We the thank ICAZ taphonomy working group 2016 conference organizers for inviting us to this conference where we presented this paper. Travel and accommodation was generously financed by IFRA (Insitut Français de Recherches en Afrique, Nairobi). The study was supported by the French Government, People’s Trust for Endangered Wildlife of UK and Earthwatch’s Samburu Carnivores Project that was directed by Professors William Ogara and Nicholas Oguge and Michigan State University Shompole Striped Hyaena Project.

1 - Introduction

1In Africa, besides humans, the three species of hyaenas (striped, spotted and brown), the leopard, some canids, mongooses, the porcupine and birds (owls, eagles, crows) are the other species known to accumulate large quantities of bones in their dens, lairs or roosting sites (Skinner et al. 1998 ; Pokines & Kerbis Peterhans, 2007 ; Lansing, et al. 2009 ; Brain, 1981). The hyaenids’ bones accumulating behaviour has been the centre of discussion for over a century due to their role in fossil assemblages’ taphonomy and are used as models to interpret these assemblages found in archaeological or paleontological contexts.

2This study focused on a comparative bone remains analysis of the two sympatric hyaenas (C. crocuta (Cc), H. hyaena (Hh) of Kenya. Previous hyaena dens studies in Kenya focused on single dens that were used by spotted hyaenas (e.g., Bunn, 1983 ; Faith, 2007 ; Hill, 1983 ; Pokines & Kerbis Peterhans, 2007). The only other two striped hyaenas studies conducted in Kenya were of Leakey et al. (1999) who focussed on its bone collecting behaviour in Turkana and Wagner (2006) in Laikipia who mainly studied its behavioural ecology and evolution. In contrast more ecological studies have been made on both species in Southern Africa and other areas (e.g., Kuhn et al. 2008; Mills, 1990 ; Skinner & van Aarde 1991 ; Wiesel, 2006 ; Fourvel et al, 2012, 2015) including European fossil record (e.g., Brugal et al. 1997 ; Fosse, 1995 ; Stiner, 1994) that provide valuable comparative information on assemblages of these species. Our study therefore, provides valuable bone collection and ravaging comparative data for these free ranging hyaenas in wildlife protected and pastoralists’ areas in Kenya.

2 - Materials and methods

2.1 - Study areas and samples

3This study was carried out in two and five group ranches plus Samburu National Reserve in the Magadi division of Kajiado and in Wamba division of Samburu Counties respectively (in Kenya’s Rift Valley) (fig. 1).These areas are climatically and socio-culturally similar and pastoralism is the major land use practise. They both have community managed wildlife protected areas for wildlife conservation and eco-tourism where livestock entry is restricted. Faunal diversity of the two areas is almost similar (Behrensmeyer & Boaz, 1980 ; Williams, 1967) and the taxa representation (especially in herbivore and carnivore) patterns in the two study areas and dens is very similar (fig. 2 and tab. 1). We studied 50 hyaena dens - located both in human occupied and wildlife protected areas - that were found opportunistically, through local people or radio collared striped hyaenas. We collected and analysed 10,792 bones with 62 % of which were identifiable to genus or species level; overall 71 species were determined. Here, we will focus on the ungulate element representation and modification patterns in eight Cc and 9 Hh dens with a more detailed comparison of two of them (Lenkobei Hh den in Magadi and Ngutuk Ongironi Den1 a Cc den in Samburu) that yielded more bones. It is noteworthy that all located spotted hyaena dens in Magadi were empty without any evidence of bones or even coprolites, although local observations indicate regular occupations. Similarly, we did not include three suspected striped hyaena dens in the detailed analysis given that we were not certain they were the sole users of the dens. Most of Hh dens are located in burrows (often dug by aardvark) under bushes; but the Lenkobei den was at the base of a volcanic flow, with bones distributed inside a pile of rocks. Most of the Cc dens are found in natural rock cavity (with narrow and low entrances, relatively deep and inaccessible), but few are in burrows.

2.2 - Data collection and analysis

4Specimens were surface collected with limited sieving through a 2 mm sieve and identified using the comparative collection at the National Museums of Kenya. We recorded the taxon, as well as age, element and anatomical position, element portion and classified according size classes (e.g., Brain, 1974). Species diversity between the study areas was compared using different indices and statistical tools (Lyman, 2008; Hammer and Harper, 2006). Number of identified specimens (NISP) and minimum number of individuals (MNI) were calculated following Lyman (2008). Bone surface modifications (weathering, gnawing marks, sharp instrument cut marks, tooth notch and puncture sizes and other observable changes) were documented (see Mwebi, 2013 for details).

3 - Results

5The same number of ungulate species (n=23, see tab. 1) were represented in both study areas with slight differences in species diversity, proportions of their bone remains and total numbers of individuals (67.6% NISP and 78.9% MNI for Magadi ; 67.4% and 75.4% for Samburu, calculation based on total NISP including all species). Table 2 presents the ungulate representation per size class in the H. hyaena and C. crocuta dens. When compared between the two hyena species, the Cc dens (n=8) were overall dominated by size classes III to V with some small variations which may be as a result of prey availability at each den site. In these Cc dens, the proportions of size class III specimens’ were higher in five while those of size class I & II dominated in three (tabl. 3 ; in this study we decided to combine size classes I and II and IV and V ). However, differences were not statistically significant (U=26, z=-0.5785, p=0.5556). Size class IV & V were the least represented.

6In contrast in Hh den sites nine out of 12 had size I & II ungulate proportions consistently higher than the other size classes. The difference between size classes I & II and III was statistically highly significant (p <0.0001). Size IV & V proportions were consistently low as was the case in Cc dens. Table 3 provides size class I&II and III species list with their NISP and MNI for the nine Hh and eight Cc dens respectively. Size class I&II constitutes 80% and 72% of the NISP and MNI respectively of the Hh dens while they are represented by about 40% and 36% respectively in the Cc dens.

7Skeletal Element representation analysis of three Cc dens that had slightly more bones indicated that there was no clear patterning in the element proportions per den site. However, there appeared to be some variations in some den sites where some elements were more represented than others. Scapulae and ribs were fewer in all the three dens while the mandibles and pelvis less represented in SaNguOD1 and NkD3 respectively. A Kruskal-Wallis statistical test indicated no significant statistical differences (p>0.05) between cranial and long limb bone representation in these dens. Thus no skeletal elements were preferentially deposited over others in Cc dens. In contrast, tibia, metapodials, pes/manus, vertebrae, ribs and skull/maxilla proportions were higher than the other long limb and mandible elements in the 9 Hh dens analysed (tab. 4). This suggests that Hh selectively destroyed the other elements over skull/maxilla, ribs, vertebrae, tibia and metapodials. Alternatively the Hh might have accessed/scavenged carcasses lacking the meat rich elements (femur, humerus, scapula and pelvis). However, ribs representation seemed variable in relation to the other elements’ across these Hh dens suggesting that in some cases they were consumed at kill sites and partially consumed skeletons were transported to the dens. In some of these Hh dens more vertebrae, ribs and pes/manus mostly survived than all the other skeletal elements. These patterns’ variability may be indicative of the hunting versus scavenging strategies in the two hyaena species given that the meat deficient elements are likely to be more collected and deposited in a scavenging strategy than in a hunting one.

8Given that most dens had few bones we have just compared the ungulate element fragmentation of two dens (SaNguOD1 a Cc den and Lenkobei an Hh den) with the largest bone assemblages. There were more complete and almost complete elements in the Cc den (68%) compared to that of Hh (59.1%). Bone shafts (including cylinders) and fragments proportions were higher (21.2%) in the Hh den than in the Cc (11.3%). However, the general fragmentation trend of the two species is the same (Figure 4). A similar trend was also observed when the tooth mark frequencies and types were compared for these two den bone assemblages (cf. infra).

9Crenulated gnawing was the most common tooth mark in both Cc and Hh dens, followed by pits then scoring, punctures and notches respectively. Porcupine gnawing (2.2%) and sharp instrument cut marks (1.2%) were low. This indicates that the two collections were exclusively carnivore modified and accumulated with limited scavenging from human refuse.

10Tooth mark frequencies, locations, notch and puncture sizes on ungulate skeletal elements were very similar between the two Cc and Hh dens. However, their frequency was higher on the pelvis, tibia, femur and humerus in the Cc than in the Hh accumulations. On the other hand metapodials, scapula, mandibles and skull/maxilla were the most marked for Hh. While the variations were small, the pattern suggests a difference in the bone collecting behaviour and access to meat rich bones between the two hyaena species. Both hyaena species leave more tooth marks on the near distal, near proximal ends and shafts of limb bones (fig. 5). The proportions of tooth marks on the epiphyses and facet surfaces/edges for both species were compared to infer which of them accessed more articulated skeletal elements than the other. This was based on the assumption that: access to articulated elements will result in higher proportions of tooth marks on the epiphyses and articulation facets surfaces/edges during the disarticulation process. The spotted hyaena elements had a higher (27.2%) of its skeletal elements marked on the articular facets and epiphyses in general than those of the striped hyaena (9%). This difference is substantial and may indicate that the spotted hyaena hunted its prey and disarticulated them while the striped hyaena accessed already disarticulated remains. Additionally, the striped hyaena accumulations had the lower limb bones’ ends more marked than the spotted hyaena. A Kolmogorov-Smirnov test on the differences between the near distal tooth marks proportions of the ungulate limb bones from the two hyaena species dens was statistically highly significant (p=0.004). That further suggests that the Hh accessed poor meaty-bones that they thoroughly gnawed while extracting the meat scrapes. While the tooth mark frequencies and location suggest a difference in the bone modifications patterns in the two hyaena species, puncture and notch sizes suggests no differences.

11While they were generally few, tooth puncture and notch dimensions were measured to nearest millimetre for five dens (one Cc and four Hh) that had slightly large assemblages. The major axes ranged from 2 to 10 mm in the Cc den with those measuring about 5 mm being the most frequent at 33.3%. The majority (78%) of the major axes of the Cc den punctures measured between 3 and 5 mm. Similarly, in the four Hh dens the major axes ranged from 1 to 12 mm. The most frequent measurement of the major axis (one third in both) was 3 mm in three Hh dens and 4 mm was the most frequent in one. These measurements do not indicate a clear pattern that can differentiate assemblages made by the two hyaena species. They varied between the dens and even within a den as demonstrated above. The only pattern that appears across is that the notch and puncture major axis lengths are very similar within one bone assemblage/den, suggesting that it is the same tooth that produces both tooth mark types.

12

4 - Concluding remarks

13Bone accumulations made by hyaenas are important models for reconstruction, interpretation and distinguishing archaeological or paleontological bone assemblages. This will be useful if their bone composition and modification patterns are well discernable. Hill’s (1983) study of a Cc Amboseli den observed a clear patterning in damage where there was a disproportionate representation of elements resulting from differential transport (size class IV & V was under-represented), bone density and size. This is consistent with our study where generally size classes I to III are more frequent, size class I & II dominating in Hh dens and size III in the Cc dens. Kerbis-Peterhans (1990) suggests that small vs. large ungulate composition aid in distinguishing Cc and Hh collections. While this criterion applies to the latter in our study, it does not seem to be the case for all Cc dens. Lam’s (1988) study of the Koobi Fora Cc den had overwhelmingly high composition of the size II (mainly domestic, caprines) while the Arad Hh den of Israel (Kerbis-Peterhans, 1990) was dominated by size class IV (mainly camel). This suggests that the structure of bone assemblages are largely conditioned and vary according to environment, availability of prey (including domestic stock) and even seasonally even though measuring seasonal variation in an assemblage has not yet been attempted and may be difficult to undertake.

14Skeletal part representation has been proposed as a means of differentiating hyaenid accumulated assemblages from those of other carnivores (e.g., Kerbis-Peterhans, 1990; Klein and Cruz-Uribe 1984). It is suggested that Cc assemblages have adult small ungulate skulls dominating compared to those of large ungulates due to Schlepp effect. This was the case in Syokimau Cc den in Kenya (Egeland et al. 2008). However, Kerbis-Peterhans, (1990) reports that scavenging Hh of Israel accumulates skulls/mandibles of all size classes at equal rates. The SaNguOD1 Cc den of this study followed the Israel hyaenas’ trend. However, two other Cc (NK3 and Lod4) dens tended to follow the Syokimau Cc den’s patterning. In contrast the Lenkobei Hh den had skull/maxillae of size class III dominating those of size class I & II but the mandibles of these size class were more abundant than those of class III ungulates. This demonstrates the high degree of variability in the nature and structure of bone assemblages made by hyenids. Thus skeletal part representations in hyaena den assemblages have to be interpreted taking into account of the possible environmental conditions under which they were accumulated. For example, modern den examples in areas rich in domestic species are far from being suitable for comparative purposes with those accumulated under pristine conditions, as are observations made in artificial conditions (zoo). The den site context, differential transport and density mediated attrition influence final bone assemblage composition (Egeland et al. 2008, Faith et al. 2007). Even tooth mark patterns are not consistent in carnivore assemblages as demonstrated in this study and others elsewhere (Egeland et al. 2008 ; Pobiner, 2008 ; Egeland and Domínguez-Rodrigo 2008 ; Domínguez-Rodrigo et al. 1999 ; Pokines & Kerbis Peterhans, 2007). Our taphonomical study, only briefly reported here (Mwebi, in prep.), tried to characterize the Cc and Hh bone assemblages based on their bone modification patterns, ungulate prey sizes, element representation and species composition. It contributes in bringing additional data useful for interpretation of bone assemblages of unknown accumulators both in zooarchaeological and paleontological studies as well as in modern ecology and conservation issues. As in this work, such studies elsewhere show the existence of variability, probably related to duration of bone accumulation and seasonal pattern not yet fully understood. In any case, such modern referential can be fruitfully tested to some extent with fossils in Pleistocene and Palaeolithic context from Africa or Eurasia.