Arribas Hiena

q 2001 The Paleontological Society. All rights reserved. 0094-8373/01/2703-0007/$1.00

Paleobiology, 27(3), 2001, pp. 512530

Taphonomic decoding of the paleobiological information lockedin a lower Pleistocene assemblage of large mammals

Paul Palmqvist and Alfonso Arribas

Abstract.The processes of fossilization have usually been perceived by paleontologists as destruc-tive ones, leading to consecutive (and in most cases irretrievable) losses of paleobiological infor-mation. However, recent developments of conceptual issues and methodological approaches haverevealed that the decrease in paleobiological information runs parallel to the gain of taphonomicinformation. This taphonomic imprinting often makes it possible to decode the fraction of paleo-biological information that was lost during fossilization, and may also contribute new data for de-ciphering paleobiological information that was not originally preserved in the assemblage, such aspaleoethology. A good example is the study of the macrovertebrate assemblage from the lowerPleistocene site at Venta Micena (Orce, southeastern Spain). Taphonomic analysis showed that thegiant, short-faced hyenas (Pachycrocuta brevirostris) selectively transported ungulate carcasses andbody parts to their maternity dens as a function of the mass of the ungulates scavenged. The frac-turing of major limb bones in the dens was also highly selective, correlating with marrow contentand mineral density. Important differences in bone-cracking intensity were related to which speciesthe bones came from, which in turn biased the composition of the bone assemblage. The analysisof mortality patterns deduced for ungulate species from juvenile/adult proportions revealed thatmost skeletal remains were scavenged by the hyenas from carcasses of animals hunted by hyper-carnivores, such as saber-tooths and wild dogs. Analytical study of the Venta Micena assemblagehas unlocked paleobiological information that was lost during its taphonomic history, and has evenprovided paleobiological information that was not preserved in the original bone assemblage, suchas the paleoethology of P. brevirostris, which differed substantially from modern hyenas in being astrict scavenger of the prey hunted by other carnivores.

Paul Palmqvist. Departamento de Geologa y Ecologa (Area de Paleontologa), Facultad de Ciencias, Uni-versidad de Malaga. 29071 Malaga, Spain. E-mail: [email protected]

Alfonso Arribas. Museo Geominero, Instituto Geologico y Minero (I.G.M.E.), c/ Ros Rosas, 23. 28003Madrid, Spain. E-mail: [email protected]

Accepted: 13 March 2001

Introduction

Prior to the 1980s most vertebrate taphon-omists emphasized the incompleteness of thefossil record, because the processes of fossil-ization were envisioned as destructive, lead-ing to loss of paleobiological information. Asa result, taphonomy came to be associatedwith the documentation of information lossand bias in the transition of organic remainsfrom the biosphere to the lithosphere. How-ever, in their important paper Behrensmeyerand Kidwell (1985) envisioned taphonomy asthe study of processes of preservation andhow they affect information in the fossil rec-ord. They were following the approach of nu-merous invertebrate paleontologists who wereengaged since the mid-1980s in more posi-tive aspects of comparative taphonomic re-search intended to establish that postmortemprocesses (e.g., weathering, transport, and

sorting) leave signatures that are useful anddiagnostic of various paleoenvironmental andsedimentary conditions (Kidwell and Bosence1991; Kidwell and Flessa 1995). Additionally,time-averagingviewed negatively by mostpaleontologists in the pastis widely recog-nized as advantageous, because short-termecological noise is dampened and longer-term signals from a biological community arepreserved. In fact, bone assemblages from sur-face environments are considered comparablein some respects to repeated ecological sur-veys in assessing the long-term dynamics ofthe potentially preservable fraction of terres-trial communities (for review and references,see Behrensmeyer and Hook 1992; Cutler et al.1999; Martin 1999).

Our analysis of large mammals preservedat Venta Micena shows that it is possible to re-cover significant paleobiological informationfrom a taphonomically altered assemblage.

513BEHAVIOR OF AN EXTINCT HYENA

Such information is obtainable from quanti-tative study of the preservational bias intro-duced by the behavior of the large, extinct hy-ena Pachycrocuta brevirostris, the bone-collect-ing agent at this site (Palmqvist et al. 1996; Ar-ribas and Palmqvist 1998). Until now little hasbeen known about the relative importance ofhunting and scavenging for this extinct bone-cracking carnivore, its role as a bone-accu-mulating agent during the lower Pleistocene,and the bias it introduced in the compositionof the assemblages of large mammals. Herewe evaluate the nature and consequences ofsuch bias for the composition of the Venta Mi-cena assemblage, paying special attention toseveral aspects not reported in detail before,such as the transport by hyenas of carcassesand bone remains to their maternity dens andthe differential breakage of limb bones fromvarious ungulate species.

The Venta Micena Site

Venta Micena (Orce, Granada, southeasternSpain) is located in the eastern sector of theGuadix-Baza intramontane basin. The basinwas endorheic (i.e., characterized by interiordrainage) until late Pleistocene times, thus fa-cilitating an exceptional record of Plio-Qua-ternary taphocenoses of large mammals pre-served in swampy and lacustrine sediments(Fig. 1A). This site is dated by biostratigraphyto the early Pleistocene, with an estimated ageof 1.3 6 0.1 Ma (Arribas and Palmqvist 1999).The 80120-cm-thick Venta Micena stratum(VM-2, Fig. 1A) is one of the various fossilif-erous units in the Plio-Pleistocene sedimen-tary sequence of Orce, whose surface can befollowed along ;2.5 km and stands out to-pographically in the ravines of the region.This stratum has the following vertical struc-ture from bottom to top (Arribas and Palm-qvist 1998: Fig. 2):

1. A basal unit (first lacustrine stage) that isone-third to one-half the total thickness,formed by homogeneous micrite with somecarbonate nodules (520 cm thick) andsmall mud banks. The sediment preservesabundant shells of freshwater mollusks andis sterile in vertebrate fossils, thus attestingto a first generalized lacustrine stage in the

region, in which the micrite was precipitat-ed in water of variable depth; the absenceof pyrite and carbonate facies rich in organ-ic matter are evidence that the lake was noteutrophic.

2. A 415-mm-thick calcrete paleosol (hard-pan) developed on the surface of the mi-crite sediments deposited during the pre-vious lacustrine stage. The calcrete formsan irregular surface, subparallel to the bed-ding plane, following the preexisting limn-ic microtopography, and is thicker at to-pographic highs. This surface defines astratigraphic unconformity, indicating amajor drop of the Pleistocene lake level andthus the emergence of an extensive plainaround the lake.

3. An upper unit of micrite (second lacustrinestage) deposited in a subsequent rise of thelake level, which continues up to the top ofthe stratum, showing root marks, mudcracks, and a high density of fossil bones oflarge mammals resting on the paleosol.

The sedimentary environment of the fossilassemblage was characterized by wideemerged zones (;4 km width) around thelake, with small shallow ponds (,1 m depth,220 m diameter) (Arribas 1999). The bonesare embedded in a porous micrite matrix (9899% CaCO3) with mud cracks and root marks,which precipitated during a period of partialexpansion of the ponds (i.e., restrictedswampy biotope of carbonate facies, withplants colonizing the border of the ponds). Itis capped by a massive micritic limestone,produced during a subsequent phase of waterlevel rise (i.e., second lacustrine stage) thatwas rather slow, as indicated by the absence ofterrigenous, erosive structures and of any ev-idence of sediment traction.

The Venta Micena quarry has an area of;300 m2 (Fig. 1B,C). This surface was dividedin square meters and excavated systematicallyfrom 1979 to 1995, providing a rich collectioncomposed of 5798 identifiable skeletal re-mains from 225 individuals belonging to 19taxa of large ($5 kg) mammals, 655 anatom-ically identifiable bones of mammals thatcould not be determined taxonomically (e.g.,diaphyses and small cranial fragments), and

514 PAUL PALMQVIST AND ALFONSO ARRIBAS

FIGURE 1. A, Geographic location of the paleontological site at Venta Micena (Orce, Granada, southeastern Spain)in the intramontane basin of Guadix-Baza, and stratigraphic section of the lower Pleistocene deposits in the VentaMicena (VM) sector. The location of two other paleontological and archaeological sites (BL 5 Barranco Leon, FN5 Fuente Nueva) is also indicated. B, Bones outcropping at high density in one grid of the surface excavated atVenta Micena. C, Density plot for the abundance of skeletal remains per square meter (z-axis) in the quarry.

;10,000 unidentifiable bone shafts. Completeelements and bone fragments range in sizefrom isolated premolars and third phalangesof Vulpes to complete mandibles of Mammu-thus. Fossil remains of micromammals, includ-ing teeth and elements from the axial skeleton,are also present in small numbers and werenot included in the taphonomic study; theywere probably deposited as fecal droppings ofsmall carnivores. Table 1 summarizes the rawdata on the abundances of large-mammaltaxa.

The longitudinal axes of major longbonesshow no preferred orientation, which suggeststhat the bones were not aligned by current.

The stratigraphy also indicates the absence ofchanneled currents in the area in which fossilswere accumulated. The bones lie horizontallyon the paleosurface, and there is no evidenceof trampling, as no skeletal element was foundin vertical or diagonal position (Arribas 1999).Surfaces of the bones are well preserved; nosigns of abrasion or polish are present andonly four elements show evidence of slightdissolution. The concentration of fossils on theexcavated surface is very high, with a meandensity of elements of ;60 bones/m2 (Fig.1C), and .90% of skeletal elements are in con-tact with other bones. Two areas had 8090bones/m2 of up to 50 cm in length, such as tib-


TABLE 1. Abundance of taxa of large mammals ($5 kg) identified in the Venta Micena assemblage (data from Ar-ribas and Palmqvist 1998). MNI 5 minimum number of individuals (juveniles/adults), based on counts of decid-uous and permanent teeth. NISP 5 number of identifiable specimens (teeth/bones). C 5 carnivore (Hy 5 hyper-carnivore, .70% vertebrate flesh in diet; Co 5 carnivore/omnivore, ,70% flesh in diet; Bc 5 bone cracker). H 5herbivore (Br 5 browser, ,10% grass in diet; Mf 5 mixed-feeder, 1090% grass; Gr 5 grazer, .90% grass). O 5omnivore.

SpeciesTrophichabits

MNI(juv./adults)

%Juv.

NISPtotal(teeth/bones)

%Total

NISPsample(teeth/bones)

%Sam-ples

Mass ofadults(kg)

Mammuthus meridionalisHippopotamus antiquusBovini cf. DmanisibosSoergelia minorPraeovibos sp.Hemitragus albaCaprini gen. et sp. indet.Eucladoceros giuliiDama sp.Stephanorhinus etruscusEquus altidensVulpes praeglacialisCanis falconeriCanis etruscusLynx aff. issiodorensisMegantereon whiteiHomotherium latidensPachycrocuta brevirostrisUrsus etruscus

H (Mf)H (Gr)H (Gr)H (Mf)H (Gr)H (Gr)H (Mf)H (Br)H (Mf)H (Br)H (Gr)C (Co)C (Hy)C (Co)C (Hy)C (Hy)C (Hy)C (Bc)O

5 (4/1)5 (3/2)

27 (16/11)13 (3/10)1 (0/1)

14 (2/12)1 (0/1)

36 (15/21)20 (3/17)6 (2/4)

70 (32/38)1 (0/1)3 (0/3)4 (0/4)1 (0/1)3 (0/3)2 (0/2)

10 (4/6)3 (1/2)

80.060.059.323.1

0.014.3

0.041.715.033.345.7

0.00.00.00.00.00.0

40.033.3

48 (16/32)58 (19/39)

775 (382/393)334 (215/129)

6 (3/3)305 (209/96)

1 (0/1)962 (557/405)417 (231/186)90 (55/35)

2562 (1183/1379)24 (19/5)65 (40/25)33 (20/13)6 (2/4)

16 (7/9)7 (6/1)

62 (34/28)27 (15/12)

0.81.0

13.45.80.15.30.0

16.67.21.6

44.20.41.10.60.10.30.11.10.5

21 (4/17)14 (0/14)97 (48/49)74 (51/23)3 (2/1)

30 (12/18)1 (0/1)

121 (70/51)55 (30/25)27 (16/11)

457 (210/247)12 (10/2)33 (21/12)16 (10/6)3 (1/2)8 (3/5)3 (2/1)

31 (17/14)14 (8/6)

2.11.49.57.30.32.90.1

11.95.42.6

44.81.23.21.60.30.80.33.01.4

60003000

450225320

7510

38095

1500350

5301010

100250100375

iae of Equus and metapodials of Eucladoceros(Arribas and Palmqvist 1998: Figs. 7, 8). Ar-ticulated bones are relatively scarce, repre-senting less than 20% of all elements in thesample; however, there is a low degree of hor-izontal dispersion, with abundant groups ofdisarticulated but associated elements, such asskulls with mandibles and metapodials withphalanges. The most frequently preserved ar-ticulations are those formed by tibiae-tarsal-metatarsal-phalanges, humerus-radius/ulna,radius-carpal-metacarpal-phalanges, and ver-tebrae.

The age estimated for individuals pre-served in the assemblage included two majorgroups: immature or juvenile individualswith deciduous teeth, and adults with fullyerupted permanent dentition (Table 1). Bodymass estimates for adults were obtained fromPalmqvist et al. (1996), who used taxon-freeregression equations of mass on craniodental/postcranial measurements from modern spe-cies (Damuth and MacFadden 1990).

Inspection of data in Table 1 shows that her-bivore taxa dominate the assemblage in bothnumber of identifiable specimens (NISP) and

estimates of minimum number of individuals(MNI). More common herbivorous species(those with higher NISP and MNI values),such as the horse Equus altidens and the largedeer Eucladoceros giulii, have high percentagesof juveniles, .40% in both cases (32/70 and15/36, respectively). Among carnivores, onlyadult individuals are recovered, with the ex-ception of the hyaenid and the ursid. Fortypercent (4/10) of the individuals of P. breviros-tris are juveniles, represented by deciduousteeth contained within the maxilla or mandi-ble, indicating that these cranial elementswere produced not by tooth replacement butas a consequence of the death of immature in-dividuals.

An Overview of the Taphonomy of VentaMicena

Previous research on the taphonomy of Ven-ta Micena (Palmqvist et al. 1996; Arribas andPalmqvist 1998; Arribas 1999; Palmqvist andArribas 2001) focused on the analysis of size/abundance patterns in ungulate species usingthe model of Damuth (1982), and on the abun-dance of preserved epiphyses and complete


TABLE 2. Abundance of skeletal elements of large mammals grouped according to their potential for water dis-persal (Voorhies groups) in the subset used for taphonomic analysis (n 5 1231), and in the three better-representedtaxa in the assemblage, the horse (Equus altidens; n 5 488), the buffalo (Bovini cf. Damanisibos; n 5 95), and themegacerine deer (Eucladoceros giulii; n 5 138).

Voorhiesgroups Skeletal element ntotal % nhorse % nbuffalo % ndeer %

Group I Isolated teethFragments of deer antlersVertebraeRibsScapulaeUlnaeCalcaneiAstragaliPhalanges

16819

1783631

8365172

13.71.5

14.52.92.50.62.94.15.9

47511612

3142338

9.6

10.53.32.50.62.94.77.8

321

0208

107

3.2

22.10.02.10.08.4

10.57.4

131413

011

0885

9.410.1

9.40.08.00.05.85.83.6

Group II HumeriRadiiPelvis fragmentsFemoraTibiaeMetapodials

7834233287

258

6.32.81.92.67.1

21.0

247

122547

145

4.91.42.55.19.6

29.7

122207

11

12.62.12.10.07.4

11.6

9720

1225

6.55.11.50.08.7

18.1

Group III Cranial elements 120 9.7 24 4.9 10 10.5 11 8.0

limb bones of ruminants. The results obtainedindicated that most losses of paleobiologicalinformation during the taphonomic history ofthe assemblage were a consequence of the se-lective destruction of skeletal remains duringthe period when the bones were exposed onthe surface before burial, and that the effect ofthis preservational bias was more pronouncedin those species of smaller body size (Arribasand Palmqvist 1998; Arribas 1999).

The role of hyenas in the bone accumulationprocess at Venta Micena was determined bycomparing the frequencies of different typesof postcranial bones in this assemblage (e.g.,vertebrae, ribs, limb and girdle bones, phalan-ges) with the corresponding figures for sev-eral recent and archaeological deposits accu-mulated by carnivores, rodents and hominids(Arribas and Palmqvist 1998: Table 2). Resultsindicated that P. brevirostris was the mainagent responsible for the bone accumulationat Venta Micena, because the composition ofthe fossil assemblage is strikingly similar tobone accumulations produced by modern hy-enas in which major limb bones predominatewhereas ribs and vertebrae are comparativelyscarce. Specifically, the relative abundances oflimb bones and vertebrae/ribs in Venta Mi-cena (79.3% and 14.3%, respectively) are sim-ilar to the frequencies of these elements in as-

semblages collected by spotted hyenas (Cro-cuta crocuta), 69.476.2% and 12.224.2%, re-spectively (Behrensmeyer and Dechant Boaz1980; Brain 1981; Skinner and Van Aarde 1981;Bunn 1982; Skinner et al. 1986), but differentfrom those accumulated by striped hyenas(Hyaena hyaena) and brown hyenas (Parahyaenabrunnea), in which higher frequencies of limbbones (90.893.2%) and lower frequencies ofvertebrae/ribs (4.27%) are found (Maguire etal. 1980; Skinner et al. 1980, 1995; Skinner andVan Aarde 1981, 1991; Kerbis Petherhans andKolska-Horwitz 1992). However, Venta Mi-cena resembles assemblages from dens ofbrown and striped hyenas in its high densityof bones (Leakey et al. 1999). Spotted hyenasare efficient hunters, owing to their greaterbody size and strong social behavior, and pro-duce a highly enriched milk; thus, they do notregularly carry carrion to their maternity densto feed their cubs (Kruuk 1972; Ewer 1973;Mills 1989). Nonetheless, there are some re-ported cases of spotted hyenas accumulatinghuge amounts of skeletal remains (e.g., seeHill 1981 for a dense accumulation of boneswithin a breeding den in Amboseli NationalPark, Kenya), and this was certainly the casewith the cave hyena (Crocuta crocuta spelaea) inthe late Pleistocene of Europe (Fosse 1996).The differences in composition between the


assemblage accumulated by Pachycrocuta atVenta Micena and those collected by otherspecies suggest that the life habits of the short-faced hyenas were not identical to those of anyliving hyaenid.

The quantitative study of differential pres-ervation of major limb bones of ruminants inthe assemblage (Palmqvist et al. 1996; Arribasand Palmqvist 1998) showed that bone-gnaw-ing and -crushing behavior by hyenas of thoseruminant carcasses transported to the mater-nity den resulted in the preferential consump-tion of longbone epiphyses with high fat con-tents, and thus the differential breakage ofmajor limb bones according to their marrowyields. Such a selective pattern rules out thepossibility that other processes (e.g., ungulatetrampling) were responsible for bone fractur-ing.

The recovery of several deciduous teeth of P.brevirostris belonging to four individuals alsosuggests that the assemblage originatedthrough accumulation of skeletal parts nearshallow breeding dens excavated by the hye-nas in the plains surrounding the Pleistocenelake (Arribas and Palmqvist 1998). The abun-dance of unworn deciduous hyena teeth rulesout the possibility that bones were accumu-lated in open feeding places located at hunt-ing sites distant from maternity dens, if wepresume that, like modern hyenas, the cubsdid not accompany adults on their search forungulate carcasses.

A comparison of the Venta Micena assem-blage with those from other Plio-Pleistocenelacustrine sites from the Guadix-Baza Basin(Arribas 1999) revealed that Venta Micenashows the highest diversity of large mam-mals, mainly because of the high diversity ofcarnivores, from opportunistic scavengers tolarge predators. The taxonomic richness oflarge mammals at Venta Micena (19 taxa) issimilar to that recorded from a modern spot-ted hyena den developed on a calcrete paleo-sol in Amboseli (18 taxa) (Hill 1981). Similarly,Leakey et al. (1999) identified 15 species ofmammals from skeletal remains collected bystriped hyenas in Lothagam, Kenya, and Skin-ner et al. (1991, 1998) identified 11 mammali-an taxa at two maternity den sites of brownhyenas from the west coast of Namibia. These

counts of taxonomic richness from hyena ac-cumulations accurately reflect the composi-tion of the available vertebrate prey commu-nity in these areas.

Biostratinomic Analysis of the Assemblage

We have used here several biostratinomicvariables to further characterize the bone as-semblage from Venta Micena, following inpart the procedure described by Behrensmey-er (1991) for studying vertebrate assemblages.Descriptive analysis was based on a subset of1339 specimens, which includes 1020 identi-fiable skeletal remains (distributed amongtaxa in Table 1) as well as 211 bones and 108bone shafts that could not be determined tax-onomically. This sample comprises the well-restored specimens housed at the Museum ofPaleontology of Orce (Palmqvist et al. 1996).The high value obtained for the Pearson prod-uct-moment correlation between the relativefrequencies of taxa in the whole assemblageand in the subset (Table 1; r 5 0.985; p ,0.0001) indicates that the latter represents arandom sample of the entire assemblage.

Table 2 shows the abundances of differentskeletal elements in the subset of large mam-mals and in the three most abundant taxa, thehorse (E. altidens), the buffalo (Bovini cf.Dmanisibos), and the megacerine deer (E. giu-lii).

The ratio of isolated teeth to vertebrae (0.94:1) is close to the value expected in the absenceof hydrodynamic sorting (1:1), indicating thatthe skeletal remains were not transported byfluvial processes prior to deposition (Behrens-meyer and Dechant Boaz 1980; Shipman1981). The frequencies of bones grouped ac-cording to their potential for dispersal by wa-ter (i.e., Voorhies groups) are as follows:48.6% for Group I (isolated teeth, deer antlers,vertebrae, ribs, scapulae, ulnae, calcanei, as-tragali, phalanges), 41.7% for Group II (fem-ora, tibiae, humeri, metapodials, pelvis, radii),and 9.7% for Group III (cranial elements);such a degree of skeletal completeness rulesout the possibility of hydraulic sorting (Voor-hies 1969).

Limb elements clearly dominate (57.7%) thesample, followed by vertebrae, cranial ele-ments (cranial vaults, maxillae, and mandi-


bles; Fig. 2GI), and ribs. Scapulae are mostlyrepresented by proximal fragments. Diaphy-ses and distal epiphyses predominate amonghumeri (Fig. 2A). Femora are mainly pre-served as fragments of diaphyses, and tibiaeas distal epiphyses (Fig. 2B). The pelvis is rep-resented only by fragments that preserve theacetabulum.

Analysis of weathering stages for the bonesin the subset indicates exposure to the ele-ments for only a relatively short time: 89.3%(784/878) of the skeletal elements showweathering stage 0 (Behrensmeyer 1978) andonly 10.7% of the bones (of which two-thirdsare metapodials) show weathering stage 1,with a few, shallow, small split-line cracks dueto insolation (Fig. 2C) and without flaking ofthe outer surface (Arribas and Palmqvist 1998;Arribas 1999). Although low degrees of phys-ico-chemical weathering could reflect protec-tion by vegetation in moist conditions untilburial, this was not the case here because mostbones show no evidence of root marks. On theother hand, bones that were preserved com-plete lack sedimentary infilling, even in areasof the medullary cavity that are close to nu-trient foramina, indicating that they were bur-ied in fresh condition, with the periosteum in-tact (Arribas and Palmqvist 1998). Thus, theseresults suggest a very short period of subaer-ial exposure before burial (less than one yearin most cases).

The detailed study of horse remains hasshown that biostratinomic fractures are veryabundant (Fig. 2), as only 29.1% of major limbbones are complete; metapodials are the mostabundant bones preserved as complete ele-ments, 82.2%. Among the fractured elements,type II spiral fractures (Shipman 1981; Lyman1994) predominate (100% of fragmented hu-meri, femora, and radii; 74.4% of tibiae). Othertypes are longitudinal fractures in tibiae, un-differentiated fractures (all ribs and vertebrae,with the exception of some vertebrae that lackonly apophyses), and maxillary bones withboth cheek-tooth rows (33.3% of cranial ele-ments). Gnaw-marks are very frequent on thehorse remains: all cranial fragments, scapulae,humeri, radii, pelves, femora, and tibiae showstriations and gnaw-marks produced by car-nivores; the preserved epiphyses have fur-

rows and punctures; and the diaphyses, aswell as the skull bones, show scoring and pit-ting. These marks are also observed in all oth-er taxa identified at Venta Micena. Coprolites(36 cm thick) are relatively common.

Evaluation of Taphonomic Bias in theAssemblage

The taphonomic analysis of the large-mam-mal assemblage preserved at Venta Micenahas revealed the existence of the followingpreservational biases, which took place con-secutively during the biostratinomic stage andaffected its original composition (Palmqvist etal. 1996; Arribas and Palmqvist 1998; Palm-qvist and Arribas 2001): (1) scavenging by hy-enas of ungulate prey hunted by hypercarni-vores; (2) selective transport of carcasses andbone remains to their maternity dens; and (3)differential breakage of major limb boneswithin the dens. In the following sections weevaluate the importance of these biases andtheir consequences for the composition of theassemblage, with a special focus on the trans-port of carcasses and the breakage of bonesfrom horse (E. altidens) and buffalo (Bovini cf.Dmanisibos), which are two of the better-rep-resented ungulate taxa in the assemblage.

Bias I: Scavenging of Ungulate Prey SelectivelyHunted by other Predators. Previous researchon the composition of the bone assemblage(Palmqvist et al. 1996; Arribas and Palmqvist1998) has shown that the overwhelming ma-jority of skeletal remains preserved in VentaMicena were scavenged by hyenas from car-casses of ungulates preyed upon by hypercar-nivores (i.e., species in which vertebrate fleshrepresented .70% of diet). The selection byhypercarnivores of specific ungulates was ba-sically a function of differences in the bodymass of the preybetween juveniles andadults as well as between the sexes.

The evidence of prey selection at Venta Mi-cena is the following: (1) U-shaped (i.e., bi-modal) attritional mortality profiles deducedfrom crown height measurements for thoseherbivore species that are well represented inthe assemblage, indicating a strong selectionby predators of very young and old individ-uals (Palmqvist et al. 1996: Fig. 8); (2) the in-terspecific analysis of the relative abundance


of juveniles with deciduous teeth, and adultswith permanent dentition, which shows thatjuveniles represent 16.7% (8/48) of all indi-viduals in ungulate species ,300 kg, yet theproportion of juveniles increases to 48.0%(72/150) in those species .300 kg (these per-centages are significantly different accordingto a one-tailed t-test: t 5 4.63; p , 0.0001); (3)the presence of many metapodials with severeosteopathologies (Palmqvist et al. 1996: Fig.11A,B), such as arthrosis, which limited thelocomotor capabilities of the ungulates andtherefore their ability to escape from preda-tors; and (4) the sex ratio deduced from thesize distribution of metapodials in large preyspecies, such as horse and buffalo, which is bi-ased in favor of females in both cases (ap-proximately 1 male : 34 females [Palmqvist etal. 1996: Fig. 11C]). This sex ratio suggests thatfemales were more vulnerable to predationbecause of their smaller body size.

Given that most carnivores usually huntherbivores within a narrow range of bodymass around the same size as that of the pred-ator (Kruuk 1972; Schaller 1972 and referencestherein), the wide range of body mass repre-sented by the ungulate taxa preserved in theassemblage (106000 kg) suggests that in mostcases these animals were preyed upon by dif-ferent carnivore species (Palmqvist et al.1996).

Hypercarnivores are represented in the as-semblage by four speciestwo saber-tooths(Homotherium latidens and Megantereon whitei),a felid (Lynx aff. issiodorensis), and a wild dog(Canis falconeri).

M. whitei (Fig. 3) had an intermediate bodysize (;100 kg), similar to that of a jaguar,Panthera onca (Martnez-Navarro and Palm-qvist 1995, 1996). Judging from the low valueestimated for the brachial index (i.e., radiuslength : humerus length, ;80%) it was an am-bush predator, hunting in closed, forestedhabitats and presumably preying on browsingand mixed-feeding ungulates of intermediateto large body mass (Lewis 1997). This recon-struction of its predatory behavior is corrob-orated by the fact that the metapodials werecomparatively shorter than those of largemodern felids and other saber-tooths. Thisdirk-toothed machairodont had a strong body

with a short back, powerfully developed fore-limbs with large claws, and extremely long,sharp, laterally compressed (and inherentlyfragile) upper canine teeth. The brain wassmall in relation to Homotheriums, showing ol-factory lobes that were well developed. Allthese features give the strong impression of ananimal built for capturing prey using a shortrush and then using its considerable strengthto bring down and hold prey with the fore-limbs, before killing with a slashing bite to thethroat (Turner and Anton 1998; Arribas andPalmqvist 1999). A similar hunting behaviorwas inferred by Anyonge (1996) for the closelyrelated genus Smilodon, a possible descendantof Megantereon in the New World.

Homotherium (Fig. 3) was a scimitar-toothedmachairodont with relatively long and slenderlimbs, which provided considerable leverage(Turner and Anton 1998; Martin et al. 2000).According to regressions of body massagainst postcranial measurements in moderncarnivores (Anyonge 1993), it was similar insize to a modern male lion, Panthera leo (150220 kg). The regression of body mass on lowercarnassial length in modern felids (Van Val-kenburgh 1990), however, provides a largersize estimate for the Venta Micena H. latidens(Palmqvist et al. 1996), 250 kg. The upper ca-nines were comparatively shorter and broaderthan those of Megantereon, bearing coarse ser-rations in the enamel of the posterior margin.The forelimb was more elongated than thehindlimb, indicating that the animal probablyhad a sloping back. The claws of Homotheriumappear to have been small, with the exceptionof a well-developed dewclaw in the first digitof the forefoot. The elongated forelimb andsmaller claws suggest increased cursorialityand less prey-grappling capability than othersaber-tooths (Rawn-Schatzinger 1992; Turnerand Anton 1998; Arribas and Palmqvist 1999).Both the comparatively high brachial index($100% [Lewis 1997]) and the results ob-tained by Anyonge (1996) in a multivariateanalysis of the postcranial skeleton of extantand extinct felids, indicate that Homotheriumwas a pursuit predator, which presumablyhunted very large grazing and mixed-feedingungulates in open habitats. Homotherium had alarge brain relative to other saber-tooths, with


FIGURE 2. Selected examples of equid bones from Venta Micena, with evidence of modification by hyenas: A, B,Humeri and tibiae, respectively, showing gnawing of epiphyses, spiral and longitudinal fractures. C, Third meta-tarsals, complete and fractured, showing longitudinal and spiral fractures made by hyaenid crushing, and orthog-


FIGURE 3. Skulls and reconstructions of the life appearance of the three largest carnivore species preserved at VentaMicena, the dirk-tooth Megantereon whitei, the scimitar-tooth Homotherium latidens, and the giant hyena Pachycrocutabrevirostris. All drawn to scale, with a typical height at shoulder of 110 cm for Homotherium. Specific coat patternsare unknown but typical of those seen across the range of living felids and hyaenids. Drawings by Mauricio Anton.

onal diagenetic fractures in the diaphyses due to sediment compaction. D, Astragali. E, Calcanei; one calcaneumshows marks made by insect larvae. F, Third phalanges; one phalanx is partially gnawed by hyenas. G, Maxillaegnawed by hyenas, indicating an extreme destruction of the splancnocranium. H, I, Mandibles gnawed and frag-mented by hyenas. Typical sequences of bone modification by hyenas for postcranial elements of Equus (Arribas1999; Arribas and Palmqvist 1998) are also indicated.


an enlargement of the optic center, a conditionsimilar to that of the cheetah, Acinonyx jubatus(Rawn-Schatzinger 1992). Turner and Anton(1998) suggest that such a cursorial lifestyleand hunting strategy would imply some de-gree of group activity to bring down and re-strain prey. In addition, given that a pursuitstrategy for hunting can be deployed only inrelatively open terrain, group behavior may beneeded to repel the inevitable attention ofscavengers. The likelihood of group activity issuggested by the similarly proportionedAmerican species, H. serum, which is knownin some numbers (NISP .250, MNI 5 33)from the late Pleistocene site of Friesenhahncave, Texas. At this site, H. serum is associatedwith numerous remains (NISP .900 [Rawn-Schatzinger 1992: Table 38]; MNI 5 34 [Ma-rean and Ehrhardt 1995: Fig. 1]) of mam-mothsone adult and the remainder juve-nilesand it has been suggested that success-ful predation of mammoths most likely wouldrequire group hunting.

C. falconeri was a hypercarnivorous canid of;30 kg, according to the results of multivar-iate analysis and multiple regression of bodymass on craniodental measurements in mod-ern canids (Palmqvist et al. 1999). The secondmetacarpal has a very reduced articular facetwith the first metacarpal, which indicates thatthe latter bone was vestigial if not absent, acondition similar to that of African painteddogs (Lycaon pictus); this suggests increasedcursoriality for C. falconeri. This predatorprobably hunted small to medium-sized graz-ing ungulates (50300 kg) in open to inter-mediate forested country.

Pachycrocuta brevirostris (Fig. 3) was a bone-cracking carnivore with a body 1020% largerthan the modern spotted hyena and was welladapted for dismembering carcasses and con-suming bone (Palmqvist et al. 1996; Turnerand Anton 1996; Arribas and Palmqvist 1998;Saunders and Dawson 1998). Apart from itssize, this short-faced hyena differed from oth-er species in the relative shortening of its dis-tal limb segments (Turner and Anton 1996):the brachial index is close to 88%, whereas inmodern hyaenids the values range between99% and 106%; the crural index (i.e., tibialength : femur length) is 74%, whereas the cor-

responding figures for modern species rangebetween 80% and 89%. These differences sug-gest a less cursorial lifestyle for P. brevirostris.It is also possible that the shortening of thedistal limb segments could provide greaterpower and more stability to dismember andcarry large pieces of carcasses, which perhapscould be obtained from aggressive scavenging(i.e., kleptoparasitism).

The proportion of juveniles in a populationof a given species depends on two factors(Palmqvist et al. 1996): the reproduction rate(i.e., the annual birthrate) and the duration ofinfancy (i.e., the time spent as a juvenile in-dividual). Rates of birth and death scale to the20.3 power of adult body mass (M), whereasgeneration time (measured by life expectancyat birth, duration of infancy, or age at death)is interspecifically related to body mass by apower of 0.3 (Damuth 1982; Peters 1983; Cal-der 1984). As a result, larger species have low-er birth and mortality rates per unit of abso-lute time but not per unit of biological time(i.e., relative to maximum life span), becauserates of birth and death per generation are sizeindependent. A third factor, differences inage-specific mortality rate, could be relevanthere, as high adult mortality would increasethe proportion of juveniles in the populationand vice versa. However, data on cohort anal-ysis and survivorship curves for African her-bivores ranging in adult body mass between,50 kg and .3500 kg (Western 1979, 1980)show no differences among species in the age-specific mortality rate (e.g., the life expectancyat birth fluctuates around 30% of total lifespan, with small variations not related tobody size). Moreover, several studies on thetiming of ontogeny in eutherian mammals(summarized in Peters 1983) indicate that, re-gardless of size, a given developmental phaserequires a constant proportion of the mam-mals life; thus, the relative time spent as a ju-venile individual does not scale with bodymass.

The ratio of juvenile to adult individuals ina population would be the product of annualbirthrate (Br) and duration of infancy (Di):

20.3 0.3 0% juveniles ; B D 5 M M 5 M .r i

This relationship implies that the proportion


TABLE 3. Differences between primary bone assemblages collected by predators, such as leopards, and non-pri-mary, secondary assemblages accumulated by scavenger carnivores, such as hyenas (Maguire et al. 1980; Richardson1980; Skinner et al. 1980, 1986, 1995; Skinner and Van Aarde 1981, 1991; Vrba 1980; Brain 1980, 1981; Hill 1981;Shipman 1981; Klein and Cruz-Uribe 1984; Behrensmeyer 1991; Kerbis Petherhans and Kolska-Horwitz 1992;Palmqvist et al. 1996; Arribas and Palmqvist 1998). Data for Venta Micena also shown (a 5 estimated from thewhole collection, b 5 estimated from the subset used for taphonomic analysis).

Characteristics of thebone assemblage

Primary assemblage,collected by predators

Secondary assemblage,collected by scavengers

Venta Micenaassemblage

Proportion of vertebrae andribs in relation to girdle andlimb bones

High, 1:4 (range 5 1:35)

Low, 1:9 (range 5 1:4.525) 16.9% (429/2544)a

Abundance of articulatedbones, in anatomical connec-tion

Articulated elementsare quite abundant

Articulated bones are scarce(exceptions: metapodialsand phalanges, vertebrae)

20.0% (204/1020)b

Abundance of major longbonespreserved complete

High and not relatedto their marrowcontent

Low, inversely related tomarrow yield; spiral andlongitudinal fractures areabundant

27.6% (137/497)b

Abundance of limb boneepiphyses in relation to di-aphyses

High (2:1), withoutpreferential destruc-tion of skeletal partsof low structuraldensity

Comparatively low (1.51:1), with evidence of pref-erential consumption oflow-density epiphyses

139.4% (693/497)b

Carnivore/ungulate index, cal-culated from MNI counts

High (2550%) or veryhigh (.50%, indeath traps)

Low (515%), similar to thatfound in modern commu-nities

13.6% (27/198)a

Relative abundance of juvenileungulates, with deciduousteeth

High proportion(.25%)

Low proportion (,25%) 40.4% (80/198)a

Proportion of young/adult in-dividuals for ungulate spe-cies

Increases as a functionof species bodymass

Not related with the size ofspecies

Positively corre-lated with spe-cies mass

Range of body mass coveredby the species preserved inthe assemblage

Narrow, usuallyaround the samesize as that of thepredator

Wide, in general more thantwo orders of magnitude(from ,10 kg to .1000kg)

56000 kga

Richness of species (largemammals)

Comparatively low(only prey species)

High diversity (all scav-enged species)

19a

FIGURE 4. Least-squares regression analysis of the pro-portion of juvenile individuals (estimated from MNIcounts) on adult body mass (in kg) for ungulate species(n 5 9) of the Venta Micena assemblage (data from Table1). Separate analyses were conducted for two groups ofprey species, the first of which (,1000 kg of estimatedmass for the adult individuals) were presumably hunt-ed by Megantereon whitei and Canis (Xenocyon) falconeri,and the second one (.1000 kg) by the large saber-toothHomotherium latidens.

of juveniles in any ungulate population is ap-proximately constant (3040%) and indepen-dent of species body size (Palmqvist et al.1996).

As previously indicated, in the Venta Mi-cena assemblage the juvenile/adult ratios ofungulates (estimated from MNI counts basedon deciduous and permanent teeth) as a func-tion of adult body mass suggests that mortal-ity age profiles differed depending on the sizeof the prey. This would be the consequence ofselection by predators, which increased theproportion of young and more vulnerable in-dividuals of those ungulate species of largersize. This interpretation is in accordance withavailable data on prey selection by Recent car-nivores as a function of size and age of theirpreferred prey (Palmqvist et al. 1996: Fig. 7;Arribas and Palmqvist 1998: Table 3). Figure 4


FIGURE 5. Comparison between the relative abundanc-es of ungulate size classes in the prey hunted and scav-enged by modern spotted hyenas (Crocuta crocuta) in theSerengeti National Park (data from Kruuk 1972) and thefrequencies of such categories in the ungulate assem-blage preserved at Venta Micena (data from Table 1,MNI counts).

shows the increase in the value of the juve-nile/adult ratio in relation to the mass esti-mated for the ungulate species from Venta Mi-cena.

Therefore, the positive slope for the rela-tionship between the proportion of juvenilesand the mass estimated for the adults indi-cates that the Venta Micena assemblage wasnot formed through catastrophic mortalityevents during droughts (in such case theabundance of juveniles of different specieswould be approximately constant and size-in-dependent). We can conclude that the vast ma-jority of skeletal elements accumulated by hy-enas came from attritional mortality in un-gulate populations, caused by selective choiceof carnivores.

Bias II: Selective Transport of Carcasses andSkeletal Parts. According to field data collect-ed by Kruuk (1972) in the Serengeti and Ngo-rongoro National Parks (Tanzania), modernspotted hyenas are efficient hunters that hunttheir prey in 58.3% of cases and scavenge un-gulate carcasses in the remaining 41.7% of cas-es. Of those ungulates scavenged, individualsdead by illness or accident represent 19.4%,whereas the rest are carcasses of prey huntedand partially defleshed by lions and painteddogs. The relative abundances of ungulateprey of different body size classes hunted bylions and painted dogs correlate well with thefrequencies of ungulate populations (Kruuk1972; Schaller 1972).

The distribution of specimens among sizeclasses in the ungulate assemblage from VentaMicena (Fig. 5; frequencies estimated fromMNI counts in Table 1) is different from thefrequencies of ungulates hunted by spottedhyenas according to a x2 test for the cumula-tive differences (x2 5 148.2; df 5 4; p ,0.0001), but remarkably similar to those inwhich spotted hyenas scavenge carcasses ofanimals killed by lions and wild dogs (x2 517.8; p , 0.01 for all size classes; x2 5 4.3; df5 3; p . 0.1 for ungulates weighing .50 kg).The only significant difference between thedistribution of ungulate size classes in VentaMicena and in the prey scavenged by spottedhyenas is the proportion of small species (,50kg), which are underrepresented in the fossilassemblage (one individual of Caprini indet.,

1/198, Table 1) but represent 14.5% (80/551[Kruuk 1972: Table 26]) of the carcasses scav-enged by spotted hyenas (one-tailed t-test: t 59.68; p , 0.0001). This indicates that short-faced hyenas preferentially consumed smallungulates in situ and selectively transportedcarcasses and body parts of larger species totheir maternity dens.

These results suggest that the predatory be-havior of Pachycrocuta differed from that ofCrocuta, because modern spotted hyenas bothhunt and scavenge ungulates, whereas theshort-faced hyenas seem to have relied moreheavily on prey hunted by other predators;therefore, the behavior of Pachycrocuta wasprobably more similar to that of modernbrown and striped hyenas, which are predom-inantly scavengers (Mills 1989). This trophicdependence was facilitated by the fact that un-gulate carcasses left by machairodonts andhypercarnivorous canids would retain vari-able amounts of flesh and all nutrients withinthe bones, given that the slicing dentition ofthese carnivores made them incapable of bone


FIGURE 6. Abundance of ungulate species in Venta Mi-cena, according to minimum number of adult individ-uals (MNI), calculated from craniodental elements (per-manent teeth and antlers or horn bases) and from min-imum number of elements (MNE) of postcranial bones(independent estimates for forelimb and hindlimbbones). Caprini indet., Praeovibos sp., and species .1000kg were excluded from this analysis owing to their lowsample sizes. Species ordered by decreasing values inthe ratio MNI(teeth) : MNI(bones).

cracking (Marean 1989; Arribas and Palm-qvist 1999).

The bone assemblage of Venta Micena canthus be considered as mixed, showing somefeatures typical of primary assemblages, col-lected by predators, and others that are char-acteristic of non-primary, secondary assem-blages accumulated by scavenger carnivores(Table 3).

The bias produced by the selective transportof ungulate remains is particularly evidencedby the differential representation of preservedskeletal parts. The abundance of each taxoncan be estimated by MNI counts obtainedfrom teeth and cranial elements (i.e., antlersand horn bases in the case of ruminants), aswell as from MNI counts based on minimumnumber of elements (MNE) estimated frompostcranial remains (i.e., forelimb and hin-dlimb bones, complete elements or those rep-resented by isolated epiphyses). Figure 6shows that for small ungulates, such as the

goat (Hemitragus alba; 75 kg of estimated massfor adult individuals) or the fallow deer (Damasp.; 95 kg), MNI(teeth) gives a higher estimate ofabundance than MNI(bones) (x2-test: x2 5 5.33and 5.88; p , 0.05 in both cases; MNIteethcounts used as expected frequencies). For Soer-gelia minor, a bovine of intermediate mass (225kg), the two MNI counts are similar (x2 5 0.10;p . 0.5). Finally, ungulate taxa of larger size,such as the horse (E. altidens; 350 kg) and thebuffalo (Bovini cf. Dmanisibos; 450 kg), are bet-ter represented by postcranial elements (x2 530.42 and 20.45, respectively, p , 0.0001 inboth cases).

These differences are in large part an indi-cation of how the hyenas handled the carcass-es they scavenged. In the case of species thatwere preferentially transported as completecarcasses, the original abundances, estimatedfrom MNI counts based on teeth and bones,would be approximately the same as the abun-dances in the accumulated assemblage. How-ever, because hyenas selectively fracture majorlongbones and destroy limb bone epiphyses toget at the marrow and fat, MNI estimates forthe assemblage that are based on teeth shouldbe higher than those based on postcranial el-ements. Therefore small ungulate species(,100 kg), which are represented by anMNI(teeth) : MNI(bones) ratio of approximately 2:1 in the Venta Micena assemblage (H. alba andDama sp.), were probably transported in mostcases as complete carcasses. In the case oflarger species (.300 kg), selective transport ofmarrow-rich body parts (i.e., the forelimb inbuffalo and the hindlimb in horse) is suggest-ed by the reverse ratio (MNIteeth : MNIbones 5 1:2). Finally, for ungulate species of intermedi-ate size (100300 kg), such as S. minor, post-cranial elements were transported by hyenaswith a somewhat higher frequency than heads(after preferential consumption of major long-bones, MNI counts calculated from teeth andbones are similar, 1:1).

The only exception to this trend is the largemegacerine deer E. giulii. Although its bodymass is estimated at 380 kg for adult individ-uals, MNI counts from teeth and bones arequite similar (Fig. 6; x2 5 0.05; p . 0.5). Thismay be due to two reasons: (1) major limbbones are relatively slender in this species and


TABLE 4. Number of forelimbs and hindlimbs (calculated from MNE counts for each major limb bone) of Equusaltidens and Bovini cf. Dmanisibos from Venta Micena. Total marrow yields and flesh weights of forelegs and hindlegs estimated from values for modern Equus caballus (Outram and Rowley-Conwy 1998) and Bison bison (Brink1997).

Equus altidens

Forelimbs Hindlimbs

Bovini cf. Dmanisibos

Forelimbs Hindlimbs

Number of legs (right/left)Total marrow yields (g)Flesh content (kg)

118 (59/59)77.114.0

141 (72/69)115.4

46.2

44 (18/26)622.0

13.0

41 (24/17)558.4

43.9

were presumably more easily fractured by hy-enas; and (2) the antlers of the males were par-ticularly large (;1.5 m in width), and the hy-enas might have transported the heads to theirdens to exploit mineral phases and hemopoi-etic tissues supplied by the antlers; interest-ingly, fragments of deer antlers are well rep-resented in the assemblage (Table 2).

The evidence that hyenas selectively trans-ported certain parts from the carcasses oflarge ungulate species suggests that eachshort-faced hyena foraged alone in search ofscavengeable carcasses, as do modern brownhyenas (Mills 1989). If they had foraged ingroups, as spotted hyenas often do (Kruuk1972; Mills 1989), the members of the hyenaclan would have transported all the anatomi-cal regions of each carcass scavenged to theirmaternity den; large ungulate taxa would thenbe represented in the assemblage by similarnumbers of postcranial bones and cranioden-tal elements, rather than the skewed ratio weobserved.

Table 4 shows the number of forelimbs andhindlimbs calculated from MNE counts foreach limb bone in two taxa well representedin the assemblage, the equid E. altidens and thebuffalo Bovini cf. Dmanisibos. The correspond-ing values for flesh and marrow contents, es-timated from data for major longbones in twomodern, similarly sized herbivoresthehorse, Equus caballus (Outram and Rowley-Conwy 1998) and the North American plainbison, Bison bison (Brink 1997), are also pro-vided. The ratio of forelimbs to hindlimbs is0.837 in the Venta Micena horse, which isclearly different from that of flesh yields pro-vided by forelimbs and hindlimbs, 0.303, andcloser to the corresponding ratio estimated formarrow contents, 0.668. In the case of the buf-

falo, the ratio of forelimbs to hindlimbs is1.073, again a value different from that esti-mated for flesh contents, 0.296, but very closeto the value obtained for marrow, 1.114. Thissuggests that marrow content was the mainreason hyenas transported limb bones to theirmaternity dens.

Bias III: Consumption of Epiphyses and the Re-duction of Major Limb Bones. Typical bone-consuming sequences for each postcranial el-ement of Equus were described recently forVenta Micena by Arribas and Palmqvist (1998)and Arribas (1999). Three distinct types ofbone-consuming activities by hyenas were es-tablished (Fig. 2), depending on the positionof the bone in the horse skeleton (which is re-lated to the hyenas pattern of disarticulation),as well as on the amount of within-bone nu-trients (i.e., grease and marrow content) andmineral density:

1. Humerus, radius, tibia, ulna, and calca-neum: these are consumed following an in-variant proximodistal pattern. The reductionof these bones by hyenas starts with gnawingthe proximal epiphysis, then is followed byfracturing the diaphysis, and is finished bygnawing of the distal epiphysis, which usu-ally shows abundant tooth marks.

2. Femur: this is the only element in whichthe sequence of consumption follows a vari-able direction (i.e., from the proximal epiph-ysis to the distal epiphysis or vice versa) andboth epiphyses are lost.

3. Third metacarpal and metatarsal: thesebones are modified by crushing, with a vari-able direction of activity, and they tend to bemore abundantly preserved as complete ele-ments than other major limb bones, owing totheir higher mineral density and lower mar-row yields.


FIGURE 7. Least-squares regression analysis of the rawabundance at Venta Micena of preserved major limbbone epiphyses of horse (Equus altidens) (A) and buffalo(Bovini cf. Dmanisibos) (B) on their mean marrow con-tent, estimated from data for modern horse (Equus ca-ballus) by Outram and Rowley-Conwy (1998) and bison(Bison bison) by Brink (1997) (variables log-trans-formed).

Therefore, these results indicate that theskeletal elements preserved in the fossil as-semblage are those remaining once all within-bone nutrients were consumed by hyenas. Toevaluate quantitatively this taphonomic biason the preservational completeness of thebone assemblage, we performed a compara-tive analysis of the preservational state andabundance of postcranial elements in E. alti-dens and Bovini cf. Dmanisibos. We hypothe-sized that there would be differences in theabundance of postcranial elements becausethere are differences in their within-bone nu-trients (Emerson 1990; Brink 1997; Arribasand Palmqvist 1998; Outram and Rowley-Conwy 1998). Figure 7A shows the raw abun-dance in which major limb bone epiphyses ofhorse are preserved in the assemblage, as afunction of their mean marrow content (esti-mated from values for E. caballus in Outramand Rowley-Conwy 1998). A least-squares re-gression revealed an inverse relationship be-tween both variables, which is statistically sig-nificant (r 5 20.60; p , 0.05). This indicatesthat hyenas selectively consumed the epiphy-ses of bones having higher within-bone nutri-ent content (e.g., proximal and distal femur,proximal tibia), and acted less intensely onthose yielding lower marrow values (e.g., me-tapodials, distal tibia).

The raw abundance of major longboneepiphyses of buffalo in relation to their mar-row weight is shown in Figure 7B (data for B.bison in Brink 1997). The regression line alsoshows a negative slope (r 5 20.85, p ,0.0001), which indicates that the skeletal partsbetter represented among the survivingepiphyses are those with lower marrowyields. However, the regression obtained forthis species is statistically more significantthan in the case of E. altidens, owing to thehigher marrow contents of bovine epiphyses(six-fold on average). This suggests a great se-lectivity in the bone-cracking behavior of hy-enas, which was in turn translated into a dif-ferential preservation of the skeletal elementsof both taxa in the bone assemblage.

Three major factors therefore appear tohave biased the composition of the Venta Mi-cena assemblage: the scavenging by adult hy-enas of ungulate prey hunted by hypercarni-

vores (bias I); the selective transport of wholecarcasses or certain anatomical parts, depend-ing on the size of the ungulate species scav-enged (bias II); and the preferential breakagein the dens of bones with higher marrow con-tent (bias III). Although these biases decreasedthe amount of paleobiological informationpreserved in the assemblage, the representa-tion of the original mammalian community isvalid, thanks to the scavenging behavior of hy-enas. A collection of bones from the prey of asingle predator may differentially sample par-ticular species because of the predators preypreferences, and such accumulation wouldprovide a poor estimate of standing diversity


in the paleocommunity (Vrba 1980; Brain1981; Shipman 1981; Behrensmeyer 1991).This is not the case at Venta Micena, however,where the skeletal remains of a wide spectrumof ungulate prey hunted by several species ofhypercarnivores in different habitats were col-lected by hyenas, thus providing a detailedpicture of the diversity of large mammals thatinhabited southern Spain during early Pleis-tocene times. This is corroborated by severalstudies of recent bone assemblages collectedby hyenas (Maguire et al. 1980; Skinner et al.1980; Hill 1981; Skinner and Van Aarde 1981,1991; Skinner et al. 1980, 1986, 1995, 1998; Ker-bis-Petherhans and Kolska-Horwitz 1992;Leakey et al. 1999), which indicate that the as-semblages accurately reflect the compositionof the mammalian fauna in areas adjacent tothe maternity dens.

The assemblage from Venta Micena wasprobably accumulated over a very short timespan; thus it is evidently not time-averagedand retains a relatively high degree of envi-ronmental resolution. The fact that most skel-etal elements are unweathered suggests this,as do inferences on hyaenid mortality pat-terns. According to data on population den-sities of modern spotted hyenas obtained byKruuk (1972), the mean numbers of adult andjuvenile spotted hyenas per den in Serengetiare 55 and 12, respectively. With Kruuks es-timate of mean annual mortality at 16.7%, ap-proximately 11 individuals of the hyena clandie each year, a figure remarkably similar tothe MNI calculated for P. brevirostris in the fos-sil assemblage (10 individuals, 6 adults and 4juveniles; Table 1). In fact, the juvenile short-faced hyenas from Venta Micena can be clas-sified within two age groups: two newborn in-dividuals, with unworn milk teeth, and an-other two that show deciduous dentition se-verely worn and being replaced by permanentteeth, indicating that they were at the end oftheir first year of life. This suggests that dur-ing a single season, probably summer, all fourof these individuals died.

Conclusions

Taphonomic processes have previouslybeen interpreted as solely destructive forces.Information loss in terrestrial and fluvial bio-

tas results largely from such processes astransport, disarticulation, sorting, and break-age of skeletal parts by water, predators, scav-engers, and trampling. However, such biostra-tinomic processes imprint a taphonomic sig-nature that often provides new data useful fordecoding paleobiological information (Wilson1988; Fernandez-Lopez 1991; De Renzi 1997).The assemblage of large mammals from VentaMicena constitutes a good example of how an-alytical studies can contribute toward re-cre-ating a significant fraction of paleobiologicalinformation lost during the taphonomic his-tory.

As we have discussed here, it is even pos-sible to infer information that was not origi-nally preserved in the bone assemblage, suchas the behavior of the extinct hyena P. breviros-tris, a species that differed from the modernspotted hyena in being a strict scavenger ofungulate carcasses selectively preyed upon byhypercarnivores. This inference was based onthe quantitative study of the preservationalbias introduced by the scavenging behavior ofthis giant hyena, which is shown to have beenhighly specialized.

However, similar paleobiological inferencesmay be obtained only in assemblages thatwere collected during the biostratinomic stageby biological agents, like hyenas, hominids, orporcupines. Other types of terrestrial accu-mulations, where the bones were accumulatedexclusively by physical agents (e.g., fluvial as-semblages), would reveal useful sedimento-logical and paleoenvironmental data (e.g.,strength and direction of water currents), butbecause the skeletal remains of such assem-blages are frequently mixed, hydrodynami-cally sorted, and even reworked, decoding thetaphonomic information locked in these as-semblages would contribute little reliable pa-leobiological information about the structureand composition of the original paleocom-munity from which they were derived. In thiscontext, the macrovertebrate assemblage fromVenta Micena constitutes an exceptional win-dow for the detailed study of the mammaliancommunities that inhabited Western Europeduring the early Pleistocene and the relation-ships among the species that lived withinthem.


Acknowledgments

Thanks to M. De Renzi, M. Foote, A. Miller,R. A. Reyment, and L. Spencer for suggestionsthat led to significant improvements in thiscontribution. We gratefully acknowledge M.Anton for providing us the reconstructions ofcarnivores used in Figure 3. R. Bobe, J. Saun-ders, S. L. Wing, and an anonymous reviewerprovided insightful comments and helpfulcriticism of the manuscript. And, last but notleast, N. Atkins improved the style of this ar-ticle.

Literature Cited

Anyonge, W. 1993. Body mass in large extant and extinct car-nivores. Journal of Zoology 231:339350.

. 1996. Locomotor behaviour in Plio-Pleistocene sabre-tooth cats: a biomechanical analysis. Journal of Zoology 238:395413.

Arribas, A. 1999. Analisis y modelizacion de la tafonoma delyacimiento de Venta Micena (Orce, Granada) y su estudiocomparativo con otras localidades espanolas del Plio-Pleis-toceno continental. Ph.D. dissertation. Universidad Complu-tense de Madrid, Madrid.

Arribas, A., and P. Palmqvist. 1998. Taphonomy and paleoecol-ogy of an assemblage of large mammals: hyenid activity inthe lower Pleistocene site at Venta Micena (Orce, Guadix-BazaBasin, Granada, Spain). Geobios 31(Suppl.):347.

. 1999. On the ecological connection between sabre-tooths and hominids: faunal dispersal events in the lowerPleistocene and a review of the evidence for the first humanarrival in Europe. Journal of Archaeological Science 26:571585.

Behrensmeyer, A. K. 1978. Taphonomic and ecologic informa-tion from bone weathering. Paleobiology 4:150162.

. 1991. Terrestrial vertebrate accumulations. Pp. 291327in P. A. Allison and D. E. G. Briggs, eds. Taphonomy: releasingthe data locked in the fossil record. Plenum, New York.

Behrensmeyer, A. K., and D. E. Dechant Boaz. 1980. The recentbones of Amboseli Park, Kenya, in relation to East African Pa-leoecology. Pp. 7293 in Behrensmeyer and Hill 1980.

Behrensmeyer, A. K., and A. P. Hill, eds. 1980. Fossils in themaking: vertebrate taphonomy and paleoecology. Universityof Chicago Press, Chicago.

Behrensmeyer, A. K., and R. W. Hook. 1992. Paleoenvironmentalcontexts and taphonomic modes. Pp. 15136 in A. K. Beh-rensmeyer, J. D. Damuth, W. A. DiMichele, R. Potts, H. D.Sues, and S. L. Wing, eds. Terrestrial ecosystems throughtime: evolutionary paleoecology of terrestrial plants and an-imals. University of Chicago Press, Chicago.

Behrensmeyer, A. K., and S. M. Kidwell. 1985. Taphonomyscontributions to paleobiology. Paleobiology 11:105119.

Brain, C. K. 1980. Some criteria for the recognition of bone-col-lecting agencies in African caves. Pp. 108130 in Behrensmey-er and Hill 1980.

. 1981. The hunters or the hunted: an introduction to Af-rican cave taphonomy. University of Chicago Press, Chicago.

Brink, J. W. 1997. Fat content in leg bones of Bison bison, and ap-plications to archaeology. Journal of Archaeological Science24:259274.

Bunn, J. T. 1982. Meat-eating and human evolution: studies onthe diet and subsistence patters of Plio-Pleistocene hominids

in East Africa. Ph.D. dissertation. University of California,Berkeley.

Calder, W. A. 1984. Size, function, and life history. Harvard Uni-versity Press, Cambridge.

Cutler, A. H., A. K. Behrensmeyer, and R. E. Chapman. 1999.Environmental information in a recent bone assemblage: rolesof taphonomic processes and ecological change. Palaeogeog-raphy, Palaeoclimatology, Palaeoecology 149:359372.

Damuth, J. 1982. Analysis of the preservation of communitystructure in assemblages of fossil mammals. Paleobiology 8:434446.

Damuth, J., and B. J. MacFadden, eds. 1990. Body size in mam-malian paleobiology: estimation and biological implications.Cambridge University Press, Cambridge.

De Renzi, M. 1997. Taphonomic information and paleobiologi-cal information: a wrong dilemma? Cuadernos de GeologaIberica 23:183208.

Emerson, A. M. 1990. Archaeological implications of variabilityin the economic anatomy of Bison bison. Ph.D. dissertation.Washington State University, Pullman.

Ewer, R. F. 1973. The carnivores. Cornell University Press, Ith-aca, N.Y.

Fernandez-Lopez, S. 1991. Taphonomic concepts for a theoreti-cal biochronology. Revista Espanola de Paleontologa 6:3749.

Fosse, P. 1996. La grotte n8 1 de Lunel-Viel (Herault, France):repaire dhye`nes du Pleistoce`ne moyen. Etude taphonomiquedu materiel osseux. Paleo 8:4781.

Hill, A. P. 1981. A modern hyena den in Amboseli National Park,Kenya. Pp. 137138 in Pan-African Congress on Prehistoryand Quaternary Studies, Proceedings.

Kerbis Peterhans, J. C., and L. Kolska-Horwitz. 1992. A bone as-semblage from a striped hyaena (Hyaena hyaena) den in NegevDesert, Israel. Israel Journal of Zoology 37:225245.

Kidwell, S. M., and D. W. J. Bosence. 1991. Taphonomy and time-averaging of marine shelly faunas. Pp. 116209 in P. A. Allisonand D. E. G. Briggs, eds. Taphonomy: releasing the datalocked in the fossil record. Plenum, New York.

Kidwell, S. M., and K. W. Flessa. 1995. The quality of the fossilrecord: populations, species, and communities. Annual Re-views of Ecology and Systematics 26:269299.

Klein, R. G., and K. Cruz-Uribe. 1984. The analysis of animalbones from archaeological sites. University of Chicago Press,Chicago.

Kruuk, H. 1972. The spotted hyena: a study of predation andsocial behavior. University of Chicago Press, Chicago.

Leakey, L. N., S. A. H. Milledge, S. M. Leakey, J. Edung, P.Haynes, D. K. Kiptoo, and A. McGeorge. 1999. Diet of stripedhyaena in northern Kenya. African Journal of Ecology 37:314326.

Lewis, M. E. 1997. Carnivoran paleoguilds of Africa: implica-tions for hominid food procurement strategies. Journal of Hu-man Evolution 32:257288.

Lyman, R. L. 1994. Vertebrate taphonomy. Cambridge Univer-sity Press, Cambridge.

Maguire, J. M., D. Pemberton, and M. H. Collet. 1980. The Mak-apansgat Limeworks grey breccia: hominids, hyaenas, hystri-cids or hillwash? Palaeontologia Africana 23:7598.

Marean, C. W. 1989. Sabertooth cats and their relevance for earlyhominid diet and evolution. Journal of Human Evolution 18:559582.

Marean, C. W., and C. L. Ehrhardt. 1995. Paleoanthropologicaland paleoecological implications of the taphonomy of a sa-bretooths den. Journal of Human Evolution 29:515547.

Martin, L. D., J. P. Babiarz, V. L. Naples, and J. Hearst. 2000.Three ways to be a saber-toothed cat. Naturwissenschaften87:4144.

Martin, R. E. 1999. Taphonomy: a process approach. Cambridge


Paleobiology Series, No. 4. Cambridge University Press, Cam-bridge.

Martnez-Navarro, B., and P. Palmqvist. 1995. Presence of theAfrican machairodont Megantereon whitei (Broom, 1937) (Fe-lidae, Carnivora, Mammalia) in the lower Pleistocene site ofVenta Micena (Orce, Granada, Spain), with some consider-ations on the origin, evolution and dispersal of the genus.Journal of Archaeological Science 22:569582.

. 1996. Presence of the African saber-toothed felid Me-gantereon whitei (Broom, 1937) (Mammalia, Carnivora, Ma-chairodontidae) in Apollonia-1 (Mygdonia basin, Macedonia,Greece). Journal of Archaeological Science 23:869872.

Mills, M. G. L. 1989. The comparative behavioral ecology of hy-enas: the importance of diet and food dispersion. Pp. 125142in J. L. Gittleman, ed. Carnivore behavior, ecology, and evo-lution. Chapman and Hall, London.

Outram, A., and P. Rowley-Conwy. 1998. Meat and marrow util-ity indices for horse (Equus). Journal of Archaeological Science25:839849.

Palmqvist, P., and A. Arribas. 2001. An overview of the taphon-omy and paleoecology of the assemblage of large mammalspreserved in lower Pleistocene deposits from Venta Micena(Guadix-Baza basin, Spain). Paleontologia i Evolucio 3233:3758.

Palmqvist, P., B. Martnez-Navarro, and A. Arribas. 1996. Preyselection by terrestrial carnivores in a lower Pleistocene pa-leocommunity. Paleobiology 22:514534.

Palmqvist, P., A. Arribas, and B. Martnez-Navarro. 1999. Eco-morphological analysis of large canids from the lower Pleis-tocene of southeastern Spain. Lethaia 32:7588.

Peters, R. H. 1983. The ecological implications of body size.Cambridge University Press, Cambridge.

Rawn-Schatzinger, V. 1992. The scimitar cat Homotherium serumCope: osteology, functional morphology, and predatory be-havior. Illinois State Museum Reports of Investigations 47:180.

Richardson, P. R. K. 1980. Carnivore damage to antelope bonesand its archaeological implications. Palaeontologia Africana23:109125.

Saunders, J. J., and B. K. Dawson. 1998. Bone damage patternsproduced by extinct hyena, Pachycrocuta brevirostris (Mam-malia: Carnivora) at the Haro River Quarry, NorthwesternPakistan. In Y. Yomida, L. J. Flynn, and L. L. Jacobs, eds. Ad-vances in vertebrate paleontology and geochronology. Na-tional Science Museum Monographs 14:215242. Tokyo.

Schaller, G. B. 1972. The Serengeti lion: a study of predator-preyrelations. University of Chicago Press, Chicago.

Shipman, P. 1981. Life history of a fossil: an introduction to ta-phonomy and paleoecology. Harvard University Press, Cam-bridge.

Skinner, J. D., and R. J. Van Aarde. 1981. The distribution andecology of the brown hyaena Hyaena brunnea and spotted hy-aena Crocuta crocuta in the central Namib Desert. Madoqua 12:231239.

. 1991. Bone collecting by brown hyaenas Hyaena brunneain the central Namib Desert, Namibia. Journal of Archaeolog-ical Science 18:513523.

Skinner, J. D., S. Davis, and G. Ilani. 1980. Bone collecting bystriped hyenas, Hyena hyena, in Israel. Palaeontologia Africana23:99104.

Skinner, J. D., J. R. Henschel, and A. S. Van Jaarsveld. 1986. Bone-collecting habits of spotted hyenas Crocuta crocuta in the Kru-ger National Park. South African Journal of Zoology 21:303308.

Skinner, J. D., R. J. Van Aarde, and R. A. Goss. 1995. Space andresource use by brown hyaenas Hyaena brunnea in the centralNamib Desert. Journal of Zoology 237:123131.

Skinner, J. D., M. A. Haupt, M. Hoffmann, and H. M. Dott. 1998.Bone collecting by brown hyaenas Hyaena brunnea in the Na-mib desert: rate of accumulation. Journal of ArchaeologicalScience 25:6971.

Turner, A., and M. Anton. 1996. The giant hyena, Pachycrocutabrevirostris (Mammalia, Carnivora, Hyaenidae). Geobios 29:455468.

. 1998. Climate and evolution: implications of some ex-tinction patterns in African and European machairodontinecats of the Plio-Pleistocene. Estudios Geologicos 54:209230.

Van Valkenburgh, B. 1990. Skeletal and dental predictors ofbody mass in carnivores. Pp. 181205 in Damuth and Mac-Fadden 1990.

Voorhies, M. R. 1969. Taphonomy and population dynamics ofthe early Pliocene vertebrate fauna, Knox County, Nebraska.Contributions to Geology University of Wyoming, Special Pa-per No. 1.

Vrba, E. S. 1980. The significance of bovid remains as indicatorsof environment and predation patterns. Pp. 247271 in Beh-rensmeyer and Hill 1980.

Western, D. 1979. Size, life history and ecology in mammals. Af-rican Journal of Ecology 17:185204.

. 1980. Linking the ecology of past and present mammalcommunities. Pp. 4154 in Behrensmeyer and Hill 1980.

Wilson, M. V. H. 1988. Taphonomic processes: information lossand information gain. Geoscience Canada 15:131148.

Documents

Arribas Hiena