Sex bias in Neolithic megalithic burials

Abstract Objectives A statistical study comparing osteological and ancient DNA determinations of sex was conducted in order to investigate whether there are sex biases in United Kingdom and Irish Neolithic megalithic burials. Materials and Methods Genetic and osteological information from human individuals from 32 megalithic sites in the UK and Ireland dating from 4000 to 2500 cal. BCE was collected and statistically analyzed to test whether there is a true over‐representation of males at these sites. The published dataset from the study by Sánchez‐Quinto et al. in 2019 was initially analyzed before being refined and included in a larger dataset. Osteological analysis of sex bias was limited to adults with available sex estimations, and genetic analysis limited to published data Results Two sites consistently returned significant p‐values suggesting a potential over‐representation in osteological males at one site (Knowe of Midhowe, Orkney) and genetic males in the other (Primrose Grange, Ireland). Cumulative statistical analyses point towards a male bias in the representation of sexes in Neolithic megalithic burials, but these results do not reflect the site‐by‐site and regional variation found in this study. Discussion The interpretation of sex bias, that is, the over‐representation of one sex over another ‐ depends on other socio‐cultural variables (e.g., kinship) and the emphasis placed on statistical significance. The trend towards males being over‐represented in Neolithic megalithic burials is not as clear as previously thought, and requires further testing and data collection to uncover.


| INTRODUCTION
In recent years, ancient DNA (aDNA) research has greatly enhanced our understanding of human demographic history. Analysis of genomewide data has allowed researchers to investigate the unwritten history of ancient cultures and societies. Several studies have focused on and provided insights into the movement and integration of prehistoric peoples in Western Eurasia (Brace et al., 2019;Cassidy et al., 2016;Olalde et al., 2018), as well as into paleodemographic structures (Sánchez-Quinto et al., 2019;Fowler, 2022). In their study of genome-wide data from selected ancient individuals from Neolithic megalithic sites in modern-day United Kingdom, Ireland, and Sweden, Sánchez-Quinto et al. (2019) highlighted a possible male sex-biased admixture, that is, a bias in genetic contribution in a population, evidenced by limited Y-chromosome haplogroup lineages and a greater diversity in mtDNA haplogroups. This was particularly evident in the samples from words, utilizing combined genetic and osteological demographic data can help us answer whether or not these megalithic burials represent a possible manifestation of a male-based kinship structure such as a patrilineal society. The methods for determining sex estimates are outlined below.
To further investigate the relationship between burial practices and cultural interpretation presented in recent research on the Neolithic, we collected genetic and osteological information from human individuals with absolute and relative radiocarbon dates from megalithic sites in the United Kingdom and Ireland dating from 4000 to 2500 cal. BCE to test whether there is a true over-representation of males in megalithic burials or if it is a result of sample bias and/or missing data. This paper starts from the dataset published by Sánchez-Quinto et al. (2019) in order to test if their hypothesis held true in osteological context, then adds sites with osteological and genetic sex data to curate a larger dataset for further analysis. Statistical significance analysis was then conducted to test whether the male sexbias hypothesis held true given more data and if there is a difference in sex representation based on osteological versus genetic data.

| MATERIALS AND METHODS
Neolithic megalithic burials are defined herein as depositions of one or more human remains within above-ground stone monuments dating to the period 4000 to 2500 cal. BCE. Megalithic burials appeared as early as 4500 cal. BCE in continental Europe, but current radiocarbon dates indicate that the phenomenon began much later in the UK and Ireland (Schulz Paulsson, 2019). These monuments range from long barrows to stone cairns, each with their own subsets and typologies (see Barrett, 1988).

Sánchez-Quinto et al.'s study initially comprised only 24 individuals
from five sites; however, we excluded the Gotland site to focus on the United Kingdom and Ireland. Their study also included individuals from Balintore, a short cist burial in Scotland; Fussell's Lodge, which, while in the United Kingdom, is a wooden structure; and the Kolin Rondel site in the Czech Republic (9470). As these two sites are, respectively, not megalithic in nature and not located in the United Kingdom and/or Ireland, they are also not included in the current analysis. Sánchez-Quinto et al. compared their data with published genome-wide data of 36 individuals from 17 megalithic sites and found a higher male:female ratio in megalithic burials in the British Isles than in non-megalithic contexts (31 out of 42 samples) (op.cit.). The sites listed below, included in Sánchez-Quinto et al., represent Dataset 1 in this study, comprising 36 individuals from 17 sites, with a total minimum number of individuals (MNI) of 320, 32 of which were genetically sexed and 71 of which were osteologically sexed (see Table 1, Figure 1).
We also compiled an expanded dataset (Dataset 2) which includes sites from Dataset 1 with sufficient published osteological sex data, as well as additional sites that have undergone recent genetic studies and which also have osteological sex data. In total, Dataset 2 comprises 32 sites with a total MNI of 442, 78 of which were genetically sexed and 149 of which were osteologically sexed (see Table 2 and Figure 2). Ultimately, analysis focused on the 78 and 149 individuals with sex data-these are not mutually exclusive, as some, but not all, Ireland, but many of these remain unsexed and/or unanalyzed. Further research establishing their sex through osteological and/or genomic analysis may allow for more far-reaching conclusions in regards to sex demographics.
The osteological and demographic data from reports and catalogs of the 17 sites in Dataset 1 were reviewed in conjunction with the genetic data published by Sánchez-Quinto et al. (2019). In Dataset 2, the 32 sites were analyzed in terms of osteological age and sex distribution, both individually and geographically, to check for regional patterns. Based on published reports, the remains were categorized as either "Genetic Male," "Osteological Male," "Osteological Possible Male," "Genetic Female," "Osteological Female," "Osteological Possible Female," "Osteological Unknown Adult," and "Osteological Subadult" (the latter referring to individuals under approximately 18 years of age at death according to the "Updated Standards" guidelines on osteological sex assessment [Brickley & Buckberry, 2017, 33-34]).
Osteological sex is determined from sexually dimorphic skeletal traits present on the cranium, mandible, and os coxa (Buikstra & Ubelaker, 1994;Cox & Mays, 2000), while genetic sex is determined through extraction of aDNA usually from the petrous portion of the temporal bone or from dentine within the tooth root by calculating the ratios of X and Y chromosomes or the ratios of autosomal chromosomes (Bauer, Niederstätter, McGlynn, Stadler, & Parson, 2013, 583;Skoglund, Storå, Götherström, & Jakobsson, 2013;Fu et al., 2016;Lamnidis et al., 2018;Mittnik, Wang, Svoboda, & Krause, 2016  , but accuracy is often decreased in highly fragmented material due to differing preservation levels. As a result of variable preservation and/or the normal variation in the skeletal presentation of secondary sex characteristics, some individuals may be given an "indeterminate" sex estimation. Both datasets in the current study include sites where the skeletal remains are often highly fragmented and commingled, and where osteological reports varied from the recent and highly detailed (such as O'Donnabhain and Tesorieri's (2014) report on Poulnabrone in Lynch, 2014) to the brief and antiquarian (such as the description of Upper Swell by Greenwell & Rolleston, 1877). A large proportion of individuals in our datasets were not assigned a sex (discussed further below), and furthermore not every "indeterminate" or "unsexed" individual underwent genetic analysis.
Furthermore, osteological sex can only be determined for individuals who have reached puberty and have sufficiently visible sexually dimorphic traits (Buikstra & Ubelaker, 1994;Cox & Mays, 2000;White et al., 2012), and therefore can only be discussed in terms of adult populations, although some novel methods have been proposed for adolescents (see Lewis, Shapland, & Watts, 2016). Therefore, in some cases, genetic sex determination was successfully undertaken on subadult remains and included in sex-distribution analyses, but those same individuals could not be included in the analysis for the osteological data. As a result, sample size for genetic sex in some studies differs from sample size for osteological sex, which itself differs from estimated site MNI, leading to the decision in the current study to study genetic and osteological sex separately. It should be noted that Binomial significance tests were conducted on sites with more than one genetically analyzed individual and where macroscopic analyses of age and sex of skeletal remains were undertaken and published; sites with only one genetically analyzed individual were included as part of an overall regional analysis of the sites. Four sites from Dataset 1 and nine sites from Dataset 2 ultimately had sufficient genetic and osteological data to undertake binomial testing. "Possible Male" and "Possible Female" counts were removed during a second round of analysis to test if their presence impacted statistical significance. Bonferroni multiple test corrections were then conducted in order to check for false positives.
In Dataset 1, the percentage of adults of unknown sex ranged from 40.7% in the Scotland data to 50% in the Ireland data, with the England data falling closer to the Scotland percentage at 42.1%. This inability to determine sex for a large number of individuals, either due to preservation or normal variation in sexually dimorphic traits, is a substantial limitation that should be taken into account, as this represents missing data that could significantly impact any conclusions drawn. Additionally, sites in Scotland and Ireland had more genetic samples than England in both datasets, which further complicates the statistical interpretations.

| RESULTS
Given the null hypothesis that there is an equal representation of sexes in Neolithic megalithic burials in the United Kingdom and Ireland, we used the standard, but arbitrary, cut-off value of p < 0.05 for determining significance.
Only three of the 17 sites in Dataset 1 had both sufficient osteological and genome-wide data to determine p-values based on the null hypothesis (see Table 3). Nine of the 32 sites in Dataset 2 had sufficient osteological and genetic data for binomial testing (see Table 5). None of the nine sites had significant p-values for both osteological and genetic sex. Interestingly, the two sites that did have significant p-values were also included in Dataset 1: Primrose Grange, and the Knowe of Midhowe.
Four of the 32 sites were either single or double burials, and will be discussed separately, as they were not used for binomial testing but may be socially significant as a representation of a specific burial rite. When the data from all 32 sites were combined, 91/149 (61.1%, p = 0.00427) adult individuals were osteologically determined to be males or possible males (see Table 7). When possible males and females were excluded, the total decreased to 75/127 (59.1%, p = 0.00427).
Out of the 78 individuals with genetic sex data, 60 were male (76.9%, p = 0.00000099). It appears, therefore, that the null hypothesis cannot be rejected unless all sites are collated -only then can an overall possible male sex bias be proposed, based on significance testing. Therefore, there is a degree of regional and site-based variation, either due to sample and/or testing bias or to genuine differences in prehistoric cultural practices, which needs to be explored further.
T A B L E 4 Regional analysis of Dataset 1 sites through binomial p-values and sex ratios; significant values notated with an asterisk (*) T A B L E 6 Regional analysis of Dataset 2 sites through binomial p-values and sex ratios; significant values notated with an asterisk (*) Of the sites able to undergo binomial analyses from both Data- Out of the 32 sites in Dataset 2, 15 had more than 50% osteological male adults represented, including four sites with single or double burials that will be discussed below. However, approximately 36% (106/294) of the total adult skeletal remains in Dataset 2 could not be assigned an osteological sex determination, and there is currently insufficient genetic data to make up this disparity. In order to increase the amount of known data, therefore, more aDNA testing on previously untested and tested sites should be undertaken, as well as a reevaluation of the skeletal remains from these sites, especially those that are commingled. At present, there simply is not enough data to make any firm interpretations regarding sex representation at these sites. There is no trend for male sex bias, as the results differ depending on the dataset and the region analyzed, but there is a tendency for sex-specific bias in the male direction that needs to be investigated further.
This conclusion contrasts the recently published study on Hazleton North, England . From an estimated 41 MNI at the site, approximately 22 of which are adults, genome-wide data was generated for 35 individuals. Surprisingly, 27 of these individuals formed a "virilocal patrilineal descent system" (Fowler, 2022, p. 7), meaning that the individuals were descended from a single male ancestor (NC1M) and the four women with whom he reproduced. The four women may have determined their kin's placement within the chambered tomb (Fowler, 2022;Fowler et al., 2022). The study is an exceptional example of what we should be striving for, as the genetic data for almost an entire site allows for a full contextualization of the archeological data that could previously not be carried out.

| KINSHIP
Patrilineal kinship, where lineage based on male descent is prioritized, has been argued to be a defining parameter of the organization of Neolithic megalithic burials (Cassidy et al., 2020;Fowler et al., 2022). In the two sites with statistical sex bias in the current study, Primrose Grange has perhaps the most evidence for kinship, but that is only for five individuals -a father/daughter pair (Primrose 2 and 17) and a second-degree relationship between Primrose 17 and 18 and between Primrose 6 and 7, both of whom are also second-degree related to a male individual at Carrowmore (Carrowmore 4) (Cassidy et al., 2020;Fowler et al., 2022;Sánchez-Quinto et al., 2019, 9472).
The Carrowmore individual, in turn, is distantly related to individuals at Newgrange (NG10, male), Carrowkeel (CAK533, female; CAK532, male; CAK530, female), and Millin Bay (MB6, male), which together form a distinct clade within passage tomb burials in Ireland, suggesting an identity-by-descent social structure based on biological relationships (Cassidy et al., 2020). This connection between megalithic burials-Primrose Grange, Carrowmore, Newgrange, Carrowkeel, and Millin Bay-albeit through increasingly distant relationships, could hint at a particular family line connected to a single male ancestor, Carrowmore 4. One family line does not, however, provide enough evidence of widespread patrilineal structure that shaped all Neolithic societies across the United Kingdom and Ireland.
Perhaps evidence of a social emphasis on males and male lineages can be found in burials that focus on a sole individual. Single and double burials appear to be exclusively part of a megalithic burial type specific to south-eastern Ireland between 3700 and 3200 cal BCE known as Linkardstown-type cists. Diagnostically, these burials include the deposition of a single individual or small number of individuals in a central stone cist along with a specific set of grave goods including decorated bipartite bowls (Cassidy et al., 2020, p. 24 Linkardstown-type cists are a minority region-specific burial rite, they should be seen as outliers within the megalithic-type burials data set. Interestingly, two individuals, those buried at Jerpoint (JP14) and Baunogenasraid (BG72), were found by Cassidy et al. to share a rare Ychromosome haplogroup (H2a) (Cassidy et al., 2020, SI 27). This does not necessarily indicate patrilineality, but does suggest, along with the predominance of males in the four Linkardstown-type cists in Dataset 2, that this might be a rite that emphasizes the importance of certain male individuals who were buried in certain types of tombs. Further data, however, is required.
A potential example of a similar rite in England is the site of Trumpington Meadows, Cambridgeshire, which was not included in the present study due to typological reasons, as it is technically not a megalith. Two of the three male individuals buried in one of the earth and timber Long Barrows at the site were determined to be brothers who died several years apart, and thus may represent the specific selection of male kin to be buried in a socially significant monument Scheib, personal communication, 2022). This interpretation should be taken with caution, however, since the remains found in the contemporary adjacent monument could not be analyzed osteologically , which results in a lack of contextual data.
Many of these interpretations of genetic kinship are predicated on the idea of a monolithic megalith culture, itself an oversimplification of prehistoric cultures that should be taken with caution. It is entirely plausible that some cultural groups used megalithic funerary structures to bury and/or commemorate specific familial groups or lineages, or entire communities with varying kinship relationships, while others used them for single or pairs of individuals. These structures span 2500 years across a large geographical range with a great degree of morphological differences; variation in burial rite and usage, even between contemporaries, is to be expected, and the relationship between the structures and the people(s) that used them to be dynamic and negotiable. This is not to invalidate interpretations and arguments for patrilineality or patrilocality or even genetic relationships between and within megaliths, but rather to put these arguments into archeological context. There is, therefore, another variable to consider in the analysis of these megalithic burials: time.

| TEMPORAL VARIATION
Not only is there a great variability in the periods these structures were in use, but there is also a large variation in the estimated length of time they were used for. were found to be significant. The regional p-values varied when the multiple test corrections were applied, with an overall trend towards not significant results. On a site-by-site and even regional basis, there is a degree of variation that cannot be overlooked and may hint at either a larger societal explanation or a more familial reflection (see the discussion of Hazleton North by Fowler et al., 2022). The question then becomes how we can further add to the available data in order to build a larger dataset that could better fit statistical parameters of significance testing.

| CONCLUSION
It has been argued that evidence of a bias towards the burial of males in Neolithic megalithic tombs in northern Europe supports a patrilineal system of social organization (Cassidy et al., 2020;Fowler et al., 2022;Sánchez-Quinto et al., 2019). In this paper, we assembled an enlarged database and subjected it to statistical significance testing on an individual, regional, and combined basis, considering both skeletally sexed data and genetically sexed data. Our findings make the argument of patrilineality, or at least a male-dominated society, less certain while not entirely rejecting it. Regionally, outside of the Irish data, there does not appear to be a statistically significant bias in male representation in the sites in Dataset 2, according to binomial p-values. The Irish megalith tradition may represent a specific, local male-centered social structure, one where sex and/or gender along with social status and kinship influences the type of tomb and the funerary rites and traditions that surround it.
The overall percentage of males, genetic and osteological, does appear to outweigh the percentage of females, although these numbers should be viewed with a degree of caution due to missing data, in the form of unsexed and subadult individuals in the osteological data, and small sample sizes in the genetic data. This does not mean that there is no male sex bias, but rather that it cannot be proven given the variable nature of the osteological data as a result of preservation or excavation conditions and the currently small number of genetic samples.
Further research is therefore needed to determine whether there is an actual sex bias present in these and other contemporary burials. Osteological collections need to be revisited and a more efficient, low cost option for genetic analysis pursued so that more data can be collected from archeological remains. Elliott would also like to acknowledge all the help and support given by the aDNA lab at the Institute of Genomics, University of Tartu, and the members of the "Making Ancestors" project.