Archaeogenetics – the study of ancient human DNA – has become an important contributor to cross-disciplinary endeavours to understand our past. It can provide illuminating insights into long-vanished human groups – particularly in time periods with no or only few written sources – complementing the information that we can gain from archaeological material and through other research methods like osteological assessments and isotope analysis.
Ancient DNA (aDNA) plays an especially important role, though, in questions of human mobility and migration. When people move, they bring with them their individual genetic profile, which gets preserved in aDNA extracted from their remains. Thereby, population movements undertaken hundreds or even thousands of years ago can be detected as changes in genetic profiles, seen in aDNA sampled through time. In the last decade, archaeogenetic research has shed new light on the spread of farming across Europe during the early Neolithic (c.6,000-4,000 BC), revealing that it was driven by multi-generational waves of people from the Near East, and not (only) by the transmission of ideas and cultural diffusion from that region. Similarly, and more recently, such analyses have demonstrated that the emergence of the Bell Beaker cultural package in Britain around 2500 BC was dominated by such large-scale movements of people from Central Europe that some 90% of the British gene pool was replaced within a few hundred years (see CA 338 and Olalde et al. in ‘Further reading’ at the end).
Besides the Beaker phenomenon, Britain in particular has undergone several other well-documented transitions – perhaps most prominently the shift from Common Brittonic and Latin to the Old English language, changes in burial tradition, and the rise of new political formations during the early and middle Anglo-Saxon periods in England. While genetics have been helping to make sense of these events for almost three decades, it was not until 2016 (and the work of Schiffels et al. and Martiniano et al. – see ‘Further reading’) that the first genome-wide aDNA evidence from this period demonstrated clear genetic differences between early and middle Anglo-Saxon individuals on the one hand, and Roman-period and Iron Age individuals on the other. This was a breakthrough in recognising population demographics, but unfortunately the small sample size used in these two studies – comprising only eight early medieval genomes – did not allow for a detailed understanding of the dynamics of mixing and immigration. Nor did their spatial coverage permit a more finely resolved comparison to the archaeological record.
Thanks to the recent research (initially published in Nature) that we are exploring in this special issue, though, that picture has changed dramatically. We have now increased the genomic coverage from the original eight samples to 278, and have increased the number of cemetery sites represented by these to 38, covering most of the south and east coasts of England. We have also filled some critical gaps in the archaeogenetic record for this period, in what today is Germany, the Netherlands, and Denmark, with 142 newly generated ancient genomes from these areas.
Mapping past populations
How does this analysis work? We can visualise the genetic profiles of groups, whether ancient or modern, through a technique called Principal Components Analysis. This is a statistical method to lay out genetic data on to a two-dimensional plane, much like a map, where distances match genetic distances between samples. On this genetic map, the genetic profiles of individuals from Britain’s past, spanning from the Bronze Age through to the Roman period, were found to overlap largely with those of many people in present-day Wales, Northern Ireland, and Scotland, at least those with local European descent (see below for details of comparisons to the present-day population). In other words, over that period of nearly 3,000 years, from 2500 BC up until the Roman period, the genetic profiles in Britain appear not to have changed that much. This view can be misleading, though – in fact, a recent study by Patterson et al. (see ‘Further reading’) indicated that, during the late Bronze and Iron Age, there was actually a substantial influx of newcomers from south-western Europe, probably from what today is France, which generated genetic changes in Britain over the course of many centuries – albeit very subtle ones, due to the relative genetic similarity of these incoming groups.
In stark contrast to this relative genetic stability, though, when we examine individuals from the early medieval period we find a surprisingly strong signal of population shift, and a large spread of genetic diversity encompassing the previous Iron Age groups all the way to contemporary groups from the continental North Sea zone. By differences in ‘genetic profiles’, we mean subtle differences in how common genetic variants segregate in different populations. Indeed, these differences are so subtle that they require hundreds of thousands of genetic markers (called ‘single nucleotide polymorphisms’, or SNPs), and thousands of present-day individuals, to become visible – see the box opposite for more on this process.
At this resolution, classification into clusters becomes statistically possible, and Principal Components Analysis – as well as the more quantitative modelling that we will describe below – are able to zoom in on the only 0.9% of genetic variation that indeed correlates with historical and geographic patterns. This striking shift of genetic profiles in early medieval Britain can only be explained by an unprecedented increase in mobility of groups migrating across the North Sea, at a larger scale than seen in the entire preceding 3,000 years (since the Bell Beaker transition).
In order to understand this shift more quantitatively, we created a model of mixture between two distinct groups – one named CNE (for Continental North European) and the other WBI (Western Britain and Ireland) – which correspond to the cluster overlapping with the North Sea countries on the continent, and the cluster of pre-medieval samples respectively. We then used this admixture model to tease apart the genetic profiles of every single sample in both the newly and previously published dataset. This process revealed that CNE ancestry was present in Britain before the early medieval period, but only sporadically – 1% during the Bronze/Iron Age, rising to 15% during the Romano-British period – followed by a dramatic spike up to 76% during the early and middle Anglo-Saxon periods.
Beyond this large change in average genetic ancestry, our research has also revealed detailed insights into how this incoming ancestry was distributed across different sites. Many cemeteries that we sampled reveal major amounts of CNE ancestry, sometimes involving many unadmixed direct migrants of CNE ancestry, sometimes with many mixed individuals of both ancestries. Within many sites we find individuals with zero CNE ancestry too, as well as many people with mixed ancestry.
By analysing these patterns, we can see that ancestry played a key and complex role in how people were buried. Duncan Sayer explores this in greater detail in his feature but, for example, taking all early Anglo-Saxon cemeteries we surveyed together, women with immigrant ancestry were more likely to be buried with grave goods than women with local ancestry. However, wherever both ancestries were present, people from both groups were buried together, suggesting a high degree of interaction despite any differing funerary customs. In some cases, reconstructed family trees have even allowed us to identify locals and newcomers producing mixed-ancestry children.
Looking at the strong change in genetic ancestry seen here, the obvious question is: where did it come from? Our additional samples – taken from areas in what is now northern Germany, the Netherlands, and extra sites in England – combined with already published samples from these locations, provide the means to study this in detail. To this end, we focused in particular on samples from England that contained exclusively continental European ancestry without admixture from the local British gene pool, and compared these to various groups from the Continent. We found that the greatest similarity is seen in a region spanning Friesland (present-day Netherlands), Niedersachsen and Schleswig-Holstein (present-day Germany), and modern Denmark up to the southern tip of Sweden. Strikingly, all of these areas had a remarkably homogeneous genetic profile during the period that we are studying, making all three probable source regions for the migration process into early medieval England.
How does this clear and strong shift in early medieval genetic profiles translate into present-day Britain? Britain today is inhabited by people with ancestries from all across the world, and for comparison with our ancient samples we focus here on a very specific subset, which does not represent present-day Britain but tries to approximate the historical population of local descent. This subset was defined and sampled in a study from 2015 (see Leslie et al. in ‘Further reading’), in which people from Britain were recruited as representing a specific region if all four grandparents were born within 80km of each other. When applying our modelling of WBI and CNE ancestry on this narrowly defined subset of the present-day populations of England, Wales, Scotland and Northern Ireland, we find that this strong shift towards continental northern Europe seen in the early medieval period did not in fact endure until today. Rather, a significant portion of present-day, European-descent England appears to fall between the two early medieval ancestry poles, and rather than being a perfect mix of these two main gene pools, it also includes a third component. This can be seen in the Principal Components Plot, where present-day England has shifted a bit to the lower left side of the early medieval cline.
Indeed, when modelling this, we find that samples from present-day English people, as defined in the subset above, have moved towards an ancestry pole best approximated by Iron Age samples from what is now France, and to a lesser degree also by samples from modern (European-descent) France itself. In fact, this ancestry makes up 15-40% of present-day English ancestry in England today, especially in the south-east, while being nearly absent in the west and north. At the same time, CNE ancestry decreased from the 75% observed during the early medieval period to 25-50% today, perhaps due to a resurgence of the previous local WBI ancestry and by the additional French-related ancestry. However, more-recent phases of high mobility across the English Channel like the Viking Period or the Norman conquest of England may have contributed additional ancestry from the north and south of Europe to England, further changing the proportions.
It appears, then, that we can see two main migration/mixture processes in England. One involves people with CNE ancestry arriving in England during the early and middle Anglo-Saxon periods; the other has people with more southern continental ancestry arriving in England at some point between the late Bronze/Iron Age and today. These two processes are markedly different in nature, though. While the first process can be clearly delineated in time to a few centuries after the end of Roman administration in Britain until the Viking period, the second arguably reflects a much more continuous long-term process which is evident in multiple periods – beginning in the late Bronze and Iron Age, observed in some individuals in the south of England in early medieval samples, and definitely increasing further until today.
While genetic analyses like this can precisely quantify the genetic impact of migration processes, though, they cannot tell us the human stories behind these shifts. What drove so many people to move to lands far from their home, and what sort of numbers were involved in these migrations? With regard to the more punctuated CNE influx during the early and middle Anglo-Saxon period, it is clear that this movement was on a large scale, but it is less easy to translate its impact into more precise numbers of people on both sides of this process. The strong shifts seen in our study are certainly in disagreement with previously proposed models of a limited, elite takeover – but if large numbers of people arrived, what did this mean for the local population, which must have already been relatively large and stable during the Roman period?
As John Hines discussed in the article preceding this one, traditional views of the Adventus Saxonum often involve violence as an explanation, rooted in the writings of Gildas and Bede. But could the motivation behind this movement have been more microbial than martial? The bacterium Yersinia pestis, the causative agent for plague, is known to have been present during this period. Plague is also believed to have contributed to a decline of Europe’s Neolithic population, and to have thereby aided the spread of Steppe-related ancestry and the Indo-European languages in Europe, something that mirrors our observation of genetic and cultural transformation in early medieval England.
We might be seeing echoes of internal migrations within Britain, too – a westward movement of the local population, and perhaps a later partial return could explain the observed decrease of CNE ancestry that we see after the Anglo-Saxon period. This is plausible, since linguistic and genetic research have found evidence of substantial movements of Brittonic speakers from Britain to Brittany. Present-day genetic data, at least, shows that Wales, and western and northern England, still harbour substantially higher levels of WBI ancestry and were probably less affected by early medieval migrations from the Continent.
In summary, these speculations point to two directions for future research. First, more sampling of aDNA in the centre and west of Britain will help us to understand the spread and decrease of the two types of incoming ancestry we find. Second, it would be illuminating to create detailed modelling of migration and admixture processes that also include sociological aspects like integration in families and communities, to fully understand the observed genetic patterns in light of human history.
Analysing ancient DNA
Human remains preserve DNA molecules that can be extracted and used to reconstruct an individual’s genetic profile, much like we can today genetically test living people. One of the main differences between DNA from living people and aDNA , though, is that in the latter case, typically only a few per cent or even less of the DNA is human, with the rest originating from bacteria and the environment. In recent years, the petrous bone (a bone of the inner-ear part of the skull) has proven to be one of the best sources for aDNA in humans, because it is among the densest tissues in the human body, which means that it mostly protects aDNA molecules from water and air. But human DNA has been extracted successfully also from hair, teeth, and even sediments and cave stones.
Once in the clean room, petrous bones are opened with a saw so that material can be extracted from the inside – using a dental drill in order to minimise contamination with modern DNA from the outside. The actual analysis is carried out with only 50mg of bone powder, which undergoes a series of molecular biological treatments, so that the aDNA fragments can be read from a modern sequencing platform.
Sequencing data consists of millions of short DNA fragments, which are mapped to the human reference genome. Specific positions in the human genome, so-called single nucleotide polymorphisms (SNPs), are variable in human populations, and around 1.2 million of those SNPs have typically been targeted for comparative analysis in population genetics studies involving aDNA – this is also the SNP set analysed in our study. Most of the variation seen between individuals along these 1.2 million SNPs is unsystematically distributed among human groups and simply reflects differences between individuals’ unique family histories. A few per cent of variation, however, reflects systematic differences and population structure along space, time, and cultural distinctions. This can be visualised using multivariate statistical approaches, such as Principal Components Analysis.
Further reading Rui Martiniano, Anwen Caffell, Malin Holst, Kurt Hunter-Mann, Janet Montgomery, Gundula Müldner, Russell L McLaughlin et al. (2016) ‘Genomic signals of migration and continuity in Britain before the Anglo-Saxons’, Nature Communications 7 (January): 10326. Iain Mathieson, Iosif Lazaridis, Nadin Rohland, Swapan Mallick, Nick Patterson, Songül Alpaslan Roodenberg, Eadaoin Harney et al. (2015) ‘Genome-wide patterns of selection in 230 Ancient Eurasians’, Nature (November), www.nature.com/doifinder/ 10.1038/nature16152. Iñigo Olalde, Selina Brace, Morten E Allentoft, Ian Armit, Kristian Kristiansen, Thomas Booth, Nadin Rohland et al. (2018) ‘The Beaker phenomenon and the genomic transformation of north-west Europe’, Nature 555 (7695): 190-196. Nick Patterson, Michael Isakov, Thomas Booth, Lindsey Büster, Claire-Elise Fischer, Iñigo Olalde, Harald Ringbauer et al. (2021) ‘Large-scale migration into Britain during the Middle to Late Bronze Age’, Nature (December): 1-14. Stephan Schiffels, Wolfgang Haak, Pirita Paajanen, Bastien Llamas, Elizabeth Popescu, Louise Loe, Rachel Clarke et al. (2016) ‘Iron Age and Anglo-Saxon genomes from east England reveal British migration history’, Nature Communications 7 (January): 10408. Stephen Leslie, Bruce Winney, Garrett Hellenthal, Dan Davison, Abdelhamid Boumertit, Tammy Day, Katarzyna Hutnik et al. (2015) ‘The fine-scale genetic structure of the British population’, Nature 519 (7543): 309-314.