Bolomor Cave is located on the southern slope of the Valldigna valley, approximately 2 km southeast of the town of Tavernes (Valencia, Spain) (Fig. 1). The site is a karstic rock shelter in the Monduver Mountain. The cave is situated approximately 100 m above sea level. The sedimentary sequence of Bolomor Cave consists mainly of allochthonous material of colluvial origin. In addition to these deposits, there are other autochthonous sediments of gravitational origin from the roof subsidence, due to tectonics or weathering processes. Seventeen geoarchaeological levels have been identified (Fumanal, 1995). Level XII consists of a clayey sediment with subangular limestone clasts. The sedimentary characteristics display a marked brecciation to the east due to the constant precipitation that produces the drip line of the cave. The karstic deposit of Bolomor Cave has been dated by Amino Acid Racemization (AR) and Thermoluminescence (TL) to between MIS 9 and MIS 5e (Fernández Peris, 2007). Levels are numbered from the top of the deposit. Several TL dates have been obtained: 121 ± 18 Ky for Level II, 225 ± 34 Ky for Level XIII and 233 ± 35 Ky for Level XIV. AR dates have also been made: 228 ± 53 Ky for Level XIII and 525 ± 125 Ky for Level XVII. Specifically, an AR date has been obtained for Level XII of approximately 180 Ky (Fig. 1).

The lithic industry recovered from Bolomor Cave is classified as a Middle Palaeolithic techno-complex. The raw materials used consist of flint, limestone and quartzite, which come from marine, colluvial and fluvial stones from the immediate area around the site and from areas further afield up to 15 km from the site (Fernández Peris, 2007). The Level XII industry consists almost exclusively of limestone flakes and sporadic flakes of flint.

The fauna from Bolomor Cave includes 20 species of macromammals. The biostratigraphic sequence is mainly characterized by the presence of Hemitragus cedrensis, Hemitragus bonali, Equus ferus, Cervus elaphus and Dama sp. and by more specific records at certain times of other species such as Megaloceros giganteus, Sus scrofa, Macaca sylvana, Equus hydruntinus, Bos primigenius, Stephanorhinus hemitoechus, Paleoloxodon antiquus, Hippopotamus amphibius and Castor fiber. The presence of carnivores in the cave is sporadic, but fossil remains of Ursus arctos, Ursus tibetanus, Canis lupus, Panthera leo, Lynx pardina, Vulpes vulpes and Meles meles have been recovered (Blasco, 2008, Blasco and Fernández Peris, 2009, Martínez Valle, 2001 and Sarrión and Fernández Peris, 2006). It is important to note also the presence of small prey such as lagomorphs, tortoises and birds throughout the entire sedimentary sequence (Blasco, 2008, Blasco and Fernández Peris, 2009 and Sanchís Serra and Fernández Peris, 2008).

Investigations carried out to date at the site have yielded remains of combustion structures at Levels II, IV, XI and XIII (Fernández Peris, 2007). For this last level, there is an AR dating on malacofaunal remains from the area close to the hearths of 228 ± 53 ky. Therefore, Bolomor Cave provides the earliest evidence for the controlled use of fire on the Iberian Peninsula.

To date, seven human bones and dental fragments have been recovered from Bolomor Cave. According to Arsuaga et al. (2001), the morphology of the human remains from Bolomor Cave is compatible with that of the European human fossils of the Middle Pleistocene.

All the archaeological remains recovered during the excavation of level XII at Bolomor Cave were three-dimensionally recovered and mapped. This allows spatial reconstructions both on horizontal and vertical axes.

The faunal analysis of level XII of the Bolomor Cave site was carried out following standard archaeozoological methods (Lyman, 1994 and Reitz and Wing, 1999) and includes all fossil material recovered until the 2005 season. Anatomical, taxonomic and modification details were recorded.

To assess completeness of the sample, Number of Identified Specimens (NISP), Minimum Number of Elements (MNE), Minimum Number of Individuals (MNI) and skeletal survival rate (Brain, 1981 and Lyman, 1994) were calculated. Skeletal survival rate estimates the proportion between the elements recovered and those expected (Brain, 1969). To determine the age at death of the animals, the best indicator in this study has been tooth replacement and dental wear (Azorit et al., 2002, Bökonyi, 1972, Mariezkurrena and Altuna, 1983 and Silver, 1969). Nevertheless, the degree of development and fusion of the epiphyses of the limb bones has also been used in some cases (Silver, 1969), as well as the type of cortical tissue (compact in adults or more porous in young animals).

Fractures and breaks on bones were analyzed and classified according to Villa and Mahieu (1991). The outline (transverse, curved/V-shaped, longitudinal), fracture angle (oblique, right, mixed) and fracture edge (smooth, jagged) were recorded. The high proportion of broken and partial bones can make specific identifications difficult. Bone fragments were included in size categories related to the estimated body weight of individuals identified in the assemblage, following the criteria established by Díez et al. (1999): large, medium, small and very small-sized animals.

Alterations observed on bone surfaces were treated at both macroscopic and microscopic level. For the microscopic study an Olympus SZ11 stereoscopic (magnification up to 110) and ESEM (FEI QUANTA 600) were used. Damage observed on the faunal remains included cutmarks, intentional bone breakage and carnivore toothmarks.

Incisions, scrapes and chopmarks have been identified at level XII (Binford, 1981, Bromage and Boyde, 1984, Potts and Shipman, 1981 and Shipman and Rose, 1983). The analysis of cutmarks took into account the number of striations, location on the anatomical element, distribution over the surface (isolated, clustered, crossed), orientation with respect to longitudinal axis of the bone (oblique, longitudinal, transverse) and delineation (straight or curved).

Surface damage caused during the bone breakage was also analyzed and the diagnostic elements of anthropic breakage were documented on faunal remains from the level XII sample: percussion pits or percussion marks (Pickering and Egeland, 2006), percussion notches, impact flakes (Capaldo and Blumenschine, 1994 and Pickering and Egeland, 2006), adhering flakes (Díez et al., 1999 and Fernández-Jalvo et al., 1999) and peeling (White, 1992).

The carnivore toothmarks identified on bone remains at level XII were mainly pits, punctures and scores (Binford, 1981, Blumenschine, 1995, Haynes, 1980, Haynes, 1983 and Stiner, 1994, inter alia). The analysis of toothmarks also took into account the number, the location and the distribution on the anatomical element.

The level XII faunal assemblage includes all fossil remains from the excavation seasons 2000-2004 and from the soundings made in 1989 and 1996. Level XII provided 2059 faunal remains (Table 1). From these, 699 have been identified to taxonomic level. The MNE is 528, mostly represented by tibiae (43), mandibles (40) and humeri (39), while basipodials and acropodials are represented by few elements (Table 2). The greatest representation of skeletal elements is due to red deer, rabbit and horse. The MNI is 48, established from the most common skeletal element according to species and age at death. Cervus elaphus is the predominant species (11), followed by Oryctolagus cuniculus (10), Equus ferus (9) and Bos primigenius (4). These animals represent 72% of the total MNI of the assemblage. All the other species are nominal in representation, and do not exceed 2 specimens in any cases.

Regarding age at death of the animals, one group is clearly predominant over the rest. Adults represent 65% of the specimens. We stress that only in Equus ferus are all age groups represented, with the exception of neonates (absent in the assemblage). The skeletal survival rate assesses the proportion between the recovered elements and those expected from the MNI. In this way, the anatomical representation obtained is valid in order to observe the elements absent in the assemblage (Brain, 1969). This rate indicates a biased skeletal representation in all the taxa in the assemblage. This selection of elements is essentially characterized by cranial elements (mainly mandibles and maxillaries) and by stylopodials and zeugopodials. On the other hand, the low proportions of autopodials and the almost total absence of the axial skeleton and phalanges are characteristic for most of the species documented at this level (Fig. 2). The elements of the greatest representation are those with the greatest nutritional and medullary value (Binford, 1981 and Emerson, 1993). As for very small-sized animals (mainly represented by O. cuniculus), the standard representation varies significantly. The axial skeleton (vertebrae and ribs) shows a greater representation than in any other size class of superior weight.

Analysis of bone breakage shows that curved/V-shaped fractures predominate (55.1%), along with oblique angles (44.9.8%), and smooth edges (75.5%). This degree of bone breakage at level XII indicates fresh bone breakage, according to Villa and Mahieu (1991). Surface damage caused during breakage of the bones has also been analyzed and diagnostic evidence has been documented: percussion pits, percussion notches, impact flakes, adhering flakes and peeling (Fig. 3) (Table 3).

Cutmarks, which are found on 12.3% of the bone remains at level XII, establish an association between hominids and other animals at the site. These were identified mainly on the long bones of medium and large-sized animals with a predominance of the incisions on the shaft and the sawing marks on the metaphysis. The action performed (viscera removal, defleshing, disarticulation and periosteum removal) was recorded according to morphology, emplacement and distribution of striations (Table 4). The consumption of small prey, specifically of Oryctolagus cuniculus and Cygnus olor has also been documented from cutmarks and anthropogenic bone breakage (Fig. 4).

However, hominids are not the only agents that play a part in the faunal assemblage. Carnivore toothmarks have been identified mainly on the proximal appendicular skeleton of medium and large-sized animals (which are the most abundant in the record). However, the incidence of activity of these carnivores is low (3% of the analyzed remains). The dimensions of the predominant toothmarks suggest small and medium-sized carnivores, such as canids (Andrews and Fernández Jalvo, 1997, Domínguez-Rodrigo and Piqueras, 2003 and Selvaggio and Wilder, 2001).

On the other hand, 29 bone remains show evidence of both human and carnivore activity on the cortical surface. From these, 6 faunal fragments allow us to establish the access sequence of both kinds of predators to animal nutrients (overlapped marks): 4 bone remains show carnivore toothmarks on cutmarks (primary human access) and 2 bone fragments describe the opposite phenomenon (secondary human access) (Blasco and Rosell, 2009).

Several types of evidence provide information about the methods used to obtain animal nutrients. Some researchers suggest that the main factor to be considered is the skeletal representation of the different taxa identified in the assemblage (Binford, 1981, Brain, 1981, Lyman, 1994 and Stiner, 1994 inter alia). This must be considered in conjunction with the age at death of the animals (Gaudzinski and Roebroeks, 2000) and the location of processing marks on the carcasses: location and purpose of the cutmarks, presence of carnivore toothmarks, overlapping marks and fractures of both anthropogenic and animal origin (Binford, 1981 and Blasco and Rosell, 2009, Bunn, 1981, Domínguez-Rodrigo and Pickering, 2003, Domínguez-Rodrigo and Barba, 2006, Marean and Kim, 1998 and Shipman and Rose, 1983; inter alia).

The level XII faunal assemblage shows a systematic proportion of skeletal elements with high nutritional value, a predominance of adult animals, cutmarks related to the viscera removal and oblique and longitudinal incisions on the diaphysis of the limb bones, which is associated with defleshing large muscle masses. All these elements suggest that the access of hominids to these animals was mainly primary and immediate. The pattern of these lines of evidence suggests the development of hunting techniques among the level XII hominids.

As already mentioned, for Gaudzinski and Roebroeks (2000) the age at death of prey animals can help determine the anthropogenic strategies for obtaining these faunal resources. According to them, adult individuals are generally healthy and strong animals and therefore could have been the main objective in hunting by human groups. At level XII, adult specimens predominate; however, in the case of the horses 44.5% are non-adult (infantile, juvenile and of advanced age). Nevertheless, the skeletal representation of these ungulates and the location and types of cutmarks suggest primary and immediate access.

On the other hand, the archaeostratigraphic distribution of the horse fossils describes a continuous vertical horizon in both north-south and east-west sections of the excavation (Fig. 5). This particular distribution is not a consequence of the slope layer or of the intervention of post-depositional agents in the assemblage: (1)

preferential inclines or orientations of the faunal remains are not observed;

According to Todd and Stanford (1992b) and Todd and Frison (1992a), a faunal refit is the reassembly of a skeletal element from several fragments recovered in the same archaeological site. For Waguespack (2002), refitting skeletal elements of the same animal and/or fragments of individual skeletal elements provides a direct method of linking skeletal segments or fragmentary elements to individual carcasses or anatomical units across an archaeological site. From this perspective, a minimum of 10 bone refits belonging to this species has been documented: 4 maxillae, 2 mandibles and 4 ulnae. This fact in addition to the single line observed in the vertical projection, indicates a certain synchrony in the deposition of horse remains. Therefore, we can distinguish at least one single event related to horses. Only in this case, and based on general characteristics of the fossils belonging to this species, we could propose the possibility of a different hunting strategy. Within the wide diversity of game techniques, perhaps the most appropriate to explain this phenomenon is those based on multiple predation. According to Steele and Baker (1993), multiple predation consists of the capture of various individuals of a herd (regardless of their age) in a single hunting event. Theoretically, this type of hunting would be reflected, at the archaeological sites that were used as a place of habitat, in: (i) a high diversity of age at death of the species captured by this technique (if these animals were hunted in different events, they would show the same age pattern as a consequence of anthropogenic selection of the individuals of the herd: adults only); (ii) a predominance of anatomic elements with high nutritional content; (iii) incisions located mainly on the shafts of the limb bones; (iv) at a given archaeostratigraphic level, a single line of fossils belonging to the same species; (v) presence of refits among the faunal remains at this archaeostratigraphic line. From this perspective, all the characteristics documented from the horse remains at level XII could correspond to a hunting strategy based on multiple predation. An important fact to consider is the behaviour of wild horses and the location of the site itself. We must bear in mind that horses are grouped in large herds, which have a gregarious and seasonal character (Nowak, 1991). Their behaviour, in addition to the high view above the plain offered by Bolomor Cave, made locating herds easier and could have favoured the development of these practices. Few archaeological examples of multiple predation are documented in early chronologies. Nevertheless, cooperative hunting is suggested at the Lower Palaeolithic site of Qesem Cave in Israel (Stiner et al., 2001). Another example of social implication in hunting could be the human use of natural traps to capture animals. At the Coudoulous I site (MIS 8-6) some researchers raise the hypothesis of a possible human intervention in the ungulates’ fall into natural traps, specifically bison (Bison priscus) (Brugal and Jaubert, 1991 and Brugal et al., 2006). For these authors, this fact probably provides the basis to use these karstic sinkholes for the development of communal hunting strategies. Nevertheless, these cases are more frequent later in human history. Some examples are recorded at the South African site of Namaqualand in the Later Stone Age (Dewar et al., 2006) and at the Neolithic site of Tell Abu Hureyra in Syria (Legge and Rowley-Conwy, 1987). Nevertheless, we have to take into account that in the case of Bolomor, multiple predation may not have been used on all the individuals of this species; the possibility that some adult horses were hunted individually cannot be excluded.

Despite this predominance of primary and immediate access at the level XII assemblage, secondary access has also been documented. There are anatomical elements that characterize the skeletal parts left by carnivores or other hominids after primary and immediate access (Binford, 1981, Brain, 1981, Lyman, 1994 and Stiner, 1994; inter alia): vertebrae, ribs and pelves of large and medium-sized animals have also been identified. Although these elements are not abundant, they are present in the fossil record. Occasionally, cutmarks on the dorsal surfaces of ribs in the form of prominent sawing and deep incisions are also observed. This evidence is recorded on large and medium-sized animals and represents 2% of the total analyzed remains. In addition, overlapped marks made by carnivores and hominids are also documented. Specifically, two of these overlapping marks coincide with those described at an experimental level in the cases of cutmarks on carnivore toothmarks (Blasco and Rosell, 2009): (1)

variations in the delineation, thickness and depth of incisions in the area closest to the puncture;

Furthermore, the cortical tissue located on the puncture base is not altered by carnivore action in the case shown in Fig. 6a. If the incisions preceded the toothmarks, these would have been conserved on the puncture base. Therefore, these overlapping marks suggest a sporadic secondary access to the carcasses by hominids. Nevertheless, the main type of access used by the hominids to obtain animal resources was mainly primary and immediate, although, occasionally, scavenging might be another option to obtain nutrients.

On the other hand, the human consumption of small prey is also documented at level XII in the form of cutmarks and anthropogenic bone breakage of birds (Cygnus olor) and lagomorphs (Oryctolagus cuniculus). The use of small prey for food is common throughout the entire stratigraphic sequence of Bolomor Cave (Blasco, 2008, Blasco and Fernández Peris, 2009 and Sanchís Serra and Fernández Peris, 2008). Stiner (2001) distinguishes among slow-moving, easily collected types (mostly tortoises and shellfish), fast-running mammals (mostly lagomorphs), and quick-flying animals (birds). The hunting of fast-running and quick-flying small prey involves complex trapping methods and a sophisticated technology. On the other hand, slow-moving small animals can essentially be captured with limited technology. Isaac and Crader (1981) consider that acquiring small prey is closer to gathering than hunting. In any case, it is evident that their capture requires techniques different from those used for large and medium ungulates.

From this perspective, several strategies to obtain animal resources are documented at level XII of Bolomor Cave. This diversity is common in current hunter-gatherer groups, such as Hadza (Bunn et al., 1988, O’Connell et al., 1988a and O’Connell et al., 1988b), Kua (Bartram and Marean, 1999) or Efe pygmies of the Ituri Forest (Coon, 1971). These groups are able to practice strategies of social hunting and to take advantage of carcasses left by large carnivores in early stages of consumption. From this perspective, it is not rare that the complex hunting suggested at Bolomor Cave, the cooperative hunting identified in Qesem Cave (Stiner et al., 2009) and the specialized hunting documented at several European Middle Palaeolithic sites, such as Mauran in France (Farizy et al., 1994), Biache-Saint-Vaast in Belgium (Auguste, 1995), Salzgitter Lebenstedt in Germany (Gaudzinski and Roebroeks, 2000) or Chokurcha, Buran Kaya, Starocelié and Kabazi II in Crimea (Patou-Mathis, 2006) coexist with episodes of opportunistic scavenging or with small game. In general, this diversity can be related to the highly adaptive subsistence strategies of these hominids, who could take advantage of the benefits offered by a diverse and rich environment, which might allow the development of diverse strategies to obtain food resources.

The zooarchaeological study of level XII of Bolomor Cave has allowed us to identify the coexistence of different anthropogenic strategies to obtain faunal resources in the late Middle Pleistocene, specifically during MIS 6. In the case of horse remains, the combination of the skeletal representation, their ages at death, the location and purpose of cutmarks, the archaeostratigraphic distribution of their fossils within the assemblage and bone refits might suggest the practice of complex techniques of hunting such as multiple predation. In this sense, the location of the cave at a high point above the plain and the gregarious character of horses could favour the development of this type of strategy.

On the other hand, individual hunting seems to have been the principal technique to obtain the other ungulates identified in the assemblage, mainly red deer. In this case, the same age pattern (mainly adults) is documented as a result of an anthropogenic selection of the individuals. Also, the human consumption of small prey, specifically Oryctolagus cuniculus and Cygnus olor, is identified in the assemblage. The capture of these animals implies the development of other techniques different from those used to acquire large and medium ungulates.

From this perspective, these different practices of hunting coexist at level XII with sporadic events of secondary access to the carcasses (scavenging). Although not abundant, some anatomical elements that characterize the skeletal parts left by carnivores after primary and immediate access, cutmarks on the dorsal surfaces of ribs in the form of prominent sawing and deep incisions and overlapping of cutmarks on carnivore toothmarks are documented.

In general, a variety of strategies to obtain faunal resources is documented at the Bolomor Cave site. For us, this fact might suggest an important human capability not only for improvisation (sporadic scavenging), but also for organization (complex hunting) during the late Middle Pleistocene. The development of all these techniques at the same site might show a sophisticated human knowledge of animal behaviour and the features of the territory during the MIS 6.