Neandertal Foraging Adaptations in Southwestern France: Implications of Mousterian Faunal Remains at Gatzarria Cave
A Thesis Submitted to the Committee on Graduate Studies in Partial Fulfillment of the Requirements for the Degree of Master of Arts in the Faculty of Arts and Science
Trent University
Peterborough, Ontario, Canada
(c) Copyright by Elspeth Ready 2010 Anthropology ?.?. Program
September 2010 Library and Archives Bibliothèque et 1*1 Canada Archives Canada Published Heritage Direction du Branch Patrimoine de l'édition
395 Wellington Street 395, rue Wellington OttawaOIMK1A0N4 OttawaONK1A0N4 Canada Canada
Your file Votre référence ISBN: 978-0-494-68237-1 Our file Notre référence ISBN: 978-0-494-68237-1
NOTICE: AVIS:
The author has granted a non- L'auteur a accordé une licence non exclusive exclusive license allowing Library and permettant à la Bibliothèque et Archives Archives Canada to reproduce, Canada de reproduire, publier, archiver, publish, archive, preserve, conserve, sauvegarder, conserver, transmettre au public communicate to the public by par télécommunication ou par l'Internet, prêter, telecommunication or on the Internet, distribuer et vendre des thèses partout dans le loan, distribute and sell theses monde, à des fins commerciales ou autres, sur worldwide, for commercial or non- support microforme, papier, électronique et/ou commercial purposes, in microform, autres formats. paper, electronic and/or any other formats. The author retains copyright L'auteur conserve la propriété du droit d'auteur ownership and moral rights in this et des droits moraux qui protège cette thèse. Ni thesis. Neither the thesis nor la thèse ni des extraits substantiels de celle-ci substantial extracts from it may be ne doivent être imprimés ou autrement printed or otherwise reproduced reproduits sans son autorisation. without the author's permission.
In compliance with the Canadian Conformément à la loi canadienne sur la Privacy Act some supporting forms protection de la vie privée, quelques may have been removed from this formulaires secondaires ont été enlevés de thesis. cette thèse.
While these forms may be included Bien que ces formulaires aient inclus dans in the document page count, their la pagination, il n'y aura aucun contenu removal does not represent any loss manquant. of content from the thesis.
¦+¦ Canada Abstract
Neandertal Foraging Adaptations in Southwestern France: Implications of Mousterian Faunal Remains at Gatzarria Cave Elspeth Ready
This study describes Neandertal foraging adaptations at Gatzarria, a Middle to Upper Palaeolithic transition site in southwestern France, through an analysis of faunal remains from level Cj. Stratigraphie problems relating to the division of the Mousterian level Cj and Upper Palaeolithic industries at Gatzarria were encountered during this research. However, a new study of lithic index fossils indicated sufficient stratigraphie separation of the artifacts to permit the selection of a secure sample of Mousterian faunal remains. This sample suggests that red deer were by far the main species encountered near Gatzarria. However, even the lowest-ranked ungulate species, roe deer and chamois, were sometimes included in the maximum diet breadth. No small game component is in evidence. The long bone assemblage suggests less stringent culling decisions at kill sites relative to other late Mousterian sites. Utilization of marrow was intensive. The Gatzarria data conform with general patterns of resource use during the Mousterian in western
France.
Keywords: Zooarchaeology, subsistence, foraging theory, Neandertals, Mousterian, Middle Palaeolithic, Gatzarria Cave, Atlantic Pyrenees
11 Acknowledgements
To begin, I would like to thank my supervisor Eugène Morin for his guidance and instruction over the past two years. Above all, I must express my gratitude to Eugène for knowing that only by continually pushing my limits would I continually improve my skills as a researcher and writer—and for being willing to invest the time and effort required to nurture my progress. Secondly, my committee members, James Conolly and Susan Jamieson, provided helpful comments, direction, and encouragement as my research developed. Besides supplying the maps in Chapters 3 and 7, James also provided a steady flow of interesting questions about zooarchaeological sampling and statistics that encouraged me to think more deeply about my treatment of faunal data. Dr. R. Lee
Lyman, from the University of Missouri, graciously accepted to serve as my external examiner and provided a number of helpful suggestions that have been incorporated into this final draft. My friends and colleagues here at Trent University, including, among many others, Chelsee Arbour, Kimberly Jankuta, Cat Robertson, Melissa Wallace, and Kristine Williams also deserve thanks for their support—both academic and social. In particular, however, I am indebted to Miranda Brunton, who has been my steadfast companion during long hours in the office over the past two years, and, outside of work, a genuinely considerate friend, and confidante; as well as Celise Chilcote, with whom I shared not only an apartment wall but also many frustrations and successes. Further away, in Edmonton, I would like to thank Elizabeth Sawchuk and Benjamin Flanagan for stimulating conversations about anthropology (and other things), and Elizabeth a second time for scrupulously editing scholarship proposals and more.
in Rob Losey, at the University of Alberta, was instrumental in getting me interested and involved in zooarchaeological research. Carleen, Mom, and Dad—thank you for your
emotional support, for keeping me company, for listening to me talk endlessly about archaeology, and (especially Carleen) for making me laugh. As for my summer in France, I would like to thank everyone who was part of the
productive and enriching experience I spent in Les Eyzies. Eugène once again deserves thanks for his help and support, especially for a trip across France during a stratigraphie crisis. Everyone at the Musée National de Préhistoire was welcoming and extremely helpful to me. I would like to thank Jean-Jacques Cleyet-Merle for the permission to work with the collections; Stéphane Madeleine for his assistance with many identifications; André Morala for his contribution to the study of the lithic material (and
for his patience with the bones scattered all over the lab during the summer); Nicolas Audebert for his help washing the bones; and Bernard Nicolas and Peggy Jacquement for their logistical support. I also greatly appreciated and learned much from my discussions with the researchers and students who spent time at the museum and at "la Maison
Bordes" over the summer. Laura Eizenberg kindly contributed her data on the Proto- Aurignacian lithic industry at Gatzarria, and Marie-Cécile Soulier shared her insights regarding the fauna of the Pyrenean Middle Palaeolithic sites. Finally, I am grateful to
Julien Alibert, Marie Fibule, Yann Goasguen, Nicolas Hecht, and Thibault Morala who warmly welcomed me in Dordogne and with whom I shared an unforgettable summer.
This study would not have been possible without financial support from a number of sources. My research was funded by a Social Science and Humanities Research Council (SSHRC) Joseph-Armand Bombardier Canada Graduate Scholarship (Master's)
iv to myself, and a SSHRC Standard Research Grant to Eugène Morin. I would also like to thank Trent University and Sandi Carr for the scholarships I received during my time at
Trent.
? Table of Contents
Abstract ii Acknowledgements iii Table of Contents vi List of Tables ix List of Figures xii
Chapter 1 : Introduction 1 Overview of the research objectives 2 Outline of the study 3
Chapter 2 : Neandertal Ecology and Formulation of the 4 Research Problem Hunting versus scavenging and the history of Neandertal subsistence 4 research Neandertal diet breadth 8 A Neandertal niche? 14 Ecology and the replacement or continuity debate 17 The research problem 18
Chapter 3 : Gatzarria Cave—Description and Previous 20 Research Geographic context 20 Excavation history and method 22 Stratigraphy 25 Geological stratigraphy 26 Archaeological stratigraphy 28 Stratigraphie problems 3 1 Previous archaeological research 33 Lithic industries 33 Bone industry and ornaments 34 Human remains 36 Palaeontological work 36
vi Ecological context 40 Palaeoenvironment 40 Ecology of the ungulate species at Gatzarria 42 Summary 52
Chapter 4: Theory and Method 53 Theoretical approach 53 The diet breadth model 54 The marginal value theorem 56 The central place forager prey choice model 59 Critiques of foraging theory 6 1 Application of foraging theory in this study 64 Defining resource patches 65 Prey rankings 66 Methodology 71 Quantification techniques and identification procedures 71 Bone modification and taphonomy 8 1 Summary 90
Chapter 5 : Sample Selection and Taphonomy 92 Sample selection 92 Stratigraphie integrity of Mousterian remains from Ej 93 Sample selection procedures 97 Stratigraphie integrity: Refit analysis 100 Overview of the Cj Mousterian faunal sample 102 Taphonomy 105 How did recovery methods affect the sample composition? 106 Density 112 The impact ofbone burning on sample composition 114 Post-depositional destruction 117 Biological agents of accumulation and destruction 119 Bone surface preservation 125 Intra-taxonomic identification bias 128 Inter-taxonomic identification bias 133 Summary 137
Chapter 6 : Ungulate Resource Exploitation at Gatzarria 139 Cave
VIl Taxonomic abundances 139 Skeletal part representation 142 Mortality profiles 145 Seasonality 148 Patch use and diet breadth 149 Foraging currencies and transport decisions 152 On-site processing of skeletal parts 162 Discussion 164
Chapter 7 : Neandertal Subsistence at Gatzarria in Regional 1 68 and Temporal Context Late Middle Palaeolithic foraging in France and northern Iberia 168 Close to home: Middle Palaeolithic sites in the Basque region and 169 Cantabria Looking north: Comparisons with Saint-Césaire and Grotte XVI 175 Discussion: Reconsidering the 'Neandertal niche' 187 Implications for the early Upper Palaeolithic at Gatzarria 190 Summary 192
Chapter 8 : Conclusion 193 Limitations of the present study and directions for further research 193 Summary of results 197 Final remarks 199
Bibliography 200
Vili List of Tables
Table 3.1: Sedimentary and archaeological stratigraphy at Gatzarria 3 1 Table 3.2: Large mammal remains identified by Lavaud (1980) 38
Table 3.3: Microfaunal, avian, and amphibian remains identified by Lavaud 38 (1980)
Table 4.1: Body mass, fatness, and total fat mass for ungulate and carnivore 69 species present at Gatzarria
Table 4.2: Size classes used in the faunal analysis 76 Table 5.1: Summary of studied faunal remains 103 Table 5.2: Identified taxa and NISP in the initial and the Cj Mousterian 104 samples
Table 5.3: Mesh sizes used in excavations at Gatzarria 106
Table 5.4: Size distribution of bone specimens, in a measured sub-subsample 108 of décapages, according to year of excavation Table 5.5: Frequency of burning in the study samples 110
Table 5.6: Frequencies of long bone portions in the Gatzarria Cj Mousterian 113 sample compared to shape-adjusted density values for reindeer
Table 5.7: Frequencies of human and carnivore marks in the Cj Mousterian 119 sample
Table 5.8: Mean lengths of long bone fragments for different taxon at 123 Gatzarria, two primarily anthropic Middle Palaeolithic deposits (Jonzac and Saint-Césaire), and a hyaena den (Bois Roche)
Table 5.9: Comparison of frequencies of long bone ends (epiphysis with shaft) 123 versus shaft-only fragments
Table 5.10: Overall surface state of NISP and NSUTS in the Cj Mousterian 126 sample
IX Table 5.11: Percentage observable surface ofNISP and NSUTS in the Cj 126 Mousterian sample
Table 5.12: Comparison ofbone surface states in the front and back of the cave 127 Table 5.13: Comparison of frequencies of cutmarks by overall surface state 128 Table 5.14: Representation of skeletal elements in NISP and NSUTS 129 Table 5.15: Fragmentation of red deer long bone shaft specimens in the Cj 132 Mousterian sample Table 5.16: Frequencies of refits on long bone shaft fragments 132
Table 5.17: Average and minimum lengths and areas for bison and red deer 135 long bone shaft fragments Table 5.18: Comparison of long bone identification rates for large and 137 medium-sized ungulates
Table 6.1: Red deer skeletal part frequencies in the Cj Mousterian sample 143 Table 6.2: Bison skeletal part frequencies in the Cj Mousterian subsample 144 Table 6.3: Abundances of selected red deer skeletal portions compared with 153 three utility indices Table 6.4: Spearman's rank order correlations between skeletal part 159 representation and the FUI, Nunamiut grease rendering episode, and UMI
Table 7.1: Percentages of ungulates in Mousterian levels at Gatzarria (Cj 170 Mousterian and preliminary data on Cjr), Axlor, and El Castillo Table 7.2: Abundances of ungulates in the Gatzarria Cj, Saint-Césaire EGPF, 178 and Grotte XVI Couche C Mousterian samples Table 7.3: Skeletal element abundances for selected high-density portions of 179 red deer at Gatzarria and reindeer at Saint-Césaire
Table 7.4: Skeletal element abundances for selected high-density portions of 181 red deer at Gatzarria and reindeer at Grotte XVI
Table 7.5: Spearman's rank order correlations between skeletal part 183 representation at Grotte XVI Couche C and the FUI, Nunamiut grease rendering episode, and UMI
? Table 7.6: Diversity of the Gatzarria, Grotte XVI and Saint-Césaire samples, 184 as measured by the reciprocal of Simpson's Index
Xl List of Figures
Figure 3.1: Location of Gatzarria Cave 21
Figure 3.2: Layout of Gatzarria Cave and excavation grid 24 Figure 3.3: Excavation sub-units and axes of coordinates for plotted artifacts 24
Figure 3.4: Theoretical stratigraphie profile at Gatzarria 26 Figure 4.1: Rankings of ungulate and carnivore species at Gatzarria according 68 to body mass and fat mass Figure 5.1: Original identification of layers within Ej during excavations at 94 Gatzarria
Figure 5.2: Two-dimensional projections of lithic artifacts from Ej in sagittal 96 profile Figure 5.3: Two-dimensional projection of lithic artifacts from Ej in frontal 97 profile Figure 5.4: Summary of the distribution of the Cj Mousterian sample 99 Figure 5.5: Green-bone refit on a red deer metatarsal 101
Figure 5.6: Size distributions of burnt and unburnt specimens in a measured 110 sub-sample of décapages from the Cj Mousterian sample Figure 5.7: Comparison of size distributions of burnt specimens from the 111 sub-sample of décapages in which all specimens were measured and all burnt specimens from the Cj Mousterian sample Figure 5.8: Frequency of long bone portions in the Gatzarria assemblage 113 versus bone density
Figure 5.9: Length distribution of identified bone specimens (NISP) at 122 Gatzarria compared to Bois Roche (a hyaena den) and Jonzac Level 22 (a Middle Palaeolithic deposit) Figure 5.10: Long bone shaft fragment lengths and circumferences in the Cj 124 Mousterian sample
XIl Figure 5.11: Comparison of long bone shaft fragment size distributions for 134 bison, red deer, and small ungulates
Figure 6.1: Fetal bones from the Cj Mousterian sample 149 Figure 6.2: Comparison of red deer skeletal part abundances with the caribou 156 FUI; the percentage of caribou parts saved for grease rendering by a Nunamiut woman; and the caribou UMI
Figure 7.1 : Locations of selected sites discussed in the text 1 69 Figure 7.2: Comparison of frequencies of high density bone specimens of 180 red deer in the Gatzarria Cj Mousterian sample and reindeer in the Saint-Césaire EGPF sample Figure 7.3: Comparision of frequencies ofhigh density bone specimens of 182 red deer in the Gatzarria Cj Mousterian sample and reindeer in the Grotte XVI Couche C sample
xiii 1
Chapter 1 : Introduction
For over 200,000 years, Neandertals survived in the variable and sometimes extremely harsh climate of Pleistocene Europe and, while doing so, left a remarkable archaeological record. Consequently, the Neandertals have received much attention in the study of hominin evolution. However, the question of where Neandertals end and modern humans begin is an increasingly complex problem, from biological, cultural, and ecological points of view. For instance, genetic data now point towards a contribution by Neandertals to the genome of some contemporary human populations (Green et al. 2010).
In other words, Neandertals likely played a role in the genesis of modern human diversity. In these respects, our understanding of the Neandertals (variably labelled archaic Homo sapiens or Homo neanderthalensis) is fundamental to our understanding of the evolutionary history of our species. Evolutionary change, regardless of whether it occurs through genetic drift, loss of genetic diversity, hybridization, assimilation, or extinction sensu stricto, involves ecological processes. A thorough understanding of a population's ecology is therefore essential to understanding its evolution. Resource acquisition, or subsistence behaviour, is a fundamental component of ecological adaptation (Kaplan and Hill 1992). Although substantial progress has been made in recent years, there remains much to be said regarding Neandertal subsistence behaviours. This study focuses on Neandertal foraging in the Atlantic Pyrenees, a region that has significant importance for studies of the Middle to Upper Palaeolithic transition, particularly with regards to chronology (e.g., Zilhäo and d'Errico 1999, 2000; Meilars 2000), but where Neandertal subsistence has not been studied in depth. In this study, Neandertal foraging is examined using Mousterian faunal remains from layer Cj at Gatzarria Cave, a stratified cave site in the French Basque country with deposits spanning the Middle to Upper Palaeolithic transition. Gatzarria was excavated over thirty years ago, but this research is the first detailed zooarchaeological analysis of remains from the
site.
Overview ofthe research objectives The general objectives of this analysis of the Gatzarria faunal remains are as
follows:
(1) To assess the stratigraphie integrity of the Cj Mousterian deposits at Gatzarria; (2) To carry out a taphonomic analysis of the faunal remains in order to determine
how different factors, such as burning, post-depositional destruction, carnivore ravaging, and excavation procedures, have shaped the assemblage; (3) To analyze Neandertal subsistence at Gatzarria during the Cj Mousterian occupation; (4) To consider the significance of the subsistence patterns documented at Gatzarria
by placing the results of the analysis within a broader comparative framework. This study will constitute a base line for understanding Neandertal subsistence at Gatzarria and, through comparisons with other sites, in the inland Atlantic Pyrenees. As a starting point for future analyses of subsistence during the Middle to Upper Palaeolithic transition at Gatzarria, this study also has relevance for understanding transition processes in the region. 3
Outline ofthe study The examination of Neandertal foraging at Gatzarria begins with a brief review of current understandings ofNeandertal ecology, which leads to the formulation of more specific research questions regarding Neandertal foraging behaviour (Chapter 2). A description of Gatzarria and of previous research on the site follows (Chapter 3). Next, the foraging models used to interpret the faunal remains are presented, along with the methods required to put this approach into action (Chapter 4). The analysis of the Gatzarria remains starts with a consideration of the site stratigraphy and taphonomy (Chapter 5), before proceeding on to a zooarchaeological analysis ofNeandertal foraging at the site (Chapter 6). The subsistence strategies documented at Gatzarria are then considered within a broader context, along with the implications of this research for our understanding ofNeandertal ecology (Chapter 7). Finally, the results of the study are summarized (Chapter 8). 4
Chapter 2: Neandertal Ecology and Formulation of the Research Problem
This chapter reviews current understandings ofNeandertal ecology, specifically regarding the exploitation of animal resources. What was the relationship between
Neandertals and their environment, and did this relationship differ in any significant way from that of modern Homo sapiens? How might the subsistence practices of the
Neandertals have been a factor in their eventual disappearance? This discussion prompts the formulation of research questions regarding Neandertal subsistence that the analysis of the Gatzarria Mousterian faunal remains will address.
Hunting versus scavenging and the history ofNeandertal subsistence research During the 1980s, a major controversy about Neandertal subsistence arose, largely as a result of the development of taphonomy as a necessary component of zooarchaeological study (Binford 1985). This controversy, which regarded the role of hunting versus scavenging in Neandertal subsistence, provided the impetus for substantial research.
Despite the rise of the ecologically-oriented New Archaeology in the 1960s, the 'Man the Hunter' paradigm, which held that hunting and meat-eating were developments necessary to early hominin evolution (Lee and DeVore 1968), dominated hominization theory through the 1970s. Because this model emphasized the hunting abilities of even the earliest hominins, many researchers considered Neandertal hunting abilities to be 5
highly advanced and similar to those of modern humans (e.g., Howell 1965; Pfeiffer 1969; Wood 1976). By the early 1980s, however, the taphonomic work of researchers such as Binford (1981) and Brain (1981) had severely undermined the hypothesis of hunting among early hominins. The method of procurement used by hominins during the Middle Palaeolithic soon came into doubt as well (e.g., Straus 1982; Dennell 1983).
Binford (1984:248) argued that until 40 to 35 kya (kilo-years ago), the carnivorous tactics of Middle Stone Age hominins at Klasies River Mouth, South Africa, involved
"scavenging coupled with opportunistic killing of small animals." Binford (1985) further suggested that Neandertals relied solely on scavenging to acquire large prey and that large game hunting was a unique ability of modern humans. This contentious argument
provided an important stimulus for interest in Neandertal subsistence and ecology, as their presumed inferior hunting abilities became a potential cause of their demise. Much research into Neandertal ecology during the 1980s and 1990s therefore focused on the
identification of scavenging among Neandertals as evidence for 'non-modern' behaviour. For example, Binford (1988) attempted to prove his hypothesis ofNeandertal scavenging through an analysis of faunal remains from Grotte Vaufrey, France. He argued that red deer, horse, and large bovids at the site were only accessed by hominins after partial consumption by carnivores. Subsequently, Grayson and Delpech (1994) reanalyzed
material from Grotte Vaufrey and found that none of Binford's conclusions regarding Neandertal scavenging were supported.
Stiner's research on Mousterian faunal remains from Italy was an essential component of the hunting/scavenging debate. Based on the skeletal part profiles produced by extant carnivores and modern human hunters, Stiner (1991, 1994) argued, following 6
Binford (1984), that head or head-and-foot dominated assemblages, as well as assemblages dominated by old individuals, could be indicative of scavenging. In her comparison of Middle and Upper Palaeolithic assemblages from Italy, Stiner suggested that this 'scavenging' pattern was present in some Middle Palaeolithic assemblages older than 55 kya, but not in younger assemblages. Therefore, Stiner argued that, prior to 55 kya, Neandertals regularly scavenged and hunted, while subsequent Neandertals and modern humans in the region exclusively hunted. Stiner (1991 :475) concluded that there was a "fundamental difference in how foraging systems were organized" between these two time periods. This difference involved a shift from generalized hunting and occasional scavenging to hunting of prime-aged adults. Stiner' s conclusions prompted a flurry of debate concerning the reality of the skeletal patterns she observed. Her conclusions were questioned on two main grounds. First, Marean and Kim (1998) argued that long bone shaft fragments were absent from some of the sites included in her analysis. These authors pointed out that many of Stiner' s collections were from old excavations whereonly 'identifiable' specimens were collected—in other words, long bone shaft fragments, ribs, and burnt bone fragments from these sites likely had been discarded. Because of the impact of density-mediated destruction (including carnivore gnawing) on long bone ends in archaeological assemblages, the absence of long bone shaft fragments may have created artificial head- and-foot patterns. Secondly, Pickering et al. (2003) questioned Stiner' s quantification techniques, suggesting that, even when long bone shafts were present in her collections, her counting methods underestimated the frequencies of long bone shafts. Stiner's MNE counts for long bones were based on lower-density bone ends, and therefore, they may 7
have been significant underestimates of the true frequencies of long bones in her samples. Furthermore, the sample sizes on which her conclusions were based were sometimes quite small (see Stiner 1994:389-397, 2005:242-281). Despite these shortcomings, Stiner's conclusion that Neandertals were not obligate scavengers has been a point of
agreement. Partially as a result of this debate, it became clear that although Neandertals may have scavenged opportunistically, they had the ability to hunt large game. Scavenging could no longer be taken as an indicator of 'primitive' subsistence behaviour among Neandertals (Burke 2000). The amount of scavenging undertaken by Neandertals remains unclear, as there is little agreement on what constitutes unequivocal archaeological evidence for scavenging. For instance, Steele (2004) suggests that although scavenging may have occurred at certain Middle Palaeolithic sites (Lazaret and Combe-Grenal), the possibility that carnivore ravaging subsequent to hominid involvement caused the 'scavenging' (old-dominated) mortality profiles cannot be eliminated. Steele emphasizes the importance ofreconstructing the taphonomic histories of assemblages prior to drawing behavioural inferences. The consensus emerging from the hunting/scavenging debate was that although Neandertals may have scavenged, they were also capable of hunting moderate to large- sized ungulate fauna. Neandertals in Western and Eastern Europe, outside of the Mediterranean rim, primarily hunted large-bodied terrestrial herbivores (Burke 2004;
Hoffecker 2009). Main prey species in these regions included red deer, horse, bison/aurochs, and in some periods, reindeer (Gaudzinski and Niven 2009). Problems that succeeded the hunting/scavenging debate involved the issue of hunting specialization and 8
the exploitation of animal resources besides moderate to large ungulates, particularly in Mediterranean Europe. These issues are addressed in the following sections.
Neandertal diet breadth
The extent to which Neandertals from different regions and different time periods incorporated animal resources other than highly profitable moderate- to large-sized ungulates into their diet is a matter of contention. Did Neandertals hunt or collect difficult-to-obtain faunal resources, such as megaherbivores (those >1000 kg; Owen- Smith 1988) and fish? Or were potentially less-profitable resources, such as small, fast game, always excluded from the optimal diet? The remains of extremely large game (e.g., mammoths, elephants, rhinoceros) are frequently present in small quantities in Middle Palaeolithic sites across Europe (Gaudzinski et al. 2005), but whether these remains are present due to opportunistic scavenging of carcasses or hunting is open to question (e.g., Hoffecker and Cleghorn 2000). However, the same ambiguity applies to proboscidean remains in most Upper Palaeolithic contexts (Gaudzinski et al. 2005). It is often difficult to prove—or disprove—that hominids, including modern humans, butchered the carcasses of large animals such as mammoths. Because of the thickness of the periosteum and cartilage, butchery marks are infrequently observed on proboscidean bones (Crader 1983; Haynes 1991, 2002), so their absence does not necessarily imply that butchering did not occur (Mussi and Villa 2008). The identification of cultural versus natural bone breakage patterns in proboscideans is also beset with difficulties (Haynes 2002). The difference between scavenging and hunting of megaherbivores is important because of the high risk that may be involved in this type of hunting (i.e., a low 9
probability of a successful kill after encounter; Bird et al. 2009), versus the very large potential returns of a scavenged carcass. Even if it is clear that humans processed an elephant or mammoth carcass, generally it is not possible to distinguish hunting from
scavenging behaviours, regardless of whether bone assemblages consist of a single carcass or of a large number of individuals (Gaudzinski et al. 2005). For example, between 200 and 130 kya, at la Cotte de St. Brelade (Jersey Islands), mammoth and
woolly rhinoceros fell into a natural trap and were subsequently butchered by hominids, apparently on two separate occasions (Scott 1980, 1989). Whether the remains are the result of a co-ordinated drive (as suggested by Scott) or scavenging of fallen animals is ambiguous. More compelling evidence of hunting comes from the site of Lehringen, Germany, where a wooden spear was found in conjunction with elephant remains (Gaudzinski 2004; Gaudzinski et al. 2005). Gaudzinski (2004, 2006) also mentions
probable evidence for hunting of young rhinoceros at Taubach, Germany. However, these remain isolated examples. Overall, there is little clear evidence for hunting of
megaherbivores, even in Eastern Europe, where mammoth remains are common, and relatively abundant, in Middle Palaeolithic sites (Hoffecker 2009). At least, Middle
Palaeolithic foragers would probably have scavenged the carcasses of megaherbivores when they were available, given the very high returns of such finds. At the opposite end of the scale, the use of small game by Neandertals is also a matter of debate. Taphonomic problems, including archaeological recovery methods, and a rarity of anthropogenic marks contribute to the difficulty in establishing clear evidence for the exploitation of smaller species. In considering small game hunting, Stiner and colleagues (1999; Stiner et al. 2000) usefully differentiate between slow-moving 10 terrestrial prey, quick-moving terrestrial prey, and flying prey. Exploitation of small game during the Middle Palaeolithic predominantly involves slow-moving prey, which can essentially be gathered rather than hunted (Stiner 1994). For example, 'slow-moving prey' often refers to tortoises, which were consumed by Neandertals at a number of sites in the Mediterranean region. At Moscerini Cave, Latium (Italy), aquatic and land tortoise bones are present, with suggestive impact and fracture patterns, as well as occasional burning, in four Middle Palaeolithic levels dating to over 100 kya (Stiner 1994). Middle Palaeolithic use of tortoises is also documented at Hayonim Cave, Israel, between 200 and 70 kya (Stiner et al. 2000) and at Kebara Cave, Israel, between 60 and 48 kya (Speth and Tchernov 2002). Tortoise bones showing cutmarks have been found in Middle Palaeolithic deposits, dating between 150 and 120 kya, at Bolomor Cave in eastern Spain (Blasco 2008). Thus, Middle Palaeolithic use of tortoises is consistent, but limited to the warm, coastal Mediterranean climates that probably affect the distribution of these animals.
Cutmarked remains of a cold-adapted, relatively slow species, marmot {Marmota marmota), have been observed in late Mousterian levels at Riparo Tagliente in the Italian Alps (Alhaique et al. 2004; Fiore et al. 2004). Marmot remains have also been found at other Middle Palaeolithic sites in the Italian Alps (Fiore et al. 2004), as well as Mesmaiskaya Cave, a high-elevation site in the Caucasus (Hoffecker and Cleghorn 2000), and Lezetxiki, in the Spanish Basque country (Altuna 1990). However, carnivore involvement in the accumulation of marmot remains is possible for many of these sites. As with tortoise, the distribution of marmot remains may suggest that Neandertals did exploit small, slow species in the circumscribed regions where they were available. 11
The role of small, fast-moving mammals and flying prey in Neandertal subsistence is less well-established than that of slow-moving prey. Remains of leporids, small
rodents, and birds are found at numerous sites, but it is difficult to prove that they were introduced by Neandertals rather than naturally (e.g., by carnivores, raptors, or natural presence at the site). Many analysts agree that most leporid and avian bones found in
Middle Palaeolithic sites are of non-anthropic origin (e.g., Laroulandie 2004; Pérez Ripoll 2004). Rather, habitual exploitation of small vertebrates is regularly considered to be a phenomenon of the late Upper Palaeolithic or Epipalaeolithic (Cochard and Brugal
2004; Costamagno and Laroulandie 2004; Jones 2004; Stiner 2004). However, in Portugal and Spain, leporids were exploited during the Gravettian, and possibly earlier
(Hockett and Haws 2002; Haws 2003; Manne et al 2005). Nevertheless, there exist only isolated, and sometimes questionable, examples of the consumption of small, quick prey during the Middle Palaeolithic. Cochard (2004) argues that rabbit remains at les Canalettes, in southern France,
were accumulated and consumed by Middle Palaeolithic humans. Cochard notes the presence of leporid remains at other Lower and Middle Palaeolithic sites in southern
France, but emphasizes that it is not yet clear whether Les Canalettes is an exceptional case. Digested rabbit and other small mammal remains have been found in late Middle
Palaeolithic deposits at Vanguard Cave in Gibraltar (Fernández-Jalvo and Andrews 2000), but whether these remains are correctly associated with Neandertal or carnivore activity is unclear.
Regarding birds, Laroulandie (2004) notes the presence of avian remains with cutmarks at two Middle Palaeolithic sites in France; however, these bones might have 12
been procured by scavenging. Blasco and Fernándes Peris (2009) suggest that birds were consumed during the Middle Pleistocene (MIS 6) at Bolomor Cave in eastern Spain. Bird
remains are also present in Mousterian deposits spanning from 80 to 38 kya at Grotta di Fumane, Italy (Fiore et al. 2004). These include one cutmarked golden eagle talon in a late Mousterian level, although Fiore et al. (2004) suggest that this may reflect ornamental, rather than subsistence, use. In this respect, it is interesting to note that one of the cutmarked specimens mentioned by Laroulandie (2004), from Grotte de l'Hortus, southern France, was also a golden eagle talon. At Pech de l'Azé IV, a raptor talon from the earliest Mousterian level is cutmarked in the same way as the Fumane specimen (Gaudzinski and Niven 2009). These authors suggest that birds were occasionally consumed by Neandertals at Pech de l'Azé IV.
Overall, clear indications of the consumption of small fast mammals and birds during the Middle Palaeolithic are rare. Even in exceptional cases where consumption of these animals is suggested (e.g., Blasco and Fernández Peris 2009), it is not known whether the animals were obtained through scavenging or through more active methods of hunting or trapping. At the present time, examples of leporid and bird consumption are most often restricted to the Mediterranean rim. The coherence and significance of this pattern remains to be established more firmly. Stiner's (2001:6994) proposal that "Middle Paleolithic foragers seldom bothered with small prey unless they could be obtained easily" may not be universally applicable, but, at present, has only a few exceptions. Aquatic resource exploitation by Mediterranean Neandertals has also been documented, although some of the evidence remains weak. Molluscs are the main marine resource apparently consumed by Neandertals in coastal regions. In the Balzi Rossi 13
caves, northern Italy, Mousterian marine faunas primarily include mussels and limpets, while at Moscerini Cave in west-central Italy, and at a number of sites in Puglia, southern Italy, mussels and marine clams are the dominant marine species (Stiner 1994). At Moscerini Cave, highly-fragmented burned mollusc remains were found in association with flint tools in Middle Palaeolithic levels dating between 110 and 65 kya. In contrast, shellfish in a sterile beach level at Moscerini are not burnt, are less fragmented than those in tool-bearing levels, and show evidence of wave wear not seen on shells from the tool- bearing levels. Accordingly, the Moscerini remains constitute reasonably strong evidence for collection of marine molluscs by Neandertals, even though the burning observed on the remains is not necessarily cooking-related (Stiner 1994:180-186). Mollusc exploitation may have been primarily influenced by the suitability of local marine habitats (Stiner 1994).
Consumption of molluscs during the late Middle Palaeolithic in Gibraltar has also been suggested (Barton 2000; Fernández-Jalvo and Andrews 2000). Stringer et al. (2008) recently reported the presence of marine mammals along with molluscs in Middle Palaeolithic levels at Vanguard and Gorham's Caves, Gibraltar, dating to over 41 800 and 30 000 radiocarbon years BP, respectively. They suggest that, at Vanguard Cave, monk seals were hunted when on the beach during their breeding period, while beached dolphins may have been scavenged. Stiner (1994) also noted the presence of monk seal at the Mousterian site of Sant'Agostino in Italy. Stringer et al. (2008) argue that Gibraltar Neandertals regularly and deliberately exploited coastal resources in a manner requiring knowledge of the seasonal availability of these resources. However, according to Klein and Steele (2008), the samples of molluscs and marine mammals from Vanguard and 14
Gorham's Cave are too small to support the arguments presented by Stringer and colleagues. The use offish by Neandertals appears to have been very limited. Although Stringer et al. (2008) suggest that Neandertals consumed fish at Gorham's Cave, they do not discuss the evidence in detail. Le Gall (2000) contends that late Neandertals occasionally practiced freshwater fishing on a very small scale, with limited technological investment. Analyses of carbon and nitrogen stable isotopes from Neandertal skeletal remains confirms that herbivore meat was the predominant source of
their dietary protein (Bocherens et al. 1991; Richards et al. 2000; Richards et al. 2001). Overall, small animals and marine resources appear to have occasionally
supplemented rather than supplanted reliance on ungulate prey during the Middle Palaeolithic. This stands in contrast to some Upper Palaeolithic, especially Magdalenian and Epipalaeolithic sites, where small taxa sometimes constitute a majority of remains (e.g., Costamagno and Laroulandie 2004; Jones 2006). Some authors have referred to this
apparent dependence on a broader range of resources by modern humans as an extension
of Flannery's Broad-Spectrum Revolution (e.g., Stiner et al. 2000; Richards et al. 2001; Stiner 2001), and imply that modern humans were either capable of exploiting a greater range of resources than Neandertals, or that higher population densities forced them to use costlier, or lower-ranked, resources. This suggestion logically leads to a consideration of whether Neandertals exploited animal resources differently than modern humans.
A Neandertal niche?
Binford's (1985) argument that Neandertals were obligate scavengers was an early proposal of a specific Neandertal niche that proved to be inaccurate. However, as 15
pointed out by Bar-Yosef (2004), past analyses of Neandertal ecology have had a tendency to homogenize Neandertal behaviour during the entire Middle Palaeolithic, instead of recognizing the potential for variability during this extended period. Burke (2000) has posited that the Neandertal niche is variable and complex, like that of modern humans. Indeed, Grayson and Delpech (2003, 2008; Grayson et al. 2001) argue that differences in prey abundances between the Mousterian, Châtelperronian, and Aurignacian assemblages at the French sites of Grotte XVI and Roc de Combe can be entirely explained by climate change, rather than by differences in Neandertal and modern human foraging practices. Morin (2004, 2008) makes a similar argument for Middle to Upper Palaeolithic transition assemblages from Saint-Césaire, in west-central France. Beyond Western Europe, Hoffecker and Cleghorn (2000) argue that, in the Caucasus, the Neandertal niche overlapped substantially with that of Upper Palaeolithic modern humans.
Stable isotope evidence provides an additional source of information on the Neandertal niche. As mentioned earlier, herbivores provided nearly all of the protein in European Neandertal diets (Bocherens et al. 1991; Richards et al. 2000; Richards et al. 2001). Richards and colleages (2001) suggest that the absence of an aquatic isotopie signature among Neandertals implies that these archaic populations were unable to exploit freshwater aquatic resources. These authors (see also Richards and Trinkaus 2009) argue that, unlike Neandertals, early modern humans in Europe had broader, more varied diets, including freshwater and marine resources. However, only one modern human on which stable isotope analyses have been carried out is roughly contemporary with Neandertals (Peçstera cu Oase; Richards and Trinkaus 2009), and this skeleton is 16
from southeastern Europe, while most Neandertal values are from Western European
specimens. In view of this small sample size (and difference in location), it is relevant that stable isotope differences between Neandertals and mid-Upper Palaeolithic modern humans in Western Europe are not statistically significant (Lee-Thorp and Sponheimer 2006; Lee-Thorp 2008). Somewhat paradoxically, however, both a broad-spectrum diet and 'specialized' hunting (a pattern characterized by a focus on a single species) have been interpreted at times as evidence for modern behaviour, implying that the failure ofNeandertals to either regularly exploit a wide diversity of species or to focus narrowly on one preferred prey constitutes evidence for adaptive inferiority. For instance, Mellars (1973, 1996, 1998) has argued that the preponderance of reindeer at Upper Palaeolithic sites in France reflects "large-scale co-operative hunting strategies" unique to modern humans (Mellars 1973:271). However, specialization on a single species would not be an adaptive behaviour if other, similarly profitable, prey species were available. The so-called specialization observed in many Upper Palaeolithic assemblages is more likely related to a climatic downturn coincident with the Aurignacian, which probably increased the abundance of reindeer while decreasing the diversity of available prey species (Grayson et al. 2001; Grayson and Delpech 2003; Morin 2004, 2008; Faith 2007). Climate-induced change cannot be used as a criterion for differentiating Neandertal and modern human foraging adaptations. Thus, both zooarchaeological and biogeochemical data point to general similarities in Neandertal and early Upper Palaeolithic modern human diets. For the moment, it appears that the Neandertal and modern human niches were broadly similar, 17
as both groups exploited many of the same resources in similar relative abundances
(Grayson et al. 2001; Grayson and Delpech 2003, 2008; Morin 2004, 2008; Faith 2007). Consequently, Neandertals and modern humans would probably have competed for the limited resources available to them in the late Pleistocene environment (O'Connell 2006).
Ecology and the replacement or continuity debate
The question of competition between Neandertals and modern humans is one that has direct relevance for understanding the Middle to Upper Palaeolithic transition in Europe. Two theories dominate the transition debate today: the replacement and continuity models. The replacement model holds that fully modern humans originated in Africa, and subsequently spread throughout the world, replacing any pre-existing hominid populations (e.g., Cann et al. 1987; Stringer and Andrews 1988; Stringer and Gamble 1993; Mellars 2005, 2006). In contrast, the multiregional model holds that pre- existing archaic populations throughout the Old World evolved into modern humans through gene flow between regional populations (e.g., Brose and Wolpoff 1971; Thorne and Wolpoff 1992; Wolpoff et al. 2001). In this model, archaic human populations, including Neandertals, are considered to have been part of a polytypic species engaged in a reticulate network of genetic exchange. Thus, local archaic populations evolved into modern form through selective pressures that favoured the spread of adaptive traits, which could arise in any regional population. At the same time, geographic differentiation—such as that characteristic ofNeandertals—could develop through the spread of selectively neutral traits in small regional populations (Wolpoff et al. 2000). Compromise scenarios, such as the 'hybridization' or 'assimilation' models generally argue that modern humans first originated in Africa and then expanded and interbred with 18
regional archaic populations. Consequently, they can essentially be viewed as variants of the replacement model, since these models also postulate a significant emigration of modern humans from Africa into Europe (e.g., Duarte et al. 1999; Smith et al. 2005;
Zilhäo 2006; Trinkaus 2007). Because of the niche overlap between late Neandertals and modern humans, if
and when these populations co-existed, as implied by the replacement model, resource competition would, in all likelihood, have impacted foraging behaviours. O'Connell (2006) and Morin (2004, 2008, 2010) have discussed this issue in depth and suggested
possible foraging implications of both the replacement and continuity models. A presentation of their arguments is left for a later chapter (Chapter 7). However, it should
be emphasized that a thorough understanding oflocalforaging behaviours during the final Mousterian is imperative as a startingpointfor understanding regional changes during the Neandertal to modern human transition.
The research problem
In light of the preceding discussion, there are two main issues that an analysis of the Mousterian fauna from Gatzarria Cave could address. First of all, the Gatzarria
assemblage could contribute substantial insight into late Neandertal foraging behaviour in the Atlantic Pyrenees, a region where, as in much of Iberia, Italy, and Greece, faunal remains have, until quite recently, been under-studied and more significantly under- published (Jaubert and Bismuth 1996). This study will contribute to the understanding of broader regional patterns in Neandertal diet breadth during the late Middle Palaeolithic. Secondly, by providing a baseline for assessing change during the Middle to Upper Palaeolithic transition in the Atlantic Pyrenees, an analysis of the Gatzarria Mousterian 19
faunal remains will contribute to the understanding of transition processes in the region. Accordingly, while addressing the general research objectives identified in Chapter 1, this study also focuses on answering the following more detailed questions about Neandertal ecology: (1) What animal resource use patterns characterized late Neandertal subsistence in the Atlantic Pyrenees? This question is addressed through an examination of foraging behaviour at Gatzarria, with regards to diet breadth, patch use, seasonality, transport of skeletal parts, and utilization of those parts, particularly marrow- bearing elements. (2) How might the early Upper Palaeolithic faunal assemblages at Gatzarria be expected to differ from the late Mousterian pattern, as a result of possible climate change or competition between Neandertals and modern humaris? Gatzarria Cave, the site that is the focus of this analysis and provides the necessary materials for answering these questions, will now be presented. 20
Chapter 3: Gatzarria Cave—Description and Previous Research
Gatzarria is a karstic cave in the Atlantic Pyrenees, southwestern France, that contains an archaeological sequence documenting the Middle to Upper Palaeolithic transition. This chapter first summarizes the geographic context, the history and method of the archaeological excavations, and stratigraphy of the site, as well as prior archaeological research carried out on the excavated materials. Subsequently, the palaeoenvironmental context of the site is presented, along with a review of the ecology of ungulate species likely to have been exploited by the occupants of Gatzarria.
Geographic Context Gatzarria is located at an elevation of 290 m on Mont Hargagne, a mountain rising to 855 m above sea level in the eastern part of the Arbailles massif (Figure 3.1). The closest modern centre is the village of Suhare, in the commune d'Ossas-Suhare, French department of the Pyrénées Atlantiques. Gatzarria was long used by the local Basque people as a refuge for livestock (Saenz de Buruaga 1991). The cave also figures in Basque mythology, as it was considered to have been the home of laminak—blond female génies with goosefeet (Saenz de Buruaga 1991; Laplace and Saenz de Buruaga 2003). The two major published works concerning Gatzarria are Saenz de Buruaga (1991) and Laplace and Saenz de Buruaga (2003). The following discussion of the site is based on these works, with additional references cited where applicable. 21
> y'
.-j^aç-i» ... r r.~— ¦—¦ VW-s : · : <-
^ 1 ^s Z*^-"-
-~;?
¡ì m-m "*> S ' ì ??
O 100 200 400 Kilometers I ? ? ? I ? ? ? I
Figure 3.1: Location of Gatzarria Cave.
Gatzarria opens towards the northeast out of a small cliff of Urgonian limestone, in a zone where the limestone contacts Albian schists (Laplace 1966a). The cave itself is oriented southwest to northeast. It is approximately 13 metres long, with a width of five metres at the entrance, and a maximum height of two metres, also near the entrance. The surface of the cave inclines gently downwards towards the rear, where the slope rapidly accelerates towards a small tunnel on the right-hand side. A number of archaeological sites are located on Mont Hargagne, including Etxeberri and Sasiziloaga Caves, which contain Magdalenian parietal art, and Harregui Cave, which contains Mousterian and Solutrean deposits (Laplace 1966a). Because 22
Gatzarria is relatively well-connected with nearby valleys, Sáenz de Büruaga (1991:79- 80) suggests that Gatzarria was an ideal site location during much of the Palaeolithic and, furthermore, was probably never an isolated habitation site.
Excavation history and method What might be considered the first excavation at Gatzarria occurred sometime in the late 1940s, when a group of youngsters tried, but failed, to clear the opening of the small tunnel at the rear of the cave in an attempt to discover additional galleries. This activity significantly disturbed the archaeological layers in this portion of the cave. It was this disturbance that led, in part, to P. Boucher and P. Bouillon's discovery of Aurignacian and Mousterian artifacts in the cave in 1950 (Laplace 1966a; Sáenz de Buruaga 1991:82). In 1951, Boucher excavated a preliminary test pit in unit 8D, under the direction of L. Méroc. However, Boucher did not record the stratigraphie provenience of the finds from this excavation, which led G. Laplace and F. Bordes, along with
Boucher, to excavate additional test pits in units IE and IF during 1952-1953. Following a number of problems with the property owner which delayed additional excavations, work was resumed in 1956-57, under the sole direction of Laplace.1 At this time,
Boucher excavated units 7D, 8D, 9D, and a small portion of 10D. Excavations were halted for the following two years, but resumed again in 1960. The cave surfaces were cleaned in preparation for additional excavation in 1960, and an extended excavation program was carried out under Laplace from 1961-1976, during which time a large part of the cave area was excavated.
In 1987, the materials from excavations prior to 1960 were in the possession of P. Boucher at his home in Mauléon (Sáenz de Buruaga 1991:84, footnote 15). These materials are not included in this study. 23
Laplace's excavation method was highly innovative for the time, and more importantly, meticulous. Saenz de Buruaga (1991:92) refers to the fourteen years required to excavate Gatzarria as proof of Laplace's painstaking care in excavation and artifact recovery. Artifacts, including bone, were piece-plotted, and material recovered from sieved sediments was collected. Long bone shaft fragments at Gatzarria were not subject to selective recovery, in contrast with many other excavations carried out during and even after the 1960s (see Turner 1989; Grayson and Delpech 2008). The careful recovery of the Gatzarria bone assemblage makes it well suited for a detailed zooarchaeological study. The site was excavated following the 'Laplace-Méroc methodology' (Laplace 1966a). A datum point was established, the site divided into a one-metre grid, and each unit given an alphanumeric coordinate. This involved seven frontal 'sectors' (A-G) extending across the width of the cave, and 15 sagittal 'sections' extending across its length (Figure 3.2). Sections 1-12 are within the cave, increasing towards the rear, and sections 21-23 lie at the cave entrance, increasing away from the cave opening. Each unit was further divided into nine 3 3-cm2 sub-units, labeled 1-9 as shown in Figure 3.3.
Excavation proceeded by 5 or 10 centimetre spits {décapages) of homogeneous sediments. Décapage depth was conditioned by geological and archaeological stratigraphie sub-divisions. Depth below datum (z), as well as frontal (x) and sagittal (y) coordinates, were recorded for significant artifacts (such as retouched tools and large teeth or bone specimens) found during excavation (Figure 3.3). 24
Sectors
Front of Cave
Figure 3.2: Layout of Gatzarria Cave and excavation grid. Modified from Saenz de Buruaga (1991:93, Figure 14). Each square = 1 m2.
Back of cave
X (frontal) Front of cave
Figure 3.3: Excavation sub-units and axes of coordinates for plotted artifacts. Based on G. Laplace's field notebooks. 25
Frontal coordinates refer to the position in the unit along the breadth of the cave, while sagittal coordinates refer to position along the length of the cave.2 Artifacts not plotted during excavation and recovered in sieves were recorded to unit, sub-unit, and
décapage. Sieving of sediments in water, washing and marking of plotted artifacts, and analysis of thefraction grossière of the sediments, stratigraphie profiles, and artifact maps generally took place during the course of excavations. In total, 50 units were excavated, of which 33 were dug over their entire surface; however, not all layers were reached in every unit.
Stratigraphy
The published stratigraphie sequence for Gatzarria is based on both sedimentary and archaeological evidence (Laplace 1966a; Saenz de Buruaga 1991). Unfortunately, no Chronometrie dating of the site has been carried out to date, although dating of the Proto- Aurignacian and Cjr Mousterian layers is currently being undertaken by Carolyn Szmidt. A theoretical stratigraphie profile is shown in Figure 3.4. The sediments at Gatzarria are divided into three groups {ensembles), primarily distinguished by sediment colour, each of which contains more than one archaeological and/or sedimentary layer. These are described here from bottom to top of the sequence (after Saenz de Buruaga 1991 :96-97, 109-181; Laplace and Saenz de Buruaga 2003:84-86), first focusing on sedimentary divisions, and secondly, on archaeological ones, as these divisions do not always coincide.
Artifacts were recorded and labeled as follows: Site.UnitiSubUnitìDécapage.Layer.Depth.Frontal. Sagittal. Measurements were recorded in centimetres. Underlining or superlining of the décapage indicates the bottom or top half of a 10 cm décapage. For example: G.7C(4)7.Cb.69.27.49. 26
Cbn Cbcs 000^0° Cbr c^> vO \ Cb ¿~r> Eb
Cbf Cbci
Cjf
Cj Ej
Cj Moust Cjm
Cjr EJr I
Figure 3.4: Theoretical stratigraphie profile at Gatzarria, modified from Laplace (1966a: 11 9, Figure 2). Stratigraphie units are described in the text. Note that lower layers in Ejr are not shown in this figure.
Geological stratigraphy
The lowest sedimentary group, the yellow-red group or Ensemblejaune-rouge (Ejr), contains two layers that are frequently indurated: 1) The lower part of Ejr is a red clay layer, the Couche rouge (Cr), which contains occasional concentrations of manganese, designated Couche rouge à manganèse (Crm). In places, the manganese lenses are associated with high
concentrations of bone, designated Couche rouge à manganèse et os (Crmo). Cr rests on the stalagmitic substrate of the cave floor. 27
2) The upper part of Ejr, the Couchejaune-rouge (Cjr), is characterized by yellow and red clay sediments, interspersed with concentrations of manganese (Couchejaune-rouge à manganèse; Cjrm) or of grey-hued clays (Couche gris-rouge; Cgr). In some places, the grey clay also contains traces ofmanganese, in which case, the contents are indicated as Couche gris-rouge à manganèse (Cgrm). These grey clay lenses are intercalated in the upper and middle parts of layer Cjr. The intermediate yellow group, the Ensemblejaune (Ej), contains yellowish sediments designated as the Couchejaune (Cj). In the lower part of the group, Cj is characterized by clayey-sandy yellow sediment, with patches of manganese labelled Couchejaune à manganèse (Cjm or Cjmg). In the upper part of Ej, Cj is characterized by a generally homogeneous yellow silty sediment with some small fragments of shale and relatively few large clasts. Hearth traces are also present in the upper part of Cj, and are designated as the Couchejaune defoyers (Cjf).
Finally, the upper brown group, the Ensemble brun (Eb), contains brown sediments divided into four main units:
1) The lowest level in Eb, the Couche brune à cailloutis inférieur (Cbci), contains brown clay-sand sediments with an importantfraction grossière and numerous cryoclastic components. Similar sediments containing hearth traces heterogeneously distributed in the superior and inferior portions of Cbci are referred to as the Couche brune defoyers (Cbf). Often Cbci and Cbf were not
distinguished and were simply designated as Cbci-Cbf. 28
2) Above Cbci-Cbf, the Couche brune (Cb) also consists of brown clay-sand sediments, but with finer particles, mainly 2-5 mm in diameter. Some small to medium-sized angular cryoclasts and larger cryogenic blocks are also
present. 3) The penultimate layer, the Couche brune à cailloutis supérieur (Cbcs), contains brown clay-sand sediments with angular cryoclasts larger than those in Cb, as well as limited cryogenic blocks. This layer was disturbed during the (historic) reoccupation of the cave and is present only in certain
undisturbed lenses.
4) The superficial layer, consisting of dark brown humic sediments, is referred to as the Couche brune noirâtre (Cbn), or where clear evidence of recent
disturbance is present, as the Couche brune remaniée (Cbr). However, as with Cbci-Cbf, these components were often not distinguished and were simply designated as Cbn-Cbr.
Archaeological stratigraphy
Although the sedimentary sequence detailed above generally corresponds closely with the archaeological levels, this does not hold for all layers, especially those in Ej.3 The lower group, Ejr, contains two separate Mousterian industries:
1) The lower, red clay levels of Ejr (Cr, Crm, and Crmo) contain a limited Mousterian industry, which provided insufficient lithic material for
typological study—only four tools according to Laplace and Saenz de
3 The lithic industry attributions referred to here are those identified by Laplace (as described by Saenz de Buruaga 1991 and Laplace and Saenz de Buruaga 2003), and referred to in publications on Gatzarria until the present time. It is fully recognized that these may be subject to some refinement and change when new stratigraphie and lithic studies are carried out, hopefully in the near future. 29
Buruaga (2003). However, Sáenz de Buruaga (1991:97) refers to it as a Mousterian/Tayacian industry. 2) The upper portion of Ejr, consisting of Cjr, Cjrm, Cgr, and Cgrm, contains a Mousterian industry with cleavers (hachereaux) on flakes, heavily dominated by carenoid sidescrapers and denticulates, primarily made of quartz and quartzite. Despite the sedimentary homogeneity of Ej, several archaeological industries are present: 1) The lower part oísedimentary level Cj, as well as the manganese lenses (Cjm and Cjmg), contain a Mousterian industry with tools based on troncatures, beaks, and burins. It may be a Denticulate Mousterian (see Laplace 1966a: 124, but note that this suggestion refers to a sample consisting of both Mousterian and Châtelperronian elements). This is archaeological level Cj. Quartzite is the most important raw material in Cj, but quartz is less common than in Cjr, with flint and lydite being relatively more important. 2) Above this Mousterian industry is a Châtelperronian level referred to as Couchejaune niveau 3 (Cjn3). The sediment of this layer is essentially indistinguishable from that of the underlying archaeological layer Cj. Cjn3 and the Cj Mousterian were mostly distinguished a posteriori using archaeological data. Laplace considered a green-white quartzite and "blackish flint" to be materials that distinguished the Châtelperronian from the Cj Mousterian, although these may be problematic criteria. 30
3) Superimposing the Châtelperronian deposit is the Couchejaune niveau 2 (Cjn2), also sedimentologically similar to Cj and Cjn3. According to Laplace (1966a), it contains a Proto-Aurignacian industry, predominantly ofblackish flint, with marginally backed blades and cylindro-conical bone points. 4) The uppermost archaeological level within Ej is Couchejaune niveau 1 (Cjnl), which according to Saenz de Buruaga (1991) contains a Proto- Aurignacian industry with carenoid scrapers and flat bone points. Laplace (1966a) suggests that the lithic industry in this level is generally made of a whitish flint. Cjnl is distinguished from Cjn2 by Cjf, the non-homogeneous level of hearths in Cj. Cjf forms the inferior portion of Cjnl. In comparison to the middle group (Ej), archaeological levels in the upper Ensemble brun correspond well with the sedimentological divisions: 1) Cbci-Cbf corresponds to an Early Aurignacian with split-base bone points. 2) Cb contains an Evolved Aurignacian industry. 3) Cbcs contains a Gravettian industry with Noailles burins. 4) Finally, Cbn-Cbr contains evidence of historic activity. Correspondences between the sedimentary levels, the archaeostratigraphy, and the lithic industries are summarized in Table 3. 1 . It should be noted that some descriptions of the site stratigraphy (e.g., Leveque and Miskovsky 1996) make no mention of the Cj Mousterian level. Laplace (1966a) noted that the 'Châtelperronian' level Cjn3 contained artifacts primarily Mousterian in nature, but argued that there was no definitive evidence for the presence of two separate industries. It appears that during later years, the presence of separate Mousterian and Châtelperronian industries in the lower part of Ej was 31
. Sedimentary layers Archaeostratigraphy Time Period Cbn-Cbr Cbn-Cbr Historic Cbcs Cbcs Gravettian Cb Cb Evolved Aurignacian Cbci-Cbf Cbci-Cbf Early Aurignacian Cj, Cjf Cjnl Proto-Aurignacian Cj Cjn2 Proto-Aurignacian Cj Cjn3 Châtelperronian Cj, Cjm/Cjmg Cj Mousterian Cjr, Cjrm, Cgr, Cgrm Cjr Mousterian (Vasconian) Cr, Crm, Crmo Cr Mousterian/Tayacian?
Table 3.1: Sedimentary and archaeological stratigraphy at Gatzarria and corresponding lithic industries. recognized: Lavaud (.1980), Saenz de Buruaga (1991), and Laplace and Saenz de Buruaga (2003) all make clear reference to the presence of both industries in separate levels. The omission of the Cj Mousterian by Leveque and Miskovsky (1996) is, in all likelihood, because the stratigraphy presented in their study was based on Lévêque's (1966) work.
Stratigraphie problems The archaeological sequence at Gatzarria detailed above is not without problems. Relatively recent disturbances in the cave include a variety of animal occupations, a good number of which appear to have been badgers denning in the rear of the cave (Saenz de Buruaga 1991:98). In certain areas (particularly units in section 10), these occupations resulted in disturbance of the archaeological layers. The attempt to clear the opening of the tunnel at the rear of the cave in the 1940s also contributed to stratigraphie disturbance in units 1 IF-G and 12F-G, although these units were not a focus of Laplace's excavations. However, the impact of local peoples on the sediments was not limited to this incident. The cave was used at times as a shèepfold, and the presence of livestock in the cave structurally altered the sediments by trampling. This alteration affects the 32 superficial layer Cbn-Cbr, as well as a large part of the Gravettian layer (Cbcs), and some resultant mixing of Cbcs and Cb (the Evolved Aurignacian) cannot be excluded (Saenz de Buruaga 1991:99). Human activity around the cave also resulted in the removal of blocks from the cave surface to the exterior for use in construction (Saenz de Buraga 1991).
The lack of sterile layers between occupations, the variability in the thickness and distribution of the layers, and the homogeneity of the sediments posed problems for distinguishing the archaeological levels. First of all, because the transition between Cb and Cbci-Cbf was not always distinct, artifacts found in areas where these layers could not be clearly isolated were assigned to 'Cb-Cbci-Cbf.' The problem of distinguishing layers in the sedimentologically homogeneous Ej was more serious. Although Cjnl was distinguished by the presence of hearth traces (Cjf), the archaeological levels Cjn2, Cjn3, and Cj Mousterian were less easily differentiated. According to Saenz de Buruaga (1991), water run-off in the cave between the Mousterian (Cj) and Châtelperronian (Cjn3) occupations led to the formation of a depression in the centre of the cave such that, near the cave walls, the Mousterian deposits are closer to datum level (i.e., are higher) than the early Upper Palaeolithic deposits, which fill the depression. However, this phenomenon is not apparent in the units considered in this study (see Chapter 5). Laplace attempted to distinguish Cjn2, Cjn3, and the Cj Mousterian subsequent to excavation by plotting the distribution of Châtelperronian points. These points were found to follow the natural profile of the cave floor, and therefore, this distribution was extrapolated and used to define the theoretical limits of Cjn3. The presence of bone retouchers (or retouchoirs, ad hoc tools likely used for retouching stone tools) was also 33
used to delimit the boundary between the Châtelperronian and Mousterian, although this was a questionable criterion as these tools are known to be present in both Mousterian and early Upper Palaeolithic assemblages (Chase 1990). Laplace also identified 'intrusive' Mousterian components in décapages assigned to the Upper Palaeolithic on the basis of morpho-technology and raw material. Quartzite, Iydite, shale, and ophite artifacts were considered as belonging to the Mousterian, while a white-green quartzite and black flint were judged to be characteristic of the Châtelperronian. As we will see, these associations are highly dubious. The possible mixing of Middle and Upper Palaeolithic artifacts in Cj, and in particular, the reassignment of lithic artifacts to different archaeological layers subsequent to excavations posed a significant challenge for this research. Further discussion of these problems, and the resolution adopted for this study, are presented in Chapter 5.
Previous archaeological research
Lithic industries
The Mousterian lithic industries from Gatzarria have been studied by Laplace and Saenz de Buruaga (2003), and the Upper Palaeolithic stone industries have been examined by Laplace (1966a) and Saenz de Buruaga (1991) using Laplace's analytical typology approach. Both Laplace and Saenz de Buruaga emphasized the evolutionary potential of, and continuity between, the lithic industries at Gatzarria, which fit into Laplace's theories of leptolithization and of the Aurignacian synthétotype. These theories essentially envision the development of Upper Palaeolithic technology from the Mousterian through the invention of blade reduction sequences. The Cjr Mousterian lithics from Gatzarria are currently the subject of a doctoral dissertation by Marianne Deschamps (Université de Toulouse, France). Hopefully, future studies will not only improve our understanding of the Gatzarria lithics, which at present is limited (see Tartar 2009), but also help to resolve some of the ambiguities in the stratigraphie sequence.
Bone industry and ornaments The bone industries in the Upper Palaeolithic assemblages at Gatzarria consist predominantly ofbone points in the Aurignacian levels, along with polishers, awls, and retouchers. Retouchers are also present in the Mousterian levels. Although the bone industries of all the Upper Palaeolithic levels have been studied by Saenz de Buruaga (1991), Tartar (2009) reports that a large portion of the tools identified in the Early
Aurignacian layer (Cbci-Cbf) by Saenz de Buruaga may be pseudo-tools. Consequently, his bone tool counts are not repeated here. Tartar (2009) recently undertook a study of the Early Aurignacian bone industry.
Tartar examined bone tools identified by Laplace and Saenz de Buruaga, as well as tools she identified in a sub-sample of the unstudied faunal remains from Cbci-Cbf. Tartar studied a total of 166 bone implements, including 14 polishers and 5 polisher preforms, mainly on ribs of large animals (i.e., large bovines such as bison), which were possibly used in skin preparation, and 1 1 awls, also mostly on rib fragments. Tartar suggested that awls at Gatzarria were made mainly on rib fragments because horse remains are infrequent at the site. According to Tartar, horse metapodials are frequently transformed into awls at other Early Aurignacian sites. Retouchers were the most abundant bone tool type (n = 92) identified by Tartar. Such tools are distinguished by discrete zones of L-shaped indentations occurring at right angles to the bone surface, slightly inset from the extremities of a bone fragment, 35
suggesting that a sharp edge was driven into the bone surface (Chase 1990). Long bone shaft fragments of large animals (again, probably mostly large bovines) appear to have being preferred for use as retouchers at Gatzarria. These were probably collected from available animal remains and then discarded (Tartar 2009). Seven of the retouchers were found on bone fragments that had also been used for other purposes, such as polishers. Tartar identified 65 of what she termed 'intermediary tools:' bone specimens showing evidence of use, such as blunting and crushing of edges and negative flake scars, but that were otherwise un-worked. However, possible taphonomic explanations for the appearance of these specimens, such as trampling shortly after deposition, were not considered. Finally, although Tartar's study focused on bone tool production, she identified 64 worked pieces of cervid antler in the Gatzarria Early Aurignacian, 42 of which are split-base bone points or point fragments. Tartar argues that the selection ofbone fragments for tool use at Early Aurignacian sites such as Gatzarria was conditioned by the technical possibilities offered by bone morphology, as well as opportunism related to the availability of different bone types. She suggests that although bone fragments for tool use were probably collected from alimentary refuse, bone raw material cannot be considered an unlimited resource, as bone elements preferred in tool making were limited to certain species, certain skeletal elements, and were of optimal condition for tool use during a limited time frame. Tool production was therefore conditioned by the availability of animals with regards to seasonal movement, gregariousness, and condition (e.g., seasonal absence/presence of antlers). 36
Descriptions of ornamental objects in the Upper Palaeolithic levels at Gatzarria are provided in Saenz de Buruaga (1991). The lower Proto-Aurignacian level (Cjn2) yielded one large cylindrical fragment of worked amber. The upper Proto-Aurignacian (Cjnl) contained a small number ofpendants made of worked bone and perforated teeth, an antler or horn fragment with linear incisions, and two bone fragments with light linear and curvilinear traces. The Early Aurignacian layer (Cbci-Cbf) contained 12 perforated teeth, mostly fox (Granger and Leveque 1997) and pearls made of steatite and amber (Saenz de Buruaga 1991). Decorated bird long bone tubes, several bones with linear and/or curvilinear incisions, and one perforated salmonid vertebra were also recovered from the Early Aurignacian deposit. Tartar (2009; citing personal communication with R. White) reports that, besides the pieces identified by Saenz de Buruaga, the Early
Aurignacian contains additional ivory beads at different stages of manufacture. A small number of decorated bird bone tubes, as well as three perforated teeth, an antler pendant, and incised bone and antler fragments were found in the Evolved Aurignacian layer.
Human Remains
Human remains at Gatzarria are limited to one molar, probably an upper second molar, from the Early Aurignacian layer (Saenz de Buruaga 1991). Morphological and, hopefully, DNA analysis of this tooth will be carried out in the near future.
Palaeontological work Preliminary palaeontological work on the Gatzarria fauna was carried out by Leveque (1966) and Chaline (micromammals only; 1972), however, these studies were completed prior to the end of Laplace's excavations at the site, and are consequently of limited use. Lavaud (1980) carried out a more extensive study of the faunal remains from 37
Gatzarria, studying all the archaeological levels except the Gravettian layer, as part of her
doctoral dissertation. Lavaud used her faunal data to discuss climate in southern France
during the interpleniglacial. Lavaud's sample of large mammalian taxa is detailed in Table 3.2. According to her, the main herbivore species throughout the Gatzarria sequence is red deer (Cervus elaphus), while large bovids {Bos/Bison) and chamois {Rupicapra rupicapra) are of
secondary importance. Horse (Equus caballus) was rare, but nonetheless present throughout the sequence. Ibex {Capra ibex) was present only from the Proto-Aurignacian onwards, while reindeer {Rangifer tarandus) was nearly absent, even in the Aurignacian layers. Lavaud observed that the Gatzarria faunal remains contained a large proportion of carnivores, especially fox {Vulpes vulpes), but was unable to conclude if the fox remains at the site were due to human activities or natural causes such as denning or prédation by other carnivores. Hyaena {Crocuta crocuta) and cave bear {Ursus spelaeus) remains were present in most of the layers. Remains of cave lion {Panthern leo), giant deer {Megaloceros giganteus), woolly mammoth {Mammuthus primigenius), and woolly rhinoceros {Coelodonta antiquitatis) were more sporadic. Other taxonomic groups rarely present at Gatzarria include lagomorphs, mustelids {Mustela nivalis), moles, shrews, bats, and amphibians. Lavaud suggested that all the small taxa accumulated naturally. Eurasian water vole {Arvicola terrestris) is the only micromammal present nearly continuously throughout the sequence. Meadow voles {Microtus arvalis, M. agrestis, and M. nivalis), southwestern water vole {Arvicola sapidus), and long-tailed field mouse {Apodemus sylvaticus) appear most consistently in the Early and Evolved Aurignacian layers. Over fifteen bird species (listed in Table 3.3) 38
/i
é
Level V?f$ Table 3.2: Large mammal specimens (NISP) identified by Lavaud (1980). Small mammals Birds /¦$ $ ¦$ £ ¦$ ¿ '//////ffffMi/ftYf/f/ffW (g/ Level Cb XXXXX X X XXY Y YY ?YY? Cbf X X Cjnl Cjn2 Cjn3 Cj Cj (Moust.) Cjm Cjr Cjrm Crmo Table 3.3: Microfaunal, avian, and amphibian remains identified by Lavaud (1980). Presence/absence only. were present at the site, representing a variety of ecological habitats. Notably, bird remains were virtually absent from Lavaud's samples prior to the second Proto- Aurignacian level (Cjnl). Lavaud's interpretations of the Gatzarria faunal data were limited to general and qualitative suggestions regarding climate change during the late Pleistocene. She suggested that during the Palaeolithic, Gatzarria was a permanently humid location, with a nearby running water source. Local habitats included rocky terrain, wooded, and open areas. Overall, Lavaud suggested that the climate at Gatzarria was moderately cool, and that the area may have served as a refugium for temperate species such as red deer during cold periods. This suggestion was supported by the near absence of reindeer at the site, especially during the Aurignacian. Comparison of the Gatzarria fauna to species lists from the nearby site oflsturitz suggested that the faunal spectrum present at Gatzarria was strongly conditioned by local ecology. Lavaud concluded that the climate near Gatzarria was relatively temperate during the deposition of the Mousterian layers, but deteriorated through the Châtelperronian until the Early Aurignacian, with a slight improvement during the Evolved Aurignacian. Lavaud's study appears to have been hindered by small sample sizes, probably due to the fact that her study focused on piece-plotted faunal remains, mostly teeth. Furthermore, Lavaud's interpretations were exclusively climatological, and did not consider taphonomic processes or foraging behaviours. The present study of the Gatzarria faunal remains, focusing on taphonomy and subsistence behaviours, will contribute to filling this gap. 40 Ecological context The ecological context of Gatzarria, alluded to in the previous paragraphs, will now be discussed in more detail. Both the climate experienced by the Palaeolithic occupants of Gatzarria, and the ecology of locally present ungulate species are considered. Palaeoenvironment The Pleistocene glacial history of the Pyrenees is not identical to that of the Alps and other more northerly regions of Europe (Gonzalez-Sampériz et al. 2006). The recent consensus view of the timing of the last déglaciation in the Pyrenees is that the Maximum Ice Extent (MIE) in the region occurred sometime before 45 000 BP (Hérail et al. 1986; Calvet 2004; Delmas 2005), with the onset of glacial retreat by at least 38 000 BP (Jalut et al. 1992). Pollen sequences studied by Jalut et al. (1992) and Gonzalez-Sampériz et al. (2006) both suggest that the climate in the Pyrenees during the last glacial cycle was dry and relatively cold, leading to a steppe vegetation of Poaceae, Artemisia, and Chenopodiaceae, with Pinus constituting the main arboreal element, until roughly 15 000 BP when the transition to a woodland environment began. Pleistocene glaciations in the Pyrenees were sharply affected by local relief and by the important influences of the Atlantic in the west and the Mediterranean in the east parts of the range (Calvet 2004; Delmas et al. 2008). The importance of topography in conditioning the local climates provided some stability in climatic conditions and circulation patterns in the region. In the western part of the Pyrenees, the influence of the Atlantic contributes to a more humid and temperate climate than elsewhere in the range (Jalut et al. 1992; Calvet 2004). An increase in precipitation associated with the humid Atlantic air is most pronounced on the northern slopes of the chain. In the pre-Pyrenean 41 massifs, glaciations were more severe than in the western Basque country, where glaciation was limited to isolated cirques (Calvet 2004). During the last MIE, which, as mentioned, may have occurred sometime before 45 000 BP, glacier tongues in the northern massifs descended from high peaks down into major valleys, sometimes down to 350-400 m. At this time, true piedmont glacial lobes were present at Lourdes and Arudy, 30-70 km west of Gatzarria, while southwest of the site, an ice field was present on the Arres d'Anie plateau (Calvet 2004). This might suggest relatively severe climatic conditions at Gatzarria during the final Mousterian and early Upper Palaeolithic. However, due to the variability in the extent and timing of glaciation within the Pyrenees, an independent monitor of the local climate at Gatzarria will be critical to a successful interpretation of subsistence changes among hominins at the site. Given pollen profiles indicating a steppe and pine forest environment during the last glaciation in the Pyrenees (Gonzalez-Sampériz et al. 2006; Jalut et al. 1992), plant foods would likely have been uncommon, at least during the cold season. Ethnographic data suggest that, besides a few seasonally collected items, such as berries, plant food intake is minimal among subarctic and boreal forest foragers (Jarvenpa 1980; So 1980; Brumbach and Jarvenpa 1997a; Johnson 2008). More importantly, Winterhaider's (1981) analysis of foraging return rates among the Cree demonstrates that plant foods have much lower return rates than hunted foods in northern environments. Although plant foods such as tubers may have been utilized by Neandertals, even in cold environments (Hardy 2010), such resources were probably not only seasonally limited in abundance around Gatzarria, but low-ranked in comparison with animal alternatives. During the winter, and during much of fall and early spring, plant resources would be unavailable, making 42 animal meat virtually the only food source. Stable isotope analyses confirm that Neandertals were mainly carnivores (Bocherens et al. 1991; Richards et al. 2000; Richards et al. 2001). Plant resources are excluded from following discussions of diet and foraging at Gatzarria. However, for the above reasons, the faunal data should be a good indicator of the overall diet. Ecology ofungulate species at Gatzarria This section considers the ecology of the different ungulates on the landscape around Gatzarria, and serves as a basis for defining resource patches, interpreting taxonomic representation, and understanding seasonality in subsequent chapters. In this and later sections, the term "ungulates" is broadly defined to include artiodactyls, perissodactyls, and proboscideans. Before proceeding, a general caution regarding the use of modern analogues should be expressed. Although many of the species that were present in Pleistocene France are extant today, some are extinct (e.g., giant deer, Megaloceros sp.), some are often managed (red deer), some are domesticated and poorly known in the wild (horse), and others have been brought back from the brink of extinction (both American and European bison). The ecology of extant wild populations of these species may have been modified through these processes. The absence of large predatory animals such as lions and hyaenas in Europe today may also impact the reproductive success, population structure, and behaviour of contemporary ungulates. Furthermore, populations of different ungulate species no longer co-occur in the manner that they did during the Pleistocene. 43 These gaps in our knowledge of the ecology of many ungulate species mean that a high degree ofprecision in interpretations of seasonality and age, and exact definitions of resource patches may not be possible. For instance, the seasonal variability of an environment and the severity of prédation may influence the timing of birth periods or the synchrony of birthing in ungulate species (Rutberg 1987; Sinclair et al 2000). Nevertheless, the ecological adaptations of "extinct forms can only be inferred from analogies drawn from their closest living relatives" (Guilday 1984:254). There are strong grounds for assuming temporal stability in the habitats and development patterns of the majority of ungulate species considered here. First, on an evolutionary time scale, the time period considered in this study is relatively recent. In this regard, modern analogues retain considerable strength for the analysis of carcass exploitation, as the anatomical structure of ungulate species such, as red deer and reindeer is virtually unchanged since the time period in question. Furthermore, patterns of skeletal morphology are strongly consistent between ungulate species (Lam et al. 1999). Second, despite the presence of some non-analogue communities in the Pleistocene (Stewart and Lister 2001), palaeoecological data suggest that the habitat associations (both climatic and topographic) and the trophic ecology of the species considered in this paper are generally similar to those observed today (Bocherens et al. 1991; Dari and Renault- Miskovsky 2001; Grayson et al. 2001; d'Errico and Sánchez Goni 2003; Grayson and Delpech 2005; Morin 2008; García García et al. 2009). Using predictive ecological modeling, Banks et al. (2008) argue that the niches of red deer and reindeer have changed little since the Last Glacial Maximum. A further example is the presence of ibex and chamois in faunal remains from Gatzarria, in Pyrenees region, but not at the site of Saint- Césaire, which is located further north in a less rugged landscape. This pattern is consistent with the modern distribution of these species. Therefore, although some differences in the ecology of Pleistocene and modern faunas can be expected, it is assumed here that the basic ecological requirements and tolerances, habitat associations, seasonal cycles including birthing seasons, and growth patterns of these animals are broadly similar. Red deer (Cervus elaphus) are generalists with plastic ecological requirements and may be found in a variety of habitats in temperate climates (Geist 1998). Clutton- Brock and Albon (1989) suggest that red deer evolved as a low-ground species, occupying areas covered by heaths, woodlands, and forest margins. Like most cervids, red deer prefer highly nutritious, protein-rich foods (Geist 1998), and therefore generally choose to forage in new growth, particularly grasses (Clutton-Brock and Albon 1989). Thus, the distribution of red deer throughout the year may be predictable. For instance, red deer often occupy lower slopes in winter and hilltops during the summer (Darling 1963; Clutton-Brock and Albon 1989). In summer, this behaviour is related to the maturation of grasses and insect avoidance. Typically, cervids occupying open areas form larger groups than forest species, and red deer are no exception. Red deer groups generally consist of four to seven adults, excluding calves, often maternally-related hinds (Clutton-Brock et al. 1982). Much larger groups may form, particularly during the summer and in open spaces. Although generally non-migratory, some red deer populations do move more than 50 km between summer and winter ranges (Clutton-Brock et al. 1982). Some populations of red deer primarily 45 inhabit forested areas, but their ecology is less well known. Forest red deer are larger and more solitary than their field counterparts (Clutton-Brock and Albon 1989). The yearly cycle of red deer is similar to that of other cold- and temperate-climate ungulates. Stags are in best condition during the summer, and put on substantial weight prior to the October rut (Clutton-Brock et al. 1982; Clutton-Brock and Albon 1989). Among field red deer, harems form on traditional rutting grounds for the duration of the rut. Stags may lose up to 20% of body weight and 80% of body fat during the rut, and remain in poor condition through the winter until June when they begin to put on weight once again (Clutton-Brock and Albon 1989). Hinds maintain a roughly constant weight through the winter while pregnant, but do not gain weight during the summer while lactating. However, the constant weight during the winter does not mean that the condition of pregnant hinds remains static; rather, that the loss of weight is compensated by fetal growth. YeId hinds (those not becoming pregnant during the fall rut) rapidly lose weight during the winter, but gain weight during the summer due to their lower energetic requirements in comparison to lactating hinds (Clutton-Brock and Albon 1989). Most calves are born in the first two to three weeks of June (Clutton-Brock and Albon 1989), and birthing is highly synchronous. Clutton-Brock and Guinness (1975) observed that, among red deer on the Isle of Rhum (Scotland), 69% of calves were born in a three-week period between May 25th and June 14th, and a quarter of calves (26%) were born during the first week of June. Finally, the vulnerability of red deer to large population fluctuations, such as the boom-and-bust population cycles characteristic of reindeer, is not clear. In a twelve-year study, Clutton-Brock and Albon (1989) observed no population crashes among the red 46 deer they studied in the Scottish highlands, even in the absence ofpredators, which might be expected to maintain populations at lower densities and thus prevent overshooting of carrying capacity. These authors suggest that red deer populations are strongly limited by the availability of suitable forage (habitat quality), resulting in lower fecundity and higher infant mortality among high-density populations. However, density-independent factors, such as environmental stochasticity, may also cause significant mortality and lowered fecundity (Saether 1997). Environmental variability may also exaggerate the impact of density-dependent mortality. For instance, an unusually severe winter may cause greater mortality among a high-density population in comparison to a low-density one. The long lifespan of large ungulates such as red deer makes the effect of year-to-year recruitment on population size difficult to study and understand (Saether 1997). The role ofprédation in red deer population dynamics is also poorly understood. If red deer populations do fluctuate significantly in size, such 'cycles' are likely to be both irregular and unpredictable (Saether 1997). Not unlike red deer, bison are ecologically plastic and able to adapt to a variety of climatic and geographic environments (Reynolds et al. 2003). Brugal (1983) suggests that the steppe bison (Bison priscus) of the European Pleistocene would have resembled the American plains bison in its gregariousness and seasonal movements. However, Pleistocene bison in North America may have been more generalist (mixed grazers and browsers) than contemporary American bison (Rivals et al. 2007), and this observation may be true for the ecology of Pleistocene bison in Europe as well. The steppe bison appears to have inhabited steppe, permafrost/tundra, as well as taiga (boreal forest) environments (Brugal 1983). As with red deer, bison populations likely occupied both 47 grassland and woodland environments. Both plains and wood bison are predominantly grazers, feeding in open grassland or meadow areas, but may consume shrubs and forest lichens depending upon forage availability (Reynolds et al. 2003). The reproductive cycle of bison is similar to that of red deer, although the rut occurs earlier, peaking in late-July/mid-August. Among plains bison, herd size increases markedly during the rut, while herd size declines among wood bison (Reynolds et al. 2003). This variability may be due to environmental differences. Most calving occurs between late April and early June, during a 3-4 week period in the spring (Reynolds et al. 2003). The condition ofmale and female bison adults throughout the year roughly parallels the cycle described for red deer (Speth 1983). Bison are gregarious animals, usually forming matriarchal groups of variable size and membership (Reynolds et al. 2003). Group sizes may range from a few individuals to a few hundred, with groups tending to be larger in open areas. Breeding herds on the American Plains may have, at times, reached several thousands (Isenberg 2000), although these appear to have been exceptional, even for plains bison which are more gregarious than forest-dwelling populations. Outside the breeding season, small groups are the norm. Annual migrations, either altitudinal or long-distance, are common, and permit bison to spend the winter in areas with better winter grazing and, possibly, to avoid insects during the summer (Reynolds et al. 2003). Isenberg (2000) suggests that plains bison population sizes would have been extremely variable, as a result of the bison's potentially high rate of increase during good years, in combination with the possibility of catastrophic mortality of herds. The same conclusion probably applies to wood bison as well 48 (Reynolds et al. 2003). Causes of catastrophic bison mortality include drought, fire, drowning, heavy snowfall, or other adverse winter conditions (Reynolds et al. 2003). It should not be forgotten that some of the large bovids at Gatzarria could be aurochs. Aurochs were presumably less ecologically flexible than bison (Brugal 1983). However, it is assumed that their annual reproductive cycles would have been similar (van Vuure 2005). Aurochs were grazers adapted to warm, humid climates, and were unable to support cold or dry conditions. They possibly inhabited grassy meadows within forested areas, or marshy areas with sedges. Van Vuure (2005) suggests an ecological partitioning between European bison and aurochs, with bison occupying high, cold, dry, more open habitats, and aurochs occupying lower, wet, warm areas. Due to their ecological requirements, aurochs were probably rare in France during the late Pleistocene until the postglacial period (Brugal 1983). During the last glaciation, aurochs did not occur above 500 m elevation in the Pyrenees (van Vuure 2005). For these reasons, it is probable that bison would have been more common at Gatzarria. Horses typically inhabit open areas and form harem bands consisting of a small number of adult animals of each sex, along with the young offspring of mares (Jenkins and Ashley 2003). Although domestic horses conceive and give birth at any time of year, breeding and foaling in feral horses generally occur during restricted periods. Most foaling among a feral horse population on Sable Island, Nova Scotia, occurred in April and May, with 95% of births between April and July (Lucas et al. 1991). A study in Nevada suggested a slightly longer birthing period, extending from March to July (Jenkins and Ashley 2003). These data suggest that birth synchrony among horse would 49 probably increase in environments with more severe winters, such as Pleistocene France (Morin2010). Like bison, horse may be vulnerable to considerable fluctuations in population size. Although horse populations have a high potential growth rate, observed growth is strongly dependent on the population age structure (Jenkins and Ashley 2003), which in turn may be shaped by catastrophic mortality of calves in certain years. Berger (1983) suggests that horses are vulnerable to catastrophic mortality due to factors such as severe winter storms, high snow depths, mud traps near water sources, or drought. Density- dependent mortality due to competition and prédation also affects horse populations (Jenkins and Ashley 2003). Chamois are small alpine ungulates generally occupying high-altitude rocky habitats, although Miracle and Sturdy (1991) suggest that chamois are primarily adapted to craggy environments unsuitable for other ungulate species, regardless of altitude. The altitude at which chamois may be found depends on food availability, as the species often follows the snow line up and down in search of fresh vegetation (Garcia-Gonzalez and Cuartas 1996; Herrero et al. 1996). Chamois prefer to graze in meadows, although they may descend into woodlands to browse during the winter (Garcia-Gonzalez and Cuartas 1996). However, where the ranges of roe deer and chamois meet, the two species may be mutually exclusive, with chamois at higher, and roe deer at lower altitudes (Herrero et al. 1996). Competition from livestock (sheep, horse, and cattle) in high pastures may push chamois to higher and/or lower altitudes in the summer (Herrero et al. 1996). These observations suggest that competition, as well as an adaptation to different terrain, would tend to segregate chamois from ungulates such as bison, horse, red deer, and roe deer. 50 Breeding takes place in November and December, and female chamois give birth in May and June (Herrero et al. 1996). Like chamois, ibex are adapted to rocky, mountainous terrain; including rocky prairies, cliffs, and steep slopes (Villaret et al. 1987). Alpine ibex are restricted to relatively high elevations, as their distribution is limited by the availability of steep cliffs and rocky habitats. In general, Alpine ibex occupy habitats over 1800 m in elevation (Parrini et al. 2009)—well above the elevation of Gatzarria, at 290 m. Ibex undertake a seasonal altitudinal migration, moving down in winter, and upwards in the summer (Acevedo and Cassinello 2009; Parrini et al. 2009). Ibex are mixed grazers and browsers, and vary within that spectrum depending on food availability (Acevedo and Cassinello 2009), although they generally avoid woodlands (Parrini et al. 2009). Today, Spanish ibex may share summer high-mountain pasture with domestic sheep, goats, horse, and cattle (J.M. Pérez et al. 2002). In winter, ibex graze on steep rocky slopes with little snow (Parrini et al. 2009). Conception occurs in December, and parturition in May (Fandos 1989). As with red deer, ibex are in their best condition in autumn and their poorest in April, at the end of winter (Parrini et al. 2009). Roe deer are small, cryptic, forest cervids. They are territorial and live in small groups (often pairs), at low densities (Geist 1998). Given these characteristics, it can be expected that roe deer would be encountered only rarely in comparison to more gregarious species that tend to feed in open areas, such as red deer or bison. Although field roe deer are present in Europe today, Geist (1998) suggests that this ecotype would be unlikely under intense carnivore prédation, which would likely have characterized the Pleistocene. 51 The remaining ungulate species (mammoth, giant deer, reindeer, and woolly rhinoceros) found at Gatzarria contributed only minimally to the Cj Mousterian assemblage. However, these species deserve a few comments. First, mammoths, like modern elephants, were probably keystone species in the ecological communities they inhabited (Haynes 2006). The feeding habits of mammoths, like elephants, might have encouraged the development of small-scale ecological mosaics by preventing the spread of woody growth—thus maintaining open grassy areas as well as shrublands (Haynes 2006; Owen-Smith 1988). Pruning of woody growth by mammoths may have promoted coppicing and new growth that would be attractive to other herbivores. Overall, the impact of mammoths may have been beneficial to other ungulate species such as bison, red deer, and horse. Other megaherbivores such as woolly rhinoceros might have had similar impacts on the landscape (Owen-Smith 1988). Geist (1998) contends that giant deer would have been similar to moose (Alces alces) not only in size but also in ecological requirements, although the giant deer was mòre warm-adapted than moose. Due to their nutritional requirements, male giant deer would probably often be found in wetland habitats (Barnosky 1985, 1986; Geist 1998), a suggestion possibly supported by the frequent finds of male giant deer in bogs. Finally, reindeer is only a minor element at Gatzarria. Although cold-adapted, reindeer are a highly plastic species whose ecological behaviour strongly depends on learning as calves (Geist 1998). As such, generalizations about reindeer ecology must be treated with caution, as not all reindeer form migratory herds inhabiting open environments. Besides tundra environments, reindeer may occupy woodland, as well as high altitude, habitats (Miller 2003). Thus, woodland or mountain reindeer ecotypes may 52 have occupied some areas around Gatzarria. Alternatively, the northern slopes of the Pyrenees may have been the fringe of migratory reindeer habitat or an area into which reindeer made incursions during colder periods. Therefore, Tartar's (2009) suggestion that the reindeer antler in the Early Aurignacian of Gatzarria was imported from a considerable distance may not be the most parsimonious explanation of the presence of reindeer antler at the site, particularly given that reindeer abundance in the complete Early Aurignacian sample has not been estimated. Although reindeer were probably only infrequently encountered by Mousterian foragers from Gatzarria, they likely would have been taken upon encounter, as their larger body size would provide higher returns than ibex, roe deer, or chamois. Summary This chapter has presented the background necessary to an interpretation of archaeological materials from Gatzarria Cave, including the geographic context, the history of excavation, and previous archaeological research at the site. In particular, the stratigraphy of the site has been outlined in detail, from both sedimentary and archaeological perspectives, and some problems with the correlation between these two sources of evidence were identified. Palaeoenvironmental information was also briefly reviewed. Notably, few plant resources would have been available to foragers at Gatzarria during the late Pleistocene, so animal resources probably constituted the bulk of the diet. Important ecological characteristics of ungulate species present in the Gatzarria assemblage were presented for their relevance to interpreting subsistence behaviours. The next chapter turns to the theory and zooarchaeological methods that are used to examine the faunal remains from the Cj Mousterian deposit. 53 Chapter 4: Theory and method This chapter begins with a presentation of the foraging models that will guide the interpretation of the Gatzarria Cj Mousterian faunal assemblage: the diet breadth model, the marginal value theorem, and the central place forager prey choice model. Matters relevant to applying these models in the analysis, including the identification of resource patches and the construction of prey rankings, are then addressed. In the second part of the chapter, the zooarchaeological methods used in this study are discussed. More specifically, this involves a description of the types of data collected and a consideration of why these data are imperative to zooarchaeological analysis. Quantification measures, refit analysis, skeletal determinants of age and sex, bone modifications such as butchery and tool use, and taphonomic processes (burning, carnivore ravaging, fracture patterns, and fragmentation) are briefly reviewed. Theoretical approach Human behaviour is highly diverse, and consequently, often resists theoretical generalization. When examining human subsistence, however, it makes sense to use a theoretical perspective that is both evolutionary and ecological, because all human groups must meet basic subsistence requirements in order to survive and reproduce. Foraging theory is a subset of Human Behavioural Ecology (HBE), a body of theory in which the principles of natural selection are assumed to shape human behaviour (Smith and Winterhaider 1992). Foraging theory has substantial predictive power, as mathematical models of expected behaviour can be derived using economic principles of optimization. Specifically, foraging theory predicts that foragers will attempt to maximize their evolutionary fitness (i.e., reproductive success) by maximizing or minimizing a measurable proxy for fitness, such as energy efficiency in foraging. For zooarchaeologists, foraging theory allows tests of formal hypotheses regarding resource abundance, transport decisions, environmental variability, and human demographic change (e.g., Broughton 1994, 2002; Cannon.2000, 2003; Stiner et al. 2000; Nagaoka 2001, 2006; Lyman 2003; Munro 2004; Faith 2007; Morin 2010). Two fundamental classes of foraging models are discussed here, along with a third model of central place foraging which incorporates elements of the previous two models. The diet breadth model Diet breadth, or prey choice, models predict which foods foragers will include in their diets and which they will ignore, based on the net gain provided by different resources (Stephens and Krebs 1986). The diet breadth model was first proposed by MacArthur and Pianka (1966), and has since been described and refined by a number of authors (e.g., Stephens and Krebs 1986). Diet breadth models predict that foragers will adjust the content of their diet in order to maximize the net return rate of subsistence activities (Kaplan and Hill 1992). Foragers are expected to ignore any prey types with an on-encounter return rate that would decrease the overall foraging return rate including search time. Consequently, the diversity (number of prey types) in the diet should be a relative measure of foraging efficiency. If the relative abundance of the highest-return prey types in the foraging environment decreases significantly over time, then the overall foraging return rate decreases. As a result, previously ignored resources, beginning with 55 the highest-return prey type not already included in the diet, are expected to be added to the diet until the foraging return rate is maximized for the new conditions of prey abundance. The return rate of a resource may also be affected by changes in technology, site function, procurement technique, and environmental change (Grayson and Cannon 1999). Diet breadth models require that the return rates ofprey items be established a priori, at the very least on an ordinal scale. In archaeology, it is generally assumed that prey body size is a good proxy for the return rate of animal resources, with the possible exception of some small animals that can be collected en masse (Ugan 2005). However, prey mobility also has a significant impact on return rates (Stiner et al. 2000; Bird et al. 2009), so the presence of particularly fast or slow prey types in the foraging environment should be considered, along with body size, in establishing prey rankings. Hill (1988; Kaplan and Hill 1992) has stressed that for human foragers, diet breadth models should only be applied to resources with similar macronutrient composition. In zooarchaeological analyses, diet breadth models are usually applied exclusively to animal prey, which are primarily sources of protein and fat. Archaeologically, diet breadth models are often used to assess temporal changes in foraging efficiency in a single archaeological sequence or at a group of succeeding sites. Animal resource types, each ranked by relative return rate, are used as measures of diet breadth. Variation in foraging efficiency may be signaled by changes in the number of prey types exploited, or by changes in the relative abundance of different prey types. Temporal changes in diet breadth may be associated with changes in mobility patterns, technological advance, climate change, or other changes in resource availability, 56 including predation-induced resource depression (Grayson and Cannon 1999). In order to control for these different possibilities, additional lines of evidence are usually required to eliminate, or support, alternative explanations for observed changes in diet breadth. The marginal value theorem ¦ The marginal value theorem (or MVT) is a model of patch time-allocation which predicts how foragers should allocate their time in an environment in which resources are clumped, or located in patches, as opposed to when resources are randomly distributed, as assumed in the diet breadth model (Charnov 1976; Stephens and Krebs 1986; Kaplan and Hill 1992). Different applications of the original theorem have been proposed to manage different analytical problems (e.g., Burger et al. 2005). In general, the return rate of a patch is considered to diminish over time. A forager is expected to abandon a patch when the instantaneous return rate ofthat patch falls below the average return rate of the habitat including travel time between patches (Charnov 1976). The relative importance of different patches can be assessed in archaeological assemblages by dividing prey types into appropriate patches according to habitat, prey characteristics, and/or procurement technique. For example, Broughton (2002) divided the fauna from the Emeryville Shellmound, California, into three main patches: terrestrial mammals, estuarine fishes, and waterfowl. Patch types must be exclusive of different prey types, but at the same time, must fulfill the assumption of random encounter with prey types within each patch. The "prey-as-patch model," which applies the MVT at a smaller scale, may be particularly useful for archaeologists because it can be used to interpret skeletal part patterns (Burger et al. 2005). Since large prey can be subdivided into parts with differing energy yields, a carcass can be considered as a patch with a return rate that diminishes 57 with increased processing time. Thus, processing intensity can be considered as a proxy of residence time in a patch, and thus provides a relative measure of return rate. Foragers are expected to move on to another patch (i.e., to exploit another animal), when the return rate of continued exploitation of a carcass drops below the expected return rate of obtaining and exploiting another animal (Burger et al. 2005). Patterns of skeletal remains deposited at a site are therefore expected to reflect forager decisions about which skeletal elements were worth transporting to a site and processing. In particular, using the prey- as-patch model, patterns of anthropogenic fracturing for marrow extraction can be considered to be a measure of overall foraging return rates. Bone grease extraction represents the lower limit of return rates that can be measured using this model (Burger et al. 2005). In some assemblages, it is possible to determine which elements were not ' cracked for marrow. The unprocessed element with the highest marrow yield is termed a "stop element" for marrow extraction by Burger et al. (2005). A comparison of these stop elements through time can be used to assess changes in foraging return rates. The fact that the prey-as-patch model easily tracks marrow exploitation makes it particularly suited to studying foragers whose goals include fat maximization (Burger et al. 2005). This is important because fat appears to be a limiting factor for temperate, subarctic, and arctic foragers (Speth and Spielmann 1983), and possibly for foragers in Europe during the Middle to Upper Palaeolithic transition as well (Morin 2004). Some authors have used the diet breadth model, rather than the prey-as-patch model (the MVT), to discuss the transport and exploitation of different skeletal elements (e.g., Jones and Metcalfe 1988), suggesting that different body parts can be conceived of as different prey types. This approach obtains the same prediction as the prey-as-patch 58 model: skeletal parts with lower return rates are added to the diet last. Therefore, why should the prey-as-patch model be preferred over the diet breadth model to understand patterns of carcass exploitation, or vice versa? The answer may be related to the assumptions of the two models and the structure of vertebrate prey. When a whole animal is encountered, or when a carcass at a residential site is not totally disarticulated, groups of different skeletal elements are encountered as a package. Therefore, if different skeletal parts are considered as prey types, the assumption that prey are encountered sequentially and following a Poisson process (Stephens and Krebs 1986) is violated. In contrast, this assumption is not violated if the patch model is used as the basis for interpreting skeletal part patterns, since in this case the assumption applies to the initial encounters with patches (whole animals). If skeletal elements are considered as different prey types, then the assumption of exclusivity of search and handling of the diet breadth model may also be violated. For example, while handling a prey type (e.g., a humerus), a forager may encounter another prey type (such as the scapula), and may in fact handle them simultaneously (for instance, if disarticulation is necessary to access either part). Because whole carcasses are considered in the prey-as-patch model, the assumption of exclusivity of travel to, and exploitation of, a patch is not violated. Consequently, the prey-as-patch model, not the diet breadth model, should be used as the basis for considering skeletal part patterns. The use of the whole carcass as a patch is not unreasonable in most archaeological circumstances, because hunters initially encounter whole animals. The whole skeleton model is, out of necessity, already a starting point for most zooarchaeological analyses (Lyman 1994:289). 59 The centralplaceforagerprey choice model The basic diet breadth model, originally derived by biologists for non-human foragers, has been criticized for not accurately representing how humans foragers behave (e.g., Broughton 1994). Indeed, some of the requirements of the diet breadth model, such as the assumption that all prey are searched for simultaneously and encountered randomly, are often violated by human foragers. This problem can sometimes be accommodated by dividing the environment into patches in which encounters with different prey types are more likely to be random (e.g., Broughton 1994; Smith 1991). This is the basic procedure that will be used in considering diet breadth at Gatzarria. However, the diet breadth model does not accommodate the fact that people generally travel to and from a central place in their foraging activities. This means that, when large prey are captured, decisions about which parts can be brought home, and which will be left behind, often need to be made. Although the logic of the MVT suggests how the selection of parts would be made, the MVT does not relate processing decisions to transport distance. In order to accommodate how humans who forage from a central place search for prey and field-process resources that come in large packages, Cannon (2003) constructed the central place forager (CPF) prey choice model. Unlike the classic diet breadth model, the expected costs of searching for a prey type, handling and, if necessary, field- processing that prey, as well as transporting it back to the central place, are integrated in this model along with the energy yield of a prey type into a single measure of foraging efficiency. Because different search costs are associated with each prey type, there is no 60 fine-grained search assumption and the patchiness of the environment does not affect the model predictions. The inclusion of search and travel costs in the calculation of post-encounter return rates in the CPF prey choice model means that if these costs increase significantly for an initially highly-ranked prey type (for instance, due to predation-induced resource depression), it may become a new, lower-ranked prey type. This leads to the important prediction that an increase in the abundance of a large-bodied prey, accompanied by an increase in field-processing intensity, may under certain circumstances signal decreased, rather than increased, foraging efficiency. However, this prediction may not apply to highly mobile foragers, such as Palaeolithic foragers in Europe, who would be expected to move residential bases closer to patches rich in high-ranked prey rather than incur the cost of longer search times and higher transport distances (Morin 2010). In this study, the prediction of concern is the relationship between field-processing intensity and average search time for a prey type. The CPF prey choice model assumes that there is a maximum load size that a forager can transport. If a prey exceeds the maximum load size, then under certain conditions, foragers may be able to increase the net delivery rate of food to the central place by field-processing that prey prior to transport. The relationship between the time spent field-processing and the additional gain in utility from that work is assumed to approximate a diminishing-return function. Therefore, as transport distances and/or average search times increase (i.e., when encounter rates decrease), foragers attempting to maximize the net delivery rate should invest more effort into field-processing. This relationship means that patterns of skeletal part transport can provide a means of 61 assessing relative foraging return rates. Skeletal profiles including a wide range of parts, including low-utility portions, should reflect lower average search times, while profiles heavily dominated by highly-ranked portions should reflect higher average search times. It should be noted that human groups do not always behave as central place foragers. For instance, when a large kill is made, human foragers may elect to move the location of their camp to the kill site, and thus avoid any need for transport (Brumbach and Jarvenpa 1997b). However, when considering faunal remains at central places, where the remains of numerous individuals of different species were deposited and where kills are unlikely to have been made (e.g., in caves; Binford 1984), the deposited remains must have been transported to the site from the original location of the carcass. Although central places may have been only one type of campsite among many, one such site is focus of the present analysis. Critiques offoraging theory Numerous critiques have been advanced regarding behavioural ecology (e.g., Gould and Lewontin 1979; Joseph 2000, 2002) and foraging theory (e.g., Martin 1983, 1985; see Pyke 1984 for a critical review of foraging theory in biology). A full discussion of these issues is beyond the scope of this work (for a cogent defense of the approach see Smith 1991; for some rebuttals to specific critiques see Smith and Winterhaider 1985; Yesner 1985; Winterhaider 2002). One particularly important concern that has been leveled at foraging theory will be discussed here. This is the suggestion that behavioural ecology, including foraging theory, is guilty ofbiological determinism. In this view, foraging theory is considered to be an overly simplistic view that ignores important 62 dimensions of human behavior, such as social goals, and places an unrealistic emphasis on economic optimization. Smith (1991) provides a compelling defense of foraging theory against this critique, and his major points are reviewed here. Most importantly, although foraging models may not explicitly consider culture, this does not imply that the actors are acultural. Rather, it is assumed that cultural behaviours, like genetically-encoded behaviours, tend to be adaptive. This is known as the phenotypic gambit, which posits that natural selection acts on both genes and learned behaviours (Smith and Winterhaider 1992). Furthermore, there are several reasons why foraging behaviours are likely to be adaptive and match the predictions of optimality models, more so than other classes of behaviour. First of all, because of the critical role of foraging success in survival and reproduction, the selective pressures on cultures for the adaptive value of foraging behaviours are likely to be quite strong. It is reasonable to assume that natural selection has shaped foraging in adaptive ways through biological means (such as taste and other sensory abilities) as well as in learned or cultural ways. Ethnographic and archaeological data suggest that human resource preferences do vary in predictable ways, and these preferences tend to reflect foraging efficiency (Smith 1991). The Middle Palaeolithic faunal record suggests that Neandertals may have also conformed to this pattern, at least as regards animal resources: species with (presumably) higher return rates appear to have been more often exploited than smaller-bodied, swift taxa (see Chapter T). Secondly, foraging behaviours in humans are often learned from close kin, and therefore are "subject to selection primarily on the basis of their effects on biological 63 survival" (Smith 1991:23), as opposed to many social behaviours which are more likely to be reproduced through non-biological means and thus become decoupled from natural (but not cultural) selection. It is probable that among earlier hominins, including Neandertals, many behaviours requiring learned skills, and food acquisition behaviours in particular, were learned from close kin (Shennan and Steele 1999). This is the case among most modern hunting and gathering groups (Shennan and Steele 1999). Finally, foraging behaviours are repetitive and have short-term pay-offs. Therefore, they give ample opportunity for actors to directly experience the consequences of individual decisions and to adjust their behaviour in response to these outcomes (Smith 1991). Although the fit between foraging model predictions and actual behaviour is often imperfect, many ethnographic studies have confirmed the usefulness of these models for predicting human behaviours (e.g., Winterhaider 1981; Hill 1988; Smith 1991; Bird and Bliege Bird 2000; Bird et al. 2009). Mismatches between observed and predicted behaviour in ethnographic studies are often due the wide range of variables affecting foraging behaviours, which are not always included in the models (Lupo 2007). For example, the prestige value of large prey is a potentially important, although controversial, factor that may affect men's foraging decisions (Bliege Bird and Bird 1997; Hildebrandt and McGuire 2002; Smith et al. 2003; Bird and O'Connell 2006; Codding and Jones 2007; McGuire et al. 2002; Gurven and Hill, 2009; Hill and Kintigh 2009). This type of costly signaling may also have been one of the objectives of Neandertal males (O'Connell 2006; Morin 2010). However, despite its simplicity—or perhaps because of it—the optimality modeling approach has permitted a better understanding of the role of many variables in affecting archaeological patterning (e.g., 64 Lupo and Schmitt 2002). Overall, the narrow focus of foraging models compromises some realism, but in doing so, provides clarity and tractability in dealing with one part of a complex system (Smith 1991). Frequently, this approach generates new insights and hypotheses that can be further tested using more complex models. Application offoraging theory in this study The use of foraging models permits zooarchaeologists to test hypotheses about changes in hunter-gatherer subsistence by providing predictions about foraging behaviours. However, formal archaeological tests of hypotheses derived using foraging theory usually require a comparison of foraging patterns through time or space. How can foraging models be productively used to examine a single occupation of a single site, in this case, the Cj Mousterian occupation at Gatzarria Cave? Although no formal test will be carried out, applying foraging theory concepts to the analysis of the Cj Mousterian fauna will allow: (1) a characterization of diet breadth and patch use during the late Mousterian in the Atlantic Pyrenees, using the concepts of the diet breadth and patch-time allocation models; (2) a relative assessment of subsistence security among Neandertals in the region, through a consideration of transport selectivity (using the reasoning of CPF prey choice model) and of marrow-cracking intensity (using the MVT). In particular, questions of diet breadth and carcass exploitation intensity at archaeological sites documenting the Middle to Upper Palaeolithic transition are of interest for understanding foraging change during the transition process. Finally, the analysis of the Cj Mousterian material should be considered as a starting point for a hypothesis-testing-oriented analysis of foraging behaviour across the full sequence of Middle and early Upper Palaeolithic occupations at Gatzarria. 65 Defining resource patches In order to discuss patch use and diet breadth within those patches, it is first necessary to identify the resource patches that would have been available. Based on the general sketches of the ecology of the ungulate species presented in Chapter 3, it is possible to broadly define the resource patches available to Mousterian foragers at Gatzarria. The coarse grain of these patches means that encounters with prey types within these patches may not have been entirely random; however, this approach is a necessary compromise between the requirements of the patch model and our imprecise knowledge of ungulate community ecology during the time period under study. To some extent, an altitudinal gradient provides a convenient way of conceptualizing patch types. The main resource patch can be termed a 'lowland patch,' consisting primarily of both grassy open areas and forests at lower altitudes. Horse might have been encountered most frequently in grassland areas, while roe deer were probably solely located in forested areas. However, the flexible ecology of red deer and bison, the primary species at Gatzarria, means that the wooded and grassy habitats cannot be considered exclusive patches. Grassland areas of the lowland patch would also have been the primary habitat of mammoth and woolly rhinoceros. Wetlands represent an additional lowland habitat that may have been available to human foragers from Gatzarria. Wetlands may have been the primary habitat of male giant deer (Barnosky 1985) and, possibly, aurochs (van Vuure 2005). However, giant deer is poorly represented at Gatzarria, and the presence of aurochs is uncertain. Other ungulates may also have been encountered in wetland areas. For instance, bison may forage for fresh growth in chest-deep waters (Reynolds et al. 2003). For these reasons, wetlands are included in the lowlands patch. 66 A secondary 'mountain patch' can also be defined. This patch consists of cliffs and craggy or rocky terrain, as well as meadows and pastures at higher elevations. The primary species that would be encountered in this patch are chamois and ibex, as the rocky habitats and steep slopes utilized by these species for escape and feeding purposes are not favourable to other ungulate species. Possibly, mountain reindeer living at higher altitudes could have belonged to this patch as well. Ungulates such as red deer, bison, and horse might have been occasionally encountered in the open meadow areas of this patch, particularly during the summer. However, during the winter, at very high altitudes, or on extremely rugged terrain, chamois and ibex (and possibly reindeer, if present) would probably be exclusively encountered. Although the lowland and mountain patches may not have been entirely exclusive in terms of the species encountered, by and large the site occupants probably would have evaluated them differently. Different travel times and differences in vertical travel distance might have distinguished these patches and influenced patch rankings, for instance, if mountain patches required substantial climbs or negotiating craggy terrain. The lower ungulate diversity and smaller body size of the main prey in the mountain patch, in addition to potentially greater travel costs, suggests that the exploitation of ibex and chamois may have been less profitable than lowland options. Prey rankings To assess the diet breadth of foragers at Gatzarria, rankings of available prey options need to be established. Generally, body size has been used as a proxy of prey profitability by archaeologists employing the diet breadth model; however, prey defense mechanisms and mobility may confound body size rankings (Bird et al. 2009; Stiner et al. 67 2000). Marin (2010) compared the body masses and escape velocities of a broad range of mammals, and suggests that body size rankings work well within the 50-700 kg size range. Around 700 kg and beyond, it is likely that the probability of a failed pursuit after encounter increases steadily. For artiodactyls within this size range, the body size rule appears to be quite strong because maximal velocity decreases with body size, suggesting that smaller artiodactyls are not only less profitable in terms of gross energy content but also because they are faster. Morin (2010) further suggests that rank inversions are less likely when prey are well-separated in size and that, consequently, the body size rule may be stronger in cool climates where a lower diversity of mammals, with a large range of body masses, would have been available to human foragers. Figure 4.1(a) illustrates the ranking of ungulate prey species at Gatzarria using the body size rule, all patches combined. Following Morin (2010), body mass was estimated using weights of modern animals from MacDonald and Barrett (2001) or, for extinct species, from Pushkina and Raia (2008). The data used in creating Figure 4.1 are shown in Table 4.1. Although it appears that carnivores were not consumed at Gatzarria (see Chapters 5 and 6), they are included in the rankings for comparison. Bison are the highest-ranked species by a considerable margin, followed by giant deer, horse, and red deer. Reindeer, ibex, and chamois—the species forming the mountain patch—constitute a second, lower-ranked group, while roe deer is the lowest-ranked ungulate species. However, net energetic (caloric) return may not necessarily have been the main target of human foraging at Gatzarria. In high-protein diets consisting primarily of animal foods, such as those that probably characterized much of Pleistocene Europe (Bocherens et al. 1991; Richards et al. 2000; Richards et al. 2001), fat can be a limiting factor 68 a) bison giant deer cave bear horse red deer cave lion hyaena reindeer ibex chamois wolf roe deer red fox P arctic fox 0 100 200 300 400 500 600 700 Mass (kg) b) giant deer bison cave bear cave lion reindeer red deer horse/ibex hyaena chamois wolf roe deer arctic fox red fox Fat Mass (kg) Figure 4.1: (a) Ranking of ungulate and carnivore species at Gatzarria according to body mass, (b) Ranking of ungulate and carnivore species at Gatzarria according to fat mass. Data from Table 4.1. 69 Species Mass (kg) Fatness (%) Fat mass (kg) bison 700.0 8.43 59.0 giant deer 388.0 19.10 74.1 cave bear 275.0 16.00 44.0 horse 270.2 2.77 7.5 red deer 202.5 4.97 10.1 cave lion 183.0 13.25 24.2 hyaena 102.0 6.70 6.8 reindeer 90.0 19.10 17.2 ibex 50.0 14.97 7.5 chamois 40.0 14.97 6.0 wolf 37.0 5.27 2.0 roe deer 25.5 4.97 1.3 red fox 6.0 2.23 0.1 arctic fox 3.6 17.00 0.6 Table 4.1: Body mass, fatness, and total fat mass for the ungulate and carnivore species present at Gatzarria. All except ibex and chamois values follow Morin (2010). Morin's mass values from MacDonald and Barrett (20001), fatness estimations derived using data provided by Caroline Pond (personal communication to Morin, 2009) and Pond and Mattacks (1985:188, Figure 2); except reindeer and giant deer from Pond et al. (1993:22); bison from Emerson (1990:513); and fox values from Pond et al. (1995:598). Ibex and chamois mass values from MacDonald and Barrett (2001) and fatness estimated using the average of two Ovis aries (Caroline Pond, personal communication, 2010). essential to survival, because without sufficient fat or carbohydrate, human digestive efficiency and health are compromised (Spetti 1983; Speth and Spielmann 1983). For this reason, foragers would need to consume fatty tissue, such as bone marrow and bone grease, along with lean meat. Fatty tissue being more limited in ungulate bodies than muscle tissue, fat acquisition might therefore have been the first priority of foraging. Morin (2010) has considered this problem, and constructed a profitability ranking for potential prey species in Pleistocene France using total fat content of modern individuals of the same or a closely related species. Table 4.1 and Figure 4.1(b) reproduce Morin's rankings for the species at Gatzarria, excluding fish and birds, but with the addition of chamois and ibex. Giant deer fatness was approximated using values for reindeer because giant deer were ecologically similar to moose (Geist 1998), which, 70 like reindeer, accumulate large fat deposits in winter (Schwartz 1992). Fat values for ibex and chamois were estimated using the average of two domestic Ovis aries (Pond and Mattacks 1985 and Caroline Pond, personal communication, 2010), the most closely related species for which fatness values are available. Like reindeer and moose, ibex survives the winter primarily by accumulating large fat reserves, and although less well- studied, chamois may use the same strategy (Lovari et al. 2006). Therefore, the relatively high fatness value for domestic Ovis aries may be a good estimate for wild ibex and chamois in prime condition. Projecting modern fatness values into the past (and using proxies to estimate fatness in species for which data is unavailable) necessarily involves some degree of error. However, given the skeletal similarities between the Pleistocene and modern forms considered here, as well apparent consistencies in ecological adaptation's for many of these species (e.g., Miracle and Sturdy 1991; Banks et al. 2008), it is assumed that the estimates of fat values are representative, at least on an ordinal scale. In the fatness scale, the rankings of giant deer and bison are reversed due to the high estimate of giant deer fat content. Red deer fall one rank in the fat mass scale, while reindeer, in contrast, gain three ranks. Despite their difference in body mass, horse and ibex contain roughly equal quantities of fat. The differences between the two scales reflect variability in winter survival strategies, with species such as reindeer, giant deer, and possibly, ibex and chamois accumulating larger fat reserves during the summer in preparation for the cold season. However, overall differences between the two scales are minimal (rs = 0.92, ? < 0.0001). This suggests that regardless of whether energetic or fat acquisition was the underlying currency of foraging, ungulate prey types would have had 71 similar rankings. If fat acquisition was important to the foragers at Gatzarria, there may have been some seasonal variation in prey rankings, with the aforementioned species being higher-ranked in the late summer through winter when they may have contained proportionally greater quantities of fat. In Chapter 6, the patch types and the prey ranking scales defined here will be used with the taxonomic abundance data to assess foraging returns at Gatzarria. Methodology In order to interpret human foraging behaviours from an archaeological assemblage, the relationship between the recovered assemblage and the deposited assemblage must be assessed, because numerous post-depositional processes, such as taphonomic filters and recovery methods, may have altered the assemblage. Furthermore, a reliable archaeological stratigraphy is required because stratigraphie mixing may create spurious patterns (Bordes 2003; Grayson and Delpech 2008; Zilhäo et al. 2008). This section considers the methods necessary to a robust analysis of the Cj Mousterian assemblage at Gatzarria. Quantification techniques and identification procedures Despite inevitable disagreements, faunal analysts have come to a remarkable level of consensus on basic definitional issues. While scientists from other fields may express shock and possibly some amusement at this statement, such agreement on foundational issues in archaeology is often rare (Aldenderfer 1998:97). The reason for the consensus on definitions in zooarchaeology is that faunal remains, unlike a lithic flake or a pottery sherd, have unambiguous biological referents. Although "zooarchaeologists might argue about whether a particular object is a fragmentary shaft of a reindeer femur or of a red deer humerus, [ . . . ] the classes that 72 underly [sic] such an argument—reindeer versus red deer, humerus versus femur—are not up for debate" (Grayson and Frey 2004:27). In terms of quantification, the basic units that are counted by faunal analysts are the skeletal element and specimen. An element is a "complete discrete anatomical unit" (Lyman 2008:5), such as a humerus or a femur. A specimen, on the other hand, is a single remain, not necessarily a complete element (Grayson 1984; Lyman 2008). The analysis of the Gatzarria faunal remains was carried out at the Musée National de Préhistoire, les Eyzies-de-Tayac, France. Identifications were made using the museum's zooarchaeological reference collection. Elements or taxa for which modern reference materials were not available were identified whenever possible using archaeological specimens in the museum collections, or with photographs and illustrated anatomical guides (e.g., Pales and Garcia 1981). All specimens were examined under a 1Ox magnification hand lens to observe surface modifications. MSP The Number of Identified Specimens (NISP) is the fundamental measure in zooarchaeological analysis of vertebrates (Reitz and Wing 1999). NISP is a direct tally of identified specimens. Its main strengths are in its simplicity: assuming that there are no interobserver differences in identification, NISP is an easily replicable measure (Grayson 1984; Lyman 2008). Because NISP is additive, there is no problem of aggregation between units or layers (Grayson 1984)—one specimen always equals one specimen, regardless of where it is found. In this study, some taxonomically identified specimens that were not identified precisely to element (e.g., metapodials) were included in NISP. 73 NISP has a number of weaknesses (see Grayson 1984; Lyman 2008 for detailed lists), many of which can be grouped into a single general category: differential representational of taxa and skeletal elements due to inter-taxonomic and inter-element variation in anatomical structure. Some animals have more bones, some taxa or elements are more readily identified, some may be more easily destroyed post-depositionally, and very small or fragmented bones are less likely to be recovered archaeologically and identified taxonomically. Because of this, NISP counts may over- or under-represent certain elements or taxa relative to their initial abundance in the deposited assemblage. At a basic level, these problems apply to all zooarchaeological quantification methods, because differential fragmentation and post-depositional destruction are the nature of the archaeological record; however, they are mentioned here because NISP is the primary measure used in this study. When differences in skeletal anatomy are the source of differential representation, the use of a normed NISP (NNISP), in which NISP counts are standardized by the frequency of skeletal elements in an individual, may help clarify patterns of skeletal representation (Grayson 1984; Grayson and Frey 2004). Differential identifiability of taxa, elements, and small fragments is a more serious concern, and indicates that NISP values should not always be taken at face value. Two means of addressing these problems (through additional quantification techniques and refitting) are discussed in a subsequent section. The problem of differential identification is dealt with in some depth in the taphonomic analysis. A second major criticism ofNISP is the problem of interdependence (Grayson 1984; Lyman 2008). The problem of interdependence is simple: each identified specimen adds one to the NISP count, so fragmented skeletal elements and individual animals may 74 be counted many times in the total NISP. Interdependence leads to problems for statistical operations (Grayson 1984), but may also cause difficulties for interpretations of faunal abundances. For example, an element may be represented by 200 specimens, but whether the specimens come from one individual or 200 individuals is not specified in a simple NISP count. Marrow-cracking activity may also inflate the number of long bone shaft fragments relative to epiphyseal portions. Because NISP is highly sensitive to fragmentation, increased fragmentation will increase the interdependence problem (Klein and Cruz-Uribe 1984; Marshall and Pilgram 1993)—until fragmentation becomes so severe that few specimens (i.e., only highly diagnostic bone portions) can be identified. Derived quantification measures have been developed to help deal with the issue of interdependence associated with the use ofNISP. Finally, an important criticism that has been leveled at NISP is that processes such as differential preservation, fragmentation, and identification render NISP counts unreliable as ratio-scale data (Grayson 1984). Fortunately, when the abundances of taxa are well separated, NISP provides good ordinal data (Grayson 1984). Ordinal-scale statistics are more than satisfactory for many zooarchaeological analyses and are used in this study. MNI, MNE, AND MAU Besides the fundamental abundance measure of NISP, three derived measures of abundance have been regularly employed in recent zooarchaeological analyses: the Minimum Number of Individuals (MNI), the Minimum Number of Elements (MNE), and the Minimal Animal Unit (MAU). The characteristics of these measures, and different calculation methods, have been discussed at length in the literature (Grayson 1984; 75 Marshall and Pilgram 1993; Lyman 1994a, 2008; Reitz and Wing 1999; Grayson and Frey 2004). Usually, MNI, MNE, and MAU are used in conjunction with NISP because the strengths and weaknesses of derived measures are complementary to those ofNISP (Klein and Cruz-Uribe 1984). Unfortunately, due to stratigraphie problems (see Chapter 5), calculations of MNI, MNE, and MAU were not possible in this study. Since NISP, MNE and MAU do, to a large extent, provide redundant information (Grayson and Frey 2004), this should not seriously affect the results of this analysis. Nevertheless, MNI, MNE, and MAU values should be calculated in future studies of the Gatzarria fauna, in order to address the problem of interdependence. For the moment, the refit analysis carried out on the sample (see below) provides some insight into specimen interdependence in the assemblage. NSUTS The problem of differential identifiability is one that was alluded to several times in the discussion of NISP. Because not all skeletal elements are equally diagnostic, some are more likely to be left out of NISP counts than others. Examples of this are ribs and vertebrae, which, when highly fragmented, may be recognizable to element but not to species. Morin (2004) has recently proposed a new quantification measure, to be used alongside NISP and MNI counts, which accommodates these specimens and differentiates them from 'true' NISP. The Number of Specimens of Uncertain Taxonomic Status (NSUTS) is intended to include "fragments for which identification is limited to skeletal part and, sometimes, body size class, without precise information regarding taxonomic status" (Morin 2004:122). NSUTS does not include specimens that cannot be determined to skeletal element (e.g., long bone shaft fragments without diagnostic 76 landmarks), but permits sample size to be increased for certain analyses and, when combined with NISP, reduces the skewing of skeletal part abundances towards more diagnostic skeletal parts. Size classes used for many NSUTS specimens in this study are detailed in Table 4.2. Remains were sometimes identified only to size class and taxonomic class (e.g., size 3 mammal), and at other times to size class and taxonomic order (e.g., size 1 carnivore) or to size class and taxonomic family (e.g., size 2-3 cervid). One particularly common category is size 3-4 ungulate (Costamagno 1999), because horse, red deer, aurochs/bison, and giant deer bone specimens may overlap in size and when fragmented, can be difficult to identify precisely. Size Class 12 3 4 5 roe deer ibex horse aurochs/bison mammoth chamois reindeer cave lion woolly rhinoceros wolf hyaena red deer giant deer fox cave bear '_ Table 4.2: Size classes used in the faunal analysis (modified from Morin 2004). Refits The problem of differential identifiability of bone fragments can, to some extent, also be remedied through refitting studies. Refitting is the process of piecing back together specimens belonging to the same element (fracture refits), or the reassembly of elements belonging to the same individual (anatomical refits). Although refitting studies have a number of analytical advantages (Hofman 1992), two reasons for carrying out refit studies in bone assemblages are addressed here. First, by piecing together small 77 fragments into larger units, refitting can increase the number of identified bones in an assemblage, and therefore decrease identification biases, particularly those against long bone shafts (Marean and Kim 1998). Secondly, through the analysis of the provenience of refitted fragments, the problem of interdependence in NISP and the stratigraphie integrity of a site can be assessed (Morin et al. 2005). Although refitting, especially total systematic refitting of an assemblage, is a time-consuming process (Marean and Kim 1998), even partial refitting of an assemblage can provide substantial information on occupation mixing (Morin et al. 2005). Given the problems identified in the archaeological stratigraphy of Gatzarria, a thorough study of refits across all layers should eventually be carried out in order to assess the stratigraphie integrity of the faunal assemblages. In this study, refits were carried out on all long bone shaft fragments, rib fragments, and taxonomically identified specimens in the total studied sample. Unidentified long bone shaft and rib fragments were refitted only with other specimens in the bag in which the bones were stored, which usually consisted of one demi-taille (i.e., a 5 cm deep pass) from one 33-cm2 sub-unit. Identified specimens were refitted with other fragments in the original bag, as well as with all the specimens belonging to the same element and taxon. Age and sex The analysis of prey age and sex provides important indications of site seasonality and human prey choice. Sexing techniques rely on skeletal evidence of sexual dimorphism, primarily size, but also on the presence/absence of organs, such as antlers, when these are present in only one sex. The main non-destructive aging techniques for mammalian skeletal remains involve tooth eruption/wear sequences and epiphyseal 78 fusion. Stages of antler development and fetal bone lengths may also provide information on age. Unfortunately, the collection of age and sex data is often limited by archaeological preservation. Tooth eruption sequences and visual assessments of tooth wear are subjective, but relatively simple, aging techniques (Klein and Cruz-Uribe 1984; Reitz and Wing 1999). However, the timing of tooth eruption and rates of tooth wear may vary by population and, in some species, by sex (e.g., caribou; Miller 1974). Furthermore, in archaeological contexts, teeth are often found isolated rather than in mandibles, a problem which also limits the resolution of age data. The maximum crown height of cheek teeth (deciduous or permanent) is frequently used as a standard for establishing age in worn teeth, as a substitute for complete mandibles. The use of this technique for red deer, the dominant species at Gatzarria, has been extensively discussed in the literature (Klein and Cruz- Uribe 1984; Steele 2002). Unfused or fusing epiphyses are also an important distinguishing feature of juvenile skeletons. However, once an epiphysis is fused there are usually no further changes to an element which permit the distinction of younger versus older animals. Epiphyseal age classes may provide only coarse information on mortality, especially for late-fusing elements in which the timing of fusion may be variable, and species for which the sequence and timing of fusion is poorly known (Klein and Cruz-Uribe 1984). In this study, small sample size limited the collection of age data. All NISP were assigned to either juvenile or adult classes, with adult being the default category when there was no evidence of young age (such as an unfused epiphysis). Even for the most common species in the assemblage, (red deer and bison), sample sizes of teeth were not 79 sufficiently large to warrant measurement of crown heights, but degree of wear was qualitatively assessed using the following code system, developed by Cédric Beauval, which will provide a relative assessment of age distribution: 0 = unworn, 1 = less than half of the crown is worn, 2 = more than half of the crown is worn, and 3 = the crown is completely worn. Regarding sex determination, one of the simplest techniques involves recording the presence or absence of antlers or horns in bovid and cervid species in which only males carry these appendages. However, this technique is limited to the frontal bone of these species. Seasonal loss of antlers in cervids, the collection of shed antlers for tool- making, and transport biases against heavy crania (Binford 1978) are additional serious limitations to the use of this method. Other techniques for sexing archaeological remains involve morphologic or Osteometrie differences (Weinstock 2000). However, a lack of sexual dimorphism, or overlap in the size of male and female animals (for instance, between young males and adult females), may limit the possibility of determining sex for some species. In species with antlers/horns, the additional weight on male forelimbs due to the greater load bearing of these limbs contributes to greater dimorphism in these elements (Klein and Cruz-Uribe 1984), so males and females can sometimes be distinguished osteometrically using forelimb elements, particularly the metacarpal. Sexual dimorphism in the innominate bone of cervids can be sometimes be used to determine sex as well (e.g., Edwards et al. 1982). Unfortunately, fragmentation and the destruction of low density bone portions through burning, carnivore ravaging, marrow extraction, or bone grease production, for example, may prevent the collection of the required observations and measurements. 80 Climate control In any archaeological analysis of hunter-gatherer subsistence, it is important to consider that the availability of animals on the landscape may be strongly conditioned by both the regional climate and more specific local factors, such as topography and humidity. In order to understand temporal changes in the range ofprey selected by hunter-gatherer groups, as opposed to environmentally-driven change, a tight control of local climate changes is required. Climatic indicators from an archaeological site itself probably represent the only strictly contemporaneous records of climate. However, human-hunted faunas may not always be a reliable indicator of climate change, for a number of reasons. First, assemblage composition may not always be affected by environmental change. For example, some species that frequently dominate assemblages from the Palaeolithic in Europe, such as bison and red deer, may tolerate a wide range of environmental conditions (Rivals et al. 2007; Banks et al. 2008). Secondly, changes in the composition of human-hunted faunal assemblages do not necessarily reflect climatic change. Resource depression, variability in site seasonality or site function, technological innovation, or changes in the characteristics of prey are among the variables that may affect the range of animals included in the diet (Grayson and Cannon 1999), and perhaps more importantly, whether those prey are transported to a particular site. Consequently, an independent source of climatic information is necessary. Pollen sequences from karstic cave sites in France, Iberia, and the Mediterranean region are also frequently beset with problems which may render palaeoclimatic indications from these sites inaccurate (Turner and Hannon 1988; Straus 1990; Carrion et al. 1999; Woodward and Goldberg 2001). Besides general problems that affect all pollen 81 studies, such as variable preservation and differences in pollen productivity and weight between vegetation types, cave pollen records may be unreliable due to discontinuous deposition ofpollen, low concentrations of pollen, reworking of sediments, percolation of pollen through layers, transport ofpollen by animals, and biased deposition of pollen due to cave orientation and prevailing wind direction (Turner and Harmon 1988; Straus 1990; Carrion et al. 1999; Woodward and Goldberg 2001). The frequent difficulties associated with cave pollen records leaves environmentally sensitive and rapidly evolving microfauna as one of the best local palaeoenvironmental indicators for cave deposits (Straus 1990). When microfaunal remains are not associated with human foraging activities, they represent an independent source of climatic information from the human-accumulated deposit (Morin 2004). Identification of micromammals should therefore be carried out in conjunction with macrofaunal zooarchaeological analyses. When present, naturally-introduced birds and amphibians may also represent a good proxy for local climate conditions. Unfortunately, due to time constraints and a lack of required expertise, micromammals, birds, and amphibians were not identified during this study. However, because this study is not examining temporal change at Gatzarria, a monitor of climate change through time is not as essential, although these data would still be useful for site comparisons. Lavaud's (1980) data on microfauna in the Cj deposits are too sparse to provide useful information on climatic conditions (Table 3.3). Bone modification and taphonomy A multitude of factors—anthropic, biological, chemical, and physical—affect skeletal remains between the moment an animal is killed and the moment the remaining 82 fragments are analyzed by an archaeologist. Without an understanding of these factors, inferences regarding human behaviours may be distorted. Burning Burnt specimens are a good indicator of human involvement in the formation of a bone assemblage (Stiner et al. 1995; Villa et al. 2004). Anthropogenic burning ofbones can result from cooking of meat, accidental scorching of bones due to proximity to a fire, outright disposal of bone in fires, the deliberate use ofbone as combustible, or cremation (Lyman 1994b). Non-anthropogenic burning may result from natural fires or in situ burning of heavily organic sediment matrix (Lyman 1994b), and may result in less intense burning than intentional fires (Stiner et al. 1995). Burning of bones due to human activity may also be indicated by discrete hearths (Villa et al. 2004), the co-occurrence of cut marks and burning, or regular anatomical patterns of burning (Lyman 1994b). Colour is often used as an indicator of burning intensity (Lyman 1994b; Stiner et al. 1995; Mayne Correia 1997). Burnt bone is generally brownish when exposed to light burning, black if carbonized, and shades from black to gray then white, sometimes passing through blue or green, as burning intensity increases to calcination (Stiner et al. 1995; Mayne Correia 1997). Susceptibility to fragmentation increases with the degree of burning, so fragment size is another important variable regarding burning intensity (Stiner et al. 1995). The use of bone as a combustible material appears to have occurred frequently in the European Palaeolithic (Castel 1999; Costamagno et al. 1999; Théry-Parisot 2002; Morin 2004, 2009; Villa et al. 2004). This process results in extensive fragmentation and, in some cases, analytical destruction of greasy bone portions, namely spongy bone 83 portions such as long bone epiphyses and vertebrae (Costamagno et al. 1999). It is therefore essential to examine unidentified specimens as well as identified specimens for evidence of burning (Castel 1999; Morin 2004; Villa et al. 2004). The use of bone as combustible may have influenced the skeletal transport decisions of past human groups (Costamagno et al. 1999). The colour ofburnt bones (brown, black, white, grey, blue, or green) was recorded for bones in the NISP and NSUTS samples. All unidentified burnt bone specimens were counted and the greatest dimension measured. This will provide a measure of bone fragmentation in relation to burning, and may be informative about collection techniques used during excavation. The longest dimension was chosen for measurement for simplicity and consistency in measuring fragments of varying shapes. Fractures Despite early hopes that bone fracture patterns could be used as a direct indicator of particular hominid activities (e.g., Dart 1957), it is now abundantly clear that identification of the agent of bone breakage in an archaeological assemblage can rarely be made solely on the basis of fracture 'types' (Binford 1981; Lyman 1994b). In fact, fracture type categories are difficult to clearly delineate due to the complexity of breakage patterns (Lyman 1994b). Although attempts have been made to distinguish carnivore and human fracturing of marrow bones using mean fracture angles, there is considerable overlap in the angles produced by the types of force-loading characteristic of both agents (see Pickering et al. 2005:253-254, Figure 3). Nevertheless, general inferences concerning the timing of bone fracturing (pre- versus post-depositional) is important for understanding both human behaviours and assemblage formation processes. 84 In this study, fractures were recorded in several ways, following Morin (2004; but see also Morían 1983). First of all, highly unambiguous fractures on compact bone were classified as green or dry, based on the shape (spiral vs. tranverse), surface texture (smooth vs. irregular), and sometimes the colour of the fracture surface, which should match the exterior colour of the bone surface in green bone fractures, but may not necessarily match for dry bone fractures. Fracture shape was also observed and assigned to one of the following categories: angular, curved, irregular, oblique, ragged, transverse, or v-shaped. Only proximal and distal fractures were recorded (Villa and Mahieu 1991). Finally, the circumference of long bone shaft fragments was recorded on a simple scale (Bunn 1983; Villa and Mahieu 1991): 1 = less than half of the original circumference, 2 = more than half, and 3 = complete. This measure is also useful for considering fragmentation. Fragmentation Bone fragmentation, like fracture patterns, can be an indicator of the behaviour of humañs and animals at archaeological sites, as well as of taphonomic disturbance. For instance, human activities such as marrow extraction tend to fragment long bones intensively, while carnivores concentrate their activities on the ends of long bones and therefore tend to leave longer fragments (Binford 1981; Villa et al. 2004). In this analysis, several measures of fragmentation were used. First, the maximum length and maximum width were measured for all NISP. The length of long bone shaft fragments was also coded as a fraction of the total shaft length: 1 = less than one quarter of the length of a complete shaft, 2 = less than half, 3 = less than three quarters, and 4 = three quarters to complete (Villa and Mahieu 1991). Lengths of all burnt bone 85 specimens in the assemblage, identified and unidentified, were measured. Finally, a random subsample of unidentified, unburnt specimens from the total study sample was selected for measurement, as a basis for size comparisons with burnt specimens and identified specimens. Because fracturing and fragmentation are the norm rather than the exception in archaeological assemblages, a system to identify which portion of a bone is represented by a fragment also is required. These systems can be quite complex. For instance, some analysts (e.g., Nagaoka 2006) record the presence/absence of density scan sites on each element. A simple system that provides adequate information to assess part representation in long bones is to record whether the fragment represents the proximal epiphysis, proximal shaft, mid-shaft, distal shaft, and/or distal epiphysis (Marean and Spencer 1991). The latter procedure was employed in this study. The presence of important bony landmarks (e.g., the anterior groove of the metapodials) was recorded when applicable. Butcheryand tool use Although cut marks are related to processes of skinning, dismemberment, and meat removal, their meaning in terms of intensity of butchery or culture-bound butchery patterns is ambiguous. The frequency and placement of marks may be related to a number of contingent variables such as technology, time of day, the experience of the butcher, and condition of the carcass (Lyman 2005). However, when positively identified, stone tool marks are an unambiguous indicator of human involvement in assemblage formation (Lyman 1994b; Villa et al. 2004), and are therefore an important factor for taphonomic consideration. Since the anthropogenic contribution to assemblage formation is a primary focus of the taphonomic analysis carried out here, only the 86 presence or absence of tool marks on specimens was noted. In addition, cutmarks (short incisions created by slicing with a tool) were distinguished from scrape marks (long, parallel marks created by dragging the edge of a tool sideways along the bone surface). As discussed earlier, traces of marrow extraction are important for assessing foraging return rates. Unfortunately, it is often difficult to distinguish percussion, or human-induced, fracturing from carnivore-induced notches in archaeological material. Capaldo and Blumenschine (1994) suggest that the shape of impact notches can be used to statistically distinguish hammerstone and carnivore-tooth fractures, due to the dynamic versus static nature of force loading involved in these two processes. Unfortunately, the molding method they use to measure notches is not practical for the analysis of large collections. Furthermore, the pattern they observe applies only weakly to large bovids (1 15-340 kg), and this problem probably holds for other large ungulates as well. In this analysis, the presence of percussion pits and notches was recorded, but they are not considered to be strictly diagnostic of anthropogenic fracturing. In Europe, the use of bone as a raw material for tool manufacture takes on particular importance from the Upper Palaeolithic onwards. Antler, in particular, was an important source of raw material for bone tools in the Upper Palaeolithic, and as such should be eliminated from skeletal part frequencies when considering subsistence activities (Lyman 1994b). Formal bone tools (such as points or awls) are generally easy to identify, but the use of minimally modified bones as ad hoc or expedient tools is more difficult to discern (Lyman 1994b). In European Palaeolithic contexts, including the Middle Palaeolithic (Chase 1990), the most important ad hoc bone tools are retouchers (retouchoirs). These tools are characterized by repeated indentations in discrete zones on 87 the bone surface and may have been used in retouching of stone tools. Retouchers often may occur on long bone shaft fragments (Chase 1990), as well as on long bone epiphyses, phalanges, scapulae and occasionally teeth (Valensi 2002a, 2002b). Morin (2004) suggests that retouchers show a high frequency of scrape marks, which may indicate the removal of the periosteum prior to tool use. No formal bone tools or worked antler were found in the Cj sample. Retouchers that were identified to element and taxon were included in NISP. Animal activity Carnivores, especially hyaenas, were important agents of assemblage formation and transformation during the Palaeolithic. The role of carnivores in assemblage formation has been a long-standing controversy among archaeologists interested in the origin of hunting. Consequently, the effect of large carnivores on bone assemblages has been well studied, both archaeologically and in actualistic studies (e.g., Binford 1981; Marean and Spencer 1991; Blumenschine 1995; Domínguez-Rodrigo 1997). Some large carnivores, such as wolves and hyaenas, may have a significant, and frequently diagnostic, impact on bone assemblages. Carnivore gnawing produces punctures, pits, scoring, furrowing, crushing of spongy bone, and crenulated bone edges (Binford 1981). Since marrow and grease are the most nutritionally attractive bone portions (after the flesh has been consumed), carnivores focus their attention on extracting these substances. This may result in the destruction of long bone ends, especially in hyaena-ravaged assemblages (Blumenschine 1988; Marean and Spencer 1991), and in a high frequency of bone cylinders or 'scooped out' long bones (Binford 88 1981). Some carnivores, particularly hyaenas, will frequently ingest bone, resulting in digested bone that appears corroded (Lyman 1994b). Like their larger relatives, small carnivores, such as foxes, accumulate and modify bone assemblages, although they have little impact on large prey. Many fox remains are present at Gatzarria, especially in the Early Aurignacian (Table 3.2). The contribution of foxes to an assemblage as accumulators, rather than as prey, may be indicated by gnaw marks on animals such as hares, ground squirrels, and some birds (Chauvière and Castel 2004). Lyman (1994b) suggests that small carnivore assemblages are characterized by extensive fragmentation (fragments generally <1 cm) of the bones of their small prey and by high frequencies of isolated teeth and broken mandibles. The pits, grooves, and puncture marks made by foxes are smaller than those of large carnivores (Lyman 1994b; Chauvière and Castel 2004). Animals other than carnivores may also deposit or alter bone assemblages. Rodents, such as porcupines, are known to gnaw and transport bones. Rodent gnawing is typically conspicuous due to the distinctive shape of their incisors (Binford 1981). In some cases, carnivorous birds may disturb or accumulate bone assemblages (Lyman 1994b). For instance, Eurasian Eagle Owls (Bubo bubo) are known to have produced lagomorph accumulations during the Palaeolithic in the Iberian peninsula (Yravedra Sainz de los . Terreros 2004). Raptor involvement in assemblage accumulation may be signaled by fracture patterns restricted to a certain size range of prey (particularly rodents), by patterns of bone loss, and by traces of digestion lighter than those produced by mammalian carnivores (Lyman 1994b; Yravedra Sainz de los Terreros 2004). Bearded vultures, which are present in the Pyrenees, are also known to frequent caves and produce 89 bone assemblages (Marín Arroyo et al. 2009). Bearded vulture action may be suggested by a presence of heavily digested bone without gnaw marks, particularly small and medium-sized ungulate phalanges (Marin Arroyo et al. 2009). The impact of animals on the Gatzarria archaeological assemblage was recorded as the presence/absence and extent of punctures, grooves, tooth pits, scooping, ragged edges, tooth notches, and traces of digestion. Data on fragment size, completeness, and circumference (already discussed) is also required for understanding animal activity. Villa et al. (2004) suggest that a comparison of long bone shaft lengths with shaft circumferences provides a diagnostic measure of human versus carnivore assemblages, as humans tend to produce assemblages dominated by short fragments representing less than one half of the original shaft circumference, while carnivores tend to produce a greater proportion of long fragments more complete in circumference (see also Binford 1981). The frequency of carnivore remains, particularly juvenile remains, is an additional indicator of the significance of carnivore use of a site (Villa et al. 2004). When studying assemblages severely impacted by carnivore action or in situ attrition, a focus on the identification and quantification of long bone shafts may often provide the highest MNE (and NISP) values, because carnivores are less attracted to these bone portions (Bunn and Kroll 1 986; Marean and Spencer 1 99 1 ) . Additional taphonomc modifications Post-depositional modifications to bone may affect whether traces of human and animal behaviour such as butchery and gnaw marks are preserved. The alterations recorded during this analysis include manganese staining, cracking, scratching, abrasion, calcification, sheeting, and exfoliation. Sheeting involves fracturing of cortical bone into 90 sheets usually parallel to the cortical surface, while exfoliation is the loss of small flakes from the cortical surface. This type of damage is more commonly found on foetal or very young rather than adult bone specimens (Morin 2004). Behrensmeyer's (1978) weathering stages were not used in this study, because the types of cracking, flaking, and splintering used to define her weathering stages were infrequently observed. This may be due to the cave environment which may have offered more stable temperature and moisture conditions in comparison to the southern Kenyan climate to which the bones in Behrensmeyer's experiment were exposed (Morin 2004). Besides these specific alterations, overall bone surface condition was observed and assigned to one of four categories: intact, relatively good, somewhat damaged, or poor condition (Morin 2004). An estimate of the percentage (by 10% intervals) of the original bone surface that could be observed was also recorded. The condition of proximal and distal long bone fracture edges was also observed and classified as fresh, slightly abraded, abraded, or very abraded, according to the definitions used by Paola Villa in her analysis of faunal remains from Bois Roche (Villa et al. 2004). The relationship between surface damage and the preservation of marks from burning, butchery, and carnivore ravaging is explored in Chapter 5. Summary The theoretical perspective adopted in this study is that of foraging theory, specifically, the diet breadth model, marginal value theorem and the central place forager prey choice model. These models will assist in interpreting the subsistence behaviours of Mousterian foragers from Gatzarria. Other studies employing foraging theory have recently provided interesting insights into Palaeolithic population dynamics (e.g., Stiner 91 et al. 2000; Munro 2004; Morin 2008). However, in order for studies applying behavioural ecology to provide meaningful results, detailed taphonomic and quantitative analyses of archaeological faunal remains are necessary. The methods required to operationalize this approach, and used in the analysis of the Gatzarria fauna, have been presented in this chapter. The following chapter will examine the stratigraphie and taphonomic integrity of the study sample. Chapter 5: Sample selection and taphqnqmy A number ofpotential problems with the archaeological stratigraphy in layer Cj at Gatzarria were identified in Chapter 3. The present chapter delves deeper into stratigraphie issues encountered during the analysis of the faunal remains, and analyzes the stratigraphie distribution of the Mousterian industry in layer Cj. Subsequently, a taphonomic analysis, an essential prerequisite for any interpretation of archaeological faunal remains, is carried out in order to assess the biases that may have been introduced into the assemblage, through human and carnivore action, post-depositional processes, and recovery methods. Sample selection A robust interpretation of the subsistence behaviours of Neandertals at Gatzarria requires a clean division between the Upper Palaeolithic (Châtelperronian and Proto- Aurignacian) layers and the Middle Palaeolithic (Mousterian) layer of interest. Both Middle and Upper Palaeolithic deposits are found within the homogeneous sediments of the middle ensemble Ej. Previous publications (Lavaud 1980; Laplace and Saenz de Buruaga 2003; Sáenz de Buruga 1991) have not presented extensive details on the archaeological stratigraphy of Ej. Although published stratigraphie profiles (e.g., Figure 3.4) suggest that the archaeological levels are more or less well sub-divided, Sáenz de Buruaga (1991) alludes to a number of disquieting issues. These include possible mixing of Middle and Upper Palaeolithic artifacts in some parts of the cave due to water run-off, 93 and the reassignment of lithic materials originally assigned to the Cj Mousterian to the Châtelperronian (Cjn3) subsequent to excavation on the basis of morphological, technological, and/or raw material characteristics. These problems raise the question of the integrity of the archaeological levels in Ej. Stratigraphie integrity ofMousterian remainsfrom Ej This study began with all faunal remains from décapages labeled Cj, Cjm, and Cjmg, with the goal of isolating the Mousterian component of Ej. However, during initial sorting of the faunal material, it was observed that there was very little material from Cjn'l, Cjn2, and Cjn3, whereas there was a large amount of material in Cj. This suggested that the spatial extent or thickness of the archaeological levels varied significantly, or that there might have been some discrepancies in the recognition of archaeological layers within the sedimentary layer Cj during excavation or the a posteriori analysis. In order to examine these possibilities, a database of all the excavated décapages, based on the original excavation notebooks, was created. This information confirmed the suspicion that the archaeological levels Cjnl, Cjn2, and Cjn3 were identified in only a few units (Figure 5.1). In fact, the Châtelperronian (Cjn3) was only identified in four of the 29 units in which the Ensemblejaune was excavated. To address this complication, a cursory analysis of piece-plotted lithic artifacts in Ej was carried out with the help of André Morala (Musée National de Préhistoire, France) and Eugène Morin. Lithic artifacts in boxes labeled Cjnl, Cjn2, Cjn3, and Cj Mousterian were examined. In this sample, diagnostic elements of Aurignacian complexes (e.g., Dufour bladelets), the Châtelperronian (e.g., Châtelperronian points), and the Mousterian (e.g., Levallois cores) were identified by André Morala. Additional data on the Proto- 10 6 5 4 3 2 1 21 D No data on Cj D Cj only recognized 22 ¦ Cjnl and/or Cjn2 recognized D ¦ Cjn3 recognized Figure 5.1: Original identification of layers within Ej during excavations at Gatzarria. Cjnl and Cjn2 are Proto-Aurignacian layers; Cjn3 is the Châtelperronian. In this figure, Cj refers to the sedimentary, not the archaeological, level. Row G was excavated along only one third of the units. Aurignacian lithics were contributed by Laura Eizenberg (Université de Toulouse). This resulted in a database of 424 diagnostic lithic artifacts with three-dimensional provenience. This brief analysis made it clear that the majority of the Upper Palaeolithic artifacts in Ej had initially been assigned to sedimentary layer Cj (demonstrated by ink labeling of lithic pieces), and that it was only subsequent to excavation that most pieces were reattributed to finer chronological sub-divisions—the archaeological 'layers' Cjnl, Cjn2, and Cjn3. It was also observed that many objects assigned to the Châtelperronian appeared tobe Mousterian in nature. This may result from Laplace's assumptions about the Châtelperronian, which he considered to have originated in the Mousterian and 95 evolved into the Proto-Aurignacian. However, were Laplace's attributions consistent with the stratigraphie distribution of the objects? This question was examined by plotting the collected data points in three- dimensional rotary plots Using Data Desk 6.2.1. Two-dimensional sagittal projections of these plots are shown in Figure 5.2. Figure 5.2(a) shows the distribution of the Proto- Aurignacian and Châtelperronian industries according to Laplace's attributions, while Figure 5.2(b) shows the distribution of the lithic industries in Ej following Morala's reclassification. In Figure 5.2(a), Upper Palaeolithic artifacts are present quite deep in the stratigraphie profile. A comparison with Figure 5.2(b) shows that the majority of these objects are in fact characteristic of the Mousterian. In the latter plot, the distribution of artifacts more closely replicates the expected top-to-bottom sequence of lithic industries: Mousterian, then Châtelperronian, and finally Proto-Aurignacian. The discrepancies between these plots suggest that Laplace's reattributions were not entirely consistent with stratigraphie divisions. Despite some overlap between the industries, the new attributions shown in Figure 5.2(b) demonstrate that there is a general vertical distribution of the archaeological industries in Ej, which will help redefine the limits of the archaeostratigraphic units. Our analysis also suggests that the black flint that Laplace considered to be characteristic of the Châtelperronian at Gatzarria (Saenz de Buruga 1991) may not be a valid criterion for determining the limits of the Châtelperronian because many probable Mousterian artifacts were made of this material—i.e., the same raw materials were probably used in both occupations. Furthermore, no clear evidence of the channeling mentioned by Saenz de Buruaga (1991) is apparent in the sagittal profiles in Figure 5.2. The distribution of lithic artifacts 96 a) o -20 -40 ó> o -60 -80 °°J -100 r i>o ¦ ?" e« .5 -120 °§ °o?l °° -140 e -160 ? ex Q b) o -20 -40 ,& -60 SP Front of cave <- -> Back of cave O Upper Palaeolithic (Proto-Aurignacian) H Châtelperronian Mk Mousterian Figure 5.2: Two-dimensional projections of lithic artifacts from Ej in sagittal profile, (a) distribution of the Proto-Aurignacian and Châtelperronian according to Laplace; (b) distribution of the Proto-Aurignacian, Châtelperronian, and Mousterian according to this study. Note that the cave floor slopes towards the back of the cave. identified in this study is shown once again, this time in frontal profile, in Figure 5.3. Only artifacts from sections 2-6 are shown in this diagram, due to the steeper slope in sections at the rear of the cave (particularly rows 7 and higher), and a lack of relevant data for sections 1 and 22-21. This figure shows a relatively uniform horizontal 97 -10 -30 -50 o o S O ? A ? ? ? -110 ? ? ?? O ? ^B ? ? ? O ? A -130 ? ? A ? ?? ? ?a ? # ? , -150 À D 0 Upper Palaeolithic (Proto-Aurignacian) 1 Chàtelperronian A Mousterian Figure 5.3: Two-dimensional projection of lithic artifacts from Ej in frontal profile (attributions from this study). Only sections 2-6 included in this figure. separation of the industries, with limited evidence of deeper channels containing Upper Palaeolithic artifacts or areas where the Middle Palaeolithic is relatively closer to datum level. The slope of the cave floor explains much of the overlap between the industries seen in this profile (for instance, the deepest Upper Palaeolithic points in sectors E and F). This suggests that the channeling noted by Sáenz de Buruaga did not affect the units included in this study, although this problem may be present elsewhere in the cave (e.g., in the units not considered because insufficient data was available). Sample selection procedures Having established that archaeostratigraphic units Cjnl, Cjn2, Cjn3, and Cj Mousterian were often not differentiated during excavations, and that there appears to be some overlap between the archaeological industries in Ej, the composition of the Cj 98 faunal sample needs to be reconsidered. For a reliable analysis of the Mousterian component of Cj, Upper Palaeolithic (Proto-Aurignacian and Châtelperronian) elements in the Cj faunal sample should be eliminated. In addressing this problem, a number of units were immediately removed from consideration because insufficient lithic data was available to assess the stratigraphie distribution of the artifacts. A further three units (6E, 6F, and 7E) were eliminated because the lithic industries showed little vertical separation and appeared to be mixed. Poor vertical separation of the industries appears to be the general case in the rear part of the cave (row 6 and higher). This pattern may be a result of several factors: (1) the strong slope of the cave floor in this part of the cave; (2) the occurrence of badger burrows in this area; (3) mid-twentieth-century digging during an attempt to unblock the tunnel in the rear right corner of the cave (Saenz de Buruaga 1991). For the remaining units, the three-dimensional plot of the lithic attributions shown in Figures 5.2(b) and 5.3 was used to determine an approximate lower limit of Upper Palaeolithic industries on a unit-by-unit basis. A moderately conservative approach to elimination of décapages was adopted, in order to preserve a sufficiently large sample size for analysis. The selected units and depth cut-offs are shown in Figure 5.4. All faunal materials found below these depth cut-offs are used in the zooarchaeological analysis. This sample will be referred to as the Cj Mousterian sample. Materials below the unit depth cut-offs belonging to Cjm and Cjmg are also included in the sample, because these represent small manganese lenses within Cj. It should be noted, however, that not all subunits in each unit were excavated to the same depth, or excavated at all, so Figure 5.4 is not a representation of the total surface area analyzed. 99 D No data on Cj El Units included in subsample B Units eliminated (no stratigraphie separation) M Units eliminated (insufficient data) Figure 5.4: Summary of the distribution of the Cj Mousterian sample. Numbers indicate the depth cut-off point (in cm) for each unit included in the sample. Only décapages below the indicated depth are included in the subsample. Although some Upper Palaeolithic material may remain in the Cj Mousterian sample, this strategy should have removed the large majority. Given that ivory and antler working are unknown in the Middle Palaeolithic, a low representation of ivory and antler and an absence of worked antler or ivory pieces in the subsample would support this inference. All but one fragment of antler, and 80% of ivory NISP, including the only incised ivory fragment in the assemblage, were removed from the Mousterian sample by the selection procedure outlined above. These observations support the inference that Upper Palaeolithic materials are mainly restricted to the upper portion of sedimentary layer Cj, and that the selected subsample is more appropriate than the original archaeostratigraphic units for the analysis of Mousterian fauna from Ej. 100 Stratigraphie integrity: Refit analysis The degree of stratigraphie mixing can also be assessed using bone refits. A total of 45 refit sets were found during the analysis of the Cj fauna. All but four of these sets are composed of two bone fragments. The remaining four sets contain three fragments each. A total of 27 refit sets remained in the Cj Mousterian sample after the removal of questionable units and décapages. No refit sets were found between the Mousterian subsample and the eliminated décapages, but the implications of this observation are limited, as the majority of refits are on post-depositional dry-bone fractures. Of the 45 refit sets, 41 are on dry-bone fractures. Intra-décapage refits on fractured ivory and single teeth are not included in the refit sets, because green- versus dry-bone fractures cannot be differentiated in these materials, as they are brittle and prone to fracturing. Green-bone fracture and anatomical refits may be more informative about site formation processes than dry-bone fractures, which occur post-depositionally (Morin et al. 2005). Only one refit is on a green-bone fracture, while two are anatomical refits. The single green-bone refit identified is a red deer metatarsal, located in 4E(6) and 4E(9) 105-1 10 cm (Figure 5.5). Both of the anatomical refits are found in the eliminated décapages. One is a radius fragment of red deer refitted with an ulna fragment, and the other consists ofjuvenile red deer third and fourth lower deciduous premolars. The anatomically-refitted specimens were found in the same décapages. For dry-bone refits, 38 of 41 sets were found within the same décapages of the same unit. The remaining three refit sets are inter-décapage refits. The first inter- décapage refit set is a dry-bone refit on a bison rib. The fragments were found in adjacent décapages (1 10-115 cm and 1 15-120 cm) of adjacent sub-units (6D(7) and 6D(8) 101 ^Èmmm fP^.4V.rii wsêmà Figure 5.5: Green-bone refit on a red deer metatarsal respectively). The second set is a dry-bone refit on a bison humerus. These fragments were located in adjacent décapages (105-1 10 cm and 110-115 cm) of subunit 4E(6). The third inter-décapage refit, a dry-bone refit on a fox pelvis, was found in adjacent décapages of the same unit (4D(6) 70-80 cm and 80-90 cm). None of these refit sets seems to have moved more than a few centimetres after breakage. A final, highly noteworthy, refit set is an inter-décapage refit consisting of two mammoth molar blade fragments located in 3F(9) 105-1 10 cm and 4G(7) 1 15-120 cm. This is the only refit for which the fragments were not from the same or adjacent sub- units. The tooth pieces are separated vertically by a minimum of five and a maximum of 15 cm, and horizontally by approximately one metre. This refit is interesting, because the tooth fragment from unit 3F was labeled 'Cj', and then relabeled 'Cjn2-3.' It is the only faunal specimen encountered that provides direct evidence of the layer reattributions carried out by Laplace, who focused mainly on lithic artifacts. An examination of the lithic database provides interesting detail on this reattribution. The décapage yielding this fragment contains a number of characteristic Mousterian artifacts such as Levallois nuclei and flakes. No evidence of Upper Palaeolithic industries was found in this décapage during our analysis. Furthermore, the fragment from unit 4G, to which this piece was refitted, was not reattributed to Cjn2-3. All this suggests that indeed these pieces are part of a Mousterian deposit. Unfortunately, this refit could not be classified as a green- or dry-bone fracture, so its meaning in terms of post-depositional movement is ambiguous. The sample size of refits is quite small. However, all refits are on fragments found in close proximity to one another. No positive evidence of significant mixing of faunal remains in Cj is present, although very little faunal material from sections 6 through 10— where lithic remains in some units appeared to be mixed (Figure 5.2)—was available for analysis. Further refit studies at Gatzarria, including materials from the Cjnl-3 faunal samples, will eventually provide greater detail on the integrity of archaeological layers at Gatzarria. Overview ofthe Cj Mousterianfaunal sample All remains from sedimentary levels Gj and Cjm/Cjmg were initially analyzed, including remains from décapages where Upper Palaeolithic artifacts were present, where the occupations may have been mixed, or where there was limited evidence for the distribution of the lithic industries. This sample consists of 1 1 162 specimens, 630 of which are taxonomically-identified at least to the genus level (after refitting). The smaller Cj Mousterian sample consists of a total of 4 907 specimens, including 376 identified specimens (after refitting) (Table 5.1). The sample to be used in the zooarchaeological analysis therefore represents 44% of the initially studied fauna, although the NISP and NSUTS sample represents 59% of the original material. 103 ______Initiai sample Cj Moust sample NISP 630 376 NSUTS 119 67 indeterminate long bone shaft fragments 572 318 indeterminate rib fragments 45 1 5 unidentified specimens 9796 4131 Total 11162 4907 Table 5.1: Summary of studied faunal remains. The initial sample includes all remains from sedimentary levels Cj and Cjm/Cjmg, whereas the Cj Mousterian sample incorporates only materials remaining after the elimination of décapages containing Upper Palaeolithic artifacts or where there was insufficient data on the stratigraphie distribution of lithic industries. Indeterminate long bone shaft and rib fragments includes only those specimens that could be not be assigned to a more precise anatomical or taxonomic category. The NISP samples are presented in Table 5.2. Taxonomic abundances in the Cj Mousterian sample and the eliminated portion of the sample (i.e., the initial sample minus the Cj Mousterian sample) are highly correlated (rs = 0.90, ? < 0.001), indicating that differences between the Cj Mousterian and the eliminated material are minimal. This suggests that the reduction in the sample size did not significantly affect species representation. However, the most abundant species, red deer, is relatively more frequent in the Mousterian sample, while large bovines (aurochs or bison) are less abundant. Mammoth remains are also less common, due to the removal of most of the ivory fragments. Similarly, fox are less well-represented, as numerous fox teeth were removed from the sample. Chamois are slightly more common in the Mousterian sample. The taxonomic attribution of three species requires some elaboration. The distinction between aurochs (Bos primigenius) and steppe bison {Bison priscus) is a problem that plagues virtually all Pleistocene faunal analyses in Europe. At the sample level, the dominance of one or the other of the two species can be estimated using tooth morphology and a variety of osteological characters and measurements (e.g., Brugal 1983; Slott-Moller 1988). Unfortunately, the Bos/Bison sample analyzed here is Species Common Name Initial NISP % Cj Moust NISP % Cervus elaphus red deer 363 57.6 268 71.3 Bos primigenius/Bison priscus aurochs or bison 89 14.1 28 7.4 Mammuthus primigenius mammoth 36 5.7 10 2.7 Rupicapra sp. chamois 35 5.6 26 6.9 Equus caballus horse 29 4.6 1 1 2.9 Vulpes vulpes/Alopex lagopus red or arctic fox 19 3.0 2 0.5 Capreolus capreolus roe deer 18 2.9 12 3.2 Capra sp. ibex 14 2.2 8 2.1 Megaloceros giganteus giant deer 6 1.0 2 0.5 Rangifer tarandus reindeer 6 1.0 3 0.8 Ursus spelaeus cave bear 6 1.0 3 0.8 Crocuta crocuta hyaena 4 0.6 1 0.3 Canis lupus wolf 3 0.5 0 0.0 Coelodonta antiquitatis woolly rhinoceros 1 0.2 1 0.3 Panthera leo lion 1 0.2 1 0.3 Total 630 100.0 376 100.0 Table 5.2: Identified taxa and NISP in the initial and the Cj Mousterian samples. small and very fragmented. Most skeletal parts are represented by only a few specimens. In addition, complete teeth are infrequent and epiphyseal fragments are virtually absent. Because of the inadequacy of the sample, as well as time constraints, the relevant observations needed to characterize the Gatzarria Mousterian bovids as bison or aurochs were not collected. For these reasons, Bos/Bison will be referred to collectively as large bovines. Secondly, it is not clear how or when the modern species C. ibex and C. pyrenaicQ arose (Acevedo and Cassinello 2009; Parrini et al. 2009). The species of Capra present at Gatzarria during the Mousterian might be C. pyrenaica, C. caucásica, or C. camburgensis. The question of whether the Pyrenean chamois is a separate species {Rupicapra pyrenaica) from the Alpine chamois (R. rupicapra) is also unresolved (T. Pérez et al. 2002; Crestanello et al. 2009; Rodríguez et al. 2009). Ungulates compose 98.1% of the Mousterian sample, with carnivores constituting only 1.9%. Red deer is by far the most common species, with 71.3% ofNISP, distantly followed by bison (7.4%) and chamois (6.9%). No lagomorph or fish remains were identified. Although birds were not taxonomically identified in this study due to my lack of familiarity with these taxa, very few bird remains were encountered during the faunal analysis. For the same reason, microfauna and amphibians are also not included. It is unlikely that the small rodents and amphibians would have constituted human diet items, due to their small size, which would give them a very low-ranking according to the body size rule (see Chapter 4), as well as the fact that there is virtually no evidence for the consumption of these animals during the Middle Palaeolithic of Europe. Taphonomy As discussed in Chapter 4, a number of processes may intervene to alter the composition of faunal assemblages between the arrival of a carcass, or parts of a carcass, at a living site and eventual zooarchaeological analysis. Some of these biases are post- depositional, while some occur at the point of analysis. Additionally, some behaviours, such as burning of bone or grease extraction, may obscure evidence of prior subsistence decisions through destruction and fragmentation of bone (Costamagno et al. 1999; Morin 2009). The following sections consider taphonomic factors relevant to the interpretation of the faunal sample, including recovery methods, fragmentation and burning, post- depositional destruction, carnivore ravaging, surface preservation, and identification biases. For most analyses, the Mousterian sample is preferred because of its more stratigraphically- and archaeologically-secure nature. However, in cases where the Mousterian sample provided inadequate sample size, the entire studied sample is used. How did recovery methods affect the sample composition? All excavated sediments at Gatzarria were sieved; but mesh size was not kept constant over the years, nor were mesh sizes recorded in every excavation report (Table 5.3). However, the role of small unidentifiable fragments in providing valuable taphonomic information has received substantial attention in some recent zooarchaeological analyses (e.g., Outram 2001; Morin 2004). In particular, recovery methods may affect the composition of the sample with regards to the frequency of burning, because burnt remains are often highly fragmented (Villa et al. 2004). Therefore, the impact of recovery techniques on abundances of small fragments should be considered. Year Top mesh Bottom mesh 1961 5 mm 3.8 mm 1962 5 mm 3.8 mm or 3.3 mm 1963 Not specified 1964 Not specified 1965 5 mm 2 mm 1966 5 mm 2 mm 1967 5 mm 2 mm 1968 5 mm 2 mm 1969 5 mm 2 mm 1970 5 mm 2 mm 1971 Not specified 1972 Not specified 1 973 Not specified 1974 Not specified 1 975 Not specified 1976 Not specified Table 5.3: Mesh sizes used in excavations at Gatzarria. In 1961 and 1962, mesh sizes were originally recorded as size 4 and size 7 or 9, referring to the number of openings per cm2. If sieve sizes were the primary factor responsible for the differential recovery of small bone fragments, different mesh sizes alone could pose difficulties for analysis of the small fraction of the sample. However, at Gatzarria, selective recovery of very small bone fragments, particularly those less than a centimetre long, is probably a more serious issue, and may be responsible for the observed low abundances of small fragments. Although long bone shaft specimens are abundant in the Gatzarria sample (constituting over 50% of the Cj Mousterian NISP, not including an additional 318 indeterminate long bone shaft fragments), and were apparently not subject to selective discard, over the course of the laboratory analysis, it was observed that small bone specimens (i.e., <1 cm long) were relatively infrequent in the sample. Evidence from Laplace's excavation reports may explain this observation. Laplace (1966b:2) noted with regards to sieving of the sediments: "[Sieving] permits the collection of the micro- industry, minuscule ornaments and microfauna."4 This may suggest that the purpose of sieving during Laplace's excavations was not to recuperate fragmented macrofaunal remains, but only lithic objects, small ornaments, and microfauna. Consequently, some proportion of small macrofaunal remains captured in sieves may not have been recovered. To assess this possibility, fragment length distributions of faunal remains from Gatzarria, grouped by excavation year, are presented in Table 5.4, along with similar data from Saint-Césaire, a transition site in west-central France, where 5 mm and 2 mm sieves were consistently used during excavations (Morin 2004). These measurements are from sub-samples of décapages at both sites in which all identified and unidentified bone "[Le tamisage] permet de recueillir le micro-outillage, les minuscules éléments de parure et la microfaune." 108 Sample <1 cm 1-2 cm 2-3 cm 3—4 cm 4-5 cm 5+ cm total n_ Gtz 1963-1964 2.7 28.8 32.7 14.8 7.0 14.0 100.0 257 Gtz 1965-1970 11.4 32.6 30.3 10.6 9.1 6.1 100.0 132 Gtz 1971-1976 20.8 48.7 20.0 5.0 1.8 3.7 100.0 600 Saint-Césaire 19.4 47.7 22.0 7.4 1.8 2.1 100.4 1353 Table 5.4: Size distribution of bone specimens in a measured sub-sample of décapages, in percentages, according to year of excavation. specimens were measured. All measured décapages from Gatzarria are included, because the way in which décapages were selected for the Cj Mousterian sample was not foreseen when décapages were selected for measurement. As a result, insufficient measured décapages were preserved in the Cj Mousterian sample for this analysis. No décapages from 1961 or 1962 were present in the sample. The data in Tablé 5.4 suggest that small fragments (2 cm and less) are strongly under-represented in the Gatzarria samples from prior to 1971. The difference between the 1963-1964 and 1965-1970 distributions is not statistically significant (Kolmogorov- Smirnov Z= 1.045,/? = 0.224). This suggests that even if different sieve sizes were used in 1963 and 1964, they did not have a significant effect on the representation of small bone fragments. In contrast, the difference between the 1965-1970 and 1971-1976 samples is statistically significant (K-S Z = 2.552, ? < 0.001). When the data from Gatzarria is compared to Saint-Césaire, they suggest that, prior to 1971, much of the highly-fragmented bone material captured in sieves at Gatzarria may not have been kept. Conversely, the similarity between Saint-Césaire and the most recent Gatzarria sample may suggest that recovery of bone material at Gatzarria was more thorough after 1971. However, even for décapages excavated from 1971 onwards, counts of very small fragments may not be reliable. Villa et al. (2004) correctly point out that most fragments less than one centimetre in length are lost during sieving and sorting even when relatively small mesh sizes (e.g., 2 mm) are used. For this reason, counts of this size class may not be representative. Although it was observed that size distributions from Saint-Césaire and Gatzarria 1971-1976 are very similar, this does not guarantee that collection procedures at the two sites were the same. For instance, it is not known for certain what sieve sizes were used at Gatzarria during this time period (Table 5.3). The possibility that mesh sizes were variable, or, more likely, that some discard of small bone material occurred during this period, cannot be ruled out. Additional sources of error in counts of very small bone fragments include post- excavation breakage, crushing, and loss during transport, handling, and storage. Faunal material from Gatzarria was bagged by entire décapages, including large bone fragments along with material from both large and small sieves. Possibly as a result of storage conditions, recent bone fractures (indicated by the lighter colour of fracture surfaces) were commonly observed in the samples. Some small fragments may have been produced (or broken into more pieces) by recent breakage. For very small fragments, this created some ambiguity regarding which fragments should be counted and measured. Consequently, abundances of small bone fragments at Gatzarria may not only be variable due to differential recovery, but also subject to some error due to post-excavation breakage. Analysis of the unidentified portion of the sample will need to accommodate these possibilities. In particular, sieve sizes and/or discard of small bone fragments may have had a pronounced effect on the recovery of burnt bone, because burning increases bone fragmentation (Stiner et al. 1995; Costamagno et al. 1999; Villa et al. 2004). 110 Burning affects 3.3% of the entire Gatzarria Cj Mousterian sample, while a slightly higher percentage (3.6%) of non-identified specimens are burnt (Table 5.5). The frequency of burning increases dramatically if the focus is enlarged to the total sample. Differential recovery of small fragments may have reduced the representation of burnt specimens in both of these samples. Specifically, how might it have affected the composition of the Mousterian sample? Figure 5.6 compares the fragment size distributions of burnt and unburnt specimens from measured décapages in the Cj Mousterian sample. Indeed, much greater proportions of burnt fragments are observed in the two smallest size categories, and no burnt specimens are longer than 3 cm. This confirms that burnt specimens are more fragmented than unburnt specimens. The observed difference is nearly statistically significant. (K-S Z= 1.334,/? = 0.057). All specimens Non-identified specimens ? burnt ? total % ? burnt ? total % Total sample 1747 11162 15.7 1730 10413 16.6 CjMoust 163 4907 3.3 162 4464 3.6 Table 5.5: Frequency of burning in the study samples 60 50 lunburnt(n = 714) • burnt (n = 20) i? 20 <1 cm 1-2 cm 2-3 cm 3-4 cm 4-5 cm 5+c Length Figure 5.6: Size distributions of burnt (black) and unburnt (grey) specimens in a measured sub-sample of décapages from the Cj Mousterian sample. Ill However, is the size distribution of specimens in this small sample representative of the entire Cj Mousterian sample? This should be considered because it is possible that, due to inconsistent recovery methods, the measured subsample is not characteristic of the entire sample. Figure 5.7 compares the size distribution of burnt specimens in the sub- sample of measured décapages to the size distribution of all burnt specimens in the Cj Mousterian sample. The similarity in the distributions suggests that the proportions of burnt specimens in the measured subsample are representative of the total faunal sample, and this is confirmed by statistical analysis (K-S Z = 0.376,/? = 0.999). Consequently, it can be concluded that burnt specimens are more fragmented relative to unburnt specimens in complete Cj Mousterian sample. This means that, if small fragments are indeed missing from the samples, the proportion of burnt bone in the assemblage is under-estimated. Unfortunately, it is not possible to estimate how much burnt material may have been lost. Measured sample (? = 20) CjMoust. (? =161) ^ 20 <1 cm 1-2 cm 2-3 cm 3-4 cm 4-5 cm 5+ cm Length Figure 5.7: Comparison of size distributions of burnt specimens from the sub-sample of décapages in which all specimens were measured (grey) and all burnt specimens from the Cj Mousterian sample (black). 112 Density A positive correlation between element abundance and the structural density of bone portions has often been used by zooarchaeologists to assess post-depositional destruction, given that less dense bone is more fragile and therefore more easily destroyed (see discussion in Lyman 1994b:23 5—281). However, positive correlations with density may also be caused by human behaviours, such as bone grease extraction or the use of bone as fuel, and by carnivore ravaging (Lyman 1994b; Cleghorn and Marean 2004; Morin 2009). This is because the mineral structure of bone is homogeneous—bone 'density' is in fact bone 'porosity' (Lyman 1994b), and the pores in 'low-density' bones are filled with grease (Brink 1997). Therefore, the same bones that are fragile because of high porosity are attractive to carnivores and people searching for fat (Blumenschine 1988; Brink 1997). In addition, transport decisions can also produce correlations with density, as for instance, when low-density axial elements, such as the vertebral column, are discarded at the kill site (Binford 1978, 1981). Is there any indication that density- mediated attrition processes (not including transport bias) affected the composition of the Gatzarria Mousterian assemblage? Because it is unlikely that long bone shafts and epiphyses would often be transported separately, this question can be addressed by comparing the representation of high- and low-density portions within the long bones (Binford 1981). Table 5.6 and Figure 5.8 compare the frequencies of long bone portions in the Cj Mousterian sample to shape-adjusted bone density values from Lam et al. (1999). Density values for reindeer are used in this comparison, as this species most resembles the dominant species at Gatzarria (red deer). However, all ungulates are considered together Bone portion Scan site ? Density humerus, proximal HUl 0 0.26 humerus, shaft HU3 18 1.12 humerus, distal HU5 1 0.48 radius, proximal RAl 1 0.53 radius, shaft RA3 15 1.09 radius, distal RA5 1 0.49 metacarpal, proximal MCl 4.5 0.92 metacarpal, shaft MC3 40 1.10 metacarpal, distal MC6 2.5 0.68 femur, proximal FEl 4 039 femur, shaft FE4 19 1.15 femur, distal FE6 6 0.32 tibia, proximal TIl 3 0.35 tibia, shaft TI3 53 1.13 tibia, distal TI5 3 0.73 metatarsal, proximal MRl 4.5 0.90 metatarsal, shaft MR3 46 1.08 metatarsal, distal MR6 1.5 0.59 Table 5.6: Frequencies of long bone portions in the Gatzarria Cj Mousterian sample compared to shape-adjusted (BMD2) density values for reindeer (Lam et al. 1999: 351-353, Table 1, BMDi values used where shape-adjustments not applicable). Selected scan sites illustrated in Lam et al. (1999:348, Figure 1). All ungulate specimens (NISP and NSUTS) are pooled in this table. Values for the metatarsals and metacarpals were adjusted by adding half the number of undistinguished metapodials to each. shtib 50 shmt · 40 shmc · P, 30 ? 20 shfem shhum · shrad · 10 dsfem.pxtib . · femdsrad pxrad dsnjc #dst.b pxmt..Dxmc" P*mc pxhum dshum «·_·_ · d=mt 0.00 0.20 0.40 0.60 0.80 1.00 1.20 Density Figure 5.8: Frequency of long bone portions in the Gatzarria assemblage versus bone density. Data from Table 5.6. 114 in these analyses because inter-taxonomic variation in bone density among ungulates is limited (Lam et al. 1999). These data demonstrate that high-density bone portions (shafts) are much better represented than lower density bone ends (both proximal and distal epiphyses). The frequency and density data are positively, and significantly, correlated (rs = 0.76, ? < 0.001). However, these are NISP, not MNE data. Therefore, it might be expected that the representation of long bone shafts would be inflated, as marrow- cracking activities may produce more shaft fragments relative to epiphyses. Nevertheless, the imbalance between epiphyseal and shaft fragment representation seems too large to be accounted for by differential fragmentation. For example, for some elements, such as the humerus and radius, epiphyses are nearly absent despite the relative abundance of shaft fragments (Table 5.6). Consequently, the correlation between bone density and frequency in the Gatzarria assemblage would likely persist even if MNE data for long bone portions were available. Destruction of spongy bone portions has quite probably occurred at Gatzarria. However, a simple correlation with density does not identify the attritional agent responsible for bone destruction (Lyman 1994; Cleghorn and Marean 2004). Three possible explanations for this pattern—bone burning, post-depositional destruction, and carnivore ravaging—are now considered in turn. The impact ofbone burning on sample composition The probable under-representation of burnt specimens in the Gatzarria sample may pose difficulties for understanding the taphonomic history of the Gatzarria assemblage, as bone burning is one possible explanation for the poor representation of low density bone portions. As mentioned in Chapter 4, numerous European Upper Palaeolithic sites show patterns of selective burning of spongy bone such as long bone 115 epiphyses. This appears to be a signature of the use of bone as fuel, because the high grease content of spongy bone portions permits them to burn well (Castel 1999; Costamagno et al. 1999; Théry-Parisot 2002; Morin 2009). This pattern is, at present, less well documented for the Middle Palaeolithic, although at Saint-Césaire, evidence for bone burning during the Mousterian is present (Morin 2004). Burning of bone might therefore also be expected during the Mousterian at Gatzarria. Indeed, the single burnt specimen in the Cj Mousterian NISP sample is in fact a spongy bone element (a thoracic vertebra fragment). However, the overall frequency of burnt specimens in the Cj Mousterian at Gatzarria is relatively low (3.3%; Table 5.5), and seems inconsistent with burning as a major taphonomic agent. In comparison, at Saint-Césaire, all assemblages are between 21.6% and 37.8% burnt (Morin 2004), and at the Upper Palaeolithic sites of Combe Sauniere, Castanet, and Cuzoul de Vers, where bone burning is also in evidence, frequencies of burning are on the order of 30% (Villa et al. 2004). On the other hand, the under-representation of burnt remains, due to fragmentation and probable discard of small fragments, may partially explain the low frequency of burning at Gatzarria. Furthermore, the Cj Mousterian sample covers only a small area of the site, and may not capture spatial variation in the distribution of burnt specimens (cf. Cuzoul du Vers; Villa et al. 2004). For example, 59% of fragments from layer Cj in unit 22G at Gatzarria are burnt. Although this unit is only represented by four décapages, and was eliminated from the Cj Mousterian sample because of insufficient lithic data, it may indicate that the frequency of burning varies substantially throughout the site—notably, unit 22G is located outside, rather than inside, the cave. When the total Cj sample, which covers a 116 greater proportion of the site area, is considered, the frequency of burning drastically increases to 15.7%. The percentage ofburning in the total sample remains lower than the frequency of burning at the sites mentioned above, but, when the under-representation of small fragments at Gatzarria is considered, is probably consistent with the hypothesis of bone burning. However, the total sample includes some Upper Palaeolithic material, and therefore this conclusion may apply only to the Upper Palaeolithic component of Cj. Although the frequency of burning in the Cj Mousterian sample (3.3%) is probably an underestimate, the available evidence is inadequate, on its own, to make a strong case for burning of bone as fuel in the Cj Mousterian sample. Again, the low frequency of burnt specimens may have been affected by relatively small sample size, the small area sampled, and limited recovery of highly fragmented bone. To answer the question of bone burning more firmly, further analyses of the Gatzarria bone assemblages might consider quantification, and possibly measurement, of all spongy bone fragments in the non-identified portion of the sample. A qualitative impression of the sample examined in this study suggests that spongy bone is very rare in the non-identified portion of the sample, but when present, is often burnt. Given the limited evidence for burning, alternative explanations for the general absence of long bone epiphyses and other spongy bone portions should be considered. There are two primary additional, not mutually exclusive, factors that could explain the poor representation of spongy bone portions at Gatzarria: post-depositional destruction and carnivore ravaging. The role of each of these in altering the Gatzarria assemblage is now considered in turn. However, if neither of these alternatives provides sufficient evidence for the destruction of spongy bone in the sample, it will suggest that the use of 117 bone as fuel may have been one of the factors shaping the assemblage, despite the poor representation of burnt bone fragments. Post-depositional destruction As mentioned earlier, low density bone portions that are more susceptible to post- depostional breakage and destruction may also be destroyed by human or carnivore action. How can post-depositional destruction be assessed if correlations with density may equally be a behavioural signature? Marean (1991) has proposed a completeness index that assesses the impact ofpost- depositional breakage by considering the fragmentation of small compact elements. These include the carpáis, tarsals (except the calcaneus), the lateral malleolus, and sesamoids, which have very small or non-existent marrow cavities. Marean suggests that these elements are rarely fragmented by humans or carnivores attempting to extract nutritive (marrow) value; although, in rare cases, they may be fragmented in the course of bone grease production (Binford 1978). Villa et al. (2004) suggest that the third phalanx can also be included in the index, because of the very small size of its marrow cavity. This element is slightly less dense than the others included in the index, meaning that it may be a somewhat more sensitive indicator of post-depositional breakage. Villa et al. (2004) calculate Marean' s index using the proportion of compact elements coded as complete (CO) or almost complete (ACO), divided by the total number of specimens. Burnt and carnivore-marked specimens must be removed from the index, since these processes may cause fragmentation that is not post-depositional. Although the sample of compact elements in the Cj Mousterian sample is small, all six specimens are complete or almost complete, and none have traces of burning or 118 carnivore action (i.e., %CO+ACO = 100). This may suggest that post-depositional breakage at Gatzarria is minimal. Given that the same bones included in this index are consistently burned at sites where bone was used as fuel, such as Saint Césaire (Morin 2004, 2009) and Cuzoul de Vers (Villa et al. 2004), the absence of burnt carpáis and tarsals at Gatzarria is notable, and suggests that bone burning was not significant at the site. However, this same line of evidence demonstrates limited post-depositional breakage. Villa et al. (2004:715) argued that high values for the completeness index, along with high proportions (=70%) of green bone fracture morphologies, at the Palaeolithic sites they studied "suggest that post-depositional processes have not significantly altered the main characteristics of these assemblages." In the Gatzarria Mousterian sample, 71% of coded bone fractures were classified as green, and 29% as dry (n = 293), suggesting a proportion of post-depositional breakage highly comparable to the sites studied by Villa and colleagues. Yet, mandibles and maxillaries at Gatzarria are extensively fragmented, and it seems likely that some of this breakage was post-depositional. Only 3 of 72 tooth specimens are attached to a bone portion, and only one or two teeth are present in each of these specimens. Overall, it is probable that post-depositional breakage and destruction at Gatzarria was moderate, although the effect of these processes on high-density and/or compact bone portions may be minimal. The precise extent to which this has affected the assemblage is difficult to specify. Regardless, the degree of post-depositional breakage does not seem sufficient to explain, on its own, the very low representation of spongy bone. 119 Biological agents ofaccumulation and destruction Both humans and animals, particularly carnivores, but also rodents and birds, may accumulate bones in caves. Consequently, faunal assemblages from caves may not be exclusively produced by one type of agent. To what degree has the Gatzarria assemblage been accumulated and transformed by non-human animals? Evidence of human and carnivore action in the Cj Mousterian assemblage is summarized in Table 5.7. Notches are not included in this table because percussion and carnivore tooth notches overlap in morphology and may be confused (Capaldo and Blumenschine 1994). Anthropic marks affect 1 1.3% of the Cj Mousterian assemblage. However, when teeth and antler are excluded from the tally, evidence ofhuman action including cutmarks, scrape marks, burning, and retouchers is found on 1.4.1% ofNISP (42/297), or 14.3% of ungulate remains (42/293). The percentage of human-marked ungulate remains is highly comparable to other, primarily anthropic, French Palaeolithic sites, such as Saint-Césaire (Morin 2004), Combe Sauniere, Castanet, and Cuzoul de Vers (Villa et al. 2004). Mark Type NISP % ungulate NISP % Anthropic 42 11.3 42 11.5 Cutmarks 33 8.8 33 9.0 Scrape marks 5 1.3 5 1 .4 Retouchers 6 1.6 6 1.6 Burning 1 0.3 1 0.3 Carnivore 21 5.6 18 4.9 Gnawing 5 1.3 3 0.8 Digestion 16 4.3 15 4.1 Total NISP 373 366 Table 5.7: Frequencies of human and carnivore marks in the Cj Mousterian sample. Specimens with non-observable (damaged) surfaces and percussion or tooth notches excluded. Frequencies of anthropic mark types do not add up to the total because some specimens showed more than one type of mark. 120 Carnivore remains constitute roughly 2% of the Cj Mousterian sample (« = 7; see Table 5.1). No large carnivore remains bear anthropic traces. None of the carnivore specimens are juvenile remains, which can be an important indicator of carnivore denning. Although coprolites are occasionally present in the bone samples, they cannot be quantified due to biased recovery. No rodent marks were observed in the assemblage. Carnivore damage, including gnawing and digestion, affects 5.6% of the Cj Mousterian NISP, or, if only ungulate remains are considered, 4.9% of NISP. This level is higher than carnivore mark frequencies at Middle Palaeolithic sites such as level 22 at Jonzac, Charente-Maritime, France (0.4% of ungulate NISP; Villa et al. 2004), and the Denticulate Mousterian at Saint-Césaire (0.1% of total NISP; Morin 2004). The frequency of carnivore damage at Gatzarria is more comparable to the Solutrean levels (Unit IV) at Combe Sauniere (Dordogne, France), where gnawing affected 3.4% of ungulates (Villa et al. 2004). These authors suggested that, because carnivore marks were rare on shaft fragments at Combe Sauniere, carnivores had secondary access to human- deposited remains and probably did not contribute a significant amount of bone to the assemblage. Less than 1% of ungulate long bone shafts are carnivore-marked at Combe Sauniere. At Gatzarria, 2.6% (5/189 ungulate diaphyses) show carnivore marks. Gnawing damage (tooth pits, scooping, etc.) affects 1.3% ofNISP, but only 0.8% when only ungulates are considered (in comparison, two of seven, or 29% of carnivore remains are gnawed). Although carnivore remains are infrequent, this suggests that they are more likely to be gnawed than ungulate specimens. Digestion is more common than gnawing, affecting 4.3% of total NISP and 4.1% ofungulates in the Mousterian sample. Although sample size is also small, there is an interesting pattern of anatomical representation in the digested remains: 39% (7/18) of digested ungulate specimens are phalanges, another three (17%) are teeth, and one additional specimen is an antler fragment. Because teeth, phalanges, and antler are only of marginal nutritional utility to human foragers, this pattern suggests secondary access by carnivores to a previously accumulated deposit. However, phalanges and teeth may be easier to identify after digestion than other skeletal parts. Traces of digestion were also found on 2.5% of unidentified specimens. The frequency of carnivore damage at Gatzarria is somewhat high for a presumably anthropogenic assemblage, but it remains much closer in line with anthropogenic deposits than with hyaena dens, for example, Bois Roche (Charente, France), where 81.7% of ungulate remains show carnivore damage (Villa et al. 2004). Patterns of fragmentation may also provide insights into the depositional history of faunal assemblages. In general, hyaena—or carnivore—created assemblages are less fragmented than human assemblages showing evidence of systematic marrow-cracking (Villa et al. 2004). In Figure 5.9, the length distribution of identified specimens in the Gatzarria sample is compared to length distributions from two of the sites mentioned above: Bois Roche, a hyaena den, and level 22 at Jonzac, an anthropically-accumulated Middle Palaeolithic deposit (Villa et al. 2004). At the lower end of the distribution, Gatzarria lies intermediate between the hyaena den and Jonzac; however, in the upper range of fragment lengths, it much more closely resembles the anthropic site. However, differences between these size distributions may be taxon-dependent, if certain taxa are more heavily fragmented or are more difficult to identify from small fragments. In fact, the primary taxon at Bois Roche is Bos/Bison, that at Gatzarria is the smaller red deer, and that at Jonzac, the even smaller reindeer; a pattern which may 122 40 -"-Gatzarria (? = 300) 30 -0-BoisRoche(n = 715) a. "^" Jonzac (? = 1299) ?? 20 10 x>-o-o. -o--?-· 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 Length (cm) Figure 5.9: Length distribution of identified bone specimens at Gatzarria compared to Bois Roche (a hyaena den), and Jonzac Level 22 (a Middle Palaeolithic deposit). Antler, teeth, and ivory are excluded. X-axis categories: 1 = 1-3 cm, 3 = 3-5 cm, etc. Data from Villa et al. (2004:717, Figure 4). explain the differences between these sites in the proportions of smaller specimens. To examine this possibility, mean fragment lengths for long bones of different-sized ungulates at these sites are shown in Table 5.8, along with comparable data from the Denticulate Mousterian level at Saint-Césaire. These data suggest that, within a single taxon, long bones at anthropic sites are much more fragmented than those from the hyaena den. Although the sample size ofBos/Bison long bones from Gatzarria is very small, the mean fragment length at this site clusters closely with Jonzac and Saint- Césaire. Smaller taxa—red deer and reindeer—are also highly fragmented at Gatzarria and Jonzac. Therefore, the important feature of Figure 5.9 may be not the height of the peak in the distributions, which might be affected by taxonomic composition and recovery methods, but their overall shape—specifically, the long tail of the hyaena den distribution. Assemblage Taxon ? mean length UNG Size 4 Gatzarria (Cj Moust) Bos/Bison 5 82.5 Jonzac (Level 22) Bos/Bison 81 83.0 Saint-Césaire (EGPF) Bos/Bison 87 80.5 Bois Roche (Layer 2) Bos/Bison 173 148.7 UNG Size 2-3 Gatzarria (Cj Moust) Cervus 181 68.3 Jonzac (Level 22) Rangifer 856 49.5 Table 5.8: Mean lengths of long bone fragments for different taxa at Gatzarria, two primarily anthropic Middle Palaeolithic deposits (Jonzac and Saint-Césaire), and a hyaena den (Bois Roche). Data from Villa et al. (2004:717, Table 10) and Morin (2004:199, Table 36). Assemblage fragmentation, as measured by the proportion of long bone shaft fragments to shaft fragments retaining a portion of the epiphysis (Table 5.9), also shows a strong similarity between Gatzarria and the anthropic sites studied by Villa et al. (2004). This pattern of fragmentation strongly contrasts with both the hyaena den (Bois Roche) and Sarrians, a mass grave where all elements were broken post-depositionally. Site Site type ? % limb ends % shafts only Gatzarria Middle Palaeolithic 189 3.2 96.8 Jonzac Middle Palaeolithic 1146 6.9 93.1 Castanet Upper Palaeolithic 2529 3.4 96.6 Combe Sauniere Upper Palaeolithic 228 5.3 94.7 Bois Roche Hyaena den 361 29.9 70.1 Sarrians Mass grave 300 54.7 45.3 Table 5.9: Comparison of frequencies of long bone ends (epiphysis with shaft) versus shaft-only fragments. Epiphysis-only fragments excluded. Data from Villa et al. (2004:722, Figure 6). Finally, the lengths and shaft circumferences of long bone specimens in the assemblage are another diagnostic measure of carnivore versus human accumulated assemblages (Villa et al. 2004). Figure 5.10 shows that long bone shafts in the Cj 124 70 ? = 212 60 50 40 s; 30 20 10 ¿SB m¡m ShCl JEEP / i*!!? mS¡? ShC2 1 «?? Sh C 3 Length Figure 5.10: Long bone shaft fragment lengths and circumferences in the Cj Mousterian sample. Sh C = shaft circumference, 1 = Mousterian sample are highly fragmented, with most specimens belonging to the smallest size category (shaft circumference France, the suggestion that burning may have been responsible for substantial destruction of spongy bone portions in the assemblage remains plausible as a partial explanation for this pattern of skeletal representation. Regardless of what factor was the primary cause of the poor representation of spongy bone at Gatzarria, interpretation of the skeletal part profiles will be unreliable if based on low-density elements. Consequently, the zooarchaeological analysis focuses on compact bone portions of similar density, which appear to have been relatively unaffected by both burning and carnivore ravaging (for example, long bone shafts). This approach circumvents most problems ofpost- depositional breakage, since compact bone is generally more resistant to post- depositional fragmentation (Marean 1991). Bone surface preservation The presence ofbone surface damage may obscure or destroy important taphonomic signatures such as cutmarks and gnaw marks. Is there any evidence that, due to post- burial bone surface damage, the frequencies of human and carnivore marks presented above are not representative of their frequencies prior to burial? In the Mousterian sample, overall preservation quality is fair to moderate. The most common alterations of bone surfaces, other than human and carnivore marks, include cracking (1 1.0% ofNISP and NSUTS), manganese staining (8.7%), pitting (4.7%), and exfoliation (4.2%). Qualitative assessments of overall bone surface state and estimates of observable (i.e., undamaged) surface area are shown in Tables 5.10 and 5.1 1. Digested specimens are excluded from these counts because digestion damage, which unevenly corrodes the bone surface, makes it difficult to assess surface condition. Surface state ? % Intact 0 0.0 Relatively good 169 47.2 Somewhat damaged 131 36.6 Poor 58 16.2 Total 358 100.0 Table 5.10: Overall surface state of NISP and NSUTS in the Cj Mousterian sample. Observable surface n % 0% 12 3.6 10% 14 4.1 20% 14 4.1 30% 4 1.2 40% 15 4.4 50% 1 1 3.3 60% 24 7.1 70% 29 8.6 80% 40 11.8 90% 54 16.0 100% 121 35.8 Total 338 100.0 Table 5.11: Percentage observable surface of NISP and NSUTS in the Cj Mousterian sample. For overall surface state, approximately half of the bone specimens fall into the relatively good category, indicating that for most specimens, damage is minimal, although none are in perfect condition (i.e., "intact"). Similarly, for the majority of bone specimens in the assemblage, 80% or more of bone surface is observable. However, all categories of observable surface are consistently represented, and a substantial proportion of specimens are in somewhat damaged or poor condition. Table 5.12 compares surface preservation quality in rows 6 and higher with rows 5 and lower. This division point was selected according to the accelerating slope in the back of the cave, which begins roughly in row 6 (see Figure 5.2). This table shows that while surface condition in the front units is generally good, with few specimens in poor condition, a greater percentage of specimens in the back of the cave are in somewhat damaged or poor condition. This suggests that the different taphonomic factors—such as the steeper slope and badger burrowing—to which specimens in the rear portion of the cave have been subjected have affected their preservation state. This should not significantly impact this analysis, as only one unit from rows 6 or higher (unit 6D) is included in the Cj Mousterian sample. However, this pattern might be more significant for future analyses, which may include a greater portion of the cave's surface area. Area ? % Relatively good %Somewhat damaged %Poor Front (Rows 5 and lower) 308 51.9 35.1 13.0 Back (Unit 6D) 51 17.6 47.1 35.3 Table 5.12: Comparison of bone surface states in the front and back of the cave. Cj Mousterian sample only. Digested specimens excluded. The effect of surface damage on the frequency of traces is considered using cutmarks in Table 5.13. The increase in the number of cutmarks from the relatively good to somewhat damaged categories indicates that cutmark frequencies have probably not been significantly affected by surface damage for these categories. This does not appear to be the case for specimens in poor condition, however, as they show a much lower incidence of traces. Thus, the total frequencies of anthropic marks may be slightly underestimated in the sample, particularly for bones from unit 6D, given that this unit has a greater proportion of specimens in poor condition. Overall surface state « ? (Cut) %Cut Intact 0 0 0.0 Relatively good 169 14 8.3 Somewhat damaged 131 15 11.5 Poor 58 __2 3.4 Total 358 31 8.7 Table 5.13: Comparison of frequencies of cutmarks by overall surface state. Digested specimens excluded. Intra-taxonomic identification biases Some elements may have artificially low NISP counts because, when fragmented, they are more difficult to anatomically identify than others. As a preliminary step to assessing intra-taxonomic identification bias, Morin (2004) compared the representation of skeletal elements in NISP and NSUTS. This procedure determines whether some skeletal portions are under-represented in NISP, but not in NSUTS, simply because they are difficult to identify taxonomically (rather than anatomically). Table 5.14 shows NISP and NSUTS counts by anatomical elements or groups of elements. This comparison indicates that although the cranium (including the maxillary), mandible, vertebrae, and 129 Element NISP %NISP NSUTS %NSUTS "/»difference antler 1 0.3 0 0.0 0.3 cranium 1 0.3 1 1.3 -1.0 mandible/maxillary bone 10 2.7 12 15.2 -12.5 tooth 69 18.6 21 26.6 -7.9 hyoid 3 0.8 0 0.0 0.8 cervical 9 2.4 1 1.3 1.2 thoracic 6 1.6 1 1.3 0.4 lumbar 5 1.4 3 3.8 -2.4 sacrum 1 0.3 1 1.3 -1.0 vertebrae 0 0.0 4 5.1 -5.1 ribs 14 3.8 18 22.8 -19.0 scapula 1 0.3 0 0.0 0.3 humerus 18 4.9 1 1.3 3.6 radius/radio-ulna 18 4.9 0 0.0 4.9 metacarpal 37 10.0 1 1.3 8.7 pelvis 8 2.2 0 0.0 2.2 femur 25 6.8 4 5.1 1.7 tibia 58 15.7 1 1.3 14.4 malleolus/fibula 0 0.0 1 1.3 -1.3 talus 2 0.5 0 0.0 0.5 calcaneum 2 0.5 1 1.3 -0.7 greater cuneiform .1 0.3 0 0.0 0.3 metatarsal 43 11.6 0 0.0 11.6 metapodial 13 3.5 5 6.3 -2.8 phalanges 23 6.2 2 2.5 3.7 vestigialphalanges 0 0.0 1 1.3 -1.3 sesamoids 2 05 0 O1(J (?5_ Total 370 100.0 79 100.0 Table 5.14: Representation of skeletal elements in NISP and NSUTS. AU taxa combined. Elements better represented in NSUTS than in NISP are shown in italics. Ivory and fetal remains excluded. ribs are under-represented in NISP, this is compensated to some degree by an over- representation in NSUTS, especially for ribs. Teeth are under-represented in NISP in comparison to NSUTS, although this is probably due to high levels of fragmentation of teeth rather than lesser identifiability. The sample sizes of vestigial phalanges, calcanea, and lateral malleoli are too small for the differences between NISP and NSUTS to be considered important. Metapodials are over-represented in NSUTS while metacarpals and metatarsals are over-represented in NISP. The difference in representation between metapodials and metacarpals/tarsals is probably related to the fact that the anatomical landmarks that specifically identify metapodials as metacarpals or metatarsals are the same as those that most clearly identify them taxonomically—for example, the anterior groove, as opposed to the posterior surface. Overall, however, metapodial fragments are generally highly identifiable as metapodials regardless of whether they can be assigned to species. Metapodials, including metacarpals and metatarsals, are probably slightly over- represented in the total assemblage when both NISP and NSUTS are considered. Table 5.14 also shows that a number of elements that are poorly represented in NISP are not better represented in NSUTS. These include spongy bone elements such as the pelvis and vertebral sections. These observations are highlighted here to further substantiate prior claims that spongy bone is poorly represented at Gatzarria. To reiterate previous conclusions regarding this pattern, it probably results from some combination of burning, carnivore attrition, and post-depositional destruction. Carpáis and tarsals are also under-represented to a degree comparable to that of low-density axial elements—in fact, carpáis are totally absent from the Mousterian sample. Long bones are generally under-represented in NSUTS in comparison to NISP, suggesting that when identifiable to element, long bones are usually also identifiable to species. A partial exception to this observation is the femur, which, in relation to other long bones, has a relatively high NSUTS, meaning that it may be more difficult to identify taxonomically. It is nevertheless clear that some long bones are better represented than others in NISP. To examine the role of intra-skeletal differential 131 identification in long bone representation, the fragmentation of different long bones can be considered. If some elements tend to be more fragmented than others, then they may be identified less frequently. Table 5.15 compares the frequencies of red deer long bone shaft specimens in the Cj Mousterian sample with two measures of fragmentation: average length in centimetres and average length relative to the complete shaft length. The average lengths of the different elements are similar, with a maximum difference of 1.5 cm between the humerus and the metacarpal. There is no apparent relationship between average fragment size and element frequency (rs = 0.09, ? = 0.92), suggesting no relationship between identifiability and absolute fragment size. The relationship between element frequency and the average length of specimens relative to the complete bone is negative, but not significant (rs = -0.54, ? = 0.29). The inverse relationship between element frequency and the latter measure of fragmentation contradicts the hypothesis that increased fragmentation of certain elements may be depressing their abundance in the Gatzarria assemblage, because in fact the most abundant elements tend to be more fragmented relative to the complete element. Rather, what these data likely mean is that, as fragmentation increases, the identifiability of certain elements—specifically the metapodials and tibia—decreases less quickly relative to other elements (the humerus, radius, and femur). This difference can be attributed to bone-specific morphology and landmark distribution. For instance, both the tibia and the metapodials of red deer have highly distinctive landmarks, notably the angled borders of the tibia and the anterior groove of the metapodials, which facilitate their identification even when a smaller portion of the total bone is represented. avg. length avg. length Element ? (cm) (proportion) tibia 42 7.7 1.17 metatarsal 36 6.8 1.33 metacarpal 32 6.3 1.34 humeras 15 7.8 1.53 femur 14 6.7 1.19 radius 12 1_A 1.42 Total 151 7.1 1.33 Table 5.15: Fragmentation of red deer long bone shaft specimens in the Cj Mousterian sample. Proportional average lengths calculated using values for shaft length coded using Villa and Mahieu's (1991) system (described in Chapter 4 and in Figure 5.10). The suggestion that certain elements are more readily identified when fragmented can be further tested using the refit data. When fragment size is held constant, if more portions of an element are diagnostic, then the likelihood that more than one specimen from a single bone will be identified increases, and so does the probability that any two identified pieces ofthat element are from adjacent areas of a single bone. Accordingly, there is a greater probability that fragments of those elements will refit (Morin 2004). Refits of long bone shafts are summarized in Table 5.16. Because the sample size of refits is small, data for both the Cj Mousterian sample and the total sample are presented. Element Total refits Total ? % refitted Refits (Cj Moust) ? (Cj Moust) % refitted tibia 9 88 10.2 7 55 12.7 metatarsal 6 55 10.9 4 41 9.8 humerus 2 24 8.3 1 14 7.1 radius 2 25 8.0 1 18 5.6 femur 1 29 3.4 1 20 5.0 metacarpal 2 45 AA 1 36 2£_ Table 5.16: Frequencies of refits on long bone shaft fragments. Metatarsal and tibia shafts, the most frequent elements in the assemblage, are by far the elements with the greatest proportion of refitted fragments. Although the proportion of refits of the metacarpal is lower, this may be an effect of small sample size, as this element is similarly identifiable to the metatarsal. However, the low frequency of refits of the femur is consistent with Morin's (2004) conclusion that this element may be less identifiable when highly fragmented, and thus under-represented in NISP counts. Overall, there does appear to be a relationship between element identifiability and NISP counts, meaning that some skeletal elements are indeed over- or under-represented in NISP. Furthermore, the refit frequencies also suggest a relationship between element identifiability and specimen interdependence. MNE data would help assess the latter problem. Inter-taxonomic identification biases The three most abundant species at Gatzarria (red deer, bison, and chamois) are artiodactyls. Given the similarities in the skeletons of these species, differences in morphology are not likely to be responsible for identification biases, as the key landmarks for identification are generally the same. The main possible cause of differential identification between ungulate species at Gatzarria is therefore also what differentiates them the most skeletally: size. Possible exceptions to this generalization are the metapodials of cervids, which are unusually diagnostic. The possible over- representation of red deer metapodials has already been discussed as a factor in intra- taxonomic differences in skeletal identification. Morin (2004) has suggested that there may be a relationship between taxonomic identifiability and the proportion of a complete bone represented by a fragment. In other words, if absolute fragment sizes are held constant, then the size of fragments from larger animals is decreased relative to body size, and so they will be less identifiable. To examine this possibility, Figure 5.11 compares long bone shaft circumferences and lengths for three ungulate size classes: bison, red deer, and small ungulates (reindeer, ibex, chamois, and roe deer). The complete sample is considered in this figure because the number ofbison long bone specimens in the Cj Mousterian sample was too small for comparison. a) Bison #1 = 29 b) Red deer « = 206 70% - 70% j" 60% ¦! 60% 1 '" 50% - 50% J ' «¡% ¦ 4C-^iI - 3C% 3C% ': 20% 20% Y 10% » -' Sh C ! 10% T / ShC 1 0% / Sh C 2 0% -i- Sh C 2 Sh C 3 / Sh C 3 Length Length c) Small ungulates ,1 = 27 70% ? 60% f 50% G 40% ) 30% i 20% 10% ? / ShCl 0% -> / Sh C 2 Sh C 3 Length Figure 5.11: Comparison of long bone shaft size distributions for bison, red deer, and small ungulates. Sh C = shaft circumference, 1 = ______Red deer (n =206) Bison (w =29) % diff avg. length (mm) 72 99 27.9 min. length (mm) 20 51 60.8 avg. area (mm2) 1860 4236 56.1 min, area (mm2) 2OJ 1142 82.2 Table 5.17: Average and minimum lengths and areas for bison and red deer long bone shaft fragments (ulna excluded). Percent different calculated using: ([bison-red deer]/bison)xl00%. These data confirm that bison long bone shaft specimens are much bigger than red deer specimens, particularly when fragment surface area, estimated using measured maximal length by breadth, is considered. The differences between the species are even more pronounced for minimum fragment sizes, as would be expected ifproportionally smaller bone fragments are harder to identify—that is, if the minimum size of an identifiable fragment is generally bigger for larger animals. However, do the differences observed here simply reflect that bison long bone shaft specimens in the sample are larger than those of red deer, or do they reflect a real identification bias, meaning that smaller bison fragments have been relegated to the unidentifiable category? A comparison of the proportions ofNISP, NSUTS, and unidentified specimens by size class can be used to assess this problem. Table 5.18 compares the NISP and NSUTS of long bone specimens from large-to-very-large sized (bison, horse, and giant deer) and medium sized ungulates (red deer, reindeer, and ibex) with the number of unidentified long bone specimens assigned to 'large' and 'medium' animal size categories. Once again, the total sample is used here because of the small sample size of large ungulates in the Cj Mousterian sample. Table 5.18 suggests that the remains of large animals are less frequently identified, and consequently, that they may be under-represented in NISP relative to smaller animals. This suggestion is supported by statistical analysis: the difference in the frequencies of identified (NISP+NSUTS) and unidentified (NID) specimens for the medium and large size classes is highly significant (?2 = 27.93,/? < 0.0001). Unfortunately, this last comparison provides only a very coarse assessment of size representation, since there is overlap between the size categories, and therefore the unidentified 'large' category may contain some red deer. However, when taken as a Size class NISP NSUTS NID long bones Total % NISP %NISP+NSUTS Medium 217 3 357 577 37.6 38.1 Large 35 0 163 198 17.7 17.7 Table 5.18: Comparison of long bone identification rates for large- (bison, horse, and irish elk) and medium-sized ungulates (red deer, reindeer, ibex) with unidentified (NID) long bones assigned to large (size 3+) and medium (size 2-3) size classes. whole, the three preceding analyses suggest, albeit with a note of caution, that there is a relationship between relative fragment size and taxonomic identifiability. Thus, large species, such as bison, may be under-represented in the NISP sample. Summary This chapter has shown that the Cj Mousterian sample is not a perfectly undisturbed layer consisting of purely anthropically-deposited remains associated with a single stone tool industry. However, sites which display such ideal conditions for archaeological study are probably exceedingly rare, or at least substantially more rare than much of the archaeological literature may suggest. Despite this drawback, a number of informative conclusions about the history of the assemblage have been drawn in this chapter. The approach adopted here has allowed many stratigraphie and taphonomic problems to be identified, understood, and in some cases, circumvented. Although not all archaeological layers in the middle ensemble were systematically recognized during excavation, their limits can be approximately traced using the distribution of diagnostic lithic artifacts. Some overlap between the industries exists, and some parts of the cave appear to be disturbed. To deal with these problems, a subsample of the initially studied fauna, associated with décapages containing predominantly Mousterian artifacts, has been selected for the analysis of subsistence during the Cj Mousterian occupation. The taphonomic analysis can be summarized in the following points. Recovery of the small bone portion of the assemblage was not consistent during the excavations. As a result, the frequencies of small fragments in the assemblage, especially burnt specimens, may be under-estimated. Spongy bone is poorly represented in the sample, probably due to a combination of the use of bone as fuel, carnivore ravaging subsequent to human occupation, and some post-depositional destruction. Humans were responsible for the deposition of most remains in the Cj Mousterian sample, although carnivores contributed and/or modified a small proportion of the remains. Finally, there are differences in the representation of species and skeletal elements that are due to differential identifiability. For example, elements such as the femur and humerus may be under-represented in NISP. To address the problems of differential destruction of low-density bone portions, the analysis of faunal patterns at the site will primarily focus on compact bone, because of its high density, which confers to compact bone both resistance to post-depositional destruction and poor value as food and fuel (Morin 2004). It is now possible to proceed to an analysis of Mousterian subsistence behaviours at Gatzarria. 139 Chapter 6: Ungulate resource exploitation at Gatzarria Cave This chapter examines how the Mousterian occupants of Gatzarria utilized animal resources, and in doing so, starts to address the research questions outlined in Chapter 2. The chapter begins with a description of the assemblage in terms of taxonomic representation, skeletal part patterns, mortality patterns, and seasonality. Secondly, the decisions made by foragers at Gatzarria in terms ofprey choice, patch use, carcass transport, and on-site marrow processing are analyzed using the perspective of the foraging models discussed in Chapter 4. Assemblage composition Before proceeding, the relationship between sample size and taxonomic diversity should be considered. Small samples can be problematic for two reasons: first, rare species may not be represented, and secondly, observed proportions of taxa may reflect sampling bias (Grayson 1984; Lyman 2008). The Cj Mousterian sample is relatively small, but although it is not very even (i.e., it is heavily dominated by one species), the sample is relatively rich in terms of species—or more precisely, it is taxonomically dense (Lyman 2008). For example, in comparison to the Cj Mousterian sample (n = 376), the larger total Cj sample (« = 630) contains only one additional species (wolf), which is represented only by one specimen. This suggests that few (if any, as potentially the wolf specimen could have been part of an Upper Palaeolithic deposit) species were omitted from the Cj Mousterian sample because of sampling bias. A related concern is whether the proportions of different species, particularly of rare species, in the Cj Mousterian sample are representative of their true frequency in the deposited assemblage. One means of examining whether abundance data provide good ordinal-scale data is to compare the rank order of species using NISP and MNI (Grayson 1984). The role of sample size in affecting relative abundances may also be determined by testing the relationship between sample size and relative abundances of taxa across multiple archaeological assemblages (Cannon 2001). Unfortunately, these approaches cannot be applied here. Data from additional layers will eventually help clarify the relationship between sample size and assemblage composition at Gatzarria. Many of the species in the Cj Mousterian assemblage might best be considered as present but rare, with their true proportional contribution to the assemblage considered small but uncertain (Grayson 1984). This treatment of the data should probably apply, minimally, to all species representing less than 1% of the sample (see Table 5.2), which would include reindeer, woolly rhino, giant deer, and all carnivores. The Cj Mousterian assemblage is dominated by red deer, which constitute 71.3% of identified specimens, followed distantly by large bovines which form 7.4% of the sample, and chamois, at 6.9% (Table 5.2). However, the taphonomic analysis indicates that large bovines may be analytically under-represented in the sample, and that carnivores may have had a somewhat greater role in the accumulation of small ungulates such as chamois. Thus the relative importance of large bovines to human subsistence at the site may be greater, and that of chamois lesser, than suggested by the NISP figures. Roe deer is the next best-represented ungulate in the assemblage, at 3.2% ofNISP. Other small and moderate-sized ungulates, including horse, reindeer, and ibex are infrequent in the assemblage. Besides large bovines, very large ungulates (giant deer, mammoth, and rhinoceros) are represented only by isolated tooth specimens. It should be considered that, given the bulkiness of their skeletons, the remains of these very large species may seldom have been transported to the site. Carnivore remains are rare (n =7) in the Mousterian sample. Their remains do not show anthropic marks and are frequently gnawed (2/7 or 29%). As argued in Chapter 5, it is unlikely that the human occupants were responsible for the accumulation of the large carnivore remains. Regarding fox, few remains are present in the Mousterian subsample, and these may represent natural or anthropic accumulations. Thus, carnivores could have played, at most, a minor role in Middle Palaeolithic human diets at Gatzarria. No rabbit or fish remains, and very few bird remains, were encountered during the analysis. Possibly these species are under-represented in the sample due to the recovery bias discussed in Chapter 5. However, fox remains, similar in size to hares, are present in the Cj Mousterian sample, and leporids and birds were identified by Lavaud (1980) in other levels at Gatzarria (see Table 3.2 and 3.3). This suggests that the complete absence of these species in the Mousterian sample is not simply an artifact of recovery bias. Small fish bones may be more likely to have escaped collection. However, although there is some limited evidence for fishing during the Mousterian in France (Le Gall 2000), Costamagno and Laroulandie (2004) have suggested that small vertebrates did not form an important part of Palaeolithic diets in the Pyrenees region until the late Magdalenian. Morin (2010) has shown that lagomorph, fish, and bird species available during the Middle to Upper Palaeolithic transition at Saint Césaire, west-central France, ranked well below ungulate alternatives and were not exploited by Middle or early Upper Palaeolithic foragers. This may be the case at Gatzarria as well. These observations do not preclude the possibility that birds, fish, or carnivores were important dietary components during •the early Upper Palaeolithic occupations of Gatzarria, but for the above reasons, the following discussion of Mousterian subsistence at the site concentrates on ungulates. Ungulates clearly constituted the vast majority of animal resources consumed at the site and were the main thrust of transport effort. Skeletalpart representation Skeletal element representation for the two most common taxa (red deer and large bovids) at Gatzarria are briefly described here. All other species have insufficient sample sizes for discussing skeletal part representation, and many are primarily represented by teeth. Normed NISP (NNISP) are used in this and subsequent analyses in this chapter to correct for the fact that some elements are expected to be more frequent because they occur numerous times in the skeleton (Grayson and Frey 2004). Cranial NISP values are divided by two to obtain NNISP because all cranial fragments are teeth or sided elements. %NNISP values, which are calculated by standardizing the NNISP counts for all elements to the highest NNISP count, are also provided in the tables and used in graphs, because this standardization facilitates comparisons between different assemblages (Chapter 7). However, raw NNISP counts are used in all statistical calculations. For red deer, the tibia and the metapodials are the most common elements (Table 6.1). However, even relatively under-represented long bones (such as the femur and humerus) are better represented than any element of the axial skeleton, with the exception of the mandible. ,Bone portion NISP NNISP %NNISP maxillary teeth 5 2.5 11.9 mandibular teeth 19 9.5 45.2 other teeth 10 N/A N/A mandible (no teeth) 6 3.0 14.3 hyoid 2 1.0 4.8 cervical vertebrae 8 1.1 5.4 thoracic vertebrae 3 0.2 1.1 lumbar vertebrae 5 0.8 4.0 rib 7 0.3 1.3 scapula (glenoid) 1 0.5 2.4 humerus, shaft 15 7.5 35.7 humerus, distal 1 0.5 2.4 radius, proximal 1 0.4 1 .9 radius, shaft 12 6.0 28.6 radius, distal 1 0.5 2.4 ulna, shaft 2 1.0 4.8 metacarpal, shaft 34 17.0 81.0 sacrum 1 1.0 4.8 innominate 6 3.0 14.3 femur, proximal 2 1.0 4.8 femur, shaft 15 7.5 35.7 femur, distal . 5 2.5 11.9 tibia, proximal 3 1.5 7.1 tibia, shaft 42 21.0 100.0 calcaneum 1 0.5 2.4 greater cuneiform 1 0.5 2.4 metatarsal, proximal 1 0.5 2.4 metatarsal, shaft 37 18.5 88.1 metapodial, shaft 12 3.0 14.3 phalanx 1 7 0.9 4.2 phalanx 2 2 0.3 1 .2 large sesamoid 1 OA 0.6 Total 268 Table 6.1: Red deer skeletal part frequencies in the Cj Mousterian sample. Bone portions not shown have an NISP of 0. As discussed in Chapter 5, the low representation of axial elements, carpals, and tarsals is probably a result of some combination of burning, post-depositional destruction, carnivore attrition, and possibly, transport bias. Phalanges are also under-represented, and this issue is addressed in a later section. Another interesting pattern in the skeletal representation data is the over-representation of the mandible relative to the cranium (%NNISP of 45.2% versus 1 1.9% respectively, tooth fragments only), which may reflect transport of the tongue. These patterns are further explored when considering carcass transport strategies. Bison and/or aurochs are primarily represented by teeth (Table 6.2). The apparent over-representation of teeth may be a consequence of the size-related difficulty in identifying large bovid bone specimens, an issue that is probably less applicable to teeth. Otherwise, the pattern of large bovid skeletal remains is generally consistent with that of red deer: the tibia is the most abundant long bone, and all represented long bones are more frequent than axial elements (with the exception of teeth). As with red deer, the majority of teeth are from the mandible rather than the cranium. The small sample size precludes a more in-depth discussion of large bovid skeletal part frequencies at present. Element NISP NNISP %NNISP cranium (no teeth) 1 0.5 11.1 maxillary teeth 2 1.0 22.2 mandibular teeth 9 4.5 100.0 other teeth 2 ?/? N/A cervical vertebrae 1 0.1 3.2 thoracic vertebrae 1 0.1 1.7 rib 4 0.2 3.4 humerus 1 0.5 11.1 femur 1 0.5 11.1 tibia 3 1.5 33.3 phalanx 1 1 0.1 2.8 phalanx 1 or 2 2 OJ ZS_ Total 28 Table 6.2: Bison skeletal part frequencies in the Cj Mousterian subsample. Mortalityprofiles As at many Palaeolithic archaeological sites in Western Europe, data concerning prey age and sex are very limited in the study sample. Data from deciduous teeth and permanent tooth wear are the primary basis of the discussion of age profiles offered here. Few long bone epiphyses are present in the sample; and those present give only very coarse information on prey age. For these reasons, epiphyseal fusion data is not presented. The highly fragmented nature of the assemblage and the lack of articular ends prevented the collection ofOsteometrie data that might permit a quantitative analysis of sex-ratios at the site. Therefore, the sex structure of the Gatzarria assemblage also remains unanalyzed for the time being, due to the lack of relevant data. Substantial size variation observed in metatarsals in the Cj sample, as indicated by cortical thickness and/or size of the posterior ridges, suggests that both large and small red deer are present in the assemblage. This variation may reflect the presence of males and females, juvenile remains and/or, possibly, of two different forms of Cervus (C. elaphus and C. simplicidens; see Guadelli 1996). Although antler may sometimes provide data on the sex and age composition of an assemblage, only one small fragment of reindeer antler is present in the Cj Mousterian sample. This specimen provides no indications as to the age or sex of the animal. Regardless, given that reindeer is only a very minor component in the assemblage, this piece might not be very informative about the most common hunting activities undertaken from the site. Red deer antlers are completely absent in the sample. This could be the result of a number of factors, including biases against males in the hunted assemblage, against the transportation of antlers (which have little, if any, food value), or against either male crania or red deer crania in general at any time of year. Deciduous teeth are infrequent in the assemblage. For red deer, deciduous teeth include one lower incisor ( Seasonality Seasonality is a critical component of foraging strategies in any climate. Prey availability can be strongly dependent on seasonal migrations or aggregations of animals, changes in the distribution ofplant foods consumed by herbivores, or in the distribution of precipitation, including snow and ice cover. This variability impacts human choices of which prey and which patches to exploit (e.g., Smith 1991). Data on the age of animals, including fetal remains, in the assemblage can provide insight into the timing of foraging activities in the annual cycle. Three fetal specimens are present in the assemblage. The specimens (Figure 6.1) appear to belong to an artiodactyl, but in the absence of a reference collection they could not be assigned to species. Nevertheless, since the wild ungulates in the assemblage give birth in the spring, the presence of these specimens probably indicates procurement during the winter. More specifically, if the fetal remains belong to red deer, they may indicate capture or death in the early winter, given their small size. ¦ I I Figure 6.1: Fetal remains from the Cj Mousterian sample. The bar represents one centimetre. One deciduous fourth premolar from a red deer permits a rough estimate of seasonality. This tooth, which is slightly worn, probably comes from an individual older than two months but younger than eight months (Lowe 1967; Mariezkurrena 1983). Given a probable birth in early June, this would suggest death or capture between August and February. Therefore, the site would likely have been occupied during the fall or winter. Both indicators of seasonality present in the sample suggest that, minimally, Gatzarria was occupied during the winter, although an occupation during the fall may also be signaled. However, with the scant information available, the possibility of occupations in other seasons cannot be ruled out. Patch use and diet breadth The previous sections established that Mousterian foragers at Gatzarria were exploiting, at minimum, young adult and prime-adult individuals of various ungulate species, and that the site was probably occupied, at least on one occasion, during the fall or winter. This section will first consider how the two resource patches defined in Chapter 4—the lowland and mountain patches—were utilized, and secondly, how the taxonomic data relates to foraging yields, as can be assessed based on diet breadth. The vast majority of foraging activities from Gatzarria were concentrated on the lowland patch. Species that would have been encountered primarily in this patch constitute 90% of ungulate NISP. The remaining 9-10% (depending on whether reindeer is considered a mountain species) of identified ungulate specimens belong to taxa from the mountain patch. Therefore, it appears that hunting activities were quite heavily concentrated at lower elevations, on terrain that was more easily negotiated, and/or where larger ungulates were more abundant. Focusing on diet breadth within the lowland patch, it is apparent that the ungulate contingent available to Mousterian foragers at Gatzarria was diverse and spanned a wide range of profitability in terms of total mass and fat mass. Excluding megaherbivores (mammoth and rhinoceros), ungulate body sizes ranged from the massive 700 kg bison to roe deer, weighing only 26 kg, while body fat content ranged from 74 kg in giant deer in peak condition to about 1 kg in roe deer (Table 4.1). Within this patch, encounter rates with higher-ranked animals, such as bison and red deer, must have occasionally been low enough that even small lean ungulates, such as roe deer (which constitute 3.2% of the assemblage) were taken. Assuming that the body mass and body fat scales reasonably approximate how preys were ranked by foragers, relatively high-ranked species such as giant deer and horse (0.5% and 2.9% respectively; see Table 5.1) were apparently infrequently encountered. 151 Turning to the mountain patch, the entire spectrum of ungulate species available appears to have been exploited as well. Chamois is the best-represented species, even though it was the lowest-ranked species in the patch. Ibex and reindeer, the higher-ranked species that may have been part of the mountain patch, were probably less regularly encountered. Although chamois is the third most abundant species in the total assemblage (6.9%), after red deer and bison, it is clear that the mountain patch was infrequently utilized. Alternatively, foragers at Gatzarria may have occasionally encountered animals from the mountain patch in the lowland patch. For example, chamois may sometimes have been encountered in forested areas at lower elevations during the winter. In either case, species from the mountain patch were included in the diet breadth. Although many carnivores would have been high-ranked according to both the body mass and fat mass scales, the lack of evidence for human exploitation of carnivores in the Gatzarria assemblage is not unexpected. First, as consumers ranked higher on the trophic chain, carnivores would probably be encountered much less frequently than herbivores, and for this reason would be less visible in the archaeological record. Second, carnivore defense mechanisms, a parameter not incorporated into the ranking scales above, might have caused these animals to be relatively costly to exploit, and hence excluded from the optimal diet (referring here to prey types that were pursued for food because of their high post-encounter return rates). Possibly, successful pursuit of these more dangerous animals conferred prestige onto hunters, but again, evidence for human involvement in the accumulation of carnivores during the Mousterian at Gatzarria is lacking. In summary, the taxonomic data suggest that local encounter rates resulted in a diet breadth that, on average, may have been relatively narrow, and focused on one species with relatively high profitability (red deer). However, it is probable that no ungulate species present in the local environment was, permanently excluded from the diet breadth during the time period sampled. Although the relative frequencies of taxa in the Cj assemblage sample might change somewhat with a larger sample size, the available data demonstrate that maximum diet breadth (Grayson and Delpech 1998) at Gatzarria during the time period sampled included the lowest-ranked ungulate species. If encounter rates with ungulate taxa in the lowland and/or mountain patches declined significantly at Gatzarria, a widening of maximum diet breadth to compensate for reduced foraging returns (in terms of including more species) would have to involve animals from other taxonomic groups (i.e., leporids or birds), or other patches (e.g., freshwater fish). However, there are means of increasing energy extraction without adding new species to the diet or exploiting new patch types—for instance, by relying more heavily on lower-ranked species already exploited, by adjusting carcass transport strategies, or by processing carcasses more intensively. Carcass transport and processing decisions in the Gatzarria Mousterian are analyzed in the following section. Foraging currencies and transport decisions In the following sections, the abundances of red deer skeletal elements are considered in order to assess foraging currencies and transport decisions. As discussed in Chapter 5, due to the general absence of spongy bone in the assemblage, these analyses focus on bone portions of similar density. Specifically, only high-density elements, as determined using shape-adjusted reindeer bone density data (Lam et al. 1999), are used in the following analyses. These high-density portions include teeth, limb bone shafts, and the glenoid portion of the scapula. Although mandibular bone has a high density similar to the long bones, it is excluded from the correlations so that the representation of the mandible is not inflated relative to the cranium, which is only represented by teeth. First phalanges are also included, because they have similar density to long bone shafts and because of their potential importance in the analysis of marrow-cracking and transport of low-utility parts. Second and third phalanges are excluded because they have lower density and may have been more severely affected by carnivore ravaging (this issue is further discussed in the section of carcass transport). The element abundance data used in these comparisons are provided in Table 6.3. Because of the possible over- and under- representation of certain long bone elements in NISP due to differential identifiability, the abundance data is treated ordinally for statistical tests. Element NNISP %NNISP FUI %FUI %saved UMI %UMI tibia 21.0 100.0 5492 53.4 62.5 51.1 100.0 metatarsal 18.5 88.1 . 1795 17.5 50.0 46.5 91.0 metacarpal 17.0 81.0 825 8.0 25.0 19.6 38.4 mandibular teeth 9.5 45.2 1600 15.6 femur 7.5 35.7 10278 100.0 83.3 34.0 66.5 humerus 7.5 35.7 4186 40.7 60.0 22.8 44.6 radius 6.0 28.6 2362 23.0 46.7 26.3 51.5 maxillary teeth (cranium) 2.5 11.9 235 2.3 - - phalanx 1. 0.9 4.2 443 4.3 0.0 3.7 7.2 scapula 0.5 2.4 2295 22.3 0.0 - - phalanx 2 0.3 1.2 443 4.3 0.0 1.8 3.5 phalanx 3 0.0 0.0 443 4.3 0.0 0.9 1.8 Table 6.3: Abundances of selected red deer skeletal portions compared with three utility indices: the caribou FUI (Food Utility Index; Metcalfe and Jones 1988:492, Table 2); the percentage of caribou parts saved for grease rendering in the spring by a Nunamiut woman (Binford 1978:36, Table 1.13); and the caribou UMI (Unsaturated Marrow Index; Morin 2007:77, Table 4). FUI and % saved values calculated following Morin (2010): FUI values for complete long bones calculated by summing proximal and distal values, %saved values calculated using the average ofproximal and distal portions. Indices of nutritional or economic utility are foundational to the application of foraging theory in zooarchaeological interpretation. The basic principle of utility indices is that different parts of an animal have different yields of food products such as meat, marrow, or fat. Thus, skeletal parts can be ranked according to their food value, and as food value increases, it is expected that the frequency with which body parts are transported to the central place and consumed will increase as well. Unfortunately, no utility indices have been derived specifically for either red deer or its North American counterpart, the wapiti. However, the anatomical structure of ungulates is strongly consistent between species (Lam et al. 1999), so utility models constructed even from a small sample ofRangifer tarandus (caribou/reindeer) by Binford (1978), Metcalfe and Jones (1988), and Morin (2007) should provide a reasonable ordinal-scale approximation for red deer. Food utility indices, such as the MGUI (Binford 1978) and FUI (Metcalfe and Jones 1988), rank elements according to the total quantity of useable tissue associated with different skeletal parts. However, Ringrose's (1993) suggestion that ratios of useable weight of skeletal elements may be a better measure of utility than absolute useable weight is an interesting point that may have relevance for central place foragers with load constraints. To evaluate this suggestion, a "%useable weight" index for caribou skeletal elements was calculated and compared to the MUI (Metcalfe and Jones 1988:489, Table 1). Following Ringrose's suggestion, %useable weight was calculated using the formula: [(gross weight - dry weight)/(gross weight)] ? 100, from the weights provided by Binford (1978:16, Table 1.1). When only the high-density elements listed in Table 6.3 are considered, Spearman's rank-order correlation between the MUI and %useable weight is very high (rs = 0.93, ? < 0.001), and is perfect (rs = 1.0) if the skull is excluded, because the MUI value of this element includes a large amount of unuseable cartilage (Metcalfe and Jones 1988). However, this is not the case when all elements are included (rs= 0.34, ? = 0.18, phalanges included only once, following the MUI). This difference is due to elements such as the ribs and pelvis, which have high MUI values, but low proportions of useable tissue. In contrast, long bones (excluding the metapodials) have relatively high useable weight values. Thus, Ringrose' s point is well-taken and may be worth exploring more completely, ethnographically and archaeologically. However, the above calculations demonstrate that a %useable weight index would not provide different insight than the MUI into long bone transport for the present analysis, which focuses on long bones. The FUI, which is the MUI adjusted for the impact of anatomical riders (Metcalfe and Jones 1988), probably better deals with transport considerations for the present analysis, given that it incorporates some aspects of long-bone field-processing costs. As argued in the discussion ofprey rankings, foragers may seek different products from the carcasses they acquire (meat, fat, raw material for tools or clothing production, or fuel for fire). When considering animal use strictly from a subsistence point of view, fat acquisition plays a critical role in survival among highly-carnivorous human foragers (Speth 1983; Speth and Spielmann 1983), and the possible importance of fat was considered in the construction of prey rankings in Chapter 4. What currency did the occupants of Gatzarria seek to maximize when they brought animal parts back to the site? Table 6.3 and Figure 6.2 compare NNISP values for red deer elements of similar density to the caribou FUI (Food Utility Index; Metcalfe and Jones 1988), the frequencies 100% ttib 80% O. (? 60% ,ç 40% H • hum •fern trad 20% i era •phal 0% >scp 0% 20% 40% 60% 80% 100% %FUI 100% • tib • mt 80% • me a U) 60% M Z Z ,P 40% a* hum fem rad 20% phal < sep 0% 0% 20% 40% 60% 80% 100% %saved 100% ? tib mt 80% ,Ç 40% hum 1 fem • rad 20% phal 0% 0% 20% 40% 60% 80% 100% o/oUMI Figure 6.2: Comparison of red deer skeletal part abundances with the caribou FUI; the percentage of caribou parts saved for grease rendering by a Nunamiut woman; and the caribou UMI. Data and references from Table 6.3. of caribou parts saved by Alaskan Nunamiut for bone grease production (Binford 1978), and the caribou UMI (Unsaturated Marrow Index; Morin 2007). The first of these is a compound index measuring the quantity of meat, marrow, and bone grease represented by different skeletal parts. This represents an intuitive measure for utility at Gatzarria, because the occupants of the site presumably consumed meat (as indicated by the presence of cut marks), extracted marrow from long bones (see below), and may have burnt spongy bone portions containing grease (Chapter 5). The second measure approximates a 'Fuel Index' (Morin 2009), as elements selected for grease production would also burn well due to their fat content. The spring, rather than summer, episode of bone selection (Binford 1978:36, Table 1.13) was used in this analysis because these data are more consistent with the seasonality estimate for Gatzarria, which suggested a winter occupation (caribou and red deer would both be in poorer condition in the winter and early spring than in the summer). The final index measures the total quantity of unsaturated fatty acids in marrow-bearing bones, which are a major source of fat in ungulates. If the correlation between the red deer element abundances and the UMI or Fuel Index is higher than the correlation with the FUI, it will suggest that fat content was the major factor in skeletal transport decisions. If the correlation with the Fuel Index is highest, it will suggest that fat content within trabecular bone, used as fuel, was a more important currency; whereas if the correlation is highest with the UMI, it will suggest that unsaturated fat calories in marrow were of greater concern (Morin 2007; 2010). Figure 6.2 suggests that there is a much stronger relationship with the UMI than with the FUI or the grease-rendering episode. The femur appears to be an outlier in the FUI chart, possibly because it is a poorly identifiable element when only present as shaft fragments (Marean and Kim 1998; Morin 2004). However, the femur is under- represented to a much lesser degree in the UMI chart, suggesting that if unsaturated marrow fat content was the basis for transport decisions, femur representation may not be significantly underestimated in the sample. The skeletal abundance data are also poorly predicted by the Nunamiut grease-rendering model, as the tibia and metapodials are strongly over-represented. This is likely because these elements have little trabecular bone and would therefore have little value for grease production. The only major discrepancy between the predictions of the UMI and the red deer skeletal part data involves the metacarpal. This element appears to be over-represented in the Gatzarria assemblage relative to its low UMI value. A possible explanation for this is that the caribou UMI does not approximate the value of the red deer metacarpal as well as other elements, because the red deer metacarpal is larger in relation to the metatarsal than the caribou metacarpal. As a result, the UMI value of red deer metacarpals would be relatively greater (closer to that of the metatarsals) than that of caribou. Despite this slight mismatch between the caribou model and the red deer data, the UMI appears to be a reasonable approximation of how skeletal elements were evaluated by Mousterian foragers at Gatzarria. The correlation with the UMI is higher than the correlations with the FUI and Fuel Index (Table 6.4). Although the rank order correlation with the UMI is not statistically significant, this is probably a result of the small number of elements considered in the correlation (marrow values are available for fewer elements), as well as the fact that caribou are not a perfect model for red deer. The Index T1 £ ? FUI 0.30 0.40 10 Grease Rendering Episode 0.64 0.09 8 UMI 0.72 0.07 7 Table 6.4: Spearman's rank order correlations between skeletal part representation and the FUI, Nunamiut grease rendering episode, and UMI. ? = the number of elements considered in the correlation. Data from Table 6.3. significance of the long bone transport patterns at Gatzarria is further investigated through comparisons with other sites in Chapter 7. Although the sample sizes for other ungulates are very small, they support the suggestion that marrow fat yield was the currency being maximized during transport. The tibia and metatarsal, the elements with the highest UMI values, are the best-represented long bones across all ungulate taxa. Excluding red deer, 49% of ungulate long bone specimens are tibiae (n = 13), and 14% are metatarsals (n = 4). In comparison, 27% of red deer long bone specimens are tibiae and 22% are metatarsals. The abundance of different skull portions can provide further insight into the transport decisions of human foragers at Gatzarria. As mentioned in an earlier section of this chapter, mandibles are much better represented in the red deer assemblage than crania: the NNISP for mandibular cheek teeth (incisors, canines excluded) is 8.0 versus 2.5 for maxillary teeth. This difference also holds for bison (NNISP of 4.5 for mandibular cheek teeth versus 1 .5 for maxillary teeth). This disparity is not related to differences in identifiability, fragmentation, or preservation (including damage due to burning), because both the mandible and cranium are mainly represented by isolated teeth. Bony portions without teeth are not included in the NNISP values listed above, and if included, would only increase the representation of the mandible relative to the cranium. Nor is the difference between the cranium and mandible compensated by NSUTS values: there are no cranial specimens, only mandibular portions, in the NSUTS sample for animals of red deer size. The greater frequency of mandibular teeth therefore suggests that mandibles and crania were subject to different transport decisions. This argument is strengthened by the complete absence of red deer antlers in the Mousterian sample. However, because both the mandible and cranium are primarily represented by isolated teeth, which are easily identified relative to fragmented bone specimens, their representation may be inflated in the NISP sample. Although mandibles may have been transported more frequently than crania, they may still have been rarely transported in comparison to major marrow-bearing elements such as the tibia or metapodials. The low NISP value for mandibular bone (n = 6) supports this suggestion. The possibility of differential transport of mandibles and crania is important. Ungulate skulls are bulky, heavy, and have low nutritional value—both absolutely and relative to weight (Binford 1978). For these reasons, the heads of large ungulates are often discarded at kill/butchery sites by contemporary foragers (Binford 1978; Bunn et al. 1988; O'Connell et al. 1988, 1990). However, the mandible (with the tongue) has a higher food value and is lighter than the skull. For caribou, the food value (MGUI and FUI) of the mandible with tongue is nearly as high as that of the lumbar vertebrae and higher than that of the metapodials (Binford 1978; Metcalfe and Jones 1988). This observation is likely true for other cervids, such as red deer. The mandible also contains a small marrow cavity. Consequently, it can be expected that foragers with limited transport capabilities will frequently disassociate the mandible and transport it alone, rather than transporting or discarding the entire skull (Monn 2010). In contrast, when transport distance is short, foragers may be more willing to accept the cost of transporting a heavy skull (Morin 2010). As demonstrated above, red deer crania were infrequently transported to Gatzarria during the Cj Mousterian occupation. To explain this pattern, it is possible to draw on the currency analysis, which suggested that marrow fat was an important criterion in transport decisions. Crania (without the mandible) have no marrow, as well as a low food value. It is suggested here that even though transport decisions may have been relatively non-selective in terms of long bone transport (see Chapter 7), the heavy, low-utility cranium still remained a very low priority for transport. Mandibles, which have both higher marrow and food values, were somewhat more frequently transported. The transport òf crania and, to a lesser extent, mandibles, may represent potential areas for intensification of carcass transport at Gatzarria. Feet, another potentially bulky skeletal portion with marginal utility (Binford 1978), are infrequent in the red deer assemblage (Table 6.1), and their poor representation requires some explanation as well. The first phalanx is reasonably well-represented (n = 7, %NNISP = 4.2%), but there are only two second-phalanx fragments in the sample, both carnivore-digested, and there are no third phalanges. One sesamoid is also present. There are several possible explanations for the low representation of these elements: differential transport, burning, carnivore ravaging, or post-depositional destruction. Because the density of phalanges decreases distally (Lam et al. 1999), it can be expected that distal phalanges would resist attrition processes—notably carnivore digestion—less frequently than proximal phalanges. Thus, the digestion of both second phalanges in the sample, and the fact that phalanges produce a bad smell when burning (Binford 1978), point towards carnivore ravaging as an explanation for the representation of intermediate and distal phalanges. However, with the available evidence, the possibility of differential transport cannot be completely ruled out. On-site processing ofskeletalparts It was argued above that the transport decisions made by foragers from Gatzarria were based on the nutritional utility, specifically, the marrow fat content of skeletal parts. However, as suggested by the CPF prey choice model (Cannon 2003), transport decisions were probably also significantly affected by transport distances and prey encounter rates. In contrast, patterns of marrow-cracking at the central place are closer to the point of consumption, and as such, reflect more proximate decisions about food needs. Furthermore, once at the central place, 'riders' with low utility, including most small foot elements, would be assessed based on their intrinsic utility, and not based on their anatomical association with high-utility parts, as might be the case for transport decisions (Binford 1978). It seems that the extraction of marrow from red deer long bones at Gatzarria was systematic. The considerable proportion of fractures typical of fresh bone (67%), which usually show a curved fracture edge and a smooth fracture surface, suggests that most of the fractures occurred around the time of assemblage deposition. There are no complete red deer long bones in the Cj Mousterian assemblage. Marrow-cracking is one possible cause of such extensive breakage, although trampling or carnivore-ravaging shortly after human occupation may have also contributed to this type of damage. A total of 27 probable percussion notches were documented on red deer specimens, all on marrow-bearing elements. Some of these may in fact be carnivore tooth notches (Capaldo and Blumenschine 1994); however, six specimens with percussion notches also bear cut or scrape marks, and these provide stronger evidence for human marrow-cracking activity. The long bones of smaller species (ibex, chamois, roe deer) also appear to have been marrow-cracked, as suggested by the presence of green fracture edges and percussion notches on some specimens. Were minor marrow-bearing elements, such as phalanges and mandibles, cracked with any regularity? The mandible is represented only by a few bone (as opposed to teeth) fragments (« = 6). Therefore, although tooth abundance data suggest that mandibles were transported to the site more frequently than crania, it is difficult to determine whether mandibles were regularly opened for the small amount of marrow they contain. One mandible fragment bears a percussion notch and green fractures, suggesting that this mandible was cracked for marrow. However, the regularity with which this behaviour took place cannot be assessed. There are no talus specimens and only one calcaneus fragment in the assemblage, so marrow extraction cannot be assessed for these elements. As discussed earlier, second and third phalanges are virtually absent in the assemblage, possibly due to carnivore ravaging. Red deer first phalanges are more frequent (n = 7), and none show carnivore damage. The possibility of marrow extraction can therefore be evaluated for this element. All of the first phalanges in the assemblage are fragmentary, except for one almost- complete specimen. Four of the broken specimens bear green bone fracture edges. It seems, therefore, that there is positive evidence for marrow-cracking of first phalanges. In summary, the degree to which small marrow-bearing bones were utilized at Gatzarria is uncertain due to small sample sizes. It appears that the site occupants may have, on occasion, been exploiting the marrow of first phalanges and mandibles. If there was room for intensification in marrow-processing during the occupation(s) that created this assemblage, it was limited. Very minor marrow bones, including the second and third phalanges, talus, and calcaneus, may have been the only potential additions to the 'marrow-cracking spectrum' at the site. Finally, due to burning and post-depositional destruction, spongy bone elements have not been considered in this section. Regarding these elements, it should be noted that there is no positive evidence of grease extraction; indeed the presence of some burnt spongy bone fragments argues against the possibility of grease extraction, as bone portions from which grease had been extracted would not burn well (Costamagno et al. 1999; Morin 2010). Discussion Unlike early Upper Palaeolithic foragers north of the Pyrenees, who depended nearly exclusively on reindeer (Grayson and Delpech 1998, 2002; Graysón et al. 2001; Morin 2008), Mousterian foragers at Gatzarria had a wide variety of ungulate fauna available to them. Numerous large ungulates inhabited the landscape, and were distributed across a wide size range, from chamois and roe deer to woolly rhinoceros and mammoth. In the lowland patch, the foragers occupying Gatzarria concentrated on some of the most profitable prey (red deer and bison). Possibly, Gatzarria was selected as a camp site to take advantage of high local abundances of red deer during certain seasons. The animal portion of the diet at Gatzarria would generally have been heavily dominated by this species. Nevertheless, alternative, potentially less profitable resources, such as roe deer, the lowest ranked ungulate in the lowland patch, were apparently included in maximum diet breadth during the time period sampled. This being considered, there was little room for dietary expansion in terms of the number of ungulate species exploited, except perhaps for the addition of mammoth and rhinoceros, depending on how the size and strength of these animals affected the probability of a successful kill. If chamois and ibex occupied a distinct mountain or rocky habitat patch, largely exclusive of other larger ungulates such as red deer, bison, and horse, then this may be considered a low-ranked patch that was exploited less frequently, perhaps when the main terrestrial patch was unproductive—for instance, if red deer were at a low point in population fluctuations. In other words, the mountain patch might have been a type of subsistence 'safety net,' or fallback resource. Carcass transport patterns for red deer suggest that transport decisions were primarily based on yields of marrow fat. High-ranking marrow bones—namely, the tibia and the metapodials—were transported more frequently than other elements. Marrow- bearing parts selected for transport were processed intensively. All red deer long bones were probably cracked for marrow, and first phalanges and mandibles appear to have been exploited for their marrow when transported. Whether smaller marrow-bearing elements, such as second and third phalanges, were transported and cracked for marrow is uncertain. Possibly, these elements were brought to the site but largely destroyed by carnivore action. Nevertheless, it is apparent that there was limited room for improvement in fat extraction at the site without resorting to marginal marrow elements or to extracting bone grease. Increased processing and, possibly, increased transport of low-ranked elements, such as crania and phalanges, could be a means of acquiring additional fat calories—again, a potential 'safety net' for when nutritional requirements were not being met, but one that would bring only very marginal increases in total energy extraction. The implication of these observations in terms of the overall subsistence security ofNeandertals in the Atlantic Pyrenees might be strongly dependent on site seasonality. Non-migratory ungulates, such as red deer, would likely have been available year-round from the site, but their condition, group size, and distribution on the landscape would have varied throughout the year. The seasonality data suggest that Gatzarria was probably occupied sometime during the fall or winter months. During the winter, male red deer would have depleted fat reserves, while the condition of females would deteriorate as the winter progressed (Clutton-Brock and Albon 1989). It might therefore be expected that late winter and spring were stressful seasons for human foragers in terms of the availability of fat calories. If Gatzarria was occupied only during the fall or early winter, the site's occupants may have faced comparatively harder times during late winter and early spring. If occupations were year-round or during seasons other than winter—possibilities that cannot be excluded with the available data—it should be considered that diet breadth, patch use, and transport decisions might have varied throughout the year. In this case, roe deer and chamois, or marrow from mandibles and phalanges, may have been seasonal components of the diet, for instance if the declining fat and body mass of red deer during the winter forced a temporary broadening of diet breadth and an intensification of carcass utilization. Additional seasonality data from other sites in the region would help to clarify this issue. In conclusion, although red deer, a relatively high-ranked species in the lowland patch, appears to have formed the bulk of the diet, low-ranked ungulate species in both the lowlands and mountain patches were, on occasion, included in the diet. Although crania were infrequently transported, exploitation of marrow bones was intensive. It appears that there was limited room for increasing energy (or fat) extraction in terms of the exploitation of red deer—opportunities for intensification would have been limited to elements of marginal utility (such as the skull), and possibly, bone grease. Beyond these measures, the occupants of Gatzarria would have had to rely more heavily on 'safety nets' such as low-ranked species and lower-ranked patches, or incorporate additional species into the diet, in order to increase their acquisition of food energy. The next chapter turns to a consideration of whether the ungulate resource exploitation behaviours of the Mousterian foragers at Gatzarria are more generally representative of late Mousterian foraging in the Atlantic Pyrenees, and what significance these patterns may have for the early Upper Palaeolithic in the region. Chapter 7: Neandertal subsistence at Gatzarria in regional and temporal context The analysis of foraging at Gatzarria carried out in the previous chapter will now be considered in the context ofNeandertal subsistence adaptation in the Atlantic Pyrenees and southwestern France, beginning with comparisons between Gatzarria and other transition sites in the Basque country and beyond. The significance of the foraging strategies pursued at Gatzarria for our conceptualization of the Neandertal niche is then explored, and the results of the faunal analysis and regional comparisons are used to discuss expectations for subsistence change during the transition at Gatzarria. Late Middle Palaeolithicforaging in southwestern France and northern Iberia Despite considerable interest over the past decades in the Middle to Upper Palaeolithic transition, the fauna from many transition sites remains poorly studied, as modern zooarchaeological studies are a relatively recent development and have sometimes been hampered by the selective discard of bone during excavations (e.g., Grayson and Delpech 2008). Numerous sites are currently undergoing renewed zooarchaeological study. However, as was the case in this study, most sites will require careful taphonomic analysis before conclusions regarding foraging behaviours can be drawn. The geographic locations of the sites considered in this chapter are illustrated in Figure 7.1. 169 y -«a. s •~<. C- < 7j> i J ÄS«v- / "> V^- / ? V j r ? ^ K..r í ? T-^r t ¦? r ^t. ¿s T f ?£? \ ^ S f**,c~Jp" S ^-*v-*^Í„¿>> S Ha ~> \jt ? >*«*~î. Q. ~~-·'?- ^ X °Y 'W V "N. (^ ( 1 K> ^ i V / . y / B »^v. / f-' ^C ^ «w ^ V* V /" >" /" / O 125 250 500 Kilometers I ? ? ? ? I ? ? ? ? I Figure 7.1: Locations of selected sites discussed in the text. A: Gatzarria, B: Isturitz, C: Les Abeilles, D: Axlor, E: El Castillo, F: Grotte XVI, G: Saint- Césaire. Close to home: Middle Palaeolithic sites in the Basque Provinces and Cantabria Other assemblages from Gatzarria will eventually form the primary basis of comparison for the Cj Mousterian assemblage. Preliminary data on the Cjr Mousterian layer (the Mousterian occupation preceding Cj) suggest minimal differences with the Cj Mousterian subsample (Table 7.1). Due to the small sample size and preliminary nature of the data, an in-depth comparison will not be undertaken at this point. Gtz Gtz Axlor Axlor Castillo Cj Moust Cjr B-C D Level 20 Species ? = 369 ? = 59 ? = 149 /? = 793 ? = 2581 red deer 72.6 76.3 21.4 35.1 60.9 bison/aurochs 7.6 11.9 27.5 32.5 21.1 chamois 7.0 3.4 0.6 0.8 1.1 roe deer 3.3 - 2.0 - 1.3 horse 3.0 1.7 20.8 5.3 13.4 mammoth 2.7 - - ibex 2.2 6.8 27.5 25.7 0.1 reindeer 0.8 giant deer 0.5 - - woolly rhino (Coelodonta) 0.3 - - - 2.1 rhino (Dicerorhinus) - - - 0.5 wild boar - - _- O1I -_ Total 100.0 100.1 99.8 100.0 100.0 Table 7.1: Percentages of ungulates in Mousterian levels at Gatzarria (Cj Mousterian and preliminary data on Cjr), Axlor (Levels B-C and D; González Urquijo et al. 2004:15, Figure 9), and El Castillo (Level 20; Dari and Renault- Miskovsky 2001:132, Table 1). Altuna (1990) provides a review of Palaeolithic faunal assemblages from the French and Spanish Basque regions, including the Middle Palaeolithic assemblages from Axlor and Lezetxiki. These sites are briefly presented here, in addition to Les Abeilles, a site in Haute-Garonne, France, with Middle and early Upper Palaeolithic deposits. It should be emphasized, however, that the Middle Palaeolithic occupations of these sites are generally not precisely dated. The Mousterian levels of several other sites in the Basque provinces—including Isturitz and Abri Olha—were eliminated from comparisons in this study because of lack of data and/or taphonomic uncertainty. For instance, both Isturitz and Olha contain high proportions of carnivores in the Mousterian levels (Straus 1982; Altuna 1990). No seasonality estimates are presently available for Mousterian sites in the Basque country, except for Gatzarria. This study has suggested that, during the period represented by the Cj Mousterian sample, the site was occupied, at least on one occasion, during the fall and/or winter. Faunal remains from the Middle and early Upper Palaeolithic levels at Les Abeilles (Laplace et al. 2006) have been briefly studied (Altuna 2006), but the Mousterian sample size is very small (n = 61). Red deer, large bovids, reindeer, and horse are present at Les Abeilles, while the smaller species—ibex, chamois, and roe deer—are absent. Les Abeilles also contains a high proportion of carnivores (at least 40% of all remains in the Mousterian and early Upper Palaeolithic levels), suggesting that they may have made a relatively important contribution to the ungulate assemblages. The archaeostratigraphy of the Middle to Upper Palaeolithic transition levels at Lezetxiki, a well-known Palaeolithic site in the Vasco-Cantabrian province of Guipúzcoa, has been the subject of some controversy. Various authors have disagreed as to whether levels III and IV at the site contain Mousterian or Upper Palaeolithic industries (Falguères et al. 2005). Hopefully, data from recent excavations, begun in 1996, will eventually clear up these issues. Dates for the Mousterian level Vb are uncertain (Falguères et al. 2005), and this level yielded very few ungulate remains during the mid- twentieth century excavations at the site. Straus (1982) suggests that Mousterian occupations at Lezetxiki were ephemeral, based on the prevalence of bear remains and the low density of lithic artifacts in levels VI and VII. The cave was apparently heavily used as a bear hibernation site during the deposition of these layers. However, levels VI and VII probably date to more than 200 kya (Falguères et al. 2005), and as such, are not useful for comparisons with the late Mousterian in the region. The Mousterian site of Axlor, originally excavated in the 1960s and 70s by J.M. de Barandiarán, has recently undergone renewed excavation and study (González Urquijo et al. 2005; González Urquijo et al. 2004). The excavation methods described by González Urquijo et al. (2004) indicate that bone was fully recovered during the recent excavations. Bone material from Level D at Axlor has been AMS radiocarbon-dated to 42 010 ± 1280 BP (González Urquijo et al. 2004), confirming that the site is appropriate for comparisons with other late Mousterian and early Upper Palaeolithic sites in the region. Level D is superposed by one other Mousterian occupation level (B-C). Lithic tools from these layers are characteristic of the Quina Mousterian (González Urquijo et al. 2004). Preliminary taphonomic analyses of levels B-C and D suggest that carnivore marks are rare and that the fragmentation pattern, notably a high frequency of long bone shaft fragments, is consistent with a primarily anthropic accumulation (González Urquijo et al. 2004). The frequency of carnivores at the site is low (only 1 of 793 identified specimens in Level D). Data on the ungulate assemblages from Axlor levels B-C and D are compared to the Gatzarria Cj Mousterian sample in Table 7.1. A greater range of species is present at Gatzarria, but the Axlor assemblages show a much more even partitioning between species. Red deer are much less common, while large bovids, ibex, and, in level B-C, horse, are better represented at Axlor. The difference in species representation between the two sites, particularly the greater representation of ibex at Axlor, may be an effect of local topography. The landscape surrounding Axlor is characterized by steep slopes, rocky surfaces, and frequent karstic features (González Urquijo et al. 2004). If the same patch divisions used in the analysis of the Gatzarria fauna can be applied, this suggests that mountain patches, and the species therein, would have been more frequently encountered near Axlor than around Gatzarria. However, the differences in taxonomic representation may also be due to differences in site seasonality, site function or chronological change. These possibilities cannot be further investigated at the present time. Middle to Upper Palaeolithic transition sites in Cantabria, such as El Castillo, Cueva Morín, and El Pendo are also relevant for comparisons with Gatzarria, given their geographical proximity to the Basque provinces and their prominence in debates over the timing of the earliest Upper Palaeolithic in Western Europe (e.g., Zilhäo and d'Errico 2000). Unfortunately, faunal data are not available for the most recent Mousterian levels at Cueva Morín (Altuna 1971), and the sample sizes from Mousterian levels XI-XIII at El Pendo are very small (Fuentes Vidarte 1980; Pike-Tay et al. 1999). However, the large mammals from the most uppermost Mousterian level at El Castillo—Level 20, which is dated to 45^13 000 BP—have recently been studied by Dari and Renault-Miskovsky (2001). These authors suggest that Level 20 is primarily an anthropic deposit, as the assemblage is highly fragmented, contains a high percentage (27%) of anthropically- marked specimens and a low percentage of carnivores (2%). Taxonomic abundances for this level are presented in Table 7.1. The faunal spectra of the Gatzarria Cj Mousterian and El Castillo Level 20 assemblages are, in fact, more similar than those at Gatzarria and Axlor. Red deer constitute the majority of both assemblages. However, bison and horse are more abundant at El Castillo, while small ungulates, particularly those from the mountain patch, are better represented at Gatzarria. Dari and Renault-Miskovsky (2001) suggest that the poor representation of mountain species at El Castillo is a result of the gentle terrain around the site. However, roe deer are also rare in the assemblage, which suggests that the poor representation of small ungulates at El Castillo might involve additional factors besides local encounter rates (such as prey return rates). Pike-Tay et al. (1999) suggest that the Mousterian occupations at El Castillo were year-round, but most animals were captured between late fall and early summer. This conclusion is similar to the estimate of a fall or winter occupation at Gatzarria. Overall, comparisons with sites in the Basque country and Cantabria are limited by a lack of data and by a need for taphonomic studies. Nevertheless, it seems that many cave sites in the Basque Provinces (Isturitz, Abri Olha, Les Abeilles, and Lezetxiki) may have been frequently occupied by carnivores during the Mousterian period. In comparison, Mousterian foragers may have utilized Axlor and Gatzarria relatively heavily, whether through longer or more repeated occupations. Disregarding the potential impact of carnivores, the faunal spectra of the Basque Middle Palaeolithic sites, as well as El Castillo, are similar in the sense that they contain a wide diversity of ungulate species. Regarding Mousterian sites in the Basque region, Altuna (1990:233) has suggested that "the faunal spectra appear to be strongly linked to the environment surrounding the site [. . .] hunting is opportunistic, not selective, and all, or nearly all, of the species in the local area were exploited."5 New data from Axlor and Gatzarria appear to confirm that most ungulate species were included in the maximum diet breadth at these sites. Consistent with Altuna's argument, there is also substantial variation between sites in the abundances of ungulate species. In particular, the differences between sites (e.g., between Gatzarria and Axlor) are much greater than differences through time at the same site (e.g,. between the Cjr and Cj deposits at Gatzarria, or between levels B-C and D at Axlor). This pattern may suggest that the 5 "El espectro faunisto parece estar mucho más ligado al medio ambiente circumdante al yacimiento [. . .] Ia caza es oportunista, no selectiva y que se explotan todas o casi todas los especies existentes en el entornado." 175 foraging strategies of Neandertals in the Pyrenees region were responsive to local shifts in the availability of ungulate species. Potentially, site functions were specialized according to these differences. However, without adéquate control over chronology, seasonality, and climatic variation, such suggestions must remain speculative. Looking north: Comparisons with Saint-Césaire and Grotte XVI How might late Middle Palaeolithic foraging have differed between the Atlantic Pyrenees and further north in western France, where the climate was more severe? Sites with sequences across the Middle to Upper Palaeolithic transition, where faunal remains were thoroughly recovered and have been recently studied in detail, are the most appropriate for comparisons with Gatzarria. Comparisons here focus on Saint Césaire and Grotte XVI. Both sites have been the subject of detailed zooarchaeological and taphonomic analyses (Grayson et al. 2001; Grayson and Delpech 2003; Morin 2004, 2008, 2009, 2010; Morin et al. 2005; Faith 2007). Morin (2010), through comparisons with Grotte du Renne and Abri Pataud, suggests that the Saint-Césaire sequence is representative of the Middle to Upper Palaeolithic transition in western France. However, sites in the Pyrenees were not included in his analysis. Grotte XVI is located further south than Saint-Césaire and, as a result, the Mousterian fauna at Grotte XVI may more closely resemble that at Gatzarria. The Mousterian sample from Saint-Césaire consists of selected décapages from layer EGPF, which contains a Denticulate Mousterian industry. Layer EGPF has been dated by thermoluminescence to approximately 40 thousand years ago (Mercier et al. 1993). The EGPF sample studied to date consists of 21 084 specimens, including a post- refit NISP of 867 (Morin 2010). Both the Saint-Césaire EGPF and the Gatzarria Cj Mousterian occupations may have occurred during the winter, although while the Gatzarria occupation may be during the fall and/or winter, that at Saint-Césaire appears to have been in the winter, spring, and perhaps, early summer (Morin 2010). In both cases, these seasonality estimates are not exclusive of other seasons and are based on limited data. The taphonomic histories of Saint-Césaire and Gatzarria are somewhat different, but do not preclude comparisons between the sites. Levels of fragmentation in the two Mousterian assemblages are similar (see Chapter 5), although bone surface preservation is better at Saint-Césaire than at Gatzarria. Spongy bone may have been used as fuel at both sites, although the evidence is much stronger in the Denticulate Mousterian of Saint- Césaire, where 35% of all bone fragments are burnt, in comparison to 3.3% in the Gatzarria Cj Mousterian sample. However, sieving and collection of small bone fragments were presumably less thorough at Gatzarria. Frequencies of carnivore remains are low in both assemblages: 1.9% of identified remains at Gatzarria and 0.4% at Saint-Césaire. Carnivore action was more significant at Gatzarria, where 4.9% of identified ungulate remains are affected by carnivores (gnawing and digestion), as opposed to Saint-Césaire, where only 0.1% of identified remains from the Denticulate Mousterian show carnivore marks. Because of the destructive effect of burning at Saint-Césaire, in Morin's studies (2004, 2009, 2010), analyses of skeletal parts focused only on parts of similar density. The same procedure has been followed in this work, due to carnivore attrition and post-depositional destruction as well as burning. As a result, skeletal part abundances at both sites should be comparable. As at Saint-Césaire and Gatzarria, close attention was paid to long bone shafts during identification of the Grotte XVI material (Grayson et al. 2001; Grayson and Delpech 2003). Couche C at Grotte XVI, which contains a Mousterian of Acheulean Tradition assemblage (Karkanas et al. 2002), is the latest Mousterian level at the site, dated by thermoluminescense to approximately 64 600 years ago (Grayson et al. 2001). A total of 540 ungulate specimens have been identified in the Couche C sample. Although cave bear remains are very abundant in Couche C at Grotte XVI (? = 585), only 1.7% of ungulate NISP are carnivore-marked, suggesting that the role of bears and other carnivores in accumulating and altering the assemblage was minimal. Evidence for bone burning at Grotte XVI has not been discussed in publications on the bone assemblages. Karkanas et al. (2002) demonstrate the presence of grass and wood phytoliths in ash from Couche C, but burnt specimens are not mentioned in their analysis. In some parts of Couche C, bone material may have been destroyed by mineral diagenesis (Karkanas et al. 2002). It should be pointed out that fish remains are abundant in Couche C at Grotte XVI (Karkanas et al. 2002). However, no taphonomic analysis of these remains has yet been published. The taxonomic compositions of the Mousterian assemblages from Saint-Césaire, Grotte XVI, and Gatzarria are compared in Table 7.2. In contrast with Gatzarria, neither the Saint-Césaire or Grotte XVI assemblages are heavily dominated by one species. However, the Gatzarria and Grotte XVI assemblages are similar in terms of species richness. The importance of low-ranked roe deer is greater at Grotte XVI than at Gatzarria. In contrast, despite the larger sample size (which should increase the probability of sampling rare species), the Saint-Césaire sample is less rich than the Grotte Gtz St.-Cés. Grotte XVI Cj Moust EGPF C Species « % ? % ? %_ red deer 268 74.7 9 1.1 183 33.9 bison/aurochs 28 7.8 329 38.4 44 8.1 chamois 26 7.2 - - 34 6.3 roe deer 12 3.3 - - 77 14.3 horse 11 3.1 295 34.5 37 6.9 ibex 8 2.2 - 12 2.2 reindeer 3 0.8 214 25.0 142 26.3 giant deer 2 0.6 7 0.8 - woolly rhino 1 0.3 2 0.2 - wild boar - - 11 2.0 Total 359 100.0 856 100.0 540 100.0 Table 7.2: Abundances of ungulates in the Gatzarria Cj, Saint-Césaire EGPF (Morin 2004:141, Table 13), and Grotte XVI Couche C (Grayson and Delpech 2003:1636, Table 2) samples. Mammoth is excluded, as information on this species is not available for Grotte XVI. XVI and Gatzarria samples. The Saint-Césaire EGPF assemblage is dominated by three species (bison, horse, and reindeer), while other species are very poorly represented (< 1% of remains). At both Grotte XVI and Gatzarria, it appears that maximum diet breadth included the full range of ungulate species available, except possibly mammoth and rhinoceros. Although ibex and chamois may not have been available around Saint-Césaire due to the local topography, it is less clear why roe deer are not represented in the sample from this site. Possibly, roe deer were excluded from the optimal diet at Saint-Césaire during the time period sampled. However, it is more likely that the absence of this taxon simply reflects its rarity in the landscape, due to a colder climate and low availability of forest cover. The prevalence of horse, a species that lives in open areas, at Saint-Césaire may support this argument. Thus, it can be suggested that diet breadth at all three sites was qualitatively similar in terms of the range of taxa exploited. Patterns of skeletal representation at Gatzarria and Saint-Césaire can be compared using reindeer, for Saint-Césaire, and red deer, for Gatzarria, as these two species have similar anatomy and are represented by adequate sample sizes at the respective sites. Comparisons with Grotte XVI also focus on the reindeer assemblages at that site. The published data on red deer skeletal part abundances at Grotte XVI (Grayson and Delpech 2003) is not sufficiently detailed for comparison with Gatzarria—mandibular and cranial elements are not counted separately, nor are metatarsals and metacarpals distinguished. The skeletal part representations of high-density bone portions of red deer and reindeer at Gatzarria and Saint-Césaire (Table 7.3) are strongly correlated (rs = 0.84, ? = 0.002). As at Gatzarria, the Saint-Césaire reindeer assemblage correlates better with the UMI than the Nunamiut grease rendering episode or the FUI (Morin 2004, 2010). This indicates that marrow fat acquisition may have been the major factor in transport decision-making for both populations of foragers. Consequently, similar patterns of skeletal representation are expected, at least for major marrow-bearing elements. Gtz Cj Moust St.Cés EGPF Bone portion NNISP %NNISP NNISP %NNISP shaft tibia 21.0 100.0 18.0 49.3 shaft metatarsal 18.5 88.1 36.5 100.0 shaft metacarpal 17.0 81.0 6.0 16.4 mandible 12.5 59.5 5.0 13.7 shaft humerus 7.5 35.7 5.0 13.7 shaft femur 7.5 35.7 4.5 12.3 shaft radius 6.0 28.6 8.5 23.3 maxillary teeth 2.5 11.9 2.0 5.5 phalanx 1 0.9 4.2 0.3 0.8 * scapula (glenoid) 05 2A U) 2.7 NISP 193 175 Table 7.3: Skeletal element abundances for selected high-density portions of red deer at Gatzarria and reindeer at Saint-Césaire (data from Morin 2004:395-396). 180 Differences between the Gatzarria and Saint-Césaire assemblages mainly involve the relative abundances of long bones with lower UMI values. The tibia and the metatarsal are the two most frequent elements in both assemblages, although the metatarsal, not the tibia, is most abundant element at Saint-Césaire. However, the most important difference between the sites may not be reflected in the rank order correlation. This difference is the elevated abundance of the metatarsal relative to other elements at Saint-Césaire (Figure 7.2). The representation of elements is much more even at Gatzarria. 100 - 90 - 80 - I "GtZCjMoUStI 70 - |ClSt.CésEGPFI 60 <*! Z 50 Z 40 - 30 - 20 10 0 ^ ^. <*> *S» .^ -4> ^ J? / /> ** ^ *'F *P ^" ^ SS^ Figure 7.2: Comparison of frequencies of high density skeletal parts of red deer in the Gatzarria Cj Mousterian sample and reindeer in the Saint- Césaire EGPF sample (data from Table 7.3). The correlation between the Gatzarria and Grotte XVI Mousterian assemblages is slightly weaker than the correlation between Gatzarria and Saint-Césaire, but is also significant (rs = 0.77, ? = 0.01). This correlation is based on a different, and slightly 181 narrower, set of element abundances than the correlation between Gatzarria and Saint- Césaire (Table 7.4). For comparability with the data provided by Faith (2007) and Grayson and Delpech (2003), phalanges and teeth had to be removed from the correlations. Following Faith's (2007) procedure, the number of metapodial shaft specimens at Gatzarria (n =12, Table 6.1) was divided by two and added to the metatarsal and metacarpal counts before re-calculating NNISP. Gtz Cj Moust Grotte XVI C Bone portion NNISP %NNISP NNISP %NNISP metatarsal 21.5 100.0 19 100.0 tibia 21 97.7 6 31.6 metacarpal 20 93.0 2.5 13.2 femur 7.5 34.9 3.5 18.4 humerus 7.5 34.9 3 15.8 radius 6 . 27.9 7.5 39.5 mandible 3 14.0 1 5.3 scapula 0.5 2.3 0 0.0 cranium 0 0.0 0 0.0 Total NISP 174 85 Table 7.4: Skeletal element abundances for selected high-density portions of red deer at Gatzarria and reindeer at Grotte XVI (data from Grayson and Delpech 2003:1638, Table 5; Faith 2007:2007, Table 3). Despite these differences in the treatment of the data, the same pattern identified in the comparison with Saint-Césaire is present. This is clearly shown in Figure 7.3. Once again, the main difference between the sites involves the abundances of low-ranked long bones. Elements other than the metatarsal are poorly represented at Grotte XVI, while at Gatzarria, there is a more even representation of elements. 182 100 - 90 - ¦GtzCjMoust 80 - El Grotte XVI C 70 - 60 - P. g 50- ?? 40 - 30 - 20 - 10 - 0 - Figure 7.3: Comparison of frequencies of high density skeletal parts of red deer in the Gatzarria Cj Mousterian sample and reindeer in the Grotte XVI Couche C sample (data from Table 7.4). None of the correlations between the Grotte XVI assemblage and the utility indices considered in this study (the UMI, the Nunamiut spring grease rendering episode, and the FUI) is significant (Table 7.5). This is probably a result of the small number of elements considered in the correlations. Nevertheless, the correlation with the UMI is higher than correlations with either the grease rendering episode or the FUI. This being considered, along with the fact that the skeletal part patterns of the Gatzarria and Grotte XVI Mousterian assemblages are significantly correlated, there is no reason to suggest that the criteria on which transport decisions were based differed at these sites. The differences in skeletal representation observed in Figures 7. 1 and 7.2 may be partially explained by anatomical differences between red deer and reindeer, which may facilitate identification or slightly alter how skeletal portions would be assessed by foragers. For instance, as suggested in Chapter 6, the red deer metacarpal may have been Index r g n_ FUI 0.50 0.17 9 Grease Rendering Episode 0.43 0.35 7 UMI 0.71 0.14 7 Table 7.5: Spearman's rank order correlations between skeletal part representation at Grotte XVI Couche C and the FUI, Nunamiut grease rendering episode, and UMI. ? = the number of elements considered in the correlation. Data used in the correlations provided in Table 7.4; %saved and UMI values not available for the mandible and cranium. relatively higher-ranked in comparison to that of reindeer. However, skeletal differences between the two species are limited. The degree of fragmentation at Gatzarria and Saint- Césaire is similar and also cannot explain the different patterns of element representation. The more even skeletal representation at Gatzarria may be explained by less stringent culling decisions at kill sites. This possibility is further explored below. The selectivity of human foragers in transporting the skeletal parts of large prey can be assessed by applying measures of evenness, diversity, and/or mean utility to skeletal abundance data (Broughton 1999; Cannon 2003; Faith 2007; Faith and Gordon 2007; Morin 2010). As a result of the anatomical differences between caribou and red deer, mean utility (specifically, mean UMI) is not used here. Measures of evenness and diversity circumvent this problem, because they make no assumptions about utility (Faith 2007). The measure of diversity used here is the reciprocal of Simpson's Index (l/D; Simpson 1949). This approach is preferred for this analysis because of its lesser sensitivity to sample size and richness in comparison to the Shannon Index (Magurran 2004). Simpson's Index is calculated using the following equation, where n¡ is the frequency of the z'th element, and N is the total of all elements: D = ZmÍMlzzXI N(N- 1) The diversity value calculated for the Gatzarria Mousterian sample is presented in Table 7.6, along with diversity values for reindeer assemblages from Grotte XVI. NNISP and diversity values for the Grotte XVI samples were calculated from Faith (2007:2007, Table 3). Diversity for the Gatzarria sample was calculated using the NNISP values provided in Table 7.4, excluding the scapula, because this element was not included in Faith's analysis. Sample 1/D XNISP 1/D(LB only) INISP GtzCjMoust 5.38 203 5.03 167 Grotte XVI Mousterian (C) 4.01 85 3.82 83 Châtelperronian (B) 5.68 141 4.47 121 Aurignacian? (Aib) 5.28 113 4.27 100 Aurignacian (Abb) 6.54 409 5.00 337 Gravettian (Abe) 6.25 462 4.69 370 Solutrean (As) 6.83 177 4.99 139 Magdalenian (0) 6.99 722 5.22 573 Saint-Césaire Mousterian (EGPF) - - 3.52 162 Châtelperronian? (Ejop inf) - - 3.72 65 Châtelperronian (Ejop sup + Ejop 8-9) - - 3.68 107 Proto-Aurignacian (Ejo sup) - - 3.01 194 Early Aurignacian (Ejf) - - 3.32 1745 Evolved Aurignacian (Ejm) - - 3.63 386 Evolved Aurignacian (Ejj) - - 3.34 143 Table 7.6: Diversity of the Gatzarria, Grotte XVI and Saint-Césaire samples, as measured by the reciprocal of Simpson's Index. LB = long bones. Procedures used to calculate diversity values described in the text, data from Tables 7.3 and 7.4, Morin (2004:395-396), and Faith (2007:2007, Table 3). Diversity values for the long bone samples (i.e., cranial and mandibular bone excluded) from Gatzarria, Grotte XVT, and Saint-Césaire are also presented, for two reasons. The first reason is for comparability with the Saint-Césaire assemblages, as counts of teeth and bony portions are not provided separately for the cranium and mandible (Morin 2004:395-396). The second reason is that skulls may have been subject to different transport decisions than major marrow-bearing elements. Morin (2010) argues that this was the case at Saint-Césaire. NNISP and diversity values for Saint- Césaire were calculated using NISP for long bone shaft fragments provided by Morin (2004:395-396). The same adjustment applied to the metapodials at Grotte XVI and Gatzarria was used in this case. Using the Shannon evenness index, Faith (2007) argued that the increasing diversity of reindeer parts at Grotte XVI from the Mousterian through to the Magdalenian was a result of increasing encounter rates with reindeer over this time period. Faith concluded that, as average search time decreased, due to increased encounter rates with reindeer, foragers became less selective in their transport decisions. This conclusion was fully consistent with the predictions of the CPF prey choice model (Cannon 2003). Using the reciprocal of Simpson's Index rather than the Shannon Index to measure the evenness of the Grotte XVI assemblages does not change the rank order of the diversity values for the assemblages with regards to Faith's (2007) evenness values (rs = 1.00). Exclusion of the cranium and mandible inverts the rank order of the Aurignacian (Abb) and Solutrean (As). However, this difference is minor, and these assemblages were only weakly differentiated in Faith's (2007) original analysis (evenness values of 0.960 for the Solutrean and 0.957 for the Aurignacian). The different calculation methods would therefore have no impact on Faith's conclusions. 186 Spearman's correlation between the diversity values and £NISP is high and significant (rs = 0.786,/? = 0.03) for the larger set of elements; however, this is driven by the Grotte XVI assemblages which increase in size and diversity through time (the Saint- Césaire assemblages are not included in this correlation). Faith (2007) has shown, however, that the pattern of diversity at Grotte XVI is not a result of sample size effects. When only long bones are considered, the correlation between diversity and sample size for all the assemblages is small and not significant (rs = 0.018, ? = 0.95). In comparison with the Grotte XVI values, the diversity value for the Gatzarria Mousterian suggests moderate selectivity in transport decisions, and when only long bones are considered, low selectivity. The Gatzarria Cj Mousterian long bone assemblage is also much more diverse than the Saint-Césaire Mousterian sample. These values are consistent with the patterns of long bone representation observed in Figures 7.2 and 7.3. It appears that Mousterian foragers at Gatzarria were much less selective in their long bone transport decisions than their counterparts at Grotte XVI and Saint-Césaire. Assuming that load sizes and the average number of carriers were similar at Gatzarria and Grotte XVI, these data could imply that transport distances and average search times for the major prey species at Gatzarria were moderately low,.at least relative to the Mousterian and Magdalenian extremes at Grotte XVI. However, there is a possible alternative interpretation of the Gatzarria diversity value. In his analysis of the Middle to Upper Palaeolithic transition assemblages at Saint- Césaire, Morin (2010) argues that long bone transport patterns at that site do not match the predictions of the CPF prey choice model. He suggests that this may reflect intensification of long bone transport when encounter rates with large ungulate prey were 187 low. This strategy would maximize the acquisition of marrow fat from available resources. Therefore, low selectivity in long bone transport at Gatzarria could also be a signature of intensified carcass exploitation under conditions of low encounter rates and longer average transport distances. This possibility is worthy of consideration, particularly when it is considered that despite the low selectivity in long bone transport, heavy red deer crania were rarely transported to Gatzarria. This fact is illustrated by the increase in the diversity of the Gatzarria assemblage relative to the Grotte XVI values when the skull is removed from consideration. However, without detailed climatic (i.e., microfaunal) data and diversity values from other occupations at Gatzarria, it is not currently possible to confirm which of these interpretations is correct. Processing of long bones for marrow during the Mousterian at Saint-Césaire was intensive, just as at Gatzarria. Unfortunately, no data on marrow-cracking is available for Grotte XVI. Morin (2010) suggests that mandibles were cracked to extract marrow at Saint-Césaire, but could not quantify this pattern because of frequent damage to the mandibular body. It was suggested in Chapter 6 that at least one mandible was cracked for marrow at Gatzarria, but the small sample size prevented a better understanding of this activity. First phalanges of red deer at Gatzarria appear to have been opened for their marrow, but again the sample size is very small. This is also the case for reindeer phalanges in the Mousterian sample from Saint-Césaire. With the available evidence, no case can be made for a difference in the intensity of marrow-cracking at these sites. Discussion: Reconsidering the Neandertal niche Both the general comparisons between Gatzarria and Middle Palaeolithic sites in the Basque country, and the more detailed comparisons with the Mousterian assemblages 188 from Saint-Césaire and Grotte XVI suggest important patterns in late Mousterian foraging behaviour in western France, including the western French and Spanish Pyrenees. First, at the majority of the sites considered, the entire range of available ungulate species was exploited, although the nature and importance of the exploitation of megaherbivores such as mammoth and rhinoceros remains poorly understood. Diet breadth at the sites considered here included a range of behaviourally diverse species, including mountain species such as chamois, small forest species such as roe deer, and more generalist species such as bison and red deer. Because of the influence of local microenvironments and climate on the distribution and abundance of many ungulate species, a wide diet breadth means that abundances of different ungulate species can be expected to vary substantially between sites even in the same region (e.g., between Gatzarria and Axlor). However, although foraging returns were sufficiently low to warrant at least occasional exploitation of low-ranked ungulates, there does not appear to have been any significant energetic and/or technological investment in the exploitation of smaller taxa at these sites. The apparent exclusion of lagomorphs, birds, aquatic resources, and, perhaps, carnivores from foraging efforts is consistent between the sites reviewed here, except possibly at Grotte XVI where many fish remains are present. This study has suggested that late Mousterian foraging patterns at Gatzarria were broadly similar to those at Saint-Césaire and elsewhere in western France. At Saint- Césaire, the focus on profitable ungulate resources resulted in a highly carnivorous diet (Bocherens et al. 2005). In this regard, it is not surprising that obtaining fat calories from ungulate prey was imperative for foragers at both Saint-Césaire and Gatzarria, at least as can be determined based on skeletal part abundances and patterns of marrow extraction at both sites. If the foragers occupying these sites were not willing to spend more time and energy in fat (or carbohydrate; Speth and Spielmann 1983) acquistion, this may have been a limiting factor on local population sizes. Indeed, alternative food sources (birds, leporids, fish, berries, tubers, bone grease) may have been quite costly to exploit. Foragers need to meet their basic energy requirements in order to survive and reproduce. For population densities to increase, additional energy is required; a fact which implies a relationship between forager population densities and the biomass of utilized food resources (Winterhaider et al. 1988). However, population densities are not linked to prey biomass in a linear fashion. Due to fluctations in population sizes resulting from both human prédation and density-independent factors such as climate variability, the "carrying capacities" of different animal resosurces are not static (Winterhaider et al. 1988; see also Stiner et al. 2000). Based on these dynamic relationships, Morin (2004:377, emphasis original) has argued that, among Palaeolithic hunter-gatherer groups who were highly dependent on ungulate fauna, "human population densities were most probably tracking climatically-inducedfluctuations in ungulate diversity and abundance relatively closely, expanding and declining in response to these changes.'''' This hypothesis has relevance for this study. Although foragers at Gatzarria primarily encountered red deer, the entire spectrum of local ungulate species was, at least on occasion, included in the diet breadth, and marrow processing of long bones was thorough. However, small game such as birds, fish, and leporids are absent from the assemblage and evidently did not make a substantial contribution to the diet at the site. This suggests that population densities may have been vulnerable to changes in the availability and reliability of the local ungulate resource base. There was probably little room for a permanent population expansion without a significant, threshold-type increase in foraging effort. Similarly, if the local availability of ungulates declined dramatically subsequent to this period, only an increase in foraging effort could even maintain human populations at the same level. Implicationsfor the early Upper Palaeolithic at Gatzarria The previous section established that the Mousterian foraging pattern observed at Gatzarria is consistent with a broader scheme of late Neandertal adaptation in the Basque Country and western France. This section considers the meaning of the Mousterian foraging pattern described above as an antecedent to the early Upper Palaeolithic—and the emergence of modern humans—in the Atlantic Pyrenees. O'Connell (2006) has suggested that modern humans arriving from Africa may have been able to replace Neandertals in Europe through competitive exclusion; specifically, by operating a more costly subsistence economy. Under this scenario, increased energy input into foraging by modern human would involve: (1) exploitation of more expensive resources not utilized by Neandertals, and/or (2) more thorough exploitation of the same resources, possibly with a more expensive toolkit. Changes in the age structure, body part representation, and processing intensity of prey assemblages could indicate more intensive use of the same resources. Morin (2010) has tested O'Connell's predictions using faunal data from Saint- Césaire. He demonstrates that there is little evidence for increased dietary breadth during the transition period at that site. Rather, as a result of deteriorating climate, diet breadth at Saint-Césaire, in terms of the number of species included in the diet, became narrower as fewer ungulate species were encountered. The fact that a probable decrease in ungulate encounter rates coincided with increased evidence for subsistence stress at Saint-Césaire, suggested not niche expansion, as stipulated by O'Connell, but rather, niche contraction for human foragers in western France during the transition period (Morin 2008, 2010). Consequently, Morin suggests that a scenario of genetic drift and bottlenecks, not replacement, is more likely for the transition in western France. Starting from the preceding statements, it is possible to suggest how the hypotheses of niche contraction induced by decreased prey availability (i.e., a population crash) versus niche expansion (i.e., operation of a more costly subsistence system to maintain higher population densities) might be tested using the Gatzarria sequence. First, necessary evidence for niche contraction would involve an apparent decrease in dietary breadth during the early Upper Palaeolithic, accompanied by an increase in cold-adapted micromammals. Cold-adapted species such as horse, and especially, reindeer would be expected to increase at the expense of temperate species including red deer and large bovids (Morin 2004, 2010). Roe deer would likely disappear from the local environment. The representation of chamois might persist at a similar level, or increase as this low- ranked species was relied on more heavily and/or their range shifted to lower altitudes as the climate worsened (Delpech 1990). Reliance on a narrower range of ungulate species, which may have highly fluctuating population sizes, could increase variability in foraging returns. Thus, continued evidence for subsistence stress, without the incorporation of novel, more expensive, technologies such as bone grease extraction, is a further condition of this scenario. 192 In contrast, niche expansion at Gatzarria would necessarily involve increased reliance on low-ranked species such as chamois, and an increase in dietary breadth involving smaller, more costly resources, in order to not only compensate for the decrease in ungulate availability, but to maintain a larger population. Marrow extraction would also be expected to increase, as it would under niche contraction (Morin 2010). An analysis of the entire Gatzarria sequence could refute either of these scenarios, although if micromammals at the site do not provide evidence for a climatic decline, the foraging implications of each scenario would have to be reconsidered. Summary This chapter has placed the results obtained in Chapter 6 within a broader context, both geographically and temporally. The pattern of Mousterian foraging at Gatzarria is consistent with a widespread pattern of intensive utilization of ungulate resources during the late Mousterian period in western France. This suggests that throughout this region, population densities may have been tied to the size and reliability of the ungulate resource base. This conclusion has implications for how foraging behaviours could change under scenarios of niche contraction or niche expansion during the early Upper Palaeolithic at Gatzarria. Chapter 8: Conclusions In this final chapter, the limitations of this work are briefly discussed along with suggestions for future research. The results of the study are then summarized in view of the research objectives outlined in Chapter 1, followed by a few concluding remarks. Limitations ofthepresent study and directionsforfurther research This research represents the first taphonomic and zooarchaeological examination of fauna from Gatzarria Cave. Faunal remains from four transition layers (the Châtelperronian, two Proto-Aurignacian and the Early Aurignacian levels), as well as from the Evolved Aurignacian, Gravettian, and Cjr Mousterian layers, remain to be studied in detail. Several suggestions for this future work, stemming from some of the challenges and limitations of the present study, can be outlined here. First, MNE and MNI values should be calculated as a means to better assess potential problems including the differential identifiability of elements and differential fragmentation of species identified in Chapter 5. MNE and MNI values could increase the robusticity of many of the analyses carried out in this study, by confirming the trends identified using NNISP. Taphonomic questions involving recovery methods, burning, and the absence of spongy bone in the assemblage could be more firmly answered with length measurements from a larger sample of unidentified specimens and with counts of unidentified spongy bone fragments. Hopefully, other layers at the site will provide larger NISP samples and thus permit more statistical analyses than were possible in the present 194 work, due to sample size limitations. Additional data may also help clarify whether the large bovid remains are indeed bison or aurochs. A close control of the local climate was not available in this study, although this did not seriously hinder interpretations as only one layer was under consideration. However, when detailed comparisons between layers are undertaken, the climatic data provided by microfaunal remains will be invaluable. Similarly, the possibility that avian resources and other small game were utilized during early Upper Palaeolithic occupations at Gatzarria cannot be ignored. Microfauna and bird remains, as well as fish remains (if at all present), must be taxonomically identified in all layers. The problems with the site stratigraphy will also need to be addressed more thoroughly in future research. The solution adopted in this study appears to be satisfactory for the Cj Mousterian, but the distribution and integrity of Cjn 1-3, the most important layers for the study of the transition, will be more difficult to establish. The early Upper Palaeolithic industries in the middle part of Ej appear to be overlapping in different parts of the cave, if not actually mixed (Figure 5.2). Detailed studies of the lithic débitage from the site, much of which remains unstudied and in the original excavation bags, might help delimit the boundaries (or lack thereof) of the archaeological industries in Ej, décapage by décapage. This would represent an enormous undertaking, but might be necessary for understanding the distribution of the archaeological industries in Ej over the entire excavated volume. As was the case in this study, large amounts of faunal material may ultimately have to be eliminated from future analyses due to uncertainty about its archaeological associations. Hopefully, the existing lithic material will provide a solution to the stratigraphie problems. However, a geoarchaeological and archaeostratigraphic study of the profiles remaining in situ after Laplace's excavations would be of considerable use, as profiles in the excavation notebooks generally do not identify the position of Cjn 1-3. An additional limitation is the lack of Chronometrie dating of the site. In this study, it has been assumed that the Cj material dates to the late Mousterian period. Based on the site stratigraphy, this is not an unreasonable assumption. The Cj Mousterian level is found in the same sedimentary deposit as the Châtelperronian and two Proto-Aurignacian levels. However, Chronometrie dating should be carried out to verify this assumption and to confirm the appropriate selection of sites for comparisons with the Cj Mousterian assemblage. Regarding methological limitations, the use of modern analogues is necessary for understanding the behaviour and ecology of ungulate species during the time period considered in this study. Some degree of uncertainty is expected in the use of such analogues; for instance, in estimates of age and seasonality. However, a general assumption of uniformity in species ecology is justified, based on palaeoecological reconstructions and anatomical similarities. This approach provides otherwise unattainable insight into human foraging behaviours in the past. Due to a lack of precise knowledge concerning ungulate community ecology in the Pleistocene, the resource patches defined in this analysis were necessarily broad. This may compromise the fine-grained search assumption of the diet breadth model. A greater emphasis on the CPF prey choice model (which ignores habitat patchiness; Cannon 2003) in future analyses could circumvent this issue. Nevertheless, when considered in view of the taxonomic abundance data, the patch divisions identified here highlight interesting patterns in resource choice and, possibly, land-use during the late Mousterian period in the Pyrenean foothills. Furthermore, the general conclusion of the diet breadth analysis— that maximum diet breadth in all patches included a broad range of ungulates, but little to no carnivores or small game—would remain robust regardless of the diet breadth model selected (i.e., the within-patch diet breadth model or the CPF prey choice model; although an appropriate choice of model would have to be made for assessing change in diet breadth through time at the site). The perspective afforded by foraging theory into the intensity of resource use has also provided interesting insight into patterning in the Gatzarria assemblage. For example, diet breadth at the site focused on ungulates, and red deer in particular, yet marrow-processing of this species was intensive. This pattern may support the argument that foragers who are heavily-dependent on animal resources may be fat-limited (Speth and Spielmann 1983). The slight differences in transport selectivity shown in the diversity analysis of Gatzarria, Saint-Césaire, and Grotte XVI may have relevance for understanding variability in foraging goals and in prey encounter rates during the late Mousterian period in southwestern France. However, the variables involved in transport decision-making and field-processing are not completely understood, and some significant factors may be ignored in current model formulations. One such incompletely- understood problem may be signaled by the different patterns of cranium versus mandible and long bone transport at Gatzarria, Saint-Césaire, and Grotte XVI. This discrepancy may relate to a broader issue involving what criteria are likely to form the basis of transport decisions under various constraints, as well as to potential problems of equifinality in determining whether changes in skeletal part patterns reflect intensification of carcass exploitation or different encounter rates (Morin 2010). This study has helped highlight some of these issues, although answers to these questions may require ethnographic investigation. Summary ofresults The results of this research can be summarized under four main headings, reflecting the objectives set out in Chapter 1 . (1) Reconsideration ofthe site stratigraphy: The archaeological levels in the Middle Ensemble at Gatzarria are weakly separated and appear to be mixed in some parts of the cave. The initially studied faunal material from Cj was associated with early Upper Palaeolithic as well as Mousterian artifacts. To resolve this problem, the distribution of diagnostic lithic pieces was plotted, and décapages associated with Mousterian artifacts were selected for the examination of Mousterian subsistence. The site stratigraphy requires further critical evaluation prior to undertaking additional faunal work. (2) Taphonomic analysis: The taphonomic analysis suggests a multifaceted history of deposition and recovery. Recovery of the small bone portion of the assemblage (fragments less than <1 cm) during excavations may not have been complete. However, there was no selective discard of larger long bone shaft fragments that would compromise zooarchaeological interpretations. Burning of spongy bone as fuel possibly altered the assemblage composition. Carnivores appear to have had secondary access to the bone deposits and were an important attritional agent at the site. Nevertheless, the impact of carnivores at Gatzarria appears to be relatively minor compared to other Mousterian sites in the Basque country. (3) Analysis offoraging behaviours at Gatzarria: Zooarchaeological analysis indicated that the Mousterian occupants of Gatzarria exploited a diversity of ungulate species, although red deer was by far the primary prey. The site may have been occupied in the fall and/or winter. Two distinct patches were defined: a lowlands patch, which was the primary focus of hunting activity, and a mountain patch, with fewer, lower-ranked species, that was less heavily utilized. Low-ranked ungulates were occasionally exploited, suggesting that they may have played a role as safety nets when more profitable prey species were less abundant. Marrow fat content appears to have been the main currency of foraging and played an important role in transport decisions. Marrow bearing-elements with high unsaturated-fat content in their marrow were the most frequently transported elements, although transport of ungulate long bones at Gatzarria was thorough in comparison with other late Mousterian sites in western France. Marrow processing at the site was intensive. (4) Implicationsfor Neandertal adaptations and the early Upper Palaeolithic in the Pyrenees: The foraging pattern documented at Gatzarria is consistent with those observed in late Mousterian assemblages elsewhere in the Basque Country and western France. The broad-spectrum use ofungulate resources, and high levels of carcass and marrow exploitation at Gatzarria suggest that Mousterian population densities may have been tracking ungulate abundances, as argued by Morin (2004, 2008, 2010). Consequently, for populations to expand in the early Upper Palaeolithic, diet breadth would have to increase to include more costly resources such as fish, birds, and lagomorphs. In contrast, if ungulate diversity decreased and diet breadth did not expand, populations would probably contract, and would be more vulnerable to fluctuations in prey availability (Morin 2008). Final remarks By offering some insight into the foraging behaviour of Gatzarria's Mousterian occupants, this research has provided a baseline for future work involving Neandertal subsistence in the Pyrenees. Additional research on the Gatzarria collections has the potential to provide further insight into Middle to Upper Palaeolithic transition processes in this region. The Neandertals at Gatzarria were engaged in complex ecological and demographic relationships with their animal resource base. The ultimate fate of the descendants of these foragers remains to be ascertained. Bibliography: Acevedo, P., and J. Cassinello 2009 Biology, Ecology and Status of Iberian Ibex Capra pyrenaica: A Critical Review and Research Prospectus. Mammal Review 39:17-32. Aldenderfer, M. 1998 Quantitative Methods in Archaeology: A Review of Recent Trends and Developments. Journal ofArchaeological Research 6:91-120. Alhaique, F., M. Bisconti, E. Castiglioni, C. Cilli, L. Fasani, G. Giacobini, R. Grifoni, A. Guerreschi, A. Iacopini, G. Malerba, C. Peretta, A. Reechi, A. Rocci Ris, A. Ronchiteli!, M. Rottoli, U. Thun Hohenstein, C. Tozzi, P. Visentini, and B. Wilkens 2004 Animal Resources and Subsistence Strategies. Collegium Antropologicum 28:23^10. Altuna, J. 1971 Los Mamíferos del Yacimiento Prehistórico de Moría In Cueva Morín: Excavaciones 1966-1968, edited by J. González Echegaray and L. G. Freeman, pp. 367-398. Patronato de las Cuevas Prehistóricas, Santander. 1990 La Caza de Herbívoros durante el Paleolítico y Mesolitico del País Vasco. Munibe (Antropologia-Arkeologia) 42:229-240. 2006 Faune de Mammifères du Gisement Préhistorique de la Grotte des Abeilles (Haute-Garonne). In Les Aurignaciens Pyrénéen des Abeilles et Mediterranean de Régismont-le-Haut. Analyses Typologique and Paléontologique, edited by G. Laplace, I. Barandiarán, A. Sáenz de Buruaga, and J. Altuna, pp. 142-153. Société des Amis du Musée National de Préhistoire et de la Recherche Archéologique, Les Eyzies-de-Tayac. Banks, W. E., F. d'Errico, A. T. Peterson, M. Kageyama, and G. Colombeau 2008 Reconstructing Ecological Niches and Geographic Distributions of Caribou (Rangifer tarandus) and Red Deer {Cervus elaphus) during the Last Glacial Maximum. Quaternary Science Reviews 27:2568-2575. Barnosky, A. D. 1985 Taphonomy and Herd Structure of the Extinct Irisk Elk, Megaloceros giganteus. Science 228:340-344. 201 1986 "Big Game" Extinction Caused by Late Pleistocene Climatic Change: Irish Elk {Megaloceros giganteus) in Ireland. Quaternary Research 25: 128-135. Bar-Yosef, O. 2004 Eat What is There: Hunting and Gathering in the World ofNeanderthals and their Neighbours. International journal ofOsteoarchaeology 14: 333-342. Barton, R. N. E. 2000 Mousterian Hearths and Shellfish: Late Neanderthal Activities on Gibraltar. In Neanderthals on the Edge: Papersfrom a Conference Marking the 150th Anniversary ofthe Forbes' Quarry Discovery, Gibraltar, edited by C. B. Stringer, R. N. E. Barton, and J. C. Finlayson, pp. 21 1-220. Oxbow Books, Oxford. Berger, J. 1983 Ecology and Catastrophic Mortality in Wild Horses: Implications for Interpreting Fossil Assemblages. Science 220: 1 403-1 404. Binford, L. R. 1978 Nunamiut Ethnoarchaeology. Academic Press, New York. 1981 Bones: Ancient Men and Modern Myths. Academic Press, New York. 1984 Faunal Remainsfrom Klasies River Mouth. Academic Press, Orlando. 1 985 Human Ancestors: Changing Views of their Behavior. Journal of Anthropological Archaeology 4:292—327. 1988 Étude Taphonomique des Restes Fauniques de la Grotte Vaufrey. Mémoires de la Société Préhistorique Française 19:535-564. Bird, D. W., and R. Bliege Bird 2000 The Ethnoarchaeology of Juvenile Foragers: Shellfishing Strategies among Meriam Children. Journal ofAnthropological Archaeology 19(4):461^7. Bird, D. W., R. Bliege Bird, and B. F. Codding 2009 In Pursuit of Mobile Prey: Martu Hunting Strategies and Archaeofaunal Interpretation. American Antiquity 74:3-29. Blasco, R. 2008 Human Consumption of Tortoises at Level IV of Bolomor Cave (Valencia, Spain). Journal ofArchaeological Science 35:2839-2848. 202 Blasco, R., and J. Fernándes Peris 2009 Middle Pleistocene Bird Consumption at Level XI of Bolomor Cave (Valencia, Spain). Journal ofArchaeological Science 36:2213-2223. Bliege Bird, R., and D. W. Bird 1997 Delayed Reciprocity and Tolerated Theft: The Behavioral Ecology of Food-Sharing Strategies. Current Anthropology 38:49-78. Blumenschine, R. J. 1988 An Experimental Model of the Timing of Hominid and Carnivore Influence on Archaeological Bone Assemblages. Journal of Archaeological Science 15:483-502. 1995 Percussion Marks, Tooth Marks, and Experimental Determinations of the Timing of Hominid and Carnivore Access to Long Bones at FLK Zinjanthropus, Olduvai Gorge, Tanzania. Journal ofHuman Evolution 29:21-51. Bocherens, H., D. G. Drucker, D. Billiou, M. Patou-Mathis, and B. Vandermeersch 2005 Isotopie Evidence for Diet and Subsistence Pattern of the Saint-Césaire I Neanderthal: Review and Use of a Multi-Source Mixing Model. Journal ofHuman Evolution 49:71-87. Bocherens, H., M. Fizet, A. Mariotti, B. Lange-Badre, B. Vandermeersch, J. P. Borei, and G. Bellon 1 99 1 Isotopie Biogeochemistry (1 3C, 1 5N) of Fossil Vertebrate Collagen: Application to the Study of a Past Food Web Including Neandertal Man. Journal ofHuman Evolution 20:481—492. Bordes, J. G. 2003 Lithic Taphonomy of the Châtelperronian/Aurignacian Interstratifications in Roc de Combe and Le Piage (Lot, France). In The Chronology ofthe Aurignacian and ofthe Transitional Technocomplexes: Dating, Stratigraphies, Cultural Implications, edited by J. Zilhäo and F. d'Errico, pp. 223-244. Proceedings of Symposium 6.1 of the 14th Congress of the UISPP. Oxbow Books, Oxford. Brain, C. K. 1981 The Hunters or the Hunted: An Introduction to African Cave Taphonomy. University of Chicago Press, Chicago. Brink, J. W. 1997 Fat Content in Leg Bones ofBison bison, and Applications to Archaeology. Journal ofArchaeological Science 24:259-21A. Brose, D. S., and M. H. Wolpoff 1971 Early Upper Paleolithic Man and Late Middle Paleolithic Tools. American Anthropologist 73 : 1 1 56-1 1 94. Broughton, J. M. 1994 Declines in Mammalian Foraging Efficiency during the Late Holocene, San Francisco Bay, California. Journal ofAnthropological Archaeology 13:371-401. 2002 Prey Spatial Structure and Behavior Affect Archaeological Tests of Optimal Foraging Models: Examples from the Emeryville Shellmound Vertebrate Fauna. WorldArchaeology 34:6-83. Brown, W. A. B., and N. G. Chapman 1991 The Dentition of Red Deer (Cervus elaphus): A Scoring Scheme to Assess Age from Wear of the Permanent Molariform Teeth. Journal ofZoology 224:519-536. Brugal, J. P. 1983 Applications des Analyses Multidimensionnelles à l'Étude du Squelette des Membres des Grands Bovidés Pleistocenes (Grotte de Lunel-Viel, Hérault): Perpectives Evolutives. Unpublished Ph.D. dissertation, Université Aix-Marseille I. Brumbach, H. J., and R. Jarvenpa 1997a Ethnoarchaeology of Subsistence Space and Gender: A Subarctic Dene Case. American Antiquity 62:414—436. 1997b Woman the Hunter: Ethnoarchaeological Lessons from Chipewyan Life Cycle Dynamics. In Women in Prehistory: North American and Mesoamerica, edited by C. C. Classen and R. A. Joyce, pp. 17-32. University of Pennsylvania Press, Philadelphia. Bugalho, M. N., J. A. Milne, and P. A. Racey 200 1 The Foraging Ecology of Red Deer (Cervus elaphus) in a Mediterranean Environment: Is a Larger Body Size Advantageous? Journal ofZoology 255:285-289. Bunn, H. T. 1983 Comparative Analysis of Modern Bone Assemblages from a San Hunter- Gatherer Camp in the Kalahari Desert, Botswana, and from a Spotted Hyena Den Near Nairobi, Kenya. In Animals andArchaeology I: Hunters and their Prey, edited by J. Clutton-Brock and C. Grigson, pp. 143-148. BAR International Series 163, Oxford. 204 Bunn, H. T., L. E. Bartram, and E. M. Kroll 1988 Variability in Bone Assemblage Formation from Hadza Hunting, Scavenging, and Carcass Processing. Journal ofAnthropological Archaeology 7:412—457. Bunn, H. T., and E. M. Kroll 1986 Systematic Butchery by Plio/Pleistocene Hominids at Olduvai Gorge, Tanzania. Current Anthropology 27:431-452. Burger, 0., M. J. Hamilton, and R. Walker 2005 The Prey as Patch Model: Optimal Handling of Resources with Diminishing Returns. Journal ofArchaeological Science 32:1 147-1 158. Burke, A. 2000 Hunting in the Middle Palaeolithic. International Journal of Osteoarchaeology 10:281-285. 2004 The Ecology of Neanderthals: Preface. International Journal of Osteoarchaeology 1 4(3^4) : 1 55- 1 6 1 . Calvet, M. 2004 The Quaternary Glaciation of the Pyrenees. In Quaternary Glaciations— Extent and Chronology. Part I: Europe, edited by J. Ehlers and P. L. Gibbard, pp. 1 19-128. Elsevier, Amsterdam. Cann, R., M. Stoneking, and A. C. Wilson 1987 Mitochondrial DNA and Human Evolution. Nature 325:3 1-36. Cannon, M. D. 2000 Large Mammal Relative Abundance in Pithouse and Pueblo Period Archaeofaunas from Southwestern New Mexico: Resource Depression Among the Mimbres-Mogollón? Journal ofAnthropological Archaeology 19:317-347. 2001 Archaeofaunal Relative Abundance, Sample Size and Statistical Methods. Journal ofArchaeological Science 28: 185-195. 2003 A Model of Central Place Forager Prey Choice and an Application to Faunal Remains from the Mimbres Valley, New Mexico. Journal of Anthropological Archaeology 22: 1-25. Capaldo, S. D., and R. J. Blumenschine 1994 A Quantitative Diagnosis ofNotches Made by Hammerstone Percussion and Carnivore Gnawing on Bovid Long Bones. American Antiquity 59:724-748. 205 Carrion, J. S., M. Munuera, C. Navarro, F. Burjachs, M. Dupré, and M. J. Walker 1999 The Palaeological Potential of Pollen Records in Caves: The Case of Mediterranean Spain. Quaternary Science Reviews 18:1061-1073. Castel, J.-C. 1999 Comportements de Subsistance au Solutréen et au Badegoulien d'Après les Faunes de Combe-Saunière (Dordogne) et du Cuzoul de Vers (Lot). Unpublished Ph.D. dissertation, Université de Bordeaux I. Chaline, J. 1972 Les Rongeurs du Pleistocène Moyen et Supérieur de la France. In Cahiers de Paléontologie. Éditions CNRS, Paris. Charnov, E. L. 1976 Optimal Foraging, the Marginal Value Theorem. Theoretical Population Biology 9:129-136. Chase, P. G. 1990 Tool-making Tools and Middle Paleolithic behavior. Current Anthropology 3 1 : 443^47. Chauvière, F.-X., and J.-C. Castel 2004 Le Statut du Renard à Combe Sauniere (Sarliac-sur-1'Isle, Dordogne) et dans le Solutréen de l'Est Aquitain. In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires, edited by J.-P. Brugal and J. Desse, pp. 389-402. Éditions APDCA, Antibes. Cleghorn, N., and C. W. Marean 2004 Distinguishing Selective Transport and In Situ Attrition: A Critical Review of Analytical Approaches. Journal ofTaphonomy 2:43-67. Clutton-Brock, T. H., and S. D. Albon 1989 RedDeer in the Highlands. Blackwell, Oxford. Clutton-Brock, T. H., and F. E. Guinness 1975 Behaviour of Red Deer (Cervus elaphus L.) at Calving Time. Behaviour 55:287-300. Clutton-Brock, T. H., F. E. Guinness, and S. D. Albon 1 982 Red Deer: Behavior and Ecology ofTwo Sexes. University of Chicago Press, Chicago. Cochard, D. 2004 Les Léporidés Dans la Subsistance Paléolithique du Sud de la France. Unpublished Ph.D. dissertation, Université de Bordeaux I. Cochard, D., and J. P. Brugal 2004 Importance des Fonctions de Sites dans les Accumulations Paléolithiques de Léporidés. In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires, edited by J.-P. Brugal and J. Desse, pp. 283-296. Éditions APDCA, Antibes. Codding, B. F., and T. L. Jones 2007 Man the Showoff? Or the Ascendance of a Just-So-Story: A Comment on the Recent Applications of Costly Signaling Theory in American Archaeology. American Antiquity 72:346-357. Costamagno, S. 1 999 Stratégies de Chasse et Fonction des Sites au Magdalénien Dans le Sud de la France. Unpublished Ph.D. dissertation, Université de Bordeaux I. Costamagno, S., C. Griggo, and V. Mourre 1999 Approche Expérimentale d'un Problème Taphonomique: Utilisation de Combustible Osseux au Paléolithique. Préhistoire Européenne 13:167- 194. Costamagno, S., and V. Laroulandie 2004 L'Exploitation des Petits Vertébrés dans les Pyrénées Françaises du Paléolithique au Mésolithique: Un Inventaire Taphonomique et Archéozoologique. In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires, edited by J.-P. Brugal and J. Desse, pp. 403-416. Éditions APDCA, Antibes. Crader, D. C. 1983 Recent Single-Carcass Bone Scattters and the Problem of "Butchery" Sites in the Archaeological Record. In Animals and Archaeology, Vol. 1: Hunters and Their Prey, edited by J. Clutton-Brock and C. Grigson, pp. 107-141 . BAR International Series 163, Oxford. Crestanello, B., E. Pecchioli, C. Vernesi, S. Mona, N. Martínková, M. Janiga, H. C. Hauffe, and G. Bertorelle 2009 The Genetic Impact of Translocations and Habitat Fragmentation in Chamois {Rupicapra) spp. Journal ofHeredity 100:691-708. D'Errico, F., and M. F. Sánchez Goni 2003 Neandertal extinction and the millennial scale climatic variability of OIS 3. Quaternary Science Review 22:769-788. Dari, A., and J. Renault-Miskovsky 2001 Études Paléoenvironnementales dans la Grotte "El Castillo" (Puente Viesgo, Cantabile, Espagne). Espacio, Tiempo y Forma, Sériel, Prehistoria y Arqueología 14:121—144. 207 Darling, F. F. 1963 A Herd ofRedDeer: A Study in Animal Behaviour. Oxford University Press, Oxford. Dart, R. A. 1957 The Osteodontokeratic Culture ofAustralopithecus prometheus. Transvaal Museum Memoir 10. Transvaal Museum, Pretoria. Delmas, M. 2005 La Déglaciation dans Ie Massif du Carlit (Pyrénées Orientales): Approches Géomorphologique et Géochronologique Nouvelles. Quaternaire 16:45- 55. Delpech, F. 1990 The Range Distribution of Some Ungulate Mammals during the Dryas I (Recent Wurm) in Aquitaine: Paleoecological Implications. Geobios 23:221-230. Dennell, R. 1983 European Economic Prehistory: A New Approach. Academic Press, London. Domínguez-Rodrigo, M. 1997 Meat-eating by Early Hominids at the FLK 22 Zinjanthropus Site, Olduvai Gorge (Tanzania): An Experimental Approach Using Cut-mark, Data. Journal ofHuman Evolution 33:669-690. Duarte, C, J. Mauricio, P. B. Pettitt, P. Souto, E. Trinkaus, H. van der Plicht, and J. Zilhäo 1999 The Early Upper Paleolithic Human Skeleton from the Abrigo do Lagar Velho (Portugal) and Modern Human Emergence in Iberia. Proceedings of the National Academy ofSciences 96:7604-7609. Edwards, J. K., R. L. Marchinton, and G. F. Smith 1 982 Pelvic Girdle Criteria for Sex Determination of White-Tailed Deer. The Journal of Wildlife Management 46:544-547. Emerson, A. E. 1 990 Archaeological Implications of Variability in the Economic Anatomy of Bison bison. Unpublished Ph.D. dissertation, Washington State University. Faith, J. T. 2007 Changes in Reindeer Body Part Representation at Grotte XVI, Dordogne, France. Journal ofArchaeological Science 34:2003-20 1 1 . 208 Faith, J. T., and A. D. Gordon 2007 Skeletal Element Abundance in Archaeofaunal Assemblages: Economic Utility, Sample Size, and Assessment of Carcass Transport Strategies. Journal ofArchaeological Science 34:872—882. Falguères, C, Y. Yokoyama, and Á. Arrizabalaga 2005 La Geocronologia des Yacimiento Pleistocènico de Lezetxiki (Arrásate, País Vasco). Critica de las Dataciones Existentes y Algunas Nuevas Aportaciones. Munibe 57:93-106. Fandos, P. 1989 Reproductive Strategies in Female Spanish Ibex (Capra pyrenaica). Journal ofZoology 218:339-343. Fernández-Jalvo, Y., and P. Andrews 2000 The Taphonomy of Pleistocene Caves, with Particular Reference to Gibraltar. In Neanderthals on the Edge: Papersfrom a Conference Marking the 150th Anniversary ofthe Forbes' Quarry Discovery, Gibraltar, edited by C. B. Stringer, R. N. E. Barton and J. C. Finlayson, pp. 171-182. Oxbow Books, Oxford. Fiore, I., M. Gala, and A. Tagliacozzo 2004 Ecology and Subsistence Strategies in the Eastern Italian Alps during the Middle Palaeolithic. International Journal ofOsteoarchaeology 14:273- 286. Frison, G. C, and C. A. Reher 1970 Appendix I: Age Determination of Buffalo by Tooth Eruption and Wear. In The Glenrock Buffalo Jump, 48C0304: Late Prehistoric Period Buffalo Procurement and Butchering in the Northwest Plains, by G. C. Frison, pp. 46-50. Plains Anthropologist Memoir 7. Fuentes Vidarte, C. 1980 Estudio de la Fauna de "El Pendo". In El Yacimiento de la Cueva de "El Pendo", edited by J. González Echegaray, pp. 215-237. Bibliotheca Praehistorica Hispana, 17., Madrid. García García, N., R. S. Feranec, J. L. Arsuaga, J. M. Bermudez de Castro, and E. Carbonell 2009 Isotopie Analysis of the Ecology of Herbivores and Carnivores from the Middle Pleistocene Deposits of the Sierra de Atapuerca, Northern Spain. Journal ofArchaeological Science 36: 1 142-1 151. 209 Garcia-Gonzalez, R., and P. Cuartas 1996 Trophic Utilization of a Montane/Subalpine Forest by Chamois (Rupicapra pyrenaica) in the Central Pyrenees. Forest Ecology and Management 88:1 5-23 . Gaudzinski, S. 2004 A Matter of High Resolution? The Eemian Interglacial (OIS 5e) in North- central Europe and Middle Palaeolithic Subsistence. International Journal of Osteoarchaeology 14:201-211. 2006 Monospecific or Species-dominated Faunal Assemblages during the Middle Paleolithic in Europe. In Transitions Before the Transition: Evolution and Stability in the Middle Paleolithic and Middle Stone Age, edited by S. L. Kuhn and E. Hovers, pp. 137-147. Springer. Gaudzinski, S., and L. Niven 2009 Hominin Subsistence Patterns During the Middle and Late Paleolithic in Northwestern Europe. In The Evolution ofHominin Diets: Integrating Approches to the Study ofPalaeolithic Subsistence, edited by J.-J. Hublin and M. P. Richards, pp. 99-1 1 1 . Springer. Gaudzinski, S., E. Turner, A. P. Anzidei, E. Alvarez-Fernandez, J. Arroyo-Cabrales, J. Cinq-Mars, V. T. Dobosi, A. Hannus, E. Johnson, S. C. Miinzel, A. Scheer, and P. Villa 2005 The Use of Proboscidean Remains in Every-day Palaeolithic Life. Quaternary International 126-128: 179-194. Geist, V. 1998 Deer ofthe World. Their Evolution, Behaviour, and Ecology. Stackpole Books, Mechanicsburg, Pennsylvania. González Urquijo, J., J. J. Ibáñez Estévez, and J. Ríos Garaizar 2004 Excavación Arqueológica en el Yacimiento de Axlor (Dima, Bizkaia). Electronic document, http://wvvw.harpea.org/Proyectos/Axlor.html, accessed online April 7,2010 González Urquijo, J., J. J. Ibáñez, J. Ríos Garaizar, L. Bourguignon, P. Castaños, and A. Tarriño 2005 Excavaciones Recientes en Axlor. Movildad y Planificación de Actividades en Grupos de Neandertales. Monografías del Museo de Altamira 20:527-539. González-Sampériz, P., B. L. Valero-Garcés, A. Moreno, G. Jalut, J. M. García-Ruiz, C. Martí-Bono, A. Delgado-Huertas, A. Navas, T. Otto, and J. J. Dedoubat 2006 Climate Variability in the Spanish Pyrenees during the Last 30,000 Yr Revealed by the El Portalet Sequence. Quaternary Research 66:38-52. Gould, S. J., and R. C. Lewontin 1979 The Spandrels of San Marco and the Panglossion Paradigm: A Critique of the Adaptationist Programme. Proceedings ofthe Royal Society of London, B. 205:581-598. Granger, J.-M., and F. Leveque 1997 Parure Castelperronienne et Aurignacienne: Étude de Trois Séries Inédites de Dents Percées et Comparaisons. Comptes-Rendus de l'Académie de Sciences 325:537—543. Grayson, D. K. 1984 Quantitative Zooarchaeology. Topics in the Analysis ofArchaeological Faunas. Academic Press, New York. Grayson, D. K., and M. D. Cannon 1999 Human Paleoecology and Foraging Theory in the Great Basin. In Models for the Millennium: Great Basin Anthropology Today, edited by C. Beck, pp. 141-151. University of Utah Press, Salt Lake City. Grayson, D. K., and F. Delpech 1994 The Evidence for Middle Palaeolithic Scavenging from Couche VIII, Grotte Vaufrey (Dordogne, France). Journal ofArchaeological Science 21:359-375. 1998 Changing Diet Breadth in the Early Upper Palaeolithic of Southwestern France. Journal ofArchaeological Science 25:1119-1129. 2002 Specialized Early Upper Palaeolithic Hunters in Southwestern France? Journal ofArchaeological Science 29: 1439-1449. 2005 Pleistocene reindeer and global warming. Conservation Biology 19(2):557-562. 2003 Ungulates and the Middle-to-Upper Paleolithic Transition at Grotte XVI (Dordogne, France). Journal ofArchaeological Science 30: 1633-1648. 2008 The Large Mammals of Roc de Combe (Lot, France): The Châtelperronian and Aurignacian assemblages. Journal ofAnthropological Archaeology 27:338-362. Grayson, D. K., F. Delpech, J. P. Rigaud, and J. Simek 2001 Explaining the Development of Dietary Dominance by a Single Ungulate Taxon at Grotte XVI, Dordogne, France. Journal ofArchaeological Science 28:1 15-125. 211 Grayson, D. K., and C. J. Frey 2004 Measuring Skeletal Part Representation in Archaeological Faunas. Journal ofTaphonomy 2:27-42. Green, R. E., J. Krause, A. W. Briggs, T. Maricic, U. Stenzel, M. Kircher, N. Patterson, H. Li, W. Zhai, M. H.-Y. Fritz, N. F. Hansen, E. Y. Durand, A.-S. Malaspinas, J. D. Jensen, T. Marques-Bonet, C. Alkan, K. Prüfer, M. Meyer, H. A. Burbano, J. M. Good, R. Schultz, A. Aximu-Petri, A. Butthof, B. Höber, B. Höffner, M. Siegemund, A. Weihmann, C. Nusbaum, E. S. Lander, C. Russ, N. Novod, J. Affourtit, M. Egholm, C. Verna, P. Rudan, D. Brajkovic, Z. Kucan, I. Gusic, V. B. Doronichev, L. V. Golovanova, C. Lalueza-Fox, M. de la Rasilla, J. Fortea, A. Rosas, R. W. Schmitz, P. L. F. Johnson, E. E. Eichler, D. Falush, E. Birney, J. C. Mullikin, M. Slatkin, R. Nielsen, J. Kelso, M. Lachmann, D. Reich, and S. Pääbo 2010 A Draft Sequence of the Neandertal Genome. Science 328(5979) :7 10-722. Guadelli, J. L. 1996 Les cerfs du Wurm ancien en Aquitaine. Palèo 8:99-108. Guilday, J. E. 1984 Pleistocene Extinction and Environmental Change: A Case Study of the Appalachians. In Quaternary Extinctions: A Prehistoric Revolution, edited by P. S. Martin and R. G. Klein, pp. 250-258. University of Arizona Press, Tucson. Gurven, M., and K. Hill 2009 Why do Men Hunt? A Réévaluation of "Man the Hunter" and the Sexual Division of Labor Current Anthropology 50(l):51-74. Hardy, B. L. 2010 Climatic Variability and Plant Food Distribution in Pleistocene Europe: Implications for Neanderthal Diet and Subsistence. Quaternary Science Reviews 29:662-679. Haws, J. A. 2003 An Investigation ofLate Upper Paleolithic and Epipaleolithic Hunter- Gatherer Subsistence and Settlement Patterns in Central Portugal. Unpublished Ph.D. dissertation, University of Wisconsin-Madison. Haynes, G. 1991 Mammoths, Mastodonts, and Elephants: Biology, Behavior, and the Fossil Record. Cambridge University Press, Cambridge. 2002 The Early Settlement ofNorth America: The Clovis Era. Cambridge University Press, Cambridge. 2006 Mammoth Landscapes: Good Country for Hunter-Gatherers. Quaternary 212 International 142-143:20-29. Hérail, G., J. Hubschman, and G. Jalut 1986 Quaternary Glaciation in the French Pyrenees. Quaternary Science Reviews 5:397-402. Herrero, J., I. Garin, A. García-Serrano, and R. García-González 1996 Habitat Use in a Rupicapra pyrenaica pyrenaica forest population. Forest Ecology and Management 88:25-29. Hildebrandt, W. R., and K. R. McGuire 2002 The Ascendance of Hunting During the California Middle Archaic: An Evolutionary Perspective. American Antiquity 67:231-256. Hill, K. 1988 Macronutrient Modifications of Optimal Foraging Theory: An Approach Using Indifference Curves Applied to Some Modern Foragers. Human Ecology 16:157-197. Hill, K., and K. W. Kintigh 2009 Can Anthropologists Distinguish Good and Poor Hunters? Implications for Hunting Hypotheses, Sharing Conventions, and Cultural Transmission. Current Anthropology 50(3):369-377. Hockett, B., and J. A. Haws 2002 Taphonomic and Methodological Perspectives of Leporid Hunting during the Upper Paleolithic of the Western Mediterranean Basin. Journal of Archaeological Method and Theory 9:269-302. Hoffecker, J. F. 2009 Neanderthal and Modern Human Diet in Eastern Europe. In The Evolution ofHominin Diet: Integrating Approaches to the Study ofPalaeolithic Subsistence, edited by J.-J. Hublin and M. P. Richards, pp. 87-98. Springer. Hoffecker, J. F., and N. Cleghorn 2000 Mousterian Hunting Patterns in the Northwestern Caucasus and the Ecology of the Neanderthals. International journal ofOsteoarchaeology 10:368-378. Hofman, J. L. 1992 Putting Together the Pieces: An Introduction to Refitting. In Piecing Together the Past: Applications ofRefitting Studies in Archaeology, edited by J. L. Hofman and J. G. Enloe, pp. 1-20. BAR International Series 578, Oxford. Howell, F. C. 1965 Early Man. Time, New York. Isenberg, A. C. 2000 The Destruction ofthe Bison: An Environmental History, 1 750-1920. Cambridge University Press, Cambridge. Jalut, G., M. Monserrat, M. Fontugne, G. Delibrias, J. M. Vilaplana, and R. Julia 1992 Glacial to Interglacial Vegetation Changes in the Northern and Southern Pyrénées: Déglaciation, Vegetation Cover and Chronology. Quaternary Science Reviews 1 1 :449-480. Jarvenpa, R. 1980 The Trappers ofPatuanak: Toward a Spatial Ecology ofModern Hunters. Canadian Ethnology Service Paper 67. National Museum of Man Mercury Series, Ottawa. Jaubert, J., and T. Bismuth 1996 Le Paléolithique Moyen des Pyrénées Centrales: Esquisse d'un Schéma Chronologique et Économique dans la Perspective d'une Étude Comparative avec les Documents Ibériques. In Pyrénées Préhistoriques Arts et Sociétés. Actes du 118e Congrès National des Sociétés Historiques et Scientifiques, edited by H. Delporte and J. Clottes, pp. 9-26. C.T.H.S., Paris. Jenkins, S. H., and M. C. Ashley 2003 Wild Horse. In Wild Mammals ofNorth America: Biology, Management, and Conservation, edited by G. A. Feldhamer, B. C. Thompson, and J. A. Chapman, pp. 1 147-1 163. 2nd edition. John Hopkins University Press, Baltimore. Johnson, L. M. 2008 Plants and Habitats—A Consideration of Dene Ethnoecology in Northwestern Canada. Botany 86:146-156. Jones, E. L. 2004 The European Rabbit (Oryctolagus cuniculus) and the Development of Broad-Spectrum Diets in South-western France: Data from the Dordogne Valley. In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire, aux Ressources Utilitaires, edited by J.-P. Brugal and J. Desse, pp. 223-234. Éditions APDCA, Antibes. 2006 Prey Choice, Mass Collecting, and the Wild European Rabbit {Oryctolagus cuniculus). Journal ofAnthropological Archaeology 25:275-289. 214 Jones, K. T., and D. Metcalfe 1988 Bare Bones Archaeology: Bone Marrow Indices and Efficiency. Journal ofArchaeological Science 15:41 5-423 . Joseph, S. 2000 Anthropological Evolutionary Ecology: A Critique. Journal ofEcological Anthropology 4:6-30. 2002 Reply to Winterhalder's Behavioral and Other Human Ecologies: Critiques, Response and Progress through Criticism. Journal ofEcological Anthropology 6:24—37. Kaplan, H., and K. Hill 1 992 The Evolutionary Ecology of Food Acquisition, hi Evolutionary Ecology and Human Behaviour, edited by E. A. Smith and B. Winterhaider, pp. 1 67-20 1 . Aldine de Gruyter, New York. Karkanas, P., J.-P. Rigaud, J. F. Simek, R. M. Albert, and S. Weiner 2002 Ash Bones and Guano: A Study of the Minerals and Phytoliths in the Sediments of Grotte XVI, Dordogne, France. Journal ofArchaeological Science 29:721-732. Klein, R. G., and K. Cruz-Uribe 1 984 The Analysis ofAnimal Bonesfrom Archaeological Sites. University of Chicago Press, Chicago. Klein, R. G., and T. E. Steele 2008 Gibraltar Data are Too Sparse to Inform on Neanderthal Exploitation of Coastal Resources. Proceedings ofthe National Academy ofSciences 105:E115. Lam, Y. M., X. Chen, and O. M. Pearson 1999 Intertaxonomic Variability in Patterns of Bone Density and the Differential Representation of Bovid, Cervid, and Equid Elements in the Archaeological Record. American Antiquity 64:343-362. Laplace, G. 1966a Les Niveaux Castelperroniens, Protoaurignaciens et Aurignaciens de la Grotte Gatzarria à Suhare en Pays Basque. Quartär 17: 1 17-140. 1966b Rapport de Fouilles 1966. Grotte Gatzarria, Commune d'Ossas-Suhare (Basses Pyrénées). Unpublished excavation report. Laplace, G., I. Barandiarán, A. Sáenz de Buruaga, and J. Altuna 2006 Les Aurignaciens Pyrénéens des Abeilles et Méditerranéens de Régismont- le-Haut. Analyses Typologique et Paléontologique. Société des Amis du 215 Musée National de Préhistoire et de la Recherche Archéologique, Les Eyzies-de-Tayac. Laplace, G., and A. Saenz de Buruaga 2003 Typologie Analytique et Structurale des Complexes du Moustérien de la Grotte Gatzarria (Ossas-Suhare, Pays Basque) et de leurs Relations avec ceux de l'Abri Olha 2 (Cambo, Pays Basque). Pyrenae 33-34:81-163. Laroulandie, V. 2004 Exploitation des Ressources Aviaires durant le Paléolitique en France: Bilan Critique et Perspectives. In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires, edited by J.-P. Brugal and J. Desse, pp. 163-171. Éditions APDCA, Antibes. Lavaud, F. 1980 Les Faunes Paléolithiques du Wurm II et III dans le Sud-ouest et le Centre-ouest de la France. Unpublished Ph.D. dissertation, Université de Poitiers. Lee, R. B., and I. DeVore 1968 Man the Hunter. Aldine, New York. Lee-Thorp, J. 2008 On Isotopes and Old Bones. Archaeometry 50:925-950. Lee-Thorp, J., and M. Sponheimer 2006 Contribution of Biogeochemistry to Understanding Hominin Dietary Ecology. Yearbook ofPhysical Anthropology 49:131-148. LeGaIl5O. 2000 Les Moustériens Étaient-Ils Pêcheurs? Bulletin de la Société d'Anthropologie du Sud-Ouest 34:3-1 1. Lev, E., M. E. Kislev, and O. Bar-Yosef 2005 Mousterian Vegetal Food in Kebara Cave, Mt. Carmel. Journal of Archaeological Science 32:475-484. Leveque, F. 1966 La Grotte Gatzarria de Suhare, Basses Pyrénées. Étude Sédimentologique et Archéologique. Unpublished D.E.S. thesis, Université de Poitiers. Leveque, F., and J.-C. Miskovsky 1996 Le Passage du Paléolithique Moyen au Paléolithique Supérieur dans la Grotte Gatzarria de Suhare: Données Sédimentologiques et Comparaisons. In Pyrénées Préhistoriques Arts et Sociétés. Actes du 118e Congrès National des Sociétés Historiques et Scientifiques, edited by H. Delporte and J. Clottes, pp. 41-60. C.T.H.S., Paris. Levine, M. A. 1 979 Archaeozoological Analysis ofSome Upper Pleistocene Horse Bone Assemblages in Western Europe. Unpublished Ph.D. dissertation, Cambridge University. Lovari, S., F. Sacconi, and G. Trivellini 2006 Do Alternative Strategies of Space Use Occur in Male Alpine Chamois? Ethology Ecology & Evolution 1 8:221-23 1 . Lowe, V. P. W. 1967 Teeth as Indicators of Age with Special Reference to Red Deer (Cervus elaphus) of Known Age from Rhum. Journal ofZoology 152: 137-153. Lucas, Z., J. I. Raeside, and K. J. Betteridge 1 99 1 Non-invasive Assessment of the Incidences of Pregnancy and Pregnancy Loss in the Feral Horses of Sable Island. Journal ofReproduction and Fertility, Supplement 44:479-488. Lupo, K. D. 2007 Evolutionary Foraging models in Zooarchaeological Analysis: Recent Applications and Future Challenges. Journal ofArchaeological Research 15:143-189. Lupo, K. D., and D. N. Schmitt 2002 Upper Paleolithic Net-hunting, Small Prey Exploitation, and Women's Work Effort: A View from the Ethnographic and Ethnoarchaeological Record of the Congo Basin. Journal ofArchaeological Method and TWy 9(2): 147-1 78. Lyman, R. L. 1994a Quantitative Units and Terminology in Zooarchaeology. American Antiquity 59:36-7 1 . 1994b Vertebrate Taphonomy. Cambridge University Press, Cambridge. 2003 The Influence of Time Averaging and Space Averaging on the Application of Foraging Theory in Zooarchaeology. Journal ofArchaeological Science 30:595-610. 2005 Analyzing Cut Marks: Lessons from Artiodactyl Remains in the Northwestern United States. Journal ofArchaeological Science 32: 1722- 1732. 2008 Quantitative Paleozoology. Cambridge University Press, Cambridge. 217 Lyman, R. L., and M. J. O'Brien 1987 Plow-zone Zooarchaeology: Fragmentation and Identifiability. Journal of FieldArchaeology 14:493^198. MacArthur, R. H., and E. R. Pianka 1 966 On Optimal Use of a Patchy Environment. The American Naturalist 100:603-609. MacDonald, D., and P. Barrett 2001 Mammals ofEurope. Princeton University Press, Princeton. Magurran, A. E. 2004 Measuring Biological Diversity. Blackwell Publishing, Oxford. Manne, T. H., M. C. Stiner, and N. F. Bicho 2005 Evidence for Bone Grease Rendering during the Upper Paleolithic at Vale Boi (Algarve, Portugal). In Animais na Pré-História e Arqueolgia da Península Ibérica. Actas do I Congresso de Arqueología Peninsular. Promontorio monográfica, edited by N. F. Bicho, pp. 145-158. Universidade do Algarve, Faro. Marean, C. W. 1 99 1 Measuring the Post-depositional Destruction of Bone in Archaeological Assemblages. Journal ofArchaeological Science 18:677-694. 1 995 Of Taphonomy and Zooarchaeology. Evolutionary Anthropology 4:64—72. Marean, C. W., and Kim, S. Y. 1998 Mousterian Large-Mammal Remains from Kobeh Cave: Behavioral implications for Neanderthals and Early Modern Humans. Current Anthropology 39:S79-S1 13. Marean, C. W., and L. M. Spencer 1 99 1 Impact of Carnivore Ravaging on Zooarchaeological Measures of Element Abundance. American Antiquity 56:645-658. Mariezkurrena, K. 1983 Contribución al Conomiento del Desarrollo de la Dentición y el Esqueleto Postcraneal de Cervus elaphus. Munibe 35: 149-202. Marín Arroyo, A. B., P. Fosse, and J.-D. Vigne 2009 Probable Evidences of Bone Accumulation by Pleistocene Bearded Vulture at the Archaeological Site of El Mirón Cave (Spain). Journal of Archaeological Science 36:284—296. 218 Marshall, F., and T. Pilgram 1993 NISP vs. MNI in Quantification of Body-Part Representation. American Antiquity 58:261-269. Martin, J. F. 1983 Review: Optimal Foraging Thory: A Review of Some Models and Their Applications. American Anthropologist 85:612-629. 1985 More on Optimal Foraging Theory. American Anthropologist 87:649-650. Mayne Correia, P. M. 1997 Fire Modification of Bone: A Review of the Literature. In Forensic Taphonomy: The Postmortem Fate ofHuman Remains, edited by W. D. Hagluund and M. H. Sorg, pp. 275-293. CRC Press, Boca Raton. McGuire, K. R., W. R. Hildebrandt, and K. L. Carpenter 2007 Costly Signaling and the Ascendance ofNo-Can-Do Archaeology: A Reply to Codding and Jones. American Antiquity 72:358-365. Mellars, P. 1973 The Character of the Middle-Upper Palaeolithic Transition in South-west France. In The Explanation ofCulture Change, edited by C. Renfrew, pp. 255-276. Duckworth, London. 1 996 The Neanderthal Legacy: An Archaeological Perspective From Western Europe. Princeton University Press, Princeton. 1998 The Fate of the Neandertals. Nature 395:539-540. 2000 The Archaeological Records of the Neandertal-Modern Human Transition in France. In The Geography ofNeandertals and Modern Humans in Europe and the Greater Mediterranean, edited by O. Bar-Yosef and D. Pilbeam, pp. 35^47. Peabody Museum of Archaeology and Ethnology, Cambridge, Massachusetts. 2005 The Impossible Coincidence. A Single-species Model for the Origins of Modern Human Behaviour in Europe. Evolutionary Anthropology 14:12— 27. 2006 Archeology and the Dispersal of Modern Humans in Europe: Deconstructing the "Aurignacian." Evolutionary Anthropology 15:167— 182. Mercier, N., H. Valladas, J. L. Joron, and J. L. Reyss 1993 Thermoluminescence Dating of the Prehistoric Site of La Roche à Pierrot, Saint-Césaire. In Context ofa Late Neandertal: Implications of 219 Multidisciplinary Researchfor the Transition to Upper Paleolithic Adaptations at Saint-Césaire, edited by F. Leveque, A. M. Backer, and M. Guilbaud, pp. 15-21. Monographs in World Archaeology 16. Prehistory Press, Madison. Metcalfe, D., and K. R. Barlow 1992 A Model for Exploring the Optimal Trade-offbetween Field Processing and Transport. American Anthropologist 94:340-356. Metcalfe, D., and K. T. Jones 1988 A Reconsideration of Animal Body-part Utility Indices. American Antiquity 53:486-504. Miller, F. L. 1 974 Biology ofthe Kaminuriak Population ofBarren-Ground Caribou, Part 2. Volume 3 1 . Canadian Wildlife Service Report Series, Ottawa. 2003 Caribou. In Wild Mammals ofNorth America: Biology, Management, and Conservation, edited by G. A. Feldhamer, B. C. Thompson, and J. A. Chapman, pp. 965-997. 2nd edition. John Hopkins University Press, Baltimore. Miracle, P., and D. Sturdy 1 99 1 Chamois and the Karst of Herzegovinia. Journal ofArchaeological Science 18:89-108. Morin, E. 2004 Late Pleistocene Population Interaction in Western Europe and Modern Human Origins: New Insights Based on the Faunal Remainsfrom Saint- Césaire, Southwestern France. Unpublished Ph.D. dissertation, University of Michigan. 2007 Fat Composition and Nunamiut Decision-Making: A New Look at the Marrow and Bone Grease Indices. Journal ofArchaeological Science 34:69-82. 2008 Evidence for Declines in Human Population Densities during the Early Upper Paleolithic in Western Europe. Proceedings ofthe National Academy ofSciences 105:48-53. 2009 Taphonomic Implications of the Use of Bone as Fuel. Palethnologie, in press. 2010 Paleolithic Foragers at Saint-Césaire, France. A Faunal Perspective on the Origins ofModern Humans. Oxbow Books, Oxford, in press. ??p?, E., T. Tsanova, N. Sirakov, W. Rendu, J.-B. Mallye, and F. Leveque 2005 Bone Refits in Stratified Deposits: Testing the Chronological Grain at Saint-Césaire. Journal ofArchaeological Science 32: 1083-1098. Morían, R. E. 1983 Spiral Fractures on Limb Bones: Which Ones are Artificial. In Carnivores, Human Scavengers, and Predators: A Question ofBone Technology. Proceedings of the Fifteenth Annual Conference of the Archaeological Association of the University of Calgary, edited by G. M. LeMoine and A. S. MacEachern, pp. 241-269. University of Calgary Archaeological Association, Calgary. Munro, N. D. 2004 Zooarchaeological Measures of Hunting Pressure and Occupation Intensity in the Natufian: Implications for Agricultural Origins. Current Anthropology 45 :S5-S33. Munson, P. J. 2000 Age-correlated Differential Destruction of Bones and its Effect on Archaeological Mortality Profiles of Domestic Sheep and Goats. Journal ofArchaeological Science 27:391—407. Munson, P. J., and R. C. Garniewicz 2003 Age-mediated Survivorship of Ungulate Mandibles and Teeth in Canid- ravaged Faunal Assemblages. Journal ofArchaeological Science 30(4):405-416. Munson, P. J., and C. W. Marean 2003 Adults Only? A Reconsideration of Middle Paleolithic 'Prime-dominated' Reindeer Hunting at Salzgitter Lebenstedt. Journal ofHuman Evolution 44:263-273. Mussi, M., and P. Villa 2008 Single Carcass of Mammuthus primigenius with Lithic Artifacts in the Upper Pleistocene ofNorthern Italy. Journal ofArchaeological Science 35:2606-2613. Nagaoka, L. 2001 Using Diversity Indices to Measure Changes in Prey Choice at the Shag River Mouth Site, Southern New Zealand. International Journal of Osteoarchaeology 11:101-111. 2006 Prehistoric Seal Carcass Exploitation at the Shag Mouth Site, New ¦ Zealand. Journal ofArchaeological Science 33: 1474—148 1 . 221 O'Connell, J. F. 2006 How Did Modern Humans Displace Neanderthals? Insights from Hunter- Gatherer Ethnography and Archaeology. In When Neanderthals and Modern Humans Met, edited by N. J. Conard, pp. 43-64. Kerns Verlag, Tübingen. O'Connell, J. F., K. Hawkes, and N. Blurton Jones 1988 Hadza Scavenging: Implications for Plio/Pleistocene Hominid Subsistence. Current Anthropology 29:356-363. 1990 Reanalysis of Large Mammal Body Part Transport Among the Hadza. Journal ofArchaeological Science 17:301-316. Owen-Smith, R. N. 1988 Megaherbivores: The Influence of Very Large Body Size on Ecology. Cambridge Unversity Press, Cambridge. Pales, L., and M. A. Garcia 1981 Atlas Ostéologique pour Servir à l'Identification des Mammifères du Quaternaire. II . Tête - Rachis Ceintures Scapulaire et Pelvienne. Herbivores. Éditions CNRS, Paris. Parrini, F., J. W. Cain, III, and P. R. Krausman 2009 Capra ibex (Artiodactyla: Bovidae). Mammalian Species 830:1-12. Pérez, J. M., J. E. Granados, R. C. Soriguer, P. Fandos, F. J. Márquez, and J. P. Crampe 2002 Distribution, status and conservation problems of the Spanish Ibex, Capra pyrenaica (Mammalia: Artiodactyla). Mammal Review 32:26-39. Pérez Ripoll, M. 2004 La Consommation Humaine des Lapins pendant le Paléolithique dans la région de Valencia (Espagne) et l'Étude des Niveaux Gravétiens de la Cova de les Cendres (Alicante). In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires, edited by J.-P. Brugal and J. Desse, pp. 191-206. Éditions ADPCA, Antibes. Pérez, T., J. Albornoz, and A. Domínguez 2002 Phylogeography of Chamois {Rupicapra spp.) Inferred from Microsatellites. Molecular Phylogenetics and Evolution 25:524-534. Pfeiffer, J. E. 1969 The Emergence ofMan. Harper and Row, New York. Pickering, T. R., M. Domínguez-Rodrigo, C. P. Egeland, and C. K. Brain 2005 The Contribution of Limb Bone Fracture Patterns to Reconstructing Early Hominid Behaviour at Swartkrans Cave (South Africa): Archaeological 222 Application of a New Analytical Method. International Journal of Osteoarchaeology 15:247-260. Pickering, T. R., C. W. Marean, and M. Domínguez-Rodrigo 2003 Importance of Limb Bone Shaft Fragments in Zooarchaeology: A Response to "On in situ Attrition and Vertebrate Body Part Profiles" (2002), by M.C. Stiner. Journal ofArchaeological Science 30: 1469-1482. Pike-Tay, A., V. Cabrera Valdés, and F. Bernaldo de Quirós 1999 Seasonal Variations of the Middle-Upper Paleolithic Transition at El Castillo, Cueva Morín and El Pendo (Cantabria, Spain). Journal ofHuman Evolution 36:283-317. Pond, C. M., and C. A. Mattacks 1985 Body Mass and Natural Diet as Determinants of the Number and Volume of Adipocytes in Eutherian Mammals. Journal ofMorphology 185:183— 193. Pond, C. M., C. A. Mattacks, R. H. Colby, and N. J. C. Tyler 1993 The Anatomy, Chemical Composition and Maximum Glycolytic Capacity of Adipose Tissue in Wild Svalbard Reindeer (Rangifer tarandus platyrhynchus) in Winter. Journal ofZoology, London 229:17-40. Pond, C. M., C. A. Mattacks, and P. Prestrud 1995 Variability in the Distribution and Composition of Adipose Tissue in Wild Arctic Foxes (Alopex lagopus) on Svalbard. Journal ofZoology, London 236:593-610. Pyke, G. H. 1 984 Optimal Foraging Theory: A Critical Review. Annual Review ofEcology and Systematics 15:523-575. Reitz, E. J., and E. S. Wing 1999 Zooarchaeology. Cambridge University Press, Cambridge. Reynolds, H. W., C. C. Gates, and R. D. Glaholt 2003 Bison. In Wild Mammals ofNorth America: Biology, Management, and Conservation, edited by G. A. Feldhamer, B. C. Thompson, and J. A. Chapman, pp. 1009-1060. 2nd edition. John Hopkins University Press, Baltimore. Richards, M. P., P. B. Pettitt, M. C. Stiner, and E. Trinkaus 2001 Stable Isotope Evidence for Increasing Dietary Breadth in the European Mid-Upper Paleolithic. Proceedings ofthe National Academy ofSciences 98:6528-6532. Richards, M. P., P. B. Pettitt, E. Trinkaus, F. H. Smith, M. Paunovic, and I. Karavanic 2000 Neanderthal Diet at Vindija and Neanderthal Prédation: The Evidence from Stable Isotopes. Proceedings ofthe National Academy ofSciences 97:7663-7666. Richards, M. P., and E. Trinkaus 2009 Isotopie Evidence for the Diets of European Neanderthals and Early Modern Humans. Proceedings ofthe National Academy ofSciences 106:16034-16039. Ringrose, T. J. 1993 Bone Counts and Statistics: A Critique. Journal ofArchaeological Science 20:121-157. Rivals, F., N. Solounias, and M. C. Mihlbachler 2007 Evidence for Geographic Variation in the Diets of Late Pleistocene and Early Holocene Bison in North America, and Differences from the Diets of Recent Bison. Quaternary Research 68:338-346. Rodriguez, F., S. Hammer, T. Pérez, F. Suchentrunk, R. Lorenzini, J. Michallet, N. Martinkova, J. Albornoz, and A. Domínguez 2009 Cytochrome b Phylogeography of Chamois {Rupicapra spp.). Population Contractions, Expansions and Hybridizations Governed the Diversification of the Genus. Journal ofHeredity 100:47-55. Rutberg, A. T. 1987 Adaptive Hypotheses of Birth Synchrony in Ruminants: An Interspecific Test. The American Naturalist 130(5):692-710. Saenz de Buruaga, A. 1991 El Paleolitico Superior de la Cueva de Gatzarria (Zuberoa, Pais Vasco). Veleia, Anejos Series Maior, Vitoria-Gasteiz. Saether, B.-E. 1997 Environmental Stochasticity and Population Dynamics of Large Herbivores: A Search for Mechanisms. Trends in Ecology and Evolution 12:143-149. Schwartz, C. C. 1 992 Physiological and Nutritional Adaptations of Moose to Northern Environments. Alces Supplement 1:139-155. Scott, K. 1980 Two Hunting Episodes of the Middle Palaeolithic Age at La Cotte de Saint-Brelade, Jersey (Channel Islands). WorldArchaeology 12:137-153. 224 Scott, K., R. Bonnichsen, and M. H. Sorg 1989 Mammoth Bones Modified by Humans: Evidence from La Cotte de St. Brelade, Jersey, Channel Islands. In Bone Modification, edited by R. Bonnichsen and M. H. Sorg, pp. 335-346. Center for the Study of the First Americans, Orono, ME. Sherman, S. J., and J. Steele 1999 Cultural Learning in Hominids: A Behavioural Ecology Approach. In Mammalian Social Learning: Comparative and Ecological Perspectives, edited by H. O. Box and K. R. Gibson, pp. 367-388. Symposia of the Zoological Society of London 72. Cambridge University Press, Cambridge. Simpson, E. H. 1949 Measurement of Diversity. Nature 163:688. Sinclair, A. R. E., S. A. R. Mduma, and P. Árcese 2000 What Determines Phenology and Synchrony of Ungulate Breeding in Serengeti? Ecology 81(8):2100-21 1 1. Slott-Moller, R. 1988 Contribution à l'Étude Paléontologique d'un Gisement Préhistorique: L'Exemple du Roc de Marcamps (Gironde). Unpublished D.E.S.S. thesis, Université de Bordeaux I. Smith, E. A. 1 99 1 Inujjuamiut Foraging Strategies. Evolutionary Ecology ofan Arctic Hunting Economy. Aldine de Gruyter, New York. Smith, E. A., R. Bliege Bird, and D. Bird 2003 The Benefits of Costly Signaling: Meriam Turtle Hunters. Behavioral Ecology 14(1): 116-126. Smith, E. A., and B. Winterhaider 1985 On the Logic and Application of Optimal Foraging Theory: A Brief Reply to Martin. American Anthropologist 87:645-648. 1992 Evolutionary Ecology and Human Behavior. Aldine De Gruyter, New York. Smith, F. H., I. Jankovic, and I. Karavanic 2005 The Assimilation Model, Modern Human Origins in Europe, and the Extinction ofNeandertals . Quaternary International 137:7-19. So, J. K. 1980 Human Biological Adaptation to Arctic and Subarctic Zones. Annual Review ofAnthropology 9:63—82. Speth, J. D. 1983 Bison Kills and Bone Counts. Prehistoric Archeology and Ecology Series. Chicago University Press, Chicago. Speth, J. D., and K. A. Spielmann 1983 Energy Source, Protein Metabolism, and Hunter-Gatherer Subsistence Strategies. Journal ofAnthropological Archaeology 2: 1-3 1 . Speth, J. D., and E. Tchernov 2002 Middle Paleolithic Tortoise Use at Kebara Cave (Israel). Journal of Archaeological Science 29:471-483. Steele, T. E. 2002 Red Deer: Their Ecology and How They Were Hunted by Late Pleistocene Hominids in Western Europe. Unpublished Ph.D. dissertation, Stanford University. 2004 Variation in Mortality Profiles of Red Deer (Cervus elaphus) in Middle Palaeolithic Assemblages from Western Europe. International Journal of Osteoarchaeology 14:307-320. Stephens, D. W., and J. R. Krebs 1986 Foraging Theory. Princeton University Press, Princeton. Stewart, J-. R., and A. M. Lister 2001 Cryptic Northern Refugia and the Origins of the Modern Biota. Trends in Ecology and Evolution 16(1 1):608—613. Stiner, M. C. 1990 The Use of Mortality Patterns in Archaeological Studies of Hominid Predatory Adaptations. Journal ofAnthropological Archaeology 9:305- 331. 1991 Food Procurement and Transport by Human and Non-human Predators. Journal ofArchaeological Science 18:455—482. 1994 Honor Among Thieves. Princeton University Press, Princeton. 2001 Thirty Years on the "Broad Spectrum Revolution" and Paleolithic Demography. Proceedings ofthe National Academy ofSciences 98:6993- 6996. 2004 Small Game Use and Expanding Diet Breadth in the Eastern Mediterranean Basin during the Palaeolithic. In Petits Animaux et Sociétés Humaines. Du Complement Alimentaire aux Ressources Utilitaires, edited by J. P. Brugal and J. Desse, pp. 499-513. Édition APDCA, Antibes. 2005 The Faunas ofHayonim Cave: A 200, 000-year Record ofPaleolithic Diet, Demography, and Society. American Schools ofPrehistoric Research Bulletin 48. Peabody Museum of Archaeology and Ethnology, Cambridge, Massachusetts. Stiner, M. C, S. L. Kuhn, S. Weiner, and O. Bar-Yosef 1995 Differential Burning, Recrystallization, and Fragmentation of Archaeological Bone. Journal ofArchaeological Science 22:223-237. Stiner, M. C, N. D. Munro, and T. A. Surovell 2000 The Tortoise and the Hare: Small-game Use, the Broad-Spectrum Revolution and Paleolithic Demography. Current Anthropology 41:39-73. Stiner, M. C, N. D. Munro, T. A. Surovell, E. Tchernov, and O. Bar-Yosef 1999 Paleolithic Population Growth Pulses Evidenced by Small Animal Exploitation. Science 283:190-194. Straus, L. G. 1982 Carnivores and Cave Sites in Cantabrian Spain. Journal of Anthropological Research 38:7'5-96. 1990 Underground Archaeology: Perspectives on Caves and Rockshelters. Archaeological Method and Theory 2:255-304. Stringer, C. B., and P. Andrews 1988 Genetic and Fossil Evidence for the Origin of Modern Humans. Science 239:1263-1268. Stringer, C. B., J. C. Finlayson, R. N. E. Barton, Y. Fernández-Jalvo, I. Cáceres, R. C. Sabin, E. J. Rhodes, A. P. Currant, J. Rodriguez-Vidal, F. Giles-Pacheco, and J. A. Riquelme-Cantal 2008 Neanderthal Exploitation of Marine Mammals in Gibraltar. Proceedings of the National Academy ofSciences 105: 143 19-14324. Stringer, C. B., and C. Gamble 1993 In Search ofthe Neanderthals. Thames and Hudson, London. Tartar, E. 2009 De VOs à l'Outil. Caractérisation Technique, Économique et Sociale de l'Utilisation de l'Os à l'Aurignacien Ancien. Unpublished Ph.D. Dissertation, Université de Paris I. Théry-Parisot, I. 2002 Fuel Management (Bone and Wood) During the Lower Aurignacian in the Pataud Rock Shelter (Lower Paleolithic, Les Eyzies de Tayac, Dordogne, France). Contribution of Experimentation. Journal ofArchaeological Science 29:1415-1421. Thorne, A., and M. H. Wolpoff 1 992 The Multiregional Evolution of Humans. Scientific American 266:76-83. Trinkaus, E. 2007 European Early Modern Humans and the Fate of the Neandertals. Proceedings ofthe National Academy ofSciences 104:7367-7372. Turner, A. 1989 Sample Selection, Schlepp Effects and Scavenging: The Implications of Partial Recovery for Interpretations of the Terrestrial Mammal Assemblage from Klasies River Mouth. Journal ofArchaeological Science 16:1—11. Turner, C, and G. E. Hannon 1988 Vegetational Evidence for Late Quaternary Climatic Changes in Southwest Europe in Relation to the Influence of the North Atlantic Ocean. Philosophical Transactions ofthe Royal Society ofLondon B 318:451-485. Ugan, A. 2005 Does Size Matter? Body Size, Mass Collection, and their Implications for Understanding Prehistoric Foraging Behaviour. American Antiquity 70:75-89. Valensi, P. 2002a Fiche Extrémités Distales d'Humérus de Grands Ongulés. In Retouchoirs, Compresseurs, Percuteurs... Os à Impressions et Éraillures, edited by M. Patou-Mathis, pp. 75-85. Fiches de la Commission de Nomenclature sur l'Industrie de l'Os Préhistorique, Cahier X. Editions Société Préhistorique Française, Paris. 2002b Fiche Phalanges d'Ongulés. In Retouchoirs, Compresseurs, Percuteurs... Os à Impressions et Eraillures, edited by M. Patou-Mathis, pp. 87-97. Fiches de Ia Commission de Nomenclature sur l'Industrie de l'Os Préhistorique, Cahier X. Éditions Société Préhistorique Française, Paris. van Vuure, T. 2005 Retracing the Aurochs: History, Morphology and Ecology ofan Extinct Wild Ox. Pensoft Publishers, Sofia. Villa, P., J. C. Castel, C. Beauval, V. Bourdillat, and P. Goldberg 2004 Human and Carnivore Sites in the European Middle and Upper Palaeolithic: Similarities and Differences in Bone Modification and Fragmentation. Revue de Paléobiologie 23 :705-730. Villa, P., and E. Mahieu 1 99 1 Breakage Patterns of Human Long Bones. Journal ofHuman Evolution 21:27-48. Villaret, J. C, R, Bon, and A. Rivet 1997 Sexual Segregation of Habitat by the Alpine Ibex in the French Alps. Journal ofMammalogy 78:1273—1281. Weinstock, J. 2000 Osteometry as a Source of Refined Demographic Information: Sex-ratios of Reindeer, Hunting Strategies, and Herd Controíin the Late Glacial Site of Stellmoor, Northern Germany. Journal ofArchaeological Science 27:1187-1195. Winterhaider, B. 1 98 1 Foraging Strategies in the Boreal Forest: An Analysis of Cree Hunting and Gathering. In Hunter-Gatherer Foraging Strategies: Ethnographic and Archaeological Analyses, edited by B. Winterhaider and E. A. Smith, pp. 66-98. University of Chicago Press, Chicago. 2002 Behavioral and Other Human Ecologies: Critique, Response and Progress through Criticism. Journal ofEcological Anthropology 6:4-23. Winterhaider, B., W. Baillargeon, F. Cappelletto, I. R. Daniel, Jr., and C. Prescott 1988 The Population Ecology of Hunter-Gatherers and Their Prey. Journal of Anthropological Archaeology 7:289-328. Wolpoff, M. H., J. Hawks, and R. Caspari * 2000 Multiregional, Not Multiple Origins. American Journal ofPhysical Anthropology 112:129-136. Wolpoff, M. H., J. Hawks, D. W. Frayer, and K. Hunley 2001 Modern Human Ancestry at the Peripheries: A Test of the Replacement Theory. Science 291:293-297. Wolverton, S. 2008 Harvest Pressure and Environmental Carrying Capacity: An Ordinal-Scale Model of Effects on Ungulate Prey. American Antiquity 73:179-199. Woodward, J. C, and P. Goldberg 2001 The Sedimentary Records in Mediterranean Rockshelters and Caves: Archives of Environmental Change. Geoarchaeology 16(4):327-354. Yesner, D. R. 1985 Archeology and Optimal Foraging Theory: Appropriate Analytical Units. American Anthropologist 87:412—414. Wood, B. 1976 The Evolution ofEarly Man. Eurobook Limited. Yravedra Sainz de los Terreros, J. 2004 Implications Taphonomiques des Modifications Osseuses Faites par les Vrais Hiboux {Bubo bubo) sur les Lagomorphes. In Petits Animaux et Sociétés Humaines. Du Complément Alimentaire aux Ressources Utilitaires, edited by J. P. Brugal and J. Desse, pp. 321-324. Éditions APDCA, Amibes. Zilhäo, J. 2006 Neandertals and Modems Mixed, and It Matters. Evolutionary Anthropology 15:183-195. Zilhäo, J., and F. d'Errico 1999 The Chronology and Taphonomy of the Earliest Aurignacian and its Implications for the Understanding ofNeandertal Extinction. Journal of World Prehistory 13:1-68. 2000 La Nouvelle 'Bataille Aurignacienne.' Une Révision Critique de la Chronologie du Châtelperronien et de l'Aurignacien Ancien. L Anthropologie 1 04: 1 7-50. Zilhäo, J., F. d'Errico, J. G. Bordes, A. Lénoble, J. P. Texier, and J. P. Rigaud 2008 Grotte des Fées (Châtelperron): History of Research, Stratigraphy, Dating, and Archaeology of the Châtelperronian Type-Site. PaleoAnthropology 2008:1^2.