Carnivore competition, bone destruction, and bone density J. Tyler Faith a, * , Curtis W. Marean b , Anna K. Behrensmeyer c a Hominid Paleobiology Doctoral Program, CASHP, Department of Anthropology, George Washington University, 2110 G St NW, Washington, DC 20052, USA b Institute of Human Origins, School of Human Evolution and Social Change, PO Box 872402, Arizona State University, Tempe, AZ 85287, USA c Department of Paleobiology, Evolution of Terrestrial Ecosystems Program, Smithsonian Institution, National Museum of Natural History, PO Box 37012, NHB, MRC-121, Washington, DC 20013-7012, USA Received 25 September 2006; received in revised form 24 December 2006; accepted 2 January 2007 Abstract In carnivore-modified archaeofaunal assemblages it is important to evaluate the degree to which carnivores have overprinted hominin behav- ioral signals. To examine the signals of increased competition for discarded bone, we present controlled experimental data on 33 simulated ar- chaeological sites subjected to secondary consumption by spotted hyenas. We examine the relationship between competition, as measured by controlled numbers of hyenas and limb bones, and resultant levels of destruction and correlations between long-bone portion survivorship and bone density. Our results indicate that levels of destruction are equivalent regardless of the numbers of hyenas or long-bones included in the experimental assemblages. Correlations between long-bone epiphyseal and near-epiphyseal portions and bone density, however, do provide an indication of the level of competition. Results from the experimental study are used to highlight divergent levels of carnivore competition for hominin-discarded bone at the Plio-Pleistocene localities FLKN-Zinjanthropus and FLKN levels 1e2 from Olduvai Gorge, Tanzania. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Carnivore competition; Carnivore destruction; Bone density; Spotted hyena; Taphonomy; Olduvai 1. Introduction Reconstructing the taphonomic history of an archaeofaunal assemblage requires an assessment of the processes that de- stroy and alter faunal remains following human discard. An evaluation of these processes, which includes, but is not lim- ited to, subaerial weathering (Behrensmeyer, 1978), carnivore modification (Binford et al., 1988; Blumenschine, 1988; Blumenschine and Marean, 1993; Marean and Spencer, 1991; Marean et al., 1992), and post-depositional alteration (Klein and Cruz-Uribe, 1984; Marean, 1991), is a crucial step toward inferring hominin behavioral patterns. It has long been recognized that differential survivorship of skeletal elements and portions thereof is mediated by physical characteristics - primarily bone density (Brain, 1967, 1969; Lam et al., 1998; Lyman, 1994). Since Lyman’s (1984) pioneering study of bone density in deer (Odocoileus spp.), bone density analyses of archaeofaunal assemblages have provided a means of gauging the effects of density-mediated attrition on skeletal element representation. The presence of a significant positive correlation between skeletal element abundance and bone density, when an assemblage has not been subjected to biasing collection procedures, is presumed to reflect density-mediated destruction of faunal remains (Grayson, 1989; Klein, 1989; Lam et al., 1998; Lupo, 1995; Lyman, 1993, 1994). Of the many destructive processes considered to be affected by bone density, carnivore destruction has been documented in experimental contexts to be at least partially density-mediated (Marean and Spencer, 1991; Marean et al., 1992). Carnivore destruction can include both primary destruction of bones en- countered from carcasses preyed upon by carnivores, and * Corresponding author. Tel.: þ1 202 994 0154. E-mail address: [email protected](J.T. Faith). 0305-4403/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2007.01.017 Journal of Archaeological Science 34 (2007) 2025e2034 http://www.elsevier.com/locate/jas
10
Embed
Carnivore competition, bone destruction, and bone density
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Journal of Archaeological Science 34 (2007) 2025e2034http://www.elsevier.com/locate/jas
Carnivore competition, bone destruction, and bone density
J. Tyler Faith a,*, Curtis W. Marean b, Anna K. Behrensmeyer c
a Hominid Paleobiology Doctoral Program, CASHP, Department of Anthropology, George Washington University, 2110 G St NW,
Washington, DC 20052, USAb Institute of Human Origins, School of Human Evolution and Social Change, PO Box 872402, Arizona State University,
Tempe, AZ 85287, USAc Department of Paleobiology, Evolution of Terrestrial Ecosystems Program, Smithsonian Institution, National Museum of Natural History,
PO Box 37012, NHB, MRC-121, Washington, DC 20013-7012, USA
Received 25 September 2006; received in revised form 24 December 2006; accepted 2 January 2007
Abstract
In carnivore-modified archaeofaunal assemblages it is important to evaluate the degree to which carnivores have overprinted hominin behav-ioral signals. To examine the signals of increased competition for discarded bone, we present controlled experimental data on 33 simulated ar-chaeological sites subjected to secondary consumption by spotted hyenas. We examine the relationship between competition, as measured bycontrolled numbers of hyenas and limb bones, and resultant levels of destruction and correlations between long-bone portion survivorship andbone density. Our results indicate that levels of destruction are equivalent regardless of the numbers of hyenas or long-bones included in theexperimental assemblages. Correlations between long-bone epiphyseal and near-epiphyseal portions and bone density, however, do providean indication of the level of competition. Results from the experimental study are used to highlight divergent levels of carnivore competitionfor hominin-discarded bone at the Plio-Pleistocene localities FLKN-Zinjanthropus and FLKN levels 1e2 from Olduvai Gorge, Tanzania.� 2007 Elsevier Ltd. All rights reserved.
Reconstructing the taphonomic history of an archaeofaunalassemblage requires an assessment of the processes that de-stroy and alter faunal remains following human discard. Anevaluation of these processes, which includes, but is not lim-ited to, subaerial weathering (Behrensmeyer, 1978), carnivoremodification (Binford et al., 1988; Blumenschine, 1988;Blumenschine and Marean, 1993; Marean and Spencer,1991; Marean et al., 1992), and post-depositional alteration(Klein and Cruz-Uribe, 1984; Marean, 1991), is a crucialstep toward inferring hominin behavioral patterns. It haslong been recognized that differential survivorship of skeletalelements and portions thereof is mediated by physical
0305-4403/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2007.01.017
characteristics - primarily bone density (Brain, 1967, 1969;Lam et al., 1998; Lyman, 1994). Since Lyman’s (1984)pioneering study of bone density in deer (Odocoileus spp.),bone density analyses of archaeofaunal assemblages haveprovided a means of gauging the effects of density-mediatedattrition on skeletal element representation. The presence ofa significant positive correlation between skeletal elementabundance and bone density, when an assemblage has notbeen subjected to biasing collection procedures, is presumedto reflect density-mediated destruction of faunal remains(Grayson, 1989; Klein, 1989; Lam et al., 1998; Lupo, 1995;Lyman, 1993, 1994).
Of the many destructive processes considered to be affectedby bone density, carnivore destruction has been documented inexperimental contexts to be at least partially density-mediated(Marean and Spencer, 1991; Marean et al., 1992). Carnivoredestruction can include both primary destruction of bones en-countered from carcasses preyed upon by carnivores, and
2026 J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
secondary consumption of skeletal portions previously dis-carded by human foragers. Low-density elements and portionsof bone, such as long-bone epiphyses, retain an abundance ofbone grease distributed within cancellous bone that is highly at-tractive to carnivores (Binford, 1978; Blumenschine, 1988;Lyman, 1985). The abundances of skeletal elements in faunalassemblages subjected to carnivore destruction are thus ex-pected to correlate positively with bone density. This is notalways the case (Carlson and Pickering, 2004; Pickering andCarlson, 2002), and it has been suggested that the absence ofa correlation in situations of undoubted carnivore interactionwith a bone assemblage may relate to the use of inaccuratebone density data or a violation of the conditions required fora correlation analysis (Lam and Pearson, 2005). Alternatively,a weak correlation between bone survivorship and bone densitymay reflect local ecological factors such as the availability ofconsumable bone and intensity of carnivore competition (Faithand Behrensmeyer, 2006). Here we present new information oncarnivore destruction of bone in an examination of experimen-tal patterns of bone destruction as they relate to bone densityand varying levels of conspecific competition.
Previous experimental studies have subjected ‘‘simulatedarchaeological sites’’ (as in Blumenschine, 1988) to spottedhyena modification to show that middle shaft portions oflong-bones provide the most accurate estimate of the originalnumber of limb elements (Marean and Spencer, 1991) and topresent controlled data on the sequence of bone choice and dif-ferential levels of destruction to axial, appendicular, and com-pact bones (Marean et al., 1992). In this analysis we expandon these studies and use Marean’s original experimental datato examine the relationship between levels of carnivore compe-tition and the differential survivorship of appendicular boneportions of varying densities. This allows us to examine skeletalelement abundance and bone density correlations under con-trolled levels of competition, as measured by a predeterminednumber of hyenas and limb elements.
In a previous paper, Faith and Behrensmeyer (2006) sug-gested that the strength of the correlation between long-boneepiphyses survival and bone density reflects the intensity of car-nivore competition and degree of modification to an assem-blage. This proposal is based on an actualistic case study ofpatterns of skeletal element survival under changing abundancesof spotted hyena (Crocuta crocuta), an effective bone-cruncher,in Amboseli Park, Kenya. At times with low abundances of Cro-cuta, long-bone epiphysis survival correlates significantly withbone density. When the Crocuta population in Amboseliincreased seven-fold, the correlation declines to insignificance.Faith and Behrensmeyer interpret this pattern as a function ofa carnivore’s ability to be selective in choosing low-density ele-ments and portions for consumption. At low levels of competi-tion (e.g., an individual carnivore consuming carcass remains),carnivores can be selective. When inter- or intra-specific compe-tition is high (e.g., a feeding frenzy), carnivores do not discrim-inate between high-and low-density portions, resulting in a poorcorrelation with bone density. In such situations, little survivesother than the highest density pieces completely lacking cancel-lous bone.
Carnivores have been implicated in the taphonomic historiesof archaeological bone assemblages in a broad range ofgeographic and environmental contexts (e.g., Assefa, 2006;Bartram and Marean, 1999; Blumenschine, 1995; Brain,1981; Chase et al., 1994; Domınguez-Rodrigo et al., 2002;Marean et al., 2000; Marean and Kim, 1998; Monahan, 1996;Mondini, 2002; Potts, 1988). Evaluating the levels of carnivorecompetition for human-discarded bone can be a powerful toolfor assessing the degree to which carnivores have overprintedhominin behavioral signals. Although the patterns in selectedarchaeological contexts closely match the patterns expectedbased on the Amboseli data (Faith and Behrensmeyer, 2006),this previous study lacks a direct observational link betweenthe processes (e.g., level of carnivore competition) and result-ing patterns (e.g., strength of correlation between limb-endabundance and bone density). A cross-validation of the rela-tionship between carnivore competition and bone density incontrolled experimental settings strengthens the applicationof the actualistic data from Amboseli to the archaeologicalrecord (Marean, 1995). With this in mind, we present new anal-yses of Marean’s experimental data (Marean and Spencer,1991; Marean et al., 1992) to enhance our understanding ofthe link between carnivore competition, bone destruction,and bone density.
2. Experimental methods
The processes involved in carnivore modification of a boneassemblage, termed ‘‘carnivore ravaging’’ by Binford (1981),include bone choice and transportation, modification by chew-ing, and destruction by consumption. Bunn (1986) and Bunnand Kroll (1986) had observed that secondary carnivore con-sumption of human-discarded bone resulted in deletion oflong-bone epiphyses. Marean and Spencer (1991) provided ex-perimental and quantitative documentation that epiphyses andnear-epiphyseal fragments are consistently consumed and de-stroyed by spotted hyenas in order to digest nutritious bonegrease within these cancellous portions. Here we test whethervariation in the destruction of long-bone portions and the rela-tionship between surviving portions and bone density is af-fected by competition for nutrients among different numbersof bone consumers.
Simulated archaeological sites were designed so as to accu-rately replicate a discarded hominin-accumulated assemblagethat was subsequently consumed by one or more carnivores(Marean and Spencer, 1991; Marean et al., 1992). Two as-sumptions were made concerning hominin carcass-processingbehaviors. First, it is assumed that bones were discarded ina defleshed state, as indicated by the presence of cutmarkson the shafts of long-bones at Pliocene (Domınguez-Rodrigoet al., 2005) and Plio-Pleistocene assemblages (Bunn, 1981;Bunn and Kroll, 1986; Domınguez-Rodrigo et al., 2002; Mon-ahan, 1996; Potts, 1988; Potts and Shipman, 1981). Likewise,it is also assumed that hominins broke limb bones to extractmarrow. This behavior is supported by fragmented shaftsand the presence of hammerstone/anvil percussion marks onlong-bones from numerous Plio-Pleistocene assemblages,
2027J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
including FLK-Zinjanthropus at Olduvai Gorge, Tanzania(Blumenschine, 1995; Blumenschine and Marean, 1993;Bunn, 1981).
Unbroken sheep (Ovis aries) bones were purchased froma meat distributor. Bones most commonly available includedhindlimb elements, the femur, tibia, metatarsal, and tarsals,as well as pelves and various vertebrae. Only the long-bonesare considered in this study. Adhering flesh, fat, and cartilagewas stripped from all bones using a metal knife. Most limbbones were broken using hammerstone and anvil percussion(as illustrated by Binford, 1981) and the marrow was subse-quently removed from the medullary cavities. An additionalgroup of limb bones were left unbroken to simulate two basictypes of sites: those with limb bones broken by hammerstonepercussion and those with unbroken limb bones. Given thatour focus is on carnivore destruction of bone, a process medi-ated by the related properties of bone density and grease con-tent, we do not distinguish between the two types of simulatedsites in the following analyses. The density and grease contentof cancellous limb bone portions remains constant, regardlessof how the bones are processed, and Marean and Spencer(1991) report no significant differences between hammer-stone-broken and unbroken assemblages.
All simulated sites involved from one to three limb units inaddition to a vertebral section. A limb unit is defined as onepelvis, femur, tibia, and metatarsal with adhering tarsals andfirst phalanges. The vertebral sections include either a sectionfrom the atlas to third thoracic, or the fifth and sixth lumbarvertebrae plus the sacrum. Fig. 1 illustrates a typical simulatedsite. The simulated sites were placed within enclosures at theBerkeley Spotted Hyena Colony. As a highly effective bone-cruncher, Crocuta provides an analogue for the more extremelevels of carnivore modification that might occur in a hominid-accumulated bone assemblage (Marean and Spencer, 1991).At the time of the experiments, three groups of hyenas werehoused within three separate enclosures, which includeda compartment with a 7 � 10 m concrete floor connected viaa tunnel to a second 7 � 20 m compartment with a naturalsoil base. All simulated sites were deposited in the soil-basedcompartment.
The hyenas were restricted to the concrete compartmentwhile the sites were set up and were not fed for 24 h priorto the experiment. All bones and bone fragments wereinscribed with a unique identification number, mapped andphotographed. One, two, or four hyenas were allowed accessto the simulated sites. Observations were made from a tower
Fig. 1. Plan of a hammerstone-broken simulated site prior to, and following, spotted hyena consumption (modified from Marean and Spencer, 1991).
2028 J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
adjacent to the enclosure and the experiments were terminatedwhen the hyenas appeared satiated, defined as no chewing orsniffing of bones for 15 min. At this point the hyenas werelured back into the tunnel and confined. All surviving bonesand bone fragments were re-numbered, mapped, described,and collected for further analysis. As in previous studies(Marean and Spencer, 1991; Marean et al., 1992), we reportthe results from an analysis of 33 of these simulated siteexperiments. In Table 1 we present the basic parameters foreach experimental trial.
It is important to note that the range of experimental param-eters (i.e., 1e4 hyenas and 1e3 limb units) provides onlya partial perspective of the possible interactions between car-nivores and discarded bone assemblages. However, the con-trolled experimental observations do afford the opportunityto directly observe taphonomic processes and their resultantpatterning on bone assemblages. A clear understanding ofthe linkage between process and pattern provides a baselinefor extrapolating to other scenarios that may characterize thearchaeological record.
Following the experimental carnivore modification, allbone fragments were identified according to element, side,portion, section and percent circumference of a completepreserved section. Bone portions include the five regionsdescribed by Marean and Spencer (1991): proximal end, prox-imal shaft, middle shaft, distal shaft, and distal end. Theseregions were defined in a manner that recognized naturalchanges in bone density along the long axis previously docu-mented by Lyman (1984). The bone section includes the partof the bone portion that is preserved, such as anterior/lateral.Bone fragments that initially proved difficult to identifywere refit so that nearly all bone fragments were identifiedto portion and section.
Minimum number of skeletal element (MNE) estimateswere calculated for each of the five bone portions followingthe fraction summation approach described by Marean et al.(2001), and used to determine skeletal element abundancesin several previously reported faunal assemblages (Frey andMarean, 1999; Marean et al., 2000; Marean and Frey, 1997;Marean and Kim, 1998). In the following analyses, we reportthe relationship between differential bone portion survivorshipand bone density as it is mediated by the number of hyenasand limb units included in each experimental trial. We usethe bone mineral density values for sheep previously calcu-lated by Lam et al. (1998) using computed tomography(CT). Shape-adjusted CT measurements provide the most ac-curate bone density values (Lam et al., 2003).
Table 1
Parameters of the 33 experimental samples
1 Hyena 2 Hyenas 4 Hyenas
No. of experiments 12 13 8
Using 1 limb 6 7 4
Using 2 limbs 6 6 3
Using 3 limbs 0 0 1
Total no. of limb units 18 19 13
3. Experimental results
Table 2 shows the combined abundance in MNE of long-bone portions for all elements and the levels of epiphyseal de-letion according to the numbers of hyenas and limb units ineach experimental trial. To assess variation in bone portionsurvival as a function of the number of limb units, we usechi-square tests to compare overall portion representation for1, 2 and 4 hyenas (Table 2). When one hyena is allowed tomodify the simulated sites, the overall bone portion survivalis equivalent regardless of the number of limb units includedin the experiment (c2 ¼ 0.268, p ¼ 0.992). In fact, as shownin Table 2, similar non-significant results are obtained acrossall of the experiments (2 hyenas: c2 ¼ 1.564, p ¼ 0.815; 4 hy-enas: c2 ¼ 5.156, p ¼ 0.741). For 1, 2, or 4 hyenas, the addi-tion of a second or third limb results in elevated levels ofepiphyseal and near-epiphyseal consumption to match theinput of new bones into the system. Thus, overall levels ofdestruction, as measured by the relative abundance of long-bone portions, are independent of the number of limbs. Achi-square test also indicates that combined long-bone portionabundances for experiments involving 1, 2, and 4 hyenas areindistinguishable (c2 ¼ 10.06, p ¼ 0.261).
These tests indicate that the degree of epiphyseal and near-epiphyseal deletion is independent of the number of hyenas(i.e., from 1 to 4) and the number of limb units (i.e., from 1to 3) included in the simulated sites. There is a slight increasein the levels of epiphyseal deletion between 1 hyena versus 2or 4, but this change is not significant.
MNE determinations for the three limb elements and sur-viving portions after consumption by 1, 2, and 4 hyenas arepresented in Table 3. In Figs. 2e4 we plot the long-bone por-tion abundance against Lam et al.’s (1998) determinations ofsheep bone density. To standardize the data, long-bone portionabundance is reported in these figures as %MNE, scaled to themost abundant portion. Dashed lines represent the leastsquares regression with middle shaft portions included andsolid lines represent the regression without mid-shafts. We re-port both Spearman’s and Pearson’s correlation coefficients forthe correlation between long-bone portion abundance andbone density in Table 4. Since carnivore consumption ofbone is a density-mediated process (Brain, 1967, 1969; Mareanand Spencer, 1991; Marean et al., 1992), it comes as no sur-prise that surviving long-bone portions correlate strongly andsignificantly with bone density. However, following Faith andBehrensmeyer (2006), we exclude the high density middleshafts from the analysis because the they survive well regard-less of levels of competition (Marean and Spencer, 1991) andbecause bone-consuming carnivores are primarily interestedin the grease-bearing epiphyseal and near-epiphyseal portions,especially when marrow within the medullary cavity has beenpreviously consumed by hominins. Thus, the relative survivalof middle shafts is insensitive to carnivore competition; theysurvive well whether competition is high or low. In Table 4we see that long-bone portion abundance, excluding middleshafts, correlates significantly with bone density only whenassemblages are subjected to secondary consumption by one
2029J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
Table 2
Combined MNE for all long-bone portions and the percentage of epiphyseal change (destroyed limb ends) for the different numbers of hyenas and limbs in each
experimental trial
1 hyena 2 hyenas 4 hyenas
1 limb 2 limbs Total 1 limb 2 limbs Total 1 limb 2 limbs 3 limbs Total
Chi-square values reflect comparisons of long bone portion MNE across varying numbers of limbs in experiments with 1, 2 and 4 hyenas.a p > 0.50.
hyena. When two or more hyenas are involved, bone portionsurvivorship no longer correlates with bone density. Althoughlevels of bone destruction are equivalent across experimentalsamples, the surviving limb epiphyses and near-epiphysealshaft portions only correlate with bone density when directcompetition for bone is absent.
4. Discussion
The experimental patterning of long-bone destruction pro-vides a means of interpreting the limb portions that survivecarnivore consumption in terms of competition intensity forconsumable bone. The experiments with captive spotted hy-enas indicate that when carnivores modify bone assemblages,levels of destruction are independent (within the experimentalrange) of the numbers of long-bones available for consump-tion and the degree of conspecific competition for thoseresources (Table 2). Levels of limb-end deletion certainlyreflect the degree of carnivore-mediated destruction of bonein a faunal assemblage, but gross levels of long-bonedestruction (i.e., the relative number of destroyed epiphysesand near-epiphyses) appear to be insensitive as a measure of
Table 3
Long-bone portion MNE after consumption by 1, 2 and 4 hyenas
Element Portion MNE
1 Hyena 2 Hyenas 4 Hyenas
Femur Proximal end 4 0 2
Proximal shaft 14.5 9.5 8
Middle shaft 16.6 17.5 12.5
Distal shaft 13 8.9 7.8
Distal end 3 0 0
Tibia Proximal end 4.2 1 0
Proximal shaft 13 14.5 10
Middle shaft 20 19 13
Distal shaft 14.9 12.2 5.5
Distal end 10 6 5
Metatarsal Proximal end 14.45 6 6.15
Proximal shaft 15.5 6.5 6
Middle shaft 15.8 15.7 12.3
Distal shaft 10 4 5
Distal end 8 3 2
increasing competition counter to previous suggestions (Blu-menschine and Marean, 1993).
Although the levels of long-bone destruction is comparable,given differing numbers of limbs, there is certainly a point atwhich a carnivore will have eaten its fill and will no longer con-sume additional bone; that limit was not reached in this experi-mental study. Nevertheless, it is reasonable to assume thatmultiple carnivores are capable of consuming absolutely morebone than a single individual. When faced with greater quanti-ties of bone and/or bones of larger mammals, the destructiveimpact of an individual carnivore will be much less than thatof a group of carnivores. Thus, an absence of competitionimplies few carnivores, lower potential for the destruction ofelements discarded by hominins and greater preservation ofpatterns relating to hominin behavior. In contrast, high levelsof competition imply a much larger potential for the destructionof bones and the alteration of pre-consumption taphonomicsignals. Under such a scenario, patterns of skeletal element rep-resentation relating to hominin behavior are likely to be severelyaltered or destroyed.
Concerning the relationship between carnivore competitionand the correlation between limb-end survivorship and bone
Fe1
Fe2Fe3
Fe4
Fe5Ti1
Ti2
Ti3
Ti4
Ti5
Mt1Mt2
Mt3
Mt4
Mt5
0
20
40
60
80
100
0.2 0.4 0.6 0.8 1 1.2 1.4
1 Hyena
Bone Density (g/cm3)
Fig. 2. %MNE of long-bone portion abundance vs. bone density (Lam et al.,
1998) following consumption by one hyena. Dashed line represents the regres-
sion with mid-shafts and solid line represents the regression excluding mid-
3, middle shaft; 4, distal shaft; 5, distal end (see also Table 4).
2030 J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
density, our observations provide experimental cross-valida-tion of the naturalistic data presented previously by Faithand Behrensmeyer (2006). A strong correlation betweenepiphyseal and near-epiphyseal abundance and bone densitysignals low levels of competition for food resources. The ab-sence of such a correlation is indicative of increasing compe-tition and potentially increased levels of destruction to otherelements in an assemblage. With the inclusion of middleshafts, long-bone portion abundance correlates significantlywith bone density given their high density and resistance to de-struction (Marean and Spencer, 1991; Marean et al., 1992;Pickering et al., 2003). Marean et al. (1992) show that the low-est density elements and portions are preferentially selectedfor consumption by the spotted hyena. As a result, hyenabone-choice correlates significantly with bone density. How-ever, such a correlation depends on whether the hyenas havethe opportunity to choose the least dense, grease-rich elements
Fe1
Fe2
Fe3
Fe4
Fe5Ti1
Ti2
Ti3
Ti4Ti5
Mt1 Mt2
Mt3
Mt4
Mt5
0
20
40
60
80
1004 Hyenas
0.2 0.4 0.6 0.8 1 1.2 1.4
Bone Density (g/cm3)
Fig. 4. %MNE of long-bone portion abundance vs. bone density (Lam et al.,
1998) following consumption by four hyenas. Dashed line represents the re-
gression with mid-shafts and solid line represents the regression excluding
mid-shafts (see also Table 4).
Fe1
Fe2
Fe3
Fe4
Fe5Ti1
Ti2
Ti3
Ti4
Ti5 Mt1 Mt2
Mt3
Mt4Mt5
0
20
40
60
80
1002 Hyenas
Mt5
0.2 0.4 0.6 0.8 1 1.2 1.4
Bone Density (g/cm3)
Fig. 3. %MNE of long-bone portion abundance vs. bone density (Lam et al.,
1998) following consumption by two hyenas. Dashed line represents the re-
gression with mid-shafts and solid line represents the regression excluding
mid-shafts (see also Table 4).
when feeding. Our experimental results indicate that in theabsence of competition (i.e., one hyena), this is indeed thecase and long-bone portion survivorship, excluding the middleshafts, correlates significantly with bone density. In competi-tive situations, however, the bone-consuming carnivores aremuch less selective. With two or more hyenas present, Mareanet al. (1992) report two patterns: (1) if one hyena is dominant,it will stand guard over the assemblage while others seek tograb bones and run off, and (2) if the hyenas are equal inrank, each would grab bones and run to opposite ends of theenclosure. Based on these observations, it is apparent that un-der direct competition, hyenas tend to go after whatever can bereadily acquired as opposed to what is least dense. When theydo not discriminate between high and low density elements,the correlation between epiphyseal and near-epiphyseal abun-dance and bone density becomes non-significant. Because wehave previously demonstrated that levels of bone destructiondo not vary according to the number of hyenas or limb unitsincluded in the experimental sample, this relationship betweenlong-bone survivorship and bone density must be mediated bywhich portions are actually selected for consumption as op-posed to the overall levels of destruction.
5. Application
The results of these experimental studies are particularlyrelevant to the interpretation of faunal assemblages fromPlio-Pleistocene localities where carnivores have been impli-cated in altering original hominin behavioral signals. Origi-nally, the accumulations of faunal remains and stone tools inEast African Plio-Pleistocene sites were heralded as ‘‘livingfloors’’ (Leakey, 1971) or what Isaac (1978) termed ‘‘homebases’’ or ‘‘central place foraging sites,’’ to which homininssupposedly transported and shared a variety of foods, particu-larly meat hunted from big game. Such interpretations pro-vided the basis for assigning numerous modern humanbehavioral characteristics to early Homo, including the emer-gence of nuclear families and sexual division of labor (Clark,1997; Isaac, 1978).
The traditional interpretations of the Plio-Pleistocene ar-chaeological record were strongly criticized by Binford(1981), who questioned the degree to which hominins were
Table 4
Correlations between long-bone portion abundance and bone density (see also
2031J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
responsible for the formation of the bone assemblages. Sincethen, archaeologists have focused on studying Plio-Pleistocenefaunal remains from a taphonomic perspective in order to de-velop better understandings of how these assemblages formed(e.g., Binford et al., 1988; Blumenschine, 1995; Bunn, 1986;Bunn and Kroll, 1986; Domınguez-Rodrigo, 1997, 1999;Isaac, 1983; Potts, 1988). Carnivores have been implicatedin the taphonomic histories of many Plio-Pleistocene assem-blages (Leakey, 1971; Potts, 1988), and archaeologists con-tinue to argue about the contribution of hominin versuscarnivore agents to these bone assemblages (e.g., Blumenschine,1995; Capaldo, 1997; Domınguez-Rodrigo, 1999; Selvaggio,1998).
In particular, the patterning of skeletal element representa-tion and degree of carnivore involvement at the Plio-Pleisto-cene sites of FLKN levels 1-2 and FLK-Zinjanthropus inOlduvai Gorge, Tanzania, have been the source of continuousinvestigation and debate (Binford, 1981, 1986, 1988;Blumenschine, 1995; Bunn, 1986; Bunn and Kroll, 1986,1987, 1988; Capaldo, 1997; Domınguez-Rodrigo and Barba,2006; Faith and Behrensmeyer, 2006; Marean et al., 1992;Potts, 1987, 1988; Selvaggio, 1994). Although it is notpossible to identify the species of carnivore responsible formodifying these assemblages, damage patterns (e.g., destruc-tion of long-bone epiphyses (Blumenschine and Marean,1993; Bunn, 1986; Domınguez-Rodrigo and Barba, 2006)) in-dicate specialized bone-crunching carnivores, such as hyenas.Fossil hyenas recovered from Beds I and II at Olduvai Gorgeinclude C. ultra and Hyaena sp. (Werdelin and Lewis, 2005;Margaret E. Lewis, personal communication, 12/22/2006).We assume that these carnivores would have had preferencesfor skeletal elements and long-bone portions similar to thoseof modern C. crocuta, since these preferences are related tothe nutritional value (i.e., grease content) and density ofbone. The assemblages from FLK-Zinjanthropus and FLKNlevels 1e2 are characterized by an abundance of limbs andvery low frequencies of axial elements. The relative rarity ofaxial remains at these sites may reflect hominin transport ofappendicular elements, carnivore destruction of axial ele-ments, or a combination of both (Marean et al., 1992). Faithand Behrensmeyer (2006) suggest that a poor correlation be-tween epiphyseal abundance and bone density at FLK-Zinjan-thropus indicates intense carnivore competition and highlevels of carnivore modification. Thus, our ability to make in-ferences about hominin behavior based on skeletal elementabundances at FLK-Zinjanthropus is limited. The experimen-tal results from our study support this interpretation of theFLK-Zinjanthropus archaeofauna. Here, we plot limb-endabundance of size 3 bovids provided by Potts (1988) againstLam et al.’s (1999) wildebeest (Connochaetes taurinus) bonedensity values (Fig. 5). Bone density values for metacarpalsand metatarsals are averaged as Potts reports these elementsas metapodials. Limb-end frequency is reported in minimalanimal units (MAU) after Binford (1984) to standardize theabundance of metapodials relative to the other elements. Thestrength of the correlation between limb-end abundance andbone density is not significant, which is consistent with the
experimental results under competitive scenarios (rs ¼ 0.506,p ¼ 0.135; r ¼ 0.237, p ¼ 0.510). In the following discussion,we consider the assemblages recovered from FLKN levels 1e2.
As with FLK-Zinjanthropus, the assemblage from FLKNlevels 1e2 is dominated by appendicular elements. This hasbeen interpreted as indicating hominin-transportation of meatylimb elements (Bunn, 1986). However, Marean et al. (1992)provide a cautionary note, demonstrating that this pattern ofskeletal element representation could result from carnivore de-struction of axial elements rather than hominin limb transport.Based on our experimental findings, we can re-examine thelevels of carnivore competition for the hominin-discarded as-semblages at FLKN levels 1e2. In Fig. 6, we plot limb-endabundances provided by Bunn (1986) for size categories 3and 4 mammals, the majority of which are bovids, againstthe bone density values for wildebeest reported by Lamet al. (1999). A very strong and positive correlation is apparent(rs ¼ 0.720, p ¼ 0.008; r ¼ 0.675, p ¼ 0.016), suggesting low
Fe1Fe5
Hu1
Hu5
Ti1 Ti5
Ra1
Ra5
Mp1
Mp5
0
2
4
6
8
10
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Bone Density (g/cm3)
MA
U
FLK-Zinj
rs = 0.506, p = 0.135
r = 0.237, p = 0.510
Fig. 5. Long-bone end MAU vs. bone density of wildebeest (Lam et al., 1999)
for size 3 bovids at FLK-Zinjanthropus (Potts, 1988). Mp, metapodial.
Hu1
Hu5Ra1
Ra5
Mc1Mc5
Fe1Fe5
Ti1
Ti5
Mt1
Mt5
0
10
20
30
40
50
MN
E
FLKN Levels 1-2
rs = 0.720, p = 0.008
r = 0.675, p = 0.016
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Bone Density (g/cm3)
Fig. 6. Long-bone end MNE vs. bone density of wildebeest (Lam et al., 1999)
for size 3 and 4 mammals at FLKN levels 1e2 (Bunn, 1986). Ra, radius; Mc,
2032 J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
levels of carnivore competition. Bunn’s report on the FLKNlevels 1e2 assemblage provides highly accurate MNE deter-minations because he included limb shafts in the counts (seeour Table 5). This allows us to calculate the total level ofepiphyseal destruction by carnivores and other agents at28.8%, a value much lower than the experimental values for1e4 captive hyenas.
Because the processes of hominin-discard and secondarycarnivore consumption contributing to the formation of theFLKN levels 1e2 archaeofauna took place over an unknownlength of time, our measurements reflect a time-averaged re-gime of carnivore competition that, overall, appears to be quitelow. Had axial remains initially been deposited at this site, it ishighly unlikely that such a low number of carnivores, perhapssolitary individuals acting sequentially on the assemblage,would have destroyed large quantities of axial elements fromsize 3e4 mammals. In their study of carnivore destructionof bone on the Amboseli landscape, Faith and Behrensmeyer(2006) show that only under extremely high carnivore densi-ties and elevated levels of competition are the axial remainsof size 2 and 3 ungulates thoroughly destroyed. A low estimateof limb-end deletion in tandem with the strong correlation be-tween limb-end abundance and bone density suggests that theFLKN levels 1e2 assemblage was subjected to minimal levelsof carnivore attrition and that the faunal remains recoveredfrom the assemblage likely retain a strong signal of homininbehavioral patterns relative to taphonomic overprinting causedby carnivore activity.
We conclude that the skeletal element representation atFLKN levels 1e2 is consistent with a pattern of hominin discardof meaty appendicular elements rather than carnivore destruc-tion of axial elements with consequent preferential survival ofappendicular elements. These findings contrast with the resultsof an equivalent analysis of FLK-Zinjanthropus, which appearsto have been modified by carnivores under high levels of com-petition (Faith and Behrensmeyer, 2006). It is possible that the
Table 5
Long-bone abundances of size 3/4 mammals at FLKN levels 1e2 as deter-
mined by Bunn (1986)
Element MNE
Humerus Proximal 7
Distal 33
Shafts included 33
Radius Proximal 30
Distal 17
Shafts included 30
Metacarpal Proximal 25
Distal 23
Shafts included 25
Femur Proximal 9
Distal 9
Shafts included 33
Tibia Proximal 13
Distal 41
Shafts included 41
Metatarsal Proximal 30
Distal 25
Shafts included 30
taphonomic history of FLK-Zinjanthropus may be atypicalrelative to other contemporaneous sites. FLK-Zinjanthropushas been the focus of numerous analyses, resulting in a widerange of interpretations for the implications of hominin behav-ioral patterns (Binford, 1981, 1986, 1988; Blumenschine,1995; Bunn, 1986; Bunn and Kroll, 1986, 1988; Capaldo,1997; Domınguez-Rodrigo and Barba, 2006; Faith and Behren-smeyer, 2006; Potts, 1988; Selvaggio, 1994, 1998). The diver-gent taphonomic histories of FLK-Zinjanthropus and FLKNlevels 1e2 stress the need for caution when making inferencesconcerning the evolutionary history of hominin foraging strate-gies based on the study of a single, unusually rich put possiblyatypical archaeofaunal assemblage.
6. Conclusions
Inferring hominin behavior from the patterns evident in thefossil record requires an understanding of the taphonomic pro-cesses that have mediated skeletal element representation. Ithas long been recognized that a skeletal element’s physicalcharacteristics, most notably bone density, play a central rolein the differential survivorship of bones and portions of bones(Brain, 1967, 1969). This is particularly relevant to carnivore-mediated attrition, which has played a significant role inpatterning the faunal remains from numerous assemblages ofpaleoanthropological interest. A well-developed understandingof how carnivore-mediated taphonomic signals relate to varia-tion in ecology, competition for nutrient resources and carni-vore behavioral patterns can significantly enhance our abilityto infer hominin behavior and general paleoecological condi-tions from archaeofaunal and paleontological assemblages al-tered by carnivores (Blumenschine and Marean, 1993; Gifford,1981).
In this study we have re-analyzed previous results fromcontrolled experimental observations of spotted hyenas in or-der to examine the relationships between carnivore competi-tion, bone destruction, and bone density. The overall levelsof carnivore-mediated long-bone destruction in simulatedhominid-discarded assemblages is unaffected by the numberof hyenas (1, 2 or 4) permitted to consume the bone, as wellas the quantity of bone discarded at each site. Thus an evalu-ation of overall levels of epiphyseal deletion at a faunal assem-blage does not provide insight into the number of carnivoresinvolved in modifying the bone, at least within the experimen-tal range. The experimental results also indicate that survivinglong-bone portions correlate strongly with bone density, pri-marily because high density middle shafts survive carnivoreconsumption regardless of levels of competition. The consis-tently high survival of mid-shaft portions act as leveragepoints, generating highly significant positive correlations inan otherwise uncorrelated scatter of points. It follows that inevaluating levels of conspecific competition for bone, middleshafts should be excluded from the statistical analysis becauseof their high survival potential. In support of previous observa-tions based on the modern landscape bone assemblage ofAmboseli Park, Kenya (Faith and Behrensmeyer, 2006) theexperimental results indicate that the correlation between
2033J.T. Faith et al. / Journal of Archaeological Science 34 (2007) 2025e2034
long-bone epiphyseal and near-epiphyseal abundance andbone density does provide a measure of overall levels ofcarnivore competition. Strong correlations signal low intensitycompetition because bone-consuming carnivores can preferen-tially select low density elements for consumption. This signalalso implies a lower potential for overall destruction of otherelements (e.g., axial elements) because few carnivores arecompeting for the skeletal remains. Such a pattern was onlyevident in experimental trials involving a single hyena. Anon-significant correlation, in contrast, signals elevated levelsof direct competition for bone. This provides evidence for theinvolvement of multiple carnivores in modifying an assem-blage. In such cases, behavioral signals from previous hominininvolvement with skeletal element representation are likely tobe overprinted or destroyed.
Acknowledgements
We are grateful to Donald Grayson, Margaret Lewis and theanonymous reviewers for helpful comments on previous ver-sions of this manuscript. JTF thanks the National ScienceFoundation for supporting this research under a Graduate Re-search Fellowship. CWM thanks the helpful staff at the Berke-ley Spotted Hyena Colony, Berkeley, California. LaurenceFrank was particularly helpful during the research with the hy-enas, and Stephen Glickman consented to the experiments.The captive colony was supported by the National Instituteof Health grant No. 5R01 MH 39917 to Stephen Glickmanduring the time of these studies. CWM also thanks HopeWilliams for assisting with the figure production, and theISSR staff for technical support.
References
Assefa, Z., 2006. Faunal remains from Porc-Epic: paleoecological and
zooarchaeological investigations from a Middle Stone Age site in south-
eastern Ethiopia. Journal of Human Evolution 51, 50e75.