Haplotypic exploration of ancient and modern populations from Britain and Ireland
Recent studies have utilised patterns of haplotypic sharing among modern British and Irish individuals to great effect, in order to explore finescale population structure within the islands (Leslie et al. 2015), Byrne et al. submitted). A similar analysis is performed on the same datasets here, with the novel addition of imputed genotypes from ancient Irish, British and northern European samples. The resulting components of variation, reveal not only spatial, but temporal trends in haplotypic affinity from the Irish Chalcolithic period onwards (Fig. 4.6). These are used here to visualise the ancestral similarities and differences found between and within Ireland’s Early Bronze Age, Iron Age and modern populations.
Isolation at the British Peripheries
Several of the major principal components (PCs) identified explain little of the variation present in the ancient dataset. This includes the first and third PCs (Fig. 4.6A), which respectively segregate the populations of Orkney and Wales, while compressing the remainder of modern and ancient variation. The presence of strong haplotypic differentiation between these populations and the rest of Britain was reported in the original publication of the dataset (Leslie et al. 2015), and, in the case of Orkney, has been explained as the result of isolation and of Norse settlement. Notably, PC5 also solely describes Orcadian variation, and, alongside PC1, is not considered in further results and discussion.
The driving factor behind haplotypic divergence of Welsh populations is less clear. However, we note here that, alongside modern individuals from the border regions of Wales, the entire Irish Early Bronze Age population (several southwestern samples excluded) is also pulled away from the main cluster and in the direction of Welsh individuals on PC3 (Fig. 4.6A), suggesting they possess some haplotypic variation found in Wales that is absent in the remainder of the dataset. Surprisingly, the same increased affinity is not seen for the Iron Age Britons of Yorkshire and southeastern England, as may have been expected given both the persistence of Brittonic Language and culture in Wales after the Anglo-Saxon migrations, and the previously demonstrated affinity of the Yorkshire individuals to the modern Welsh population (Martiniano et al. 2016). This could indicate that the prolonged regional isolation of Wales, aided by its mountainous geography, stretches into the Bronze Age period, allowing the build up of the extensive haplotypic diversity seen in PC3.
The Establishment of Irish Haplotypic Diversity
The second component of variation (Fig. 4.6B-C) is unique in that it explains a large amount of variation present in both ancient and modern individuals. This corresponds to the primary split seen in fineSTRUCTURE analysis by Byrne et al. (submitted), which segregates Ireland and Britain into two distinct genetic islands, capturing what is defined as an Anglo-Celtic cline. Western Ireland and southeastern Britain form the two extremes of this component, with the Scottish population bridging the gap between the two clusters. Strikingly, ancient samples also separate out along this axis, with Irish individuals from both the Early Bronze Age and Iron Age periods falling further towards modern Irish variation than their British and continental counterparts (Fig. 4.6B-C).
Irish Iron Age samples extend the entire range of Irish variation on PC2, suggesting substantial continuity with the modern population. Irish Early Bronze Age samples show a more constricted distribution closer to the center of the plot, but still exhibit a systemic shift towards Irish Iron Age and modern populations, particularly those from individualised burials. The most parsimonious explanation for such observations is direct continuity between the Chalcolithic/Early Bronze Age and modern period in Ireland, with much of the haplotypic variation explained by PC2 forming in the intervening millennia, in a similar manner as suggested for Wales in PC3. While migration may be partially responsible for this structure, it is worth noting that the Irish Iron Age and modern population typically extends away, rather than towards, any potential external sources of variation in the dataset, including a contemporary Iron Age population from Britain, the most likely source of migration into Ireland between the Bronze Age and Early Christian periods. However, several exceptional Irish Iron Age samples exist, returned to in later sections.
The homogenisation of British population structure through admixture
In contrast to the gentle gradient of ancient Irish variation, British and continental individuals show a more punctuated distribution along PC2 (Fig. 4.6B-C), forming two clear clusters at both ends of modern British variation. Anglo-Saxons fall with southeastern English variation in this and all other PCs considered, alongside a Nordic Iron Age sample, reflecting the large genetic contribution of Germanic migrations to this part of the island (Leslie et al. 2015; Schiffels et al. 2016). Iron Age Britons comprise another tight grouping at the opposite end of British variation, emphasising the admixed nature of the modern population (Leslie et al. 2015; Martiniano et al. 2016; Schiffels et al. 2016). Early snapshots of continental introgression events may be represented by two samples that fall midway between the two groups, one from an Anglo-Saxon context (O3), which was reported as admixed in the original study (Schiffels et al. 2016), and the second from a Roman British population (6DT23), another member of which was demonstrated to be of likely Middle Eastern origin (Martiniano et al. 2016). Notably, no Irish Iron Age samples are seen to fall into this region of the PC space.
The compression of Iron Age British haplotypic variation close to the zero coordinate, relative to that of Ireland, suggests that PC2 may not effectively explain the majority of diversity present within this group, possibly due to their lack of representation within the larger admixed modern British cohort. In this respect, PC2 is perhaps best considered as explaining the distribution of Irish-related haplotypic variation in both modern and ancient individuals, which acts as somewhat of an imperfect proxy for Celtic ancestry in the neighbouring island of Britain, counterbalancing the Anglo-Saxon input. We caution that such a phenomenon may cause similar placement of individuals for unrelated demographic reasons. For example, the placement of Northern Irish and Scottish individuals between the two islands is proposed to be the result of numerous migrations in both directions, including the Gaelicisation of Scotland circa 600 AD and the later Ulster plantations (Byrne et al. submitted). It is notable that no PC segregates Scotland from the rest of the dataset, suggesting the modern population has been mainly borne from admixture, rather than isolation, the reverse of what is proposed for Wales. Indeed, the more muted and systematic shift towards Irish variation of Welsh populations, whose diversity is better captured in PC3 and PC6, may represent more ancient shared Celtic ancestry between the groups. The tight clustering of three German Late Neolithic and Bronze Age individuals at the edge of ancient Irish variation, alongside the Iron Age British population, could also be due to a similar effect of older shared ancestry. Such an interpretation may find some temporal grounding in the differential placement of a Nordic Late Neolithic individual further towards the Germanic extreme of the plot.
Haplotypic diversification after the Iron Age
PC6 shows the reverse trend to PC4, compressing ancient variation along the zero line, while allowing modern variation from all populations to fan out across the axis (Fig. 4.6B). This spread of modern haplotypic diversity shows something of a north-south trend, as identified in Byrne et al. (submitted). South Welsh and Cornish populations exhibit the largest amount of haplotypic variation and are followed on the axis by populations from southern Ireland, Devon and Border regions of Wales. Populations from the northern regions of Wales, England and Ireland, as well as Scottish groups, form the other extreme of PC6, with compression of the eastern populations of both islands apparent due to the homogenising effects of Anglo-related admixture (Byrne et al. submitted).
The clustering of ancient samples along the zero line suggests that, in Ireland at least, the majority of the geographical variation captured by PC6 postdates the Iron Age period. That said, a subtle shift towards northern groups is apparent in the Irish Iron Age, relative to both the preceding Bronze Age and British Iron Age. This is particularly apparent for individuals falling further towards Irish modern variation on PC2, and suggests some of the diversification captured by PC6 was already underway at this point in time. Two early modern Irish individuals from the Plantation period, postdating the Late Iron Age by roughly a millennia, are also plotted here and show extremely similar placement to modern individuals from the same regions, falling further ‘north’ and ‘south’ of the preceding axis of Iron Age samples. Further sampling of the British Iron Age and Medieval periods, specifically in Wales, will be required to interpret how such patterns are related to the clear divergence of northern and southern Celtic-speaking populations on the neighbouring island of Britain. Indeed, this novel preliminary analysis highlights the powerful temporal anchors ancient genomes can provide to spatial trends of regional genetic variation.
Bookmarks