The significance of cooking for early hominin scavenging

lable at ScienceDirect

Journal of Human Evolution xxx (2015) 1e9

Contents lists avai

Journal of Human Evolution

journal homepage: www.elsevier .com/locate/ jhevol

The significance of cooking for early hominin scavenging

Alex R. Smith a, *, Rachel N. Carmody b, Rachel J. Dutton b, Richard W. Wrangham a

a Department of Human Evolutionary Biology, 11 Divinity Ave., Harvard University, Cambridge, MA 02138, USAb FAS Center for Systems Biology, 52 Oxford Street, Harvard University, Cambridge, MA 02138, USA

a r t i c l e i n f o

Article history:Received 21 August 2014Accepted 28 March 2015Available online xxx

Keywords:DietMeatMarrowBacteriaHominin ecologyChimpanzee

* Corresponding author.E-mail address: [email protected] (A.R. Smit

http://dx.doi.org/10.1016/j.jhevol.2015.03.0130047-2484/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Smith, A(2015), http://dx.doi.org/10.1016/j.jhevol.201

a b s t r a c t

Meat scavenged by early Homo could have contributed importantly to a higherequality diet. However, ithas been suggested that because carrion would normally have been contaminated by bacteria it wouldhave been dangerous and therefore eaten rarely prior to the advent of cooking. In this study, wequantified bacterial loads on two tissues apparently eaten by hominins, meat and bone marrow. Wetested the following three hypotheses: (1) the bacterial loads on exposed surfaces of raw meat increasewithin 24 h to potentially dangerous levels, (2) simple roasting of meat on hot coals kills most bacteria,and (3) fewer bacteria grow on marrow than on meat, making marrow a relatively safe food. Our resultssupported all three hypotheses. Our experimental data imply that early hominins would have found itdifficult to scavenge safely without focusing on marrow, employing strategies of carrion selection tominimize pathogen load, or cooking.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

The emergence of the genus Homo from australopithecines wascharacterized by the appearance of several distinctly human traitsincluding increased body mass, reduced masticatory structures,narrower rib cage, longer lower limbs, increased commitment tobipedalism, and increased absolute and relative brain size (Enget al., 2013; Ant�on et al., 2014). These traits became pronouncedwith the evolution ofHomo erectus around 1.9million years ago andare considered to reflect a concomitant increase in energy use anddietary quality (Wood, 1992; Aiello and Wheeler, 1995; Wranghamand Carmody, 2010).

Increased meateeating is widely thought to have been animportant contributor to increases in energy use and dietaryquality. Persistent carnivory has been confidently dated to 2.0million years ago (Ferraro et al., 2013). Occasional carnivory, bycontrast, is shown by stone tools and cutemarks dated to between2.6 and 2.5 million years ago (Domínguez-Rodrigo et al., 2010), orpossibly by cutemarks at 3.4 million years ago (McPherron et al.,2010). The original meateeating models featured hunting as thesource of meat, evidenced by animal bone assemblages exhibitingstoneetool cut marks (Dart, 1953, 1959; Washburn and Lancaster,1968; Bunn, 1981; Potts and Shipman, 1981). Later authors notedthe coeoccurrence on animal bones of marks made not only by

h).

.R., et al., The significance of5.03.013

stone tools but also by carnivores, and therefore suggested thatscavenging was a main strategy of obtaining meat (Binford, 1981;Brain, 1981; Bunn et al., 1986; Blumenschine, 1987; Cavallo andBlumenschine, 1989).

Attempts have therefore been made to use the archaeologicalrecord to determine hominin carcass acquisition strategies basedon data such as the frequency of carnivore tooth marks and thepresence or absence of skeletal elements that tend to be removedby carnivores. Some results suggest that early hominins were pri-marily engaged in hunting activities (Domínguez-Rodrigo et al.,2005, 2009; Pobiner et al., 2008), or that both hunting and scav-enging were important (Ferraro et al., 2013). Others point to adominant role for scavenging (O'Connell et al., 2002), but whileassemblages and marks on bones are often consistent with scav-enging (of both meat and marrow), none of the signals are certain(Lupo, 1998; Pante et al., 2012). The extent of scavenging by earlyHomo thus remains undecided.

Scavenging has received increased attention over the yearspartly because it has been observed among African foragers,especially in open habitats. O'Connell et al. (1988a,b) reported thatthe Hadza obtained 14% of their meat from scavenging, with 20% ofthe 54 carcasses that provided meat over a oneeyear period havingbeen scavenged. Yellen (1991) reported that 9% of 143 small an-telope eaten by !Kung San during the dry season were scavenged,and indicated an even stronger role for scavenging in the rainyseason. By contrast, the foresteliving Bofi and Aka of Central Africascavenge only “on rare occasions,” i.e., 0.3% of 650 mammals andbirds eaten over a foureyear period (Lupo and Schmitt, 2005: 337).

cooking for early hominin scavenging, Journal of Human Evolution

mailto:[email protected]

www.sciencedirect.com/science/journal/00472484

http://www.elsevier.com/locate/jhevol

http://dx.doi.org/10.1016/j.jhevol.2015.03.013



Figure 1. Schematic representation of butchery and sample preparation. Black linesindicate transverse and longitudinal cuts. The gray portions indicate the actual samplesof meat gleaned from the limb. Dotted lines indicate where the limbs were removedfrom the torso.

A.R. Smith et al. / Journal of Human Evolution xxx (2015) 1e92

There are fewer carcasses, and therefore fewer scavenging oppor-tunities, in humid forests than in open savannas (Watts, 2008).Meat scavenged by tropical hunteregatherers is apparently alwayscooked.

Although scavenging by openecountry hunteregatherers isinformative by showing that significant amounts of meat can beobtained, Ragir et al. (2000) suggested that scavenged meatwould be a costly food. When an animal is killed and its fleshexposed, bacteria accumulate on the meat (Janzen, 1977). Ragiret al. (2000) argued that bacteria are likely to be pathogenic andproduce toxins. These can indeed be costly. Even in the relativelyhygienic conditions of the United States, gastroenteritis (i.e.,foodepoisoning) annually affects 48 million people, hospitalizes128 thousand, and kills three thousand (Centers for DiseaseControl, 2014). Basing part of their argument on the fact thatchimpanzees normally fail to take advantage of scavenging op-portunities, Ragir et al. (2000) suggested that all hominoids lackthe gut morphology and digestive kinetics to cope with thesepathogens in an energetically feasible manner. They thereforeproposed that cooking would be required to minimize the risk ofingesting pathogenic bacteria on meat, and therefore that car-casses exploited before the advent of cooking must have beenprimarily hunted rather than scavenged.

The time when cooking was first practiced is not known. Onepossibility is that it was initiated by early Homo, as predicted frombiological evidence (Wrangham et al., 1999; Wrangham andCarmody, 2010). Cooking increases the effective energetic valueof meat (Carmody et al., 2011), and cooked meat is spontaneouslypreferred to raw meat by living hominoids (Wobber et al., 2008).Therefore, hominins who cooked scavenged meat could haveincreased their overall energy gain by increasing their access toanother highequality source of food. However, the extent towhich they would have benefited from the antiebacterial prop-erties of fire is unclear because to date there has been no studyquantifying the bacterial loads of meat and marrow in an envi-ronmentally relevant scavenging scenario. Nor has there been anyquantification of the effect of a simple cooking method that mighthave been available to ancestral Homo (openeflame roasting) onthese bacterial loads. It is therefore uncertain how predictablybacteria accumulate on raw wild meat, and how effectively Homocould have made scavenged meat safe with early methods ofcooking.

In this paper we describe two tests of Ragir et al.’s (2000) hy-potheses. First, we conducted experiments to assess how rapidlybacterial populations grow on freshly killed meat in the wild, andwhether openeflame roasting reduces those populations. Second,we compared the densities of bacteria on meat versus bonemarrow. Hadza hunteregatherers of northern Tanzania consumeraw marrow regularly, though how often it comes from scavengedor hunted carcasses has not been reported (Oliver, 1993). From amicrobiological perspective, marrow could be relatively safecompared with meat because the bone casing is expected to offerprotection from microbes, even though bacteria injected into thecirculatory system could in theory enter the bone through thenutrient artery (Trueta, 1959).

Accordingly, we designed an experiment in which meat anddefleshed, intact bones from wild animals were allowed todecompose in a controlled environment, after which their bacte-rial loads were compared with those of roasted counterparts. Wetested three hypotheses: (1) exposed meat surfaces accumulatehigher bacterial loads with increasing time after death, (2) cook-ing reduces the bacterial load on meat, and (3) for matchedsamples of meat and of bone marrow, the bone marrow has alower bacterial load due to the protection conferred by the bonecasing.

Please cite this article in press as: Smith, A.R., et al., The significance of(2015), http://dx.doi.org/10.1016/j.jhevol.2015.03.013

2. Materials and methods

Research was conducted from June 15 to August 5, 2013 inGraford, Texas (32�560 N, 98�140 W) at the Wagley Ranch, ownedand operated by Jay and SueWagley. Feral Eurasian boar (Sus scrofa)cause property damage on the ranch, and are therefore regularlyculled (1see note below). The Wagleys donated fresh carcasses. Wereceived exemption from the oversight of IACUC (the InstitutionalAnimal Care and Use Committee) in utilizing them.

2.1. Preparation of meat and marrow samples

After ranch workers shot and killed an adult boar, they imme-diately brought the whole carcass to our experimental station. Thetime to arrival was never more than 30 min after death. Butcheringbegan at once. The four limbs of each carcass were not touched bybullets and were the sources of meat in this experiment. Initialbutchering was performed with a large knife that was sterilizedwith 70% ethanol before and after the cutting of each individuallimb. To butcher the hind limbs, we cut across the exterior surfaceto the acetabulofemoral joint, freeing the femoral head from theacetabulum. To butcher the forelimbs, we cut across the gleno-humeral joint, freeing the humeral head from the glenoid fossa.Great care was exercised to ensure that the intestines and otherorgans were not punctured and that there was no contact betweenthe meat and the anus, feces, or other sources of contamination.

Using a new, sterile scalpel for each cut, three incisions weremade in each limb from skin to bone. One was a transverse cutacross the top of the limb, another was longitudinal down thefemur or humerus (through either the biceps femoris or triceps,respectively), and the third was the same as the first (transverse)except at the bottom of the longitudinal cut (Fig.1). The sample wasthen ‘butterflied’ so that the two halves of muscular tissue werepulled open from the longitudinal cut, exposing the adjacent sur-faces of muscle that were originally divided by the longitudinal cut.Marrow samples came from distal limb bones, which were dis-articulated and defleshed using a sterilized knife.

The limbs and bones were placed in individual, large tupper-ware containers (‘boxes,’ 72 cm L � 46 cm W � 16 cm H) fordecomposition, the aim being to protect the samples from scav-enging vertebrates or insects (a method adapted from Carter andTibbett, 2006; Spicka et al., 2011). There is no doubt that insectsplay an important role in the decomposition process. By trans-porting microbes and producing offspring that tunnel and aeratecarcass tissues, insects alter themicrobial and physical nature of thecarrion source in ways that are expected to promote bacterialgrowth (Payne, 1965; Putman,1978; DeVault et al., 2003). However,


A.R. Smith et al. / Journal of Human Evolution xxx (2015) 1e9 3

given that our aim was to standardize the process for each sample,we felt it was necessary to exclude insects from this experiment asmuch as possible. The containers had 35 1ecm wide evenlydistributed holes drilled in the walls and lid. Twenty holes weredrilled in the walls (four on each short side and six on each longside, all at equal height) and 15 holes were drilled on the lid. Thesecontainers were thoroughly cleaned before and after each use, andthen sterilized in 70% ethanol.

The limbs were placed inside the boxes on a layer of soil, 2 cmthick, which came from the same 10 m2 plot of land for all exper-iments. One limb was placed in each box so that the cut surfacefaced up. The boxes were then wrapped in mosquito nets, each netbeing washed and sterilized before and after each use. Boxes wereplaced on the ground 2.00 m apart with 0.75em tall plywoodboards positioned in between (and outside the end boxes) tominimizewindeborne bacterial transfer among samples. The boxeswere lined up in the NortheSouth plane so that each would receivesimilar radiant and ultraviolet exposure from sunrise to sunset(Fig. 2.).

The recorded decomposition time began as soon as the limbswere in the boxes. The 0 h samples came from the longissimus dorsi(‘backstrap’) muscles (a portion of the erector spinae). They werecut with a sterile knife in a fashion similar to the removal of thelimbs, with care taken to avoid contamination. The longissimus cutwas convenient because of its easy accessibility and similarity insize to the samples taken from the limbs.

Though Blumenschine (1987) noted that medium sized car-casses can persist in East African habitats for as long as two and ahalf days, it has been speculated that modern hunteregatherersmost often scavenge efour to 12 h after an animal has died (Ragiret al., 2000 citing O'Connell et al., 1988a,b). Therefore, thisresearch focused on growth throughout theefour to 12 h post-mortem interval, but we also tested samples at 0 h (analogous tofresh, ‘hunted’ meat) and up to 24 h. In the first six carcassessampled, one limb was allowed to decompose for four hours, onefor eight hours, and one for 12 h. In the second set of six carcasses, a24-h condition was added. Thus for the 0, four, eight, and 12 hconditions n ¼ 12, while for the 24 h condition n ¼ 6. In addition,

Figure 2. Four boxes, each containing one limb, aligned in the North-South (left-right) planeon either side of samples to prevent them from influencing each other. The nearby trees d


the second set of six carcasses was used to test bone marrow(n ¼ 6), which was allowed to decompose for 0, 12, and 24 h, witheach time series coming from a different bone. These were allprocessed similarly to the other caracasses except that additional‘defleshing’ cuts were utilized to slice the muscle tissue away fromthe bone. Humidity and temperature were recorded for each hourof decomposition.

A small fire was constructed 20 min prior to sampling time in alarge sandepit level with the ground. Fires were constructed uni-formly; eachwasmade of twigs and sticks (ranging in length from 5to 50 cm) that were arranged in a conical shape such that thediameter of the cone's base was never greater than 60 cm (a typicaldiameterof lowintensityfiresbuilt by theHadza;Mallol et al., 2007).

After approximately 20 min of burning, a flat bed of hot coalsremained. The butterflied pieces of meat (from the same limb) wereremoved from the skin with a sterilized knife. One of these twopieces was placed on the bed of hot coals. After 10 min the samplewas flipped onto the opposite side, at which point the temperatureof the coals was recorded. This piece cooked for 10 more minutesand was then removed from the coals and placed on a sterilizedsampling board. Bone samples were sawed in half transversely, andone half of each bone was cooked in the same way as the meat.Bone and meat samples were placed on opposite sides of the fire soas not to contaminate each other.

2.2. Bacterial sampling

Four types of 3M™ Petrifilm™ plates (targeting Escherichia coli,Enterobacteriaceae, Staphylococcus spp., and aerobic bacteria) wereused to test for the presence and quantity of bacteria on the surfaceof meat samples and within the bone cavity for marrow samples.The four tests have different levels of specificity, from species(E. coli) and genus (Staphylococcus) to family (Enterobacteriaceae)and functional grouping (aerobic). These assays were used to gaugethe potential presence of certain pathogens commonly associatedwith foodborne illness. Common pathogenic strains include: E. coliO157:H7, Salmonella enterica, Staphylococcus aureus, and Mycobac-terium tuberculosis. The 3M™ quickeswab was used to swab the

so that each would receive the same amount of sunlight. Plywood boards were placedid not cast uneven shadows on the samples.



surfaces of both raw and cooked meat. This sterile elongated swabcomes encased in 1ml of Letheen Broth, a type of liquid agar base. Asterile, flexible, metal circle 30.7 cm2 in area was placed on thesurface of the meat that had been exposed, and the quick-swabwasused to sample this area.

Two types of bone marrow swabs were used to test for an effectof external contamination on the bacterial load of bonemarrow. For‘internal marrow,’ a sterile quick-swab was used to remove the thinmarrow surface that had contact with the exterior environment(in the time that had elapsed since the bone was sawed), and wasimmediately discarded. Another quickeswab was then insertedinto the bone cavity to swab the interior marrow. For ‘externalmarrow,’ the swabbing included the external ring around the bonecanal and the marrow that had contact with the exterior environ-ment while being cut. This environmental contact should theoret-ically have been minimal because the bone saw was sterile.

The swabwas placed in Letheen Broth and shaken vigorously for10 s to remove the bacteria from the swab and create a 1 ml solu-tion. This solutionwasmixedwith 9ml of sterile peptonewater andinoculated onto a plate. In accordance with standard 3M™ Petri-film™ plate protocol, E. coli and aerobic plates were incubated for48 h and Enterobacteriaceae and Staphylococcus plates were incu-bated for 24 h before colony counting. All plates were incubated at36 �C.

At the end of the incubation period, the plates were examinedfor colony forming units (CFUs). Standard 3M™ count protocol wasfollowed, which entailed counting the individual CFUs on theplates. Each dot was recorded as a CFU of the specific bacteria testedgiven that it was the right color for that type of bacteria. AerobicCFUs were red, Staphylococcus CFUs were reddish purple and/orblack, E. coli CFUs were blue and/or reddish blue, and Enter-obacteriaceae CFUs were yellow, surrounded by gas bubbles, orboth.

2.3. Lipid content

Lower lipid concentrations in meat tend to favor bacterialgrowth (Montville andMatthews, 2008). To analyze lipid content inour experimental samples we took large portions of one of thehamstring muscles (biceps femoris) from five carcasses. Portionswere cut into cubes with volumes of approximately 15 cm3,immediately placed on ice and kept at a temperature of �20 �Cuntil they were freezeedried. We combined 0.5 g of each cube with50.0 ml of petroleum ether in a glass jar with an airetight glassstopper. Mixtures were gently swirled and placed under a fumehood for four days, duringwhich time theywere swirled once a day.The hot weights of empty beakers were recorded after heating in a60 �C oven for five hours. The mixture was filtered into thesebeakers, which were placed in the oven, and then hot weighed thenext day to calculate the weight of the lipid residue. We interpretthe mean value of the 10 samples as representative of the averagelipid content of limb muscle in the population of feral boar at theWagley Ranch.

2.4. Statistics

To determine whether bacterial growth was significantthroughout the decomposition period, we performed linear re-gressions testing for the effect of time on bacterial growth for agiven carcass. Next, we compared rates of growth for three con-trasts: 1) raw meat versus cooked meat, 2) external raw marrowversus internal raw marrow, and 3) raw meat versus external rawmarrow. Only one CFU (aerobic) was ever found on cooked bonemarrow, so it was left out of the comparison. Because of low samplesizes (n¼ 12 for zero, four, eight and 12 h meat; n¼ 6 for 24 h meat


and all marrow samples) and high variance in CFU number amongsamples of the same hour, we used noneparametric Wilcoxonsignerank tests to compare means. In accordance with the hy-potheses, which all involved directional relationships, all tests wereoneetailed. We utilized a Bonferroni correction to ensure thatperforming multiple comparisons did not erroneously inflate thechance of obtaining significance in any one comparison. Thus,comparisons required a pevalue of 0.016 (0.05/3) or lower to beconsidered significant.

Generalized linear models were used to test for the independenteffects of temperature and humidity on growth (i.e., the number ofCFUs), with the hour of decomposition being held constant. Wetested both the mean and maximum of temperature and humidityfor a given sample's decomposition period.

Fire construction was standardized and the temperature wasmeasured during cooking, but the temperature could not be pre-cisely controlled. We therefore tested for the effect of fire temper-ature by regressing temperature against the cooked meat counts ofall samples, while holding bacterial type constant.

Microsoft Excel version 12.2.8 (2007) was used to produce allgraphs and tables, and SPSS version 21 (2012) was used for allstatistical tests.

3. Results

On raw meat all bacterial populations grew rapidly within 24 h(Fig. 3). To test the sensitivity of growth rate (number of CFUs inrelation to elapsed time) to climatic variables we used generalizedlinear models to assess the independent effects of environmentaltemperature and humidity. Positive correlations with growth rateoccurred for mean or maximum temperature for Staphylococcus(mean: c2¼ 8.36, p < 0.01; maximum: c2¼ 8.13, p < 0.01), formeanhumidity for Enterobacteriaceae CFUs (c2 ¼ 3.83; p ¼ 0.05), and formaximum humidity for aerobic CFUs (c2 ¼ 12.59, p < 0.01). Thus,higher temperatures and higher humidity tended to favor increasedgrowth rates, though not sufficiently for the effects to be recordedin all conditions.

Cooking dramatically suppressed bacterial growth on meat:88.4% of the 216 cultured plates of cooked meat samples recordedno CFUs (Fig. 4), and differences between raw meat and cookedmeat were highly significant (Table 1). The strength and consis-tency of this effect is presented in Figure 3, which shows that for allfour bacterial types at every sampling time, cooked meat had veryfew CFUs compared with raw meat. Nevertheless, the mean bac-terial count on cooked meat was 9.9 CFUs rather than zero,including four cooked meat samples that exceeded 100 CFUs (all ofwhich were aerobic), and one outlier sample of aerobic bacteria at880 CFUs. Though the median and mode were both 0 CFUs forcooked meat, it was occasionally possible for bacteria to escape thedestructive effects of cooking. Alternatively, because most cookedsamples did not behave like this, these CFUs could indicate somekind of crossecontamination that occurred posteprocessing.

Therewas no evidence that variation in fire temperature affectedthe survival of bacteria. Thus, although measured fire temperaturesranged from 150� to 290 �C (mean 219 ± 43 �C, n ¼ 216), there wasno relationship between fire temperature and bacterial counts fromcooked meat (R2 ¼ 0.00, y ¼ �0.0041x þ 10.76, p ¼ 0.97) (Fig. 4).

For all bacterial types, at all sample times, the number of CFUsfrom samples of marrowwas lower than that from rawmeat (Fig. 3;Table 1). Nevertheless, Figure 3 shows that by 12 or 24 h, samplestaken from exposed marrow tended to support significant bacterialpopulations, equivalent to bacteria growing on meat for aroundeight hours.

The number of bacterial populations from samples taken frommarrow inside the bone, by contrast, was consistently low (Fig. 3).


Figure 3. Growth of CFUs of (a) Escherichia coli, (b) Staphylococcus, (c) Enterobacteriaceae, and (d) aerobic bacteria on raw meat, cooked meat, and (raw) external marrow. Note thatY-axis (number of colony forming units, CFUs) is logarithmic. X-values represent the amount of time for which a sample had decomposed: 0, 4, 8, 12 or 24 h. Each column representsthe mean CFU value of all samples tested (n ¼ 6 or 12). Standard deviation bars are shown for raw meat and marrow. Because ‘0’ cannot be graphed on a logarithmic axis, labeledbars are inserted to stand for values for which the mean ¼ 0. Values less than 1 are depicted by slightly taller bars, and are labeled. * indicates that n ¼ 6. For all other samples,n ¼ 12. ND indicates that no data were collected for those samples (i.e., marrow was not tested at four and eight hours).

0

200

400

600

800

1000

120 150 180 210 240 270 300

Col

ony

form

ing

units

Fire temperature (degrees Celsius)

Figure 4. Bacterial population sizes on meat or marrow sampled immediately afterhaving been roasted on an open fire. The number of colony forming units (CFUs) isshown without distinguishing bacterial types (E. coli, Staphylococcus, Enter-obacteriaceae, aerobic bacteria).


No CFUs of E. coli or Enterobacteriaceaewere ever found on internalmarrow, and very few for Staphylococcus (one sample at 12 h had490 CFUs; all others had 0 CFUs).

Lipid composition was assessed in the quadriceps muscles offive carcasses. The amount of lipid as a percentage of muscle dryweight ranged from 3.3 to 5.4% (mean ± standard deviation3.9 ± 0.9%).


4. Discussion

We found that bacterial populations increased rapidly on freshlykilled raw wild meat. However, roasting on coals, a particularlysimple form of cooking, was highly effective at reducing or entirelyeliminating bacteria. These results support Ragir et al.’s (2000)proposal that consumption of scavenged meat would have ledearly Homo to ingest significant populations of bacteria, and thatthe advent of cooking would have greatly reduced the exposure tomeateborne bacteria. Since bacteria growing on meat commonlyinclude pathogenic types, our data indicate that the risks of costlydisease or toxicity from eating meat would have been markedlyreduced by eating it freshly cooked.

We also considered the relative costs of meat and marrow. Wefound substantially lower bacterial loads on marrow than on meat.This indicates that early Homo would have sharply reduced theintake of potentially dangerous bacteria by selectively scavengingmarrow.

Specific features of our experiment potentially limit the signif-icance of these results. First, we did not test directly for patho-geneload or infectious dose. Although the 3M™ Petrifilm™ platesthat we used accurately quantify CFUs, their main function is toindicate the general sanitary condition of foods rather than identifyspecific risks. The high bacterial loads in our samples imply thatmost raw meat would be unsanitary. Additionally, several patho-genic strains of bacteria that fall within the groups we tested havehigh prevalence rates in feral swine. Jay et al. (2007) found that 15%


Table 1Comparisons of mean colony forming unit values using 1-tailed Wilcoxon sign-ranktests.

RMeCM RMeEBa EBeIBa

Escherichia coli0 Hours 0.020 0.090 1.0004 Hours 0.004* No test No test8 Hours 0.003* No test No test12 Hours 0.002* 0.014* 0.03324 Hoursa 0.014* 0.014* 0.034

Staphylococcus0 Hours 0.014* 0.034 0.5004 Hours 0.001* No test No test8 Hours 0.003* No test No test12 Hours 0.001* 0.038 0.03424 Hoursa 0.014* 0.014* 0.034

Enterobacteriaceae0 Hours 0.051 0.090 0.5004 Hours 0.006* No test No test8 Hours 0.004* No test No test12 Hours 0.002* 0.014* 0.09024 Hoursa 0.014* 0.014* 0.034

Aerobic0 Hours 0.002* 0.125 0.0904 Hours 0.001* No test No test8 Hours 0.001* No test No test12 Hours 0.001* 0.014* 0.014*24 Hoursa 0.013* 0.014* 0.014*

A Bonferroni correction was used to account for multiple comparisons, so a p-valueof 0.016 (0.05/3) or lower is considered significant. RM ¼ raw meat, CM ¼ cookedmeat, EB ¼ external bone marrow, IB ¼ internal bone marrow. The abbreviation onthe left side of the dash is always the one with the predicted higher mean number ofCFUs. Marrow was not tested at four and eight hours.*indicates significance at the 0.016 level.

a indicates n ¼ 6 samples. For all other tests, n ¼ 12 samples.


of wild boar in California tested positive for E. coli strain O157:H7.Decastelli et al. (1995) found 71e83% of Italian boar tested positivefor S. aureus, though Sarkis et al. (2003) found a prevalence of 11%.Chiari et al. (2013) found 25% of Italian boar tested positive forSalmonella spp., and Vengust et al. (2006) found this rate to be 47%in Slovenian boar. For aerobic bacteria, Naranjo et al. (2008) foundthat 43% of European boar tested positive for M. tuberculosis, whileRuizeFons et al. (2006) found that 30% of Spanish boar testedpositive for Brucella spp. A similar prevalence rate of 24%was foundfor Brucella spp. in Texas boar (Leiser et al., 2013). Together, theserates suggest that at the level of bacterial growth we observed inour samples, pathogenic bacteria were likely to be found in ourstudy species. Though diet can influence the gut microbiome(Muegge et al., 2011), we expect similar effects for themeat of otherwild mammals.

Given this probability of pathogenicity in our samples, some ofthe samples may have exhibited enough growth to be consideredan infective dose. According to the FDA, the infective dose of E. colistrain ETEC is 106e109 CFUs, while the infective dose for the morepathogenic strain O157:H7 is merely around 10e100 CFUs. Theinfective dose for S. aureus is 105 CFUs, anywhere from 10e106 CFUsfor M. tuberculosis, and as low as 1 CFU for some species ofSalmonella.

Second, bacterial growth is influenced by climate. At ourresearch site the mean temperature for the observation period was28.3 �C, the mean relative humidity was 51.5%, and annual rainfallwas 800 mm (Weather Underground, 2013). In northern Tanzaniawhere Hadza live, most scavenging occurs in the late dry seasonwhen the mean temperature is 25.8 �C, and the mean relative hu-midity is 34.4% (N. BlurtoneJones, Personal communication), withannual rainfall being 300e600 mm (Marlowe, 2010). Our researchtherefore occurred in a slightly warmer and wetter habitat than


where the Hadza live, and the range of temperature and humidityonly reflected a short timeeperiod. However, our climate variableswere within tropical ranges. We suggest that our results provide aninformative start for understanding the problems associated withscavenging in East African woodlandesavanna habitats.

Third, only two muscle tissues and longebone marrow weretested in this experiment. Other types of tissues may experiencedifferent patterns of bacterial growth. However, presumably bacteriaon all tissue substrates respond to cooking in the same manner.

Other factors suggest that our analysis likely underestimates thebacterial load faced by ancestral Homo or modern hunter-egatherers when scavenging. For instance, our experimental sys-tem constrained the sources of bacterial contamination bypreventing contact of edible parts with guts or feces, cutting withsterile knives, and stopping other animals, especially insects, fromsharing the meat or marrow. In swabbing the ‘internal’ marrow weused great caution to avoid contacting any outer surface. Theobserved growth of bacteria therefore represent lower levels thanexpected in nature, such that meat from a realistic scavengingopportunity would likely be substantially more dangerous thanfound in this study. The same is true for marrow. For examplemodern hunteregatherers such as Hadza will suck and gnaw onbones (Oliver, 1993). It is difficult to imagine an early homininextracting marrow with enough care that no contact is made withthe exterior surface of the bone, a surface that would normally becovered in dried blood, meat, cartilage, and other tissues.

In addition, our measured bacterial loads may have been limitedby the higher fat content of wild boar compared with many Africanungulates. The animals used in this study had limb fat levelsranging from 3.3% to 5.4%. While these were similar to USDApublished values of wild boar meat (3.3%), they are higher than thefat levels of many African ungulates, which in the case of theThomson's gazelle (Eudorcas thomsonii) in a poor season have beenmeasured as low as 0.01% (Smith, 1970). Furthermore, most Hadzascavenging takes place in the late dry season (Blumenschine, 1987;O'Connell et al., 1988a,b) when animals are likely to be nutritionallystressed and lacking in fat reserves, putting them in the lower rangeof limb fat levels (Speth and Spielmann, 1983; Speth, 1987). Thisdisparity in fat content between Texas boar and dry season Africanungulates could have implications for bacterial growth on meat.Microbes thrive with moisture, but fatty acids, because they arehydrophobic, tend to create a hostile environment for bacterialgrowth (Tamplin, 2002; Montville and Matthews, 2008). Leanermeat is therefore expected to be more conducive to microbialgrowth. Again, this suggests that we underestimated the exposureto bacteria that would occur for consumers of game meat in Africa.

Thus, although much work remains to be done to understandvariability in the growth of pathogenic bacteria and the factorsresponsible for such variation, we conclude that scavenged meatcan be expected to have high bacterial populations within a fewhours of death, and is predictably dangerous. By contrast marrowfrom a freshly broken bone is relatively safe; and cooking reducesbacterial populations to such low levels that pathogenicity isminimized, allowing humans to exploit scavenged meat relativelysafely.

4.1. Scavenging and bacterial growth in evolutionary context

Given our results, what significance do they have for the scav-enging hypothesis? First, they suggest that prior to the evolution ofcooking, marrowwould have been the safest animal source food forscavenging hominins. Paleoeecological evidence suggests thatmarrow could indeed be an important food. For instance theplacement of tooth, cut and percussion marks indicates that hom-inins were the primary consumers of marrow in at least one



Oldowan site (FLKe22, Olduvai, ~1.84 million years ago, Pante et al.,2012). Because few terrestrial carnivores have the jaw strength tocrack open large bones, marrow is both protected from exposure tothe environment and relatively abundant at carcasses. The onlycarnivores possessing jaws strong enough to crack such bones(mainly Crocuta) were probably infrequent occupants of riparianwoodlands due to territory competition with larger flesheeatingfelids (Blumenschine, 1987). Thus, marrow may have beenfrequently available. However, regular marrow consumption doesnot eliminate the possibility of hominins consuming scavengedmeat because cut marks may indicate large quantities of flesh wereremoved from carcasses (Pante et al., 2012). This means that theproblem remains of how early Homo managed the threat frombacteria on scavenged meat.

Second, our results imply that, as Ragir et al. (2000) argued,hominoids should have been very cautious about eating carrion.Ragir et al.’s predictionwas basedon evidence that bothhumans andchimpanzees are poorly adapted to ingesting potentially pathogenicbacteria, and on the inference that early hominins would have beensimilarly poorly adapted. In support of this idea, they reviewed evi-dence that susceptibility to pathogenic bacteria is enhanced by arelatively rapidmovement of food contents from the stomach to thesmall intestine. Emptying of the stomach occurs within a short timeinhumans (withinone to twohours) comparedwith carnivores (e.g.,more than four hours in dogs). Since gut kinetics are evolutionarilyconservative, this suggests that hominoids are predictably poorlyadapted to resisting the bacterial consequences of scavenging.

Events inside the stomach are similarly suggestive. Mostmammalian carnivores canmaintain a low stomach pH of 1e2 evenwhen food (which acts as a buffer) is present (Stevens and Hume,2004). The whiteback griffon vulture (Gyps africanus), almost apure scavenger, likewisemaintains a low stomach pH, which rangesfrom 1.0 to 1.5 while food is present (Houston and Cooper, 1975).Ragir et al. (2000) argued that humans and apes do not have thisability to maintain a low pH so consistently after ingesting a largemeal, so that the stomach environment becomes relatively morebasic. While their idea is plausible, it remains to be carefully tested.

The difficulties of understanding these dynamics are illustratedby the problem of scavenging by chimpanzees. Ragir et al. (2000)and Watts (2008) noted that chimpanzees tend to avoid carrion.While this is true, on rare occasions chimpanzees nevertheless havebeen seen to scavengemeat. Furthermore, meat is sometimes eatenmore than 12 h after the prey has been killed, including beingretained overnight in a sleeping nest before finally being consumedthe next day (Goodall, 1986; Boesch and Boesch-Achermann, 2000;Watts, 2008). Our data indicate that such periods are long enoughto allow substantial bacterial growth. This raises the question ofwhether chimpanzees thereby incur important risks or costs frombacteria. Further information on the tolerance of chimpanzees forconsuming bacteria is accordingly needed to test the hypothesisthat they cannot safely consume scavenged meat.

Our data showed that the rate of bacterial growth can vary evenwhen many sources of variation are controlled. This emphasizesthat it would pay a consumer of raw meat to be able to assess thequality of carrion. Janzen (1977) described this relationship be-tweenmicrobes and large organisms as antagonistic, withmicrobesevolving strategies to advertise their presence and larger organismsevolving strategies to recognize the advertisement. If microbesdeter larger scavengers by producing unpleasant compounds, wewould expect meateeating taxa to react to these compounds. Withold meat, eaters should exercise more caution (e.g., visual inspec-tion, smelling, etc.) and decide their consumption strategy accord-ingly. The Hadza, and to our knowledge all tropical hunters, areliberated from this danger by cooking all meat that they scavenge(O'Connell et al., 1988a,b; Wrangham, 2009; O'Connell, Personal


communication). Many modern humans eat pathogeneladen meatthat does not hold up to regulatory standards, but the risks fromconsumption are mitigated by cooking and/or mixing with spicesand acids thatmay have antiemicrobial effects (El-Alim et al.,1999).Chimpanzees, on the other hand, appear to utilize many inspectionsystems for carrion such as sniffing, poking, and staring (Mulleret al., 1995), often to the effect that they decide not to consumethe carcass. If chimpanzees are shown to be able to distinguish theedibility of a raw carcass according to its bacterial load, the impli-cation is that early Homo might have been able to do the same.

Ragir et al. (2000) posited that the digestive inability to processthe bacterial loads of raw meat is conserved among hominoids, butthat the mechanism for carrion avoidance is a foodeaversionresponse learned in an individual's lifetime. However, an innateaversion to the smell of diamines produced by carrion has beendemonstrated in some vertebrates and is suspected to be present inhumans (Hussain et al., 2013). Thus prior to cooking, scavengingcould have been facilitated not only by selecting primarily marrowbut also by careful assessment of meat quality, including cuttingaway and discarding bacteriaerich surfaces.

5. Conclusion

Several scenarios could explain how early Homo utilized carrionwithout suffering from pathogenic bacteria. (1) Early homininsmight have been less susceptible to pathogenic bacteria thanmodern humans. Following Ragir et al. (2000), this would be sur-prising. (2) Scavenged meat and marrow might have had suffi-ciently low populations of bacteria that the consumers would nothave suffered. Our data suggest that this would have been true onlyif they ate within a few hours of the animal being killed. (3) If earlyHomo cooked, they would have been able to scavenge both marrowand meat as a regular dietary strategy. (4) If early Homo did notcook, scavenging might not have contributed importantly to theirdiets (similar to chimpanzees). (5) Alternatively, if early Homo didnot cook, they may have exploited carcasses by a combination oftargeting marrow and being highly selective with regard to thebacterial quality of meat.

While we cannot discriminate these possibilities, we suggestthat bone marrow should have been a favored resource from car-casses eaten raw by hominins. Early Homo would have hadincreased access to the longebone remains of carcasses killed bypredatory carnivores (Blumenschine, 1987) and the relative resis-tance of marrow to bacterial growth means that this foragingbehavior would have been less energetically costly than consumingraw meat, which appears to lead to costly upregulation of the im-mune system (Carmody and Wrangham, 2009). Alternatively, earlyHomo may have had early access to freshlyekilled carcasses thatwere relatively safe (Mole�on et al., 2014).

Though the earliest dates of controlled fireeusemay extend intothe Lower Paleolithic (Gowlett and Wrangham, 2013), the role ofcooking has been largely ignored by those who have investigatedmeateeating by hominins (e.g., Bunn, 2007). The hunting versusscavenging debate is premised on the concept of a sharp distinctionbetween two behaviors that in reality combine to form a relativelyfluid and adaptable meateeating dietary strategy (DeVault et al.,2003). Our results indicate that, along with modern hunter-egatherers and chimpanzees, thefirstmembers of thehumangenuswouldhavehadaproblem incopingwithpathogens fromscavengedraw meat. How they solved this problem is an important question.

Acknowledgments

For funding we thank the W. W. Howells Fund of the Depart-ment of Human Evolutionary Biology, Harvard University.We thank



Zarin Machanda and Katherine Zink for advice with statistical an-alyses, John Speth for constructive discussion, David Pilbeam forencouragement, and Nancy Lou ConklineBrittain for helping withlipid analysis. Nicholas BlurtoneJones kindly provided climate datafrom Hadzaland, and James O'Connell commented valuably aboutcooking scavenged meat. Finally, we would like to thank theWagley family for their enormous generosity.

References

Aiello, L., Wheeler, P., 1995. The expensive-tissue hypothesis: the brain and thedigestive system in human and primate evolution. Curr. Anthropol 36, 199e221.

Ant�on, S.C., Potts, R., Aiello, L., 2014. Evolution of early Homo: an integrated bio-logical perspective. Science 345, 1236828.

Binford, L.R., 1981. Bones: Ancient Men and Modern Myths. Academic Press, NewYork.

Blumenschine, R.J., 1987. Characteristics of an early hominid scavenging niche. Curr.Anthropol 28, 383e407.

Boesch, C., Boesch-Achermann, H., 2000. The Chimpanzees of the Taï Forest:Behavioural Ecology and Evolution. Oxford University Press, Oxford.

Brain, C.K., 1981. The Hunters or the Hunted? An Introduction to African CaveTaphonomy. University of Chicago Press, Chicago.

Bunn, H.T., 1981. Archaeological evidence for meat-eating by Plio-Pleistocenehominids from Koobi Fora and Olduvai Gorge. Nature 291, 574e577.

Bunn, H.T., 1986. Patterns of skeletal representation and hominid subsistence ac-tivities at Olduvai Gorge, Tanzania, and Koobi Fora, Kenya. J. Hum. Evol. 15,673e690.

Bunn, H.T., 2007. Meat made us human. In: Ungar, P.S. (Ed.), Evolution of the HumanDiet: The Known, the Unknown, and the Unknowable. Oxford University Press,Oxford, pp. 191e211.

Carmody, R.N., Wrangham, R.W., 2009. The energetic significance of cooking.J. Hum. Evol. 57, 379e391.

Carmody, R.N., Weintraub, G.S., Wrangham, R.W., 2011. Energetic consequences ofthermal and nonthermal food processing. Proc. Natl. Acad. Sci. 108,19199e19203.

Carter, D.O., Tibbett, M., 2006. Microbial decomposition of skeletal muscle tissue(Ovis aries) in a sandy loam soil at different temperatures. Soil Biol. Biochem 38,1139e1145.

Cavallo, J.A., Blumenschine, R.J., 1989. Tree-stored leopard kills: expanding thehominid scavenging niche. J. Hum. Evol. 18, 393e399.

Center for Disease Control (CDC), 2014. Estimates of Foodborne Illness in the UnitedStates. http://www.cdc.gov/foodborneburden.

Chiari, M., Zanoni, M., Tagliabue, S., Lavazza, A., Alborali, L.G., 2013. Salmonellaserotypes in wild boars (Sus scrofa) hunted in northern Italy. Acta Vet. Scand.55, 42.

Dart, R.A., 1953. The predatory transition from ape to man. Int. Anthropol. Linguist.Rev. 1, 201e218.

Dart, R.A., 1959. Further light on australopithecine humeral and femoral weapons.Am. J. Phys. Anthropol 17, 87e93.

Decastelli, L., Giaccone, V., Mignone, W., 1995. Bacteriological examination of meatof wild boars shot down in Piedmont and Liguria, Italy. Ibex J. Mt. Ecol. 3,88e89.

DeVault, T.L., Rhodes Jr., O.E., Shivik, J.A., 2003. Scavenging by vertebrates: behav-ioral, ecological, and evolutionary perspectives on an important energy transferpathway in terrestrial ecosystems. Oikos 102, 225e234.

Domínguez-Rodrigo, M., Pickering, T.R., Semaw, S., Rogers, M.J., 2005. Cutmarkedbones from Pliocene archaeological sites at Gona, Afar, Ethiopia: implicationsfor the function of the world's oldest stone tools. J. Hum. Evol. 48, 109e121.

Domínguez-Rodrigo, M., Mabulla, A., Bunn, H.T., Barba, R., Diez-Martin, F.,Egeland, C.P., Espílez, E., Egeland, A., Yravedra, J., S�anchez, P., 2009. Unravelinghominin behavior at another anthropogenic site from Olduvai Gorge(Tanzania): new archaeological and taphonomic research at BK, Upper Bed II.J. Hum. Evol. 57, 260e283.

Domínguez-Rodrigo, M., Pickering, T.R., Bunn, H.T., 2010. Configurational approachto identifying the earliest hominin butchers. Proc. Natl. Acad. Sci. 107,20929e20934.

El-Alim, S.S.L.A., Lugasi, A., H�ov�ari, J., Dworsch�ak, E., 1999. Culinary herbs inhibitlipid oxidation in raw and cooked minced meat patties during storage. J. Sci.Food Agr 79, 277e285.

Eng, C.M., Lieberman, D.E., Zink, K.D., Peters, M.A., 2013. Bite force and occlusalstress production in hominin evolution. Am. J. Phys. Anthropol 151, 544e557.

Ferraro, J.V., Plummer, T.W., Pobiner, B.L., Oliver, J.S., Bishop, L.C., Braun, D.R.,Potts, R., 2013. Earliest archaeological evidence of persistent hominin carnivory.PloS One 8, e62174.

Goodall, J., 1986. The Chimpanzees of Gombe: Patterns of Behavior. Harvard Uni-versity Press, Cambridge.

Gowlett, J.A.J., Wrangham, R.W., 2013. Earliest fire in Africa: towards the conver-gence of archaeological evidence and the cooking hypothesis. Azania: Archae-ological Research in Africa 48 (1), 5e30.

Houston, D.C., Cooper, J.E., 1975. The digestive tract of the whiteback griffon vultureand its role in disease transmission among wild ungulates. J. Wildlife Dis 11,306e313.


Hussain, A., Saraiva, L.R., Ferrero, D.M., Ahuja, G., Krishna, V.S., Liberles, S.D.,Korschinga, S.I., 2013. High-affinity olfactory receptor for the death-associatedodor cadaverine. Proc. Natl. Acad. Sci. 110 (48), 19579e19584.

Janzen, D.H., 1977. Why fruits rot, seeds mold, and meat spoils. Am. Nat 111,691e713.

Jay, M.T., Cooley, M., Carychao, D., Wiscomb, G.W., Sweitzer, R.A., Crawford-Miksza, L., Farrar, J.A., Lau, D.K., O'Connell, J., Millington, A., 2007. Escherichiacoli O157: H7 in feral swine near spinach fields and cattle, central Californiacoast. Emerg. Infect. Dis. 13, 1908.

Lupo, K.D., 1998. Experimentally derived extraction rates for marrow: implicationsfor body part exploitation strategies of Plio-Pleistocene hominid scavengers.J. Archaeol. Sci. 25, 657e675.

Lupo, K.D., Schmitt, D.N., 2005. Small prey hunting technology and zooarchaeo-logical measures of taxonomic diversity and abundance: ethnoarchaeologicalevidence from Central African forest foragers. J. Anthropol. Archaeol 24,335e353.

Leiser, O.P., Corn, J.L., Schmit, B.S., Keim, P.S., Foster, J.T., 2013. Feral swine brucellosisin the United States and prospective genomic techniques for disease epidemi-ology. Vet. Microbiol. 166, 1e10.

Mallol, C., Marlowe, F.W., Wood, B.M., Porter, C.C., 2007. Earth, wind, and fire:ethnoarchaeological signals of Hadza fires. J. Archaeol. Sci. 34, 2035e2052.

Marlowe, F., 2010. The Hadza: Hunter-gatherers of Tanzania. University of CaliforniaPress, Oakland.

McPherron, S.P., Alemseged, Z., Marean, C.W., Wynn, J.G., Reed, D., Geraads, D.,Bobe, R., B�earat, H.A., 2010. Evidence for stone-tool-assisted consumption ofanimal tissues before 3.39 million years ago at Dikika, Ethiopia. Nature 466,857e860.

Mole�on, M., S�anchez-Zapata, J.A., Margalida, A., Carrete, M., Owen-Smith, N.,Don�azar, J.A., 2014. Humans and scavengers: the evolution of interactions andecosystem services. BioScience 64, 394e403.

Montville, T.J., Matthews, K.R., 2008. Food Microbiology: An Introduction. ASMPress, Washington, DC.

Muegge, B.D., Kuczynski, J., Knights, D., Clemente, J.C., Gonz�alez, A., Fontana, L.,Henrissat, B., Knight, R., Gordon, J.I., 2011. Diet drives convergence in gutmicrobiome functions across mammalian phylogeny and within humans.Science 332, 970e974.

Muller, M.N., Mpongo, E., Stanford, C.B., Boehm, C., 1995. A note on scavenging bywild chimpanzees. Folia Primatol 65, 43e47.

Naranjo, V., Gortazar, C., Vicente, J., de la Fuente, J., 2008. Evidence of the role ofEuropean wild boar as a reservoir of Mycobacterium tuberculosis complex. Vet.Microbiol. 127, 1e9.

O'Connell, J.F., Hawkes, K., Jones, N.B., 1988a. Hadza hunting, butchering, and bonetransport and their archaeological implications. J. Anthropol. Res. 44, 113e161.

O'Connell, J.F., Hawkes, K., Jones, N.B., 1988b. Hadza scavenging: implications forPlio/Pleistocene hominid subsistence. Curr. Anthropol 29, 356e363.

O'Connell, J.F., Hawkes, K., Lupo, K.D., Blurton Jones, N.G., 2002. Male strategies andPlio-Pleistocene archaeology. J. Hum. Evol. 43, 831e872.

Oliver, J.S., 1993. Carcass processing by the Hadza: bone breakage from butchery toconsumption. In: Hudson, J. (Ed.), From Bones to Behavior: Ethnoarchaeologicaland Experimental Contributions to the Interpretation of Faunal Remains.Occasional Paper 21. Southern Illinois University, Center for ArchaeologicalInvestigations, Carbondale, pp. 200e227.

Pante, M.C., Blumenschine, R.J., Capaldo, S.D., Scott, R.S., 2012. Validation of bonesurface modification models for inferring fossil hominin and carnivore feedinginteractions, with reapplication to FLK 22, Olduvai Gorge, Tanzania. J. Hum.Evol. 63, 395e407.

Payne, J.A., 1965. A summer carrion study of the baby pig Sus scrofa Linnaeus.Ecology 46, 592e602.

Pobiner, B.L., Rogers, M.J., Monahan, C.M., Harris, J.W., 2008. New evidence forhominin carcass processing strategies at 1.5 Ma, Koobi Fora, Kenya. J. Hum. Evol.55, 103e130.

Potts, R., Shipman, P., 1981. Cutmarks made by stone tools on bones from OlduvaiGorge, Tanzania. Nature 291, 577e580.

Putman, R.J., 1978. Patterns of carbon dioxide evolution from decaying carrion:decomposition of small mammal carrion in temperate systems, Part 1. Oikos 31,47e57.

Ragir, S., Rosenberg, M., Tierno, P., 2000. Gut morphology and the avoidance ofcarrion among chimpanzees, baboons, and early hominids. J. Anthropol. Res. 56,477e512.

Ruiz-Fons, F., Vicente, J., Vidal, D., H€ofle, U., Villanúa, D., Gauss, C., Segal�es, J.,Almería, S., Montoro, V., Gort�azar, C., 2006. Seroprevalence of six reproductivepathogens in European wild boar (Sus scrofa) from Spain: The effect on wildboar female reproductive performance. Theriogenology 65, 731e743.

Sarkis, F., Barancelli, G.V., Gallo, C.R., 2003. Avaliaç~ao das condiç~oes microbiol�ogicasde carnes de animais silvestres no município de S~ao Paulo. Hig. Aliment 17,60e67.

Smith, N.S., 1970. Appraisal of condition estimation methods for East African un-gulates. Afr. J. Ecol 8, 123e129.

Speth, J.D., 1987. Early hominid subsistence strategies in seasonal habitats.J. Archaeol. Sci. 14, 13e29.

Speth, J.D., Spielmann, K.A., 1983. Energy source, protein metabolism, and hunter-gatherer subsistence strategies. J. Anthropol. Archaeol 2, 1e31.

Spicka, A., Johnson, R., Bushing, J., Higley, L.G., Carter, D.O., 2011. Carcass mass caninfluence rate of decomposition and release of ninhydrin-reactive nitrogen intogravesoil. Forensic Sci. Int 209, 80e85.


http://refhub.elsevier.com/S0047-2484(15)00080-9/sref1








































http://www.cdc.gov/foodborneburden





















































































































































































Stevens, C.E., Hume, I.D., 2004. Comparative Physiology of the Vertebrate DigestiveSystem. Cambridge University Press, Cambridge.

Tamplin, M.L., 2002. Growth of Escherichia coli O157: H7 in raw ground beef storedat 10C and the influence of competitive bacterial flora, strain variation, and fatlevel. J. Food Protect 65, 1535e1540.

Trueta, J., 1959. The three types of acute haematogenous osteomyelitis. J. Bone Jt.Surg. Br. 41, 671e680.

Vengust, G., Valencak, Z., Bidovec, A., 2006. A serological survey of selected path-ogens in wild boar in Slovenia. J. Vet. Med. B 53, 24e27.

Washburn, S.L., Lancaster, C.S., 1968. The evolution of hunting. In: Lee, R.B.,DeVore, I. (Eds.), Man the Hunter. Aldine, Chicago, pp. 293e303.

Watts, D.P., 2008. Scavenging by chimpanzees at Ngogo and the relevance of chim-panzeescavenging toearlyhomininbehavioral ecology. J.Hum.Evol. 54,125e133.


Wobber, V., Hare, B., Wrangham, R., 2008. Great apes prefer cooked food. J. Hum.Evol. 55, 340e348.

Wood, B.A., 1992. Origin and evolution of the genus Homo. Nature 355, 783e790.Wrangham, R., 2009. Catching Fire: How Cooking Made Us Human. Basic Books,

New York.Wrangham, R., Carmody, R., 2010. Human adaptation to the control of fire. Evol.

Anthropol 19, 187e199.Wrangham, R.W., Jones, J.H., Laden, G., Pilbeam, D., Conklin-Brittain, N., 1999. The

raw and the stolen. Curr. Anthropol 40, 567e594.Weather Underground, 2013. Weather history for Graford, TX.Yellen, J.E., 1991. Small mammals: !Kung San utilization and the production of

faunal assemblages. J. Anthropol. Archaeol 10, 1e26.





































Documents

The significance of cooking for early hominin scavenging