A female Viking warrior confirmed by genomics

Abstract Objectives The objective of this study has been to confirm the sex and the affinity of an individual buried in a well‐furnished warrior grave (Bj 581) in the Viking Age town of Birka, Sweden. Previously, based on the material and historical records, the male sex has been associated with the gender of the warrior and such was the case with Bj 581. An earlier osteological classification of the individual as female was considered controversial in a historical and archaeological context. A genomic confirmation of the biological sex of the individual was considered necessary to solve the issue. Materials and methods Genome‐wide sequence data was generated in order to confirm the biological sex, to support skeletal integrity, and to investigate the genetic relationship of the individual to ancient individuals as well as modern‐day groups. Additionally, a strontium isotope analysis was conducted to highlight the mobility of the individual. Results The genomic results revealed the lack of a Y‐chromosome and thus a female biological sex, and the mtDNA analyses support a single‐individual origin of sampled elements. The genetic affinity is close to present‐day North Europeans, and within Sweden to the southern and south‐central region. Nevertheless, the Sr values are not conclusive as to whether she was of local or nonlocal origin. Discussion The identification of a female Viking warrior provides a unique insight into the Viking society, social constructions, and exceptions to the norm in the Viking time‐period. The results call for caution against generalizations regarding social orders in past societies.

approximately 1,100 have been excavated, making it one of the largest known congregations of burials in the Viking world. The graves are distributed over large burial grounds encircling the town area.
One warrior grave, Bj 581, stands out as exceptionally wellfurnished and complete (Arbman, 1941;Thålin-Bergman, 1986) ( Figure   2 and S1). Prominently placed on an elevated terrace between the town and a hillfort, the grave was in direct contact with Birka's garrison. The grave goods include a sword, an axe, a spear, armour-piercing arrows, a battle knife, two shields, and two horses, one mare and one stallion; thus, the complete equipment of a professional warrior. Furthermore, a full set of gaming pieces indicates knowledge of tactics and strategy (van Hamel, 1934;Whittaker, 2006), stressing the buried   (Stolpe, 1889) individual's role as a high-ranking officer. As suggested from the material and historical records (Jesch, 1991;Jochens, 1996), the male sex has been associated with the gender of a warrior identity. Hence, the individual in Bj 581 was considered a male based on the assemblage of grave goods (Arbman, 1941;Gräslund, 1980), and the sex was only questioned after a full osteological and contextual analysis (Kjellstr€ om, 2016) that showed that the individual was a woman (S2 and S3). The existence of female warriors in Viking Age Scandinavia has been debated among scholars (Gardeła, 2013;Jesch, 1991;Jochens, 1996).
Though some Viking women buried with weapons are known, a female warrior of this importance has never been determined and Viking scholars have been reluctant to acknowledge the agency of women with weapons (Hernaes, 1984;Moen, 2011) (S1). The osteological analysis triggered questions concerning sex, gender and identity among Viking warriors. This made it important to further investigate the biological sex and to do additional analyses to explore the genetic affinity of the individual buried in Bj 581. Here we present data including nuclear DNA and strontium isotopes of the individual.

| Osteology
The skeleton was represented by bone elements from all body regions (S2). Stored with Bj 581 was also a femur belonging to another burial which was excluded. The age and sex estimation results, presented at a conference in 2014 (Kjellstr€ om, 2016), were based on osteological standard methods for morphologic indicators (Buikstra & Ubelaker, 1994) (S2-S3). The epiphyseal union was completed on all preserved bones, and the appearance of the auricular surface of the left hip bone meets the morphologic criteria for phase 3 according to methods by Lovejoy, Meindl, Mensforth, and Pryzbeck (1985) and Meindl and Owen (1989). Furthermore, the dental wear of the lower molars was clear but moderate (stage 2-4) (Brothwell, 1981). In all, this suggests that the individual was at least above 30 years of age. The greater sciatic notch of the hip bone was broad, and a wide preauricular sulcus was present. This, together with the lack of projection of the mental eminence on the mandible, assessed the individual as female. Additionally, the long bones are thin, slender and gracile which provide further indirect support for the assessment. No pathological or traumatic injuries were observed.

| Archaeological samples
Two samples intended for DNA analyses were removed from individual Bj 581. The samples were taken from the left canine and the left humerus; see Supporting Information Appendix, Section S3 for details.

| Sequence analyses
The computations were performed on resources provided by SNIC through Uppsala Multidisciplinary Center for Advanced Computational Science (UPPMAX) under the following projects: b2013240, b2015307, and b2016056. Sequence data was analyzed following previously published procedures (G€ unther et al., 2015;Omrak et al., 2016). De-multiplexed sequencing pair-end reads were merged, trimmed and then mapped to the human reference genomes build 36 and 37 with BWA v. 0.7.13 (Li & Durbin, 2010). The PCR duplicates were removed with FilterUniqueSAMCons.py (Kircher, 2012). Obtained DNA fragments were then checked for presence of 3 0 and 5 0 degradation patterns characteristic of ancient DNA (Briggs et al., 2007, Brotherton et al., 2007, Sawyer, Krause, Guschanski, Savolainen, & Pääbo, 2012 using PMDtools . Molecular sex assignment was estimated based on the ratio of sequences aligning to the two sex chromosomes, X and Y (Skoglund, Storå, G€ otherstr€ om, &  (Green et al., 2008). The consensus mitochondrial DNA (mtDNA) sequences were called using samtools package (Li et al. 2009) while the initial haplogroup assignment was in HAPLOFIND tool (Vianello et al., 2013

| Reference population panel
The genetic data from the Birka warrior was merged with three different population reference data-sets consisting of genotype SNP data from: the Human Origins dataset (Patterson et al., 2012;Lazaridis et al., 2014), the Swedish reference (Salmela et al., 2011), and the Population Reference Sample-POPRES (Nelson et al., 2008)

| Population comparisons
The principal component analyses (PCA) was undertaken using EIGEN-SOFT v.6.0.1 (Patterson et al., 2006). The analyses were performed with pseudo-haploid genomes and excluding of transition sites. To obtain information on genetic affinities between the Birka individual FIG URE 4 Maps visualizing the results of f 3 -statistic in which the individual from grave Bj 581 was compared to (a) Human Origins population reference panel (Lazaridis et al., 2014;Patterson et al., 2012) and (b) Population Reference Sample (POPRES) (Nelson et al., 2008). (c) The Birka warrior plotted against PC1 values for 21 Swedish subpopulations representing all counties and the total of 1525 individuals (Salmela et al., 2011). The three colours represent the conventional regional division to the southern G€ otaland (red), central Svealand ( (Patterson et al., 2012) and D statistics which were calculated using qpDstat of ADMIXTOOLS (Durand, Patterson, Reich, & Slatkin, 2011;Patterson et al., 2012). The results are summarized in Figure 4 and Supporting Information Figure S4.2a-

| Strontium isotope analyses
Three molar teeth from the lower jaw of individual Bj 581 were submitted to Sr analyses. For comparison teeth of additional five individuals from Birka were also analyzed, Supporting Information Table S5

| R E SU LTS
Genome-wide sequence data was generated in order to confirm the biological sex, to support skeletal integrity and also to investigate the genetic relationship of the individual to ancient individuals and modern day groups (S4). We investigated two samples from grave Bj 581, the left canine and the left humerus, which both yielded sufficient amounts of DNA for further analyses (Supporting Information Table S4.1). The DNA was extracted following previously published procedures (G€ unther et al., 2015). The bone extract contained 0.54% endogenous human DNA and the tooth extract contained 3.88%. The obtained DNA sequences showed all the characteristics of authentic and ancient DNA (Briggs et al., 2007) (Supporting Information Figure S4.1a,b), with mitochondrial contamination estimated to 0.42% (Green et al., 2008).
The Birka warrior was sequenced to mean 0.093 nuclear and 326.53 mitochondrial genome coverage. The mt-haplogroup was assigned to T2b (Vianello et al., 2013). The total of 11312749 reads mapped to the human genome. When corrected for clonality, the number of reads mapping to X and Y chromosomes were 248,170 and 247, respectively, resulting in the proportion of the alignments (R Y ) equal to 0.001 (SE 5 0.0001). The cut-off value for identification of females is R Y 0.016, showing that Bj 581 was a female (Skoglund et al., 2013) ( Figure 3). Hence the individual in grave Bj 581 is the first confirmed female high-ranking Viking warrior. Finally, we note that both biological sex and mtDNA analyses support the single-individual origin of the analyzed cranial and postcranial remains (same sex and mt-haplogroup T) (Supporting Information Table S4.3).
The Viking warrior female showed genetic affinity to present-day inhabitants of the British Islands (England and Scotland), the North Atlantic Islands (Iceland and the Orkneys), Scandinavia (Denmark and Norway) and to lesser extent Eastern Baltic Europe (Lithuania and Latvia) (Figure 4a,b and Supporting Information Figure S4.2a,b). Furthermore, the woman is significantly more similar to these modern northern Europeans than to southern Europeans (Supporting Information Table S4 family rather than the individual (Gardeła, 2013). Male individuals in burials with a similar material record are not questioned in the same way. Furthermore, an argument can be put forward that the grave originally may have held a second, now missing, individual. In which case, the weaponry could have been a part of that individual's grave furnishings, while the remaining female was buried without any objects. However, the distribution of the grave goods within the grave, their spatial relation to the female individual and the total lack of any typically female attributed grave artefacts disputes this possibility.
Do weapons necessarily determine a warrior? The interpretation of grave goods is not straight forward, but it must be stressed that the interpretation should be made in a similar manner regardless of the biological sex of the interred individual. Furthermore, the exclusive grave goods and two horses are worthy of an individual with responsibilities concerning strategy and battle tactics. The skeletal remains in grave Bj 581 did not exhibit signs of antemortem or perimortem trauma which could support the notion that the individual had been a warrior. However, contrary to what could be expected, weapon related wounds (and trauma in general) are not common in the inhumation burials at Birka (e.g., 2 out of 49 confirmed males showed signs of sharp force trauma).
Although not possible to rule out, previous arguments have likely neglected intersectional perspectives where the social status of the individual was considered of greater importance than biological sex. This type of reasoning takes away the agency of the buried female. As long as the sex is male, the weaponry in the grave not only belong to the interred but also reflects his status as warrior, whereas a female sex has raised doubts, not only regarding her ascribed role but also in her association to the grave goods.
Grave Bj 581 is one of three known examples where the individual has been treated in accordance with prevailing warrior ideals lacking all associations with the female gender (Jesch, 2009) (S1, S2, and S3). Furthermore, the exclusive grave goods and two horses are worthy of an individual with responsibilities concerning strategy and battle tactics.
Our results caution against sweeping interpretations based on archaeological contexts and preconceptions. They provide a new understanding of the Viking society, the social constructions and also norms in the Viking Age. The genetic and strontium data also show that the female warrior was mobile, a pattern that is implied in the historical sources, especially when it comes to the extended households of the elite (cf. Steinsland, Sigurdsson, Rekdal, & Beuermann, 2011). The female Viking warrior was part of a society that dominated 8th to 10th century northern Europe. Our results-that the high-status grave Bj 581 on Birka was the burial of a high ranking female Viking warrior-suggest that women, indeed, were able to be full members of male dominated spheres. Questions of biological sex, gender and social roles are complex and were so also in the Viking Age. This study shows how the combination of ancient genomics, isotope analyses and archaeology can contribute to the rewriting of our understanding of social organization concerning gender, mobility and occupation patterns in past societies.
Then the high-born lady saw them play the wounding game, she resolved on a hard course and flung off her cloak; she took a naked sword and fought for her kinsmen's lives, she was handy at fighting, wherever she aimed her blows.

ACKNOWLEDGMENTS
This study was made possible by the generous funding from Riksbankens jubileumsfond and the Swedish Research Council (VR). We also thank Per Hall, Karolinska Institute, for sharing the genotype data from modern-day Swedish individuals.