q 2003 The Paleontological Society. All rights reserved. 0094-8373/03/2902-0005/$1.00

Paleobiology, 29(2), 2003, pp. 205–229

Paleoecological reconstruction of a lower Pleistocene largemammal community using biogeochemical (d13C, d15N, d18O, Sr:Zn) and ecomorphological approaches

Paul Palmqvist, Darren R. Grocke, Alfonso Arribas, and Richard A. Farina

Abstract.—Ecomorphological and biogeochemical (trace element, and carbon, nitrogen, and oxygenisotope ratios) analyses have been used for determining the dietary niches and habitat preferencesof large mammals from lower Pleistocene deposits at Venta Micena (Guadix-Baza Basin, Spain).The combination of these two approaches takes advantage of the strengths and overcome the weak-ness of both approaches. The range of d13Ccollagen values for ungulate species indicates that C3 plantswere dominant in the diet of these mammals. d13Ccollagen values vary among ungulates: perissodac-tyls have the lowest values and bovids the highest ones, with cervids showing intermediate values.The hypsodonty index measured in lower molar teeth and the relative length of the lower premolartooth row indicate that the horse, Equus altidens, was a grazing species, whereas the rhino, Ste-phanorhinus etruscus, was a mixed feeder in open habitats. The similar d13Ccollagen values shown inboth perissodactyls does not reflect differences in feeding behavior with other ungulates, but rathera lower isotope enrichment factor in these monogastric herbivores than in ruminants, owing to theirlower metabolic efficiency. The cervids Eucladoceros giulii and Dama sp. show low hypsodonty val-ues, indicating that they were mixed feeders or browsers from forested habitats, an ecomorphol-ogically based conclusion corroborated in the former by its low d15Ncollagen content (canopy effect).Bovid species (Bovini aff. Leptobos, Soergelia minor, and Hemitragus albus) presumably inhabitedopen environments, according to their comparatively high hypsodonty and d15Ncollagen values. Car-nivore species (Homotherium latidens, Megantereon whitei, Pachycrocuta brevirostris, Canis falconeri, andCanis etruscus) exhibit higher d15Ncollagen values than ungulates. These results record the isotopicenrichment expected with an increase in trophic level and are also supported by low bone Sr:Znratios. The elevated d15Ncollagen value for a sample of Mammuthus meridionalis, which came from anindividual with unfused epiphyses, confirms that it was a suckling animal. The d15Ncollagen value ofthe scimitar-cat H. latidens is well above that obtained for the young individual of Mammuthus,which indicates that juvenile elephants were an important part of its diet. The hippo, Hippopotamusantiquus, yielded unexpectedly high d15Ncollagen values, which suggest feeding on aquatic, non-N2-fixing plants. The high d18Ohydroxyl values of bovids Hemitragus and Soergelia and of cervid Dama in-dicate that these ungulates obtained most of their water requirements from the vegetation. Themegaherbivores and Eucladoceros exhibit the lowest d18Ohydroxyl values, which suggest increased wa-ter dependence for them. Paleosynecological analysis was based on the relative abundance of spe-cies of large mammals from different ecological categories, determined by feeding behavior andlocomotion types. The comparison of the frequencies of such categories in Venta Micena with thosefound in modern African communities indicates that the composition of the paleocommunity close-ly resembles those of savannas with tall grass and shrubs. The net above-ground primary produc-tivity estimated for the on-crop biomass of the mammalian species preserved in the fossil assem-blage also yields a figure congruent with that expected for an open environment.

Paul Palmqvist. Departamento de Geologıa y Ecologıa (Area de Paleontologıa), Facultad de Ciencias, Cam-pus Universitario de Teatinos, 29071 Malaga, Spain. E-mail: Paul.Palmqvist@uma.es

Darren R. Grocke. Department of Geology, Royal Holloway, University of London. Egham Hill, Egham,Surrey TW20 0EX, United Kingdom. E-mail: d.grocke@gl.rhul.ac.uk

Alfonso Arribas. Museo Geominero, Instituto Geologico y Minero de Espana (IGME), Rıos Rosas, 23,28003 Madrid, Spain. E-mail: a.arribas@igme.es

Richard A. Farina. Departamento de Paleontologıa, Facultad de Ciencias, Universidad de la Republica,Igua, 4225, 11400 Montevideo, Uruguay. E-mail: fari;a@fcien.edu.uy

Accepted: 1 November 2002

IntroductionVenta Micena (378449150N, 2824990W, eleva-

tion 974.5 m) lies near the village of Orce, inthe eastern sector of the Guadix-Baza Basin(Betic Chains, southeastern Spain; Fig. 1). Thissite is dated by biostratigraphy to the early

Pleistocene, with an estimated age of 1.3 6 0.1Ma (Arribas and Palmqvist 1999). The fossilswere preserved in 98–99% pure micritic lime-stone, precipitated on a caliche paleosol thatsurrounded a shallow Pleistocene lake withswampy marginal zones (Arribas and Palm-

206 PAUL PALMQVIST ET AL.

FIGURE 1. Geological map of the Betic Chains, showing the situation of the Guadix-Baza intramontane basin andthe stratigraphic profile of the early Pleistocene locality at Venta Micena. This sedimentary basin was endorheic(i.e., characterized by interior drainage) until late Pleistocene times, thus facilitating an exceptional record of Plio-Quaternary taphocenoses of large mammals preserved in swampy and lacustrine sediments. The 80–120-cm-thickVenta Micena stratum (VM-2) is one of the various fossiliferous units in the sedimentary sequence of Orce, whosesurface stands out topographically in the ravines of the region and can be followed along ;2.5 km. This stratumhas the following vertical structure from bottom to top: (1) A basal unit that is one-third to one-half the total thick-ness, formed by homogeneous micrite with some carbonate nodules and small mud banks. The sediment preservesabundant shells of freshwater mollusks and is sterile in vertebrate fossils, thus attesting to a first generalized la-custrine stage in the region, in which the micrite was precipitated in water of variable depth. The absence of bothpyrite and carbonate facies rich in organic matter is evidence that the lake was not eutrophic. (2) A 4–15-mm-thickcalcrete paleosol (hardpan) developed on the surface of the micrite sediments. The calcrete forms an irregular sur-face, subparallel to the bedding plane, and is thicker at topographic highs. It defines a stratigraphic unconformityindicating a major drop of the Pleistocene lake level, and thus the emergence of an extensive plain around the lake.(3) An upper unit of micrite deposited in a subsequent rise of the lake level, which continues up to the top of thestratum, showing rootmarks, mudcracks, and a high density of fossil bones of large mammals resting on the pa-leosol.

qvist 1998). The macrovertebrate assemblagecomprises ;5800 identifiable skeletal remainsof 225 individuals, which belong to 20 speciesof large mammals (Table 1).

Taphonomic research on the composition ofthe Venta Micena assemblage has shown thatthe skeletal remains were scavenged by the gi-ant, short-faced hyena Pachycrocuta brevirostrisfrom carcasses of ungulates preyed upon byflesh-eating carnivores (Palmqvist et al. 1996;

Arribas and Palmqvist 1998). The selection bypredators of specific ungulates was basically afunction of differences in the body mass of ju-venile and adult prey individuals, as well asin the sex of prey (Palmqvist et al. 1996). Majortaphonomic biases in the preservation of thebone assemblage are related to the selectivetransport by hyenas of ungulate carcasses andbody parts to their maternity dens, and to thepreferential consumption of low-density, mar-

207ECOLOGY OF PLEISTOCENE MAMMALS

TABLE 1. Herbivores larger than 10 kg and carnivores larger than 5 kg found in the fauna of Venta Micena, theirtrophic habits, their number of identifiable specimens (NISP), their minimal number of individuals (MNI)(data fromPalmqvist and Arribas 2001), their estimated mass (data from Palmqvist et al. 1996), their calculated on-crop bio-mass, and their mass-specific basal metabolic rate. Abbreviations: Br 5 browser, .75% leaves. Mf 5 mixed feeder,25–75% grass. Gr 5 grazer, .75% grass. F 5 flesh eater, .70% vertebrate flesh in diet. Mi 5 meat and insect eater,20–70% flesh. Mb 5 meat and bone eaters. O 5 omnivore, ,20% flesh.

SpeciesTrophichabits

NISP(teeth/bones)

MNI(juv./adults)

Body mass(adults,in kg)

On-cropbiomass

(kg km22)

Basalmetabolic

rate(J kg21 s21)

Mammuthus meridionalisHippopotamus antiquusBovini aff. LeptobosSoergelia minorPraeovibos sp.Hemitragus albusCaprini gen. et sp. indet.Eucladoceros giuliiDama sp.Stephanorhinus etruscusEquus altidensVulpes praeglacialisCanis falconeriCanis etruscusLynx aff. issiodorensisMegantereon whiteiHomotherium latidensPachycrocuta brevirostriscf. Meles sp.Ursus etruscus

MfGrGrMfGrGrMfBrMfBrGrMiFCoFFFMbOO

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)

1 (1/0)27 (15/12)

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)1 (0/1)3 (1/2)

60003000

450225320

7510

38595

1500350

5301010

100200

7010

375

840.5706.8439.8369.9403.9281.0169.8423.0298.1594.3413.1

3.66.94.74.7

10.713.8

9.44.7

17.2

0.470.550.891.060.971.392.310.931.310.660.952.931.802.432.431.301.081.442.430.91

row-rich skeletal parts (Palmqvist and Arribas2001).

The main objective of this study was to de-termine the autecology of those species oflarge mammals represented at Venta Micenausing biogeochemical and ecomorphologicaltechniques, in order to (1) interpret niche andresource partitioning by the species from eachtrophic level; (2) infer ecological relationshipssuch as those between predators and theirprey; and (3) study the synecological proper-ties of the paleocommunity. The use of thiscombined approach allows the researcher totake advantage of the strengths and overcomethe weakness of both techniques.

Ecomorphological Analyses

Autecological inferences are presented forlarge mammals from Venta Micena, followingan ecomorphological approach. Size estimatesfor these species (Table 1) range between 5 kgand 6000 kg and were obtained by Palmqvistet al. (1996) using ‘‘taxon-free’’ regressionequations for body mass on craniodental andpostcranial measurements in modern carni-

vore and ungulate species (e.g., Damuth andMacFadden 1990; Anyonge 1993).

Feeding preferences of extinct ungulatescan be determined from their craniodentalmorphology, because several features of theskull and dentition are indicative of diet (seereview in Janis 1995; see also Spencer 1995;Mendoza et al. 2002). In herbivores, the hyp-sodonty index (HI) or relative tooth crownheight (defined here as unworn molar toothcrown height divided by molar width [Janis1988]) is widely accepted as a useful indicatorof diet (Fortelius 1985; Williams and Kay2001): grazing ungulates from open habitatsthat feed upon abrasive grasses with high sil-icophytolith contents generally show higherhypsodonty values than browsers from for-ested environments that consume succulentleaves (in this article grazers are defined asspecies in which grass represents .75% ofdiet, browsers include ungulates consuming.75% of leaves, and those species taking be-tween 25% and 75% of grass are considered asmixed feeders [Mendoza et al. 2002]). Muzzleshape is also a good indicator of the specific

208 PAUL PALMQVIST ET AL.

FIGURE 2. Mean values of relative length of the lower premolar tooth row [(length P2–P4/length M1–M3) 3 100] andhypsodonty index [(unworn M3 crown height/average width) 3 100] in modern perissodactyls (data from Mendozaet al. 2002) and extinct species (horse E. altidens and rhino S. etruscus) from Venta Micena.

adaptations related to the ‘‘cropping mecha-nism,’’ which includes the shape of the pre-maxilla and the corresponding mandibularsymphyseal region, as well as the relative pro-portions of the incisor teeth (Gordon and Il-lius 1988; Solounias and Moelleken 1993;Perez-Barberıa and Gordon 2001): browsersusually show narrow muzzles (i.e., short pre-maxillary width), consisting of a rounded in-cisor arcade with the first incisor generallylarger than the third, whereas grazers havebroad muzzles with transversely straight in-cisor arcades, showing equal or subequal-sized teeth.

Although the hypsodonty index is probablythe best single variable for predicting diet inboth extant and ancient ungulates, in somecases molar crown height does not seem to bea good indicator of feeding habits (see reviewin Mendoza et al. 2002). For example, mostgrazing and mixed feeding ungulates havehypsodont teeth, but the hippo (Hippopotamusamphibius) and the rock hyrax (Procavia capen-sis) have brachyodont (i.e., short-crowned)teeth (HI , 1.4 and , 1.7, respectively). Bothspecies have relatively low metabolic rates,consuming less food per day than would beexpected for animals of their body size (No-vak 1999), which means that the total amountof wear on the teeth would correspondinglybe less.

Some second-order differences in cranio-

dental morphology are related to phylogenet-ic constraints. For example, horses have rela-tively more narrow muzzles than grazing ru-minants of similar body size (Janis and Ehr-hardt 1988; MacFadden and Shockey 1997)and not all exclusively grass-eaters are similarin their degree of hypsodonty (Williams andKay 2001). Furthermore, different ungulategroups have adopted different solutions whenfaced with the same ecological specializations(Mendoza et al. 2002). Grazers, for example,require a comparatively greater grinding areaof cheek teeth than browsers and frugivores(Janis 1995). Perissodactyls have faced thisproblem by enlarging the size of the premolartooth row, which in low-crowned, browsingand mixed feeding rhinos is comparativelyshorter than the molar tooth row (Fig. 2). Inthese species, the mesiodistal length of thepremolar row represents ;75% of the corre-sponding measurement for the molar row, butgrazing perissodactyls have lower premolarand molar cheek teeth of nearly the samelength, as can be seen in the white rhino (Cer-atotherium simum), or even have a premolarrow that is longer than the molar row, as inhorses (Fig. 2). However, ruminants and ca-melids show an opposite trend in the size ofthe premolar tooth row; in these groups, graz-ers have comparatively shorter premolar toothrows than browsers (premolars ;45% of themolar tooth row, vs. ;70%; Figs. 3, 4). This

209ECOLOGY OF PLEISTOCENE MAMMALS

FIGURE 3. Mean values of relative length of the lower premolar tooth row [(length P2–P4/length M1–M3) 3 100] andhypsodonty index [(unworn M3 crown height/average width) 3 100] in modern bovids (data from Mendoza et al.2002) and extinct species (Bovini aff. Leptobos, S. minor, and H. albus) from Venta Micena. The position in this dia-gram of nine extant species covering the whole range of values for hypsodonty and premolar tooth row length isshown, in order to facilitate the visual interpretation of the diagram. The feeding category ‘‘high level browsers’’includes those browsing bovids that feed from trees and bushes at high levels above the ground (e.g., the long-necked gazelles of the genus Litocranius).

difference is probably due to differences in theway food is orally processed in foregut andhindgut fermenters (Mendoza et al. 2002).

In carnivores, features of the dentition re-lated to diet include the morphology of theupper canine (C1), lower carnassial (M1), andfourth premolar (P4). Particularly indicative offeeding habits are the relative length of the tri-gonid blade and the morphology of cusps inthe talonid basin of M1, the shape of C1 and P4

(i.e., tooth length divided by tooth breadth),the total grinding area in the dentition, andthe smaller masseter and temporalis momentarms (Van Valkenburgh 1988, 1989; Biknevi-cius and Van Valkenburgh 1996; Biknevicius et

al. 1996). These variables provide discrimina-tion between hypercarnivores, bone crackers,and omnivores. Relevant variables in the post-cranial skeleton of carnivores include the bra-chial and crural indexes (i.e., radius length di-vided by humerus length and tibia length di-vided by femur length, respectively), the un-gual phalanx shape and manus proportions,the ratio of phalanx length to metacarpallength, the biceps brachii leverage index, andcross-sectional geometric properties of longbones such as the midshaft femoral cross-sec-tional shape (Gonyea 1976; Van Valkenburgh1985, 1987; Anyonge 1996; Lewis 1997). Thesevariables allow the estimation of different as-

210 PAUL PALMQVIST ET AL.

FIGURE 4. Mean values of relative length of the lower premolar tooth row [(length P2–P4/length M1–M3) 3 100] andhypsodonty index [(unworn M3 crown height/average width) 3 100] in modern cervids (data from Mendoza et al.2002) and extinct cerrid species (E. giulii and Dama sp.) from Venta Micena. The position in this diagram of fiveextant species is shown for illustrative purposes. The feeding category ‘‘high level browsers’’ includes the moose,Alces alces, which is the only cervid that is able to feed at a high level above the ground.

pects related to habitat preferences and hunt-ing techniques: for example, the brachial andcrural indexes (Fig. 5) are useful in felids todiscriminate between predators that ambushand those that pursue their prey.

The species of large mammals preserved atVenta Micena can be classified in the follow-ing trophic groups, according to their cran-iodental and postcranial morphology:

Carnivores. The order Carnivora is repre-sented in the assemblage by nine species, in-cluding three felids (Homotherium latidens, Me-gantereon whitei, and Lynx aff. issiodorensis);

one hyenid (P. brevirostris); three canids (Canis[Xenocyon] falconeri, Canis etruscus, and Vulpespraeglacialis); one mustelid (cf. Meles sp.); andone ursid (Ursus etruscus). They can be clas-sified in one of the following sub-categories:

1. Hypercarnivores (i.e., predators whose dietconsists of .70% flesh): saber-tooths, H. la-tidens and M. whitei; the lynx, L. issiodoren-sis; and wild dog, C. falconeri.

2. Carnivores (i.e., those species in whichflesh constitutes between 20% and 70% ofthe diet, with fruits and invertebrates mak-

211ECOLOGY OF PLEISTOCENE MAMMALS

FIGURE 5. Mean values of brachial and crural indexes in several species of extant felids (data from Gonyea 1976and Anyonge 1996) and in the saber-tooth species preserved in Venta Micena. The value for H. latidens is the meanobtained by averaging unpublished length measurements of major limb bones from the early Pleistocene site atIncarcal (kindly provided by A. Galobart) and those published for the North American species H. serum (Rawn-Schatzinger 1992). Given the absence of at least one complete radius and tibia of M. whitei, the means of the valuesestimated in M. cultridens (Lewis 1997) and in the closely related genus Smilodon (Gonyea 1976) were used. Thebrachial index is positively correlated in modern felids with their cursorial abilities, and the values obtained in bothindexes are useful to describe the transition from closed environments, with dense tree cover, to open habitats.

ing up the balance): small-to-medium-sized canids, C. etruscus and V. praeglacialis.

3. Bone crackers (i.e., species that fracturebones for access to their marrow content):the hyena, P. brevirostris.

4. Omnivores (i.e., those species consuming,20% of vertebrate flesh): the ursid, U.etruscus, and the badger, cf. Meles sp.

M. whitei was a dirk-toothed machairodontwith a strong body, similar in size to that of ajaguar, Panthera onca (Martınez-Navarro and

Palmqvist 1996). The low value estimated forthe brachial index (;80%; Fig. 5) and the shortmetapodials indicate that it was an ambushpredator, hunting in closed, forested habitats.The forelimbs are powerful, with large claws,and the upper canine teeth are extremely long,sharp, and laterally compressed (Arribas andPalmqvist 1999).

H. latidens was a scimitar-toothed cat similarin size to modern lions, Panthera leo (Anyonge1993). It had small claws and elongated fore-

212 PAUL PALMQVIST ET AL.

limbs (brachial index of ;90%; Fig. 5), whichprovided considerable leverage. These fea-tures suggest increased cursoriality in anopen habitat (Martin et al. 2000), with lessprey-grappling capability than other saber-tooths (Rawn-Schatzinger 1992). The uppercanines were shorter and broader than in Me-gantereon, bearing coarse serrations in theenamel margins.

The results obtained in a comparative eco-morphological analysis of modern and Pleis-tocene canids (Palmqvist et al. 1999, 2002) in-dicate that C. falconeri was a hypercarnivorewith a body mass of ;30 kg. The second meta-carpal shows a reduced articular facet to thefirst metacarpal, which indicates that the latterbone was vestigial if not absent. Such condi-tion is similar to that of the African painteddog, Lycaon pictus, and indicates increasedcursoriality in open to intermediate forestedcountry. C. etruscus was similar in size (;10kg) and feeding habits to the modern coyote,Canis latrans. In this species, the trigonid bladerepresents ,70% of the mesiodistal length ofthe lower carnassial, which suggests that scav-enging probably represented an importantfraction of its diet.

L. issiodorensis had a body mass comparableto extant Lynx europaeus. Given the small sizeof the prey hunted by modern lynxes, it is notlikely that this extinct species was a predatorof large mammals.

P. brevirostris was a hyena with a body andskull 10–20% larger than that of modern spot-ted hyenas, Crocuta crocuta, well-adapted fordestroying carcasses and consuming bone(Palmqvist et al. 1996; Arribas and Palmqvist1998). This giant hyena shows a relative short-ening of the distal limb segments (Turner andAnton 1996), which suggests that it was lesscursorial than other hyenas, although suchshortening provided greater power and morestability to dismember and carry large piecesof carcasses. Several features of the dentition(e.g., the relative length of the trigonid bladein the lower carnassial) indicate that Pachycro-cuta fed more heavily on carrion than spottedhyenas did.

The overall craniodental morphology of theetruscan bear, U. etruscus, is similar to that ofmodern brown bears, Ursus arctos, thus sug-

gesting that this extinct species was also om-nivorous.

Herbivores. Ungulates are represented by11 taxa in Venta Micena, including one pro-boscidean (Mammuthus meridionalis); one hip-popotamus (Hippopotamus antiquus); two pe-rissodactyls (the horse, Equus altidens, and therhino, Stephanorhinus etruscus); and seven ar-tiodactyls (five bovids, Bovini aff. Leptobos,Soergelia minor, Hemitragus albus, Praeovibos sp.,and Caprini indet.; two cervids, Eucladocerosgiulii and Dama sp.). These taxa can be classi-fied into feeding categories by using ecomor-phological variables such as the hypsodontyindex and the relative length of the lower pre-molar tooth row (Figs. 2–4).

Concerning the perissodactyls, E. altidenswas a grazer, showing high values of both thehypsodonty index (HI 5 6.1) and the relativelower premolar row length (;110%), similarto those of living equids (Fig. 2). This specieshad elongate, slender metapodials and pha-langes, very similar to those of modern Gre-vy’s zebra, Equus grevyi, which indicates thatit was adapted to open and dry habitats(Guerrero-Alba and Palmqvist 1997). S. etrus-cus was a browser or mixed feeder withbrachyodont teeth (HI 5 1.8) and a short pre-molar tooth row (;82%), similar to those ofmodern rhinos in which grass represents,50% of the diet (Fig. 2). It was clearly not agrazer, because the white rhino, C. simum, theonly living hypergrazing rhino, shows a hyp-sodonty value (HI 5 3.9) 2.5 times greater thanboth Stephanorhinus and modern mixed feed-ing rhinos, and a longer premolar tooth row(;118%). The Etruscan rhino presumably in-habited open to mid-covered habitats, as sug-gested by its slender metapodials.

Within the artiodactyls, the goat H. albus (HI5 4.4) and the bison Bovini aff. Leptobos (HI 53.9) were presumably grazers or mixed feed-ers dwelling in unforested habitats, as de-duced from their comparatively high hypso-donty values (Fig. 3), which are similar tothose of modern grazing (HI 5 3.8–6.1) andmixed feeding ruminants (HI 5 2.5–5.3) fromopen habitats (range of values for extant bo-vids estimated from data in Mendoza et al.2002). S. minor shows intermediate values inboth the hypsodonty index (HI 5 2.9) and the

213ECOLOGY OF PLEISTOCENE MAMMALS

lower premolar row length, indicative ofmixed feeding habits (Fig. 3). The dentition ofthe giant deer, E. giulii, shows a very low valueof hypsodonty (HI 5 1.6). A comparison withhypsodonty data in modern deer (Fig. 4) sug-gests that it was a mixed feeder from closedhabitats (HI 5 1.1–2.8) or more probably abrowser (HI 5 1.2–1.6 [values from Mendozaet al. 2002]). The fallow deer, Dama sp., was amixed feeder, judging from its moderatelyhypsodont teeth (HI 5 1.7) and diet of its clos-est living relative, Dama dama. Of relevance toH. antiquus, modern hippos are basicallyfresh-grass grazers, although Bocherens et al.(1996) reported carbon isotope values unusualfor a diet on terrestrial C4 grasses in Pleisto-cene hippos from localities in North and EastAfrica, which suggests that they fed predom-inantly on aquatic vegetation.

The proboscidean, M. meridionalis, was abrowser or mixed feeder like modern Africanelephants, Loxodonta africana, which consumeon average 20% grass and 80% leaves, al-though grass was probably a more significantcomponent of diet in Mammuthus according tocarbon isotopes of tooth enamel (Cerling et al.1999).

Biogeochemical Analyses:Stable-Isotope Ratios

Biogeochemical techniques have proved ex-tremely useful in determining the dietaryniche of fossil mammalian species, and in pro-viding detailed environmental reconstruc-tions (see reviews and references in Bocherenset al. 1996; Grocke 1997a; Koch 1998; andMacFadden 2000). We analyzed carbon, nitro-gen, and oxygen isotope content of collagenmaterial and hydroxylapatite from bone andtooth enamel of large mammals preserved inVenta Micena to determine dietary niches andtrophic relationships, following the method-ology described by Grocke (1997a).

Stable isotopes are useful as paleobiologicaland paleoclimatological tracers because, as aresult of mass differences, different isotope ra-tios of an element (e.g., 12C vs. 13C, 14N vs. 15N,16O vs. 18O) have different thermodynamicand kinetic properties. For elements with anatomic mass of ,40, these differences can leadto measurable isotopic partitioning between

substances during physical and chemical pro-cesses that labels the substances with distinctisotopic ratios. Natural fractionations aresmall, so isotopic ratios are reported as partsper thousand (‰) deviation in isotope ratiofrom a standard, using the d notation, wheredX 5 [(Rsample/Rstandard) 2 1] 3 1000, with X 5C, N, or O, and R 5 13C/12C, 15N/14N, or 18O/16O. Rsample and Rstandard are the high-mass tolow-mass isotope ratios of the sample and thestandard, respectively: commonly used stan-dards for d13C, d15N, and d18O are Peedee bel-emnite (PDB), atmospheric N2, and standardmean ocean water (SMOW), respectively. Pos-itive d values indicate enrichment in the high-mass isotope relative to the standard, whereasnegative values indicate depletion.

Terrestrial plants can be divided into threemain groups on the basis of their photosyn-thetic pathway (Edwards and Walker 1983):(1) C3 plants, which follow the Calvin-Bensoncycle, fixing atmospheric CO2 directly throughthe reductive pentose phosphate pathway; (2)C4 plants, which use the Hatch-Slack cycle (C4-dicarboxylic acid pathway); and (3) CAM(crassulacean acid metabolism) plants. C3

plants are all trees, temperate shrubs, andgrasses of cool/moist climates and high alti-tudes. Most C4 plants are herbaceous, tropical,arid-adapted grasses and warm/dry herbs.Succulent or CAM plants use an intermediateC3/C4 photosynthetic pathway, which is mod-ified in response to environmental changes.

All plants take up 12CO2 in preference to13CO2, but there are important differences intheir isotopic composition related to their car-boxylating enzymes. C3 plants use the ribu-lose 1,5-biphosphate carboxylase, which dis-criminates effectively against 13CO2, produc-ing a depletion of 219.5‰. Given that themean d13C value of atmospheric CO2 is 26.5‰(Koch 1998), C3 plants show an average d13Cvalue of 226 6 2‰, with a range from 234‰in dense forests to 221‰ in open areas ex-posed to water stress. C4 plants use the phos-phoenolpyruvate carboxylase, which is less ef-fective at discriminating against 13CO2, andproduces a depletion of only 25.5‰. As aconsequence, C4 plants have an average d13Cvalue of 212 6 1‰, with a more restrictedisotopic range from 215‰ to 27‰. CAM

214 PAUL PALMQVIST ET AL.

plants are capable of fixing carbon with eitherphotosynthetic pathway and therefore displayd13Cplant values covering the range for C3 andC4 plants (see reviews and references in Smithand Epstein 1971; Bocherens et al. 1996;Grocke 1997a,b; 1998; Ehleringer et al. 1997;Collatz et al. 1998; Zazzo et al. 2000; Freemanand Colarusso 2001; and Franz-Odendaal etal. 2002).

When a plant is consumed by a herbivore,the plant’s carbon is incorporated into the an-imal’s skeletal tissues with some additionalfractionation. The difference between the d13Cvalue of the animal’s diet and that subsequent-ly incorporated into collagen translates intoan increase in d13Ccollagen per trophic level from13‰ to 15‰. This implies that a mammalconsuming C3 vegetation (i.e., fruits, leaves,and roots of trees and bushes) will record anaverage d13Ccollagen value of 222‰ (range:230‰ to 217‰), whereas a herbivore feed-ing on C4 plants (i.e., tropical grass blades,seeds and roots) will exhibit a value around28‰ (range: 211‰ to 23‰). The d13Ccollagen

enrichment is similar for carnivores: a preda-tor consuming C3-browsing ungulates will re-cord an average d13Ccollagen value of 218‰,whereas a carnivore that consumes C4-grazerswill show a value close to 24‰ (DeNiro andEpstein 1978; Van der Merwe 1982; Klepingerand Mintel 1986; Grocke 1997a,b; Koch 1998).Isotope variability among mammals is notsolely a result of diet, however: for example,recycling of CO2 in closed forests (i.e., canopyeffect) causes a shift to more negative d13Cplant

values and, thus, in the collagen and hy-droxylapatite of those browsing herbivoresconsuming such plants.

The d13C value of herbivore hydroxylapatiteis also enriched with respect to that of the an-imal’s diet. d13Chydroxyl ranges from 19‰ to114‰ for wild ungulates, but the precise en-richment value probably varies among speciesbecause of differences in digestive physiology(Cerling and Sharp 1996; Koch 1998). For me-dium- to large-bodied mammals, the enrich-ment is ;114‰ (Cerling and Harris 1999;Feranec and MacFadden 2000; Zazzo et al.2000). Therefore, a continuous range ofd13Chydroxyl values is observed for mammalianherbivores, ranging from 216‰ for C3 brows-

ers in humid, closed-canopy conditions to ashigh as 13‰ for hypergrazers. Intermediated13Chydroxyl values are observed for mixed C3/C4 feeders. A cut-off d13Chydroxyl value of 28‰has been proposed to identify pure C3 feeders.d13Chydroxyl values for a pure C4 diet are locatedbetween 22‰ and 13‰ (Lee-Thorp et al.1989; MacFadden and Cerling 1996; Cerling etal. 1997a,b, 1999; Cerling and Harris 1999).

The d15N composition of collagen in mam-mals records their position within the ener-getic pyramid, because each trophic levelabove herbivore is indicated by an increase ind15Ncollagen between 11‰ and 16‰ (average;3.4‰ [Robinson 2001]). For herbivores, theprimary factors that affect the d15N value ex-pressed in collagen are (1) soil synthesis of ni-trogen, (2) the diet of the animal (i.e., if it con-sumes N2-fixing or non-N2-fixing plants), and(3) nitrogen metabolism in that animal (Koch1998). Herbivores from forests generally ex-hibit lower d15Ncollagen values than herbivoresfrom grassland environments as a conse-quence of soil acidity in densely forested hab-itats (Rodiere et al. 1996; Grocke et al. 1997).Plants that fix nitrogen have d15N values thatcluster close to the atmospheric N2 value of0‰, whereas those that do not fix nitrogenand use other sources (e.g., soil NH4

1 andNO3

2) usually show a wider range of d15Nplant

values (Robinson 2001). As a result, animalsconsuming N2-fixing plants exhibit d15Ncollagen

values between 0‰ and 14‰, whereas her-bivores feeding on non-N2-fixing plants re-cord d15Ncollagen values between 12‰ and18‰. The d15N value of plants near marineand/or salt-affected areas is also enriched, re-flecting the higher 15N content of soil nitrateand ammonium in saline environments (Sealyet al. 1987; Matheus 1995; Koch 1998).

The effect of nitrogen metabolism in ani-mals is also very important. The higherd15Ncollagen values observed in animals inhab-iting arid regions result from differences in ni-trogen recycling and/or urea loss associatedwith adaptations for drought tolerance (Am-brose and DeNiro 1986a,b; Grocke 1997a;Grocke et al. 1997; Koch 1998; Schwarcz et al.1999): in arid regions, the diets of herbivoresare low in protein content and the animals aremore dependent on recycling their urea to

215ECOLOGY OF PLEISTOCENE MAMMALS

conserve nitrogen; in such conditions, herbi-vores concentrate urine and excrete concen-trated urea depleted in 15N relative to diet,subsequently causing elevated d15N values.Similarly, the isotopic signature in cave bearshas been related to the physiology of dorman-cy: higher d15Ncollagen values are recorded inthose sites corresponding to colder periods,owing to the reuse of urea in synthesizingamino acids during prolonged periods of dor-mancy (Nelson et al. 1998; Fernandez-Mos-quera et al. 2001). In general, d15Ncollagen valuesare higher in large monogastric herbivoresthan in ruminants (Grocke and Bocherens1996). Elevated nitrogen isotope levels, how-ever, may also indicate either advanced age oryoung suckling animals. Such an elevation,caused by excretion of concentrated urea or byingestion of nutrient-enriched milk, respec-tively, can cause an apparent increase in thatanimal’s trophic level by one (Minagawa andWada 1984; Nelson et al. 1998; Witt and Ay-liffe 2001).

Oxygen isotope ratios reflect prevailing cli-matic conditions (e.g., paleotemperature), butfor a local fauna they also can be used to in-terpret the dietary water source (Ayliffe et al.1992; Kohn et al. 1996; Sponheimer and Lee-Thorp 2001; Harris and Cerling 2002).d18Ohydroxyl parallels the d18O of body water,whose composition is related to three majoroxygen sources: air oxygen, drinking water,and food. Although the oxygen isotope com-position of air is the same worldwide and lo-cal water compositions show limited variabil-ity, diet and physiology differ among animals.The d18O is more positive in plants than intheir source water, which is derived from localrain. Thus, among sympatric herbivorousmammals a more positive result would indi-cate that the species is obtaining most of itswater requirements from the vegetation that iteats rather than from drinking. This dependson the digestive physiology of ungulates andon the water content of their food, which rep-resents only 5% of dry grass but may reach upto 70% of succulent forage.

Use of stable isotopes to reconstruct animaldiets was first demonstrated with collagen(DeNiro and Epstein 1978). More recently, itwas shown that tooth enamel is a faithful re-

corder of diets in the geological record (see re-views in Cerling and Harris 1999 and in Spon-heimer and Lee-Thorp 1999a,b). Carbon andnitrogen isotope ratios in bone collagen are in-dicative of overall diet during the last fewyears of an animal’s life. The information ob-tained from enamel isotopes is more limited,as it only records the dietary preference of theanimal during the time the teeth were formed(in large mammals, around 1.5 years frombirth). Moreover, retrieval of environmentalinformation from isotopic ratios in collagen,bone carbonate, and enamel is further com-plicated by potential diagenetic overprinting.Tooth enamel is much less porous than boneand dentine, and it has greater inorganic con-tent, density, and crystallinity. For these rea-sons, enamel is less susceptible to diageneticchange, reflecting more closely the originalabundance of trace elements and stable iso-topes. Similarly, cortical bone is less prone todiagenetic contamination than porous, can-cellous bone tissue. Because the mineral inbones and teeth, carbonate hydroxylapatite,survives much longer than protein, most iso-tope analyses have been performed in enameland bone hydroxylapatite (see review inSponheimer et al. 1999).

Several methods are available to identify al-teration of collagen, including analysis of C:Nratios and amino acid composition (DeNiroand Epstein 1978; Grocke 1997a; Richards etal. 2000). The acceptable values for the C:N ra-tio are between 2.9 and 3.6. This C:N ratio cri-terion allows for the identification and exclu-sion of collagen that is heavily degraded and/or contaminated. There are no equivalent cri-teria for integrity of stable isotope measure-ments of bioapatite in bone mineral andenamel, and appropriate monitors of its iso-topic integrity are yet to be developed.

Biogeochemical Analyses:Trace Element Ratios

In both fresh and fossil bones the hydroxyl-apatite is far from being chemically pure; rath-er it is substituted with small amounts of car-bonate, citrate, and a host of minor and traceelements. Interest in them stems from the factthat their concentration in bone reflects die-tary intakes. Among the different elements,

216 PAUL PALMQVIST ET AL.

TA

BL

E2.

Rel

ativ

eab

un

dan

ceof

carb

on,o

xyg

en,a

nd

nit

rog

enis

otop

era

tios

(d13

C,d

18O

,an

dd15

N,e

xpre

ssed

in‰

)in

carb

onat

eh

yd

roxy

lap

atit

ean

dco

llag

enm

ater

ial

extr

acte

dfr

omb

one

(b)

and

toot

h(t

)sa

mp

les

coll

ecte

din

larg

em

amm

als

spec

ies

from

Ven

taM

icen

a.T

he

rela

tive

abu

nd

ance

ofst

ron

tiu

mv

s.zi

nc

issh

own

.Ast

eris

kin

dic

ates

skel

etal

rem

ain

sfr

omju

ven

ile

ind

ivid

ual

s(i

.e.,

lim

bb

one

wit

hu

nfu

sed

epip

hy

ses

orm

ilk

toot

h).

Mea

ns

for

spec

ies

( 6on

est

and

ard

dev

iati

onfo

rth

ose

inw

hic

hat

leas

tth

ree

esti

mat

esw

ere

avai

labl

e)ar

eal

sop

rov

ided

.

Spec

ies

Sam

ple

d13C

hyd

roxy

ld18

Ohy

dro

xyl

d13C

coll

agen

d15N

coll

agen

C:N

coll

agen

Sr:Z

n

Mam

mut

hus

mer

idio

nal

isM

amm

uthu

sm

erid

ion

alis

Mam

mut

hus

mer

idio

nal

isM

ean

for

Mam

mut

hus

VM

-443

9(b

)*V

M-O

X1

(t)*

VM

-358

1(b

)

27.

12

7.0

26.

22

6.77

(60.

49)

24.

52

3.9

22.

82

3.73

(60.

86)

221

.4—

220

.62

21.0

14.

7—

12.

91

3.8

3.1 — 3.3

3.2

27.8 —

70.1

48.9

5H

ippo

pota

mu

san

tiqu

us

Hip

popo

tam

us

anti

quu

sH

ippo

pota

mu

san

tiqu

us

Mea

nfo

rH

ippo

pota

mu

s

VM

-412

3(b

)V

M-O

X6

(t)

VM

-429

9(b

)

26.

42

7.8

26.

92

7.03

(60.

24)

22.

12

5.8

23.

72

3.87

(60.

62)

222

.3—

222

.62

22.4

5

17.

7—

17.

41

7.55

3.5 — 3.4

3.45

— —55

.655

.6B

ovin

iaf

f.L

epto

bos

Bov

ini

aff.

Lep

tobo

sB

ovin

iaf

f.L

epto

bos

Bov

ini

aff.

Lep

tobo

sB

ovin

iaf

f.L

epto

bos

Bov

ini

aff.

Lep

tobo

sB

ovin

iaf

f.L

epto

bos

Bov

ini

aff.

Lep

tobo

sB

ovin

iaf

f.L

epto

bos

Mea

nfo

rB

ovin

iaf

f.L

epto

bos

VM

-347

3(b

)V

M-3

503

(b)

VM

-418

5(b

)V

M-4

444

(b)

VM

-OX

7(t

)V

M-3

583

(b)

VM

-347

3(b

)V

M-5

44(b

)V

M-4

164

(b)

27.

32

6.4

27.

02

8.7

27.

72

7.6

26.

92

8.2

27.

7

27.

50(6

0.70

)

22.

22

4.1

22.

82

3.1

25.

22

3.7

23.

22

4.3

23.

3

23.

54(6

0.89

)

222

.12

21.5

— — — —2

21.4

221

.8—

221

.70

(60.

32)

13.

81

3.6

— — — —1

3.5

14.

0—

13.

73(6

0.22

)

3.1

3.3 — — — — 3.3

3.4 —

3.28

( 60.

13)

64.4

61.2

66.6

62.9 —

65.3

62.8

63.6

64.1

63.8

6( 6

1.65

)S

oerg

elia

min

orS

oerg

elia

min

orS

oerg

elia

min

orS

oerg

elia

min

orS

oerg

elia

min

orM

ean

for

Soe

rgel

ia

VM

-344

8(b

)V

M-3

867a

(b)

VM

-386

7b(b

)V

M-O

X10

(t)

VM

-398

2(b

)

28.

22

8.3

28.

52

8.1

28.

72

8.36

(60.

24)

22.

72

2.1

22.

32

2.6

22.

72

2.48

(60.

27)

—2

23.2

223

.5—

223

.02

23.2

3(6

0.25

)

—1

3.2

13.

4—

13.

81

3.47

(60.

31)

— 3.4

3.5 — 3.3

3.40

( 60.

10)

62.8

64.2

65.6 —

63.3

63.9

8( 6

1.23

)H

emit

ragu

sal

bus

Hem

itra

gus

albu

sH

emit

ragu

sal

bus

Hem

itra

gus

albu

sH

emit

ragu

sal

bus

Mea

nfo

rH

emit

ragu

s

VM

-380

2(b

)V

M-3

922

(b)

VM

-OX

11(t

)V

M-3

157

(b)

VM

-344

9(b

)

25.

92

7.3

25.

92

6.8

26.

32

6.44

(60.

61)

23.

42

1.5

23.

32

2.1

22.

92

2.64

(60.

82)

220

.92

20.1

— —2

20.4

220

.47

(60.

40)

14.

01

3.9

— —1

3.7

13.

87(6

0.15

)

3.3

3.4 — — 3.5

3.40

( 60.

10)

53.9

58.0 —

55.5

56.4

55.9

5( 6

1.71

)E

ucla

doce

ros

giu

lii

Euc

lado

cero

sgi

uli

iE

ucla

doce

ros

giu

lii

Euc

lado

cero

sgi

uli

iE

ucla

doce

ros

giu

lii

Euc

lado

cero

sgi

uli

iE

ucla

doce

ros

giu

lii

VM

-329

7(b

)V

M-4

155

(b)

VM

-311

1(b

)V

M-3

556

(b)

VM

-OX

8(t

)V

M-4

181

(b)

VM

-311

1(b

)

27.

42

7.1

26.

72

6.3

28.

32

7.0

26.

2

23.

32

2.8

24.

02

4.2

24.

62

3.7

23.

9

225

.9— —

225

.6—

225

.4—

11.

6— —

11.

3—

11.

8—

3.2 — — 3.6 — 3.4 —

96.9

89.9 — — —

91.2

93.2

217ECOLOGY OF PLEISTOCENE MAMMALS

TA

BL

E2.

Con

tin

ued

.

Spec

ies

Sam

ple

d13C

hyd

roxy

ld18

Ohy

dro

xyl

d13C

coll

agen

d15N

coll

agen

C:N

coll

agen

Sr:Z

n

Euc

lado

cero

sgi

uli

iV

M-3

124

(b)

27.

52

3.1

225

.81

1.4

3.5

94.1

Euc

lado

cero

sgi

uli

iM

ean

for

Euc

lado

cero

sV

M-3

780

(b)

26.

72

7.02

(60.

65)

24.

02

3.73

(60.

57)

—2

25.6

8(1

0.22

)—

11.

53(6

0.22

)—

3.43

(60.

17)

—93

.06

(62.

71)

Dam

asp

.D

ama

sp.

Dam

asp

.D

ama

sp.

Dam

asp

.D

ama

sp.

Dam

asp

.M

ean

for

Dam

a

VM

-305

5(b

)V

M-3

482

(b)

VM

-433

0(b

)V

M-O

X9

(t)

VM

-441

0(b

)V

M-3

047

(b)

VM

-306

0(b

)

26.

62

6.9

27.

92

7.0

27.

52

7.1

27.

82

7.26

(60.

49)

23.

22

2.6

22.

42

2.6

22.

92

3.0

22.

52

2.74

(60.

29)

—2

23.6

— — —2

23.8

223

.22

23.5

3(6

0.31

)

—1

2.5

— — —1

2.9

12.

61

2.67

(60.

21)

— 3.5 — — — 3.3

3.4

3.40

( 60.

10)

68.1

67.2

65.7 —

64.9

67.8

66.2

67.0

( 60.

7)E

quu

sal

tide

ns

Equ

us

alti

den

sE

quu

sal

tide

ns

Equ

us

alti

den

sE

quu

sal

tide

ns

Equ

us

alti

den

sE

quu

sal

tide

ns

Equ

us

alti

den

sE

quu

sal

tide

ns

Equ

us

alti

den

sE

quu

sal

tide

ns

Equ

us

alti

den

sE

quu

sal

tide

ns

Equ

us

alti

den

sM

ean

for

Equ

us

VM

-302

8(b

)V

M-3

119

(b)

VM

-316

2(b

)V

M-3

258

(b)

VM

-343

0(b

)V

M-3

529

(b)

VM

-418

9(b

)V

M-4

421

(b)

VM

-OX

2(t

)V

M-O

X3

(t)

VM

-308

9(b

)V

M-4

421

(b)

VM

-327

9(b

)V

M-3

428

(b)

27.

02

8.6

27.

32

8.1

26.

82

6.9

28.

22

8.7

27.

22

7.5

27.

72

7.1

28.

22

7.6

27.

64(6

0.63

)

22.

52

2.8

23.

02

2.9

22.

32

3.3

22.

32

3.8

23.

42

5.2

23.

52

3.9

24.

32

2.8

23.

29(6

0.82

)

226

.12

26.7

226

.0— —

226

.02

25.5

225

.0— — —

225

.82

26.6

—2

25.9

6(6

0.55

)

12.

01

3.5

13.

0— —

13.

31

3.5

12.

1— — —

12.

81

3.1

—1

2.91

(60.

58)

3.2

3.6

3.5 — — 3.4

3.5

3.4 — — — 3.4

3.5 —

3.44

( 60.

12)

—78

.282

.470

.778

.289

.973

.173

.5 — —72

.678

.482

.580

.178

.15

( 65.

59)

Ste

phan

orhi

nus

etru

scu

sS

teph

anor

hinu

set

rusc

us

Ste

phan

orhi

nus

etru

scu

sS

teph

anor

hinu

set

rusc

us

Ste

phan

orhi

nus

etru

scu

sS

teph

anor

hinu

set

rusc

us

Ste

phan

orhi

nus

etru

scu

sM

ean

for

Ste

phan

orhi

nus

VM

-357

8(b

)V

M-4

487

(b)

VM

-451

0(b

)V

M-O

X4

(t)

VM

-OX

5(t

)V

M-3

610

(b)

VM

-357

8(b

)

26.

92

7.6

27.

52

9.4

28.

02

7.8

27.

52

7.81

(60.

78)

24.

62

3.7

23.

72

3.7

25.

12

4.2

24.

82

4.26

(60.

59)

—2

26.5

226

.6— —

226

.2—

226

.43

(60.

21)

—1

3.9

13.

7— —

13.

5—

13.

70(6

0.20

)

— 3.5

3.3 — — 3.3 —

3.37

( 60.

12)

86.9

91.9

89.3 — —

87.3

90.8

89.2

4( 6

2.17

)H

omot

heri

um

lati

den

sM

egan

tere

onw

hite

iPa

chyc

rocu

tabr

evir

ostr

isC

anis

falc

oner

iC

anis

etru

scu

sU

rsu

set

rusc

us

VM

-434

0(b

)V

M-3

301

(b)

VM

-222

6(b

)V

M-2

261

(b)

VM

-225

4(b

)V

M-1

172

(b)

28.

32

8.1

26.

82

7.6

27.

22

4.9

23.

72

4.2

24.

02

4.8

24.

82

3.3

221

.72

25.2

222

.52

22.7

223

.5—

16.

81

5.8

16.

11

6.5

15.

3—

3.4

3.5

3.4

3.5

3.6 —

41.3

45.8

45.5

48.2

51.9 —

218 PAUL PALMQVIST ET AL.

strontium and zinc have proven particularlyuseful as indicators of paleodietary niches(Toots and Voorhies 1965; Wyckoff and Dob-erenz 1967; Sillen and Kavanagh 1982; Klepin-ger 1984; Nedin 1991). We analyzed the abun-dance of strontium and zinc in the bone tissueof large mammals from Venta Micena with in-ductively coupled plasma atomic emissionspectrometry at Royal Holloway University ofLondon. This was done on the resultant su-pernatant from collagen extraction, evaporat-ed to incipient dryness and re-dissolved in40 mL of 3M HCl. Triplicate analyses resultedin an internal error of ,10%.

Strontium shares many chemical propertieswith calcium, including partial replacement ofcalcium in the metabolic processes of plantsand animals. Strontium levels in bone are aconsequence of strontium levels in the soil andof biological factors (Toots and Voorhies 1965).Plants incorporate strontium in place of cal-cium but the amount is determined by the rel-ative abundance of these elements in the soil.However, different kinds and even differentparts of plants vary in strontium content: treeand bush leaves as well as the exposed partsof succulent herbaceous vegetation are usu-ally enriched, whereas grass leaves and rootsof trees, bushes and grasses appear to be de-pleted (Bowen and Dymond 1955). Animalsincorporate elements differentially; they dis-criminate against strontium, and as a resultthe proportion of Sr:Ca in the tissues and or-gans of herbivores is 50–80% less than the lev-els in the plants that are ingested (Kshirasagaret al. 1966; Schoeninger 1979). Uptake in car-nivores also discriminates against the stron-tium content of the flesh consumed. The pro-portion of Sr:Ca is thus useful for discriminat-ing between plant-eating, primary consumersand flesh-eating, secondary consumers, be-cause this proportion is a function of the initialintake (Richards et al. 2000; Balter et al. 2001).Similarly, strontium levels discriminate be-tween browsing, leaf-eating herbivores andgrazing herbivores, with the former showinghigher contents in their bones. Strontium levelsalso allow identification of specific predator-prey pairs; the Sr:Ca ratio of the carnivore is50–80% less than that of its prey (Toots and

Voorhies 1965; Sillen 1986, 1992; Grupe andKruger 1990; Sillen and Lee-Thorp 1994).

Strontium values can be used to determinepaleodiet provided the following conditionsare met: (1) only bones from the same localityare compared, because trace element abun-dance depends heavily on the geochemistry ofthe soil, which varies greatly between locali-ties; (2) only adult bones are used, because ju-veniles show markedly reduced bone stron-tium levels owing to the very small amountsof strontium in mammalian milk; (3) teethand ribs are discarded from the analysis, be-cause strontium levels in teeth do not changeduring growth and ribs, which are more met-abolically active than other bones, are prone toshort-term resetting, especially in lactating fe-males; (4) more than one species should beused for comparison; and (5) there is no evi-dence of diagenetic alteration (Toots and Voor-hies 1965; Nedin 1991; Sillen 1992; Sillen andLee-Thorp 1994).

Relative levels of zinc and strontium in bonehave also been used as a tool for predictingpaleodietary niches (Nedin 1991). However, incontrast to strontium, zinc levels increasethrough the food chain; carnivores have high-er concentrations than herbivores because zinclevels in blood and flesh are higher than inplants (Elias et al. 1982).

Paleodietary Inferences

Table 2 shows d13Chydroxyl, d18Ohydroxyl, and Sr:Zn values measured in bone and enamel ob-tained from fossil mammals preserved at Ven-ta Micena (n 5 68). Collagen, preserved inonly 37 out of 57 bone samples analyzed, pro-vided d13Ccollagen and d15Ncollagen values. Theprecision for stable-isotope analysis was,0.1‰ for both carbon and oxygen, and,0.2‰ for nitrogen. C:N ratios of the collagenmaterial extracted averaged 3.4, and the ami-no acid composition from four specimens issimilar to that of bone collagen in modernmammals, indicating good preservation (Fig.6).

The recovery of collagen in fossils olderthan one million years is not rare, and highamino acid concentrations have been found infossils from Cretaceous–Pleistocene samples(Wyckoff 1972; Ostrom et al. 1994), although

219ECOLOGY OF PLEISTOCENE MAMMALS

FIGURE 6. Amino acid spectra of the modern Tammar Wallaby (Macropus eugenii) (Grocke 1997a) and the averageamino acid abundance (hyp 5 hydroxyproline; asp 5 aspartic acid; thr 5 threonine; ser 5 serine; glu 5 glutamicacid; pro 5 proline; gly 5 glycine; ala 5 alanine; val 5 valine; met 5 methionine; ileu 5 isoleucine; leu 5 leucine;tyr 5 tyrosine; phe 5 phenylalanine; hyl 5 hydroxylysine; lys 5 lysine; his 5 histidine; arg 5 arginine) of fourherbivore specimens from Venta Micena (VM-3473, VM-3867b, VM-3297, VM-3047). The similarity in compositionof extant and extinct specimens indicates that the bone collagen preserved in the fossils was not altered duringdiagenesis.

as noted above a number of specimens ana-lyzed here yielded no collagen. Other fossilproteins (e.g., albumin and immunoglobulin)have been detected in fossil samples from Ven-ta Micena with immunological techniques(Torres et al. 2002). We have some reservationsconcerning the isotopic signature recorded inthe bone apatite samples, however, because aprevious geochemical study (Arribas andPalmqvist 1998) showed that in several fossilsa fraction of the original hydroxylapatite wasreplaced during diagenesis by other mineralphases. Therefore, the following discussion ofcarbon isotopes focuses on the analysis of col-lagen.

Carbon Isotope Ratios. Figure 7 shows arange of d13Ccollagen values from 220‰ to227‰ for large mammal species preserved atVenta Micena, which agrees well with that ofmodern mammals eating C3 plants. This in-dicates that C4 grasses were not a major con-stituent of the plant ecosystem in southernSpain during early Pleistocene times. In fact,C4 plants are poorly represented today in thevegetation of this semi-desert Mediterraneanregion, constituting only ;5% of grass species(Sagredo 1987) and representing ,1% of totalplant biomass; their growth (from April toSeptember) reflects a marked seasonality (J.M. Nieto personal communication 2002). Sim-

ilar results were reported by Bocherens et al.(1996) and Franz-Odendaal et al. (2002) in theisotopic analyses of terrestrial mammals fromthe early–middle Pleistocene site at Tighenif,Algeria, and from the early Pliocene site atLangebaanweg, South Africa, respectively. Inboth localities all ungulate species showedd13Cenamel values exclusive of a diet of C3 plants.

The inferences that can be drawn fromd13Ccollagen values are complicated by physio-logical factors. According to the hypsodontyindex, the Venta Micena perissodactyls in-clude one grazing species with high-crownedmolars, the horse E. altidens, and another thatwas clearly a mixed feeder or browser givenits low hypsodonty, the Etruscan rhino, S.etruscus. Thus, the similarity in the d13Ccollagen

values shown by both monogastric herbivores(Fig. 7) does not indicate either a similarity intheir feeding requirements or differences withthose of other ungulates that show higher 13C/12C ratios. A t-test indicates that the differencebetween the mean d13Ccollagen values for peris-sodactyls (226.09‰ 6 0.52‰, n 5 11) and ar-tiodactyls (222.95‰ 6 1.78‰, n 5 19) is sta-tistically significant (one-tailed test, adjustedfor small sample size: t 5 6.23, p , 0.001). Thisdifference may owe to the fact that perisso-dactyls, being hindgut fermenters, are lessmetabolically efficient for the extraction of nu-

220 PAUL PALMQVIST ET AL.

FIGURE 7. d13C and d15N values of collagen material extracted from bone samples of those species of large mammalspreserved in the lower Pleistocene site at Venta Micena. The lines represent one standard deviation around the meanfor those species in which three or more measurements of stable isotopes were available (data from Table 2).

trients from plant foods (;70% on average)than ruminants, which are foregut fermenters(Janis 1976; Janis et al. 1994). Of interest to thisstudy, Cerling and Harris (1999) found thatBurchell’s zebras, whose diet is composed of100% grass, are consistently 1–2‰ depleted ind13C values for tooth enamel as comparedwith sympatric ruminant hypergrazers suchas alcelaphine bovids (e.g., buffalo, wilde-beest, hartebeest, and topi) from Kenya, thusimplying a lower isotope enrichment factor forzebras. A similar difference was detected byLee-Thorp and Van der Merwe (1987) in bonehydroxylapatite between zebra and wilde-beest. The higher isotope enrichment of ru-minants is probably related to their higherrates of methane production (six times, on av-

erage) than those of non-ruminant ungulates(Crutzen et al. 1986), because ruminant meth-ane has very negative d13C values (Metges etal. 1990).

The Venta Micena bovids show the highestd13Ccollagen levels among ruminants. The threespecies show hypsodonty values that are com-paratively high, similar to those of moderngrazing and mixed feeding ruminants fromopen habitats. Thus, the more positived13Ccollagen values of bovids relative to those ofperissodactyls (Fig. 7) probably indicate thatthe former predominantly fed on grass, hav-ing a greater metabolic efficiency for the as-similation of carbon. As mentioned above, thecomparison of the hypsodonty values of cer-vids with those of modern deer suggests that

221ECOLOGY OF PLEISTOCENE MAMMALS

the cervids were mixed feeders or browsersfrom closed habitats. The difference betweenthe mean d13Ccollagen values for cervids(224.76‰ 6 1.17‰, n 5 7) and bovids(221.79‰ 6 1.17‰, n 5 10) is statisticallysignificant according to a t-test (t 5 5.15, p ,0.001). Such a difference, especially in the caseof E. giulii, could indicate the influence of a hy-perbrowsing behavior in a forested habitatand the recycling of CO2 (i.e., the canopy ef-fect). Alternatively, cervids may exhibit a low-er metabolic efficiency to assimilate carbon(and then a lower isotope enrichment factor)than bovids, which are more advanced rumi-nants with a better compartmentalized stom-ach.

Nitrogen Isotope Ratios. Figure 7 also showsd15Ncollagen values for large mammals fromVenta Micena. It is interesting to mention thatthe values recorded by all herbivore speciesexcept the hippo agree with those expected inanimals consuming N2-fixing plants. Carni-vores H. latidens, M. whitei, P. brevirostris, C. fal-coneri, and C. etruscus show higher values thanungulates except in the case of H. antiquus andone sample of M. meridionalis. The isotope en-richment in carnivores (mean d15Ncollagen for alltaxa: 16.10‰) relative to herbivores (meand15Ncollagen for all taxa, excluding the hippo:12.94‰) is in good accordance with the en-richment value expected from increasing onetrophic level (;3.4‰), thus suggesting thatthe collagen extracted from the fossils did notundergo diagenetic alteration. The very highd15Ncollagen record of H. antiquus probably im-plies that this species fed predominantly onaquatic, non-N2-fixing plants, instead of on C4

grasses as in the case of modern H. amphibius.Such a difference was also suggested by Boch-erens et al. (1996) for Pleistocene hippopotamifrom several African localities. The elephantsample that shows a high d15Ncollagen value wasobtained from a bone specimen with unfusedepiphyses, which belonged to a young indi-vidual that may still have been suckling; theother sample came from an adult individualand records a d15Ncollagen value similar to thoseof other herbivores.

Interesting differences in d15Ncollagen valuesamong the carnivore species are probably re-lated to specific predator-prey relationships in

this paleocommunity (Palmqvist et al. 1996).The large saber-tooth, H. latidens, shows thehighest d15Ncollagen value and this would indi-cate that it was the top predator in the paleo-community. Interestingly, the single d15Ncollagen

value available for this species (Table 2, Fig. 7)is well above that obtained for the young in-dividual of Mammuthus, suggesting that ju-venile elephants were an important part of thediet of this scimitar cat. The likelihood of suchspecialized hunting behavior is evident in thecase of the related American species H. serum,known in high numbers from the late Pleis-tocene site of Friesenhahn cave, Texas. This lo-cality is interpreted as a saber-tooth den andis associated with numerous skeletal remainsof juvenile mammoths (Rawn-Schatzinger1992; Marean and Ehrhardt 1995). C. falconeri,a cursorial canid that presumably hunted inan open environment, shows the second high-est d15Ncollagen value among the hypercarni-vores. M. whitei, an ambush predator thatprobably inhabited forests, records the lowestd15Ncollagen value. Such differences may indicatethat these carnivores preyed upon differentungulate species, with Megantereon huntingbrowsing and mixed feeding cervids in aclosed habitat, and the wild dogs focusing ongrazing bovids and horses from open envi-ronments. Interestingly, the large hyena P. bre-virostris, a species that scavenged the prey ofthese hypercarnivores (Arribas and Palmqvist1998), shows an intermediate d15Ncollagen value.Finally, the medium-sized canid C. etruscus,whose teeth indicate clearly a more omnivo-rous behavior than that of C. falconeri, recordsthe lowest d15Ncollagen value of all the carni-vores.

In the case of the ungulates, the ecomor-phological analyses indicated that all peris-sodactyl and bovid species presumably inhab-ited an open environment, without tree cover,a finding corroborated by differences in nitro-gen isotope ratios. These species show morepositive d15Ncollagen values (mean for all sam-ples: 13.40‰ 6 0.55‰, n 5 21) than cervids(12.01‰ 6 0.64‰, n 5 7). Such a differenceis statistically significant (t 5 4.37; p , 0.001)and indicates that the latter species tended tofeed in a closed habitat. The lower d15Ncollagen

values in the diet of cervids would have re-

222 PAUL PALMQVIST ET AL.

sulted from greater soil acidity in the forest(Rodiere et al. 1996; Grocke 1997a). This isparticularly the case for the megacerine deerE. giulii, which exhibits the lowest d15Ncollagen

values in the fauna. The high d15Ncollagen valueof S. etruscus, a mixed feeding or browsing rhi-no of open habitats, is congruent with that ex-pected from a large monogastric herbivore(Grocke and Bocherens 1996).

Oxygen Isotope Ratios. Bovids H. albus andS. minor and cervid Dama sp. show the highestd18Ohydroxyl values among herbivores (Table 2).Given that the d18O value in plants is morepositive than in their source water (Sponhei-mer and Lee-Thorp 2001; Harris and Cerling2002), the high d18Ohydroxyl values of these un-gulates suggest that they obtained most oftheir water requirements from the vegetationrather than from drinking. On the contrary,the three megaherbivores represented in theassemblage (i.e., those species with a bodymass above 1000 kg), M. meridionalis, H. antiq-uus, and S. etruscus, as well as the megacerinedeer E. giulii, exhibit the most negatived18Ohydroxyl values, which indicate an increasedwater dependence for these species. d18Ohydroxyl

values of the equid E. altidens and the bisonBovini aff. Leptobos are intermediate but closerto those of megaherbivores and large deer,suggesting that both species had moderatewater dependence. These results agree wellwith expectations from their living closest rel-atives. For example, goats are particularly welladapted for arid conditions, obtaining most oftheir water requirements from the vegetationeaten (Novak 1999), and this also seems tohave been the case for the caprine Hemitragusand the ovibovine Soergelia, according to theirhigh mean values of d18Ohydroxyl. Similarly, themodern fallow deer, D. dama, tolerates morearid climates than the red deer, Cervus elaphus,showing a lower water intake rate per kg ofbody mass (Braza and Alvarez 1987). On thecontrary, large monogastric mammals such asrhinos, hippos and elephants are obligatedrinkers, with greater water requirements(Novak 1999), a characteristic that agrees wellwith the comparatively low d18Ohydroxyl valuesmeasured for fossil Stephanorhinus, Hippopota-mus, and Mammuthus.

Trace Element Ratios. Strontium and zinc

levels also provide additional information onthe diet of large mammals from Venta Micena(Table 2, Fig. 8). As expected from their higherposition within the food chain, carnivoresshow lower Sr:Zn ratios (range 5 41.3–51.9)than herbivores (53.9–96.9, excluding the ju-venile elephant). This suggests that the origi-nal abundances of both trace elements in thefossil bones were not altered during the dia-genesis.

Homotherium records the lowest Sr:Zn value(41.3), as expected from its hypercarnivorousdiet. Megantereon shows a slightly higher ratio(45.8), which suggests a different habitat fromthat of Homotherium and, hence, leaf-browsingungulates (which show higher Sr levels) as themain prey. This interpretation agrees wellwith the results obtained in the ecomorphol-ogical analysis, which indicate that Homother-ium was a pursuit predator whereas Meganter-eon was an ambusher. The giant hyena Pachy-crocuta, a bone crusher that presumably in-habited open habitats, also shows a higher Sr:Zn ratio (45.5) than Homotherium, as expectedfrom its diet, which included large amounts ofbone material (Palmqvist and Arribas 2001).The wild dog, C. falconeri, has a dentition ex-tremely specialized for meat-slicing and re-cords a lower Sr:Zn value (48.2) than the moreomnivorous species C. etruscus (51.9), whichshows a well-developed talonid in the lowercarnassial (Palmqvist et al. 1999). Such differ-ence indicates that in the latter species bonematerial, as well as fruits and insects, proba-bly composed a greater part of the diet.

Concerning the herbivores, Sr:Zn ratios arehigher in Bovini aff. Leptobos (mean 5 63.9,range 5 61.2–66.9, n 5 8), S. minor (mean 564.0, range 5 62.8–65.6, n 5 4), and H. albus(mean 5 56.0, range 5 53.9–58.0, n 5 4) thanin the carnivores. These data agree well withthe ecomorphologically based paleodietaryinference of a grazing niche in an open habitatfor all these bovids, although it is worthy tomention that the goat, H. albus, records some-what low ratios. The medium-sized cervid,Dama sp., shows a relatively low Sr:Zn ratio(mean 5 67.0, range 5 64.9–68.1, n 5 6), butthe large deer, E. giulii, records the highest ra-tio among ungulates (mean 5 93.1, range 589.9–96.9, n 5 5), as expected from its very

223ECOLOGY OF PLEISTOCENE MAMMALS

FIGURE 8. Sr:Zn abundance (%) and d15N collagen values (‰) of bone samples of large mammals preserved inVenta Micena (data from Table 2).

low hypsodonty value, similar to that of mod-ern hyperbrowsers. Such a difference could in-dicate an ecological segregation betweenDama sp. and E. giulii, with the former inhab-iting more open habitats and feeding moregrass. The high Sr:Zn ratio obtained in thecase of S. etruscus (mean 5 89.2, range 5 86.9–91.9, n 5 5) clearly indicates its browsing ormixed feeding diet, a conclusion previouslyinferred from the low values of the hypsodon-ty index and the length of the premolar toothrow. However, in the case of E. altidens themoderately high Sr:Zn ratio (mean 5 78.2,range 5 70.7–89.9, n 5 11) could suggest thatit was a mixed feeder, a conclustion thatwould be in clear disagreement with the graz-ing niche deduced from the very high hyp-sodonty index in this species. This discrep-

ancy could be due, perhaps, to the fact thatthis species had a migratory behavior, as mod-ern zebras do, and thus moved between local-ities. Finally, the strikingly low Sr:Zn ratio es-timated in one sample from M. meridionalis(27.8) may represent a young, suckling indi-vidual; the other sample analyzed, obtainedfrom an adult elephant, records a value (70.1)in agreement with the mixed feeding diet in-ferred for this species (Cerling et al. 1999).

Taxon-Free Characterization of thePaleocommunity

Distribution by habitat type of large mam-mal species in modern African communities isshown in Figure 9. Community types were es-tablished by Reed (1998) according to the veg-etation cover, and species were assigned to

224 PAUL PALMQVIST ET AL.

FIGURE 9. Average proportion of species of different locomotor types (A) and trophic groups (B 5 herbivores, C5 carnivores) in several modern African communities of large mammals, established according to major vegetationtypes (frequencies of species estimated from data in Reed 1998). The corresponding values for the lower Pleistocenesite at Venta Micena are also included (data from Table 1). Categories of vegetative habitats are broadly character-ized and cluster forests (lowland rain, tropical temperate, and montane forests), closed woodlands (including wood-land-bushland transition), bushlands (including medium density woodland and bushland areas), open woodlands,shrublands (including scrub woodlands), and savanna (including grasslands and plains). These habitats providean ecological gradient from areas that are well watered throughout the year (forests) to those that are extremelydry or quite seasonal (shrublands and savannas). Deserts were not included in the analysis because Reed’s samplesize of this habitat type was too small. Locomotor types include species adapted to arboreal locomotion (mostlyprimates), aquatic habitats (e.g., the hippo), mixed terrestrial and arboreal locomotion (e.g., the saber-tooth Me-gantereon), and exclusively terrestrial (e.g., the ungulates except the hippo and most carnivores) and/or fossorialspecies (e.g., the badger). Feeding categories as follows: frugivores, diet .50% of fruit; browsers, .75% leaves;mixed feeders; 25–75% grass; grazers: .75% grass; flesh eaters, .70% vertebrate meat; meat/bone eaters, hyenas;meat/insects, 20–70% meat; omnivores, ,20% meat.

several ecological categories by feeding be-havior and locomotion, following the ap-proach proposed by Andrews et al. (1979).The frequencies of such ecological categoriesin Venta Micena are also shown.

When species are grouped by type of loco-motion (Fig. 9A), the relative abundance ofspecies adapted to arboreal locomotion is pos-itively correlated with the proportion of thehabitat covered by trees and bushes. Con-versely, the proportion of terrestrial species

increases as grass cover increases. The pro-portions of locomotor types estimated forVenta Micena, in which terrestrial species aredominant, are very similar to those in Africansavannas. The average frequencies of differenttypes of primary consumers (Fig. 9B) alsoshow a gradient from closed to open habitats.In forested environments the dominant spe-cies are frugivores (largely represented by pri-mates) and browsing ungulates. In open hab-itats the most abundant species are mixed

225ECOLOGY OF PLEISTOCENE MAMMALS

feeders and grazers. The proportions of suchtrophic categories in Venta Micena are againvery close to those in modern savannas, al-though frugivores are absent from Venta Mi-cena. Further, the frequencies of differenttypes of secondary consumers in Venta Mi-cena (Fig. 9C) are similar to those shown bysuch ecological categories in modern savannasand shrublands. The main difference is theproportion of omnivorous species, which arerepresented exclusively by the bear becausesuids are absent from Europe during the earlyPleistocene.

These results indicate that the region thatsurrounded the Pleistocene lake of Orce atVenta Micena was very similar to that repre-sented in modern African savannas with tallgrass and low bush/tree cover, suggesing thatthe countryside in the Guadix-Baza basin wasrelatively unforested during early Pleistocenetimes, as happens today.

Such reconstruction can be assessed alsothrough an analysis of another aspect of thetrophic relationships of this fauna, followingthe general ecological relationships betweenpopulation density and body size (Damuth1987) and between basal metabolic rate andbody size (Kleiber 1932; McNab 1980). Thistype of approach has previously been appliedto both South and North American faunas atLujan and Rancho La Brea (Farina 1996).

All of the herbivore species found at VentaMicena were classified according to theirprobable diet (those species whose masseswere estimated at ,10 kg were not consid-ered), and their population densities and basalmetabolic rates were estimated (Table 1) ac-cording to the appropriate equations of Da-muth (1987) and Peters (1983). The energy re-quirements for each species were obtained bymultiplying its on-crop biomass by its basalmetabolic rate. A typical assimilation efficien-cy of 50% (of the edible material) was as-sumed, and average actual maintenance me-tabolism was taken to be 2.5 times the basalrate, an average of the values given for modernspecies (2 to 3, according to Peters 1983). Add-ing up the requirements of all the herbivorespecies considered, and converting the units,it follows that they must have needed some683 kJ m22 year21, or 12.5 g C m22 year21, in

habitat primary productivity. The three me-gaherbivores alone account for the consump-tion of almost 186 kJ m22 year21, or 3.4 g C m22

year21.Table 1 lists also the carnivores at Venta Mi-

cena, which range in mass from 5 kg to 375 kg.The on-crop biomass for each species was ob-tained by multiplying the calculated popula-tion density by its body mass. The total on-crop biomass for these carnivores was 71.1 kgkm22, or 497 J m22. Adding up the require-ments of all the large carnivore species, andconverting the units, it follows that they musthave needed about 6.6 kJ m22 year21 as habitatsecondary productivity to sustain their basalmetabolism if an assimilation efficiency of50% was assumed (see Peters 1983). It was alsoassumed that the actual maintenance metab-olism was 2.5 times as high as the basal rate.Therefore, the energy required to carry theirmaintenance metabolism was about 16.5 kJm22 year21. For an on-crop biomass such asthat estimated here, the net above ground pri-mary productivity predicted by the equationof McNaughton et al. (1989) is 14.5 MJ m22

year21, a figure congruent with that expectedfor an open environment such as that pro-posed above. A savanna has a productivity of130 g C m22 year21, or 7.3 MJ m22 year21 (Mar-galef 1980). This savanna possibly had smallwooded areas, as deduced from the presenceof the hyperbrowsing deer Eucladoceros; in thiscase its productivity would have doubled, andthe estimate of productivity would thereforebe congruent. It should be further taken intoaccount that some of the herbivore species arepoorly represented (e.g., Praeovibos sp. and thesmall caprine), which suggests that they mayhave been transient species and thereforeshould not be completely included in the re-quirements of the herbivores. Moreover, ac-cording to nitrogen isotopes, the hippo musthave fed at least partially on aquatic vegeta-tion, which may have reduced even more theoverall herbivores requirements.

As stated earlier, the metabolic require-ments of the large carnivores would have beenfulfilled with 16.5 kJ m22 year 21. The second-ary productivity predicted by the appropriateequations of McNaughton et al. (1989) for anecosystem able to support the herbivore bio-

226 PAUL PALMQVIST ET AL.

mass inferred above is 20 kJ m22 year 21, or abit lower according to the primary productiv-ity estimated after Margalef (1980). Further, ifa growth efficiency of 0.025 is assumed (Peters1983), the secondary productivity must havereached 17.1 kJ m22 year 21. Both figures arevery close to that obtained for the require-ments of the carnivores, suggesting that alllarge carnivore species living in the originalpaleocommunity were finally represented inthe bone assemblage collected by the hyenas.This is congruent with the results obtained byArribas and Palmqvist (1998) in the compari-son of the taxonomic composition of the fau-nal assemblage preserved at Venta Micenawith those from other Plio-Pleistocene locali-ties of the Guadix-Baza Basin.

Acknowledgments

C. Nedin and S. Vizcaıno provided usefulcomments on trace element ratios and eco-morphological analyses, respectively. B. Mar-tınez-Navarro and J. L. Santamarıa providedpart of the tooth measurements from ungu-lates and also helped to collect the samplesused in the biogeochemical analyses. Fundingand analytical facilities for trace elements andstable isotopes were provided by the Depart-ment of Geology at the Royal Holloway Uni-versity of London and also by the Universityof Oxford. Thanks also to J. McDonald, C.Trueman, and an anonymous reviewer fortheir insightful comments and helpful criti-cism of the original manuscript. And, last butnot least, N. Atkins improved the style of thisarticle.

Literature CitedAmbrose, S. H., and M. J. DeNiro. 1986a. Reconstruction of Af-

rican human diet using bone collagen carbon and nitrogenisotope ratios. Nature 319:321–324.

———. 1986b. The isotopic ecology of East African mammals.Oecologia 69:395–406.

Andrews, P., J. Lord, and E. M. Nesbit-Evans. 1979. Patterns ofecological diversity in fossil and modern mammalian faunas.Biological Journal of the Linnean Society 11:177–205.

Anyonge, W. 1993. Body mass in large extant and extinct car-nivores. Journal of Zoology 231:339–350.

———. 1996. Locomotor behaviour in Plio-Pleistocene sabre-tooth cats: a biomechanical analysis. Journal of Zoology 238:395–413.

Arribas, A., and P. Palmqvist. 1998. Taphonomy and paleoecol-ogy of an assemblage of large mammals: hyaenid activity inthe lower Pleistocene site at Venta Micena (Orce, Guadix-BazaBasin, Granada, Spain). Geobios 31(Suppl.):3–47.

———. 1999. On the ecological connection between sabre-toothsand hominids: faunal dispersal events in the lower Pleisto-cene and a review of the evidence for the first human arrivalin Europe. Journal of Archaeological Science 26:571–585.

Ayliffe, L. K., A. M. Lister, and A. R. Chivas. 1992. The preser-vation of glacial-interglacial climatic signatures in the oxygenisotopes of elephant skeletal phosphate. Palaeogeography, Pa-laeoclimatology, Palaeoecology 99:179–191.

Balter, V., A. Person, N. Labourdette, D. Drucker, M. Renard,and B. Vandermeersch. 2001. Les Neandertaliens etaient-ilsessentiellement carnivores? Resultats preliminaires sur les te-neurs en Sr et Ba de la paleobiocenose mammalienne de Saint-Cesaire. Comptes Rendus de l’Academie des Sciences, serie II,332:59–65.

Biknevicius, A. R., and B. Van Valkenburgh. 1996. Design forkilling: craniodental adaptations of predators. Pp. 393–428 inJ. L. Gittleman, ed. Carnivore behavior, ecology, and evolu-tion, Vol. 2. Cornell University Press, Ithaca, N.Y.

Biknevicius, A. R., B. Van Valkenburgh, and J. Walker. 1996. In-cisor size and shape: implications for feeding behaviors in sa-ber-toothed ‘‘cats.’’ Journal of Vertebrate Paleontology 16:510–521.

Bocherens, H., P. L. Koch, A. Mariotti, D. Geraads, and J. J. Jae-ger. 1996. Isotopic biogeochemistry (13C, 18O) and mammalianenamel from African Pleistocene hominid sites. Palaios 11:306–318.

Bowen, H. J. M., and J. A. Dymond. 1955. Strontium and bariumin plants and soils. Proceedings of the Royal Society of Lon-don B 144:355–368.

Braza, F., and F. Alvarez. 1987. Habitat use by red deer and fal-low deer in Donana National Park. Miscellania Zoologica 11:363–367.

Cerling, T. E., and J. M. Harris. 1999. Carbon isotope fraction-ation between diet and bioapatite in ungulate mammals andimplications for ecological and paleoecological studies. Oec-ologia 120:337–363.

Cerling, T. E., and Z. D. Sharp. 1996. Stable carbon and oxygenisotope analysis of fossil tooth enamel using laser ablation.Palaeogeography, Palaeoclimatology, Palaeoecology 126:173–186.

Cerling, T. E., J. M. Harris, S. H. Ambrose, M. G. Leakey, and N.Solounias. 1997a. Dietary and environmental reconstructionwith stable isotope analyses of herbivore tooth enamel fromthe Miocene locality of Fort Ternan, Kenya. Journal of HumanEvolution 33:635–650.

Cerling, T. E., J. M. Harris, B. J. MacFadden, M. G. Leakey, J.Quade, V. Eisenmann, and J. R. Ehleringer. 1997b. Global veg-etation change through the Miocene/Pliocene boundary. Na-ture 389:153–158.

Cerling, T. E., J. M. Harris, and M. G. Leakey. 1999. Browsingand grazing in elephants: the isotope record of modern andfossil proboscideans. Oecologia 120:364–374.

Collatz, G. J., J. A. Berry, and J. S. Clark. 1998. Effects of climateand atmospheric CO2 partial pressure on the global distri-bution of C4 grasses: present, past, and future. Oecologia 114:441–454.

Crutzen, P. J., I. Aslemann, and W. Seiler. 1986. Methane pro-duction by domestic animals, wild ruminants, and other her-bivorous fauna, and humans. Tellus 38B:271–284.

Damuth, J. 1987. Interspecific allometry of population densityin mammals and other animals: the independence of bodymass and population energy use. Biological Journal of the Lin-nean Society 331:193–246.

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

DeNiro, M. J., and S. Epstein. 1978. Influence of diet on the dis-

227ECOLOGY OF PLEISTOCENE MAMMALS

tribution of carbon isotopes in animals. Geochimica et Cos-mochimica Acta 42:495–506.

Edwards, G., and D. A. Walker. 1983. C3, C4: mechanisms, andcellular and environmental regulation, of photosynthesis.Blackwell Scientific, Oxford.

Ehleringer, J. R., T. E. Cerling, and B. R. Helliker. 1997. C4 pho-tosynthesis, atmospheric CO2, and climate. Oecologia 112:285–299.

Elias, R., Y. Hirao, and C. Patterson. 1982. The circumvention ofthe natural biopurification of calcium along nutrient path-ways by atmospheric inputs of industrial lead. Geochimica etCosmochimica Acta 46:2561–2580.

Farina, R. A. 1996. Trophic relationships among Lujanian mam-mals. Evolutionary Theory 11:125–134.

Feranec, R. S., and B. J. MacFadden. 2000. Evolution of the graz-ing niche in Pleistocene mammals from Florida: evidencefrom stable isotopes. Palaeogeography, Palaeoclimatology,Palaeoecology 162:155–169.

Fernandez-Mosquera, D., M. Vila-Taboada, and A. Grandal-d’Anglade. 2001. Stable isotopes (d13C, d15N) from the cavebear (Ursus spelaeus): a new approach to its palaeoenviron-ment and dormancy. Proceedings of the Royal Society of Lon-don B 268:1159–1164.

Fortelius, M. 1985. Ungulate cheek teeth: developmental, func-tional, and evolutionary interrelations. Acta Zoologica Fen-nica 180:1–76.

Franz-Odendaal, T. A., J. A. Lee-Thorp, and A. Chinsamy. 2002.New evidence for the lack of C4 grassland expansions duringthe early Pliocene at Langebaanweg, South Africa. Paleobi-ology 28:378–388.

Freeman, K. H., and L. A. Colarusso. 2001. Molecular and iso-topic records of C4 grassland expansion in the late Miocene.Geochimica et Cosmochimica Acta 65:1439–1454.

Gonyea, W. J. 1976. Behavioral implications of saber-toothed fe-lid morphology. Paleobiology 2:332–342.

Gordon, I. J., and A. W. Illius. 1988. Incisor arcade structure anddiet selection in ruminants. Functional Ecology 2:15–22.

Grocke, D. R. 1997a. Stable-isotope studies on the collagen andhydroxylapatite components of fossils: palaeoecological im-plications. Lethaia 30:65–78.

———. 1997b. Distribution of C3 and C4 plants in the late Pleis-tocene of South Australia recorded by isotope biogeochem-istry of collagen in megafauna. Australian Journal of Botany45:607–617.

———. 1998. Carbon-isotope analyses of fossil plants as a che-mostratigraphic and palaeoenvironmental tool. Lethaia 31:1–13.

Grocke, D. R., and H. Bocherens. 1996. Isotopic investigation ofan Australian island environment. Comptes Rendus del’Academie des Sciences de Paris, serie II, 322:713–719.

Grocke, D. R., H. Bocherens, and A. Mariotti. 1997. Annual rain-fall and nitrogen-isotope correlation in macropod collagen:application as a palaeoprecipitation indicator. Earth and Plan-etary Science Letters 153:279–285.

Grupe, G., and H. H. Kruger. 1990. Feeding ecology of the stoneand pine marten revealed by element analysis of their skele-tons. The Science of the Total Environment 90:227–240.

Guerrero-Alba, S., and P. Palmqvist. 1997. Estudio morfometri-co del caballo de Venta Micena (Orce, Granada) y su compar-acion con los equidos modernos y del Plio-Pleistoceno de Eu-ropa y Africa. Paleontologia i Evolucio 30–31:93–148.

Harris, J. M., and T. E. Cerling. 2002. Dietary adaptations of ex-tant and Neogene African suids. Journal of Zoology 256:45–54.

Janis, C. M. 1976. The evolutionary strategy of the Equidae andthe origins of rumen and cecal digestion. Evolution 30:757–774.

———. 1988. An estimation of tooth volume and hypsodonty in-

dices in ungulate mammals, and the correlation of these fac-tors with dietary preference. In D. E. Russell, J. P. Santoro, andD. Sigogneau, eds. Teeth revisited. Memoires du Museum Na-tional d’Histoire Naturelle C 53:367–387.

———. 1995. Correlations between craniodental morphologyand feeding behaviour in ungulates: reciprocal illuminationbetween living and fossil taxa. Pp. 76–98 in J. J. Thomason, ed.Functional morphology in vertebrate paleontology. Cam-bridge University Press, Cambridge.

Janis, C. M., and D. Ehrhardt. 1988. Correlation of the relativemuzzle width and relative incisor width with dietary pref-erences in ungulates. Zoological Journal of the Linnean So-ciety 92:267–284.

Janis, C. M., I. J. Gordon, and A. W. Illius. 1994. Modellingequid/ruminant competition in the fossil record. HistoricalBiology 8:15–29.

Kleiber, M. 1932. Body size and metabolism. Hilgardia 6:315–353.

Klepinger, L. L. 1984. Nutritional assessment from bone. An-nual Review in Anthropology 13:75–96.

Klepinger, L. L., and R. W. Mintel. 1986. Metabolic consider-ations in reconstructing past diet from stable carbon isotoperatios of bone collagen. Pp. 43–48 in J. G. Olin and M. J. Black-man, eds. Proceedings of the 24th International ArchaeometrySymposium. Smithsonian Institution Press, Washington, D.C.

Koch, P. L. 1998. Isotopic reconstruction of past continental en-vironments. Annual Review of Earth and Planetary Sciences26:573–613.

Kohn, M. J., M. J. Schoeninger, and J. W. Valley. 1996. Herbivoretooth oxygen isotope compositions: effects of diet and phys-iology. Geochimica et Cosmochimica Acta 60:3889–3896.

Kshirasagar, S. G., E. Lloyd, and J. Vaughen. 1966. Discrimina-tion between strontium and calcium in bone and the transferfrom blood to bone in the rabbit. British Journal of Radiology39:131–140.

Lee-Thorp, J., and N. J. Van der Merwe. 1987. Carbon isotopeanalysis of fossil bone apatite. South African Journal of Sci-ence 83:712–715.

Lee-Thorp, J., J. C. Sealy, and N. J. Van der Merwe. 1989. Stablecarbon isotope differences between bone collagen and boneapatite, and their relationship to diet. Journal of Archaeolog-ical Science 16:585–599.

Lewis, M. E. 1997. Carnivoran paleoguilds of Africa: implica-tions for hominid food procurement strategies. Journal of Hu-man Evolution 32:257–288.

MacFadden, B. J. 2000. Cenozoic mammalian herbivores fromthe Americas: reconstructing ancient diets and terrestrialcommunities. Annual Review of Ecology and Systematics 31:33–59.

MacFadden, B. J., and T. E. Cerling. 1996. Mammalian herbivorecommunities, ancient feeding ecology, and carbon isotopes: a10-million-year sequence from the Neogene of Florida. Jour-nal of Vertebrate Paleontology 16:103–115.

MacFadden, B. J., and B. J. Shockey. 1997. Ancient feeding ecol-ogy and niche differentiation of Pleistocene mammalian her-bivores from Tarija, Bolivia: morphological and isotopic evi-dence. Paleobiology 23:77–100.

Marean, C. W., and C. L. Ehrhardt. 1995. Paleoanthropologicaland paleoecological implications of the taphonomy of a sa-bretooth’s den. Journal of Human Evolution 29:515–547.

Margalef, R. 1980. Ecologıa. Omega, Barcelona.Martin, L. D., J. P. Babiarz, V. L. Naples, and J. Hearst. 2000.

Three ways to be a saber-toothed cat. Naturwissenschaften87:41–44.

Martınez-Navarro, B., and P. Palmqvist. 1996. Presence of theAfrican saber-toothed felid Megantereon whitei (Broom 1937)(Mammalia, Carnivora, Machairodontidae) in Apollonia-1

228 PAUL PALMQVIST ET AL.

(Mygdonia basin, Macedonia, Greece). Journal of Archaeolog-ical Science 23:869–872.

Matheus, P. E. 1995. Diet and co-ecology of Pleistocene short-faced bears and brown bears in eastern Beringia. QuaternaryResearch 44:447–453.

McNab, B. K. 1980. Energetics and the limits to a temperate dis-tribution in armadillos. Journal of Mammalogy 61:606–627.

McNaughton, S. J., M. Oesterheld, D. A. Frank, and K. J. Wil-liams. 1989. Ecosystem-level patterns of primary productivityand herbivory in terrestrial habitats. Nature 341:142–144.

Mendoza, M., C. M. Janis, and P. Palmqvist. 2002. Characteriz-ing complex craniodental patterns related to feeding behav-iour in ungulates: a multivariate approach. Journal of Zoology258:223–246.

Metges, C., K. Kempe, and H. L. Schmidt. 1990. Dependence ofthe carbon isotope contents of breath carbon dioxide, milk,serum and rumen fermentation products on the value of foodin dairy cows. Brown Journal of Nutrition 63:187–196.

Minagawa, M., and E. Wada. 1984. Stepwise enrichment of 15Nalong food chains: further evidence and the relation betweend15N and animal age. Geochimica et Cosmochimica Acta 48:1135–1140.

Nedin, C. 1991. The dietary niche of the extinct Australian mar-supial lion: Thylacoleo carnifex Owen. Lethaia 24:115–118.

Nelson, D. E., A. Angerbjorn, K. Liden, and I. Turk. 1998. Stableisotopes and the metabolism of the European cave bear. Oec-ologia 116:177–181.

Novak, R. M. 1999. Walker’s mammals of the world. Johns Hop-kins University Press, Baltimore.

Ostrom, P. H., J.-P. Zonneveld, and L. L. Robbins. 1994. Organicgeochemistry of hard parts: assessment of isotopic variabilityand indigeneity. Palaeogeography, Palaeoclimatology, Pa-laeoecology 107:201–212.

Palmqvist, P., and A. Arribas. 2001. Taphonomic decoding of thepaleobiological information locked in a lower Pleistocene as-semblage of large mammals. Paleobiology 27:512–530.

Palmqvist, P., B. Martınez-Navarro, and A. Arribas. 1996. Preyselection by terrestrial carnivores in a lower Pleistocene pa-leocommunity. Paleobiology 22:514–534.

Palmqvist, P., A. Arribas, and B. Martınez-Navarro. 1999. Eco-morphological analysis of large canids from the lower Pleis-tocene of southeastern Spain. Lethaia 32:75–88.

Palmqvist, P., M. Mendoza, A. Arribas, and D. R. Grocke. 2002.Estimating the body mass of Pleistocene canids: discussion ofsome methodological problems and a new ‘‘taxon free’’ ap-proach. Lethaia 35:358–360.

Perez-Barberıa, F. J., and I. J. Gordon. 2001. Relationships be-tween oral morphology and feeding style in the Ungulata: aphylogenetically controlled evaluation. Proceedings of theRoyal Society of London B 268:1021–1030.

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-haviour. Illinois State Museum Reports of Investigations 47:1–80.

Reed, K. E. 1998. Using large mammal communities to examineecological and taxonomic structure and predict vegetation inextant and extinct assemblages. Paleobiology 24:384–408.

Richards, M. P., P. B. Pettitt, E. Trinkaus, F. H. Smith, M. Pau-novic, and I. Karavanic. 2000. Neanderthal diet at Vindija andNeanderthal predation: the evidence from stable isotopes.Proceedings of the National Academy of Sciences USA 97:7663–7666.

Robinson, D. 2001. d15N as an integrator of the nitrogen cycle.Trends in Ecology and Evolution 16:153–162.

Rodiere, E., H. Bocherens, J. M. Angibault, and A. Mariotti.1996. Paticularites isotopiques de l’azote chez le chevreuil

(Capreolus capreolus L.): implications pour les reconstitutionspaleoenvironnementales. Comptes Rendus de l’Academie desSciences, serie II, 323:179–185.

Sagredo, R. 1987. Flora de Almerıa: plantas vasculares de laprovincia. Instituto de Estudios Almerienses, Diputacion Pro-vincial de Almerıa, Spain.

Schoeninger, M. J. 1979. Diet and status at Chalcatzingo: someempirical and technical aspects of strontium analysis. Amer-ican Journal of Physical Anthropology 51:295–309.

Schwarcz, H. P., T. L. Dupras, and S. I. Fairgrieve. 1999. 15N en-richment in the Sahara: in search of a global relationship. Jour-nal of Archaeological Science 26:629–636.

Sealy, J. C., N. J. Van der Merwe, J. A. Lee Thorp, and J. L. Lan-ham. 1987. Nitrogen isotopic ecology in southern Africa: im-plications for environmental and dietary tracing. Geochimicaet Cosmochimica Acta 51:2707–2717.

Sillen, A. 1986. Biogenic and diagenetic Sr/Ca in Plio-Pleisto-cene fossils of the Omo Shungura Formation. Paleobiology 12:311–323.

———. 1992. Strontium-calcium ratios (Sr/Ca) of Australopithe-cus robustus and associated fauna from Swartkrans. Journal ofHuman Evolution 23:495–516.

Sillen, A., and M. Kavanagh. 1982. Strontium and paleodietaryresearch. Yearbook of Physical Anthropology 25:67–90.

Sillen, A., and J. A. Lee-Thorp. 1994. Trace-element and isotopicaspects of predator-prey relationships in terrestrial foodwebs.Palaeogeography, Palaeoclimatology, Palaeoecology 107:243–255.

Smith, B. N., and S. Epstein. 1971. Two categories of 13C/12C ra-tios for higher plants. Plant Physiology 47:380–384.

Solounias, N., and S. M. C. Moelleken. 1993. Tooth microwearand premaxillary shape of an archaic antelope. Lethaia 26:261–268.

Spencer, L. M. 1995. Morphological correlates of dietary re-source partitioning in the African Bovidae. Journal of Mam-malogy 76:448–471.

Sponheimer, M., and J. A. Lee-Thorp. 1999a. The alteration ofenamel carbonate environments during fossilization. Journalof Archaeological Science 26:143–150.

———. 1999b. Isotopic evidence for the diet of an early hominid,Australopithecus africanus. Science 283:368–370.

———. 2001. The oxygen isotope composition of mammalianenamel carbonate from Morea Estate, South Africa. Oecologia126:153–157.

Sponheimer, M., K. E. Reed, and J. A. Lee-Thorp. 1999. Com-bining isotopic and ecomorphological data to refine bovid pa-leodietary reconstruction: a case study from the MakapansgatLimeworks hominin locality. Journal of Human Evolution 36:705–718.

Toots, H., and M. R. Voorhies. 1965. Strontium in fossil bonesand the reconstruction of food chains. Science 149:854–855.

Torres, J. M., C. Borja, and E. Garcıa-Olivares. 2002. Immuno-globulin G in 1.6-million-year-old fossil bones from Venta Mi-cena (Granada, Spain). Journal of Archaeological Science 29:167–175.

Turner, A., and M. Anton. 1996. The giant hyena, Pachycrocutabrevirostris (Mammalia, Carnivora, Hyaenidae). Geobios 29:455–468.

Van der Merwe, N. J. 1982. Carbon isotopes, photosynthesis, andarchaeology. American Scientist 70:596–606.

Van Valkenburgh, B. 1985. Locomotor diversity within past andpresent guilds of large predatory mammals. Paleobiology 11:406–428.

———. 1987. Skeletal indicators of locomotor behavior in livingand extinct carnivores. Journal of Vertebrate Paleontology 7:162–182.

———. 1988. Trophic diversity in past and present guilds oflarge predatory mammals. Paleobiology 14:155–173.

229ECOLOGY OF PLEISTOCENE MAMMALS

———. 1989. Carnivore dental adaptations and diet: a study oftrophic diversity within guilds. Pp. 410–435 in J. L. Gittleman,ed. Carnivore behavior, ecology, and evolution. Cornell Uni-versity Press, Ithaca, N.Y.

Williams, S. H., and R. F. Kay. 2001. A comparative test for adap-tive explanations for hypsodonty in ungulates and rodents.Journal of Mammalian Evolution 8:207–229.

Witt, C. B., and L. K. Ayliffe. 2001. Carbon isotope variability inthe bone collagen of red kangaroos (Macropus rufus) is age de-pendent: implications for palaeodietary studies. Journal ofArchaeological Science 28:247–252.

Wyckoff, R. W. 1972. The biochemistry of animal fossils. Scien-technica, Ltd., Bristol, England.

Wyckoff, R. W., and A. R. Doberenz. 1967. The strontium con-tent of fossil teeth and bones. Geochimica et CosmochimicaActa 32:109–115.

Zazzo, A., H. Bocherens, M. Brunet, A. Beauvilain, D. Billiou, H.T. Mackaye, P. Vignaud, and A. Mariotti. 2000. Herbivore pa-leodiet and paleoenvironmental changes in Chad during thePliocene using stable isotope ratios of tooth enamel carbonate.Paleobiology 26:294–309.