or within the cave. Although friction against the sediment (i.e. trampling) may produce rounding, this occurs in dry conditions and does not smooth the bone edges to the extent seen in these specimens. Trampling may also be discarded because trampling marks are not more abundant on these highly rounded and smoothed edges than on the rest of the bone surfaces (Fig. 10.9b). We have observed, in fact, that earlier modifications, such as tooth mark grooves, have also been smoothed or rounded (white arrows in Fig. 10.9a, c), and so the chewing predated the rounding. Trampling, on the other hand, occurred after the rounding and produced scratches on the rounded surfaces (Fig. 10.9d) and breakage (Fig. 10.9a). Two factors emerge: firstly, examination of the rounded surfaces under high magnification microscopy showed that some areas of these highly rounded bones have a peculiar pattern of parallel microscopic striations (different from microstriations produced by friction against sediment, Fernández-Jalvo and Andrews 2003); and secondly, many of the highly rounded fossils are too large to be ingested. Our interpretation is that these fragments were abraded by oral enzymes and tongue abrasion, and that the rounding could have been the result of bear licking.

Post-Depositional Damage

The most characteristic trait of the Azokh 1 fossil assemblage is breakage, affecting 1619 fossils (86.2% of the fossils from Azokh 1). Breakage has hampered higher taxonomic (even anatomical) identifications and increased the numbers of small fragments (Table 10.1). However, biological agents such as humans or carnivores did not cause this high breakage rate. Neither of these agents has produced a high impact on the fossil assemblages of Azokh 1, as indicated by little evidence of breakage by humans (see Tables 10.6, 10.7 and 10.8) (6.7%) or by carnivores (1.6%). Fire may also contribute to bone breakage (Shipman et al. 1984; Stiner et al. 1995; Mayne 1997; Cáceres et al. 2002), but fire is infrequent at Azokh 1 (4.6%), although all burnt fossils are broken. Trampling has the highest incidence at Azokh 1, although marks on surfaces of fossils have only affected 18.1% of the fossils and 21.1% of broken fossils (Table 10.8).

Post-depositional environmental conditions indicated by mineral deposits (e.g., manganese) and surface modifications (warped-up cracking) on these fossils suggest a damp environment (see below). This is in agreement with the low number of fossils gnawed by rodents in Azokh 1 (NR16, 0.85%). Porcupines chew hard surfaces, such as plastics

(Kibii 2009) and bones when they are dry and have lost the greasy periosteum (Brain 1981), and they do this to wear out their ever-growing teeth or to gain minerals from the bone in the same way that herbivores do, but when bone is wet it does not produce the same effect. The location of these marks has no relation to muscle insertions (Rabinovich and Horwitz 1994; Klippel and Synstelien (2007).

Damp environments, high relative humidity and mild temperatures generally are characteristic of inner parts of caves, and these conditions persist today in Azokh 1. Measurements in summer of temperature and relative humidity inside and outside the cave gave a high relative humidity (average 80% compared with 32% outside), but the temperature does not vary, especially sediment temperatures (constantly 19 °C). High humidity may increase the effects of trampling, leaving bones more susceptible to breakage, but we have not investigated this yet. Traits of breakage following Villa and Mahieu’s (1991) methodology show high percentages of curved-Vshape outline and smooth edges (Fig. 10.6). These two traits would suggest breakage when bones were fresh (as shown by Villa and Mahieu 1991, for Fontbrégoua site). However, Azokh 1 fossils have a higher abundance of mixed angles on the broken edges that does not fit with Fontbrégoua, Bezouce or Sarrians sites. Broken fragments that have trampling marks have also been analyzed with the same methodology showing high predominance of mixed angles, as well as, curved and smooth edges.

The best explanation for the high breakage recorded in Azokh 1 is tentatively taken to be the combination of high humidity and trampling by large sized animals (Ursus spelaeus).

Preand Post-Burial Environmental

Conditions of Azokh 1 Fossils

Bacterial attack may also increase the fragility of bones. Forensic studies (Bell et al. 1996) indicate that removal of soft parts of the body by scavenging or predation reduces indigenous bacterial attack, stopping or slowing down the dispersal of bacteria through the vascular network. Histological analyses of fossils from Azokh 1 have showed either a very intense bacterial attack (Unit II), or, more frequently, reduced/absent microbial activity. The bacterial attack observed in Unit II (OHI = 0, no original features identifi- able other than Haversian canals) has a characteristic organization pattern around histological features (Fig. 10.11a),

which is typical of decay processes. This would suggest that, at least in Unit II, there were intact carcasses that were not eaten by carnivores or butchered by humans. Large bones were found almost complete in some areas (e.g., bear femora), but they were not articulated with the rest of the skeleton but found isolated, often close to cave walls, together with stone tools. Bacteria are absent in fossils from Units I, Vu or Vm, whether as a result of consumption of the carcasses by carnivores or humans, or by the action of fire (in Unit I). On the other hand, Units III, and Vm show evidence of corrosion by fungi.

Some indications about the cave environment may be discerned through histological modifications, although the histological studies have been carried out in a small number of fossils. Successive generations of bacterial colonies have been distinguished in some fossil bones from Unit II that provide some evidence of change in environment in Azokh 1. This succession of bacterial generations has been observed in modern bones monitored in Neuadd (UK) resting in a seasonal river (Fernández-Jalvo et al. 2010a). In the case of Azokh today, there are important variations in the water rate of the cave, becoming extremely humid in summer after the rains and very dry in winter (Dom- ínguez-Alonso, personal communication). Another type of histological damage observed in fossils from Units I, II and Vm affect the canaliculi (histological channels of connection between osteocytes/lacunae). Several fossils from Azokh 1 show enlarged canaliculi, an alteration that has been described by Jans (2005) in a medieval settlement in Moorend farm (UK). Similarly, enlarged canaliculi have also been found in modern bones monitored in open air environments (Fernández-Jalvo et al. 2010a) resting in highly acid soils (pH < 6) under constant high humidity and extensive vegetation (moss and algae). These previous cases of what has now been also observed in Azokh 1 suggest that the histological damage of enlarged canaliculli is related to acidic fluids that penetrate the cortical bone, dissolving the walls of the canaliculi.

In the case of Azokh 1, the site has an acidic environment through the combination of the urea from the bat guano, the high relative humidity in the cave and the damp ground. Fossils from Unit II (central area of excavation) are heavily corroded (flaky and heavily cracked-laminated texture) that affects the entire compact bone. The sediment also has a characteristic crumbly texture and grayish color. Fossils (flaky) and sediment (crumbly) share the presence of tinsleyite in their composition (see Table 10.10). The formation of tinsleyite has been related to the presence of bat guano (Magela da Costa and Rúbia Ribeiro 2001; Marincea et al. 2002; White and Culver 2012). The formation of a wide

variety of minerals of the apatite group is frequent in caves and related to urea and bat guano (White and Culver 2012). The presence of brushite has also been mentioned as neoformed mineral by bone decay (Molleson 1990). According to this author, the hydroxylapatite (bone mineral component) is unstable in conditions of high acidity and transform into brushite during decay. The effect of this transformation is very destructive, for brushite occupies a space much larger than the crystals of hydroxylapatite, so when this transformation occurs, the original molecular structure suffers a physical destruction (Molleson 1990). The formation of barite [Ba(SO4)] is also common in conditions of high acidity and associated with microbial activity. Its formation is less destructive to bone because barite fills histological empty spaces, or secondary porosity.

A common feature of fossils from Azokh 1, Azokh 2 and Azokh 5 is the absence of collagen (Smith et al. 2016). This absence has been observed even in bone remains from Holocene periods in Unit I, which did not contain enough carbon to be dated by 14C (Ditchfield, personal communication, see Appendix, radiocarbon). The destruction of collagen at Azokh 1 (Smith et al. 2016) occurs according to a model described by Smith et al. (2002), characterized by a high bone crystallinity, almost no histological damage, but sudden removal of collagen in a very short time. The cause of such destruction of collagen is not known (Smith et al. 2002), but in the case of Azokh it also seems to affect DNA preservation, or lack of it. Indeed, as mentioned by Bennett et al. (2016), attempts to PCR amplify DNA from many fossils and sediment samples from all units of the sites Azokh 1, 2 and 5 have failed. The common element in all these areas of the cave system is the presence of guano and, at least in the case of Azokh, the explanation for this widespread loss of collagen appears to be due to this agent.

The distribution of areas of alteration in the center of the Unit II is seen today in the interior of the cave inhabited by bats. This distribution suggests that populations of bats occupied areas closer to the entrance of the cave, and conditions would have resembled those we find today in the interior of the gallery where bats live permanently, with thick accumulations of guano may reach up to 3 m thick. Water filtering through cracks of the cave carried highly acidic fluids, rich in phosphates, sulfates and carbonates dissolved from the guano, through the sediments. Depending on the conditions of the cave (moments of aridity vs. increase in humidity), different sets of minerals would be formed, some of them very aggressive to the hydroxylapatite mineral component of the bone. Brushite, for instance, is one of the most common cave minerals in guano deposits, formed at low pH by reaction of phosphate-rich solutions

with calcite, clay and bone. Its formation affects the bone structure, and we have also seen that tinsleyite, and probably sepiolite, may also destroy the bone structure. High humidity and guano has produced the conditions for the high corrosion observed in these fossils, especially at Unit II.

The duration of these sealed conditions that led to the occupation of populations of bats in this part of the cave cannot be established, but it was long enough to allow fluid percolation, mineralization and corrosion took place, reaching several meters deep through the sediment. Evidence of corrosive fluids percolation is still recorded in the section on top of Unit II (Fig. 10.8b). The time gap between Unit II (Pleistocene *100 kyr) and Unit I (Holocene) and the laminar sedimentation observed on the limit between these two units (Fig. 10.8c) suggest the involvement of water erosion that ended by flooding the substrate. It is likely that an episode of heavy rains and flash floods, removed the upper part of the Pleistocene sediments by erosion. This event opened the cave again to the outside environment, and allowed the entry and cave occupation by animals and humans during Holocene.

Human occupations of caves are usually more frequent near cave entrances, and bears are more frequent at the back of the cave where they may hibernate. Sediments closer to the entrance had already been excavated, however, and no information was available to us. On the other hand, the back area of the cave is a more suitable area for bear hibernation, and bear fossils are more abundant in this part of the cave than at the cave entrance and were exposed to the damp acidic conditions. This situation suggests to us that the actual butchery processes were concentrated on bears, while other large mammal carcasses could also be butchered, but not in this part of the cave.

This opens an interesting discussion about the behavior and type of occupation of these extinct cave bears (Ursus spelaeus), whether for example they lived more permanently in the cave than only during hibernation. Probably, bears were not in the cave during summer, as the cave becomes wetter during this period after the rains. Humans could then shelter in the cave at the entrance in summer time and eventually penetrate into the cave interior. Another question is the capacity of bears to scavenge other bear remains or at least to chew bones. Chewing marks on the fossils of Azokh 1 and breakage by carnivore action are scarce, and there is no taphonomic evidence of hyenas taking any part of the bone accumulations. The maximum size of puncture marks on diaphyses/compact bone in two units of Azokh 1 exceeds any recorded chewing by larger carnivores, such as lions (Pinto et al. 2005; Domínguez-Rodrigo and Piqueras 2003;

Pobiner 2008). These are Unit II and Unit Vu, and in the absence of a better candidate, we propose that cave bears have damaged and chewed other bones during deposition of these two units. Unit I, on the other hand, lacks large chewing marks (<4 mm) and have a pattern suggesting scavenging by a small carnivore such as a fox or jackal. Unit Vm fossils have tooth marks larger than 4 mm, but with a similar pattern to those in Unit I, and the pattern in Unit III is close to the multispecies site of Arrikrutz (see Fig. 10.12). These patterns would suggest that together with cave bears, smaller carnivores were scavenging.

Another possibility is that bears carried bones from previous human occupations at the entrance, where abandoned bone remains were lying on the floor, into the rear of the cave where the bears were living. Comparing the rear of the cave environment in the past to what it is like today suggests that the back of the cave had almost permanent high relative humidity and seasonally (summer-autumn) damp substrate. Seasonal dampness in the cave is indicated by the formation of cave minerals characteristic of both wet and dry caves, modifications observed on the surface of the fossils (cracks by humidity) and successive stages of bacterial attack. In this context, trampling under wet conditions could greatly increase breakage. If the cave was dry in winter, as seen today, the intense trampling observed on bones (as well as on lithics, see Asryan et al. 2016), and the resulting high breakage rate, suggests that bears were occupying the cave for longer periods than just while hibernating, extending the occupation to autumn and spring. With further analysis and larger samples, it may be possible to investigate the nature of the cave bears populations, whether all males, or females with their young, were alternately using the cave as a den: there is an indication from variations in tooth mark sizes (Fig. 10.10) that it was occupied by both at different times, the presence of very large tooth marks being the product of male-only occupation, and variation in size being the product of females with yearling young (Kurten 1958; Andrews and Turner 1992).

Conclusions: Site Formation

and Background Scenario

1.The location of the fossiliferous sediments studied here is at the back of the Azokh 1 cave entrance, about 40 meters from the contact with the open air. This situation limits the taphonomic history typical taphonomic karstic agents.

2.Large mammal fossils, other than bears, recorded at the back of the cave show signs of human activity, but do not show a clear pattern that may indicate which type of human occupation took place.

3.Small and medium sized animal skeletons are sparse and incomplete, suggesting an anomalous carcass selection, skin removal and butchery technique. More complete and typical sequences of butchery have been observed on bear fossils, but even here many parts of the skeleton are absent and large bones unbroken.

4.Human occupations of caves are usually more frequent near cave entrances, and bears are more frequent at the back of the cave where they may hibernate. Sediments closer to the entrance had already been excavated, however, and no information was available to us. On the other hand, the back area of the cave is a more suitable area for bear hibernation, and bear fossils are more abundant in this part of the cave than at the cave entrance. This situation suggests to us that the actual butchery processes were concentrated on bears, while other large mammal carcasses could also be butchered, but not in this part of the cave.

5.Chewing marks on the fossils of Azokh 1 and breakage by carnivore action are scarce, and there is no taphonomic evidence of hyenas taking any part of the bone accumulations. The maximum size of puncture marks on diaphyses/compact bone in two units of Azokh 1 exceeds any recorded chewing by other large carnivores, and it is likely that cave bears have damaged and chewed other bones during deposition of Units II and Vu.

6.Unit I, on the other hand, lacks large chewing marks and have a pattern suggesting scavenging by a small carnivore such as a fox or jackal. Unit Vm fossils have tooth marks larger than 4 mm, but with a similar pattern to those in Unit I, and the pattern in Unit III is close to the multispecies site of Arrikrutz (see Fig. 10.12). These patterns would suggest that together with cave bears, smaller carnivores were scavenging the bones in these units.

7.The cave bears may have carried bones from previous human occupations at the entrance to the rear of the cave.

8.Seasonal dampness in the cave is indicated by the formation of cave minerals characteristic of both wet and dry caves, modifications observed on the surface of the fossils (cracks by humidity) and successive stages of bacterial attack. In this context, trampling under wet conditions could greatly increase breakage.

9.The distribution of areas of alteration in the center of the Unit II is seen today in the interior of the cave inhabited

by bats. Accumulations of guano in a damp environment would produce highly acidic fluids (rich in phosphates, sulfates and carbonates dissolved from the guano) percolating through the sediments. Several minerals identified in fossils, sediments and stones are associated with guano breakdown (e.g., brushite, monetite, ardealite, gypsum, sepiolite, tinsleyite).

10.Depending on the conditions of the cave (moments of aridity vs. increase in humidity), different sets of minerals could be formed, some of them very aggressive to the hydroxylapatite mineral component of the bone. High humidity and guano has produced the conditions for the high corrosion observed in these fossils, especially at Unit II.

11.The time gap between Unit II (Pleistocene *100 kyr)

and Unit I (Holocene) and the laminar sedimentation observed on the limit between these two units (Fig. 10.8c) suggest the involvement of water erosion that ended by flooding the substrate. It is likely that an episode of heavy rains and flash floods, removed the upper part of the Pleistocene sediments (formerly deposited almost reaching the roof of the cave) by erosion. This event opened the cave again to the outside environment, and allowed the entry and cave occupation by animals and humans during the Holocene.

12.With further analysis and larger samples, it may be possible to investigate if cave bears males and females with their young were alternately using the cave as a den: there is an indication from variations in tooth mark sizes (Fig. 10.10) that it was occupied by all male groups in some cases, and by females with young in others (Kurten 1958; Andrews and Turner 1992).

Acknowledgements This chapter is based in part on the PhD Thesis investigation by DMM. We are deeply grateful to the authorities of Nagorno-Karabakh for the support and permissions to work at Azokh Caves and to analyze these fossils. We are grateful to Manuel Nieto who has greatly helped with the statistical treatments of this extensive data base, as well as to M.D. Pesquero for taphonomic discussions. Thanks also to Jesús Muñoz and Fernando Señor of the Photo Unit of the Museo Nacional de Ciencias Naturales. We also thank the EMUnit, Laura Tormo, Marta Furió, and Alberto Jorge, as well as Rafael Gómez (XRD analyses) for their professional work and deep involvement in the analysis of some of these samples. The authors are grateful for constructive comments from the three anonymous reviewers and the editor in charge (Tania King) which greatly improved this chapter. These taphonomic investigations have been made possible through funded research projects by the Spanish Ministry of Science (BTE2000-1309, BTE2003-01552, BTE 2007-66231 and CGL201019825).