<<

Commentary

From molecular to archaeogenetics

Colin Renfrew

McDonald Institute for Archaeological Research, Downing Street, Cambridge CB2 3ER, United Kingdom

pplying to ques- termediate between their Irish and Welsh perhaps disappointing that Orcadian sam- Ations of early population his- samples on the one hand (which they ples were not included in the analysis tory, and hence to major issues in prehis- assume to be representative of pre-Norse undertaken for microsatellites on the X toric , is becoming so fruitful Orkney also) and, on the other, the sample chromosome, where Basque and Norway an enterprise that a new discipline— from Norway (the Viking homeland). This are differentiated in the principal archaeogenetics—has recently come into is a very suggestive finding. Note, how- component analysis. being. That many of its applications have ever, that the term ‘‘Celtic’’ with which the The paper by Wilson et al. (11) raises so far related to prehistoric (1) is authors designate the Irish and Welsh other challenging issues that have yet to be due in part to the detailed archaeological samples is a linguistic one that could also resolved. In the first place, it poses explic- attention devoted to Europe by a series of be used for the (pre-Norse) Pictish pop- itly the question of the extent to which nineteenth and twentieth century scholars ulation of Orkney whose little-understood major cultural transitions, as documented (2). It is also due in part to the early language is currently assigned to the Celtic in the archaeological record, involved the application of a specific demographic language family (15). movement of people or simply of ideas. model, the ‘‘wave of advance’’ (3), to A further important development in And of course the authors have success- explain the chronological patterning that archaeogenetics reflected in this paper is fully shown that there was indeed signifi- emerged as farming spread across Europe the remarkable long-term continuity in cant gene flow accompanying the Norse at the onset of the period (4) and the use of sur- conquest of the to elucidate the structuring resulting from names as secure Orkneys. But the ef- an early principal components analysis of indicators of pa- fectiveness of such the classical genetic markers for Europe ternal lineage, as The archaeological record bears out analysis inevitably (5, 6). The application of DNA sequenc- has previously the picture conveyed in the Norse depends on the ex- ing, permitting female lineages to be in- been observed in istence of diagnostic vestigated through mtDNA (7) and male Ireland by Hill et sagas that Viking princelings criteria that would lineages through the Y chromosome (8), al. (16), where sig- from took distinguish the pop- has already brought a series of new ques- nificant genetic ulations of the re- tions into perspective, generating lively differences were control of Orkney. ceptor and donor debate (9, 10). The time is ripe, therefore, noted between areas, here the for more closely focused regional studies, Gaelic and non- Orkneys and Nor- devoted to specific historical problems. Gaelic surname samples. In one province way, at the time in question. They have The paper by Wilson et al. (11) in this issue (Connaught) the Gaelic surname samples indeed documented that for the relevant of PNAS breaks new ground in investigat- showed a frequency of 98% for haplo- male markers in those two areas, but the ing one such early demographic episode, group 1, relating to the Atlantic Modal matter remains open on the female side in the Viking conquest of the Orkney Islands Haplotype discussed in the Orkney study view of the current lack of distinctive (Fig. 1) in the ninth century A.D. It also reviewed here (11). A consideration of parameters (as between Orkney and Nor- raises a number of general problems that Orcadian surnames, excluding those asso- way) where the mtDNA data are con- emerge when reconstructing demographic ciated with Scottish settlers subsequent to cerned. When they make the observation history. the fourteenth century A.D., allows 38% that ‘‘patterns of Y chromosome variation The archaeological record in the of the (male) chromosomes to be identi- indicate that Neolithic and Age tran- Orkney Islands (12) bears out the picture fied as Scandinavian in origin. sitions in the British Isles occurred with- conveyed in the Norse sagas (13) that It is unfortunate, however, that the gen- out large-scale male movements,’’ one is Viking princelings from Scandinavia took eral underlying similarity in the mtDNA entitled to apply the same strict criteria. If, control of Orkney, establishing the dy- haplogroup distributions in European for example, the populations situated on nasty of the Norse earls. Because that populations (17) was reflected in an ap- both sides of the English Channel had record indicates considerable continuity parent lack of structure in the samples broadly similar Y chromosome haplotype from the preceding Pictish period as well analyzed, so that no evidence is available frequencies immediately before the neo- as Norse innovations, it has always been a to indicate whether an equivalent female lithic transition, it is perfectly possible in matter for surprise that the surviving population from Norway accompanied theory that a very substantial population place names of Orkney so comprehen- the male migrants inferred from the Y movement could have taken place across sively reflect the Norse language of the chromosome haplogroup frequencies. the Channel without significantly chang- Viking incomers, with hardly any surviv- There may, however, be more work to be ing the haplotype frequencies on either ing Pictish toponyms (14). These top- done here because the principal compo- the French or the English side. These may onyms do, however, survive in the High- nents analysis undertaken on the mtDNA be difficult matters to investigate, but it lands of the Scottish mainland data (figure 2 of ref. 11) shows Orkney immediately to the south. It is thus highly more than twice as distant from the interesting that Wilson et al. (11) find their Basque sample than is the Norwegian See companion article on page 5078. Orkney Y chromosome sample to be in- sample when mtDNA is considered. It is *E-mail: [email protected].

4830–4832 ͉ PNAS ͉ April 24, 2001 ͉ vol. 98 ͉ no. 9 www.pnas.org͞cgi͞doi͞10.1073͞pnas.091084198 Downloaded by guest on October 2, 2021 with the mitochondrial data and with the conclusion that the female-mediated gene flow inferred must have occurred ‘‘since the Upper Palaeolithic.’’ This conclusion rests on the implicit assumption that much of the variability now seen in mtDNA haplogroup distributions entered Europe since the , an assump- tion developed in the original ‘‘wave of advance’’ model (4, 5) but one contested in subsequent mtDNA studies (20). These remarks are not intended as a criticism of the paper by Hill et al. (16), nor of the original ‘‘wave of advance’’ model for demic diffusion, but rather to suggest the need for a ‘‘second generation’’ wave of advance model that will take into ac- count not only the cultural interactions between the incoming farmers (initially from to Greece) and the indig- enous population, but also the genetic and demographic consequences of the inter- marriages between the two groups. Al- ready the Y chromosome data produced in the important paper by Semino et al. (21) clearly show a decline from south- east to north-west Europe in the fre- quency of the supposed ‘‘neolithic’’ hap- logroup. If we imagine that, through assimilation and intermarriage, an actively Fig. 1 The Orkney Islands located north of Scotland. Archaeogenetic data suggest that Viking settle- farming community in a region to which ments left substantial genetic as well as cultural influence on this Scottish archipelago. [Reproduced with farming had recently spread contained, T permission from www.orknet.co.uk (Copyright 1997, Orknet).] centuries after the inception there of farming, a genetic input of X% (say 10%)

should be observed that the analysis here farming may have spread in Britain more from the indigenous popula- COMMENTARY has proceeded without the use of samples through cultural transmission than tion and retained (100-X)% (i.e., 90%) from mainland Britain: the samples con- through some form of gene flow.’’ Later in genetic input deriving from the adjacent sidered are from peripheral islands the paper they observe: ‘‘This is in sharp source area from which the most recent (Orkney, Ireland, and Anglesey). Al- contrast with the mtDNA pattern [in the stage of the spread occurred, we have the though the question that they pose is an principal components analysis] in which basis for a model, the Staged Population- entirely valid one, there would clearly be the [Irish and Welsh] populations are Interaction Wave of Advance (SPIWA). need of a more ambitious sampling strat- closer to the centre of the plot, indicating Such assumptions could yield an exponen- egy to begin to formulate a definitive that they have undergone more female- tial decline across Europe (along the di- answer. The matter is underlined, so far as mediated gene flow from other European rection of spread) in the frequency of the the neolithic is concerned, by the circum- populations than the Basques have. Thus ‘‘incomer-farmer’’ genes as against the stance that the Orkney Islands may well at least one of the cultural transitions in indigenous mesolithic genes (which are have lacked any permanent population the British Isles since the Upper Palaeo- assumed as a first approximation to be until the arrival of neolithic settlers (18). lithic must have involved a demic compo- homogeneous). The SPIWA model ad- This, like the inception of the neolithic nent on the female side.’’ I suspect, how- dresses the same general problem as the period in (19), is one instance where ever, that these arguments rest on two ‘‘wave of advance’’ demic diffusion model the movement of females as well as males uncertain premises that illustrate the gen- of Ammerman and Cavalli-Sforza (3) but seems an indispensable assumption! eral difficulties in interpretation of all brings into direct consideration the inter- There is another important finding in archaeogenetic data. actions between the indigenous mesolithic the paper by Wilson et al. that is both The first problem is the inference that, population and the incoming farmers interesting in itself and leads them to an if the Irish, Welsh, and Basque Y chro- whose demographic progress was consid- argument whose inferential foundations mosome haplotype frequencies are closely ered in the original model. The ‘‘neoli- may be questioned. They rightly empha- similar today and may have been so in thization’’ process is here viewed as a size the strong similarity in the Y chro- Upper Palaeolithic times, then no signif- series of successive steps or stages, in each mosome haplogroup frequencies between icant gene flow into Ireland and Wales in of which the incoming farming population the Basque country, and the Welsh and the male line occurred at the onset of the interacts (culturally and genetically) with Irish samples. The three in consequence neolithic. As noted above, very significant the local mesolithic population. Popula- cluster closely on the principal compo- gene flow could have occurred at that time tion growth takes place with the inception nents diagram for the Y chromosome without notable impact on haplotype fre- of farming as in the original model, but the data. This observation leads the authors to quencies if the donor and receptor popu- fall-off in gene flow (and the clinal reduc- the following conclusion: ‘‘in the British lations were themselves not distinguish- tion) is exponential rather than linear, a Isles the Neolithic transition did not entail able in that respect. Such may well have pattern more in keeping with recent Y a major demographic shift. Accordingly, been the case. The second problem lies chromosome work (21).

Renfrew PNAS ͉ April 24, 2001 ͉ vol. 98 ͉ no. 9 ͉ 4831 Downloaded by guest on October 2, 2021 Moreover, it should be observed that, if the similarities or equivalence between task. The paper by Wilson et al., with its the indigenous gene frequencies hap- donor and receptor haplotype frequencies well-defined regional focus, certainly pened at the outset to be the same as those would make the process invisible to gene draws attention to many of the right of the incoming neolithic farmers, the frequency analysis. questions. With the increasing availabil- entire ‘‘wave of advance’’ could take place, It is not, of course, suggested here that ity of data for both male (Y chromo- precisely as in the original model, without such crude models could approximate to some) and female (mtDNA) variability any impact on the haplotype frequencies the complex reality of prehistoric Eu- in Europe (21, 20), the way will increas- at all. Very significant gene flow could , but simply that we need to con- ingly be open for useful regional studies indeed occur, as in the original model, but struct further models appropriate to that of this kind.

1. Renfrew, C. (2000) in Archaeogenetics: DNA and 8. Underhill, P. A., Shen, P., Lin, A. A., Jin, L., 16. Hill, E. W., Jobling, M. A. & Bradley, D.G (2000), the Population of Europe, eds. Renfrew, Passarino, G., Yang, W. H., Kauffman, E., Bonne- in Archaeogenetics: DNA and the Population Pre- C. & Boyle, K. (McDonald Institute, Cambridge, Tamir, B., Bertranpetit, J., Francalacci, P., et al. , eds. Renfrew, C. & Boyle, K. U.K.), pp. 3–12. (2000) Nat. Genet. 26, 358–361. (McDonald Institute, Cambridge, U.K.), pp. 203– 2. Trigger, B. G. (1989) A History of Archaeological 9. Barbujani, G., Bertorelle, G. & Chikhi, L. (1998) 208. Thought (Cambridge Univ. Press, Cambridge, Am. J. Hum. Genet. 62, 488–491. 17. Simoni, L., Calafell, F., Bertranpetit, J. & Barbu- U.K.). 10. Richards, M. & Sykes B. (1988) Am. J. Hum. jani, G. (2000), in Archaeogenetics: DNA and the 3. Ammerman, A. J. & Cavalli-Sforza, L. L. (1973) in Genet. 62, 491–492. Population Prehistory of Europe. eds. Renfrew, C. The Explanation of Culture Change: Models in 11. Wilson, J. F., Weiss, D. A., Richards, M., Thomas, & Boyle, K. (McDonald Institute, Cambridge, Prehistory , ed. Renfrew, C. (Duckworth, London), M. G., Bradman, N. & Goldstein, D.B. (2001) U.K.), pp. 131–138. pp. 343–58. Proc. Natl. Acad. Sci. USA 98, 5078–5083. (First 18. Saville, A. (2000) in Neolithic Orkney in its Euro- 4. Ammerman, A. J. & Cavalli-Sforza, L. L. (1984) The Published April 3, 2001; 10.1073͞pnas.071036898) pean Context, ed. Ritchie, A. (McDonald Institute, Neolithic Transition and the Genetics of Population in 12. Morris, C. (1985) in The Prehistory of Orkney, ed. Cambridge, U.K.), pp. 91–100. Europe (Princeton Univ. Press, Princeton). C. Renfrew (Edinburgh Univ. Press, Edinburgh), Cretan Stud. 5, 5. Menozzi, P., Piazza A. & Cavalli-Sforza, L. L. pp. 210–243. 19. Renfrew, C. (1996). 1–27. (1978) Science 201, 786–792. 13. Palsson, H. & Edwards P. (1981) Orkneyinga Saga: 20. Richards, M., Macaulay, V., Hickey, E., Vega, E., 6. Cavalli-Sforza, L. L., Menozzi, P. & Piazza, A. The History of the Earls of Orkney (Penguin Books, Sykes, B., Guida, V., Rengo, C., Sellitto, D., (1994) The History and Geography of Human Harmondsworth, U.K.). Cruciani, F., Kivisild, T., et al. (2000) Am. J. Hum. Genes (Princeton Univ. Press, Princeton). 14. Marwick H. (1929) The Orkney Norn (Oxford Genet. 67, 1251–1276. 7. Richards, M., Corte-Real, H., Forster, P., Ma- Univ. Press, Oxford). 21. Semino, O., Passarino, G., Oefner, P. J., Lin, A. A., caulay, V., Wilkinson-Herbots, H., Demaine, A., 15. Forsyth, K. S. (1997) Language in Pictland: The Arbuzova, S., Beckman, L. E., De Benedictis, G., Papiha, S., Hedges, R., Bandelt, H. J. & Sykes, B. Case against ‘‘Non-Indo-European Pictish’’ (De Francalacci, P., Kouvatsi, A., Limborska, S., et al. (1996) Am. J. Hum. Genet. 59, 185–203. Keltiche Draak, Utrecht, The Netherlands). (2000) Science 290, 1155–1159.

4832 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.091084198 Renfrew Downloaded by guest on October 2, 2021