|
Post by Admin on Jul 23, 2022 17:06:43 GMT
A 3D basicranial shape-based assessment of local and continental northwest European ancestry among 5th to 9th century CE Anglo-Saxons Kimberly A. Plomp ,Keith Dobney,Mark Collard Published: June 23, 2021 doi.org/10.1371/journal.pone.0252477Abstract The settlement of Great Britain by Germanic-speaking people from continental northwest Europe in the Early Medieval period (early 5th to mid 11th centuries CE) has long been recognised as an important event, but uncertainty remains about the number of settlers and the nature of their relationship with the preexisting inhabitants of the island. In the study reported here, we sought to shed light on these issues by using 3D shape analysis techniques to compare the cranial bases of Anglo-Saxon skeletons to those of skeletons from Great Britain that pre-date the Early Medieval period and skeletons from Denmark that date to the Iron Age. Analyses that focused on Early Anglo-Saxon skeletons indicated that between two-thirds and three-quarters of Anglo-Saxon individuals were of continental northwest Europe ancestry, while between a quarter and one-third were of local ancestry. In contrast, analyses that focused on Middle Anglo-Saxon skeletons suggested that 50–70% were of local ancestry, while 30–50% were of continental northwest Europe ancestry. Our study suggests, therefore, that ancestry in Early Medieval Britain was similar to what it is today—mixed and mutable. Introduction The settlement of large parts of the island of Great Britain by Germanic-speaking people from continental northwest Europe between the mid 5th and early 7th centuries CE has long been recognised as an important event, leading as it did to the formation of the ethnic group called the Angli or, more commonly, the Anglo-Saxons [1]; the development of the English language; and the formation of the Kingdom of England and, eventually, the United Kingdom [2–5]. Not surprisingly, therefore, this episode has been the subject of a considerable amount of research (see Hardland [6] for a recent review). Despite this extensive work, uncertainty remains about the number of settlers and the nature of their relationship with the preexisting inhabitants of the island, especially the Romano-British. Traditionally, knowledge of the settlement of Britain by the Angles, Saxons, Jutes, and Frisians relied on two historical texts, Historia Ecclesiastica Gentis Anglorum and the Anglo-Saxon Chronicle. Written by the Venerable Bede, Historia Ecclesiastica is thought to have been completed in 731 CE. The original version of the Anglo-Saxon Chronicle was compiled in the late 9th century CE, and copies of it were updated until at least the mid 12th century CE. Both of these documents describe a mass invasion and a rapid replacement of the indigenous population [7–9]. This picture has been challenged by archaeologists. Several researchers have argued that the archaeological record reveals that the changes associated with the arrival of the Germanic-speaking settlers in Britain occurred relatively slowly and that this is inconsistent with the idea that the settlers replaced the Romano-British [10–12]. In line with this, analyses of oxygen and strontium isotopes from Anglo-Saxon skeletons have found that only a small minority of the sampled individuals were from the Continent [13–15]. Geneticists have also sought to shed light on these issues, but the results they have obtained are highly variable. A comparison of the ancestry estimates reported by Weale et al. [3], Leslie et al. [16], and Schiffels et al. [17] illustrates this. Based on analyses of modern Y chromosome DNA, Weale et al. [3] concluded that 50–100% of men in central England have male ancestors from continental northwest Europe. Leslie et al.’s [16] analyses of modern genomes suggested that 10–40% of people from central and southern England have continental northwest European ancestry. Schiffels et al. [17] analysed the genomes of ten Iron Age and Anglo-Saxon individuals from England. They found that the Anglo-Saxon individuals were closely related to modern Danish and Dutch people and estimated that introgression from continental northwest Europe accounted for 38% of the ancestry of people currently living in eastern England. Here, we report a study designed to bring a new line of evidence to bear on the issue. We used three-dimensional (3D) shape analysis techniques to compare the cranial bases of Anglo-Saxon skeletons to those of skeletons from Great Britain that pre-date the Early Medieval period and skeletons from Denmark that date to the Iron Age. We focused on the basicranium because previous studies have shown that the 3D shape of this region of the skull can be informative about relatedness among human populations (e.g. [18–21]). The shape analysis techniques we utilised are collectively referred to as ‘Geometric Morphometrics’ [22–24]. These techniques have been used extensively by palaeoanthropologists to tackle comparable problems (e.g. [25, 26]). The goal of the analyses was to estimate the percentage of Anglo-Saxon individuals who were of British ancestry and the percentage who were of continental northwest European ancestry.
|
|
|
Post by Admin on Jul 24, 2022 16:48:49 GMT
Materials and methods The sample is summarised in Table 1 and the locations of the sites from which the remains were obtained are shown in Fig 1. Further information about the individuals in the sample can be found in the S1 File. Fig 1. Map depicting the locations of the five Anglo-Saxon cemeteries sampled in this study (red dots). The map also shows the homelands and destinations of the main Germanic-speaking groups who settled in Britain in the Early Medieval period—the Angles (pink ellipses), Saxons (green ellipses), Jutes (purple ellipses), and Frisians (blue). doi.org/10.1371/journal.pone.0252477.g001We recorded data on a total of 236 individuals, all of whom had an intact basicranium. Only adults were measured to avoid the confounding effects of ontogeny; individuals were judged to be adult on the basis of dental eruption and epiphyseal fusion. Both males and females were included in the sample, with sex being estimated primarily on the basis of pelvic morphology, especially the presence/absence of the ventral arc [27]. Cranial indicators of sex were also considered when necessary [27]. Eighty-nine of the individuals come from five Anglo-Saxon cemeteries: Brandon in Suffolk, Breedon-on-the-Hill in Leicestershire, Burwell in Cambridgeshire, Dover Buckland in Kent, and Eriswell in Suffolk. Breedon-on-the-Hill, Dover Buckland, and Eriswell date to the Early Anglo-Saxon Period (410 CE-660 CE), while Brandon and Burwell date to the Middle Anglo-Saxon Period (660 CE-889 CE). Hereinafter, we will refer to the individuals from Breedon-on-the-Hill, Dover Buckland, and Eriswell as Early Anglo-Saxons, and the individuals from Brandon and Burwell as Middle Anglo-Saxons. Another 101 of the individuals are from pre-Medieval sites in England. The sites in question are Poundbury and Maiden Castle in Dorset, and Hallet’s Garage in Kent. Poundbury and Hallet’s Garage are both of Romano-British date (43 CE-410 CE), while Maiden Castle dates to the preceding Iron Age (800 BCE-43 CE). The remaining 46 individuals are from various sites in Denmark dating to the Iron Age (800 BCE-399 CE). We focused on individuals from Denmark partly because Schiffels et al.’s [17] analyses indicated that the early English individuals in their sample from Cambridgeshire were genetically closer to modern Danish people, and partly because they are better preserved than similarly dated skeletal remains from the other main potential source areas, northern Germany and the Netherlands. We used photogrammetry to generate 3D models of the crania. Each cranium was photographed 150 times with an eight-megapixel digital single-lens reflex Canon EOS 77D camera mounted with a Canon 50mm lens. The cranium was placed on a PalaeoPi rotating table and then photographs were shot at intervals of approximately 10°, as per Evin et al. [28]. Subsequently, the photographs for a single cranium were converted into a 3D model in Metashape. The photographs were aligned at the ‘high’ accuracy level and 3D depth maps generated. After this, the 3D depth maps were used to create a mesh model of each cranium. To digitise the 3D shape of the individuals’ cranial bases, we imported the mesh models into MorphoDig [29] and recorded the 3D Cartesian coordinates of a total of 34 anatomical landmarks (Fig 2). The landmarks were chosen based on those used by Harvati and Weaver [18] to capture the shape of the basicranium. According to Bookstein’s [30] widely used scheme, seven of the landmarks are Type I landmarks and 27 are Type II landmarks.
|
|
|
Post by Admin on Jul 24, 2022 21:02:20 GMT
Fig 2. Location of the 34 landmarks used in the present study. doi.org/10.1371/journal.pone.0252477.g002A single observer recorded the data (KAP). Intra-observer error was assessed in the manner outlined by Neubauer et al. [31, 32]. A single cranium was digitised ten times and then Morphologika [33] was used to compare the greatest Procrustes distance between the ten repeated landmark configurations with the smallest Procrustes distance between the non-repeated landmark configurations representing all the crania. The smallest distance between the non-repeated cranium was almost double the greatest distance between the repeated crania. According to Neubauer et al. [31, 32], this amount of error is unlikely to influence the analysis of the shape variance of the sample. After assessing the intra-observer error, we divided the dataset by sex. We did so to avoid sexual dimorphism influencing the results. We then sought to minimise the impact of a number of other potential confounding factors on each sex-specific dataset. Following Klingenberg et al. [33], we reflected and re-labelled the landmark coordinates and subjected the data to generalised Procrustes analysis, which removes translational and rotational effects and scales the configurations to centroid size. We removed asymmetry by calculating the average Procrustes coordinates between the original and reflected landmarks. These procedures were carried out in MorphoJ [33]. Having minimised the confounding effects of translation, rotation, size, and asymmetry, we divided the individuals in the female dataset into four groups: Early Anglo-Saxons, Middle Anglo-Saxons, Pre-Medieval British, and Danish. We then did the same for the male dataset. Subsequently, we carried out three sets of analyses. In the first, we explored the shape variance among the four groups. We did this by applying canonical variates analysis (CVA) to the Procrustes coordinates. We conducted two analyses, one that concentrated on the female dataset and one that focused on the male dataset. The CVAs were performed in MorphoJ [33]. In the second set of analyses, we investigated whether there were significant shape differences among the four groups. The primary goal of these analyses was to ensure that the two potential source groups, the Pre-Medieval British and Danish, differed because this was a prerequisite for making meaningful statements about the ancestry of the Anglo-Saxon individuals. We first subjected each dataset to principal components analysis (PCA) followed by the principal component (PC) reduction procedure developed by Baylac and Frieβ [34]. This procedure aims to reduce noise from PCs that account for little variance while still retaining all relevant shape information. It tackles this optimisation problem by progressively adding PCs into the analysis until the cross-validation percentage begins to drop. Thereafter, we subjected the retained PCs to MANOVA. Beginning with the females, we assessed whether there were any significant differences among the four groups. Because this analysis returned a significant result, we proceeded to compare the populations on a pairwise basis, i.e. Danish vs Pre-Medieval British, Early Anglo-Saxons vs. Danish, Early Anglo-Saxons vs. Pre-Medieval British, Middle Anglo-Saxons vs. Danish, Middle Anglo-Saxons vs. Pre-Medieval British, Early Anglo-Saxons vs. Middle Anglo-Saxons. Subsequently, we repeated the MANOVAs with the males. The PCAs were performed in MorphoJ [33], while the MANOVAs were performed in R [35]. The PC-reduction procedure was implemented in R [35]. In the third set of analyses, we assessed the relative contribution of the two potential source groups to the ancestry of the two Anglo-Saxon groups. We did this by applying cross-validated linear discriminant analysis (LDA) to the PCs used in the previous set of analyses. Following Evin et al. [36], we designated the potential source groups as the known samples and then directed the LDA to indicate which of the potential source groups the Anglo-Saxon individuals most likely belonged. Because there are two potential source groups, the standard average percentage used to attribute an individual to one of the source groups is 50% [36, 37]. We opted for a more conservative affiliation value, and so an individual was deemed to be attributed to a given source group if the average percentage for that group was ≥55%. If both average attribution percentages for an individual were ≤54%, the individual was deemed to be unattributable. Thereafter, we calculated the percentage of Early Anglo-Saxons and Middle Anglo-Saxons that were attributed to each of the potential source groups. Once again, the female and male individuals were analysed separately. The LDAs were performed in R [35].
|
|
|
Post by Admin on Jul 25, 2022 16:00:33 GMT
Results The CVA of the female individuals yielded three CVs. CV1 accounted for 58% of the variance, CV2 for 27%, and CV3 for 14%. Fig 3A shows the values for CV1 and CV2 for the female individuals plotted against each other. The Pre-Medieval British and Danish overlap nearly completely on CV2 but only partially on CV1. On CV1, the bulk of the Pre-Medieval British individuals are located more positively than the majority of the Danish ones. The Early Anglo-Saxons overlap substantially with both the Pre-Medieval British and the Danish on CV2, but only with the Danish on CV1. The Middle Anglo-Saxons overlap substantially with the Early Anglo-Saxons on CV1 but do not overlap with the Pre-Medieval British or Danish on the same CV. In contrast, the Middle Anglo-Saxons overlap somewhat with the Pre-Medieval British and Danish individuals on CV2; the overlap with the Pre-Medieval British individuals is greater than the overlap with the Danish ones. Fig 3. Plots of CV values for female individuals when (a) CV1 is plotted against CV2 and (b) CV2 is plotted against CV3. doi.org/10.1371/journal.pone.0252477.g003The values for CV2 and CV3 for the females are plotted against each other in Fig 3B. The Pre-Medieval British and Danish individuals overlap almost entirely on CV2 but only a small amount on CV3. On CV3, the middle of the distribution of the Pre-Medieval British is closer to the negative end of the CV than the middle of the distribution of the Danish. The Early Anglo-Saxons overlap substantially with the Pre-Medieval British on both CVs. In contrast, while the Early Anglo-Saxons overlap with the Danish individuals on CV2, they barely overlap with them on CV3. The Middle Anglo-Saxons have a different relationship with the Pre-Medieval British and Danish individuals. They barely overlap with the Danish on either CV. They overlap markedly more with the Pre-Medieval British on CV2 and even more with them on CV3. The Middle Anglo-Saxons overlap with Early Anglo-Saxons on CV3 but are separated from them on CV2. The CVA of the male individuals also produced three CVs. CV1 accounted for 44% of the variation, CV2 for 37%, and CV3 for 19%. Fig 4A shows the values for CV1 and CV2 for the male individuals plotted against each other. The Pre-Medieval British and Danish individuals overlap on both CVs, although to a lesser extent on CV1. On both CVs, the centre of the distribution of Pre-Medieval British individuals is closer to the positive end of the CV than the centre of the distribution of the Danish individuals. The Early Anglo-Saxons overlap substantially with both the Pre-Medieval British and Danish on CV2. The situation is different when we look at CV1. On this CV, the Early Anglo-Saxons overlap with the Danish individuals to a considerable extent, but barely overlap with the Pre-Medieval British ones. The Middle Anglo-Saxons overlap substantially with the Pre-Medieval British, Danish, and Early Anglo-Saxon individuals on CV1. On CV2, they do not overlap at all with the Danish individuals and only slightly with the Pre-Medieval British and Early Anglo-Saxon ones.
|
|
|
Post by Admin on Jul 25, 2022 18:11:24 GMT
Fig 4. Plots of CV values for male individuals when (a) CV1 is plotted against CV2 and (b) CV2 is plotted against CV3. doi.org/10.1371/journal.pone.0252477.g004The CV2 and CV3 values for the males are plotted against each other in Fig 4B. On CV2, the Danish individuals are positioned towards the negative end of the CV and the Pre-Medieval British ones towards the centre of the CV. The Early Anglo-Saxons overlap substantially with both the Danish and Pre-Medieval British ones on CV2. On CV3, Early Anglo-Saxons overlap to a great extent with the Pre-Medieval British individuals but only partially with the Danish ones. The Middle Anglo-Saxons are distributed in such a way that do not overlap with the Danish individuals on CV2. Instead, they partially overlap with the Pre-Medieval British individuals and extend to the positive end of the CV. On CV3, the Middle Anglo-Saxons overlap substantially with both the Pre-Medieval British and Danish individuals, although the overlap is greater in the case of the Pre-Medieval British ones. The Early Anglo-Saxons and Middle Anglo-Saxons overlap extensively on CV3, but barely overlap on CV2. The PC reduction procedure retained 10 PCs for the female individuals. Seventy percent of the total shape variance was accounted for by these PCs. The overall MANOVA for the females returned a significant result (λ 0.473, F = 2.447, p<0.001, ηp2 = 0.22). The results of the pairwise MANOVAs are presented in Table 2. Only three of the pair-wise comparisons returned significant results—Pre-Medieval British vs. Danish, Early Anglo-Saxon vs. Middle Anglo-Saxon, and Early Anglo-Saxon vs. Pre-Medieval British. Table 3 shows the percentage of the Early Anglo-Saxons and Middle Anglo-Saxons for whom the predicted source population was the Pre-Medieval British group, the percentage for whom the predicted source population was the Danish group, and the percentage deemed unattributable. The individual likelihood frequencies are listed in S3 and S4 Tables in S1 File. Table 3. Summary of the results of the linear discriminant analyses. The table shows the percentage of Anglo-Saxon individuals attributed to the Pre-Medieval British and Danish groups, and the percentage deemed to be of uncertain ancestry. See the Materials and Methods for details of the criterion used to identify the latter cases. doi.org/10.1371/journal.pone.0252477.t003In the female LDA analysis, 63% of the Early Anglo-Saxons were attributed to the Danish group, and 26% to the Pre-Medieval British group. The remaining 11% were deemed to be of uncertain ancestry. Thus, nearly two-thirds of the Early Anglo-Saxons were suggested to be of non-local ancestry, while a quarter were deemed to be of local ancestry. The pattern was different for the Middle Anglo-Saxons. Fifty-four percent of these individuals were attributed to the Pre-Medieval British group; 39% were attributed to the Danish group; and 8% were deemed to be of uncertain ancestry. In other words, just over half of the Middle Anglo-Saxon females were suggested to be of local ancestry and just under half were suggested to be of non-local ancestry. In the male LDA analysis, 65% of the Early Anglo-Saxons were attributed to the Danish group, and 25% to the Pre-Medieval British group. The remaining 10% were deemed to be of uncertain ancestry. As was the case with the female LDA analysis, the pattern was different for the Middle Anglo-Saxons. Sixty-nine percent of these were attributed to the Pre-Medieval British group, and 31% to the Danish group. As such, the results of the male LDA analysis were similar to those of the female LDA analysis. They also indicated that nearly two-thirds of the Early Anglo-Saxon males were of non-local ancestry while a quarter were of local ancestry. Also like the female data, the male data indicate that the pattern reversed with the Middle Anglo-Saxons, with the majority of them appearing to have been of local ancestry and the minority of non-local ancestry.
|
|