Academia.eduAcademia.edu

Figure 1 – uploaded by Alison Sheridan

See full PDFdownloadDownload figure
Note: We pool all individuals from each period and region removing those failing goAdm modeling at p<0.01 according to Online Table 5). In the left columns are gpAdm estimates of ancestry for each group. Below diagonal are Z-scores from f4(Row population, Column population; Turkey_N, OldSteppe) (highlighted in red if |Z|>3). Above diagonal are inbreeding-corrected Fsr values (highlighted in yellow if Fsr>0.005). Extended Data Table 1: Ancestry change over time in Britain

Table 5 Note: We pool all individuals from each period and region removing those failing goAdm modeling at p<0.01 according to Online Table 5). In the left columns are gpAdm estimates of ancestry for each group. Below diagonal are Z-scores from f4(Row population, Column population; Turkey_N, OldSteppe) (highlighted in red if |Z|>3). Above diagonal are inbreeding-corrected Fsr values (highlighted in yellow if Fsr>0.005). Extended Data Table 1: Ancestry change over time in Britain

Related Figures (11)

Figure 1: Ancient DNA Dataset. Geographic distribution of sites and temporal distribution of individuals in the period 4000 BCE-43 CE. Newly reported data are in black; published data In the map of Britain we label sites that harbour individuals who are ancestry outliers relative to others of the same period. In the timeline we denote the archaeological periods accordin, chronology: Neolithic (3950-2450 BCE), Chalcolithic and Early Bronze Age (C/EBA, 2450-1550 BCE), Middle Bronze Age (MBA, 1550-1150 BCE), Late Bronze Age (LBA, 1150-750 1 Roman Iron Age (IA, 750 BCE-43 CE). To aid visualisation, we add jitter on the Y axis and randomly sample dates from their probability distributions given in Online Table 1.
Figure 1: Ancient DNA Dataset. Geographic distribution of sites and temporal distribution of individuals in the period 4000 BCE-43 CE. Newly reported data are in black; published data In the map of Britain we label sites that harbour individuals who are ancestry outliers relative to others of the same period. In the timeline we denote the archaeological periods accordin, chronology: Neolithic (3950-2450 BCE), Chalcolithic and Early Bronze Age (C/EBA, 2450-1550 BCE), Middle Bronze Age (MBA, 1550-1150 BCE), Late Bronze Age (LBA, 1150-750 1 Roman Iron Age (IA, 750 BCE-43 CE). To aid visualisation, we add jitter on the Y axis and randomly sample dates from their probability distributions given in Online Table 1.
LBA (n=23), and stabilized in the IA at 37.9+0.4% (n=273) (here and below, we quote one
LBA (n=23), and stabilized in the IA at 37.9+0.4% (n=273) (here and below, we quote one
a at — hemi fae aa! — ad Lad nical ide! — had BER ROR ACCOAERS VLR 4etst ancestry increased in the south (England and Wales) beginning with the Margetts Pit MBA outliers but hardly in the north (Scotland). Estimates from qpAdm with one standard error in four periods. (B) Detecting runs of homozygosity (ROH), we find that close kin unions were rare at all periods as reflected in the paucity of individuals harbouring >50 centimorgans (cM) of their genome in ROH segments of >12 cM (red dots in top panel). The number of ROH of size 4-8 cM per individual (bottom panel) reflects the rate at which distant relatives have children, providing information about the sizes of mate pools (N.) averaged over the hundreds of years prior to when individuals lived; thus, the broad trend of an approximately four-fold drop in N. from the Neolithic to the IA is robust, but we may miss fluctuations on a time scale of centuries. The thick black line is Ne from a Gaussian process with a 600-year smoothing kernel (gray area 95% confidence interval); horizontal lines show period averages from maximum likelihood; and we interrupt the fitted line for periods with too little data for accurate inference (<10 individuals in a 400-year interval centered on the noint).
a at — hemi fae aa! — ad Lad nical ide! — had BER ROR ACCOAERS VLR 4etst ancestry increased in the south (England and Wales) beginning with the Margetts Pit MBA outliers but hardly in the north (Scotland). Estimates from qpAdm with one standard error in four periods. (B) Detecting runs of homozygosity (ROH), we find that close kin unions were rare at all periods as reflected in the paucity of individuals harbouring >50 centimorgans (cM) of their genome in ROH segments of >12 cM (red dots in top panel). The number of ROH of size 4-8 cM per individual (bottom panel) reflects the rate at which distant relatives have children, providing information about the sizes of mate pools (N.) averaged over the hundreds of years prior to when individuals lived; thus, the broad trend of an approximately four-fold drop in N. from the Neolithic to the IA is robust, but we may miss fluctuations on a time scale of centuries. The thick black line is Ne from a Gaussian process with a 600-year smoothing kernel (gray area 95% confidence interval); horizontal lines show period averages from maximum likelihood; and we interrupt the fitted line for periods with too little data for accurate inference (<10 individuals in a 400-year interval centered on the noint).
Figure 4: Genetic change in Britain in the context of Europe-wide trends. (A) North-south ancestry convergence. We show eight ancient DNA time transects for up to four periods, 1 average of dates of individuals in periods defined for each region as in Online Table 5 (we use the N/C/EBA/MBA/LBA/IA division into archaeological periods for Britain, but avoid u most of the rest of Europe because chronological boundaries differ). Dotted lines show which points are regionally grouped and should not be interpreted as implying a smooth chang (B) The allele conferring lactase persistence made its major rise about a millennium earlier in Britain than in Central Europe suggesting different selection regimes and possibly cult differences in the use of dairy products in the two regions in the IA. We visualise the frequency trajectory of the lactase persistence allele at SNP rs4988235 by using the GaussianProcessRegressor function from the Scikit-learn library in Python with parameter alpha=0.1 and 1*RationalQuadratic kernel with parameter length_scale_bounds=(1, 100( GLIMPSE*’ software to impute diploid genotype posterior probabilities (GP), restricting to samples with max(GP)>0.9 for this SNP. The analysis includes 459 ancient individuals fic 468 from Central Europe (Czech Republic, Slovakia, Croatia, Hungary, Austria, Germany and Slovenia); to represent modern Britain we use a pool of 190 CEU and GBR individuals 1000 Genomes Project*®, and to represent modern Central Europe we use 288 individuals from the modern Czech Republic*®. Each vertical bar represents the derived allele frequenc: individual with values [0, 0.5, 1]; we use jitter on the x-axis for each vertical bar, and we show in shading the inferred 95% confidence interval for the allele frequency at each time pc
Figure 4: Genetic change in Britain in the context of Europe-wide trends. (A) North-south ancestry convergence. We show eight ancient DNA time transects for up to four periods, 1 average of dates of individuals in periods defined for each region as in Online Table 5 (we use the N/C/EBA/MBA/LBA/IA division into archaeological periods for Britain, but avoid u most of the rest of Europe because chronological boundaries differ). Dotted lines show which points are regionally grouped and should not be interpreted as implying a smooth chang (B) The allele conferring lactase persistence made its major rise about a millennium earlier in Britain than in Central Europe suggesting different selection regimes and possibly cult differences in the use of dairy products in the two regions in the IA. We visualise the frequency trajectory of the lactase persistence allele at SNP rs4988235 by using the GaussianProcessRegressor function from the Scikit-learn library in Python with parameter alpha=0.1 and 1*RationalQuadratic kernel with parameter length_scale_bounds=(1, 100( GLIMPSE*’ software to impute diploid genotype posterior probabilities (GP), restricting to samples with max(GP)>0.9 for this SNP. The analysis includes 459 ancient individuals fic 468 from Central Europe (Czech Republic, Slovakia, Croatia, Hungary, Austria, Germany and Slovenia); to represent modern Britain we use a pool of 190 CEU and GBR individuals 1000 Genomes Project*®, and to represent modern Central Europe we use 288 individuals from the modern Czech Republic*®. Each vertical bar represents the derived allele frequenc: individual with values [0, 0.5, 1]; we use jitter on the x-axis for each vertical bar, and we show in shading the inferred 95% confidence interval for the allele frequency at each time pc
Extended Data Table 2: Fine structure in Iron Age Britain. Note: This is an expanded version of Table | including not just ancestry estimates for each group but also pairwise population comparisons. We pool all individuals from each period and region removing those failing goAdm modeling at p<0.01 according to Online Table 5). In the left columns are gpAdm estimates of ancestry for each group. Below diagonal are Z-scores from /4(Row population, Column population; Turkey_N, OldSteppe) (highlighted in red if |Z|>3). Above diagonal are inbreeding-corrected Fsr values (highlighted in yellow if Fst>0.0025).
Extended Data Table 2: Fine structure in Iron Age Britain. Note: This is an expanded version of Table | including not just ancestry estimates for each group but also pairwise population comparisons. We pool all individuals from each period and region removing those failing goAdm modeling at p<0.01 according to Online Table 5). In the left columns are gpAdm estimates of ancestry for each group. Below diagonal are Z-scores from /4(Row population, Column population; Turkey_N, OldSteppe) (highlighted in red if |Z|>3). Above diagonal are inbreeding-corrected Fsr values (highlighted in yellow if Fst>0.0025).
Extended Data Figure 1: Post-MBA Britain was not a mix of earlier British populations. (A) qpAdm P-values for modeling British groups as a mix of Neolithic and Chalcolithic/EE from England and Wales, Scotland, or Ireland (outgroups OldAfrica, OldSteppe, Turkey_N, CzechRepublic.Slovakia.Germany_3800.to.2700BP, Netherlands_C.EBA, Poland_Globul Spain.Portugal_4425.to.3800BP, CzechRepublic.Slovakia.Germany_4465.to.3800.BP, Sardinia_4100.to.2700BP, Sardinia_8100.to.4100BP, Spain.Portugal_6500.to.4425BP). We hi p<0.05 (yellow) or p<0.005 (red). Both sources and target populations in this analysis remove outlier individuals (‘Filter 2” in Online Table 5); we obtain qualitatively similar resul outlier individuals are not removed (not shown). In England and Wales, the main MBA group, and to a marginal extent the main LBA grouping, can be modelled as a mixture of the N C/EBA populations; we can reject such a models for the IA (we can more weakly reject this model even for Scotland, whether using England/Wales or Scotland sources, suggesting so same stream of migration also affected Scotland albeit probably in a more subtle way). (B) To obtain insight into the source of the new ancestry in Britain in the IA, we computed Sa(England.and.Wales_IA, a(England.and.Wales_N) + (1-a)(England.Wales_C.EBA); R1, R2) for several different (R1, R2) pairs. If England.and.Wales_IA is a simple mixture of England.and.Wales_N and England.and.Wales_C.EBA without additional ancestry, then for some mixture proportion the statistic will be consistent with zero for all (R1, R2 pairs). W = (OldAfrica, OldSteppe) feasible Z-scores (Z1 in the plot) are observed when a~0.85, showing that ~85% ancestry from England.and.Wales_C.EBA ancestry is needed to contribute proportion of Steppe ancestry in England.and.Wales_IA. However, when (R1, R2) is (Balkan_N, Sardinian_8100.to.4100BP), we get infeasible Z-scores (Z2) of <-6 across the range ' other Z-score (Z1) is remotely feasible. Thus, Iron Age people from England and Wales must have ancestry from an additional population deeply related to Sardinian Early Neolithic
Extended Data Figure 1: Post-MBA Britain was not a mix of earlier British populations. (A) qpAdm P-values for modeling British groups as a mix of Neolithic and Chalcolithic/EE from England and Wales, Scotland, or Ireland (outgroups OldAfrica, OldSteppe, Turkey_N, CzechRepublic.Slovakia.Germany_3800.to.2700BP, Netherlands_C.EBA, Poland_Globul Spain.Portugal_4425.to.3800BP, CzechRepublic.Slovakia.Germany_4465.to.3800.BP, Sardinia_4100.to.2700BP, Sardinia_8100.to.4100BP, Spain.Portugal_6500.to.4425BP). We hi p<0.05 (yellow) or p<0.005 (red). Both sources and target populations in this analysis remove outlier individuals (‘Filter 2” in Online Table 5); we obtain qualitatively similar resul outlier individuals are not removed (not shown). In England and Wales, the main MBA group, and to a marginal extent the main LBA grouping, can be modelled as a mixture of the N C/EBA populations; we can reject such a models for the IA (we can more weakly reject this model even for Scotland, whether using England/Wales or Scotland sources, suggesting so same stream of migration also affected Scotland albeit probably in a more subtle way). (B) To obtain insight into the source of the new ancestry in Britain in the IA, we computed Sa(England.and.Wales_IA, a(England.and.Wales_N) + (1-a)(England.Wales_C.EBA); R1, R2) for several different (R1, R2) pairs. If England.and.Wales_IA is a simple mixture of England.and.Wales_N and England.and.Wales_C.EBA without additional ancestry, then for some mixture proportion the statistic will be consistent with zero for all (R1, R2 pairs). W = (OldAfrica, OldSteppe) feasible Z-scores (Z1 in the plot) are observed when a~0.85, showing that ~85% ancestry from England.and.Wales_C.EBA ancestry is needed to contribute proportion of Steppe ancestry in England.and.Wales_IA. However, when (R1, R2) is (Balkan_N, Sardinian_8100.to.4100BP), we get infeasible Z-scores (Z2) of <-6 across the range ' other Z-score (Z1) is remotely feasible. Thus, Iron Age people from England and Wales must have ancestry from an additional population deeply related to Sardinian Early Neolithic
Extended Data Figure 3: Frequency change over time at two phenotypically important alleles. Present-day frequency is shown by the red line. (A-D/Top) Lactase persistence allele at rs4988235. (E-H/Bottom) Light skin pigmentation allele at rs16891982. In Britain the rise in frequency of the lactase persistence allele occurred earlier than in Central Europe. This analysis is based on direct observation of alleles; imputation results are consistent (Figure 4B). Online Table 8 gives full numerical results for 107 phenotypically important alleles.
Extended Data Figure 3: Frequency change over time at two phenotypically important alleles. Present-day frequency is shown by the red line. (A-D/Top) Lactase persistence allele at rs4988235. (E-H/Bottom) Light skin pigmentation allele at rs16891982. In Britain the rise in frequency of the lactase persistence allele occurred earlier than in Central Europe. This analysis is based on direct observation of alleles; imputation results are consistent (Figure 4B). Online Table 8 gives full numerical results for 107 phenotypically important alleles.
Extended Data Figure 4: Y chromosome haplogroup frequencies change over time. We show the estimated frequency of the characteristically British Y chromosome haplogroup R1b-P312 L21/M529 in all individuals for which we are able to make a determination and which are not first-degree relatives of a higher coverage individual in the dataset. The frequency increases significantly from ~0% in the whole island Neolithic, to 8944% in the whole island C/EBA. It declines albeit non-significantly to 794+9% in the MBA and LBA (from this time on restricting to England and Wales because of the autosomal evidence of a change in ancestry in the south but not the north). It further declines to 68+4% in the IA, a significant reduction relative to the C/EBA (P=0.014 by a two-sided chi-square contingency test). There is a further reduction from this time to the present, where the proportion is 4343% in Wales and the west of England (P=5x10° for a reduction relative to the IA), and 1442% in the center and east of England (P=3x10°? for a reduction relative to the IA), potentially due to later migrations bearing a different distribution of Y chromosome haplogroup frequencies.
Extended Data Figure 4: Y chromosome haplogroup frequencies change over time. We show the estimated frequency of the characteristically British Y chromosome haplogroup R1b-P312 L21/M529 in all individuals for which we are able to make a determination and which are not first-degree relatives of a higher coverage individual in the dataset. The frequency increases significantly from ~0% in the whole island Neolithic, to 8944% in the whole island C/EBA. It declines albeit non-significantly to 794+9% in the MBA and LBA (from this time on restricting to England and Wales because of the autosomal evidence of a change in ancestry in the south but not the north). It further declines to 68+4% in the IA, a significant reduction relative to the C/EBA (P=0.014 by a two-sided chi-square contingency test). There is a further reduction from this time to the present, where the proportion is 4343% in Wales and the west of England (P=5x10° for a reduction relative to the IA), and 1442% in the center and east of England (P=3x10°? for a reduction relative to the IA), potentially due to later migrations bearing a different distribution of Y chromosome haplogroup frequencies.
Extended Data Figure 5: Version of Figure 3A contrasting Kent to the rest of southern Britain. We show the period 2450-1 BCE. All the high EEF outliers at the M-LBA are from Kent—the part of the island closest to France—and in addition all the individuals from 1000-875 BCE from the group of samples showing the ramp-up from MBA to IA levels of EEF ancestry are from Kent (5 from Cliffs End Farm and 3 from East Kent Access Road). This suggests the possibility that this small region was the gateway for migration to Britain at the M-LBA. Further sampling from the rest of Britain at the M-LBA is critical in order to understand the dynamics of how this ancestry spread more broadly, but the fact that only sample from the LBA that is not from Kent, [12624 from Blackberry Field in Potterne in Wiltshire at 950-750 BCE, already has IA levels of EEF ancestry, suggests that this ancestry began spreading more broadly by the date of this individual.
Extended Data Figure 5: Version of Figure 3A contrasting Kent to the rest of southern Britain. We show the period 2450-1 BCE. All the high EEF outliers at the M-LBA are from Kent—the part of the island closest to France—and in addition all the individuals from 1000-875 BCE from the group of samples showing the ramp-up from MBA to IA levels of EEF ancestry are from Kent (5 from Cliffs End Farm and 3 from East Kent Access Road). This suggests the possibility that this small region was the gateway for migration to Britain at the M-LBA. Further sampling from the rest of Britain at the M-LBA is critical in order to understand the dynamics of how this ancestry spread more broadly, but the fact that only sample from the LBA that is not from Kent, [12624 from Blackberry Field in Potterne in Wiltshire at 950-750 BCE, already has IA levels of EEF ancestry, suggests that this ancestry began spreading more broadly by the date of this individual.
Related topics:
GeographyArchaeologyCeltic StudiesBronze Age ArchaeologyLate Bronze Age archaeologyAncient DNA ResearchMedicineBronze Age BritainaDNA AnalysisLate Bronze Age Early Iron Age
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved