Level III of Teixoneres Cave provided 3964 leporid remains. The MNE is 1512, and the majority of elements are well represented. Of these, radii (78), humeri (77), femora (73) and calcanea (70) are best represented. A MNI of 49 were morphologically attributed to the species Oryctolagus cuniculus, considering radius, calcaneus and tibia; and just one from the genus Lepus, established by a tibia. Although the sample includes both genera, rabbits seem to be the highly represented taxon. In spite of this, we prefer to use the term leporid or lagomorph from here, as more caution is needed when classifying diaphyseal long bone fragments (without ends).

Adults seem to predominate in the assemblage (76%), while the percentages of identified juveniles (6%) and infants (18%) are lower. Following Jones’ (2006) criteria for distinguishing sex-ratio profiles in the assemblage, the data presented in a scatter plot seem to be dispersed homogeneously and it is not possible to distinguish two clear groups, implying that neither sex predominates (Fig. 2).

The % MAU was calculated, to evaluate the anatomical representation of the assemblage. This shows which elements were absent as a percentage of what one would have expected to find. According to this, the skeletal representation is biased. The best-represented elements are those from the proximal appendicular skeleton, of which the radius (79.59%) and humerus (78.57%) have the highest values. The femur and calcaneus are also well represented (74.49% and 71.43%, respectively), whereas carpal/tarsal bones and vertebrae are under-represented. Despite that, there are no significant differences in representation between the anterior and posterior limbs, while the bones of the axial and cranial skeleton are under-represented. According to Pavao and Stahl's (1999) criteria, a differential conservation index shows that the biases away from some parts of the anatomy are not due to bones being lost because of their low density (Pearson r = 0.22). The biases should therefore be caused by other biological agents unrelated to bone density (Table 1).

A high degree of fragmentation is observed; only 13.75% of the remains were recovered complete. For the long bones (stylopodials and zeugopodials) the fragmentation is even more obvious, with less than 4% of bones complete for each category. Of the fragmented bones, shaft fragments predominate for long bones (> 53% of representation in each category). Only the humeri present different tendencies, with a higher presence of distal epiphysis (51.91% of humeri fragments).

Taking the assemblage as a whole, 1974 bone fracture planes were analysed. Most of them (97.11%) were attributed to fresh breakage, presenting curved shaped fractures, oblique angles and smooth edges (Sanchis, 2010). Ten long bone shaft cylinders were also identified in the assemblage (0.3% of the assemblage), including femur (NISP = 5), tibia (NISP = 2), radius (NISP = 2) and an undetermined long bone. None of them exceed 50% of total bone length, being two femora, one tibia and one undetermined bone less than 25% of the total shaft length. All the shaft cylinders measure less than 40 mm long, and three of them have alterations caused by digestion.

A total of 640 leporid remains present modifications produced by carnivores (16.15%). Most of them (51.92% of the modified remains and 9.21% of the total remains) were caused by digestion. Of the digested bones, there is a predominance of those that have been slightly digested (85.21% of digested bones) with rounded edges, showing early stages of tissue loss. Water effect was ruled out on these bones because no homogeneous polish/rounded surfaces and micro-striations were detected (Bromage, 1984). Grade 3–4 corrosion is also present and, even at low proportions, it especially affects calcanea.

Mechanical modifications are present on 8.53% of all the remains in the assemblage. Of these, tooth-marks predominate: 3.15% are pits/punctures and 1.61% scores (Fig. 3). Notches or crenulated edges are also represented in lower proportions. Of the mechanical alterations, 37.01% are located on the diaphyses of long bones. Flat bones have also been affected, containing 31.64% of the tooth-marks in the assemblage. In this respect, the mandibles and pelvis were most affected, with around 23% of each type of bone showing mechanical modifications.

Anthropic damage is present on 2.14% of the assemblage (Table 2 and Table 3). It was mainly detected by the presence of cut-marks (0.76%), and thermo-alterations on bones (0.91%); nevertheless, other modifications, such as tooth-marks and breakage (0.48%) might be also attributed to human action because of its shallow and irregular morphology and its association with flexion fractures and crenulated edges (Landt, 2007, Sanchis, 2010 and Sanchis et al., 2011).

Cut-marks have been identified on tibia (NISP = 7) and metatarsus (NISP = 8), which together represent 50% of the remains having this type of modification (Fig. 4). Other unidentified long bones also have cut-marks on them (NISP = 2), as do the pelvis (NISP = 2), femur (NISP = 3) and scapula (NISP = 3), among others. Incisions seem to predominate, associated with skinning, disarticulation and defleshing. Skinning and defleshing are better represented (Table 2). Human breakage produced on bones by bending or tooth pressure was detected associated with notches and human tooth-marks (Landt, 2007).

With respect to burned bones (Fig. 5), these represent a significant part of those remains showing direct evidence of processing by hominins. They are in small fragments which are less than 2 cm long. Most of them (NISP = 22) are parts of long bones that could not be identified because the fragments are too small (Table 3). No burnt ends have been found in the assemblage. Grade 2 and double colorations predominate on the burned bones, being found in high proportions (on 33.33% and 25% of the burned bones, respectively). The differential burning grade is significant enough to determine anthropic consumption.

The leporid remains from level III of Teixoneres Cave seem to agree with the interpretation suggested in previous works (Rosell et al., 2010a and Rosell et al., 2010b), where it was interpreted as a carnivores’ den where hominins were occasionally active. The data presented here support the main hypothesis that both human and carnivores consumed leporid remains, forming a palimpsest. This seems to be confirmed by the presence of modifications to bone surfaces made by both of them. The possibility that leporids entered the site naturally is rejected due to the low proportion of complete bones, and also because no bones were found in anatomical connection. Additionally, no burrows were identified during the fieldwork.

To get accurate interpretations, taphonomical data obtained from level III were compared with other studies of accumulations produced by various known predators. Although previous studies proved the presence of hyena in the site, it is highly unlikely that this carnivore intervened on leporid remains. Even if it were the case, its powerful mandibles should destroy the bones and digestive corrosions must be more severe than the ones detected. M. meles was also excluded despite it being present in the site since (a) no studies exist about leporid accumulations exclusively produced by badger. Only one study is published (Maylle et al., 2008), but fox also intervened on the remains and there is no clear distinction of who produce each modification. And, (b) badgers tend to produce scores similar to gnawing and clearly distinctive, with wide grooves of variable length, flat bottoms and parallel between. On this basis, we used different criteria to distinguish between other potential predators (Table 4): •

anatomical representation;

The age of the prey has been dismissed as not being a reliable criterion for distinguishing predators, because various species do not show an exclusive behaviour pattern and there is wide intra-specific variability, conditioned by ecological circumstances (Cochard, 2004a and Sanchis, 2010).

With respect to the anatomical profile, the proportions presented for level III seem to be closer to accumulations produced by mammal carnivores (Table 4), such as fox (Cochard, 2004a, Cochard, 2004b, Hockett, 1999, Hockett and Haws, 2002, Lloveras et al., 2012b, Mondini, 2000, Sanchis, 1999, Sanchis, 2000, Sanchis, 2010 and Sanchis and Pascual, 2011) or lynx (Lloveras et al., 2008b). However, accumulations generated by hominins can also present similar proportions (Blasco, 2011, Blasco et al., 2013, Cochard, 2004a and Cochard et al., 2012) and for this reason, the skeletal representation cannot be used as the unique criterion for distinguishing predators in one assemblage.

The under-representation of cranium fragments might be explained as a result of bone density, but this is not an explanation for the bias found in vertebrae. This bias can be caused by mastication of these bones during human consumption, as Landt (2004) recorded using data from Bofi in Central Africa, but could also be due to the presence of mammal carnivores, such as foxes, who commonly eat bones while consuming prey and then defecate elsewhere (Cochard, 2004a and Sanchis, 2010). Other authors observed differences in skeletal profiles based on the functionality of the site (Charles and Jacobi, 1994 and García-Argüelles et al., 2004). For example, in localities where the skinning of lagomorphs as a regular activity, the anatomical representation consists mainly of distal appendicular elements, such as patellas and caudal vertebrae. The limited presence of acropodials in level III of Teixoneres might indicate that it is not an area where animals were skinned, but was rather focused on the consumption of carcasses. Additionally, burning should be taken into account as having potentially destroyed remains at level III of Teixoneres, and it is possible that some of the smallest leporid remains recovered from the mesh are mixed with the micro-faunal remains still in process of identification.

Regarding the fragmentation of the assemblage, data from Teixoneres level III is extremely fragmented, unlike the patterns established for raptors and closer to the values produced by lynx (Lloveras et al., 2008b), wolves (Schmitt and Juell, 1994) and humans (Cochard, 2004a and Cochard et al., 2012). Although assemblages produced by foxes normally present moderate degrees of fragmentation (Cochard, 2004a, Cochard, 2004c, Sanchis, 1999 and Sanchis, 2000); there is considerable variability (Lloveras et al., 2012b, Sanchis, 1999 and Sanchis, 2000), which is conditioned by the functionality of the place where the remains were found (Cochard, 2004b, Cochard, 2007 and Sanchis, 2000). We therefore cannot dismiss this small carnivore as a possible accumulator.

The presence of shaft cylinders has also been considered and discussed. These elements are usually associated with accumulation by humans, especially when they are present in large numbers (e.g., Allué et al., 2010, Cochard, 2004a, Cochard et al., 2012 and Sanchis, 2010). However, carnivores can also produce this morphotype, although in very low proportions (e.g., Cochard, 2004a, Hockett, 1991, Lloveras et al., 2008a, Sanchis, 2010 and Schmitt and Juell, 1994). Shaft cylinders found in natural deposits are typically short (Brugal, 2006), while they are usually larger in human context. This can be used to distinguish them. The dimensions of shaft cylinders found at level III of Teixoneres (< 40 mm), which never overate the 50% of the total bone length, together with the presence of digestive corrosion on some of these bones and their limited presence, suggest that carnivores may possibly have been the agents responsible for these modifications. However, the percentages of digested bones are low compared with those produced by mammal carnivores. Although there are some cases of strong or extreme corrosion in the assemblage (grades 3–4), the pattern fits well with the traces of digestion originated by Bubo bubo (Table 4). Nonetheless, the possibility that they were contributed by other agents, such as imperial eagle, or mammal carnivores, cannot be discarded. The percentages of bones that have been slightly digested are, however, higher than those normally produced by these nocturnal raptors (B. bubo) (Cochard, 2004a, Cochard, 2004b, Lloveras et al., 2009a, Lloveras et al., 2012a, Martínez Valle, 1996, Sanchis, 1999, Sanchis, 2000 and Sanchis, 2001). This may be due to a mixture of accumulations from various agents. It is also important to emphasize the case of calcanea, which are the element most closely associated with modification by strong corrosion. For these bones, the values are similar to those from other sites, such as the Arbreda Cave, where the accumulation was attributed to a mammal carnivore, such as lynx, fox or wildcat (Lloveras et al., 2010).

Mechanical modifications to the surfaces of bones are evidenced in similar percentages as those for foxes. Sanchis (2010) and Sanchis and Pascual (2011) carried out a study in Sitjar Baix, where mechanical modifications oscillated between 3.18 and 9.17% range. These values are like those obtained from level III of Teixoneres. Considering the data available to date, there are no other predators that produce similar values of tooth-marks on bone surfaces. Nevertheless, only a few studies have been published for other mammal carnivores and more studies should be carried out to improve this data.

In spite of the significant contribution by carnivores (mammals and raptors) to the leporid assemblage, the analysis of Teixoneres level III provides enough evidence to confirm the butchering and consumption of rabbit remains at the site by groups of humans. This is supported by the presence of: •

cut-marks on limb bones, pelvic girdles and crania;

Nevertheless, it must be taken into account that burns on bones could reflect other types of intentional actions, such as cleaning activities intended to remove waste, or could be the result of unintended processes, such as accidental cremation. It could even be a consequence of post-depositional damage, for example, thermal alteration by hearths built in the same place where bones fragments are shallowly buried. In the case of Teixoneres, no differential burning patterns have been found, so it is possible that part of the burning might be the result of non-nutritive episodes that occurred following consumption. Although the proportion of bones recorded at level III has shown direct evidence of human consumption is not high, it is important to emphasize that, because of their size, small animals do not present as many direct modifications to bone surfaces as larger ones. They can be processed using only the teeth and hands without any tools, which often makes it difficult to distinguish these modifications from those caused by other predators. Other factors must therefore be taken into account in order to establish whether humans had a significant impact on the remains or not. Nevertheless, and despite tooth-marks associated to human activities were detected on a small proportion, they have always been associated to bending and breakage. It cannot be possible to associate them with other activities, and the clearly high carnivore activity makes to dismiss the possible idea of higher role of hominids on leporid remains. To this end, the Teixoneres assemblage has been compared with other archaeological accumulations (Table 5).

The values presented for level III of Teixoneres are close to those presented at level 4 of Canalettes (Cochard, 2004a and Cochard et al., 2012), in terms of cut-marks or burned bones. The difference lies in the large number of shaft cylinders − which are associated with human consumption – that were observed at this French site (Cochard, 2004a, Cochard et al., 2012 and Rufà, 2013). The low numbers of digested bones in Canalettes is also significant when comparing it with Teixoneres, where the proportion is higher. The case of level IV of Bolomor Cave is comparable (Blasco, 2011, Blasco and Fernández Peris, 2012a, Blasco and Fernández Peris, 2012b, Blasco et al., 2013 and Sanchis, 2010). This site, like level III of Teixoneres, presents low rates of cut-marks on bones. Digested bones are only incidentally present in the Bolomor assemblage and there are a high number of thermal alterations, confirming the anthropic nature of the contribution. Level 4 of Abri du Maras can also be compared with Teixoneres level III. It presents very few lagomorph remains with butchery marks (NISP = 2) and there is no evidence of burned bones, although leporid hair found on stone tools was proposed to demonstrate that they had been processed there (Hardy et al., 2013). Other analyses carried out of use-wear on stone tools provide evidence of small animal consumption at archaeological sites where there are no cut-marks on the remains. The Middle Palaeolithic of Payre (Ardèche, France) is such a case (Hardy and Moncel, 2011). Another example to compare it with is the Mousterian assemblage of Gruta do Caldeirão, Tomar, Portugal (Lloveras et al., 2011), which also contains few bones with cut-marks and no evidence of burning. It presents a comparable percentage of digested bones to level III of Teixoneres and the origin of its accumulation seems to be similar to that of Teixoneres level III. Most of these examples show that, despite the fact that direct evidences are scarce and – in some cases – mixed up with carnivore activity, Neanderthals could access a wide spectrum of prey, including leporids (Table 5).

Although the data from Teixoneres seems to indicate a low level of human activity at the site, it exists. Comparing with other Mousterian levels in the region such as level I of Arbreda Cave, Teixoneres level III shows a higher proportion of anthropogenic activity. Contrary to Arbreda, where mammal carnivore activity clearly predominated (Lloveras et al., 2010), Teixoneres level III seems to have more assortment in its accumulation.

Data from macromammals at Teixoneres level III also suggests that both humans and carnivores used the cave and performed actions linked to their consumption sequences (Rosell et al., 2010a and Rosell et al., 2010b). In the case of hominins, the final phases of the butchery sequence took place in the cave. This has been determined from cut-marks related mainly to defleshing, although evidence of skinning and disarticulation have also been detected, as well as intentional bone breakage associated with marrow extraction. Thermal alterations are important, specifically those associated with double colorations, which suggest that bones were roasted when they still had meat adhering to them (Rosell et al., 2012).

In addition to this, the analysis of the rabbit assemblage could be linked to an occupational pattern and provide information about it. In the case of level III, it should be taken into account that leporids are just one group among a wide variety of species present at the site. This diversity may be a consequence of the hominins having better control and knowledge of the territory, which would indicate a pattern of long-term occupancy or a succession of short occupations where encounter rates play an important role (Blasco et al., 2013, Jones, 1983 and Martínez, 2009).

The age and sex-ratio profiles of leporid population proposed by Jones (2006) and applied by Cochard et al. (2012) and Blasco et al. (2013) can be of assistance in determining the procurement system, which may have been related to knowledge of environment and, indirectly, to occupational pattern. These authors assume that an assemblage with high proportions of females (and, in many cases, infant leporids) might be linked to trapping near the burrows and, consequently, a good knowledge of the landscape. In contrast, higher or equal numbers of adult males might commonly be associated with individual acquisition of prey, more closely linked to encounter rates, and occurring opportunistically. In that respect, Teixoneres patterns seem to be related to individual acquisition of prey, because no sex profile predominates (Fig. 2). This pattern is similar to that observed at levels XI and XVII of the Bolomor Cave and TD10-1 of Gran Dolina (Blasco et al., 2013) and differs from the data presented at Bolomor IV (Blasco et al., 2013) and Canalettes 4 (Cochard et al., 2012). Nevertheless, these results should be taken as an approximation, because most of the lagomorph remains recovered at level III of Teixoneres were brought there by small mammal carnivores, or probably, B. bubo.

Regardless, the high levels of carnivore activity observed not only on leporid remains but also on ungulate remains (Rosell et al., 2010a and Rosell et al., 2010b) reinforce the idea of successive short-term human occupations forming palimpsests (Sánchez-Hernández, 2013 and Sánchez-Hernández et al., 2014). This is concluded from the facts that, first, there is no evidence of interaction between hominids and carnivores and, second there are very few bones with hominid and carnivore alterations on the same surface, so any overlapping appears to have been occasional.

Eagle owl and fox activity, and possibly the activity of other mammals (such as lynx or wolf) are strongly associated with situations of alternated occupation of the cave with hominins. Additionally, the taphonomic study, linked to the occupational patterns and the presence of a high diversity of taxa that were consumed by hominins at the site – not only ungulates, but also leporids and other small prey – confirms that there was a broad spectrum of resources available for consumption by Neanderthals. This diversity of prey species also indicates the Neanderthals’ adaptability to several different landscapes and, their overall knowledge of the environment. This was previously demonstrated by Henry et al. (2011) and Hardy et al. (2012), who found evidence of the use and consumption of plants from studies of phytoliths and microfossils in dental calculus taken from Neanderthal specimens. In the case of the Spanish site of El Sidrón, this vegetable component in the diet not only had nutritional value but also medicinal properties (Hardy et al., 2012), reflecting the Neanderthals’ capabilities and their behavioral complexity.

The accumulation of leporid remains from level III of Teixoneres Cave is an example of a mixed assemblage in which different agents were active, forming a palimpsest. This study confirms that the assemblage was produced by a mix of contributions, in which carnivores played an important role. Fox (V. vulpes) and eagle owl (B. bubo) seem to have been the carnivores principally responsible for modifying the remains, as is demonstrated by modifications produced both mechanically and by digestive processes. However, other carnivores such as lynx cannot be discounted as agents that had a role in producing the assemblage.

Humans also took part in producing the assemblage, as is evidenced by cut-marks, intentional breakage to release fat/marrow, and burned bones. Human activity may possibly have been under-represented, because leporids can be processed without the use of tools. This study has proven that Neanderthals had a broad spectrum of animals available to them that they could potentially exploit. Of these, the leporids were not exceptional. Other species of ungulates were consumed at the site, and, on the basis of the encounter rates, leporids may well have been hunted. The wide variety of animals consumed by humans, together with the sex-ratio and age profiles of leporids and the presence of multiple carnivores might be linked to the idea of short-term occupational patterns at Teixoneres. These conclusions also reflect the Neanderthals’ ability to adapt to new and different environmental conditions.