We thank the Ministère de la Culture of France, the Service Régional de l’Archéologie (DRAC Poitou-Charentes), the Conseil Général de la Charente, the Conseil Municipal de Marillac-le-Franc and its residents, Princeton University, the CNRS, University of Bordeaux, UMR5199 PACEA, P. Guyomarc’h, E. Schostmans, D. Cochard, E. Discamps, B. Gravina for discussions and bibliographic help, F. Sémah and D. Grimaud-Hervé for access to the Arcy-sur-Cure material curated at the Institut de Paléontologie humaine (Paris). This work has been supported by the Research Program of the Aquitaine Region : Transitions, d’une société à l’autre processus d’adaptation et de peuplements, convention n° 20051403003AB and by the Scientific Project : NéMo of the LabEx Sciences archéologiques de Bordeaux, support provided by the ANR - n° ANR-10-LABX-52 / LaScArBx, the Research Program of the Nouvelle Aquitaine Region : Neanderthalenses Aquitanensis : Territoires, Chronologie, Humanité convention n° 2016-1R40204-00007349-00007350. We also acknowledge the editorial board of Paleo and the reviewers for their very helpful comments that improved the manuscript.

1 - Introduction

1Hominin remains are uncommon in the fossil and archaeological record ; each find represents another piece of evidence in the reconstruction of extinct patterns of biology and behavior. Without the contribution of newly developed bio-chemical and paleogenetic techniques (Buckley and Collins 2011 ; Van Door et al. 2011 ; Welker et al. 2016), it is oftentimes difficult to recognize fragmentary hominin remains, especially small ones from long bone shafts, amongst a sample of animal bones. Teeth, with their distinctive morphology, are usually easily identified. This, however, is not always the case. Two teeth from the site of Marillac had been modified in such a way that their identification as human was extremely difficult.

2Marillac is located in the village of Marillac-le-Franc, Department of the Charente, in the southwest of France, about 20 km east of Angoulême (fig. 1). The site is on the right bank of a very small stream, La Ligonne, a right tributary of the Tardoire River. Although casual activities at the site early in the last century had resulted in the discovery of a Neandertal mandibular fragment (David 1935 ; Piveteau 1957), systematic archaeological excavations at Marillac were not undertaken until 1965-1980 (Vandermeersch 1976a). More recent work at the site was carried out between 2001-2013 (Maureille et al. 2004a, 2010). Since the 1930s, 87 hominin fossils (corresponding to 109 field fragments), have been discovered, most of them cranial and dental remains plus a few post-cranial pieces. All these hominin specimens possess distinctive morphological attributes of Neandertals and many of them have been published or are available for the scientific community (Piveteau 1951, 1957 ; Vandermeersch 1976a, 1976b ; Mann et al. 2002 ; Maureille et al. 2004a, 2004b ; Mussini 2011 ; Mussini et al. 2011, 2012 ; Garralda et al. 2014a and b).

3Re-examination of the 1965-1980 collection of faunal remains resulted in the discovery of a few hominin fragments, including two anterior teeth, from the same facies, which were previously identified as deciduous incisors of an herbivore, either a bovid or a cervid (fig. 2). The existence of these teeth, with their ambiguous morphology modified in such a way as to make their recognition, as hominin, extremely difficult, has prompted the presentation of this report.

2 - Site context

4The site has a complex geological and sedimentological history. Presently it is a large rectangular depression, some 12 meters long, 9 meters wide and about 3 to 9 meters deep representing the remains of a Bajocian limestone collapsed karstic cave. Figure 3 presents the eastern site locus from which the majority of hominin fossils have been recovered, including the two teeth presented in this article. At the base of the profile, sedimentological facies 2a to 4b (which are clast-supported limestone deposits associated to a clayey matrix ; fig. 3) contain the most important archaeological remains, including lithics of the Quina Mousterian techno-complex (Meignen and Vandermeersch, 1986 ; Bourguignon, 1997). Reindeer largely dominate the associated ungulate fauna (NISP close to 95 %). A small number of carnivore bones of cave hyena, cave lion, wolf and fox have also been recovered from these levels (Costamagno et al. 2005). Some complete coprolites, gnaw marks on herbivore bones as well as some regurgitated bone fragments, reindeer petrous bones and cervid or hyena teeth in the base of facies 2a and facies 4b document some active involvement of carnivores in either primary accumulation or in secondary scavenging related to hominin activities. In much of facies 2a and facies 2b, small pieces of coprolites and a few partially digested bone fragments (no more than 37 numbered fragments from a total sample of 3423 ; this does not include the numerous very small bone/ tooth fragments from this facies) also attest to the presence of carnivores, most probably cave hyena rather other Pleistocene carnivores (Mann et al. 2005).

5The detailed chronological results of the site are presented in Frouin et al. (submitted this volume) which demonstrate that the age of the Neanderthal fossils is around ~55 ka, and dates the Quina occupations between the end of Marine Isotope Stage (MIS) 4 and into the MIS 3 (see also Royer 2013 ; Royer et al. 2013 ; Frouin 2014).

3 - Description of the two teeth (fig. 4a and 4b)

6They are an upper permanent central incisor, perhaps right side (M72 c.10 E11 # 336) and an upper permanent lateral right incisor (M70 c.10 F12 # 54), both discovered in layer 10 (equivalent to a part of facies 2a). Both possess marked occlusal attrition, stage 4 for the central incisor and 3 for the lateral one following the analysis of dental attrition published by Murphy in Smith (1984). Given the differences in occlusal wear on the two teeth, and the observed differences in wear between incisors in any Neandertal jaw, it is not possible to state with certainty that they derive from the same individual. A detailed description of the morphological traits, crown and root characteristics and diameters of these teeth as well as micro CT-scans and 3D dental tissue proportions will be presented by Garralda et al. (in preparation).

7These teeth present numerous unusual features (figure 4a and 4b). Techniques such as scanning electronic microscopy and CT scanning provide a large corpus of useful information but the amassing of this data it is not the main aim of this article. Here the focus is essentially on their gross morphology and the external aspects of the enamel and dentine ; the goal is to provide the essential macroscopic features to identify partially digested human teeth. The unusual features of the two Marillac incisors are :

1. The relative proportions between crown and root diameters, specifically for the central incisor, are not those of a permanent tooth. They are more appropriate in a deciduous tooth with an almost completely calcified root ;

2. The roots of both teeth are very flat mesio-distally and bucco-lingually. This morphology is more accentuated on the central than on the lateral incisor and is unusual for a permanent tooth ;

3. The two root apices are open and very large for teeth with such marked attrition. An almost completely calcified permanent (or deciduous) incisor does not possess an open apex of this size. Note that such a size is also inappropriate for a partially resorbed deciduous incisor root ;

4. Atypical too are the polished root surfaces lacking cementum ;

5. The dentine exposures on the occlusal surface of both teeth have a polished patina. In contrast, the borders of the enamel appears more irregular ;

6. Finally and more intriguing, the apical limits of the enamel of both incisors (figure 4a and 4b : open arrow) are not located at their normal positions, i.e. the anatomical cervix (fig. 4a and 4b : solid arrow). They terminate well away from them. Note also that the central incisor possesses a very irregular cervical border, while the border on the lateral incisor, though smooth and regular, has an abnormal convexity.

8Although this unusual suite of features is not present to the same degree of development/extension on the two incisors and does not have the same modifying impact on their normal morphology, their overall modifications are similar. This indicates they were subject to the same taphonomic processes.

9Significant research has been undertaken to more fully understand the various modifications observed on the faunal and human remains from archaeological sites (i.e. Horwitz and Smith 1988 ; Fisher 1995 ; Brugal et al. 1997 ; Selvaggio 2001 ; Egeland et al. 2004 ; Domínguez-Rodrigo and Piqueras 2003). Taphonomic modifications include abrasion, weathering, the action of vegetation, animals and soil acidity (for example, Lyman 1994 ; Hedges and Millard 1995 ; Nicholson 1998 ; Nielsen-Marsh and Hedges 2000 ; Pickering et al. 2004 ; Fernández-Jalvo and Andrews 2016). Because the three hard dental tissues (enamel, dentin and cementum) composing a tooth are characterized by distinct compositions and structures (enamel is the most mineralized of the tissues, dentin and cementum less so ; i.e. Hillson 2014), some of these processes would differentially act on a tooth.

4 - Potential pathology

10A literature review (Hillson 1996, 2014 ; Piette and Goldberg 2001) failed to discover any pathological conditions that could explain such unusual crown and root morphology on human permanent incisors or any other tooth type.

5 - Soil acidity (fig. 5)

11Soil acidity is one possible agent that might produce modifications on teeth. Enamel is the most mineralized hard tissue in the body, with its mineral structure and crystallinity organized to resist to corrosive effects of dietary and digestive acids (Dawes 2003 ; Pasteris et al. 2008). However, to assess the possibility that soil acidity might have been responsible for the modifications observed on the Neandertal incisors, modern deciduous and permanent bovid isolated molars were placed in a 10 % formic acid solution (PH = 2.91). This obviously is not a precise measure of the soil acidity at Marillac, and while formic acid at this concentration is considered an irritant and not corrosive, it nevertheless represents a reasonable way to observe the reaction of the various parts of the teeth to the action of an acid. Figure 5 illustrates the teeth before immersion in the formic acid and after 22 hours in this acid solution. Note the decrease of the mesio-distal diameter of the crown : 5.8 % for the deciduous tooth, 3.3 % for the permanent molar as well as the more limited loss of tissue at the cervix : 0.74 % for the deciduous tooth, 0.66 % for the permanent molar.

12The acid treatment produced obvious changes on the morphology of the teeth : the heavily mineralized enamel crowns have been markedly transformed by immersion in the formic acid solution while the root, composed of less mineralized dentine, shows little change. The acid worked primarily on the mineralized portion of the dental tissues. The reduction in the size of the crowns is greater than in the roots. These changes to root and crown sizes are exactly opposite from the modifications observed on the Neandertal incisors from Marillac ; that is to say, those teeth showed greater loss of root dentine tissue and less on the more heavily mineralized crown.

13Many isolated faunal and human teeth are known from Marillac (and from many Upper Pleistocene sites from the south-west of France) and virtually all of them possess normal morphology. Thus, it is extremely unlikely that soil acidity, never very important within the high pH limestone karstic caves, could be the main factor that produced these unusual dental modifications.

6 - Partial digestion by carnivores (fig. 6)

14Research (Sutcliffe 1970 ; Binford 1981 ; Andrews 1990, 1995 ; Cruz-Uribe 1991 ; d’Errico and Villa 1997 ; Dauphin et al. 2003 ; Pickering et al. 2004 ; Backwell et al. 2009 ; Camarós et al. 2017) provides evidence for another taphonomic process that results in modifications to hard tissues : the action of digestive enzymes and acids during consumption by a variety of animals including birds (eagles, vultures), reptiles (crocodiles) and mammalian carnivores (lion, wolf, hyena). The combination of digestive acids, enzymes and time (Denys et al. 1995) generally leads to the destruction to both mineral and organic components of the tooth (Fernández-Jalvo and Andrews 2016) ; the soft tissues (the organic material) will disappear first (with some variability ; Pickering 2001).

15To assess the potential that digestive processes might have on dental tissues, the two incisors were compared to a sample of partially digested faunal teeth directly associated with hyena occupations (i. e. Villa et al. 2004 ; Fourvel 2012 ; Fourvel et al. 2015) from the French Pleistocene sites of Les Plumettes (Airvaux 1987) and les Rochers-de-Villeneuve (Beauval et al. 2005). The major modifications on these faunal teeth are a differential erosion of dentine and enamel, especially at the cervix, a reduction in the diameter and length of the root, and a significant widening of the root apex. This is the result of both digestive enzymes, like pepsin, which break down the collagen component of the less mineralized dentin and cementum and digestive acids which decalcify the mineral portion of both enamel and dentin.

16It is clear from this examination of teeth from Pleistocene hyena dens that the morphological modifications produced by their partial digestion are the same as those observed on both Neandertal teeth from Marillac. Figure 6 illustrates three bovid incisors from Marillac with the same modifications, at different stages of the process, produced by carnivore partial digestion.

7 - Discussion

17Archaeological analysis of the remains from facies 2a and 2b of Marillac strongly suggests that this site served repetitively over time as a hunting camp. Many reindeer carcasses, as well as those of several horses and bisons, were butchered and prepared for transport and storage to another location for consumption (Costamagno et al. 2006 ; Meignen et al. 2010).

18There is also evidence at Marillac of human manipulation of dead Neandertals (Vandermeersch 1971, 1976a ; Le Mort 1987, 1988 ; Garralda et al. 2005 ; Garralda 2008 ; Mussini 2011). More than 35 % of the Marillac Neandertal cranial and infracranial pieces (both mature and juvenile individuals) present unquestionable cut-marks (indicative of scoring with a stone tool), percussion pits and incipient fracture lines. Internal vault releases (White 1992) are also identifiable on different skull fragments. These observations document butchery practices on these humans. The presence of butchered Neandertal bones may indicate cannibalism at Marillac (Maureille et al. 2010 ; Mussini 2011), or some other cultural practice that did not involve consumption (Garralda and Vandermeersch 2000 ; Garralda et al. 2005 ; Garralda 2008 ; Pettitt 2011).

19The most parsimonious explanation to account for the partial digestion of human teeth is that faunal elements left after butchery, as well as hominin remains, were abandoned and subsequently scavenged by carnivores.

20Since the initial identification of partially digested incisors at Marillac (Costamagno et al. 2003), five other teeth at the same site, a left lower second permanent molar (M71 c.9/10), a first left premolar (M72 c.11 I10 # 22), a first left lower deciduous molar (M73 c.10 E10), a germ of a lateral left upper incisor (M75 c.11 H9 # 3) have also been identified. Descriptions, as well as 3D and virtual reconstructions of the Marillac teeth are being prepared for publication (Garralda et al., in preparation).

8 - Conclusion

21The examination of some difficult to interpret morphological features on two isolated teeth from the Marillac site that had been previously identified as those of a bovid or a cervid deciduous teeth has resulted in their identification as partially digested Neandertal maxillary incisors.

22This contribution gives also macroscopic key features to identify such peculiar taphonomic modifications on isolated human teeth and which could help to find, within faunal collection samples, new human fossils. The description of the complete set of regurgitated Neandertal teeth from Marillac will perhaps also help to discuss the impact of the partial digestion by carnivores of partial human maxillas or mandibles.

23An implication of this research is that more attention ought to be directed to faunal samples from sites where large predators may have been involved in their accumulation. Hyena dens were a frequent feature in Western Europe at the end of Mousterian times (Discamps 2011 ; Discamps 2014). For some of them, for example Rochelot (Tournepiche and Couture 1999), Les Rochers-de-Villeneuve (Beauval et al. 2005), partially gnawed hominin bones (humerus, femur) and isolated teeth have been discovered. It is likely that additional examples of regurgitated hominin teeth are present in the faunal collections from similar sites, but may not have been considered human, identified instead as deciduous teeth of a medium or large mammal. Bovid and cervid anterior teeth with notable occlusal attrition but with an open (and large) root apex can be more carefully examined for a discrepancy between the crown’s apical border and the tooth cervix, the partial absence of enamel, an abnormal enamel thickness or the asymmetry between the breadth of the crown and the root. A polished luster over the entire tooth and a relative lack of cementum are also useful signs.

24A literature review has resulted in the discovery of similar modifications on hominin posterior teeth from the Upper Pleistocene sites of Grotte du Bison (Yonne, figure 3 in Tillier et al. 2013a) and Grotte des Hyènes at Arcy-sur-Cure (Yonne, Auy H D6 Iva, unpublished, pers. obs.). Moreover, a left deciduous central incisor of Paranthropus robustus (SWT3-2433) from Member 3 of Swartkrans (South Africa ; figure 2 in Pickering et al. 2016) also appears to possess the same modifications. Note that the authors of these articles provide crown diameters of the teeth, but apparently do not take into consideration the possible reduction of their crown dimensions. However, such a taphonomic process has a real influence on the metrical dimensions of the crown and root. This was emphasized by Tillier et al. (2013b) for an upper permanent third molar which may very well have been partially digested. It is worth noting that all these teeth were identified as hominin, indicating it is essentially incisors and canines where carnivore digestive processes have resulted in changes in crown and root morphology that make recognition as hominin difficult and confusing.