Human mobility and the prehistoric spread of farming: isotope evidence from human skeletons

Human mobility and the prehistoric spread of farming: isotope evidence from human skeletons Alex Bentley For over a century, archaeologists, linguists and, more recently, geneticists have debated whether the earliest farmers in Europe and elsewhere were migrants to new regions, whether indige­ nous hunter-gatherers adopted farming, or whether both pro­ cesses combined as the two groups intermarried. Now analysis of isotopes in archaeologically recovered skeletons is providing new evidence about the mobility of some of the earliest farmers in central Europe and Southeast Asia.

Human mobility and the prehistoric spread of fa rming: isotope evidence from human skeletons Alex Bentley For over a century, archaeologists, linguists and, more recently, geneticists have debated whether the earliest fa rmers in Europe and elsewhere were migrants to new regions, whether indige nous hunter-gatherers adopted fa rming, or whether both pro cesses combined as the two groups intermarried.Now analysis of isotopes in archaeologically recovered skeletons is providing new evidence about the mobility of some of the earliest fa rmers in central Europe and Southeast Asia.
T he aim of this research project1 is to use isotopic analysis of human skeletons from archaeo logical sites to investigate the prehistoric spread of agricul ture -one of the most remarkable transfor mations in human history, which changed the world from one of dispersed, mobile hunter-gatherers to one of settled farmers.This complex process left a clear demo graphic legacy, in geographical patterns of languages and human genes that often reflect the directions in which agriculture spread.By studying evidence from the bones, plant remains, pottery and house structures that the earliest farmers left behind, archaeologists have been able to show that agriculture spread quite rapidly through Europe about 7000 years ago and more slowly through mainland Southeast Asia about 4000 years ago. 2 One of the most interesting questions about this transition is whether agriculture spread with migrating farmers or as a new way of life adopted by indigenous hunter gatherers .Unfortunately, current evidence derived from prehistoric artefacts, genes of present-day people, and the distribution of languages, cannot answer this question directly.The archaeological, genetic and linguistic evidence can tell us when and where agriculture spread, but not whether, in any given region, it was spread by migrating farmers or adopted by indige nous hunter-gatherers; nor can it reveal how much intermarriage may have taken place between them.I aim to test, in both regions, two hypoth eses about the transition to agriculture: that it involved intermarriage between indigenous and migrant populations, and that the pace of the transition was influ enced by the prevailing customs of marital residence.The results will advance our understanding of how and why agriculture spread, and provide valuable parameters that population geneticists can use to refine their continental-scale models of demographic history.3

Isotopic analysis of archaeological skeletons
The method involves measuring ratios of isotopes of various elements in human skeletal remains.Isotopes are atoms of the same chemical element that have different masses.The element strontium (Sr), for example, has several naturally occurring isotopes, including 86Sr, which represents about 10 per cent, and 87Sr about 7 per cent, of the strontium that occurs naturally on the Earth today.The heavier isotope, 87Sr, has one more neutron than the lighter iso tope, 86Sr, but because they both have the same number of protons and electrons, the two isotopes behave chemically in the same way.
In the project we measure isotope ratios of strontium (87Sr/86Sr) in archaeological tooth enamel, which enables us to identify the geographical area where a person obtained his or her food during childhood, while the enamel was forming.'The 87Sr/ 86Sr ratio in rock minerals depends on how much rubidium (Rb) was originally in the mineral and how old the rock is.This is because a particular isotope of rubidium, 87Rb, changes slowly into 87Sr through radioactive decay.The half-life ofthis pro cess is very long (48 billion years), and so the decay of 87Rb into 87Sr occurs signifi cantly only on a geological timescale, not on a human (or archaeological) timescale.Unlike lighter elements such as carbon and nitrogen, strontium isotopes are heavy enough to be conveyed, without measura bly fractionating (a term used to describe the changing proportions of one isotope to another), through eroded rocks, soils and the food chain into human tooth enamel.Because the 87Sr/86Sr ratio varies in differ ent types of rock, it serves as a geological, and hence geographical, signature within archaeological tooth enamel.'• 5 By comparing the human tooth-isotope values with the local 87Sr/86Sr range of values for each site, we can infer that indi viduals with values within the range were local and those with values outside the range were immigrants.However, it is possible that isotopically identified local individuals might have migrated between similar geological provinces, and non locals may once have been mobile foragers who obtained their food from extensive, ecologically diverse territories.6 We hope that a new method we are developing7 will allow us to differentiate between migrants between permanent agricultural settle ments and hunter-gatherers who moved The research described here approaches these questions in a new way, by using a novel but proven scientific method to elu cidate the geographical basis of this pre historic expansion of human settlement.The project is designed to acquire evidence of human mobility directly from the skel etons of prehistoric individuals involved in the transition from hunting and gather ing to agriculture.I have selected two geo graphical regions for comparison from the many archaeologically recorded prehis toric agricultural expansions, one in cen tral Europe and the other in Southeast Asia.By comparing evidence for human mobility during the transition to agricul ture in these two widely separated regions, Figure 1 Th e author loading purified strontium onto a tungsten fi lament for analysis by thermal-ionization mass spectrometry.seasonally to exploit wild foods -two alternate explanations for a non-local iso tope signature at a site.
In the analytical procedure that I follow, several milligrams of tooth enamel fr om an archaeological skeleton are cleaned and then dissolved in acid.At the Institute of Archaeology, I then extract purified stron tium from the dissolved enamel by using a special resin to capture the strontium.The purified Sr solution is then dried on a small tungsten filament (Fig. 1), similar to the fil ament of a light bulb, which is loaded into a thermal ionization mass spectrometer (TIMS, Fig. 2) in the Isotope Geochemistry Unit, directed by Rex Taylor at the South ampton Oceanography Centre.Inside the TIMS machine, which is kept very nearly at a vacuum, the filament is heated by several amperes of current.As the ions (atoms with a +1 charge) of strontium burn off from the filament, they are accelerated through a curved flight tube by means of an extremely high voltage (8000 volts) and directed around the curve by a strong mag netic fi eld.Being slightly heavier, the 87Sr travels in a slightly wider curve, enabling one detector to catch the 87Sr atoms while a separate detector, positioned at the end of a slightly tighter curve, catches the 86Sr atoms.What is measured is the 87Sr/86Sr ratio, which we use directly for our inter pretations.

Results from Neolithic Germany
Much of my recent research, often in col laboration with T. Douglas Price of the University of Wisconsin, has focused on the spread of the Linearbandkeramik (LBK) culture, which comprises the archaeolog ical remains of the first farmers of central Europe and the pottery they introduced some 7500 years ago.Their domesticated plants and livestock clearly derive from Neolithic cultures in southeastern Europe and western Southwest Asia (the Levant), where their large wooden longhouses and so-called shoe-last adzes appear to be novel innovations of the LBK, apparently as an adaptation to life in the deciduous for ests of central Europe.
The radiocarbon dates from the earliest LBK sites tell us that the culture spread rap idly from the Hungarian plain into central Europe between about 5700 and 5500 BC.
From about 5500 to 5375 BC there appears to have been a pause in the LBK expansion, when part of the western LBK boundary lay within the upper Rhine Valley in south western Germany (Fig. 3).Because archae ological dates for the earliest LBK sites in the valley overlap with the latest dates for Late Mesolithic sites farther west, in the period from c. 5500 to c. 5200 BC, it was here that we decided to look for evidence of possible interaction between LBK groups and indigenous hunter-gatherers to the west.
Geologically, the upper Rhine Valley in southern Germany is also ideal for stron tium-isotope analysis because the Vosges and Black Forest uplands, underlain by gneisses and granites, should have signif icantly higher 87Sr/86Sr values (> 0. 715) than the Jurassic and younger sedimentary rocks (<0.710) of the lowlands.Having previously determined that, for prehistoric Germany, archaeologically recovered pig teeth represent the local values well,8 I and my colleague Corina Knipper (University ofTiibingen) then mapped the biologically available strontium, carbon and oxygen isotopic signatures of prehistoric southern Germany by analyzing tooth enamel fr om the remains of pigs from archaeological sites distributed around the region.9The mapping confirmed our expectations of a marked upland/lowland difference in bio logically available 87Sr/86Sr values.Fur thermore, carbon isotopes in the carbonate fraction of pig enamel were generally slightly more enriched in 13C in the sam ples from the uplands.
The cumulative results from southern Germany, that we have obtained over the past four years are beginning to provide a very informative picture of the changes in human mobility patterns over time in the early Neolithic.They indicate that shortly after 5500 BC, when farming began in the region, there were many non-locals at each community, mostly with upland stron tium-isotope signatures (Fig. 4).A few cen turies later, around 5000 BC, there were fewer non-locals overall, and females are more common among the non-locals with upland signatures.
The burials of non-locals show patterns that are consistent within each site but dif ferent between sites.Our studies at two LBK cemeteries, Dillingen and Flomborn (Fig. 3), showed that very few of the non-locals, male or female, were buried with a shoe last adze, the characteristic early Neolithic artefact that was present with most of the local males.10However, at Stuttgart-Miil hausen (Fig. 3) we recently found that males buried with a shoe-last adze have the non-local signatures.Also, at several sites, most non-locals were buried so that the head was orientated towards a certain cardinal direction, with that direction peculiar to each site.At Flomborn and Stuttgart-Miilhausen, for example, most of the non-locals are buried with the head . ---•,_ \  It may be that some of the former were hunter-gatherers who joined agricultural communities, but as yet we cannot be sure of this.
To address this difficult problem, I insti gated last year a collaboration with Tim Atkinson, Director of the new Bloomsbury Environmental Isotope Facility at UCL, to analyze carbon and oxygen isotopes in addition to strontium isotopes.This has added a substantial amount of information from each skeleton, for only 15 per cent additional cost per sample analyzed.For example, at Talheim (see Fig. 3), the site of a village massacre about 4900 BC, a few females have 67Sr/66Sr values that are only slightly (although significantly) higher than those from the rest of the village pop ulation.But, by measuring 180/160 ratios in the same samples, we have already dis covered that a sample fr om one of these women differs greatly from the rest, and we can tell more about her origin by compar ing the sample with the regional isotope data obtained from archaeological pig teeth.Although the values for the rest of the community fall within the range for the Neckar Valley, where Talheim is located, this one female 160/160 ratio is higher than any of our samples from the region.One probable explanation is that she came from farther west, towards the Atlantic, an infer ence based on the fact that 160/160 ratios in precipitation are progressively higher in that direction (because the heavier oxygen tends to fa ll out first from the rain clouds as they move inland from the Atlantic).If we find similar oxygen-isotope outliers in samples from earlier Neolithic sites when hunter-gatherer territory lay to the west of the Rhine Valley, they may provide the best evidence for the presence of hunter gatherer immigrants within farming com munities.

A new comparative study in Southeast Asia
In order to undertake a second case study, independent of the first, I recent!y began to investigate the spread of agriculture into Southeast Asia by analyzing samples from two Neolithic sites in Thailand.Current evidence points to the swampy lakelands of the middle Yangtze River as the home of the first rice farmers about 10,000 years ago.Although somewhat more recent, this mirrors the process of wheat and barley domestication in Southwest Asia.In both regions, agricultural communities multi plied and spread.Just like the Neolithic sites of Europe with respect to the origins of their agricultural complex in the Levant, the earliest agricultural sites in Southeast Asia show a pattern of decreasing age away from a central China heartland, and for Southeast Asia we are asking essentially the same question: how much was the spread of rice farming the result of migra tion by farmers, and how much was it the result of indigenous hunter-gatherers adopting agriculture?Interestingly, in con trast to the rapid and almost complete spread of agriculture in prehistoric Europe, hunting and gathering persisted in parts of prehistoric Southeast Asia for many centuries after agriculture became established.By comparing the results from Thailand and Germany, we hope to gain a better understanding of the spread of agri culture in both regions.
With funding from a small grant from the British Academy, I accepted a once-in a-lifetime opportunity to undertake the first isotope analyses ever carried out of samples from the two most substantial and well documented Neolithic human skele tal collections from Thailand (Fig. 5): Khok Phanom Di (2000-1500 BC) and Ban Chi ang (2100 BC to 200 AD).!found that, at each of these important sites, the range of 67Sr/ 66Sr values for males remains the same through time, but the range of female values compresses dramatically to within the local range at a particular mortuary phaseY• 13 This may indicate a cultural shift to matrilocality 14 in Southeast Asia.If so, there are anthropological grounds for hypothesizing that matrilocality slowed the conversion to farming.This preliminary project is now in the process of being expanded, in collabora tion with other archaeologists and skeletal biologists, including Charles Higham and Nancy Tayles of the University of Otago, and Michael Pietrusewsky and Miriam Stark ofthe University ofHawai'i, with the aim of analyzing over 300 human skeletons from multiple sites in Thailand and Cam bodia that span the period 2100 BC to AD 400.As in Germany, I am first mapping the biological!y available isotopic signatures around the study region using archaeologi cal tooth enamel from local animals such as rats and domestic pigs.
A site of particular interest is Ban Non Wat (Fig. 5), from which Professor Higham and Dr Tayles will provide over 100 human skeletal samples for analysis.Here, a Neolithic settlement and cemetery, strat ified below Bronze and Iron Age layers, provide a rare opportunity to examine the evidence of early farmers and also the mar riage and migration practices of a prehis toric village community over the ensuing 2000 years.Excavations at Ban Non Wat have already unearthed the largest sample of human remains so far discovered in Southeast Asia.The analysis of strontium isotopic variation in the excavated skele tons will make it possible for key questions to be addressed directly.For example, were the first settlers of Ban Non Wat 4000 years ago born and raised elsewhere and did they come to settle a new environment, or, alternatively, were some of the popula tion local hunter-gatherers who intermar ried with intrusive agriculturalists?Was this an isolated village, or did marriage partners come from kindred groups else where?If so, was it a matrilocal group, revolving around the women, or did the women move residence on marriage?How did social organization change over time?These questions will be addressed as I ana lyze comparative material from the other Neolithic sites in central Thailand.

Implications
The research described in this article is beginning to provide the kind of direct (if still inconclusive) evidence, at the scale of the individual, that is crucial to an under standing of how the great demographic transition in prehistory from hunting and gathering to agriculture took place.Com parison of the results so far obtained from Germany and Thailand point to a contrast between the presence of non-local females in early agricultural sites in Neolithic Ger many, and of a shift to local females in comparable Neolithic sites in Thailand.Because the pace of the transition to agriculture was considerably slower in Thailand than in Germany, the interesting question arises of whether marital resi dence is somehow related to how easily the transition occurs.This question invites more detailed study and I hope our con tinuing research will help to resolve it.

Figure 2 A
Figure2A thermal-ionization mass spectrometer.The sample fi laments are loaded into the instrument through the door on the righ t; the ions are directed through th e fligh t tube through electric and magnetic fi elds, and are counted by detectors at the other end on the righ t.

Fi gure 3
Th e Rhine Valley and adjacent areas, showing the extent of early agricul tural (LBK) settlement c. 5300 BC and the location of Neolithic sites mentioned in the text: D = Dillingen, F = Flomborn, S = Schweitzingen, SM = Stuttgart Mi.ilhausen, V= Vaihingen .

Fig ure 5
Fig ure 5 Th ailand, showing the location of Neolithic sites mentioned in the text.