First Dairying in Green Saharan Africa in the Fifth Millennium BC

LETTER doi:10.1038/nature11186

First dairying in green Saharan Africa in the fifth millennium BC

Julie Dunne1, Richard P. Evershed1,Me´lanie Salque1, Lucy Cramp1, Silvia Bruni2, Kathleen Ryan3, Stefano Biagetti4 & Savino di Lernia4,5

In the prehistoric green Sahara of Holocene North Africa—in con- art, long known for their rich and vivid portrayal of scenes from every- trast to the Neolithic of Europe and Eurasia—a reliance on cattle, day life4,15,16. The extensive rock art demonstrates that cattle played an sheep and goats emerged as a stable and widespread way of life, important part in the lives and ideology of ancient human groups living long before the first evidence for domesticated plants or settled in this region during the Holocene. This pictorial record contains village farming communities1–3. The remarkable rock art found countless scenes with representations of cattle, some emphasizing widely across the region depicts cattle herding among early the female’s full udders and, in a few cases, depictions of the actual Saharan pastoral groups, and includes rare scenes of milking; milking of a cow, such as at Wadi Teshuinat II15 in the Acacus or Wadi however, these images can rarely be reliably dated4. Although the Tiksatin in the Messak16. However, reliable dates for this rock art can faunal evidence provides further confirmation of the importance rarely be ascertained4. of cattle and other domesticates5, the scarcity of cattle bones makes Faunal remains from securely dated contexts indicate that domes- it impossible to ascertain herd structures via kill-off patterns, ticated animals (cattle, sheep or goats) were present in the area from thereby precluding interpretations of whether dairying was prac- the early sixth millennium BC, becoming much more common in the ticed. Because pottery production begins early in northern Africa6 fifth millennium BC. Unfortunately, these remains are highly fragmen- the potential exists to investigate diet and subsistence practices ted and poorly preserved, precluding herd reconstructions, and thus using molecular and isotopic analyses of absorbed food residues7. even indirect evidence of dairying is missing5. This approach has been successful in determining the chronology Direct evidence for the practice of dairying, beginning in the seventh of dairying beginning in the ‘Fertile Crescent’ of the Near East and millennium BC in northwestern Anatolia8, appearing in the sixth 8–11 its spread across Europe . Here we report the first unequivocal millennium BC in eastern Europe11 and reaching Britain in the fourth 13 13 chemical evidence, based on the d C and D C values of the major millennium BC9,10, has been established through the compound- alkanoic acids of milk fat, for the adoption of dairying practices by specific stable carbon isotope analysis of animal fat residues preserved prehistoric Saharan African people in the fifth millennium BC. in archaeological pottery. Notably, this research on the antiquity of Interpretations are supported by a new database of modern rumin- dairying practices has largely been confined to Europe, the Near East ant animal fats collected from Africa. These findings confirm the and Eurasia, with no attempt yet being made to identify the inception importance of ‘lifetime products’, such as milk, in early Saharan of dairying practices in the African continent. pastoralism, and provide an evolutionary context for the emer- Here we present direct chemical evidence for early dairying prac- gence of lactase persistence in Africa. tices within the central Sahara through the use of gas chromatography It is widely accepted that African pastoralism with cattle, sheep and (GC), gas chromatography–mass spectrometry (GC–MS) and gas goats emerged long before plant domestication2, in contrast to the chromatography–combustion–isotope ratio mass spectrometry process of ‘neolithization’ in the Near East, characterized by the trans- (GC–C–IRMS) analyses carried out on organic residues extracted ition from a mobile hunter-gatherer lifestyle to an increasingly settled, from archaeological pottery sampled from the Takarkori rock shelter agricultural way of life. In Saharan Africa, during the Early Holocene, located in the southwest Fezzan, Libyan Sahara, an area licensed to largely sedentary and pottery-producing hunters, fishers and gatherers Sapienza University of Rome (Supplementary Fig. 1). Four seasons of became nomadic cattle herders3, dynamically adapting to, and exploit- fieldwork identified evidence of Late Acacus (hunter-gatherer) occu- ing, different environments and resources. pation followed by Early, Middle and Late Pastoral remains (Sup- Today, it seems impossible that cattle could survive in such a hostile plementary Fig. 2), dating between approximately 8100 and 2600 BC environment as the arid desert land of the Sahara, but this region (Supplementary Table 1). This long Pastoral period, between approxi- enjoyed vastly more favourable climatic and environmental condi- mately 6000 and 2600 BC, denotes the adoption of cattle together with tions12 during the Holocene African Humid Period, which began sheep and goats, combined with intensive exploitation of wild cereals17,18. around 10,000 years ago13. Here, faunal evidence demonstrates that Analyses of absorbed organic residues focused on 81 potsherds by the early sixth millennium BC, cattle, sheep and goats were found (Supplementary Table 2), covering a wide range of decorative tech- together across the savannas of what is now the Sahara1,5. This suggests niques and motifs found on Saharan ceramics18,19 (Supplementary that the inception of dairying practices in North Africa and an early Fig. 3). These vessels were mainly excavated from securely dated and independent ‘secondary products’ economy14 seems plausible Middle Pastoral (n 5 56) levels (approximately 5200–3800 BC), with a given what we now know of the first appearance of milking in the small number originating from the Late Acacus (n 5 8) and Early Near East8. (n 5 14) and Late Pastoral (n 5 3) periods. The lipids were extracted Compelling evidence of prehistoric cattle herding in northern Africa using established protocols8–10. Many potsherds demonstrated extra- comes from the remarkable rock paintings and engravings of the ordinary preservation of lipids, containing concentrations of up to Sahara (Fig. 1), possibly the world’s largest concentration of prehistoric 6mgg21 (mean 1.2 mg g21), with one particular potsherd (TAK

1Organic Geochemistry Unit, School of Chemistry, University of Bristol, Cantock’s Close, Bristol BS8 1TS, UK. 2Dipartimento di Chimica Inorganica, Metallorganica e Analitica ‘‘Lamberto Malatesta’’, Universita` degli Studi di Milano - Via G. Venezian 21, 20133 Milano, Italy. 3African Section, University of Pennsylvania Museum of Archaeology and Anthropology, 3260 South Street, Philadelphia, Pennsylvania 19104-6324, USA. 4Dipartimento di Scienze dell’Antichita`, Sapienza, Universita` di Roma, Via Palestro, 63 - 00185 Roma, Italy. 5School of Geography, Archaeology & Environmental Sciences, University of the Witwatersrand, Johannesburg, Private Bag 3, Wits 2050, South Africa.

0 50 cm

Figure 1 | Rock art image and tracing from Teshuinat II rock shelter, South West Libya. a, b, Rock art image (a) and tracing (b) showing Saharan pastoralists with their pots and cattle (adapted with permission from ref. 15).

443) having a concentration of 17 mg g21. It is noteworthy that lipids differential routing of dietary carbon and fatty acids during the syn- were observed in every potsherd, in contrast to European archaeolo- thesis of adipose and dairy fats in ruminant animals, thus allowing gical sites, where generally ,40% of potsherds contain extractable ruminant milk fatty acids to be distinguished from carcass fats by 21 13 13 13 lipids with mean concentrations of approximately 0.1 mg g (refs calculating D C values (d C18:0–d C16:0) and plotting that against 13 10, 20). This remarkable preservation is likely to be related to the the d C value of the C16:0 alkanoic acid. Previous research has shown 13 extremely arid conditions prevailing in the region. that by plotting D C values, variations in C3 versus C4 plant consump- Lipid biomarker analyses by GC–MS showed that residues fall into tion are removed, thereby emphasizing biosynthetic and metabolic three broad categories (Fig. 2). The most common distribution characteristics of the fat source9,10. We have now confirmed this (Fig. 2a) was dominated by high abundances of the C16:0 and C18:0 through the GC–C–IRMS of a new reference collection of modern fatty acids, which derive from degraded animal fats. Also abundant ruminant animal fats from Africa collected to encompass the range 25 were branched-chain fatty acids, C13 to C18, components of bacterial of carbon isoscapes likely to have been encountered by early Saharan 21 13 origin diagnostic of ruminant animal fats . The second most common pastoralists. The d C values of the C16:0 and C18:0 components of these type of residue (Fig. 2c) contained a relatively low abundance of the modern fats, presented in Fig. 3, show d13C values of goat dairy fats C18:0 alkanoic acid, with several extracts showing high abundances of from the Acacus region, Libya (n 5 9), together with cattle dairy fats C12 and C14 homologues. Such distributions have rarely been seen in and cattle, sheep and goat adipose fats (n 5 9, 12, 7 and 12, respect- 22 13 European pottery and are more diagnostic of plant oils . Also present, ively) from Kenya. The d C16.0 values for the C16:0 alkanoic acids of and again rarely seen in European prehistoric ceramics, are a homo- the African reference fats plot in the range from 235 to 215%, indi- logous series of long-chain n-alkanes from C16–C33 (odd-over-even cating diets ranging from predominantly C3 to C4. These results con- 13 carbon number predominance), usually maximising at C25, regarded firm the global applicability of the D C proxy. as originating from epicuticular waxes of vascular plants23. A third, Of the 29 animal fat residues selected for GC–C–IRMS analyses, 22 intermediate category of residue (Fig. 2b) is characterized by a series of originate from Middle Pastoral levels, 3 from the Late Acacus, 2 from a,v-dicarboxylic acids (diacids), in the C5 to C18 carbon-chain-length the Early Pastoral and the remaining 2 from the Late Pastoral period 13 range (Fig. 2a, b) with C9 (azelaic acid) the most abundant homologue, (Table 1). The comparison of the D C values of the modern reference the latter commonly deriving from the ‘drying reaction’ of plant oils24. animal fats with those of the archaeological pottery residues from the Such residues also contained long-chain alkyl lipids of plant origin. Middle Pastoral period (approximately 5200–3800 BC) show that 50% Together such mixtures probably reflect either processing of both of these plot within, or on the edge of, the isotopic ranges for dairy fats, plant and animal products in the vessels or the multi-use of vessels. with a further 33% falling within the range for ruminant adipose fats Of the 81 potsherds, only those residues unambiguously assigned as and the remainder corresponding to non-ruminant carcass fats degraded animal fats (Table 1)—that is, those dominated by palmitic (Fig. 3). Notably, the residues originating from earlier periods do not (C16:0) and stearic (C18:0) alkanoic acids (for example, Fig. 2a)—were contain dairy fats, and plot in the non-ruminant fat range, probably selected for GC–C–IRMS analysis to determine the d13C values for deriving from wild fauna found locally. The unambiguous conclusion C16:0 and C18:0, with the aim of establishing their origins. Differences is that the appearance of dairy fats in pottery correlates with the 13 in the d C values of C16:0 and C18:0 alkanoic acids are due to the more abundant presence of cattle bones in the cave deposits, suggesting

a Figure 2 | Partial gas chromatograms displaying the trimethylsilylated lipid extract from potsherds excavated from Middle Pastoral levels in the C16:0 Takarkori rock shelter. a–c, The distributions are characteristic of degraded

C18:0 animal fat (a), a mixture of animal and plant fats (b) and plant material (c). Chromatographic peak identities denoted by filled triangles comprise straight-chain fatty acids in the carbon chain range C9:0 to C29:0, maximizing at C16:0; filled squares represent n-alkanes in the carbon-chain range C20:0 to C35:0; and filled circles indicate a,v-dicarboxylic acids (diacids) in the carbon-chain range C5:0 to C16:0. IS, internal standard, C34 n-tetratriacontane.

a full pastoral economy as the cattle were intensively exploited for their IS secondary products. b Of particular note is the wide range of d13C values exhibited by the alkanoic acids, plotting across the range 225% to 210%, which is broader even than the reference fats range (maximum 215%). This suggests that the animals giving rise to these fats had subsisted on an extensive range of different forages either composed completely of C3 plants, varying combinations of C3 and C4 plants, or a diet comprising 13 wholly C4 plants. The wide range of alkanoic acid d C values found for these African potsherds is unprecedented and points to differing pastoral modes of subsistence (such as vertical transhumance, which is still practised today) by these prehistoric Saharan groups. This is IS supported by their settlement pattern based on summer sites in the c 3

Relative intensity lowland sand seas and winter sites (such as Takarkori) in the mountains , which was probably in response to seasonal weather patterns. Our findings provide unequivocal evidence for extensive processing of dairy products in pottery vessels in the Libyan Sahara during the Middle Pastoral period (approximately 5200–3800 BC), confirming that milk played an important part in the diet of these prehistoric pastoral people. The findings are notable for three other reasons: (1) they confirm that domesticated cattle, used as part of a dairying eco- IS nomy, were present in North Africa during the fifth millennium BC, thus supporting the idea of an earlier ingression into the central Sahara1–3 and suggesting a local process of pastoral development, based 20 30 on the exploitation of secondary products; (2) the finding of dairy fat Retention time (min) residues in pottery is consistent with milk being processed, thereby

Table 1 | Subset of potsherds selected for isotopic analyses 13 13 13 Potsherd Laboratory Period Wall decoration technique Diameter at Part of vessel Lipid d C16.0 d C18.0 D C Classification number code mouth (cm) concentration (mgg21) 21 TAK21A Middle Pastoral Plain edge fishnet Not known Wall 1 base 5,830.6 214.7 220.5 25.8 Dairy 26 TAK1 Middle Pastoral APS return 20 Rim 1 wall 760.7 214.2 215.0 20.9 Ruminant adipose 45 TAK45 Middle Pastoral APS return Not known Wall 639.8 221.9 224.1 22.1 Ruminant adipose 120 TAK120 Middle Pastoral APS return Not known Wall 5,592.7 215.2 218.7 23.5 Dairy 124 TAK124 Middle Pastoral Ridged APS return Not known Wall 1,615.5 218.1 220.1 22.0 Ruminant adipose 197 TAK197 Middle Pastoral APS return Not known Wall 151.5 220.9 221.1 20.2 Non-ruminant adipose 420 TAK420 Middle Pastoral APS return, triangles Not known Wall 1,119.3 218.3 221.5 23.2 Dairy 443 TAK443 Middle Pastoral APS return Not known Wall 17,217.6 216.9 223.7 26.8 Dairy 576 TAK6 Middle Pastoral Ridged plain edge 30 Rim 1 wall 800.2 222.0 221.7 0.3 Non-ruminant adipose 748 TAK9 Middle Pastoral Ridged APS return 22 Rim 1 wall 5,650.5 213.7 219.0 25.2 Dairy 824 TAK11 Late Pastoral Continuous plain edge and Not known Wall 1 base 4,994.2 220.5 224.9 24.4 Dairy impressed dashes 873 TAK873 Middle Pastoral APS return Not known Wall 71.8 218.5 217.7 0.8 Non-ruminant adipose 896 TAK896 Middle Pastoral APS return Not known Wall 218.0 223.6 225.0 21.5 Ruminant adipose 987 TAK987 Middle Pastoral APS return Not known Rim 1 wall 4,442.6 213.6 219.3 25.7 Dairy 997 TAK15 Middle Pastoral APS return 16 Rim 1 wall 1,117.4 213.3 217.4 24.1 Dairy 1009 TAK1009 Middle Pastoral APS paired lines, banded Not known Wall 1,555.7 211.0 211.0 0.0 Non-ruminant adipose 1012 TAK1012 Middle Pastoral Irregular APS return Not known Wall 3,591.2 214.9 216.5 21.7 Ruminant adipose 1572 TAK1572 Middle Pastoral APS return Not known Wall 3,148.5 223.7 228.2 24.5 Dairy 1693 TAK21 Late Acacus Undecorated Not known Rim 1 wall 20.1 223.1 219.8 3.3 Non-ruminant adipose 1797 TAK24 Early Pastoral Combined plain edge with 30 Rim 1 wall 1,674.6 221.9 221.0 0.9 Non-ruminant adipose semilunar-motif impressed dots 1846 TAK25 Middle Pastoral Plain edge continuous Not known Wall 1 base 819.2 215.6 219.7 24.1 Dairy 1863 TAK26 Middle Pastoral Ridged APS return 16 Rim 1 wall 175.0 222.3 226.2 24.0 Dairy 1903 TAK27 Early Pastoral Cord impression 20 Rim 1 wall 308.5 222.8 221.7 1.1 Non-ruminant adipose 2028 TAK2028 Middle Pastoral APS, paired lines of dashes Not known Wall 1 base 1,931.0 224.5 228.9 24.4 Dairy 2251 TAK28 Middle Pastoral APS (row of impressed dots) 14 Rim 96.9 221.5 224.0 22.5 Ruminant adipose 2523 TAK29 Late Pastoral Undecorated Not known Rim 1 wall 445.6 218.5 219.7 21.2 Ruminant adipose 2588 TAK30 Late Acacus UNCL (impressed dots) 30 Rim 1 wall 823.3 213.9 213.8 0.1 Non-ruminant adipose 2817 TAK32 Late Acacus Rocker packed Not known Rim 6,882.8 219.3 217.5 1.8 Non-ruminant adipose 2857 TAK35 Middle Pastoral APS return continuous, triangles Not known Rim 1 wall 238.3 220.1 222.9 22.8 Ruminant adipose

APS, alternately pivoting stamp.

–30 –20 –10 –30 –20 –10 –30 –20 –10 a b c –10

C = –3.1‰ 13 C = –3.1‰ 13 Δ Δ C = –3.1‰ Δ13 δ –20 13 C 18:0 (‰)

Middle Pastoral adipose –30 Dairy Kenya adipose Late Acacus adipose Late Pastoral adipose Kenya dairy Early Pastoral adipose Dairy Libya dairy

5 d e f C3 diet Increasing C4 diet C3 diet Increasing C4 diet C3 diet Increasing C4 diet

1 Δ Non-ruminant adipose Non-ruminant adipose Non-ruminant adipose 13 C (‰) –1 adipose Ruminant adipose Ruminant adipose Ruminant –3

–5 Ruminant Ruminant Ruminant dairy dairy dairy –7 –30 –20 –10 –30 –20 –10 –30 –20 –10

δ13 δ13 δ13 C16:0 (‰) C16:0 (‰) C16:0 (‰) Figure 3 | Plots of d13C values and D13C values of alkanoic acids in modern fauna. e, The extensive processing of dairy products in pottery vessels from this reference ruminant fats and archaeological animal fat residues in region begins in the Middle Pastoral period approximately 5200–3800 BC.The 13 13 prehistoric Saharan pottery. a–c, Plots of the d C16:0 and d C18:0 values for broad array of values suggests that the animals giving rise to these ruminant fats archaeological animal fat residues in Late Acacus (hunter-gatherer) and Early subsisted on an extensive range of different diets either composed completely of Pastoral (Neolithic) pottery (a), archaeological animal fat residues in Middle C3 plants, varying amounts of C3 and C4 plants or, for some of the and Late Pastoral Neolithic pottery (b), and modern reference animal fats archaeological samples, a diet comprising wholly C4 plants. The ranges shown collected from Libya and Kenya (c). d–f, Plots denote D13C values for the here represent the mean 6 1 standard deviation of the D13C values for a global archaeological fat residues (Late Acacus/Early Pastoral) (d) and Middle/Late database comprising modern reference ruminant animal fats from Africa, the 10 28 29 Pastoral (e) and modern reference animal fats (f). Notably, the residues UK (animals raised on a pure C3 diet) , Kazakhstan , Switzerland and the originating from the Late Acacus and Early Pastoral periods (d) do not contain Near East30, published elsewhere. dairy fats, and plot in the non-ruminant range, probably deriving from wild providing an explanation of how, in spite of lactose intolerance, milk sampled and surfaces were cleaned with a modelling drill to remove any products could be consumed by these people with the practice being exogenous lipids. The potsherds were then ground to a powder, an internal adopted quickly; (3) they are consistent with the finding of the standard was added, and solvent was extracted by ultrasonication (chloroform/ 213910*T allele, associated with the lactase persistence trait in methanol 2:1 v/v, 2 3 10 ml). The solvent was evaporated under a gentle stream Europeans, across some Central African groups such as the Fulbe from of nitrogen to obtain the total lipid extract (TLE). Aliquots of the TLE were 26 trimethylsilylated (N,O-bis(trimethylsilyl)trifluoroacetamide 80 ml, 70 uC, northern Cameroon , supporting arguments for some movement of 60 min), and submitted to analysis by GC and GC–MS. Further aliquots of the people, together with their cattle, from the Near East into eastern TLE were treated with NaOH/H2O (9:1 w/v) in methanol (5% v/v, 70 uC, 1 h). Africa in the Early to Middle Holocene; and (4) they provide a context Following neutralization, lipids were extracted into chloroform and the excess for understanding the origins and spread of other, independently aris- solvent was evaporated under a gentle stream of nitrogen. Fatty acid methyl 27 ing LP-associated gene variants in sub-Saharan Africa . esters (FAMEs) were prepared by reaction with BF3-methanol (14% w/v, 70 uC, 1 h). The FAMEs were extracted with chloroform and the solvent removed METHODS SUMMARY under nitrogen. The FAMEs were re-dissolved into hexane for analysis by GC and A total of 81 potsherds were sampled from the Takarkori rock shelter, Tadrart GC–C–IRMS. Acacus Mountains, Libyan Sahara, of which 56 were excavated from the Middle FAMEs of freeze-dried reference fats (typically using 5 mg of TLEs) were pre- Pastoral period and the remainder originated from the Late Acacus (n 5 8), and pared exactly as above. Early (n 5 14) and Late Pastoral (n 5 3) periods (Supplementary Table 2). Of the 81 potsherds analysed, 29 were selected for GC–C–IRMS analysis; of these, 18 Received 16 March; accepted 8 May 2012. showed clear evidence of pure animal fat origin, with the remaining 11 comprising 1. Gifford-Gonzalez, D. & Hanotte, O. Domesticating animals in Africa: implications of lipid profiles suggestive of the mixing of animal and plant fats. genetic and archaeological findings. J. World Prehist. 24, 1–23 (2011). Lipid analysis and interpretations were performed using established protocols 2. Marshall, F. & Hildebrand, E. Cattle before crops: the beginnings of food described in detail in earlier publications8–10. Briefly, ,2 g of potsherd was production in Africa. J. World Prehist. 16, 99–143 (2002).

3. di Lernia, S. in Droughts, Food and Culture: Ecological Change and Food Security in 23. Tulloch, A. P. in Chemistry and Biochemistry of Natural Waxes (ed. Kolattukudy, P. E.) Africa’s Later Prehistory (ed. Hassan, F. A.) 225–250 (Kluwer Academic/Plenum, 235–287 (Elsevier, 1976). 2002). 24. Mills, J. S. & White, R. The Organic Chemistry of Museum Objects 2nd edn (Elsevier, 4. di Lernia, S. & Gallinaro, M. The date and context of Neolithic rock art in the Sahara: 1987). engravings and ceremonial monuments from Messak Settafet (south-west Libya). 25. West, J. B., Bowen, G. J., Dawson, T. E. & Tu, K. P. Isoscapes: Understanding Antiquity 84, 954–975 (2010). Movement, Pattern and Process on Earth through Isotope Mapping (Springer, 5. Gautier, A. in Droughts, Food and Culture: Ecological Change and Food Security in 2010). Africa’s Later Prehistory (ed. Hassan, F. A.) 195–208 (Kluwer Academic/Plenum, 26. Mulcare, C. A. et al. The T allele of a single-nucleotide polymorphism 13.9 kb 2002). upstream of the lactase gene (LCT)(C213.9kbT) does not predict or cause the 6. Huysecom, E. et al. The emergence of pottery in Africa during the tenth millennium lactase-persistence phenotype in Africans. Am. J. Hum. Genet. 74, 1102–1110 cal BC: new evidence from Ounjougou (Mali). Antiquity 83, 905–917 (2009). (2004). 7. Evershed, R. P. Organic residue analysis in archaeology: the archaeological 27. Itan, Y., Jones, B. L., Ingram, C. J., Swallow, D. M. & Thomas, M. G. A worldwide biomarker revolution. Archaeometry 50, 895–924 (2008). correlation of lactase persistence phenotype and genotypes. BMC Evol. Biol. 10, 36 8. Evershed, R. P. et al. Earliest date for milk use in the Near East and southeastern (2010). Europe linked to cattle herding. Nature 455, 528–531 (2008). 28. Outram, A. K. et al. The earliest horse harnessing and milking. Science 323, 9. Copley, M. S. et al. Direct chemical evidence for widespread dairying in prehistoric 1332–1335 (2009). Britain. Proc. Natl Acad. Sci. USA 100, 1524–1529 (2003). 29. Spangenberg, J. E., Jacomet, S. & Schibler, J. Chemical analyses of organic 10. Dudd, S. N. & Evershed, R. P. Direct demonstration of milk as an element of residues in archaeological pottery from Arbon Bleiche 3, Switzerland—evidence archaeological economies. Science 282, 1478–1481 (1998). 11. Craig, O. E. et al. Did the first farmers of central and eastern Europe produce dairy for dairying in the late Neolithic. J. Archaeol. Sci. 33, 1–13 (2006). foods? Antiquity 79, 882–894 (2005). 30. Gregg, M. W., Banning, E. B., Gibbs, K. & Slater, G. F. Subsistence practices and 12. Cremaschi, M., Zerboni, A., Spo¨tl, C. & Felletti, F. The calcareous tufa in the Tadrart pottery use in Neolithic Jordan: molecular and isotopic evidence. J. Archaeol. Sci. Acacus Mt. (SW Fezzan, Libya) An early Holocene palaeoclimate archive in the 36, 937–946 (2009). central Sahara. Palaeogeogr. Palaeoclimatol. Palaeoecol. 287, 81–94 (2010). Supplementary Information is linked to the online version of the paper at 13. deMenocal, P. et al. Abrupt onset and termination of the African Humid Period: www.nature.com/nature. rapid climate responses to gradual insolation forcing. Quat. Sci. Rev. 19, 347–361 (2000). Acknowledgements We thank the UK Natural Environment Research Council for a PhD 14. Sherratt, A. The secondary exploitation of animals in the Old World. World Archaeol. studentship to J. D. and the Life Science Mass Spectrometry Facility. Sapienza, 15, 90–104 (1983). University of Rome (Grandi Scavi di Ateneo) and The Minister of Foreign Affairs (DGSP) 15. Gallinaro, M. et al. in La Memoria dell’arte. Le pitture rupestri dell’Acacus tra passato e are thanked for funding for the Italian Archaeological Mission in the Sahara. Our Libyan futuro (eds di Lernia, S. & Zampetti, D.) 73–255 (All’Insegna del Giglio, 2008). colleagues in the Departments of Archaeology in Tripoli and Ghat, in particular S. Agab, 16. Lutz, R. & Lutz, G. The Secret of the Desert: the Rock Art of the Messak Settafet and are also thanked. The USA National Science Foundation and The Royal Society are Messak Mellet, Libya (Golf, 1995). further thanked for funding. 17. Biagetti, S. & di Lernia, S. in On Shelter’s Ledge: Histories, Theories and Methods of Rockshelter Research (eds Kornfeld, M., Vasil’ev, S. & Miotti, L.) 125–132 (British Author Contributions R.P.E. and S.d.L. conceived and planned the project. J.D., R.P.E. Archaeological Reports S1655, Archaeopress, 2007). and S.d.L. wrote the paper. J.D., M.S., L.C. and S.B. performed analytical work and data 18. Biagetti, S., Merighi, F. & di Lernia, S. Decoding an Early Holocene Saharan analysis, and S.d.L. and K.R. directed sampling of archaeological materials and stratified site. Ceramic dispersion and site formation processes in the Takarkori reference fats. All other authors either directed excavations or provided expertise in rock-shelter, Acacus Mountains, Libya. J. Afric. Archaeol. 2, 3–21 (2004). relation to pottery together with essential insights into the study region and sites. M.S. 19. Caneva, I. in Archaeology and Environment in the Libyan Sahara. The Excavations in and L.C. contributed to revising the article. All authors read and approved the final the Tadrart Acacus, 1978–1983 (ed. Barich, B.) 231–254 (British Archaeological manuscript. Reports S368, Archaeopress, 1987). 20. Charters, S. et al. Quantification and distribution of lipid in archaeological Author Information Reprints and permissions information is available at ceramics: implications for sampling potsherds for organic residue analysis. www.nature.com/reprints. The authors declare no competing financial interests. Archaeometry 35, 211–223 (1993). Readers are welcome to comment on the online version of this article at 21. Christie, W. W. Lipid Metabolism in Ruminant Animals (Pergamon, 1981). www.nature.com/nature. Correspondence and requests for materials should be 22. Copley, M. S. et al. Detection of palm fruit lipids in archaeological pottery from Qasr addressed to R.P.E. ([email protected]) or S.d.L. Ibrim, Egyptian Nubia. Proc. R. Soc. Lond. B 268, 593–597 (2001). ([email protected]).