Table 10.5 Number of large mammals fossils and coprolites, and fossil richness content (fossil/cubic metre of sediment) for each unit at Azokh 1. 1NR includes fossils and coprolites making a total number of 1935

some of the bear bones, indicating possible human activity in the accumulation of the bones (see below).

Fossil Size, Shape and Density

Most fossils recovered from Azokh 1 measure between 2 and 10 cm long, with about 10% of remains being larger than 10 cm or smaller than 2 cm (Table 10.1). With regard to the shape of these fossils, we have applied the eight categories established by Blott and Pye (2008). All eight shape categories have been recognized in nearly all units of Azokh 1 (Fig. 10.4). Unit I has no fossils in either category 4 (discoid) or category 8 (equant-spheroid), which are also scarce in the rest of the units (Fig. 10.4).

Correlations between the structural bone density of fossils from Azokh 1 and their relative abundance (Ri), have provided negative values or insignificant correlations. Spearman correlation is sensitive to non-linear relationships between variables, so that if only a slight correlation exists between two variables, r Sperman will show this better than Pearson’s correlation. Results obtained from Azokh 1 shown in Table 10.3 suggest that skeletal elements present from these units lack differential preservation due to the density or strength (robustness) of the skeletal element.

Finally, with regard to the transport groups established by Voorhies (1969), the five groups are all well represented in

all units of Azokh 1 (Table 10.4), indicating that there is no preferential accumulation of more easily transported skeletal elements. This suggests that there is no evidence of transport in the fossil assemblage of Azokh 1. This is also in agreement with field data/observations indicating the lack of any preferential orientation of fossils, and the horizontal position that most fossils were found resting in within the sediment. It has also been possible to refit fossils within most units (Units II, III and Vu), indicating lack of reworking. Reworking has only been seen in Unit I, but this was modern reworking by burrowing animals.

The abundances of fossils from each unit, and the richness per unit volume of sediment, are shown in Table 10.5. This does not take into account variations across units, and we have found that all units have higher abundances of fossils at certain parts of the excavation area. In Units II and III these coincide with stone tool spatial abundance (Fig. 10.5).

Results of breakage (Fig. 10.6) following Villa and Mahieu’s (1991) methodology show predominance of mixed angles, curved-Vshape outline and smooth edges. Most fossils have circumference1 (C1 < 1/2 of the original) and length1 (L1 < 1/4 of original length). Complete bones (C3/L4) appear more abundant in Units Vu, III and II, and very scarce in Units I and Vm, and in general breakage is high in all units.

Surface Modifications

The number of fossils that show carnivore tooth marks is low (120 fossils, and 6.4% of total NR in Azokh 1, Table 10.6) and there is a low frequency of bone splinters. Unit II yielded the highest number (NR) of chewed fossils (76), but the relative number of fossils showing carnivore damage is highest in Unit I (12.9%, see Table 10.6). The distribution of chewing categories seen in Unit I fossils also differs from the rest of the units of Azokh 1, and especially from Units III and Vm (Χ2 = 26.043; p < 0.05; df = 4).

The high post-depositional damage (mainly trampling, see below) has also damaged the edges of tooth marks, which hindered measurement of many of them. The total number of tooth marks that could be measured was 199, and the measurements provided in Table 10.6 distinguish the place on the bone where the tooth marks are located (Andrews and Fernández-Jalvo 1997). Most tooth marks are less than 4 mm wide in all categories, but there are also some tooth marks larger than 7 mm in Unit II and Unit Vu. Tooth marks linked to breakage (on spiral breaks category d, transversal break category e, or splinters category f) are scarce or absent (Unit III). Category d (tooth marks on spiral breaks) is most characteristic of predator size, but it is especially rare at Azokh 1, only present at Unit I and II on one and two specimens respectively. The most abundant damage in all units is grooves on diaphysis (compact bone, category b). These are predator-specific as well, but because the diaphyses bone is compact and the marks may have been made by anterior teeth (such as incisors) as well as posterior ones, the sizes of the grooves are generally small, even when made by large body-size carnivores. Tooth prints made by

multi-cusped teeth (category g) have only been recovered from Unit II, the measurements of which (length × breadth of both cusps, and distance between cusps) are as follows:

•(4.5 × 2.2) (7.4 × 3.0) d 8.9

•(3.9 × 1.76) (2.9 × 1.45) d 8.4

•(2.84 × 3.53) (3.73 × 4.80) d 7.5

•(17 × 8.7) no clear distance can be measured

•(16.1 × 5.92) d 11.82.

The last two multi-cusped prints above 15 mm in length, have been recorded on both sides of a Panthera pardus calcaneus (Fig. 10.7a, b). The large size of these prints may be influenced by their location on thin cortical bone and the small dimension of the anatomical element. However, the size of the tooth printed on this bone indicates the size of the carnivore animal that produced it.

The size distributions of these tooth marks based on the four chewing categories are presented graphically in box plot diagrams, showing the range of measurements (max and min), the median and the distribution of measurements (Fig. 10.8). Some outliers are present (from Unit II), shown

on the general overview in Fig. 10.8f, but they have been excluded from the single unit figures. The lowest values have been recorded in Unit I. Units II and Vu show the largest tooth marks and the most diverse tooth mark sizes, especially pits on compact bones (pc). The median values are not high, in part due to the low number of marks recorded on these fossils, but the maximum show large sizes that exceed values obtained in modern chewing cases. This is especially significant as measurements taken in Azokh correspond to breadth or minor axis taken at the narrower part of the pit or the groove. Finally fossils bearing tooth marks are dispersed over the excavation surface at each unit, except the central area of excavation in Unit II, where poor fossil preservation has obscured the evidence (see below, histological analyses).

A rib fragment from Unit I (Holocene) has the ventral end bent at the edge of the rib, which is characteristic of human chewing (see the Fig. 10.4a in Fernández-Jalvo and Andrews 2011). This is the only evidence that we have found so far at Azokh that can be assigned to human chewing.

Few fossils in Azokh 1 have rodent tooth marks (NR = 16, 0.85%). Fossils from Unit I and II have small gnawing marks (between 0.18 and 0.40 mm width). Unit III has not yielded fossils with rodent tooth marks and those from Unit Vu could not be measured. Unit Vm fossils have larger marks in a range of medium (1–2 mm) and large sized marks (3–4 mm) respectively. The latter are characteristic of porcupine (Tong et al. 2008).

Tool induced damage on fossils from Azokh 1 affect 135 fossils (7.2%) showing evidence of cut marks. Unit I exhibits the highest relative percentage (14.7%) of tool marked fossils (Table 10.7) and Unit Vm the lowest (3.8%). The most abundant types of damage recorded are cut marks (incisions) and scraping marks, although the Unit I fossils have few scraping marks. Only a small number of fossils showing stone tool marks could be taxonomically identified (41 specimens), and of these 37 were Ursus spelaeus, mainly from Units II, III and Vu. The rest were Capra hircus and Cervus elaphus from Units I, II and Vm. Incisions identified on these fossils are related to specific anatomical areas, such as joints or ligament/muscle attachments (even in unidenti- fied bone fragments) and most of them appear oblique to the

Table 10.6 Types of tooth marks distinguished in Azokh 1 (see methods). NRtm, Number of fossils bearing tooth marks; Ntm, Number of tooth marks; %tm, percentage of tooth marks (in square brackets, size ranges in mm, in brackets individual measurements, shown in bold are the outliers); %tAz, percentage of tooth marks compared to the total number of tooth marks in Azokh1

226

.al et Monfort-Marin .D.M

length of the bone. They are rarely transversal and only one case from Unit Vm has a longitudinal incision. In contrast, scraping marks usually run longitudinally to the length of the bone, rarely transversal. Although all types of anatomical elements have been affected to some extent by stone tool marks, the most abundant are limb bones with the locations of the marks indicating defleshing processes. There is also evidence of disarticulation and skinning, especially on Ursus spelaeus. Only eight fossils have surface stone tool damage (scraping) directly related to breakage for marrow extraction (four of these were on U. spelaeus). Most fossils between 2 and 10 cm long were cut marked, but in units Vu and Vm cut marks are more frequent on fragments between 2 and 5 cm, while in Units I, II, and III they are present on fragments between 5 and 10 cm. Cut bone fragments larger than 10 cm are present in all units but in low percentages. Remains of animals that have been butchered are dispersed over the excavation area, except for the areas close to the walls.

Raw materials used for stones tools are chert, limestone, basalt, hornfelts, schist, obsidian and a range of siliceous and volcanic materials from nearby river gravels (Asryan et al. 2016). Obsidian is a hard vitreous and easy to knap raw

material, but exotic to the area. Obsidian stone tools can be made with durable and thin edges, leaving thin cut marks on bone surfaces, and this sometimes makes it difficult to distinguish between trampling and cut marks as they strongly mimic each other.

Trampling striations occur over the whole bone surfaces. Trampling marks are present on fossils from all units in Azokh 1, with a higher abundance on those from Unit II and less than 10% at Unit Vu and Vm (Table 10.8). The X2 test shows significant differences between trampling recorded from both parts of Unit V (Vu and Vm) and the other units of Azokh 1, especially between them and Unit II (X 2 = 66.128; p < 0.05; df = 4). On bones with a long axis, trampling marks usually run transversal to the long axis of the bone, affecting salient angles and rarely concave areas. Apart from striations, trampling may also produce rounding, breakage and dispersal of bones (Olsen and Shipman 1988; Andrews 1990; Fernán- dez-Jalvo and Andrews 2016). The breakage traits of broken fossils bearing trampling marks, analyzed according to Villa and Mahieu’s (1991) methodology, shows predominance of mixed angles, although mixed angles are also predominant on all broken fragments at Azokh 1 (Fig. 10.6). Fossils showing

signs of trampling are dispersed over the excavation areas with no particular pattern.

Rounding is also produced by biotic agents such as digestion and licking by carnivores. Digestion produces bone polishing and rounding that differs from sediment abrasion (dry or wet) by distinct ultramicroscopic features (Andrews 1990; Fernández-Jalvo and Andrews 2003, 2016). Digested fossil bones have been recovered from Azokh 1, but in relatively low proportions (<10%). In addition, some fossils from Units II, III and Vu have extremely rounded edges and elongated shape (Fig. 10.9a). These fossils could be mistaken for highly abraded bones, but this possibility is incongruent with taphonomic traits and geological evidence from the site. Evidence of transport has only been distinguished in Unit VI (Murray et al. 2016). Trampling scratches are not any more abundant on the rounded edges than on the rest of the bone surfaces (Fig. 10.9b). Tooth grooves or scores produced by carnivores appear smooth and rounded

(Fig. 10.9c). Sharper scratches by trampling are superimposed on the smooth and rounded edges (Fig. 10.9d). A peculiar pattern of microstriations observed on some of these fossil bones also have rounded edges (Fig. 10.9e). Several of these fragments also show more recent breakage, probably by trampling, with no rounding affecting broken edges (Fig. 10.9a).

Most fossils from Azokh 1 are broken (86%). Table 10.8 shows surface modifications of known taphonomic agents on broken fragments (trampling, abrasion, human action, carnivores or fire). Carnivore tooth marks on broken fragments are not abundant (1.6%). Similarly, tool induced damage and human action related to breakage (adhered flakes, impact and percussion marks and peeling) only affects 6.7% of broken bones. Many of the complete bones have low nutritional content (e.g., ulnae or radius). Fire could influ- ence breakage (in fact, all burnt fossils are broken), for fire increases the likelihood of breakage, but the incidence of

burning is low at Azokh 1 (Table 10.8). The highest number of fossils affected by fire is observed in Unit I (a manure hearth), and low frequencies of burnt bones have been recovered from Units II and Vm (10 and 4 respectively) and none from Units III and Vu. Although the percentage of fossils bearing trampling marks is low (18% of total fossils from Azokh 1), this is the main agent of breakage on broken fossils (16%, see Table 10.8). Breakage due to diagenetic compression such as sediment compaction would occur in situ with both fragments lying close each other.

Fossils recovered from Azokh 1 show cracks with raised or warped up ridges (13.8%) similar in shape to mud-cracks in fine substrates (Fig. 10.7d). This type of cracking has been observed on fossils deep inside caves or in continental aquatic environments (lakeshores), suggesting its relationship with changes in humidity in a damp environment (Díez et al. 1999; Pesquero et al. 2010; Fernández-Jalvo and Andrews 2016). A high abundance of manganese deposits cover these fossils (57%) (Table 10.9).

Highly corroded bones have been observed at Unit II in the central area of excavation. At the same level, but near the cave walls, there is less to no corrosion, and undamaged fossils have been recovered (Fig. 10.10a). The corrosion is observed as heavy cracking, producing a laminated effect and with flaky and dusty surfaces (Fig. 10.10a). Corroded fossil bones are very fragile and require consolidants before they can be removed from the sediments. This corrosion also affects the sediment that is grayish and crumbly towards the central area of excavation, while sediment close to the cave walls is reddish and lacks this crumbly texture. In addition, stones (originally limestone and chert) found in the central part of the excavation are soft and yellow-white in color (‘decayed stones’, Fig. 10.10b, c), and when analyzed by XR-diffraction (Table 10.10) and EDS, inclusions of tinsleyite, apatites and other minerals are seen.

Fig. 10.10 a General view of the excavation area of Unit II, and the remains on the left of the fumier of Unit I. Note modern burrows in the section of Unit I, some of them also affecting the top of Unit II, that caused reworking of fossils and stone tools. The black lines along the sides of the excavation area of Unit II show the limit of crumbly grey sediment in the center and the reddish non-crumbly texture of the sediment next to the cave wall. The small inset bottom left shows the characteristic heavy corrosion of fossils recovered from the central area embedded in the crumbly grey sediment. The small inset top right shows fossils structurally undamaged near the cave wall, embedded in unaltered sediment. Note stones in the central area have a yellow color (and soft texture), white colored stones are less damaged. b Section of Unit II showing the grayish-crumbly sediment and the erosive contact between Unit II and I. c Detail of previous showing the crumbly texture of the Unit II sediment containing soft-yellow stones. Note the laminar sedimentation of Unit I at the erosive contact with Unit II

Corrosion and Chemical Composition

(Histological Analysis)

Histological analyses of 53 fossils analyzed show either heavy bacterial attack (OHI = 0) or bone unmodified by microorganisms (OHI = 5). Intense bacterial attack has only been found in Unit II affecting six of the 22 fossils that were histologically analyzed from this unit. Bacterial colonies in these fossils from Unit II are organized around canals of Havers in a characteristic way (encircling osteones, Fig. 10.11a) that suggests that bacteria acted during the body’s decay (Bell 1990). Bacterial attack can be recognized in the BSE-SEM (backscattered electron mode SEM) as more dense (hypermineralized) zones containing small pores and thin channels 0.1–2.0 microns in diameter. Superimposed generations of bacterial attack on some of these specimens (Fig. 10.11b) have been observed, suggesting fluctuation in the environmental conditions. These more dense areas (re-precipitation of amorphous calcium phosphate) have ‘resisted’ destructive effects of bone corrosion that have literally ‘eaten away’ remains of bone between bacteria colonies (Fig. 10.11a). Canaliculi (histological canals interconnecting osteocytes/lacunae) are enlarged mainly on the most external cortical layer (Fig. 10.11c).

In contrast to Unit II, light bacterial attack has been detected in one fossil from Unit III, based on a sample of five from this Unit. Microbial damage is so mild that has not affected the general histology of this bone (OHI is still 5). No bacterial attack has been observed on fossils from Units I, Vu and Vm. Fungal action has been found on fossils from Units III and Vm affecting 1 and 3 fossils respectively (Fig. 10.11d).

In addition to bacterial attack, Unit II has the heaviest corrosion observed in Azokh 1. No specific histological damage has been observed that may explain this heavily flaked and laminated texture that also affects the inner layers of the compact bone (Fig. 10.11e). Fossils contain secondary deposits of tinsleyite minerals (KAl2(PO4)2(OH) 2(H2O)), confirmed by XRD. Crumbly sediment and flaked fossils also have tinsleyite in their composition (Table 10.10). Soft ‘decayed’ stones have a highly porous texture and have been transformed into calcium phosphate (Fig. 10.11f), identified as hydroxylapatite by XRD (Table 10.10, stub sample #110). Local deposits of barite [Ba(SO4)] have also been detected by EDS in Unit II fossil bones, as well as gypsum [Ca(SO4) 2(H2O)] and bassanite [2(Ca)2(SO4) (H2O)] identified by diffraction spectrometry (XRD).

Hydroxylapatite [Ca5(PO4)2.5(CO3)0.5(OH)] is the most thermodynamically stable mineral of this group, and it forms in a short time. Brushite [Ca(HPO4)·2(H2O)] is stable in conditions of high acidity (pH < 6) and damp conditions, but it loses water readily, converting to monetite [Ca(HPO4)]. The formation of gypsum [Ca(SO4)·2 (H2O)] is common in the processes of decomposition of bat guano, where sulfur comes from organic matter and the calcium from the dissolution of the calcareous rock (or fossils). Ardealite [Ca2(SO4)(HPO4)·4(H2O)] can also be formed in caves in the presence of guano, and this mineral together with gypsum often appear in dry caves. Sepiolite [Mg4Si6O15·6(H2O)] is an authigenic mineral of caves associated with conditions of filtration of water rich in magnesium followed by extreme aridity, while pyrolusite [MnO2] and oxides of manganese are frequent in damp caves. Several of these minerals (hydroxylapatite or series

of apatite, gypsum, sepiolite and tinsleyite) have been detected by XR diffraction or by EDS analysis of the Azokh 1 fossils, indicating fluctuations of arid conditions and high humidity.

Unit Vm has also yielded laminated and flaky fossils (Fig. 10.11g). EDS analyses of these fossil bones shows a high proportion of phosphorous that anomalously exceeds the amount of calcium for the basic bone mineral composition of hydroxylapatite. None of the samples from Unit Vm analyzed by XRD yielded tinsleyite, which was observed in Unit II and which could explain the cracked-laminated surfaces. These XRD analyses, however, show the presence of sepiolite minerals [Mg4Si6O15·6(H2O)] both in fossils and sediment/stones.

Some fossils from Azokh 1 have a ‘stone-like’ texture to the naked eye, and while most of them come from Unit Vm, some also are known from Units II and Vu. These fossils show a heavy microscopic cracking (Fig. 10.11h) and anomalous quantities of chemical elements detected by EDS spectrometry. Some fossils have enrichment in phosphate and others in calcium, as well as sulfur, potassium or silica. Sediment attached or underneath these fossils are enriched in phosphorous. An amorphous secondary deposit infilling histological features (canals of Havers or Volkmann’s) of some fossils is mainly calcium phosphate (brushite, apatite or hydroxylapatite) sometimes with sulfur (possibly ardealite, monetite). Element composition obtained by EDS is not sufficiently conclusive to identify the type of mineral or substance that might deposit on these fossils. XRD analysis of one of these ‘stone like’ fossils provides 100% of hydroxylapatite. The only conclusive result that may be proposed for this peculiar heavy microscopic cracking and ‘stone-like’ texture is chemical, but it is unclear what the exact chemical process was, or which minerals were responsible for the damage to these fossils. Limestone blocks from the central area of Unit II excavation are also structurally and chemically altered. These stones have enormously increased porosity, and the original calcareous or siliceous composition (limestone or chert) has been transformed to calcium phosphate.