Structural Density of Domesticated South American Camelid Skeletal Elements and the

Journal of Archaeological Science (1999) 26, 1347–1368Article No. jasc.1998.0389, available online at http://www.idealibrary.com on

Structural Density of Domesticated South American CamelidSkeletal Elements and the Archaeological Investigation ofPrehistoric Andean Ch’arki

Peter W. Stahl

Department of Anthropology, Binghamton University, Binghamton, NY 13902–6000, U.S.A.

(Received 27 October 1998, revised manuscript accepted 30 November 1998)

This paper presents a standardized series of replicable and comparable density assays, based on the technique of photondensitometry (or absorptiometry), for two native South American camelid taxa, the domesticated llama (Lama glama)and alpaca (L. pacos). Two sets of volume density (VD) measures (in g/cm3) are provided: (1) ‘‘shape-adjusted’’ volumedensity (VDSA) which computes cross-sectional area of each scan site based upon computerized analysis of scanneddigital images; and (2) standard volume density (VDLD/BT) which norms cross-sectional area to a block form. Derivedvalue sets are compared with each other, between camelid specimens and with published figures for camelid, deer andbison skeletal elements. Patterns of structural bone densities between South American camelids and other artiodactylsmight be attributed to unique anatomical expressions of locomotor and dietary adaptations. The density data arecombined with derived meat utility indices to explore the archaeological correlates associated with the production,distribution and consumption of ch’arki, a native Andean dried meat product. The paper examines the validity of anAndean charqui effect (sensu Miller, 1979, An introduction to the ethnoarchaeology of the Andean Camelids. Ph.D. Thesis,University of California at Berkeley), based on the well-known schlepp effect, by comparing hypothetical ch’arki andchalona models of meat preservation. Faunal data from the prehistoric Peruvian site of Chavın de Huantar do notsupport the consumption of imported ch’arki, but could implicate its local production. The data could also support theconsumption of imported chalona, or the importation of live camelids which were subsequently slaughtered, locallyproduced into chalona and consumed. In both cases, the data leave open the possibility that Chavın may have been aproduction centre for either ch’arki or chalona; however, acceptance of the differing interpretations requires somepotentially problematic assumptions. � 1999 Academic Press

Keywords: TAPHONOMY, STRUCTURAL DENSITY, PHOTON DENSITOMETRY, CAMELIDAE,CH’ARKI, ZOOARCHAEOLOGY, SOUTH AMERICA.

Introduction

T he New World camelids include four closelyrelated species that can successfully interbreed:the wild guanaco (Lama guanicoe) and vicuna

(Vicugna vicugna, or L. vicugna) and, the most con-spicuous of the indigenous South American animaldomesticates, the llama (L. glama) and alpaca (L.pacos). A prolonged interest in their early domestica-tion and subsequent spread throughout prehistoricAndean South America has been supplementedrecently by a developing attention to aspects of camelidtaphonomy (e.g. Miller, 1979; Borrero, 1985, 1987,1988, 1990a, 1990b; Elkin & Zanchetta, 1991; MengoniGonalons, 1991, 1995, 1996; Olivera & Nasti, 1993;Tomka, 1994; Elkin, 1995; Kuznar, 1995; Mondini,1995).

One aspect of Andean zooarchaeology that is ofparticular interest for prehistorians is the production,distribution and consumption of camelid meat in theform of dried ch’arki. The English term ‘‘jerky’’, for

13470305–4403/99/111347+22 $30.00/0

strips of deboned and dried meat, is etymologicallyderived from the Quechuan ch’arki, through theSpanish charqui or charque. The desiccation ofdeboned camelid meat yields a product with anextended shelf-life that is easily transportable over longdistances. Both the timing and appearance of archaeo-logical correlates associated with ch’arki productionare important. They offer potential clues for a pastoralcomponent within an important economic system ofecological complementarity which is attributed tonative Andean societies prior to and during the earlyyears of Spanish conquest (Murra, 1975: 59–115).

Taphonomically, ch’arki can be treated as a special-ized Andean form of animal carcass utilization andtransport. Andeanist archaeologists have followed theinnovative suggestions of Miller (1979) who exploredthe archaeological visibility of ch’arki production andconsumption by assessing the frequency of survivingskeletal parts in excavated assemblages. A SouthAmerican ‘‘charqui effect’’ (Miller, 1979:210), coinedfrom the earlier ‘‘schlepp effect’’ of Perkins & Daly

� 1999 Academic Press

1348 P. W. Stahl

(1969:104) and originally modelled after ideas aboutprehistoric North American bison accumulations (e.g.White, 1953:162), considers the patterned over-representation of: (1) head and distal limb bones athigh altitude production centres; and (2) proximal limbbones at lower altitude consumption sites, as coupledarchaeofaunal correlates for prehistoric ch’arki manu-facture and trade. However, despite its popularity inAndean zooarchaeology, Browman (1989:263) criti-cized the applicability of the model, citing the lack ofuniformity in ch’arki production across Andeantime and space. Furthermore, specific problems withPerkins & Daly’s (1968) schlepp effect, and later deri-vations thereof, are certainly well known to zooarchae-ologists (e.g. see various discussions and references inBunn & Kroll, 1986; Turner, 1989; O’Connell, Hawkes& Blurton Jones, 1990; Bartram, 1993; Lyman, 1994;chapter 7; O’Connell, 1995; Monahan, 1998).

Zooarchaeologists must independently assess thevalidity of inferences based upon proportionate esti-mates of intra-skeletal variability. A number of earlierpapers skillfully addressed the significance of interpre-tations which inferred economic utility from survivingskeletal portions in the buried record by treatingstructural density as an ultimate factor in mediatingassemblage formation and survivorship (Lyman, 1985,1991, 1992, 1993, 1994; Grayson, 1988, 1989). It isimportant to develop precise and replicable densityassays for understanding and measuring the underlyingaffect of structural variability within skeletal portionson assemblage representation. This was attempted in anumber of pioneering efforts (e.g. Brain 1967, 1969;Voorhies, 1969; Behrensmeyer, 1975; Binford &Bertram, 1977; Miller,1979:68) and eventually refinedby Lyman (1982, 1984, and see 1994: chapter 7), whoeffectively championed the use of photon densitometryas an accurate and reliable tool for assessing boneassemblage formation and survivourship.

Empirically derived bone density estimates havebecome ‘‘an indispensable tool in taphonomic analy-sis’’ (Kreutzer, 1992:291). The potential for applicationof photon densitometry to taphonomic studies inarchaeology is hindered only by the relatively few taxafor which reliable assays are available. These include:chinook salmon, Oncorhynchus tshawytscha (Butler& Chatters, 1994); largescale sucker, Catostomusmacrocheilus (Butler, 1996); rabbits, Sylvilagusfloridanus, Oryctolagus cuniculus; hares, Lepus spp.(Pavao, 1996; Pavao & Stahl, 1999); marmots,Marmota spp. (Lyman, Houghton & Chambers, 1992);humans, Homo sapiens (Galloway, Willey & Snyder);phocid seals, Phoca spp. (Chambers, 1992; Lyman,1994: chapter 7); New World camelids, Lama spp.(Elkin, 1995; Elkin & Zanchetta, 1991); New Worlddeer, Odocoileus spp.; domestic sheep, Ovis aries;pronghorn antelope, Antilocapra americana (Lyman,1982, 1984, 1994: chapter 7); and bison, Bison bison(Kreutzer, 1992). In the absence of taxon-specificdensity values, assays from closely related taxa might

be cautiously applied; however, ‘‘where great differ-ences in body size and behavioral adaptation existbetween the model and subject taxa, the model may beentirely inappropriate or, at best, provide only a bluntinstrument capable of detecting gross patterning acrossthe assemblage’’ (Kreutzer, 1992:291; see also Lyman,Houghton & Chambers, 1992:565). It important forour taphonomic tool kit to develop a comprehensivecatalogue of precise and reliable assays for as manydifferent taxa as possible.

In this paper, I present structural density valuesobtained through photon densitometry (or absorp-tiometry) for a series of replicable and comparablescan sites on elements from 10 domesticated SouthAmerican camelid skeletons. I compare the resultsto structural bone density from previous studies ofcamelids and other large artiodactyls and offer sometentative interpretations of similarities and differences.In a subsequent section, I combine the structuraldensity values with other actualistic studies that focuson South American camelids, to explore further thearchaeological correlates for prehistoric Andeanch’arki. Hypothetical models of ch’arki and chalonaproduction, distribution and consumption are used toinvestigate alternate interpretations for the preservedcamelid archaeofaunal remains from the Peruvian siteof Chavin de Huantar.

Structural Density of Skeletal Elements fromDomesticated South American Camelids
Previous studies of camelid structural bone density
Miller (1979:68) was the first to undertake the quanti-fication of differential densities between camelidskeletal portions, by determining the specific gravity ofselected post-cranial elements obtained through sur-face collection of recent camelid accumulations. Majorlimb bones were sectioned transversely into three partsand metapodia were bisectioned through the mid-shaft.Each portion, along with intact calcanea, astragali andphalanges, was weighed and respective volumetricdeterminations calculated via water displacement. Hedemonstrated a ‘‘definite correspondence’’ between therelative structural density of long-bone epiphyses andpatterned longitudinal fracturing observed duringcamelid carcass preparation in the southern highlandsof Peru.

Elkin & Zanchetta (1991) obtained bone mineraldensity estimates of selected camelid elements throughthe use of dual energy X-ray densitometry. Analysedmaterials included skeletal elements from at least twoguanacos (L. guanicoe) and two vicunas (V. vicugna),in addition to a few archaeological specimens ofindeterminate Lama sp. Volume density determina-tions for locations considered as most representativeof each bone, or portion thereof, were derivedthrough dividing machine readings of bone mineraldensity by corresponding bone thickness. Respective

Structural Density of South American Camelid Skeletal Elements 1349

determinations for each camelid (Elkin & Zanchetta,1991: Table 2) were highly correlated with each otherand with corresponding values for cervids (Lyman,1984).

In a subsequent study, Elkin (1995) expandedher sample to include one vicuna (V. vicugna), twoguanacos (L. guanicoe) and one llama (L. glama)skeleton. An additional set of camelid bone volumedensity estimates was established using water displace-ment of skeletal elements (excluding the head andpelvis) and sectioned long-bone portions which con-formed to those examined in the earlier study. Volumedensity values for all three taxa displayed a highlypositive and significant correlation; however, onlythose for llama were published as they were con-sidered to be characteristic of all studied camelids(Elkin, 1995: table 2). These assays were used todemonstrate density-mediated survivorship of camelidremains at the early Holocene site of Pintoscayoc 1 innorthwestern Argentina.

The structural density assays published by Elkin &Zanchetta (1991) and Elkin (1995) were derived forvarious camelid taxa using the technique of photonabsorptiometry; however, they are not entirely com-parable with the parallel studies of various vertebratetaxa. Lyman (1984:272–273) originally selected specificscan sites on each element which were structurallydifferent, easy to locate and readily applicable to theanalysis of fragmented remains in archaeological sites.For consistency, and in order to examine underlyinganatomical variation by keeping conditioning factorsconstant, subsequent researchers chose similar scansites for measurement. In most cases, the existingcamelid density figures published by Elkin & Zanchetta(1991) and Elkin (1995) lack both comparability andprecision as they average values for corresponding scansites. Furthermore, published volume density valuesare based on only one adult llama (L. glama) skeleton(Elkin, 1995: table 2), measurements from one juvenilevicuna (V. vicugna) and the combined elements of twoadult guanacos (L. guanicoe) (Elkin & Zanchetta, 1991:Table 2). The research described below expands uponthese innovative foundations.

Table 1. Studied camelid skeletons

Accession number Taxon Sex Origin Remarks

6240 Lama pacos M Central Park Zoo None6362 Lama pacos M No data Skull discarded22815 Lama glama F New York Zoological Soc. Skin discarded35235 Lama glama M New York Zoological Soc. Measurements70068 Lama pacos F New York Zoo Skin and skeleton80113 Lama sp. M New York Zoo Skeleton80295 Lama sp. M New York Zoo Complete skeleton207764 Lama sp. ? New York Zoological Soc. None237998 Lama sp. F No data Skeleton only237999 Lama sp. ? No data Post-cranial only, cervicals sectioned

M, male; F, female.

Materials and methods

Following published procedures (see Lyman, 1984;Kreutzer, 1992; Lyman, Houghton & Chambers,1992), a series of standardized bone sites were scannedby dual-energy X-ray absorptiometry (DEXA) in orderto generate comparable structural density values fordomesticated camelid taxa. Although dual energy pro-jections have been employed for in vivo clinical assess-ments of both hard and soft tissue composition forover a decade, the recent introduction of DEXA hasincreased precision while decreasing radiation dosage(e.g. Mazess et al., 1990; Johnston, Slemenda &Melton, 1991; Lohmann, 1992). Utilizing quantitativedigital radiography, DEXA technology has been usedwith success to generate replicable skeletal densityvalues for bison (Kreutzer, 1992) and salmon (Butler& Chatters, 1994). A detailed description of the instru-mentation and measurement procedures has beenpublished in Kreutzer (1992:275); any departures fromthis approach are detailed below.

Ten adult domesticate camelid skeletons, includingLama sp., L. glama and L. pacos, were obtained onloan through the courtesy of the Department ofMammalogy, American Museum of Natural History(Table 1). For sake of comparability, scan sitelocations used originally by Lyman (1984: Figure 2)and subsequently by Kreutzer (1992: Figure 2) weremeasured wherever possible (Figure 1). Deviations areas follows: (1) camelids have fused radio-ulnae;(2) mandibular sites DN5 and DN6 were repositioned;(3) only two scans through the entire breadth of thescapular blade were measured; and (4) all of thesmaller carpals and tarsals were scanned, some ofwhich can be compared with their counterparts in thebison skeleton. The number of measured scan sitesvaries as some skeletons were incomplete, whereasother elements were omitted from the study due toprior modification. Portions of skeleton AMNHNo. 6362 had been previously drilled for articulateddisplay, and AMNH No. 237999 included sagittallysectioned cervical vertebrae. Following (Kreutzer1992:275), the left member of paired elements was

1350 P. W. Stahl

examined and only substituted by its right counterpartif unavailable.

The machine used in this study was a DP3 Dual-Photon Absorptiometer, with integrated Version 2·1Software, manufactured and developed by the LunarRadiation Corporation, Madison Wisconsin. Except

for collination, which refers to the opening at thebottom of the detector, all machine default values weremaintained during the analysis. A number of elementswere positioned end-to-end and oriented in a stan-dardized position under the passing beam on a back-ing plate of 1 mm thick aluminum, which imparted

DN1 DN2 DN3DN4 DN5

DN6

DN7

DN8

Left mandibleLateral view

AT2

AT3AT1

Atlas

Dorsalview

AX1

AX2

AX3

AxisVentral view

LU3

LU1

LU2

LumbarDorsal view

TH1TH2

ThoracicDorsal view

IL1

IL2

PU1

PU2

IS2

AC1

Right innominateLateral view

SC1

SC2

SacrumDorsal view

RI2 RI1

RI3

RI4

RI5

RibPosterior view

ST1

SternabraDorsal view

CE2

CE1

CervicalDorsal view

IS1

Figure 1. Selected scan sites.


readings ranging from 0·00 to 0·03 gm/cm2 of ‘‘back-ground’’ density. Upon completion of each scan, adigitized image was generated on a linked computermonitor. The machine automatically determines bonewidth (BW) as it detects the outside edges of highdensity bone and outlines its contours. The screenimage of each element could then be checked andmanually adjusted if needed, in order to ensure that thedefined analytical region of interest corresponded as

closely as possible with the contours of the elementoutline. Each scan site was then manually defined bymanipulating the selected region of interest on thescreen image. Beam width was standardized to thesmallest possible setting of 2 mm for all measurements.The machine then automatically calculated bonemineral density (BMD in g/cm2), also referred to aslinear density (LD: Lyman, 1984:273), from thedetected bone mineral content (BMC in g/cm) by

SP3

SP4

SP2

SP1

Right scapulaLateral view

PA1

Left patellaAnterior view

P13 P12 P11

First phalanxMedial view

HU5

HU4

HU3

HU2

HU1

Right humerusPosterior view

RU1

RU2

RU3

RU4

RU5

RU6

Rightradius-ulnaLateral view

MC1

MC2

MC3

MC4

MC5

MC6

MetacarpalPosterior view

P22P23 P21

Second phalanxMedial view

P31

Third phalanxMedial view

FE2FE1

FE3

FE4

FE5

FE6

Right femurAnterior view Left tibia

Anterior view

Right astragalusPosterior view

TI1

TI2

TI3

TI4

TI5AS1

AS2

AS3

Right calcaneumPosterior view

CA1

CA2

CA3

MetatarsalPosterior view

MR1

MR2

MR3

MR4

MR5

MR6

CA4

Figure 1. (cont.). Selected scan sites.

1352 P. W. Stahl

factoring in the dimensions of beam width and BW foreach scan site.

As previously discussed (Lyman, 1984:273;Kreutzer, 1992:283), this derived measure is invalid asit does not consider bone thickness (BT) at each scansite. This additional dimension is crucial for computa-tion because each scan site is a cross-sectional volumein three dimensions (beam width�BW�BT). Eachcross-section of each scan site on each different elementis, of course, variable. Originally, Lyman (1984:280)normalized these cross-sections to a block shape byfactoring in caliper measurements of BT at each site. Avalid and comparable volume density (VD) measurewas then computed through dividing LD (BMD) byBT (VD=LD/BT). However, a square or rectangularshape necessarily simplifies cross-sectional area(Lyman, 1984:280), which reduces shape variabilityand can systematically underestimate VD values. Dif-ferent methods for calculating scan site cross-sectionsand volume estimations have been subsequently used,including: caliper measurement of each site cross-section plotted on graph paper (Kreutzer, 1992); water,displacement of skeletal elements and sectioned long-bone portions (Elkin, 1995); normalizing each site to ageometric shape that most closely approximatesits profile (Pavao, 1996; Pavao & Stahl, 1998); andcomputing site diameter calculated from the bonecircumference at each scan site measured with a linentape (Galloway, Willey & Snyder, 1997). Recent appli-cation of computed tomography (CT) to bone densitydeterminations of caprid long bones appears to haveeliminated this problem. As the technique recordsbeam attenuation from many different directions, CT

provides density assays based upon precise assessmentsof cross-sectional area and scan site volume (Lamet al., 1998:561–562).

My method for calculating bone volume at scan sitesdiffers from these previous estimations. I drew thecross-sectional profiles of each scan site represented inLama sp. skeleton AMNH No. 80113 with the aid of acarpenter’s moulding tool and digital callipers (Figure2). Note that vertebral cross-sections take into accountair space of intervertebral canals, whereas all othercross-sectional estimates do not include internal porespace. Next, I digitized the sketches of each cross-section to produce compatible files in TIFF format.With digital callipers, I measured the distance betweenstandardized osteometric landmarks for each scan siteof all available skeletal elements. I then computed thearea of each site using the Analyze Menu of NIHImage Ver. 1·52, a MacIntosh-based image processingand analysis program distributed as public domainsoftware by the National Institute of Health. Thedistance between landmarks for each scan site wasmanually defined on the screen image, after which thecorresponding known distance was entered and thescale set to centimetres. This procedure was repeatedfor the entire set of scan sites, resulting in exactingareal (cm2) estimates.

I computed VD in two different ways. In order to usemy cross-sectional area estimates, I factored BW (cm)out of BMD (g/cm2) to calculate BMC (g/cm)(e.g. BMC=BMD�BW). I then computed ‘‘shape-adjusted’’ volume density (VDSA) by dividing cross-sectional area (cm2) into BMC (g/cm) (e.g.VDSA=BMC/Area cm2). As cross-sections from only

P1

PisiformMedial view

S1

ScaphoidVentral view

CuneiformVentral view

LunarVentral view

C1 L1

M1

MagnumVentral view

T1

TrapezoidVentral view

U1

UnciformVentral view

T11

First tarsalMedial view

E1

EntocuneiformVentral view

N1

NavicularVentral view

LM1

Lat. malleolusMedial view

CD1

CuboidVentral view

Figure 1. (cont.). Selected scan sites.


one camelid skeleton were drawn, I am assuming thatthey are morphologically representative of all camelidscan site cross-sections; a time-saving assumptionthat necessarily reduces variability and introduces itsown degree of error. I present a second set of VDcalculations in which cross-sectional areas are simplynormalized to a square or rectangular shape, throughdividing BMD by BT for each scan site [e.g. VDLD/BT=LD (BMD)/BT]. This measurement conforms tothe original VD calculations offered by Lyman (1984)

and facilitates comparison with his figures and certainothers.

Results and discussionTable 2 records the averaged VDSA (g/cm3) andVDLD/BT (g/cm3) values for 101 camelid scan sites,including corresponding sample sizes and rankings forthe different volume estimation techniques. The re-lationship between the two sets of averaged volume

TI3 –––

TI4 –

TI5 –––

TI2 –

TI1 ––––

VDSA VDLD/BT

Left tibia

Sacrum

SC2 –––––––

SC1 ––––

Figure 2. Comparison of volume density (VDSA and VDLD/BT) cross-sectional areas.

1354 P. W. Stahl

*I follow Lyman, Houghton & Chambers (1992:564) in usingPearson’s r for interval scale data where the numerical distancebetween values is known to be equivalent, and Spearman’s rho (rs)for ordinal scale data where this is unknown and/or cannot beassumed.

Table 2. Average structural bone density values for selected camelid scan sites

Scan site N

MeanVDSA

(g/cm3) Rank

MeanVDLD/BT

(g/cm3) Rank Scan site N

MeanVDSA

(g/cm3) Rank

MeanVDLD/BT

(g/cm3) Rank

AC1 9 1·89 50 0·76 25AS1 10 1·91 53 1·17 52AS2 10 2·08 60 1·20 54AS3 10 2·14 67 1·44 70AT1 9 1·79 42 0·97 39AT2 9 1·80 43 0·53 10AT3 9 1·94 54 0·58 12AX1 10 0·69 2 0·47 7AX2 10 1·66 37 0·75 24AX3 10 1·35 18 0·46 6C1 10 1·66 36 1·16 51CA1 10 1·54 31 1·02 43CA2 10 3·75 94 2·37 94CA3 10 1·60 32 0·84 29CA4 10 2·73 83 1·56 79CD1 10 1·49 28 0·87 30CE1 10 1·04 10 0·42 4CE2 10 1·33 16 0·41 3DN1 9 2·43 73 1·33 64DN2 9 3·87 96 2·50 97DN3 9 4·38 97 2·84 98DN4 9 3·85 95 2·37 95DN5 9 3·09 87 2·38 96DN6 8 5·00 98 3·26 101DN7 8 7·19 100 3·11 99DN8 8 7·23 101 3·26 100E1 10 2·45 74 1·85 90FE1 10 1·41 22 0·96 38FE2 9 1·03 9 0·90 32FE3 9 1·35 17 0·93 35FE4 9 1·50 29 0·99 41FE5 9 1·36 19 0·83 28FE6 9 0·86 5 0·49 8HU1 9 0·61 1 0·44 5HU2 9 0·84 4 0·61 14HU3 9 1·42 23 0·92 34HU4 9 1·39 21 0·98 40HU5 9 1·30 15 0·81 27IL1 9 3·29 91 1·45 71IL2 9 3·18 89 2·06 91IS1 9 5·04 99 2·25 93IS2 9 2·12 65 1·34 65L1 10 1·86 47 1·13 48LM1 10 2·84 84 1·51 77LU1 10 1·84 45 0·62 16LU2 10 1·91 52 0·72 21LU3 10 3·02 86 1·60 83M1 10 2·59 78 1·35 66MC1 9 1·46 26 1·22 55MC2 9 2·39 71 1·30 63MC3 9 2·12 66 1·50 75

MC4 9 3·43 93 1·85 89MC5 9 2·50 76 1·78 86MC6 9 1·78 41 1·10 46MR1 9 1·47 27 0·94 36MR2 9 2·08 59 1·37 67MR3 9 1·89 49 1·51 76MR4 9 2·92 85 1·82 88MR5 9 2·48 75 1·64 84MR6 9 1·71 39 1·16 50N1 10 2·39 70 1·24 58P1 10 2·61 80 1·59 82P11 9 1·53 30 1·26 60P12 9 3·20 90 2·13 92P13 9 1·90 51 1·27 62P21 9 1·88 48 1·27 61P22 9 1·61 33 1·05 44P23 9 2·41 72 1·24 57P31 8 3·10 88 1·45 72PA1 10 1·45 25 1·07 45PU1 9 1·83 44 1·00 42PU2 8 1·03 8 0·52 9RI4 10 2·56 77 1·67 85RI1 10 1·61 34 1·15 49RI2 10 2·12 64 1·10 47RI3 10 3·36 92 1·80 87RI5 10 2·60 79 1·58 80RU1 9 2·10 62 1·53 78RU2 9 1·26 14 0·61 15RU3 9 1·86 46 1·19 53RU4 9 2·06 57 1·26 59RU5 9 1·75 40 0·88 31RU6 9 1·18 12 0·74 23S1 10 1·98 56 1·23 56SC1 9 1·71 38 0·59 13SC2 9 1·65 35 0·79 26SP1 10 1·10 11 0·72 20SP2 10 2·22 68 0·71 19SP3 10 1·44 24 0·29 1SP4 10 2·12 63 0·71 18ST1 10 0·83 3 0·63 17T1 10 2·34 69 1·47 73T11 7 2·62 81 1·59 81TH1 10 0·91 6 0·40 2TH2 10 1·97 55 0·91 33TI1 9 0·96 7 0·57 11TI2 9 1·25 13 0·73 22TI3 9 2·09 61 1·43 69TI4 9 2·07 58 1·42 68TI5 9 1·36 20 0·95 37U1 10 2·68 82 1·50 74

density values is characterized by a high positive andsignificant correlation (Pearson’s r=0·90, P�0·001)*,graphically illustrated in Figure 3. Variations in rankordering between the two sets of values are affected by

the differing techniques for estimating cross-sectionalarea. This is clearly demonstrated in many of the scansites whose cross-sectional profiles most radicallydepart from a block outline (e.g. AC1, IL1, SP2, SP3,SP4) and/or from which internal air space wassubtracted (e.g. most of the vertebral body sites).Conversely, relative agreement in rank ordering isfound amongst those scan sites whose cross-sectionalprofiles most closely approximate a block outline (e.g.


mandibular, phalangeal and many of the carpal/tarsalsites) and from which internal air space was notsubtracted. Of course, the density value for any scansite cross-section which does not take into accountinternal air space or structural heterogeneity incor-porates a degree of inaccuracy. This is particularlyimportant for long-bone diaphyses with pronouncedmedullary cavities. Marean and colleagues (Marean &Frey, 1997; Lam et al., 1998; Marean, 1998; Marean& Kim, 1998) critically assess the effect of con-sistently underestimating long-bone mid-shaft densityon archaeological interpretations of assemblageaccumulation involving skeletal frequencies. Althoughthe method for estimating scan-site cross-section in thisstudy is offered as a procedure for increasing theaccuracy of bone mineral density calculations, likemost other density studies it inevitably underestimatesthe density of those scan sites for which signifi-cant cross-sectional heterogeneity was not taken intoconsideration.

Comparison of scan site values between camelids. Con-sidered separately, the two sets of density values areinternally consistent. The average of the correlationcoefficients between VDSA values of scan sites from amaximum of 10 camelid skeletons is highly positiveand significant (r=0·79; P�0·001), as is the corre-sponding average of the correlation coefficients forVDLD/BT (r=0·82; P�0·001). The slightly higher cor-relation for normalized cross-sections is understand-able, as these values are inclined toward greaterhomogeneity. This reduction in variability is producedby normalizing all cross-sectional outlines to a blockprofile (see Figure 2).

Despite their internal consistency, however, theaveraged values mask a certain degree of sampleheterogeneity. The skeletal sample upon which thevalues are based includes a mixture of adult males andfemales from at least two taxa of domesticated animalsraised in zoos. There is no way to assess the extent towhich unique handling has affected the computed

values. The effects of special diet or unusual exerciseregime in a zoo environment may be important,especially when considering that the underlying BMC(g/cm) values are much higher than those for anypreviously examined taxon. This is the first structuralbone density study undertaken exclusively on domestictaxa. One inevitable outcome of domestication canbe to increase the degree of individual variation withinan overall population structure when compared tothe wild condition. In particular, camelid AMNHNo. 207764 appears as a potential outlier. Its corre-sponding VDSA values are less clearly, yet significantly,correlated with three other camelids (AMNH No.237998: r=0·49; P�0·001; AMNH No. 35235: r=0·53,P�0·001; AMNH No. 22815, r=0·62, P�0·001) incomparison with all other possible permutations.Sample heterogeneity is certainly an issue when consid-ering the averaged values of adult males and femalesfrom two species. The fact that these two interbreedingtaxa are notoriously difficult to differentiate on thebasis of osteological criteria certainly mitigates some ofthe potential problems created by lumping togetherL. glama and L. pacos scan site values, i.e. they tend tobe morphologically indistinguishable. Nevertheless,despite some degree of overlap, the distinction betweenliving llamas and alpacas is usually clear. Llamas aretypically bred for meat consumption and for wool usedin the production of coarse bags, ropes and rugs. Theirtolerance to a wider variety of elevational conditionsand forage types makes them better suited as mobilebeasts of burden. Alpacas tend to be found only athigher elevations where they are exploited more fortheir finer wool than for meat consumption (Franklin,1982:486, Table 1). Five of the scanned skeletons areidentified only to the level of genus; therefore, thesample of sexed llamas and alpacas is minimal. A veryhigh correlation (r=0·95; P�0·001) characterizes therelationship between VDSA values of two male alpacas(AMNH Nos. 6362 and 6240); however, the number ofscan sites available for comparison is low (N=58), asone skeleton had been previously drilled for articulateddisplay. The available llama sample consists of onemale and one female, but is also characterized by ahigh correlation (r=0·82; P�0·001) between respectiveVDSA values. A comparison between skeletons ofsimilar sex, regardless of taxon (including Lama sp.individuals), reveals little in the way of patterningexcept for an overall highly positive and significantstatistical correlation.

Comparison of scan site values with previous camelidstudies. How do these scan site values compare withthose generated by previous studies of camelidskeletons? Miller’s (1979:68) original density calcula-tions were estimated for selected skeletal elements andportions, including: (1) long-bone end portions of thehumerus, radio-ulna, femur and tibia; (2) metapodialends including shaft portions; and (3) intact calcanea,astragali and first phalanges. Ten examples of each

8

3.5

VD (SA)

VD

(B

T)

5

2.5

3

2

1.5

1

0.5

1 2 3 4 6 70

Figure 3. Scatterplot comparison of volume density (VDSA andVDLD/BT) assays in g/cm3.

1356 P. W. Stahl

Comparison of scan site values with other artiodactyls.The camelid structural density assays compare favour-ably with corresponding values for deer (Lyman, 1984)and bison (Kreutzer, 1992) which were generated by anapproach similar to the one used here. Sample sizesvary as the discrepancies between measured scan sitesfor camelids and other artiodactyls necessitated thefollowing departures: (1) the fused camelid radio-ulnarsites are omitted as they have no counterparts in deer,and were measured somewhat differently in bison; (2)carpal and tarsal comparisons vary as camelids do nothave a fused naviculocuboid and comparable cunei-form, lunar, lateral malleolus and unciform values areavailable for bison only; (3) scapular blade values forcamelid sites SP3 and SP4 are considered to be com-parable to sites SP4 and SP5 respectively in bison anddeer; and (4) the bison study does not include valuesfor the second phalanx mid-toe measurement, thus siteP22 is comparable to site P23 and omits assays for thesternum and patella.

Linear density (LD, or bone mineral density, BMD)values of camelid scan sites are correlated with corre-sponding assays of both deer (r=0·76; P�0·001;N=84) and bison (r=0·77; P�0·001; N=85). UsingKreutzer’s (1992) technique of comparing the 10highest and 10 lowest ranking LD scan sites for each

taxon, it appears that: (1) highest ranking sites similarlycluster in proximal foreleg, distal hind leg and meta-podial elements of camelids, with the addition of twosites on the mandibular body; and (2) lowest rankingsites tend to cluster in scapular and axial elements ofcamelids, with one mandibular site on the coronoidprocess (Table 3). Nevertheless, as LD measures onlytwo-dimensional density of bone mineral (g/cm2, calcu-lated by BMC/BW), no adjustment is made for cross-sectional volume at different scan sites. As Kreutzer(1992: 285), also Lyman, 1984:271) points out, thesecorrelations reflect bone morphology more than theyreflect actual differences in bone mineral content.

Correlations of VDLD/BT are also positive and sig-nificant for camelids and deer (r=0·44; P�001;N=83). However, similar VDLD/BT correlations forcamelids and bison are less strong yet significant(r=0·37; P�0·001; N=84). The 10 most- and 10 least-dense VDLD/BT scan sites are presented in Table 4. Aweaker positive yet significant correlation is noted forcomparisons between scan sites of camelids and deer(r=0·28; P=0·009; N=83) and camelids and bison(r=0·36; P�0·001; N=85) when using VDSA. This iscertainly a predictable outcome of employing cross-sectional volume estimations that increasingly divergefrom those normalized to a block shape. It is alsounderstandable that the differences between correla-tions employing VDSA and VDLD/BT values are far lessdramatic for the bison sample than those for the cervidsample, considering that the bison volumes were calcu-lated using more accurate angular plots on graph paper(Kreutzer, 1992: Figure 4). Although the VDSA valuespotentially offer increased accuracy, comparisonsof previously published structural bone densityvalues between taxa should use VDLD/BT values whereappropriate.

Interpretation of bone density. I suspect that much ofthe patterned similarity and dissimilarity in structuralbone densities between the South American camelidskeletons and their counterparts in deer and bison canbe attributed to anatomical expressions of locomotorand dietary adaptations that emerged during thecourse of a unique evolutionary history (Franklin,1982; Dagg, 1974; 1979; Gautier-Pilters & Dagg, 1981;Webb, 1965; 1972, 1974). Long before the latePliocene dispersal of camelids from their originalNorth American homeland, a number of distinctivechanges in locomotor and feeding adaptations hadoccurred, most notably: (1) the appearance of equallyproportioned fore- and hind-limbs with partially fusedmetapodia and padded digitigrade feet for support onsoft substrates; and (2) the development of hypsodontcheek teeth as an adaption for grazing abrasive vegeta-tion, with the loss of upper incisors and the emergenceof a cropping mechanism in which the lower incisorsbit against split upper lips.

These Miocene progenitors developed the typicalfaster gait of all modern camels in their unique

element were collected from surface scatters of recentcamelid accumulations and respective mean specificgravities were estimated using weight and volumedetermined by water displacement. Each of Miller’s 15specific gravity measurements incorporates a numberof scan sites (Miller, 1979: Figures 2–15), therefore Icompared them with an average of correspondingstructural density values. Rank order correlationsbetween Miller’s specific gravity determinations andboth VDSA (Spearman’s rs=0·795; P�0·001) andVDLD/BT (rs=0·81; P�0·001) values are highly positiveand significant.

Elkin & Zanchetta (1991; (Table 2) provide VD(g/cm3) estimates for selected element portions andcomplete bones, determined by dividing machine-derived measurements by BT. VDLD/BT values wereused, as the respective volumes were normalized to ablock shape, and averaged where appropriate. Aver-aged VDLD/BT assays are correlated with 27 corre-sponding measurements for L. guanicoe (rs=0·68;P�0·001). Averaged VDLD/BT assays are also corre-lated with 26 corresponding measurements for a juven-ile specimen of the smaller and more gracile wild V.vicugna (rs=0·45; P=0·002). Elkin’s (1995) subsequentstudy includes VD (g/cm3) estimates for selected ele-ment portions and complete bones including shaftportions from a llama skeleton. Respective bone vol-umes were estimated via water displacement. AveragedVDSA values are correlated with 32 correspondingmeasurements (rs=0·51; P=0·0026), whereas their cor-relation with corresponding VDLD/BT assays (rs=0·69;P�0·001) is much stronger.


Table 3. Ten most-dense and 10 least-dense LD scan sites held in common, ranked from highest to lowest in mineralcontent

Camelid Bison Deer

Scan site Rank Scan site Rank Scan site Rank

DN4 1 HU3 1 FE6 1FE6 2 HU4 2 MR1 2TI1 3 HU5 3 TI1 3DN5 4 HU1 4 MR2 4HU5 5 MR1 5 CA3 5HU1 6 FE6 6 TI3 6TI5 7 MR2 7 HU5 7MR1 8 MR3 8 HU4 8TI3 9 FE4 9 MR3 8MC1 10 TI3 10 TI5 10

P22 76/77 SP4 76 RI1 75RI3 77/78 SP3 77 P31 76RI4 78/79 DN8 78 DN6 76DN8 79/80 RI1 79 RI4 78LU3 80/81 TH2 80 DN6 78SP3 81/82 RI5 81 DN8 80RI5 82/83 LU3 81 RI5 81PU2 83/82 RI5 83 TH2 81TH2 84/83 SP5 84 PU2 83SP4 85/84 PU2 85 LU3 84

rs=0·77; P�0·001; N=85 rs=0·79; P�0·001·; N=84

Table 4. Ten most-dense and 10 least-dense VDLD/BT scan sites held in common, ranked from highest to lowest inmineral content

Camelid Bison Deer

Scan site Rank Scan site Rank Scan site Rank

DN6 1 AX3 1 TI3 1DN8 2 AT2 2 MR3 2DN7 3 CA2 3 MC3 3DN3 4 DN7 4 MC2 4DN2 5 DN8 4 MR2 5DN5 6 TI3 6 CA2 6DN4 7 AS1 7 HU4 7CA2 8 MC3 8 DN8 8IS1 8 MR3 9 AS3 9P12 9 CA4 9 AS2 9

AT2 76/75 RI1 76 SC1 75PU2 77/76 SC1 76 AX1 75FE6 78/77 SC2 78 SC2 77AX1 79/78 FE6 78 IS2 77AX3 80/79 HU2 80 AX3 77HU1 81/80 HU1 81 AT2 80CE1 82/81 IL1 82 CE2 80CE2 83/82 IS2 83 RI5 82TH1 84/83 SP5 84 AT1 83SP3 85/84 LU2 85 AX2 84

rs=0·39; P�0·001; N=85 rs=0·52; P�0·001; N=84

propensity for a unilateral pace in which both legs ofone side move forward simultaneously and alternatewith their counterparts on the other side. In this way,stride length can be extended and efficiency of energyexpenditure increased, as moving limbs on one side ofthe animal never overlap (Webb, 1972:100). Dagg

(1974, 1979; Gauthier-Pilters & Dagg, 1981:102)observes that in modern camels the sole of the foot isplaced flat on the ground during stride as weight isshifted from one side to the other, followed by forwardpropulsion through pushing off from the toes. There-fore, locomotion involves variable plantigrade to

1358 P. W. Stahl

digitigrade posture, as opposed to the typical unguli-grade stance employed in most artiodactyl locomotion.Foot structure is accordingly unique, including solarpads and two anterior toenails rather than clovenhooves. Lower limbs are characterized by reducedfibulae, extensively fused radio-ulnae, and partiallyfused metapodia whose distal ends resemble aninverted ‘Y’ (Figure 1). Furthermore, unlike mostartiodactyls, the articular ridges of the distal meta-podia do not completely encircle the distal articulation;rather they are restricted to the plantar surface. Thisenables the phalanges to spread widely during weightsupport, the ridges being contacted only when weight isremoved (Webb, 1972:106–107).

The unilateral pace of camelids is well suited forlong-legged animals that move about in flat and openterrain without any need for exceptional manoeuvr-ability. A pacing gait necessarily reduces lateralstability, which is exacerbated by a resulting side-to-side body sway. Manoeuvrability is further decreasedas the pacing animal’s feet tend to be on the groundless in accordance with increased stride length, there-fore a pacer is less capable of immediate directionalchange (Webb, 1972:102). In contrast, cervids avoidpredators by utilizing uneven paths in dense vegetationand require the stability and balance afforded bydiagonal support, along with superior manoeuvrability(Dagg, 1979:1162). The instability created by longperiods of unilateral support and side-to-side sway iscountered to some extent by placing: (1) the limb nearthe body midline; and (2) the weight of the camelid’shead at the end of a relatively long neck which iscarried low during faster gait. The typically wide andsplay-toed feet tend to be larger in the forelimbs as theysupport this heavier counterbalancing weight (Webb,1965:33, 1972:104). Also, the forequarters generate themajor force of locomotion as modern camels tend topull, rather than push themselves (Gauthier-Pilters &Dagg, 1981:105–106).

It appears that by the beginning of the Pleistocene, allama-like ancestor radiated into the South AmericanAndes and pampas from its original homeland in thesouthern portions of North America. The typicallylong-limbed Hemiauchenia appears to have been muchbetter suited for a cursorial existence than its descend-ants, Palaeolama and the recent llamas. During thePleistocene, these latter forms display a marked reduc-tion in metapodial length and an increase in epipodiallength. This suggests that the speed afforded by length-ened metapodia and shortened pro- and epipodia of theplains- and pampa-adapted Hemiauchenia had becomeless emphasized in the more recent New Worldcamelids as they radiated into the more rugged terrainof the Andes, where manoeuvrability and jumping be-came crucial. Further developments in the masticatoryapparatus of the most recent llamines include a shal-lowing of the jaw and the appearance of low-crownedcheek teeth and cervoid premolars. This may indicatean increased consumption of browse and a decreased

consumption of grass, which is typical of the grazingplains adapted Hemiauchenia (Webb, 1974:207–211).Amongst the contemporary New World camelids, L.glama tends toward generalized browsing and grazingin alpine grass and shrub land, while L. pacos inclinesmore toward obligate grazing of alpine grasslands,meadows and marshes (Franklin, 1982:465).

The most striking pattern in Lama spp. structuraldensity values is the consistently high ranking of man-dibular scan sites. These scan sites in deer and bisonalso tend to be relatively high density sites; however,regardless of which method for volume estimation isused, the seven highest density camelid sites on averageare mandibular (Table 4). It is usually the anterior-most scan site (DN1) which is not included amongstthe highest density sites. Scan sites high up on thecoronoid process (DN8) and just below the mandibularcondyle (DN7) are routinely the most dense. Thesesites are also locations of high mineral density for theother examined artiodactyl taxa. It is unclear exactlywhy nearly all camelid mandibular sites have suchconspicuously high values in bone mineral density.The pattern is consistent between sample skeletonsidentified as L. glama and L. pacos, the former agrazer/browser, the latter a browser. Minor variationsin the relative ordering of mandibular scan sites can bedetected between male and female specimens acrossspecies, otherwise the pattern remains very consistent.One possibility for high bone mineral density may bea need for increased weight in the camelid head asit counterbalances the inherent instability of a uni-lateral gait. Also, unlike ruminants, the mandible ofhypsodont camelids bears canines and is characterizedby a completely co-ossified symphysis (Webb, 1965:12–13). The anterior portion of the mandible supportslong, procumbent incisors that crop grass and drybrowse against a unique pad formed by a deeply splitand mobile upper lip which also serves as a prehensilegrasping device (Webb, 1965:6, 1972:108). Anotherpossibility is that the pattern may be unique in zoospecimens due to special care and upkeep.

Excluding the seven high density mandibular sitesfor the moment, the majority of next highest densityscan sites tend to be found in the lower limbs. Portionsof the metapodial shafts and calcaneum are character-istically high in structural bone density for all studiedArtiodactyla. It may be important to note that thehighest lower limb values are found in the distalmetaphyses of camelid metapodia, as opposed to theproximal metaphysis and mid-shaft portions in deer,and the mid-diaphysis portions in bison. However,Lam et al. (1998:564) indicate that these differencesmay simply be an artefact of measurement, as theprocedure mistakes an essentially heterogenous distalcross-section as homogenous. Additionally, the mid-shaft of the camelid first phalanx is relatively high instructural density, as are most of the phalangeal scansites in comparison with deer and bison. This is theprobable reflection of a wide, splay-toed plantigrade to


digitigrade stance used by camelids in unilateral gait asan adaptation to stable substrates. Unlike the obligateunguligrade stance of most cursorial animals, whichplaces enormous compressive load on the metapodium(e.g. Hildebrand, 1985:56), the splayed toes andpadded digitigrade feet of camelids evenly distributeload to distal members including the phalanges andtoe-like partially fused distal metapodia. As expected,the distal articular surfaces of camelid metapodia tendto be relatively less dense than comparable scan sites indeer and bison. This is very likely due to the absence ofencircling ridges on upper articular surfaces, whichenable the toes to spread widely when planted (Webb,1972:102).

New World camelid and deer skeletons are verysimilar in that relatively low structural density sites aredominated by vertebral sites in the neck. Both aretypically long-necked taxa, but for different reasons asthe extended camelid neck is useful for carrying thehead in a low position in order to counterbalanceinstability during faster unilateral gait. Kreutzer(1992:287) attributes a pattern of high density in theanterior cervical vertebrae of bison to nuchal muscula-ture and robusticity needed to support a massive skull.Otherwise, bison and camelids share a number of lowdensity sites, including portions of the scapular blade,the large proximal humeral head and the broad distalarticular portion of the femur. It may be pertinent inthis regard to emphasize the expansion of certain bonesurfaces on the upper portions of camelid shoulderand hip joints. These include an extended spine andacromion process on the scapula, a prominent deltoidcrest on the lateral portion of the humerus and atransverse expansion of the proximal femur. Eachserves as an attachment for powerful abductor musclesthat promote lateral stability as they pull the body in asideways motion over the planted feet, thereby facili-tating control of lateral movement (Webb, 1972:104).Increased limb abductor muscle involvement incamelids requires expanded bone surface area formuscle attachment, which in turn may lead to a relativedecrease in structural density for specific portions ofthe upper appendages.

Archaeological Investigation of PrehistoricAndean Ch’arkiAs a specialized Andean form of animal carcassutilization and transport, ch’arki (also charque,charqui, carki, tsarke, xarque, xarqui, carne seca) isthe most celebrated indigenous meat preservationtechnique in Andean South America. Archaeologicalvisibility of its production and consumption is gener-ally predicated on the basis of camelid skeletal partfrequencies preserved in excavated assemblages. Thesuggested correlates of prehistoric ch’arki provide ameans to explore the combined use of camelid bonemineral density assays and utility indices as a way of

assessing the validity of inferences about economicutility based upon skeletal part frequencies.

Ch’arki production, distribution and consumptionThe literature contains frequent, yet casual, referencesto ch’arki; frustratingly little is devoted in any detail toits actual production. The latter involves desiccation asthe primary means of preserving fresh meat whichwould otherwise rapidly spoil in sub-tropical and tropi-cal climates. The principal outcome of desiccation is toreduce water content which: (1) inhibits microbialgrowth; (2) decreases overall bulk and weight; and(3) causes a proportionate increase in salt, protein, ashand fat content by weight (Norman & Corte, 1985:6,Figure 5). Salted and fully dried ch’arki can have aneffective shelf life of at least 3–4 months (Norman& Corte, 1985:2; see also Garcilaso, 1871:603) andusually much longer (La Barre, 1948:62; Tomka,1994:241). The dried product can have over twice thecaloric yield because the ratio of fresh meat to ch’arkivaries between 2:1 and 4:1 by weight (Norman &Corte, 1985:6; Tomka, 1994:119). Preserved ch’arkican be later rehydrated through prolonged water soak-ing and is most commonly consumed as reconstitutedchips or small pieces in soups and stews.

As a nutritious and easily transportable meat prod-uct with a prolonged shelf life, ch’arki was an import-ant pre-Columbian Andean subsistence resource.Considered somewhat of a luxury in the Incan empire,meat was made available for common consumptiononly during ceremonial occasions and military service.Ch’arki, in particular, was deposited as tribute in statestorehouses to provision imperial armies (Garcilaso,1871:459; Murra, 1980:49). The location and timingfor ch’arki production and consumption in Andeanprehistory is significant. Relevant material correlatescan serve as potential signatures for the inclusion of apastoral component within a pre-Hispanic economicsystem that incorporated complimentary ecologicalzones in the highly heterogenous Andean environment.Today, camelids are commonly bred in the high punagrasslands which lie beyond the elevational limits ofeffective agriculture. High elevation pastoral productsare integrated into a network of production andexchange with harvested or manufactured productsfrom the lower elevational settings of agricultural,jungle, and coastal lands. Ethnically based networks ofproduction and exchange within and between ecologi-cally complimentary zones, whether imposed by stateauthority or enduring as de facto arrangements, con-stituted the lifeblood of human existence in a highlyvariable landscape (Murra, 1975, 1980, 1985; Shimada,1985; Stanish, 1992; Van Buren, 1996).

The majority of references which include any detailof ch’arki production, especially among recent high-land Quechuan and Aymara populations, explicitlymaintain that its manufacture involves thin strips orsheets of deboned meat (e.g. Rowe, 1944:221; La Barre,

1360 P. W. Stahl

1948:62; Cardozo Gonzalez, 1954:98, 198; Hickman,1963:132; Ravines, 1978:186; Norman & Corte,1985:4–5; Johnsson, 1986:40; Flannery, Marcus &Reynolds, 1989:85; Barton et al., 1990:65; Cobo,1990:198; Tomka, 1994:119, 241, 383; Jones, 1996:82).At high altitudes, thin, deboned strips are exposed tothe elements during the driest and coldest months,usually between May and August. They are oftenhung on lines, specially constructed poles, or simplyplaced on roof tops, out of the reach of scavenginganimals (e.g. Hickman, 1964:132; Flannery, Marcus &Reynolds, 1989:58), from as little as 4–5, to as many as25 days (Ravines, 1978:186; Norman & Corte, 1985:22;Tomka, 1994:119). It is important for ch’arki produc-tion that meat cuts be of uniform thickness to controlthe rate and consistency of surface tissue desiccation.Norman & Corte (1985:5, 10–13) describe optimalmantas or large, flat muscle blocks no greater than5 mm thick to ensure uniform salt penetration andconsistent drying. Commenting on Peruvian Indians inthe early half of the 17th century, Father BernabeCobo ([1653]1990:198) clearly details many of thesefeatures.

The plebeians ate very little meat and when they did it wasat festivals and banquets. They ate more dried meat thanfresh, and they prepared the dried meat without salt in thefollowing way. They cut the meat in wide, thin slices. Thenthey put these slices on ice to cure. Once the slices weredried out, they pounded them between two stones to makethem thinner. They called this dried meat charqui.

Despite the contemporary and historic accounts ofch’arki production, we must heed Browman’s(1989:263) caution that the product and its manufac-ture were likely not uniform in prehistoric time andspace. Indeed, recent accounts also mention related,but different techniques. For example, contemporaryAymara use chalona (calona) to refer to an inferiorkind of ch’arki (a term which they properly apply onlyto desiccated llama meat) which is made by splitting,flattening and drying a whole, eviscerated sheep afterremoval of the skin and head (e.g. Forbes, 1870:243;La Barre, 1948:62, 75). Browman (1989:263) describesthin sheets of ch’arki which include flattened head andfoot elements in contemporary Bolivian Aymara mar-kets. Ravines (1978:186) clearly mentions that ch’arkiis reserved for deboned meat, whereas chalona is usedfor meat with the bone left in. Some authors variablyreserve the term chalona for dried mutton (e.g. LaBarre, 1948:58; Flores Ochoa, 1979:41; Johnsson,1986:44), whereas others include mutton as ch’arki (e.g.Barton et al., 1990:65). Cardozo Gonzalez (1954:198)refers to a salted and air-dried meat which differs fromch’arki as it uses deboned meat prepared in a solutionof capsicums, vinegar, cumin, salt and pepper. Also,some authors variably refer to the Spanish cecina(hung, or salted and dried beef), either as a synonymfor ch’arki (Latcham, 1922:81; Maccagno, 1932:41), orspecifically for use with beef (Johnsson, 1986:44) or

pigs (La Barre, 1948:62). Norman & Corte (1985:10)consider pork as unsuitable for ch’arki production as itis overly rich in unsaturated fatty acids. Conversely,meat with too little carcass fat produces an inferior andless palatable product, due to excessive dehydration.

Archaeological correlates of prehistoric Ch’arkiproductionIn his original formulation of the South American‘‘charqui effect’’, Miller (1979:99–100) explicitlyacknowledges the possibility of both geographical andtemporal variation in Andean ch’arki production.However, archaeological visibility of the effect is clearlypredicated on a type of ‘‘ch’arki’’ that is produced withthe bones left in muscle tissue and subsequently distrib-uted as ‘‘joints of meat’’ along with embedded bones.

All parts of the carcass are utilized with the exception ofthe lower legs and the head which are never used. The term‘‘lower leg’’ in this case includes cannons, phalanges andperhaps some carpals or tarsals, depending upon thebutchery technique (or alternate methods used to separatelower from upper limbs). The lower legs are not includedbecause they lack sufficient meat to make them desirable,while the head will not keep for a long time and is setaside for immediate home consumption (Miller, 1979:99,italics added by P. W. S.).

The archaeological visibility of Andean ch’arki produc-tion and distribution would then be potentially visiblethrough a coupled set of archaeofaunal correlatesinvolving: (1) an over-abundance of cranial and distallimb elements at high altitude production centres;and (2) proximal limb elements at lower altitudeconsumption sites.

While heeding Browman’s (1989:263) caveat regard-ing the lack of a spatio-temporal uniformity of Andeanllamoid meat preservation, the validity of Miller’s‘‘charqui effect’’ for archaeological inference clearlydepends upon the manufacture and distribution of ach’arki more akin to the chalona of recent literature;that is, a desiccated product with most meat-bearingbones left in. It is of course the ultimate fate of camelidbones in ch’arki or chalona manufacture that generatesprima facie evidence for zooarchaeologists. Not only isthe exact nature of meat preparation crucial for under-standing initial bone accumulation, but the locationof bone deposition within a sequence of camelidslaughter, butchering/processing and consumption isalso essential. This is illustrated in two hypotheticalversions of ‘‘ch’arki’’ (Figure 4) and ‘‘chalona’’ (Figure5) manufacture and consumption. Each is based upona relatively consistent pattern of contemporary camelidbutchery which divides the animal into seven or eightcarcass portions (e.g. La Barre, 1948:74; Miller,1979:43; Flannery, Marcus & Reynolds, 1989:85;Tomka, 1994:59, 241). Non-meat-bearing lower limbsare separated at the ankle joint, with variably com-plete, partial or no carpal/tarsal riders (e.g. Miller,1979:39; Tomka, 1994:59). Furthermore, descriptions


Figure 4. Spatial and temporal variability of bone assemblage accumulations based upon a hypothetical model of ch’arki production andconsumption.

of Andean butchery suggest that the head is oftenafforded special treatment because, of rapid spoilage,low meat content or brain removal (e.g. Forbes,1870:243; Cardozo Gonzalez, 1954:128; Hickman,1964:132; Nachtigall, 1966:221; Miller, 1979:44;Weismantel, 1988:100; Tomka, 1994:241).

Lower limbs and heads usually remain at thelocation of slaughter (Tomka, 1994:239). These boneassemblages are characterized by elements with rela-tively higher mineral density and lower meat utility:VDSA values range from 1·46 to 7·23 g/cm3 with amean value of 2·67 g/cm3 and meat utility values range

1362 P. W. Stahl

† With the exception of cranial values, which are based on Tomka’s(1994:64) whole body part or element meat utility index (WSEMUI),all meat utility values are derived from Tomka’s (1994:65) adjustedfragmentary skeletal element meat utility index (FSEMUI). Secondand third phalanx values are based on first phalanx values. Allcarpals and tarsals are included in the lower limb segments, withsmaller elements considered as a unit and assigned values derivedfrom adjacent metapodial elements. Ranked ordering of high meatutility is negatively correlated with ranked ordering of high bonemineral density (rs= �0·47; P=0·019; N=24).

from 143 to 1613 g with a mean value of 583 g forbutchery segments†. The transported counterparts canremain at the location of production and/or consump-tion, depending upon which model is used. These boneassemblages are characterized by elements with lowermineral density and higher meat utility: VDSA valuesrange from 0·61 to 5·04 g/cm3 with a mean value of1·75 g/cm3 and meat utility values range from 768 to6912 g with a mean value of 3331 g. A third assem-blage, consisting of the complete skeleton, can variablyaccumulate and deposit at locations of slaughter, pro-duction and consumption. These bone assemblages arecharacterized by elements with intermediate mineraldensity and meat utility: VDSA values range from 0·61to 7·23 g/cm3 with a mean value of 2·15 g/cm3 and meatutility values range from 143 to 6912 g with a meanvalue of 2714 g.

In four out of five hypothetical ch’arki scenarios(Figure 4, cases A, B, D and E), entire skeletal profilesare potentially deposited in areas of production, cer-tainly as bones are an undesirable waste product ofmanufacture. Consumed ch’arki, introduced from anon-local production site, is archaeologically invisible.The potential for a ch’arki schlepp effect (sensu Miller)is possible when camelids are slaughtered at highelevation sites, and head and lower limb elements areretained in the area of slaughter. Major meat-bearingportions are subsequently removed to a nearby site ofproduction (which can also be a site of consumption),yielding the potential deposition of transportedelements (Figure 4, case C).

Chalona production has the potential for introduc-ing greater variability into assemblage accumulationbecause bones are an integral part of the traded andconsumed product (Figure 5). In all locations whereproduction and consumption are carried out, entireskeletons are deposited (Figure 5, cases A and E). Onlyin those cases where production and consumption areseparated, could a chalona schlepp effect be possible.These include three different scenarios: (1) live trans-port to a production site with subsequent trade andconsumption elsewhere (Figure 5, case B); (2) move-ment of a slaughtered carcass to a production site withsubsequent trade and consumption elsewhere (Figure5, case C); and (3) high elevation slaughter and pro-duction with subsequent trade and consumption else-where (Figure 5, case D). In each case transportedbones are deposited at consumption sites.

A comparison of the two hypothetical models sug-gests that determining a precise mechanism for bone

accumulation and deposition at neighbouring high-elevation pastoral and lower-elevation residential/agricultural sites is potentially equifinal in at least twocases. The first (Figure 4, case C) is when the animal isslaughtered at a high-elevation pastoral site wherehead and lower limbs are retained and meat-bearingcarcass portions are transported to a nearby ch’arkiproduction site with subsequent consumption in thelocation of production and/or trade to a non-local site.The second (Figure 5, case D) is when the animal isslaughtered and chalona is produced at a high elevationpastoral site, after which processed meat-bearing car-cass portions are transported to a nearby consumptionsite. In either case, it is difficult to identify the siteof major skeletal deposition as exclusively a site ofproduction or a site of consumption.

Recently, Miller & Burger (1995) have argued thecase for ch’arki consumption at the famous Peruviansite of Chavın de Huantar. Located at 3150 milesabove sea level (masl) on the floor of the Mosna valley,the site lies directly adjacent to an area suitable forhigh altitude agriculture above 3300 masl and within4 km of nearby high puna grasslands above 3800 masl(Miller & Burger, 1995:424). Camelid remains areimportant throughout each occupational phase atChavın. An archaeofaunal research sample (N=12,672) representing three temporal phases: Urabarriu(900–500 ); Chakinani (500–400 ); and Janabarriu(400–200 ), was selected for analysis (Miller &Burger, 1995:427). Miller & Burger (1995:435) suggestthat the Urabarriu sample includes the mixed exploi-tation of domesticated llama and wild vicuna, with asubsequent increase of llama exploitation in the subse-quent Chakinani phase and eventual domination bydomesticated llama in the Janabarriu sample.

Miller & Burger (1995:442) suggest that a highsurvivorship of limb bones and poor representation ofcranial and podial elements in the camelid assemblagefrom Chakinani and Janabarriu contexts indicate con-sumption of ch’arki acquired through trade with localhigh-altitude puna production areas. This contrastswith a relatively high representation of cranial andpodial elements in the Urabarriu camelid assemblage.They do not attribute this to ch’arki production, how-ever, suggesting instead that the relative survivorshipof cranial and lower foot elements in these earlierdeposits is inflated by an attrition of limb shafts fromwild and domestic taxa which were modified into tools(Miller & Burger, 1995:445).

Figure 6 presents scatter plots comparing percentagesurvivorship (Lyman, 1994:256) of camelid remainswith respective bone mineral density and meat utilityvalues from each occupational phase at Chavın. Inter-esting differences are revealed between the earliestUrabarriu phase and its later counterparts. The Ura-barriu camelid assemblage indicates a weak tendencytoward density-mediated survivorship and inclinesnegatively with meat utility; however, neither trend issignificant. Although the sample (N=462) does include


Figure 5. Spatial and temporal variability of bone assemblage accumulations based upon a hypothetical model of chalona production andconsumption.

relatively higher frequencies of cranial and first phalan-geal fragments, I suspect that these weak and insignifi-cant correlations reflect the overall survivorship ofmost identifiable elements within the camelid skeleton.It must be stressed that a significant portion of higher

meat-yielding elements, especially vertebrae and ribs,are difficult to identify osteologically to a higher taxo-nomic level than zoological order. These data are notincluded in the analysis; however, the authors statethat indeterminate artiodactyl elements include ‘‘large

1364 P. W. Stahl

4.5

100

Density (VDSA)

Su

rviv

al (

%)

2.5

80

60

40

20

0.5 1 1.5 2 3 40

Urabarriu

3.5

(rs = 0.24; P = 0.26)

6000

100

Meat utility (FSEMUI)

4000

80

60

40

20

1000 2000 3000 50000

(rs = –0.14; P = 0.53)

4.5

100

Density (VDSA)

Su

rviv

al (

%)

2.5

80

60

40

20

0.5 1 1.5 2 3 40

Chakinani

3.5

(rs = –0.61; P = 0.001)

6000

100


4000

80

60

40

20

1000 2000 3000 50000

(rs = 0.59; P = 0.002)

4.5

100

Density (VDSA)

Su

rviv

al (

%)

2.5

80

60

40

20

0.5 1 1.5 2 3 40

Janabarriu

3.5

(rs = –0.32; P = 0.13)

6000

100


4000

80

60

40

20

1000 2000 3000 50000

(rs = 0.48; P = 0.02)

Figure 6. Scatterplot comparisons of volume density (VDSA in g/cm3) with meat utility (FSEMUI in g) for camelid bone subassemblagesrecovered from three occupational phases at Chavın de Huantar.

numbers of ribs and vertebrae’’ (Miller & Burger,1995:458). The small Chakinani (N=177) and largerJanabarriu (N=1155) camelid samples display dis-tinctly different survivorship from the earliest Urabar-riu phase, although virtually all of the identifiableelements within the camelid skeleton are again present.Both samples tend to be negatively correlated withbone mineral density; however, only the Chakinaniassemblage is statistically significant. Both are signifi-cantly and positively correlated with meat utility. Withthe data at hand, it might be argued that assemblage

survivorship in the later Chakinani and Janabarriusamples at Chavın indicates the local accumulation anddeposition of transported skeletal remains, reflecting abulk utility strategy (e.g. see discussions in Grayson,1988:70; Lyman, 1991:126, 1992, 1994:228; Bartram,1993:116). From an Andean perspective, this couldsuggest: (1) the production and consumption of ch’arkifrom camelids slaughtered at nearby pastoral sites(Figure 4, case C); (2) the consumption of a chalonalocally produced from transported carcasses ofcamelids slaughtered at high elevation sites (Figure 5,


case D); and/or (3) the consumption of chalona tradedfrom high elevation production and slaughter sites(Figure 5, case D).

DiscussionThe use of a chalona model might support Miller &Burger’s (1995) claim that after the earliest Urabarriuphase occupations, the inhabitants of Chavın began toconsume packages of preserved meat that were pro-duced and transported from highland puna sites (Fig-ure 5, case D). However, such a claim would still haveto account for a relatively high occurrence of lower legelements, especially in the Janabarriu context (Miller &Burger, 1995: Table 4). This is acknowledged by theauthors, who suggest that the principal diet of pre-served meat was occasionally supplemented by freshmeat introduced during winter herd culling, or duringseasons when dried meat could not be produced (Miller& Burger, 1995:444). On the other hand, the depositsmight also potentially suggest a chalona pattern inwhich camelids were slaughtered at high-elevationsites with subsequent transportation of higher utilitycarcasses to a lower-elevation production site whichwould necessarily have to be the location of consump-tion and bone deposition as well (Figure 5, case A).

In either case, the interpretation is based upon theproduction of an item akin to contemporary chalona inwhich the major meat-bearing bones remain in themuscle mass. The interpretation of prima facie evidenceis equifinal. The use of a chalona model might suggestthat the inhabitants of the later two Chavın occupa-tions began to consume packages of preserved meatthat were produced and transported from elsewhere or,conceivably, one in which camelids were transportedlive to Chavın, which thereafter served as the site ofslaughter, production, chalona consumption and even-tual bone deposition. However: (1) why would theyhave manufactured a dried meat product with largebones still in place, as this subtracts from portabilityand use in long-distance trade?; (2) wouldn’t bones beprocessed and fragmented further after rehydration ofchalona at the consumption site?; and (3) why wouldthe prehistoric inhabitants of Chavın have gone to thetrouble of importing an elaborately preserved meatproduct, when puna pastoral lands lay within less thanan hour’s walk from the consumption site?

Using a ch’arki model, the nature of skeletal sur-vivorship in the Chakinani and Janabarriu phases canbe accommodated in situations where camelids wereslaughtered in puna pastoral locations, with transpor-tation of high meat yield carcass portions to the valleybottom for production and/or consumption (Figure 4,case C). Taking into account the significant survivor-ship of lower limb elements, especially in Janabarriucontexts, then live camelids could have been trans-ported from puna pastoral sites and subsequently con-sumed, or produced into ch’arki for local consumptionor trade (Figure 4, cases A, B). It would, however, be

doubtful that camelids were only consumed at Chavınas ch’arki in the strict sense of deboned meat, becausethis pattern of consumption would leave few if anybone fragments for the archaeofaunal record. The onlycase where this could agree with the preserved archaeo-faunal record is one in which ch’arki was necessarilyproduced, either for local consumption and/or trade(Figure 4, case C).

It is interesting that in most of the hypotheticalscenarios considered, Chavın would have been a pre-historic centre for the production of preserved camelidmeat. Miller & Burger (1995:449) argue that the sud-den rise of domesticated camelids in later depositsreflects an economic transformation through the needfor more cargo llamas. The eventual abandonment ofhunting afforded more time for valley-centred activitieslike trade, craft work and farming, while placing ahigher dependency on the products of high-altitudeherders and farmers. From earliest times, Chavın wasan important trade centre linking wide areas of high-land, coast and jungle, and by the Janabarriu phase ithad expanded into a large proto-urban centre withevidence for specialist craft production and socialhierarchy (Miller & Burger, 1995:445). It is certainlypossible that one of these trade commodities may havebeen deboned and dried ch’arki produced for export atthe site of Chavın.

The authors suggest that it is impossible to processch’arki at the site’s altitude of 3150 masl (Miller &Burger, 1995:444) as they consider freezing tempera-tures to be necessary for its production. The authorsstate that ‘‘during the dry season, nightly frosts arecommon at elevations above 3300 m and on rareoccasions even affect the valley floor’’ (Miller &Burger, 1995:422). The higher elevation is similar tothat of the Incan capital of Cuzco, where Garcilaso(1871:603) wrote in the mid-16th century that theprevailing climate was sufficiently dry and cold enough,that a joint hung up in a room with open windowswould render a preserved product which could then bekept for up to 100 days. It is important to consider thatlands above 3300 masl which are conceivably subject tosufficient frost for ch’arki production lie directly adja-cent to Chavın. Nevertheless, freezing should not beoveremphasized as a necessary prerequisite for ch’arkimanufacture. Freezing, salting and exposure to sun orwind all can serve to draw moisture out of fresh meat;therefore, freezing temperatures alone need not be vitalfor the production of a dried meat product. In manylow elevation tropical areas today, it is effectivelyproduced with only the aid of salt, sun and sometimeswind (e.g. Norman & Corte, 1985; and see Bartram,1993 for a discussion of African biltong). It is thereforeentirely feasible that deboned llamoid meat could berendered into a dried product either at the site ofChavın or into freeze-dried ch’arki in the immediatevicinity of the site.

Nevertheless, each of these hypothetical assemblageinterpretations requires some potentially problematic

1366 P. W. Stahl

assumptions. Each assumes temporally precise andunvarying analogues for the production and consump-tion of ch’arki and chalona and treats the preservedassemblages as accurate and reliable reflections oforiginal bone accumulation and deposition. Identifica-tion of the two key component sets of skeletal elements(i.e. crania and lower limbs spatially separated fromthe rest of the skeleton) is potentially compromisedthrough significant density-mediated attrition and ob-literation of depositional resolution during assemblageaccumulation, deposition and burial. The discovery ofspatially discrete higher density/lower utility cranialand lower limb packages can be confounded by aspectsof differential destruction. Light density crania may beheavily fragmented through human or carnivore con-sumption and further destroyed or rendered unidenti-fiable through various post-depositional processes.Mandibulae, metapodial shafts and phalanges can beremoved and curated for adornment and tools, poten-tially leaving assemblages composed of upper cheekteeth and distal articular ends of metapodia. By thesame token, highly fragmented limb shafts may not berecovered, curated or identified to a sufficiently fine-scale taxonomic level for inclusion in assemblageinterpretation. This is particularly significant for osteo-logically indistinguishable camelids, as the eliminationof undiagnostic mid-shaft fragments could artificiallyproduce an assemblage seemingly dominated by cra-nial and lower limb fragments. This is, of course, verysimilar to the assemblage patterns of Middle Paleo-lithic collections, which Marean and colleagues(Marean & Frey, 1997; Lam et al., 1998; Marean &Kim, 1998) demonstrate are analytical artefacts pro-duced through the consistent underestimation of theoccurrence of long-bone mid-shafts in archaeologicalinterpretation. Also, the exclusion of highly significantmeat-bearing elements such as vertebrae and ribswhich are osteologically difficult to identify as camelid,further contributes to analytical ambiguity. Indeed, thesame processes that can potentially obscure identifica-tion of higher density/lower utility assemblages canlead to interpretational ambiguity in all of the potentialscenarios offered for ch’arki or chalona production. Inany case, successful interpretation would necessarilyhinge on: (1) the repeated deposition and preservationof similar assemblages at fixed locales through time; (2)accurate resolution of distinct depositional events atcoterminous and functionally linked sites; (3) a tem-porally and spatially enduring ch’arki productionwhich is identical to contemporary descriptions; and(4) the high-resolution identification of skeletal frag-ments with relatively poor taxonomic acuity and theirsubsequent inclusion in assemblage interpretation.

SummarySince their introduction to zooarchaeology, standard-ized density assays derived through the technique of

photon densitometry have proved to be useful tools forassessing important aspects of archaeofaunal assem-blage formation. The accuracy of these measurementsfor archaeological interpretation relies upon thedependability and comparability of standardizedmeasurements obtained from a comprehensive range ofdifferent skeletons. Although the list of studied taxa isgrowing, it remains relatively meagre. This study addsmeasurements for domesticated llama (L. glama) andalpaca (L. pacos) to the list and demonstrates howthese assays can be used with other important tapho-nomic tools to increase the resolution of archaeologicalinference. In conjunction with meat utility indices,volume density measurements are used to examineskeletal frequencies of preserved camelid archaeofau-nal remains in an attempt to elucidate certain ambigui-ties involved in the identification of prehistoric ch’arkiproduction and consumption; an Andean version ofthe famous ‘‘schlepp effect.’’

From a taphonomic perspective, temporal changesin the frequency of the camelid subassemblages at thePeruvian site of Chavın de Huantar may implicate thisimportant pre-Hispanic religious and trade site as aproduction, rather than as simply a consumption,centre for the distribution of preserved ch’arki. Never-theless, any interpretation of the record strongly de-pends upon: (1) the exact definition of ch’arki used; (2)the acceptance of a prehistoric preservation techniquewhich is invariable and constant through time andspace; (3) an assumption that the taphonomic historyof assemblage accumulation, deposition and preserva-tion is understood; (4) the precise identification ofimportant meat-bearing elements with a normally lowtaxonomic acuity; and (5) the unambiguous recog-nition of temporally discrete depositional events inthe archaeological record. Before any of these import-ant problems can be satisfactorily resolved, it ismost probably unwise to accept any behaviouralinterpretation as definitive.

AcknowledgementsThe photon densitometry study was undertaken duringthe academic year 1995–1996 and made possiblethrough National Science Foundation Grant SBR-9421751 awarded to Peter W. Stahl and J. StephenAthens. The camelid skeletons were obtained from theDepartment of Mammalogy, American Museum ofNatural History. I thank Dr Ross McPhee for grantingme permission to transport study specimens and owe adebt of gratitude to various staff members, especiallyFiona Brady, Dr Bryn Mader and the late Karl-HeinzFuchs, who were crucial to the success of this project.I am deeply grateful for continued access to the differ-ent collections housed at the AMNH and thank thevarious curatorial staff, especially Drs Allison Andors(Ornithology) and Robert Voss (Mammalogy), fortheir continued kindness and interest in my research.


Mr Ed Lis, Head of Radiology, United Health ServicesHospitals, generously granted access to the bone den-sitometer and Thomas Cummings RT(R), LRT isthanked for his technical advice, assistance and goodhumour throughout the research. Barnet Pavao cheer-fully aided in all phases of the photon densitometrystudy and Josh Trapani’s keen analytical skills regu-larly rescued me from seemingly hopeless situations. Igratefully acknowledge the additional input of LuısBorrero, Charlie Cobb, Dolores Elkin, Clark Erickson,Ann Hull, Bill Isbell, Michael Little, Lee Lyman, WillieMengoni, Jim Zeidler and an anonymous reviewer;however, I alone must remain responsible for anyerrors of omission or commission.

ReferencesBarton, S. A., Castro Williams, N., Barja, I. & Murillo, F. (1990).

Nutritional characteristics of the Aymara of Northern Chile. In(W. J. Schull & Rothhamer F., Eds) The Aymara: Strategies inHuman Adaptation to a Rigorous Environment. Dordrecht: KluwerAcademic Publishers, pp. 63–74.

Bartram, L. E. Jr. (1993). Perspectives on skeletal part profiles andutility curves from eastern Kalahari ethnoarchaeology. In J.Hudson, Ed.) From Bones to Behavior. Ethnoarchaeological andExperimental Contributions to the Interpretation of Faunal Re-mains. Center for Archaeological Investigations Occasional PaperNo. 21. Carbondale: Southern Illinois University, pp. 115–137.

Behrensmeyer, A. K. (1975). The taphonomy and paleoecology ofPlio-Pleistocene vertebrate assemblages east of Lake Rudolf,Kenya. In Bulletin of the Museum of Comparative Zoology 146,473–578.

Binford, L. R. & Bertram, J. B. (1977). Bone frequencies andattritional processes. In (L. R. Binford, Ed.) For Theory Building inArchaeology. New York: Academic Press, pp. 77–153.

Borrero, L. A. (1985). Taphonomic observations on guanaco skel-etons. Current Research in the Pleistocene 2, 65–66.

Borrero, L. A. (1987). Sites in action: the meaning of guanaco bonesin Fuegian archaeological sites. Archaeozoologia 3, 9–24.

Borrero, L. A. (1988). Estudios tafonomicos en Tierra del Fuego: surelevancia para entender procesos de formacion del registro ar-queologico. In (H. D. Yacobaccio, Ed.) Arqueologia Contempora-nea Argentina: Actualidad y Perspectivas, Buenos Aires: Busqueda,pp. 13–32.

Borrero, L. A. (1990a). Fuego-patagonian bone assemblages andthe problem of communal guanaco hunting. In (L. B. Davis &B. O. K. Reeves, Eds) Hunters of the Recent Past. London: UnwinHyman, pp. 372–399.

Borrero, L. A. (1990b). Taphonomy of guanaco bones in Tierra delFuego. Quaternary Research 34, 361–371.

Brain, C. K. (1967). Hottentot food remains and their bearing on theinterpretation of fossil bone assemblages. Scientific Papers of theNamib Desert Research Station 32, 1–7.

Brain, C. K. (1969). The contribution of Namib Desert Hottentots toan understanding of Australopithecine bone accumulations. Scien-tific Papers of the Namib Desert Research Station 39, 13–22.

Browman, D. L. (1989). Origins and development of Andean pasto-ralism: an overview of the past 6000 years. In (J. Clutton-Brock,Ed.) The Walking Larder. London: Unwin Hyman, pp. 256–268.

Bunn, H. T. & Kroll, E. M. (1986). Systematic butchery by Plio/Pleistocene hominids at Olduvai Gorge, Tanzania. CurrentAnthropology 27, 431–452.

Butler, V. L. (1996). Tui Chub taphonomy and the importance ofmarsh resources in the western great basin of North America.American Antiquity 61, 699–717.

Butler, V. L. & Chatters, J. C. (1994). The role of bone density instructuring prehistoric salmon bone assemblages. Journal ofArchaeological Science 21, 413–424.

Cardozo Gonzalez, A. (1954). Los Auquenidos. La Paz: EditorialCentenario.

Chambers, A. L. (1992). Seal bone mineral density: its effect onspecimen survival in archaeological sites. B.A. Honors Thesis,University of Missouri, U.S.A.

Cobo, B. (1990). Inca Religion and Customs. Austin: University ofTexas Press.

Dagg, A. I. (1974). The locomotion of the camel (Camelus dromedar-ius). Journal of Zoology, London 174, 67–78.

Dagg, A. I. (1979). The walk of large quadrupedal mammals.Canadian Journal of Zoology 57, 1157–1163.

Elkin, D. C. (1995). Volume density of South American camelidskeletal parts. International Journal of Osteoarchaeology 5, 29–37.

Elkin, D. C. & Zanchetta, J. R. (1991). Densitometrıa osea decamelidos—aplicaciones arqueologicas. Shincal 3 1, 195–204.

Flannery, K. V., Marcus, J. & Reynolds, G. (1989). The Flocks ofthe Wamani. A study of llama Herders on the Punas of Ayacucho.San Diego: Academic Press.

Flores Ochoa, J. A. (1979). Herders of Paratia. Philadelphia:Institute for the Study of Human Issues.

Forbes, D. (1870). On the Aymara indians of Bolivia and Peru.Journal of the Ethnological Society of London 2, 193–305.

Franklin, W. F. (1982). Biology, ecology, and relationship to manof the South American camelids. In (M. A. Mares & H. H.Genoways, Eds) Mammalian Biology in South America. SpecialPublications, no. 6. Pittsburgh: Pymatuning Laboratory ofEcology, University of Pittsburgh, pp. 457–489.

Galloway, A., Willey, P. & Snyder, L. (1997). Human bone mineraldensities and survival of bone elements: a contemporary example.In (W. D. Haglund & M. H. Sorg, Eds) Forensic Taphonomy.The Postmortem Fate of Human Remains. Boca Raton: CRC Press,pp. 295–317.

Garcilaso de L Vega, El Inca (1871). First Part of the RoyalCommentaries of the Yncas. 2 vols. London: Hakluyt Society.

Gauthier-Pilters, H. & Dagg, A. I. (1981). The Camel. Its Evolution,Ecology, Behaviour and Relationship to Man. Chicago: Universityof Chicago Press.

Grayson, D. K. (1988). Danger Cave, Last Supper Cave, andHanging Rock Shelter: the faunas. American Museum of NaturalHistory Anthropological Papers 66, 1–130.

Grayson, D. K. (1989). Bone transport, bone destruction, andreverse utility curves. Journal of Archaeological Science 16, 643–652.

Hickman, J. M. (1963). The Aymara of Chinchera, Peru: persistenceand change in a bicultural context. Ph.D. Thesis, Cornell Univer-sity. Ithaca.

Hildebrand, M. (1985). Walking and running. In (M. Hildebrand,D. M. Bramble, K. F. Liem & D. B. Wake, Eds.) FunctionalVertebrate Morphology. Cambridge: The Belknap Press ofHarvard University, pp. 38–57.

Johnsson, M. (1986). Food and Culture among the Bolivian Aymara:Symbolic Expressions of Social Relations. Uppsala Studies inCultural Anthropology 7. Uppsala: Uppsala University.

Johnston, C. C., Slemenda, C. W. & Melton III, L. J. (1991). Clinicaluses of bone densitometry. New England Journal of Medicine 324,1105–1109.

Jones, K. L. (1996). Charqui. Encyclopedia of Latin AmericanHistory and Culture 2, 82.

Kreutzer, L. A. (1992). Bison and deer bone mineral densities:comparisons and implications for the interpretation of archaeo-logical faunas. Journal of Archaeological Science 19, 271–294.

Kuznar, L. A. (1995). Awatimarka. The Ethnoarchaeology of anAndean Herding Community. Fort Worth: Harcourt Brace CollegePublishers.

La Barre, W. (1948). The Aymara Indians of the Lake TiticacaPlateau, Bolivia. Memoir Series 68. New York: American Anthro-pological Association.

Lam, Y. M., Che, X., Marean, C. W. & Frey, C. J. (1998). Bonedensity and long bone representation in archaeological faunas:comparing results from CT and photon densitometry. Journal ofArchaeological Science 25, 559–570.

1368 P. W. Stahl

Latcham, R. E. (1922). Los Animales Domesticos de la AmericaPrecolombiana. Santiago: Imprenta Cervantes.

Lohmann, T. G. (1992). Advances in Body Composition Assessment.Current Issues in Exercise Science Series, Monograph, no. 3.Champaign: Human Kinetics Publishers.

Lyman, R. L. (1982). The taphonomy of vertebrate archaeofaunas:bone density and differential survivorship of fossil classes. Ph.D.Thesis, University of Washington.

Lyman, R. L. (1984). Bone density and differential survivorship offossil classes. Journal of Anthropological Archaeology 3, 259–299.

Lyman, R. L. (1985). Bone frequencies: differential transport, in situdestruction, and the MGUI. Journal of Archaeological Science 12,221–236.

Lyman, R. L. (1991). Taphonomic problems with archaeologicalanalyses of animal carcass utilization and transport. In (J. R.Purdue, W. E. Klippel & B. W. Styles, Eds.) Beamers, Bobwhitesand Blue Points. Tributes to the Career of Paul W. Parmalee.Scientific Papers, no 23. Springfield: Illinois State Museum,pp. 125–138.

Lyman, R. L. (1992). Anatomical considerations of utility curves inzooarchaeology. Journal of Archaeological Science 19, 7–22.

Lyman, R. L. (1993). Density-mediated attrition of bone assem-blages: new insights. In (J. Hudson, Ed.) From Bones to Behavior.Ethnoarchaeological and Experimental Contributions to the In-terpretation of Faunal Remains. Center for Archaeological Inves-tigations Occasional Paper, no. 21. Carbondale: Southern IllinoisUniversity, pp. 324–241.

Lyman, R. L. (1994). Vertebrate Taphonomy. Cambridge:Cambridge University Press.

Lyman, R. L., Houghton, L. E. & Chambers, A. L. (1992). The effectof structural density on marmot skeletal part representationin archaeological sites. Journal of Archaeological Science 19,557–573.

Maccagno, L. (1932). Los Auquenidos Peruanos. Lima: Ministerio deFomento, Direccion de Agricultura y Ganaderıa, Secion deDefensa y Propoganda.

Marean, C. W. (1998). A critique of the evidence for scavenging byNeandertals and early modern humans: new data from KobehCave (Zagros Mountains, Iran) and Die Kelders Cave 1 Layer 10(South Africa). Journal of Human Evolution 35, 111–136.

Marean, C. W. & Frey, C. J. (1997). Animal bones from caves tocities: reverse utility curves as methodological artifacts. AmericanAntiquity 62, 698–711.

Marean, C. W. & Kim, S. Y. (1998). Mousterian large-mammalremains from Kobeh Cave. Current Anthropology 39 (Suppl.),S79–S113.

Mazess, R. B., Barden, H. S., Biseck, J. P. & Hanson, J. (1990).Dual-energy X-ray absorptiometry for total-body and regionalbone-mineral and soft-tissue composition. American Journal ofClinical Nutrition 51, 1106–1112.

Mengoni Gonalons, G. L. (1991). La llama y sus productos prima-rios. Arqueologia (Revista de la Seccion Prehistoria, Institutode Ciencias Antropologicas, Facultad de Filosofia y Letras,Universidad de Buenos Aires) 1, 179–196.

Mengoni Gonalons, G. L. (1995). Importancia socio-economica delguanaco en el perıodo precolombino. In (S. Puig, Ed.) Tecnicaspara el Manejo del Guanaco. Buenos Aires: Union Mundial paraLa Naturaleza, pp. 13–25.

Mengoni Gonalons, G. L. (1996). La domesticacion de los camelidossudamericanos y su anatomıa economica. Zooarqueologıa deCamelidos 2, 33–45.

Miller, G. R. (1979). An introduction to the ethnoarchaeology of theandean camelids. Ph.D. Thesis: University of California, Berkeley.

Miller, G. R. & Burger, R. L. (1995). Our father the cayman, ourdinner the llama: animal utilization at Chavın de Huantar, Peru.American Antiquity 60, 421–458.

Monahan, C. M. (1998). The Hadza carcass transport debaterevisited and its archaeological implications. Journal of Archaeo-logical Science 25, 405–424.

Mondini, N. M. (1995). Artiodactyl prey transport by foxes in punarockshelters. Current Anthropology 36, 520–524.

Murra, J. V. (1975). Formaciones Economicas y Politicas del MundoAndino. Lima: Instituto de Estudianos Peruanos.

Murra, J. V. (1980). The Economic Organization of the Inka State.Research in Economic Anthropology, Supplement 1. Greenwich,Connecticut: JAI Press.

Murra, J. V. (1985). The limits and limitations of the ‘‘verticalarchipelago’’ in the Andes. In (S. Masuda, I. Shimada & C.Morris, Eds.) Andean Ecology and Civilization. Tokyo: Universityof Tokyo Press, pp. 15–20.

Nachtigall, H. (1966). Indianische Fischer, Feldbauer und Viehzuchter.Beitrage zur Peruanischen Volkerkunde. Marburger Studien zurVolkerkunde, Band 2. Berlin Dietrich Reimer.

Norman, G. A. & Corte, O. O. (1985). Dried Salted Meats: Charqueand Carne-de-Sol. FAO Animal Production and Health Paper, no.51. Rome: Food and Agriculture Organization of the UnitedNations.

O’Connell, J. F. (1995). Ethnoarchaeology needs a general theory ofbehavior. Journal of Archaeological Research 3, 205–255.

O’Connell, J. F., Hawkes, K. & Blurton Jones, N. (1990). Reanalysisof large mammal body part transport among the Hadza. Journalof Archaeological Science 17, 301–316.

Olivera, D. & Nasti, A. (1993). Site formation processes in theArgentine Northwest Puna: taphonomic researches on archaeo-faunistic record preservation. Arqueologıa Contemporanea 4, 85–98.

Pavao, B. (1996). Toward a taphonomy of leporid skeletons:photondensitometry assays. Senior Honors Thesis. Binghamton:Binghamton University.

Pavao, B. & Stahl, P. W. (1999). Structural density assays of leporidskeletal elements with implications for taphonomic, actualistic andarchaeological research. Journal of Archaeological Science 26,53–66.

Perkins, D. & Daly, P. (1968). A hunter’s village in Neolithic Turkey.Scientific American 219, 96–106.

Ravines, R. (1978). Tecnologıa Andina. Lima: Instituto de EstudiosPeruanos.

Rowe, J. H. (1944). Inca culture at the time of the Spanish conquest.In (J. H. Steward, Ed.) Handbook of South American Indians, Vol.2. Bulletin 143. Washington D.C.: Bureau of American Ethnology,pp. 183–330.

Shimada, I. (1985). Introduction. In (S. Masuda, I. Shimada & C.Morris, Eds.) Andean Ecology and Civilization. Tokyo: Universityof Tokyo Press, pp. xi–xxxii.

Stanish, C. (1992). Ancient Andean Political Economy. Austin:University of Texas Press.

Tomka, S. A. (1994). Quinua and camelids on the Bolivian Altiplano:an ethnoarchaeological approach to agro-pastoral subsistence pro-duction with an emphasis on agro-pastoral transhumance. Ph.D.Thesis. University of Texas, Austin.

Turner, A. (1989). Sample selection, schlepp effects and scavenging:the implications of partial recovery for interpretations of theterrestrial mammal assemblage from Klasies River Mouth. Journalof Archaeological Science 16, 1–11.

Van Buren, M. (1996). Rethinking the vertical archipelago. Ethnic-ity, exchange, and history in the south central Andes. AmericanAnthropologist 98, 338–351.

Voorhies, M. R. (1969). Taphonomy and Population Dynamics of anEarly Pliocene Vertebrate Fauna, Knox County, Nebraska. Contri-butions to Geology Special Paper, no. 1. Laramie: University ofWyoming.

Webb, S. D. (1965). The osteology of camelops. Bulletin of the LosAngeles County Museum 1, 1–54.

Webb, S. D. (1972). Locomotor evolution in camels. Forma et Funcio5, 99–112.

Webb, S. D. (1974). Pleistocene llamas of Florida, with a brief reviewof the Lamini. In (S. D. Webb, Ed.) Pleistocene Mammals ofFlorida. Gainesville: University of Florida, pp. 170–213.

Weismantel, M. J. (1988). Food, Gender, and Poverty in the Ecuado-rian Andes. Philadelphia: University of Pennsylvania Press.

White, T. E. (1953). Observations on the butchering technique ofsome aboriginal peoples: no. 2. American Antiquity 19, 396–398.

Structural Density of Domesticated South American Camelid Skeletal Elements and the

Documents

Transcript of Structural Density of Domesticated South American Camelid Skeletal Elements and the