are elevated—it nistic use of fire” and “habitual and planned” and suggest that may still be opportunistic use of natural fi re (Sandgathe et al. “habitual” means that fire was “a consistent element in behav- 2011b) but in situations where the opportunity to access nat- ioral adaptations” (196). While it may very well be the case that ural fire is frequent or even constant. Ultimately, our under- different researchers intend different meanings when they use standing of the development of pyrotechnology will need to the term, “habitual use” typically means (or at least implies) rely even more heavily on the analysis of site-level data (sed- “regular,”“persistent,”“continuous,” or “perpetual.” In ap- iments and residues; see Goldberg, Miller, and Mentzer 2017), plying it to prehistoric fire use, would this mean daily use and but we also need to seriously consider any apparent geographic at every occupation of a site by the hominins in question? It and temporal patterning of fire residues in the archaeological certainly seems to be strongly implied (though not explicitly record. stated) that at this level of use, hominins were making fire themselves and not relying on natural sources. Recognizing What the Patterns Mean However, without some understanding of the actual fre- quency of fire use at these sites and whether these groups are When it comes to interpreting the available data on Paleolithic actually creating fire at will, this becomes problematic. While fire use, of course we recognize that there are some general at a coarse level the fire sequences may be described as “reg- considerations. These include some basic taphonomic reali- ular,”“successive,” or even “continuous,” there may still be ties (see Aldeias 2017; Goldberg, Miller, and Mentzer 2017; decades, centuries, or in some cases even millennia between fire- Gowlett and Wrangham 2013 for more on this): use events. The types of data necessary to achieve the necessary resolution between fire events recorded in a site sequence can • A general loss of archaeological sites over time: the older the only potentially come from micromorphology, and even with time period, the lower the percentage of sites that have been this there are often going to be cases that are not definitive (see, preserved because of simple geologic and erosional attrition. e.g., Aldeais et al. 2012; Goldberg et al. 2012). • A loss of ephemeral fire residues over time: the greater the passage of time, the fewer the fire residues that tend to survive even in sites that have been preserved. “Habitual” Use of Fire across a Region • Fire residues are likely to survive better in protected (cave) Roebroeks and Villa (2011) take a broader geographical sites than in open-air sites. This is especially the case with (continental) approach and discuss the evidence for the ap- ephemeral residues such as ash and charcoal and less so with pearance of habitual fire use in Europe. They make the case that residues such as burned bone or burned lithics. increased frequencies of fire residues in Europe after 400 kya reflect the appearance of habitual use. The current evidence The result is that there will be a general loss of evidence for seems to show that any real use of fire only began after 400 kya fire in successively older sites. Because of this we must rely (potential fire use at Cueva Negra is the single example that more heavily on types of evidence that are not typically af- predates the late Middle Pleistocene; Walker et al. 2016). fected by such taphonomic processes, for example, frequen- However, in higher latitudes—Europe in particular (although cies of burned flints, magnetic susceptibility, Fourier transform data from East and South Asia are very limited)—the evidence infrared spectroscopy, and micromorphological analysis (see suggests that while some hominins in some places and at some Dibble et al. 2017; Goldberg, Miller, and Mentzer 2017). times were definitely using fire, this use clearly remained in- There are also some important general considerations in termittent and spotty even quite late in the Paleolithic. There not only the types of data but how the data should be compiled. is compelling evidence that as recently as the latter half of the For example, Gowlett and Wrangham (2013) correctly argue Late Pleistocene, at least some hominin populations were not that the simple use of presence-absence data severely limits always using fire during significant occupations of cave sites our potential understanding of early fire use. Simple presence- (e.g., Aldeias et al. 2012; Dibble et al. 2017; Goldberg et al. 2012; absence comparison of fire residues between different sites will Sandgathe et al. 2011a, 2011b). However, even in this context, not be particularly informative. For example, a single site with the term “habitual” has come to be used to describe European 10 stratigraphic layers, one of which has fire residues, will count S364 Current Anthropology Volume 58, Supplement 16, August 2017 in the literature as a single example of positive evidence of fire • Major climatic change: we can expect to see the gradual use, while a site with 10 layers, none of which has fire residues, appearance of intermittent fire use at higher latitudes as counts as a single example of lack of evidence of fire use. Such hominins began to use fire more frequently, mainly corre- a scenario, with a total of 20 distinct strata potentially span- sponding to warm and wet climatic periods when natural ning many millennia only one of which has evidence for fire fire was more readily available. use, would be presented as 50% of the sites having evidence for fire. This was the approach that Roebroeks and Villa (2011) While much of the literature has presented arguments for evi- took. This is not intended as a criticism of their paper, as it was dence of “control of fire,” the current data on Paleolithic fire use the first to even attempt to carry out broader regional analysis may simply be a reflection of a reliance on natural fire sources, of fire-use patterns based on a large database, and they should which would be dependent on lightning frequencies through be credited for this, but we are at the point where we should time and across geographic space. For example, although hom- begin taking more sophisticated approaches to the available inins had been occupying higher, cooler latitudes in Europe data. and Asia since well before 1.0 mya, the very earliest potential It does seem to be the case that we are now developing an evidence for fire use (claims dating from 1.6 mya to 800 kya) appreciation for the limitations of such basic presence-absence is in equatorial or subtropical latitudes (Africa and the Middle approaches, and there are some recent important exceptions East), where lightning frequencies would have remained rela- where researchers have provided robust frequency data on tively high throughout the Pleistocene because these regions fire residues or proxy fire data from individual sites. Two such would not have been as affected by global climatic cycles as examples (and undoubtedly there are others) are the recent higher latitudes were (fig. 1). work on evidence for fire use at Tabun (Shimelmitz et al. Assuming all the current claims for very early hominin use 2014) and many of the recent publications on the fire residues of fire in these regions are correct, the data still reflect a very at Gesher Benot Ya‘akov (Alperson-Afil 2008, 2017; Alperson- spotty, intermittent record (e.g., Koobi Fora FxJj20 [Hlubik et al. Afil, Richter, and Goren-Inbar 2007; Karkanas et al. 2007), 2017], Chesowanja in Kenya, Gadeb in Ethiopia, and Swart- where interpretation is based on quantifiable data that will krans and Wonderwerk Cave in South Africa). Furthermore, the arguably be very little affected by taphonomic processes (spe- first evidence of regionally based, repetitive or successive fire use cifically, burned flints; see Aldeias 2017; Dibble et al. 2009, is again restricted to subtropical latitudes beginning only be- 2017; Sandgathe et al. 2011a). We would argue that our basic tween 800 and 400 kya (Gesher Benot Ya‘akov, Tabun, Qesem, understanding of when and where fire was first used and when and Hayonim, all in Israel, and Cueva Negra in Spain). In lat- and where its use became regular is still, obviously, going to itudes above 357 north, the earliest potential evidence for any depend heavily on detecting the presence of early examples fire use is quite late, at ca. 400 kya (e.g., Beeches Pit in the United of clear anthropogenic fire use. What needs to be done is to Kingdom, Bilzingsleben in Germany, and VértesszőlősinHun- use a more sophisticated version of what constitutes presence gary). Even after fire does start appearing in Europe, the inter- and absence along with the use of more quantitative data and mittent nature of the evidence throughout the late Middle and analyses. We also need a more sophisticated understanding early Late Pleistocene shows strong patterns of correspondence of what we might view as the “background noise” of natural to warmer climatic periods, which could still simply be reflecting fire residues that presumably make up the majority of fire a reliance on natural fire sources. These patterns are presented residues in the larger depositional record of a region (although in table 1, which is based on a limited literature review of evi- the fires being of natural origin does not preclude them being dence for fire in 377 stratigraphic levels from 52 Lower and exploited by hominins). Middle Paleolithic sites across Europe, with presence or absence We can, perhaps, suggest some general expectations about of fire residues at all the components of these sites following how the overall evidence might present. At least initial fire use Roebroeks and Villa’s (2011) qualitative criteria. was probably dependent on access to and exploitation of nat- While these data must be viewed with caution (see table 1 ural fire sources, which will typically be caused by lightning. notes), they appear to show three important things. The first Because temperature and humidity are the biggest factors in is that there is currently a single site with potential evidence lightning frequencies, presence and absence and frequency of for hominin use of fire in Europe before marine isotope stage use were probably spatially and temporally dependent because (MIS) 11, although we have clear evidence that hominins had access to natural fire was dependent on climate and environ- arrived there before MIS 35 and potentially by MIS 45 (Car- ment. Therefore, until people developed fire-making techniques, bonell et al. 2008; Moyano et al. 2011). The second is that aside the pattern was probably one of intermittent fi re use depend- from a complete lack of evidence from MIS 10 and 9, after this ing, in large part, on the following. there is a general, though not entirely consistent, trend toward increasing frequency of fire residues. The third is that while • Latitude: we can expect to see initial, more frequent, and there are examples of fire use in later cold periods, there more regular fire use in warmer latitudes, where natural appears to be a strong correlation between fire frequencies and fire frequency was not (or was less) affected by global cli- warm periods. Taken at face value, this could be reflecting an mate variability. ongoing reliance on natural fire sources. Intermittent fire use Sandgathe Describing the Evolution of Hominin Use of Fire S365

Figure 1. Location of the earliest sites (by region) with potential evidence for hominin fire use mapped onto the frequency distri- bution of lightning strikes today. In spite of the earliest appearance of hominins in different regions, the earliest evidence for fire use appears to follow a latitudinal pattern with the earliest sites located closer to the equator. The farther from the equator, the later the dates of the earliest sites with potential hominin fire use. This indicates that not only is there temporal patterning to the appearance of fire use, there is also a tentative geographic pattern that could be reflecting a reliance on natural fires, the frequency of which will have been strongly influenced by latitude. Darker areas indicate increased frequency of lightning strikes. Image modified from http:// geology.com/articles/lightning-map.shtml).

in Europe (beginning in the late Middle and continuing through cannot necessarily be argued to be clear evidence for hominins early Middle Late Pleistocene) and examples of longer-term, having fire-makingtechnologyduringthesetimeperiods.Cur- successive fire use in southwest Asia starting in the Middle rently the archaeological evidence does not support a scenario Pleistocene could be argued to be part of the same pattern: in which any hominins (in Europe, Asia, or Africa) were using in both cases, hominins relying on natural fire that is pre- fire regularly enough to suggest that it was an integral part of dominantly a product of lightning frequency, which is strongly their adaptation until sometime in the Late Pleistocene. Nei- associated with warm climatic conditions. If fire-making tech- ther does the current evidence support a scenario in which the niques had been developed at this point and were widely em- existence of fire-making technology can be recognized or in- ployed, then there should be more examples in high-latitude ferred until very late as well. The implication is that the ac- regions with clear evidence for long-term, successive fire use. tual process of the development of pyrotechnology was more In fact, if hominins could make fire then, we might anticipate complicated than has been presented in the literature so far. a strong positive correlation between fire residues and cold pe- We are at the point where we need to move beyond the con- riods. The argument here is that an early, long-term reliance cept of a point in time where all hominins have “control” of on natural fire sources is a very plausible explanation given the fire, which eventually leads to fire use becoming “habitual” available data. The main point that should be taken from this among all hominins. So, when it comes to evidence for fire use, is that even in the aggregate, the quality of our data and its po- how do we describe what we are finding in the archaeological tential to provide concrete interpretations are very limited and record, and how do we describe our interpretations of it? S366 Current Anthropology Volume 58, Supplement 16, August 2017

Table 1. Percentage of stratigraphic components per marine isotope stage (MIS)

MIS Evidence for fire 34–3456789101112–45 0 60.0 25.0 55.0 58.8 81.8 45.0 66.7 100.0 .0 50.0 95.7 1 10.8 .0 25.0 2.1 13.6 20.0 33.3 .0 .0 .0 .0 2 6.9 7.1 12.5 20.6 .0 30.0 .0 .0 .0 8.3 4.3 3 22.3 67.9 7.5 18.6 4.5 5.0 .0 .0 .0 41.7 .0 Total 100 100 100 100 100 100 100 100 0 100 0 Total number of components 130 28 40 97 22 20 3 3 0 12 23 Note. No evidence for fire p 0; possible evidence but insufficiently described or no supporting evidence p 1; good evidence of fire p 2; clear evidence of fire p 3. The total number of components included for each stage is in the bottom row. Some components could only be placed in either MIS 3 or MIS 4 (not one or the other), but the number was significant, and their exclusion from the table would have biased the results. It needs to be stressed that there are a number of potential issues with this type of data:

• The dating of sites and site components and their placement in specific MISs is often difficult, and sometimes there are conflicting data. Because of this, many more sites and site components were reviewed in the literature but were not included here because of issues with their dating. • Different researchers use different criteria for distinguishing discrete layers, some reporting for lithostratigraphic units and some for archaeological levels. • For some sites, certain potential evidence for fire might not have been reported, e.g., percent frequencies of burned lithics and burned bone. • Some researchers might include mention of fire residues for a specific site because they were found in at least one component of the site, but they do not necessarily make it clear whether the residues occur in one, some, or all components. • For some excavations (especially earlier ones) there might not be confirmation that sediments/residues reported as charcoal or ash were actually such residues. • We can expect differential preservation patterns at closed versus open sites.

Developing Theoretical Models and Associated Terminology tation, a commonly used term is “hearth,” which brings with it inherent implications that may or may not be supported by the A persistent topic in the symposium on which this supple- actual fire residues identified at a site. The terms “control” and mental issue of Current Anthropology is based was the nature “habitual,” as they have generally been used in the literature, are of the various levels of analysis and interpretation inherent in examples of attempts to describe stages of hominin use of fire; trying to develop an understanding of the process of the that is, they relate mainly to the third category, the process of the development of hominin use of fire. However, any discussion development of hominin interaction with fire. These terms ei- is immediately limited by the lack of any established theo- ther need to be replaced or at least be better defined when they retical model(s) for such a process and a lack of common are used. I would argue for the former because both terms have terms that researchers can use to explain their points of view. alreadycometobesobaggageladen. Such discussions are necessarily going to involve different categories of terminology depending on at what point we are in the research process. Specifically, we can identify three basic levels at which we might develop specific terminologies: Describing the Theoretical Process of Development of Hominin Interaction with Fire 1. Describing archaeological phenomena: what terminology do we use to describe the actual residues of fire recovered As discussed above, we can imagine a theoretical process of from archaeological sites? the development of the interaction between fire and people. 2. Interpretation of archaeological phenomena: what termi- Realistically, there had to have been a development of in- nology do we use to describe in what specific context we creasing complexity of hominin association with fire begin- think those residues were created? This would be site- ning from simple interaction with natural fire in the environ- specific behavior. ment (a very common thing in grassland environments) to 3. The theoretical process of development of hominin in- the eventual invention of fire-making technologies (Clark and teraction with fire: what do we think is the broader be- Harris 1985; Pruetz and LaDuke 2010; Rolland 2004). Dif- havioral context of the theoretical development of fire use ferent researchers will undoubtedly have different ideas about in which these residues were created? This involves our the pace and temporal scale of this development and the in- interpretation of the role of fire in hominin adaptations. terpretation of what it means in terms of hominin adapta- tion. We should also take care not to intentionally couch this Each of these levels already includes its own regularly used development in terms of stages that imply directional progress terms, and at least some of these are also problematic. For ex- or linear movement. However, we are realistically talking about ample, among archaeological phenomena and their interpre- levels of increased complexity. If individual “stages” in this in- Sandgathe Describing the Evolution of Hominin Use of Fire S367 creasing complexity are not viewed as necessary preconditions We would not expect to see Solutrean points in an Oldowan for other stages, then we can avoid any implication of inherent assemblage. However, it is also the case that at some times and in linear progress. In some cases, hominin groups may potentially some places, groups may not have followed this sequence. For skip stages in the development of their use of fire that other example, for some groups, initial use and maintenance may hominins had gone through. However, practically speaking, it have developed at more or less the same time. will probably be the case that in most (pre)historic circum- This sequence does not necessarily have to have occurred stances there were common stages in the development of the just once in prehistory or at the same time and the same rate use of fire. among different hominin populations. For some populations, Another very important part of this discussion is the rec- simple use (with or without maintenance) may have been the ognition that the use of fire can be completely unrelated to limit of their fire use for very long periods of time before the the maintenance or manufacture of fire and could even involve ability to manufacture fire developed—if it did. In some re- no real control of fire. We can easily imagine such scenarios: gions (and time periods) high frequencies of natural fires may hominins simply cooking a piece of meat over burning vege- have provided some hominin groups with constant, reliable tation resulting from natural fire or, in an even less proximate access to fire, limiting any pressure to develop fire-maintenance interaction with fire, hominins intentionally foraging in burned- techniques or fire-manufacture technologies. In other regions, out areas shortly after a natural fire has passed (Herzog et al. perhaps due to pressures resulting from low frequencies of 2014; Pruetz and LaDuke 2010). natural fires or the importance of fire in exploiting certain The goal should be to use terms that express logical levels resources, the development of fire-manufacture technology of increasing complexity or sophistication in the degree of might occur very shortly after habituation. And in some re- interaction between hominins and fire. Starting at some point gions and time periods, very low natural fire frequencies and a in the past before which hominins were not interacting with lack of fire-making technology could have meant that the use fire at all, they probably began some sort of simple interaction of fire was simply not an important part of some hominin with fire, which may be as basic as the suppression of flight in populations’ adaptations (Henry 2017). face of natural fire (Clark and Harris 1985). This has been ob- This terminology allows us to be interpretive about the ar- served among chimpanzees (Pruetz and Herzog 2017; Pruetz chaeological record while avoiding unsupported presumptions and LaDuke 2010). This would logically be followed by simple (an issue that, I would argue, exists with the terms “control” use or application of fire, such as simply using fire for a task and “habitual”). For example, depending on the specificde- regardless of how complicated the task, where that use occurs, tails, we might be able to argue that fire residues in a hominin and how that fire was acquired. The example, given above, of occupation are evidence of use and perhaps even maintenance, hominins cooking food over naturally burning vegetation would but the terminology makes it clear that in the event that there is represent this level of interaction. Presumably, at some point no positive evidence that the hominins actually manufactured some hominins could begin to maintain fire regardless of its the fire, the interpretation ended there. We have the ability to original source. This would involve adding fuel to vegetation deal with the strong disassociation between questions we might that had been set on fire by a natural cause. Eventually hominins want to ask—could the hominins at this site make fire?—and would have developed techniques for the actual manufacture our ability to answer them. and ignition of fire, creating fire where there was none. Based on this (and on the work of others such as Pruetz and LaDuke Conclusions 2010 and Monica L. Smith, personal communication, 2015) we can suggest some concise, specificterminologythatreflects these: We do not yet have proper evidence to make big claims about either the earliest fire use or about when fire use became a 1. Habituation to natural fire, regular component of technological repertoires and hominin 2. Use of fire, adaptations came to depend on it. What is becoming clear is 3. Maintenance of fire, that our terminology, the approaches we take in our research, 4. Manufacture of fire. and the interpretations we arrive at from our analysis should start with some basic expectations about the course of the Over the long-term course of hominin evolution, there might development of pyrotechnology. have been a consistent or common sequence to the appearance of these levels of interaction with fire. While not implicit or • The development of pyrotechnology must be assumed to necessary in all circumstances, logically, there is a certain degree have been a long, drawn-out process that was probably rel- of directionality to this list as presumably a hominin species had atively complex. to become habituated to fire before it could achieve the other • Initial fire use was probably intermittent with frequent fits levels of interaction. Or, if it had the technology and know-how and starts, and this might have been the situation for a to manufacture fire, it had probably already spent some time significant part of subsequent prehistory. using and maintaining fire. The use of fire has to have been a • Initial fire use was probably based on the exploitation of process like, for example, the development of lithic technologies. natural fire sources (mainly lightening-ignited vegetation S368 Current Anthropology Volume 58, Supplement 16, August 2017

where and when available) and perhaps included simple be the result of simple use of fire. Determining at what point fire maintenance at some times and in some places. hominins began fire maintenance becomes more problematic, • Before the development and discovery of fire-making tech- and this is even more the case for manufacture of fire. Before we nology, it is unlikely that regular (“habitual”?) use of fire can seriously develop an understanding of the role of fire in appeared among all hominins in all regions or even a single hominin biocultural evolution and adaptation, we need a better region at the same time—it probably became more regular understanding of the nature of these stages, including when and in certain regions or with certain populations for periods of where they appeared and their subsequent durations. time. A final note is that the development of fire-making tech- • The discovery of fire-manufacturing technology probably niques may also be dependent on biology and the emergence of occurred in multiple places and potentially even multiple hominin species with the requisite cognitive abilities. However, times in any one region. it should be stressed that fire making is a learned behavior (and, • Fire-manufacturing technology could well have been a rel- based on my own experience, very difficult to accomplish using atively late development. traditional methods even if one knows theoretically exactly how • Such technology may very well have been lost and redis- to do it), and the lack of fire-making techniques is not an a covered multiple times as well, either through group fission- priori indication of reduced cognitive abilities any more than a ing events or through local or regional extinction events of modern human society lacking computers would be. hominin populations. • The evidence might suggest that fire had come to be used Acknowledgments repeatedly and successively at a single site over a significant period of time, but this cannot be seen as de facto evidence I thank the other participants in the symposium “Fire and the for the regular, constant use by a population over an entire Genus Homo” for their knowledge, insight, sense of humor, region, never mind a species. and lively discussion. I would also like to thank my fellow guest editor, Francesco Berna. Thank you to the Wenner-Gren The available data make it clear that before at least the Late Foundation, its president Leslie Aiello, and conference program Pleistocene, hominins are not using fire all the time. If the associateLaurieObbinkformakingthesymposiumpossible evidence suggests that frequent fire use among Middle Pleis- and for ensuring that it ran smoothly. Thank you to the two tocene European hominins (that is to say, high-latitude groups) anonymous reviewers whose constructive suggestions im- does not appear before 400 kya, and these hominins were not proved this paper. Finally, many of the ideas in this paper are, using fire regularly (especially during cold periods), then this in large part, the product of discussions with my close col- necessarily has implications for claims for any earlier fire use in leagues Harold Dibble, Paul Goldberg, Shannon McPherron, Africa. If Lower and early Middle Pleistocene African hominins and Vera Aldeias over the last several years, and they deserve were using fire regularly (e.g., Tabun, Qesem) and knew how to my thanks for this. create it, then at least some groups would presumably take this technology with them when they moved out of Africa into higher, cooler latitudes. The bottom line is that the evidence References Cited might be reflecting a much simpler scenario of fire-use devel- Aldeias, V., H. L. Dibble, D. Sandgathe, P. Goldberg, and S. J. McPherron. opment: 2016. How heat alters underlying deposits and implications for archaeo- logical fire features: a controlled experiment. Journal of Archaeological Sci- ence 67:64–79. • Very intermittent and strictly opportunistic use of naturally Aldeias, V., P. Goldberg, D. M. Sandgathe, F. Berna, H. L. Dibble, S. P. available fire during the Lower and early Middle Pleistocene McPherron, and R. Zeljko. 2012. Evidence for Neandertal use of fire at Roc de Marsal (France). Journal of Archaeological Science 39(7):2414–2423, doi: (e.g., Roebroeks and Villa 2011; Shimelmitz et al. 2014), org/10.1016/j.jas.2012.01.039. • More regular use of natural fire sources beginning in the latter Aldeias, Vera. 2017. Experimental approaches to archaeological fire features half of the Middle Pleistocene—still mainly opportunistic and their behavioral relevance. Current Anthropology 58(suppl. 16):S191– fi S205. exploitation of natural re where and when it was regularly Alperson-Afil, Nira. 2008. Continual fire-making by hominins at Gesher available—with perhaps the occasional (local?) development Benot Ya‘aqov, Israel. Quaternary Science Reviews 27:1733–1739. of fire-making technology. ———. 2012. Archaeology of fire: methodological aspects of reconstructing fire history of prehistoric archaeological sites. Earth-Science Reviews 113: 111–119. We need more objective and general terms to allow us to de- ———. 2017. Spatial analysis of fire: archaeological approach to recognizing early fire. Current Anthropology 58(suppl. 16):S258–S266. scribe these presumed increases in the complexity of hominin fi fi Alperson-A l, Nira, Daniel Richter, and Naama Goren-Inbar. 2007. Phantom use of re over time. Some of the very early associations of hearths and the use of fire at Gesher Benot Ya‘akov, Israel. Palaeoanthro- hominin occupations with fire residues in south and east Africa pology 2007:1–15. fl fi Attwell, Laura, Kris Kovarovic, and Jeremy R. Kendal. 2015. Fire in the Plio- may simply re ect either the ubiquitous nature of natural res in fi fi Pleistocene: the functions of hominin re use, and the mechanistic, de- Africa or, perhaps, some level of habituation to (natural) re. velopmental and evolutionary consequences. Journal of Anthropological However, some of these hominin fire-residue associations might Sciences 93:1–20. Sandgathe Describing the Evolution of Hominin Use of Fire S369

Barbetti, M. 1986. Traces of fire in the archaeological record, before one Gowlett, J. A. J., J. Hallos, S. Hounsell, V. Brant, and N. C. Debenham. million years ago? Journal of Human Evolution 15(8):771–781. 2005. Beeches Pit: archaeology, assemblage dynamics and early fire his- Barkai, Ran, Jordi Rosell, Ruth Blasco, and Avi Gopher. 2017. Fire for a tory of a Middle Pleistocene site in East Anglia, UK. Eurasian Prehistory reason: barbecue at Middle Pleistocene Qesem Cave, Israel. Current An- 3(2):3–38. thropology 58(suppl. 16):S314–S328. Gowlett, J. A. J., J. W. K. Harris, D. Walton, and B. A. Wood. 1981. Early Bar Yosef, O., A. Belfer-Cohen, P. Goldberg, S. L. Kuhn, L. Meignen, B. Van- archaeological sites, hominid remains and traces of fire from Chesowanja, dermeersch, and S. Weiner. 2005. Archaeological background to Hayonim Kenya. Nature 294:125–129. Cave and Meged Rockshelter. In The faunas of Hayonim Cave: a 200,000-year Gowlett, J. A. J., and R. W. Wrangham. 2013. Earliest fire in Africa: towards record of Paleolithic diet, demography, and society. M. C. Stiner and O. Bar- the convergence of archaeological evidence and the cooking hypothesis. Yosef, eds. Pp. 17–38. Cambridge, MA: Peabody Museum of Archaeology and Azania: Archaeological Research in Africa 48(1):5–30. Ethnology. Henry, Amanda G. 2017. Neanderthal cooking and the costs of fire. Current Bellomo, R. 1993. A methodological approach for identifying archaeological Anthropology 58(suppl. 16):S329–S336. evidence of fire resulting from human activities. Journal of Archaeological Herzog, N. M., C. H. Parker, E. R. Keefe, J. Coxworth, A. Barrett, and K. Science 20:525–554. Hawkes. 2014. Fire and home range expansion: a behavioral response to ———. 1994. Methods of determining early hominid behavioral activities burning among savanna dwelling vervet monkeys (Chlorocebus aethiops). associated with the controlled use of fire at FxJj 20, Main, Koobi Fora, American Journal of Physical Anthropology 154(4):554–560. Kenya. Journal of Human Evolution 27:173–195. Hlubik, Sarah, Francesco Berna, Craig Feibel, David Braun, and John W. K. Bentsen, Silje Evjenth. 2014. Using pyrotechnology: fire-related features and Harris. 2017. Researching the nature of fire at 1.5 Mya on the site of FxJj20 activities with a focus on the African Middle Stone Age. Journal of Ar- AB, Koobi Fora, Kenya, using high-resolution spatial analysis and FTIR chaeological Research 22(2014):141–175, doi: 10.1007/s10814-013-9069-x. spectrometry. Current Anthropology 58(suppl. 16):S243–S257. Berna, Francesco, Paul Goldberg, Liora Kolska Horwitz, James Brink, Sharon Isaac, G. 1982. Early hominids and fire at Chesowanja, Kenya. Nature 296:870. Holt, Marion Bamford, and Michael Chazan. 2012. Microstratigraphic James, Steven R. 1989. Hominid use of fire in the Lower and Middle Pleis- evidence of in situ fire in the Acheulean strata of Wonderwerk Cave, tocene. Current Anthropology 30(1):1–26. Northern Cape Province, South Africa. Proceedings of the National Acad- Karkanas, Panagiotis, Ruth Shahack-Gross, Avner Ayalon, Mira Bar-Matthews, emy of Science of the USA 109:7593–7594. Ran Barkai, Amos Frumkin, Avi Gopher, and Mary C. Stiner. 2007. Evidence Blasco, Ruth, Jordi Rosell, Pablo Sanudo, Avi Gopher, and Ran Barkai. 2015. for habitual use of fire at the end of the Lower Paleolithic: site-formation What happens around a fire: faunal processing sequences and spatial dis- processes at Qesem Cave, Israel. Journal of Human Evolution 53(2):197– tribution at Qesem Cave (300 ka), Israel. Quaternary International 398: 212. 190–209. March, Ramiro J. 1992. L’utilisation du bois dans les foyers préhistoriques: Brown, K. 2009. Fire as an engineering tool of early modern humans. Science une approche expérimentale. Bulletin de la Société Botanique de France: 325:859, doi: 10.1126/science.1175028. Actualités Botaniques 139(2–4):245–253. Canti, M. G., and N. Linford. 2000. The effects of fire on archaeological soils Moyano, I. T., D. Barsky, D. Cauche, V. Celiberti, S. Grégoire, F. Lebegue, and sediments: temperature and colour relationships. Proceedings of the M. H. Moncel, and H. De Lumley. 2011. The archaic stone tool industry Prehistoric Society 66:385–395. from Barranco León and Fuente Nueva 3 (Orce, Spain): evidence of the earli- Carbonell, E., J. M. B. de Castro, J. M. Parés, A. Pérez-González, G. Cuenca- est hominin presence in southern Europe. Quaternary International 243(1): Bescós, A. Ollé, M. Mosquera, et al. 2008. The first hominin of Europe. 80–91. Nature 452(7186):465–469. Parker, C. H., E. R. Keefe, N. M. Herzog, J. F. O'Connell, and K. Hawkes. Chazan, Michael. 2017. Toward a long prehistory of fire. Current Anthro- 2016. The pyrophilic primate hypothesis. Evolutionary Anthropology: Issues, pology 58(suppl. 16):S351–S359. News, and Reviews 25(2):54–63. Clark, J. D., and J. W. K. Harris. 1985. Fire and its roles in early hominid Peters, M. E., and P. E. Higuera. 2007. Quantifying the source area of mac- lifeways. African Archaeological Review 3:3–27. roscopic charcoal with a particle dispersal model. Quaternary Research 67: Daniau, A.-L., F. d’Errico, and M. Fernanda Sanchez Goni. 2011. Testing the 304–310. hypothesis of fire use for ecosystem management by Neanderthal and Pickering, T. R., C. P. Egeland, M. Domínguez-Rodrigo, C. K. Brain, and A. G. Upper Palaeolithic modern human populations. PLoS ONE 5(2):e9157. Schnell. 2008. Testing the “shift in the balance of power” hypothesis at Dibble, H., F. Berna, P. Goldberg, S. McPherron, S. Mentzer, L. Niven, D. Swartkrans, South Africa: hominid cave use and subsistence behavior in the Richter, D. Sandgathe, I. Thery Parisot, and A. Turq. 2009. A preliminary Early Pleistocene. Journal of Anthropological Archaeology 27:30–45. report on Pech de l’Azé IV, Layer 8 (Middle Palaeolithic, France). Paleo- Pruetz, J. D., and T. C. LaDuke. 2010. Brief communication: reaction to fire by Anthropology 2009:182–219. savanna chimpanzees (Pan troglodytes verus) at Fongoli, Senegal: concep- Dibble, Harold L., Aylar Abodolahzadeh, Vera Aldeias, Paul Goldberg, tualization of “fire behavior” and the case for a chimpanzee model. American Shannon P. McPherron, and Dennis M. Sandgathe. 2017. How did hominins Journal of Physical Anthropology 141(4):646–650. adapt to Ice Age Europe without fire? Current Anthropology 58(suppl. 16): Pruetz, Jill D., and Nicole M. Herzog. 2017. Savanna chimpanzees at Fongoli, S278–S287. Senegal, navigate a fire landscape. Current Anthropology 58(suppl. 16):S337– Goldberg, P., and O. Bar-Yosef. 1998. Site formation processes in Kebara and S350. Hayonim Caves and their significance in Levantine prehistoric caves. In Roebroeks, W., and P. Villa. 2011. On the earliest evidence for habitual use of Neandertals and Modern Humans in Western Asia. T. Akazawa, K. Aoki, fire in Europe. Proceedings of the National Academy of Sciences of the USA and O. Bar-Yosef, eds. Pp. 107–126. New York: Plenum. 108(13):5209–5214. Goldberg, P., H. L. Dibble, F. Berna, D. M. Sandgathe, S. J. P. McPherron, and Rolland, N. 2004. Was the emergence of home bases and domestic fire a punc- A. Turq. 2012. New evidence on Neandertal use of fire: examples from Roc tuated event? a review of the Middle Pleistocene record in Eurasia. Asian de Marsal and Pech de l’Azé IV. Quaternary International 247(1):325–340. Perspectives 43(2):248–280. Goldberg, P., S. Weiner, O. Bar-Yosef, Q. Xu, and J. Liu. 2001. Site formation Rowlett, Ralph M. 2000. Fire control by Homo erectus in East Africa and Asia. processes at Zhoukoudian, China. Journal of Human Evolution 41:483–530. Acta Anthropologica Sinica 19(suppl.):198–208. Goldberg, Paul, Christopher E. Miller, and Susan M. Mentzer. 2017. Recog- Sandgathe, D. M., H. L. Dibble, P. Goldberg, S. P. McPherron, A. Turq, L. nizing fire in the Paleolithic archaeological record. Current Anthropology 58 Niven, and J. Hodgkins. 2011a. On the role of fire in Neandertal adaptations (suppl. 16):S175–S190. in Western Europe: evidence from Pech de l’Azé IV and Roc de Marsal, Goren-Inbar, N., N. Alperson, M. E. Kislev, O. Simchoni, Y. Melamed, A. France. Paleoanthropology 2011:216–242. Ben-Nun, and E. Werker. 2004. Evidence of hominin control of fire at ———. 2011b. Timing of the appearance of habitual fire use. Proceedings of Gesher Benot Ya‘aqov, Israel. Science 304:725–727. the National Academy of Sciences of the USA 108(29):E298. Goudsblom, Johan. 1986. The human monopoly on the use of fire: its origins Sandgathe, Dennis M., and Francesco Berna. 2017. Fire and the genus Homo: and conditions. Human Evolution 1(6):517–523. an introduction to supplement 16. Current Anthropology 58(suppl. 16): Gowlett, J. A. J., J. S. Brink, Adam Caris, Sally Hoare, and S. M. Rucina. 2017. Ev- S165–S174. idence of burning from bushfires in southern and east Africa and its relevance Schiegl, S., P. Goldberg, O. Bar-Yosef, and S. Weiner. 1996. Ash deposits in to hominin evolution. Current Anthropology 58(suppl. 16):S206–S216. Hayonim and Kebara Caves, Israel: macroscopic, microscopic and miner- S370 Current Anthropology Volume 58, Supplement 16, August 2017

alogical observations, and their archaeological implications. Journal of Archae- Stahlschmidt, M. C., C. E. Miller, B. Ligouis, U. Hambach, P. Goldberg, F. ological Science 23:763–781. Berna, D. Richter, B. Urban, J. Serangeli, and N. J. Conard. 2015. On the Shahack-Gross, R., F. Berna, P. Karkanas, C. Lemorini, A. Gopher, and R. Barkai. evidence for human use and control of fire at Schöningen. Journal of 2014. Evidence for the repeated use of a central hearth at Middle Pleistocene Human Evolution 89:181–201. (300 ky ago) Qesem Cave, Israel. Journal of Archaeological Science 44:12–21. Walker, M. J., D. Anesin, D. E. Angelucci, A. Avilés-Fernández, F. Berna, A. T. Shimelmitz, R., S. L. Kuhn, A. J. Jelinek, A. Ronen, A. E. Clark, and M. Buitrago-López, Y. Fernández-Jalvo, et al. 2016. Combustion at the late Weinstein-Evron. 2014. “Fire at will”: the emergence of habitual fire use Early Pleistocene site of Cueva Negra del Estrecho del Río Quípar (Murcia, 350,000 years ago. Journal of Human Evolution 77:196–203. Spain). Antiquity 90(351):571–589. Forthcoming Current Anthropology Wenner-Gren Symposium Current Anthropology Supplementary Issues (in order of appearance)

VOLUME 58 SUPPLEMENT 16 AUGUST 2017 Human Colonization of Asia in the Late Pleistocene. Christopher J. Bae, Current Katerina Douka, and Michael J. Petraglia, eds. The Anthropology of Corruption. Sarah Muir and Akhil Gupta, eds. Cultures of Militarism. Catherine Besteman and Hugh Gusterson, eds. Anthropology

Previously Published Supplementary Issues

Working Memory: Beyond Language and Symbolism. Thomas Wynn and THE WENNER-GREN SYMPOSIUM SERIES Frederick L. Coolidge, eds.

Engaged Anthropology: Diversity and Dilemmas. Setha M. Low and August 2017 Sally Engle Merry, eds. FIRE AND THE GENUS HOMO Corporate Lives: New Perspectives on the Social Life of the Corporate Form. Damani Partridge, Marina Welker, and Rebecca Hardin, eds. GUEST EDITORS: DENNIS M. SANDGATHE AND FRANCESCO BERNA The Origins of Agriculture: New Data, New Ideas. T. Douglas Price and Ofer Bar-Yosef, eds. Fire and the Genus Homo The Biological Anthropology of Living Human Populations: World Histories, Recognizing Fire in the Paleolithic Archaeological Record National Styles, and International Networks. Susan Lindee and

Volume 58 Experimental Approaches to Archaeological Fire Features and Their Ricardo Ventura Santos, eds. Behavioral Relevance Human Biology and the Origins of Homo. Susan Antón and Leslie C. Aiello, eds. Evidence of Burning from Bushfires in Southern and East Africa and Potentiality and Humanness: Revisiting the Anthropological Object in Its Relevance to Hominin Evolution Contemporary Biomedicine. Klaus Hoeyer and Karen-Sue Taussig, eds. Ethnoarchaeology of Paleolithic Fire: Methodological Considerations Alternative Pathways to Complexity: Evolutionary Trajectories in the Middle Aboriginal Use of Fire in a Landscape Context: Investigating Presence and Paleolithic and Middle Stone Age. Steven L. Kuhn and Erella Hovers, eds. Absence of Heat-Retainer Hearths in Western New South Wales, Australia Supplement 16 Crisis, Value, and Hope: Rethinking the Economy. Susana Narotzky and Researching the Nature of Fire at 1.5 Mya on the Site of FxJj20 AB, Koobi Niko Besnier, eds. Fora, Kenya, Using High-Resolution Spatial Analysis and FTIR Spectrometry The Anthropology of Christianity: Unity, Diversity, New Directions. Joel Robbins Spatial Analysis of Fire: Archaeological Approach to Recognizing Early Fire and Naomi Haynes, eds. Evidence of Hominin Use and Maintenance of Fire at Zhoukoudian Politics of the Urban Poor: Aesthetics, Ethics, Volatility, Precarity. Veena Das and How Did Hominins Adapt to Ice Age Europe without Fire? Shalini Randeria, eds. Technologies for the Control of Heat and Light in the Vézère Valley The Life and Death of the Secret. Lenore Manderson, Mark Davis, and Aurignacian Chip Colwell, eds. Control of Fire in the Paleolithic: Evaluating the Cooking Hypothesis Reintegrating Anthropology: From Inside Out. Agustin Fuentes and Fire for a Reason: Barbecue at Middle Pleistocene Qesem Cave, Israel Polly Wiessner, eds.

Pages S163–S370 Neanderthal Cooking and the Costs of Fire New Media, New Publics? Charles Hirschkind, Maria José A. de Abreu, and Savanna Chimpanzees at Fongoli, Senegal, Navigate a Fire Landscape Carlo Caduff, eds. Toward a Long Prehistory of Fire Current Anthropology is sponsored by The Identifying and Describing Pattern and Process in the Evolution of Wenner-Gren Foundation for Anthropological Hominin Use of Fire Research, a foundation endowed for scientific, educational, and charitable purposes. The Foundation, however, is not to be understood as endorsing, by virtue of its financial support, any of Sponsored by the Wenner-Gren Foundation for Anthropological Research the statements made, or views expressed, herein.

T H E U N I V E R S I T Y O F C H I C A G O P R E S S