Fig. 12.5 a Distribution of sequence matches in 619,848 reads analyzed from high throughput sequencing of the Azokh coprolite extract. b Distribution of the mapped reads of the same sequencing experiment

Discussion

Bone and Coprolite Morphometry

Digestion observed in the fossil bone fragment from coprolite 5153 is no higher than moderate in Andrews’ (1990) classification. Although this author did not include hyenas, bears or any other omnivorous predators in his study, experimental work on these predators (Denys et al. 1995; Matthews 2000, 2006; Mondini 2002; Montalvo et al. 2007) indicates that they produce highly digested bones, showing a “torn-like” damaged surfaces when exposed to strong gastric acids (Andrews and Fernández-Jalvo 1998). Effects of digestion on bones ingested and regurgitated by Crocuta crocuta show rounding of the broken edges of digested bone fragments and the characteristic “torn-like” damaged surface (Fernández-Jalvo et al. 2010b). Similar damage is observed on bones digested and excreted by Hyaena brunnea in Fig. 12.2d–f, but it is absent in the smooth surface bone from 5153 (Fig. 12.2b). Since we did not have bone specimens ingested by striped hyena (Hyaena

hyaena), we could not analyze the effects of its digestion under high magnification electron microscopy (3,000×). Lower grades of digestion in the only bone that could be studied from the coprolite of Azokh1 Unit II could be pointing to an animal with weaker gastric action than hyenas, but differences in the degree of bone digestion may occur simultaneously during digestion depending on the position that the bone had in the stomach (Andrews 1990). The absence of a “torn-like” damaged surface is, however, unexpected, but further studies are needed using bones digested by Hyaena hyaena at high magnification electronic microscopy.

Kolska Horwitz (1990) observed signs of digestion on large mammal bones from recent striped hyena scats as well as coprolites from Kebara and Fazel 6 Levant sites. This author provided a size range of bone chips and splinters from 17 to 3 mm (62% smaller than 2 mm long). The small bone fragments found in the Azokh coprolite (6 and 5 mm long) fall within this range. However, higher abundance of bone splinters and larger pieces than the bones in the Azokh coprolites have been found in hyena scats/coprolites (Kolska Horwitz and Goldberg 1989; Kolska Horwitz 1990). Similarly, other carnivore feces referred to by Binford (1980), Haynes (1980), Maguire et al. (1980), Payne and Munson (1985) contain small sized bone splinters showing heavier signs of digestion. This is not the case with the small piece of fossil bone from 5153, in which digestion is moderate.

Kolska Horwitz and Goldberg (1989) indicate the breadth measurement (maximum diameter) of the scat may distinguish between spotted and brown hyena, with that of the spotted hyena being significantly wider. Crocuta crocuta is the largest extant hyaenid with a weight ranging between 45 and 85 kg. Hyaena brunnea weighs on average around 45 kg with exceptional cases reaching up to 72.6 kg (Roberts 1954), and Hyaena hyaena weighs between 30 and 35 kg. The major and minor diameters are the most consistent measurements, because length depends on the number of segments attached (see Fig. 12.1b). Most authors, however, refer to the length and maximum diameter (major axis) and we have used these dimensions to include most coprolite and modern scat measurements available. The results of these measurements are plotted in Fig. 12.1a. Unfortunately, other papers display results of coprolite measurements as average and range values of length and width which cannot be included here.

The larger Azokh coprolites (Fernández-Jalvo et al. 2010a) are still smaller than three of the coprolites from the West Runton Freshwater Bed site in Norfolk (WRFB-UK) measured by Larkin et al. (2000). Parfitt and Larkin (2010) mention an Early Pleistocene site (Untermassfeld, Germany, Keiler 2001) where the dimensions of the coprolites are even larger than those from Norfolk. The exceptionally large coprolites from Untermassfeld, which exceed the dimensions

of modern and fossil hyena scats are considered by Keiler (2001) to have been produced by adults, whilst the smaller and more abundant coprolites derive from hyena pups. However, this would suggest that all coprolites found in other sites (see Table 12.1) are produced by pups and this would need further study. In addition, the size of modern brown hyena scats is also subject to individual diversity (E. M.G. and personnel from the “Fauverie du Mont Faron”, personal observation). On the other hand, the shape of the large coprolites from Norfolk (see Fig. 2b in Larkin et al. 2000) and modern bear (Ursus arctos) scats are alike, both being segmented (Fig. 12.6), and, there is fossil evidence of the presence of Ursus sp. at West Runton (see bear taxonomic discussion in Lewis et al. 2010). Therefore, coprolite morphometry is indicative, but is not conclusive.

In this respect, there is the contention that Ursus spelaeus cannot produce the coprolites because hibernation and fasting would limit bear scats in caves (Nelson et al. 1973; Fernández et al. 2001; Grandal d’Anglade and Fernández-- Mosquera 2008). However, we find evidence in the site that bears were not only hibernating in Azokh 1, but living for longer periods in the cave (Marin-Monfort et al. 2016, see also discussion on Ursus spelaeus diet in that chapter). The fossil bones recovered from Unit II show no characteristic crushed bone hyena traits, nor do they have extensive gnawing or intensive bone digestion (Marin-Monfort et al. 2016). Bone splinters and breakage linked to chewing is low in Azokh 1. Chewing affected only 6.4% of the total number of fossils of Azokh1 and 7.24% in Unit II. These values

seem too low for hyena (Skinner et al. 1998; Pickering 2002; Pokines and Peterhans 2007; Pobiner 2008; Diedrich 2012). The maximum size of tooth marks recorded on bones from Unit II is much larger than those of (known) hyenas, or even lions (Selvaggio and Wilder 2001; Domínguez-Rodrigo and Piqueras 2003; Pobiner 2008; Delaney-Rivera et al. 2009, see discussion in Marin-Monfort et al. 2016).

Chemical Analyses of the Coprolites

Results from XRD and XRF chemical analyses of coprolites obtained by other authors (Kolska Horwitz and Goldberg 1989; Larkin et al. 2000; Lewis 2011; Pesquero et al. 2011) are similar to those obtained here from Azokh coprolites (Tables 12.1 and 12.2, and Fig. 12.3). Percentages of the amorphous phase, however, are higher in Azokh coprolites than in other sites (see Pesquero et al. 2011). Most coprolites and modern hyena scats analyzed by Kolska Horwitz and Goldberg (1989) contained apatite minerals (except two, likely to have been produced by striped hyena). Poor crystallization and better crystallized apatites were indistinctly obtained from both modern and fossil specimens. Coprolites analyzed from West Runton Norfolk and Boxgrove sites give peaks of calcium phosphate in similar proportions to recent Crocuta crocuta scats from Colchester Zoo, all of which were chemically analyzed in Larkin et al. (2000) and Lewis (2011). Other elements (Fe, Mn, Al, Si) detected in these two

Fig. 12.7 X-Ray diffraction diagram obtained from the sediment of Unit II of Azokh cave site (#120), two decayed stones found in Unit II (#110 and 115), and a corroded fossil bone from Unit II (#123) all of them affected by diagenesis probably caused by bat guano deposits (see Marin-Monfort 2016, label # refers to the stub number shown in Table 10.10)

British fossil sites are probably due to diagenetic alteration of the host sediment. Similarly, La Roma site (Pesquero et al. 2011) shows hydroxyapatite below 50%, with other minerals (calcite, quartz, gypsum and moscovite) formed in the calcareous marginal lake environment of this site. The sample from the interior of the coprolite (5153B-inner) contains hydroxyapatite (72.9%), and quartz (10.1%) and other minerals obtained from the other portions or coprolites from Azokh show the influence of the host sediment. Lewis (2011) observed a shift between the hydroxyapatite and fluorapatite phases in the range between 2q 44–54 region of the XRD coprolite diagrams. This shift was interpreted as result of diagenetic alterations frequently occurring in hydroxyapatite minerals. Some of the double peaks observed in this 2q region in Fig. 12.3 may suggest some influence of similar diagenetic alteration, though standard hydroxyapatite positions are better conformed in Azokh coprolites than in coprolites analyzed by Lewis (2011). Therefore, the diagenetic historical context has to be considered when analyzing the chemical composition of fossil materials.

Hydroxyapatite found in Azokh coprolites could in principle come from bones ingested by the animal that produced these coprolites. However, we cannot assert that hydroxyapatite identified in coprolites is exclusively the result of digested bones because Azokh Unit II has been intensively affected by diagenesis due to fluid percolation enriched in acidic bat guano. A wide variety of secondary minerals are associated to bat guano diagenesis in caves, and hydroxyapatite is the most common and stable neo-formed mineral, which has actually been formed in the geological materials of Unit II, such as stones and sediments (Tables 12.3 and 12.4). Hydroxyapatite together with quartz and tinsleyite have been identified as neo-formed (secondary) minerals in Azokh 1 (see Marin-Monfort et al. 2016, Table 10.10 and discussion in that chapter, and Murray et al. 2016). The peaks of samples analyzed in Fig. 12.7 are less broad than in coprolites, but still form short (except for the sediment) and irregularly shaped curves. As said above, the higher and thinner the peak of a mineral is, the better crystallized it is. Thus, the short, broad and irregular peaks suggest poor crystallinity in stones, which is abnormal and indicates the presence of abundant secondary minerals due to diagenesis. The percentages of amorphous phases in these highly diagenetically altered samples, are higher than in coprolites (up to 30%) and the LOI is variable with high values in fossil and sediment and low values in decayed stones.

In this context, the relative high abundance of amorphous phase in Azokh coprolites (above 7% up to 17.1%, Table 12.1), absent in modern scats, suggest the presence of poor-crystallized minerals rather than organic matter content, which may agree better with secondary neo-formed minerals during diagenesis. This is also in agreement with the bone diagenesis results obtained by Smith et al. (2016) who concluded that fossil materials from Azokh 1 (Units II and III) show typical ACH (Accelerated Collagen Hydrolysis), with only small amounts of collagen remaining and often extreme mineralogical changes. Similarly, no preserved DNA could be PCR-amplified from any of the numerous bones analyzed from various locations and layers in Azokh (Bessa-Correia and Geigl, unpublished results).

Table 12.4 Fluorescence (XRF) results from fossil and damaged (decayed) stone and sediment from Azokh 1 Unit II

Paleogenetic and Paleogenomic

Analyses

The paleogenetic analysis of the coprolite using a targeted, highly sensitive quantitative PCR approach revealed the presence of hyena DNA. The mitochondrial cytochrome B/ND2 gene sequences obtained matched those that were produced from modern brown hyena hair (Hyaena brunnea, formerly Parahyaena brunnea) rather than the extant spotted hyena (Crocuta crocuta) or the extinct cave hyena (Crocuta crocuta spelaea). Brown hyenas have never been recorded outside of Southern Africa, and it appears surprising and highly unlikely that the range of this species could have

extended 100,000 years ago as far as to the Caucasus without any prior evidence of the past presence of this species on a wide geographical area (Rohland et al. 2005). Before proposing a profound reappraisal of the past distribution of brown hyenas, it is worthwhile to consider an alternative hypothesis: sample contamination.

There are several arguments in favor of contamination: first, the cytochrome B/ND2 sequences of the Azokh coprolite are identical to those of modern brown hyenas that presently show a reduced diversity of the cytochrome B gene. It would be surprising that the putative brown hyenas that would have lived in the Caucasus 100,000 years ago would have had a mitochondrial DNA identical to extant brown hyenas from South Africa. Indeed, a past population size that would cover such a wide geographical range should have a higher genetic diversity. Second, high throughput sequencing revealed that the coprolite contains essentially environmental DNA and traces of contaminating human DNA. Thus, the DNA from the scat producer is extremely rare, if present at all. The traces of human DNA having a higher mean fragment size than the bulk of DNA can most confidently be attributed to contamination. Indeed, the coprolite was identified as cave bear or hyena scat and originally intended to be solely subject to pollen and taphonomic analyses. Therefore, no contamination prevention procedures were applied. In contrast, the coprolite was extensively manipulated prior to the genetic analysis. In the high throughput next-generation sequencing data, there is no evidence of DNA from another species. This indicates that the hyena DNA sequences obtained in the PCR approach correspond to minute traces of DNA that can be detected only due to the high sensitivity that PCR can achieve when it is optimized. When downstream procedures of high sensitivity must be used, one must ensure that extreme precautions of contamination prevention are used at every stage of the analysis, especially upstream of the high sensitivity analyses, starting from sample collection in the field. These precautions were not applied to this sample prior to the paleogenetic analysis.

It is particularly striking that the hyena species that was identified via a genetic analysis is present only in Southern Africa where the sample was prepared for pollen analysis. Contamination occurred most likely at this stage of the analysis. Contamination in the paleogenomics laboratory in Paris is unlikely since contamination prevention is routinely practiced at all stages of sample analyses: strict physical separation of the different experimental steps in positive air pressure laboratories, as well as multiple contamination prevention procedures, carry-over contamination prevention and reagent decontamination that have been developed and optimized in the laboratory and are used without exception (Pruvost et al. 2004, 2005; Champlot et al. 2010). Furthermore, in the paleogenomics laboratory no hyena DNA of any species had ever been analyzed