Introduction and main issues

1The domestication of the wolf marks a veritable revolution in the history of human societies and is a fundamental issue in studying the cognitive and technical evolution of recent Palaeolithic hunter-gatherer groups. The first animal to be domesticated was the dog, and this undeniably opened up new possibilities for human groups (e.g., group protection, prestige symbol, hunting assistant, waste elimination sensu Clutton-Brock 1977; Morey 1996; Müller 2005; Morey 2010; Vigne 2011). The chronological modalities of this process have raised a lot of controversy (e.g,. Germonpré et al. 2009; Morey 2010; Ovodov et al. 2011; Crockford and Kuzmin 2012; Germonpré et al. 2012; Napierala and Uerpmann 2012 for the most recent references). Although some authors consider this process to have begun very early, at the beginning of the Upper Palaeolithic (Germonpré et al. 2009; Ovodov et al. 2011; Germonpré et al. 2012), the domestication of the dog is widely documented at several Tardiglacial sites (e.g., Davis and Valla 1978; Morel and Müller 1997; Chaix 2000; Vigne 2005; Morey 2010; Pionnier-Capitan et al 2011; Napierala and Uerpmann 2012).

2In the archaeological record, it must be noted that distinguishing between dog remains (Canis familiaris) and wolf remains (Canis lupus) is a delicate task as the morphology and the size of these two canids are similar during the Palaeolithic period. The domestication process results in several morphological modifications which are represented by the conservation of neotenic characteristics in domesticated adults and a significant decrease in size (e.g., Hemmer 1990; Morey 1992; Clutton-Brock 1995; Morey 1996, 2010). These changes can occur over a period of several decades (Trut 1999). Size thus generally represents the most diagnostic criterion.

3The analysis of canid remains discovered in Palaeolithic contexts thus represents an essential aspect of prehistoric research in order to describe the modalities of the emergence and the diffusion of the domestication process throughout time and space. Using a considerable biometric reference system made up of western European fossil and modern wolves (Boudadi-Maligne 2010), the reassessment of the canid remains from Abri du Morin is presented here.

The reassessment of the Morin rock-shelter series

4Abri du Morin is located in the commune of Pessac-sur-Dordogne, in Gironde (fig. 1). It was excavated between 1954 and 1958 by Deffarge who described two main sedimentary complexes (Lenoir 1970 - pl. 18), covered by a superficial level containing several pottery shards and other post-glacial remains. Complex A is subdivided by Deffarge into four subcomplexes (from the top to the bottom: I to IV) (Bordes 1959). It yielded a large quantity of lithic and faunal remains attributed to the Final Magdalenian (Bordes and Sonneville-Bordes 1979). The underlying complex B, subdivided into two subcomplexes (I and II), yielded remains ascribed to the Upper Magdalenian (ibid. Lenoir 1970, 1978, 2003). The hypothesis of the Azilianization of the Magdalenian through an internal evolution of the latter was advanced, based on the evolution of the lithic industry (Bordes 1959; Lenoir 1978; Bordes and Sonneville-Bordes 1979). Abri du Morin also yielded an abundant antler industry, with numerous decorated harpoons, a diversity of decorated objects (Deffarge et al. 1975) and a rich fauna (Chauviré 1965; Delpech 1967; Donard 1982; Delpech 1983). The ongoing revision of the lithic and antler series from Le Morin, in the context of the ANR Magdatis, has already yielded evidence of typical Laborian hunting projectiles (Langlais et al. in press). The current hypothesis thus reveals the presence of a Laborian complex, which had not been recognized during previous research.

The fauna from Le Morin

5The dominant species represented at Le Morin include the reindeer (Rangifer tarandus), bovids and the horse (Equus caballus) (Delpech 1983; Kuntz in progress). The presence of the red deer in levels A-IV (Cervus elaphus), the wild boar (Sus scrofa) and the roe deer (Capreolus capreolus) could indicate a more temperate and humid climate during the accumulation of the remains in level B-I.

6The carnivores in complex A include the brown bear (Ursus arctos) and the lion (Panthera sp.). Small carnivores are represented by the fox (Vulpes vulpes), the Arctic fox (Alopex lagopus), the lynx (Lynx sp.), the wild cat (Felis silvestris), the badger (Meles meles), the marten (Martes martes) and the weasel (Mustela nivalis). Three leporidae species are present at the site: the mountain hare (Lepus timidus), the brown hare (Lepus europaeus) and the European rabbit (Oryctolagus cuniculus). Among the rodents, the presence of the beaver (Castor fiber), the gopher (Spermophilus sp.), the dormouse (Glis glis), the large vole (Arvicola sp.) and the field mouse (Apodemus sylvaticus) is noteworthy. Lastly, hedgehog (Erinaceus europaeus) and mole (Talpa europaea) insectivore remains were identified. Only one rabbit and one hare remain come from level B. With the exception of the dormouse, the large vole, the field mouse and the mole, the archaeozoological analysis shows that all the other species bear marks of human modification. They were thus brought to the site by the hunter-gatherers.

7Nineteen different bird species were identified at the site (Mourer-Chauviré 1975; Delpech 1983) in two levels. The archaeozoological analysis of the snowy owl remains (Bubo scandiaca) shows that this bird of prey was exploited by the Magdalenians (Gourichon 1994). The analysis of the other species is in progress and will reveal their role in Magdalenian exploitation systems (Laroulandie in progress).

8Lastly, salmon (Salmo salar), trout (Truta fario) and pike (Esox lucius) remains were also identified (Delpech 1967, 1983).

9Until now, the absolute chronology of the Morin site consisted of a single conventional date of 10,480 ± 200 BP (Gif-2105) (Delibrias and Evin 1974). This date was nonetheless to be treated with caution as the stratigraphic location of the sample was unknown (Lenoir 1983). During the revision of the sequence, several species underwent direct dating in order to refine the site chronology (fig. 2). In the subcomplex B-I a reindeer remain and a snowy owl remain were dated to 15,800 and 15,000 cal BP. Dating carried out on a red deer and another snowy owl yielded ages ranging between 15,200 and 14,000 cal BP (Szmidt et al. 2009; Langlais et al. 2012). In order to test the assertions concerning more recent occupations (e.g., from the Azilian and the Laborian), direct dates on temperate species are in progress.

10In the context of the reassessment of the fauna and site chronology, the canid remains were reexamined in order to revisit the question of the presence of the dog raised several decades ago (Suire 1969).

Analysis of the canid remains

11Several studies have previously been carried out on the canid remains. They were first described by Delpech (1967, 1983) and were ascribed to the wolf. Then they were analyzed by Suire (1969) who observed that a small sized lower carnassial tooth could come from a dog, although he expressed doubt concerning the stratigraphic position of this tooth among the other remains. The remains were later studied by Langlois (2000) who reattributed all the dental remains to the wolf (Canis lupus) on the basis of the talonid morphology of the lower carnassial. According to Langlois, the small size of the Canis present at Morin constituted evidence of the decrease in size of this species after the “Würm maximum”.

12Our analysis concerns all the previously studied remains (NR = 24). In addition, during the analysis of small game and carnivores by one of us (J.-B. Mallye), new canid remains were identified (different from the vulpine remains) bringing their total number to 32. These elements are mostly from complex A (tabl. 1) and two thirds of them are from subcomplexes IV and III.

13Different skeletal parts are represented. However, the excavation methods used at the time doubtlessly contributed to the anatomical bias of the collection. Taking account of the identified remains and the distinction between complex A and complex B, the minimum number of individuals is two.

Generic attribution

14As the presence of the dhole has been attested in several Upper Palaeolithic sites (e.g,. Altuna 1983; Perez Ripoll et al. 2010), it is essential to confirm the presence or absence of this species in the assemblage studied (Pionnier-Capitan et al. 2011). The morphology of the lower carnassials leaves no room for ambiguity. Indeed, on MOR72, the presence of two cusps on the talonid of the first molar (Fig. 3-A-a) as well as the presence of an alveolus for the lower third molar (fig. 3-A-b) very clearly indicates that this hemi mandible is attributable to the Canis genus and not the Cuon. The latter is characterized by a monocusped lower carnassial talonid (fig. 3-B-a) and the absence of the lower third molar (fig. 3-B).

15The morphological study of the dental remains allows us to affirm that they are exclusively from the genus Canis

16Although some of the post cranial elements are small in size, the recently published criteria for determining the dhole (Perez-Ripoll et al. 2010; Pionnier-Capitan et al. 2011) were not identified on the canid remains from Le Morin. We thus attribute all of the 32 remains to the genus Canis.

Morphometric analysis

17Considering the small dimensions of certain remains, we conducted a biometric analysis in order to define their specific attribution. We used the measurements described by von den Driesch (1976) in the context of a PhD study (Boudadi-Maligne 2010).

18The measurements of bone and dental remains were compared to three datasets, namely:

• fossil dogs (Nobis 1979; Tchernov & Valla 1997; Chaix 2000; Pionnier-Capitan 2011; Napierala and Uerpmann 2012);

• fossil wolves dating from the Upper Pleniglacial and the Tardiglacial (Boudadi-Maligne 2010, unpublished);

• present day populations of European wolves (Boudadi-Maligne 2010, unpublished).

19Among the 32 remains attributed to the genus Canis, 19 can clearly be ascribed to the wolf (Canis lupus) because of their large size (tabl. 2) which fits well with the known variability of wolves at the end of the Upper Pleistocene (Boudadi-Maligne 2010).

20Seven remains (fig. 4) can be ascribed to the dog (Canis familiaris). These are made up of two isolated teeth, a lower carnassial (fig. 4-3) and a third upper incisor (fig. 4-5) and five long bone fragments (fig. 4-1, 2, 4, 6, 7). Bivariate projections of the measurements of the lower carnassial and the upper incisor (fig. 5) indicate that they are both outside the known range of variability for present day and fossil wolves.

21In order to quantify these differences, the probabilistic distances (Maureille et al. 2001) were calculated between these two teeth and their counterparts from our reference collection of present day and fossil wolves. This distance is calculated according to the formula adapted to small samples (Santos in Scolan et al. 2011). These data (tabl. 3) confirm that the lower carnassial and the upper third incisor from Le Morin rock-shelter are outside the known variability range for fossil and present day wolves. This distance is even more marked if only the robustness index (ROB) of these teeth is taken into account: product of the mesio-distal diameter (DMD) and the vestibulo-lingual diameter (DVL).

22In the same way, for three long bone shaft fragments, the biometric data shows significant differences for fossil and modern wolves (fig. 6). Two other remains, distal shaft fragments from a radius and a humerus, not measurable in the strict sense of von den Driesch (1976), present very small transverse and anterior-posterior diameters. In addition, the thin cortical thickness of these bones is not compatible with that observed on wolf long bones. On this type of element, biometric reference data are extremely rare, which prevents any quantitative comparison. Nonetheless, the proportions of these pieces associate them with the dog rather than the wolf.

23Lastly, six remains were not attributed to either species, as we did not have access to comparative elements. These remains are three distal extremities of a metapodial, two incisors and a fibula shaft fragment.

Archaeozoological analysis

24All the pieces were observed with a stereomicroscope LEICA Z16 APO, with minimum magnifications of x40 in order to identify any marks likely to provide information concerning the origin and the status of the remains at the site. Ten of the wolf remains bear human-induced marks. These are scraping marks identified on the root of an upper right canine (fig.7 - A) and a lower right canine. These marks can be interpreted as preparation of the tooth before perforation. Other marks were also identified on phalanges and on the mandible, signifying skinning activities. Lastly, cutmarks identified on a radius demonstrate defleshing of the bone.

25Cutmarks were also identified on canid remains (wolf or dog) from two metapodials (fig.7 – B) as well as on the crowns from two incisors. These marks can be linked to skinning when the skin is pulled over the head (Mallye 2011). Lastly, a fibula fragment bearing cutmarks can be linked to defleshing.

26Dog long bones bear anthropogenic marks. These are cutmarks, located on the posterior face of a right femur (fig.7 – D), but also on the posterior face of a radius and a tibia shaft. Lastly, a 3 mm long flint fragment was also found wedged in a radius shaft (fig.7 – C).

27These marks demonstrate the exploitation of both the dog and wolf by Tardiglacial human groups.

Dating the Morin dog remains?

28During the reassessment of the site, an analysis of systematic refits of ungulate dental series was carried out (J.-B. Mallye, fig. 8) resulting in connections and dental near connections. Thirty-six of these were pieces from the same complexes. In the other cases, these connections and near connections concern mainly the pieces from the different A subcomplexes. Some rare connections link the two complexes A and B. Coupled with a revision of the lithic industries (M. Langlais), the results obtained from the analysis of the dispersal of the connected faunal pieces show that the different levels recorded during excavations correspond to artificial constructs rather than to stratigraphic subdivisions. Thus, in order to discuss the cultural attribution of the identified canid remains and thereby the spread of domestication among Palaeolithic groups, it appeared essential to carry out direct dates on the Morin dog remains.

29Four samples were sent to the ORAU (Oxford Radiocarbon Accelerator Unit). These four remains, made up of two bones and two dental remains, were first molded following the protocol established by the ORAU and photographed on all sides. These remains had not been coated with conservation or consolidation products. The first two, a femur shaft and a radius shaft, did not contain enough collagen to be dated and two teeth attributed to Canis familiaris (lower M1 and upper I3) were thus submitted instead.

30They were treated using the standard collagen extraction method, which includes a pretreatment phase of ultrafiltration (Bronk Ramsey et al. 2002, 2004; Brock et al. 2010). The radiocarbon ages expressed in non-calibrated BP were obtained using the Libby value for the half-life of carbon 14 and corrected with natural isotopic splitting. The collagen level, elementary measurements and the stable isotope measurements for both the dated samples are within the normal margins (Van Klinken 1999). The calibration of the dates was conducted using the OxCal v4.1.7 program developed by Bronk Ramsey (2009, 2010), based on the atmospheric data from IntCal09 (Reimer et al. 2009).

31The lower carnassial from A III (OxA-23628) is dated to 12,450 ± 55 BP or 15,005-14,155 (95.4%) cal BP and the third upper incisor found in A I (OxA-23627) to 12,540 ± 55 BP or 15,114-14,237 (95.4%) cal BP.

The domestication of the wolf in Western Europe

32In spite of widespread interest in the domestication of the wolf, very few dog remains have been directly dated. Indeed, only four Tardiglacial sites can be cited (fig. 9): Saint-Thibaud-de-Couz in Savoie (Chaix 2000), le Pont d’Ambon in Dordogne (Célérier et al. 1999), Kesslerloch in Switzerland (Napierala and Uerpmann 2012) and Bonn-Oberkassel in Germany (Baales 2006). The dog remains from Pont d’Ambon and Saint Thibaut de Couz yield dates from the first half of the recent Dryas for the first and from the recent Dryas-Preboreal for the second. They are associated with complexes from the end of the Palaeolithic, probably Laborian (Célérier et al.1999; Chaix 2000). The remains from Bonn-Oberkassel and Kesslerloch dated to the end of the Bölling and the Alleröd were found in a Magdalenian context (Baales 2006, Napierala and Uerpmann 2012). Dog remains elsewhere which have not benefited from direct dating are associated with the same technocomplexes: Hauterive-Champréveyres in Switzerland (Morel and Müller 1997), Erralla in Spain (Altuna et al. 1985; Vigne 2005) and Montespan in Haute-Garonne (Pionnier-Capitan et al. 2011). The dates obtained on the Morin dog remains confirm the presence of this canid in an Upper Magdalenian context and the spread of this zootechnical innovation in Western Europe from the Bölling onwards.

33The biometric analysis of the Morin dog remains reveals the small size of the specimens, pointing towards profound transformations in comparison to their wild ancestors. Such modifications were described elsewhere following experimental domestication research on canids (Trut 1999). In controlled conditions this type of modification is visible very early on, after about 20 to 100 generations, which represents several decades (Trut 1999; Arbuckle 2005). Therefore, given the radiometric dates currently available and the presence of several C14 plateaux (fig. 9), it is not easy to perceive the different stages of this wolf to dog transformation.

34However, the size of the Taridglacial dogs shows that the domestication process is already fully complete. Certain authors suggest that domestication appeared during the Upper Palaeolithic (Germonpré et al. 2009; Ovodov et al. 2011; Germonpré et al. 2012), but it is noteworthy that, as other authors have suggested (Detry and Cardoso 2010; Morey 2010; Napierala and Uerpmann 2012), no decrease in size has been recorded in canid remains from pre-Magdalenian periods. Arguments in favor of early domestication must thus be treated with caution. Domestication seems to be rooted either in Middle Magdalenian societies (cf. Erralla or Montespan), or in a different cultural substratum nearly contemporaneous with the Magdalenian.

Discussion et conclusion

Early dates

35Several remains ascribed to the genus Canis were identified at Morin. Among these, seven present dimensions clearly outside the range of variability of fossil and modern wolves. These are not small Tardiglacial wolves but genuine dogs living alongside Upper Magdalenian hunter-gatherers from the Bölling onwards. Moreover, the dates obtained are relatively similar even though the remains are from different subcomplexes. It is thus reasonable to assume that they are from the same specimen.

Waste elimination

36At Morin, the presence of ungulate remains with tooth marks made by a medium-sized carnivore could also indicate the proximity of Man and dog. However, contrary to expectations (e.g., Klippel et al. 1987; Morey and Klippel 1991), there are no digested bones at the site. It is nonetheless possible that unidentifiable bones such as these were considered to be of little or no archaeological interest during excavations. If such bones existed, they were not collected.

Man’s best friend?

37The examination of dog bone surfaces revealed the presence of butchery marks. This animal was thus treated as a meat resource in the same way as other species. This calls into question the status of this animal in hunter-gatherer groups. Indeed, during the Upper Magdalenian, the treatment of human remains indicates a complex relationship with death. Tombs are rare and human remains are frequently mixed with faunal remains. In addition, they can also bear marks of human intervention (e.g., Gambier 1990/1991; Le Mort and Gambier 1992; Gambier 1995).

38Can we therefore imagine that the marks described on the Morin dog remains demonstrate an identical status to that of humans, or at least different from the hunted species? If the dog was a food resource during the Upper Magdalenian, did the presence of dog represent planned resource management by keeping a live stock of meat?

Other changes in societies: the dog as a hunting aid?

39Although the dog is traditionally associated with hunting activities, ethnography provides contradictory insights (Poplin et al. 1986; Gautier 1990; Vigne 2004). Some groups use canids as hunting aids, others for accompanying hunters, whereas others send the canids back to camp when the hunters leave (synopsis in Digard 2006, 2009).

40If Magdalenian groups possessed dogs, did new hunting strategies accompany this zootechnical innovation? The reply to this question is complex given the paltry data provided by the archaeological record. However, two observations can be recalled here: the evolution of hunting arms and the integration of small game in the diet. In addition to the presence of dogs, hunting weapons during the Upper Magdalenian demonstrated a redeployment of lithic points. Taking the example of Morin, the successive development of shouldered points then Laugerie-Basse type leaf points underlines an internal Magdalenian dynamic which initiates in a way the emergence of Azilian backed points. Shaft types or point hafting remain widely unknown and underestimated due to the absence of taphonomical conditions propitious to wood preservation. It would be interesting to conduct an experimental program to tackle the thorny question of the use of the spear thrower and/or the bow. However, it would be simplistic to consider the dog as an integral element of the hunter with a bow. If it difficult to directly correlate the hypothesis of the dog as hunting aid with the evolution of weapons, it is worth recalling that at the same time, small game, such as birds, leporids, small carnivores and large rodents are more intensively exploited (e.g., Stiner et al. 1999, 2000; Laroulandie 2000; Cochard 2004; Hocket and Haws 2005; Costamagno et al. 2008). It is thus legitimate to question the dog’s role in the acquisition strategies of these small animals.

Towards a systemic vision

41Correlating these events does not necessarily signify that there is a causal relationship between them and these hypotheses remain difficult to prove. The systematic search for clues such as the presence of eaten and digested remains in archaeological sites where dog remains have been identified as well as research concerning the association between dog remains and hunted species will lead to a better understanding of the role played by the dog among the first human groups to have domesticated it and will entail a more systemic vision of the domestication process.

The authors wish to thank the UMR 5199 PACEA as well as the Treilles foundation for financing this research. We also wish to thank the Magdatis project, n° ANR 2011 BSH3 005 02 and all the personnel involved as well as the Bordeaux Muséum d’Histoire naturelle. We are grateful to Michel Lenoir for the documentation and information that he supplied and to an anonymous reviewer and Françoise Delpech for different comments that contributed to improving this paper.