Skip to main content
NCBI home page
Science Advances logoLink to Science Advances
. 2024 Jan 3;10(1):eadj6958. doi: 10.1126/sciadv.adj6958

Genetic variants underlying human bisexual behavior are reproductively advantageous

Siliang Song 1, Jianzhi Zhang 1,*
PMCID: PMC10796114  PMID: 38170769

Abstract

Because human same-sex sexual behavior (SSB) is heritable and leads to fewer offspring, how SSB-associated alleles have persisted and whether they will remain in human populations are of interest. Using the UK Biobank, we address these questions separately for bisexual behavior (BSB) and exclusive SSB (eSSB) after confirming their genetic distinction. We discover that male BSB is genetically positively correlated with the number of offspring. This unexpected phenomenon is attributable to the horizontal pleiotropy of male risk-taking behavior–associated alleles because male risk-taking behavior is genetically positively correlated with both BSB and the number of offspring and because genetically controlling male risk-taking behavior abolishes the genetic correlation between male BSB and the number of offspring. By contrast, eSSB is genetically negatively correlated with the number of offspring. Our results suggest that male BSB–associated alleles are likely reproductively advantageous, which may explain their past persistence and predict their future maintenance, and that eSSB-associated alleles are likely being selected against at present.

Unlike exclusive same-sex sexual behavior, human bisexual behavior is genetically positively correlated with the offspring number.

INTRODUCTION

Same-sex sexual behavior (SSB) is not uncommon in animals and has been observed in mammals, birds, reptiles, amphibians, fish, and invertebrates (1). Across human societies, 2 to 10% of individuals perform SSB (2). Human SSB is heritable (3–7), with a broad-sense heritability of ~30% (2). Using genome-wide association study (GWAS), Ganna et al. (2) discovered five genomic loci that are associated with SSB. Because SSB leads to fewer children (2, 8, 9), it has been a long-standing evolutionary puzzle why natural selection has not eliminated the genetic variants underlying SSB. Many hypotheses have been proposed to explain the genetic persistence of SSB, including, for example, sexual antagonism (10, 11), antagonistic pleiotropy (12, 13), kin selection (14, 15), and overdominance (13, 16). However, empirical evidence supporting any of these hypotheses had been scanty until recently. Specifically, in support of the antagonistic pleiotropy hypothesis, Zietsch et al. (17) detected a positive genetic correlation between SSB and the number of sexual partners among individuals who perform exclusive opposite-sex sexual behavior (OSB). That is, OSB individuals carrying SSB-associated alleles tend to have more sexual partners than OSB individuals not carrying such alleles. Because the number of sexual partners of an OSB individual positively predicts the number of children in premodern societies (18–22), the above finding could, in principle, explain the genetic maintenance of SSB in the past (17). That is, although SSB alleles predispose an individual to SSB, which leads to fewer children, these alleles are associated with more sexual partners and thereby more children among OSB individuals. However, widespread contraception in modern societies decouples the number of children from the number of sexual partners and consequently terminates the above mechanism for the genetic maintenance of SSB (23). SSB is genetically negatively correlated with the number of children in the contemporary UK population (23), suggesting that, at least in this population, SSB as a whole is no longer genetically maintained. Note that, although the reproductive output of a person in a modern society is influenced by many nonbiological factors in addition to biological factors, the mean number of children of a genotype remains a primary determinant of the fitness and evolutionary fate of the genotype.

To investigate the potential genetic maintenance of subtypes of SSB, here, we separately examine bisexual behavior (BSB) and exclusive SSB (eSSB). It is debated whether sexual orientation is continuous (24–27) or discrete (28, 29). The existence of bisexuality was historically controversial because some considered self-reported bisexuals either homosexuals or heterosexuals (30), but recent evidence supports bisexuality as a sexual orientation class distinct from both homosexuality and heterosexuality (31–33). Ganna et al. (2) further suggested genetic discontinuity of SSB (stratified by the proportion of same-sex sexual partners). Furthermore, we investigate sexual behaviors for the two sexes separately because of the known disparities between male and female sexual orientation patterns. For instance, women are more likely to report a bisexual than an exclusively homosexual orientation, while men show the opposite (34, 35). In addition, men’s sexual orientations closely match their sexual arousal patterns, while women’s do not (28). Analyzing the UK Biobank (UKB) (36), which harbors the genotypic and phenotypic information from approximately 0.5 million British participants, we provide evidence that BSB and eSSB are genetically nonidentical, justifying a separate analysis of these two traits. Our analyses suggest that male BSB–associated alleles are reproductively advantageous, whereas eSSB-associated alleles are reproductively disadvantageous. The findings in this paper are meant to add to a better understanding of human sexuality. Because they deal with a specific sample group, they may not be representative of a general pattern across populations with diverse cultural, social, economic, and/or political environments.

RESULTS

BSB and eSSB lead to fewer children

We started by verifying that SSB is associated with fewer children because this condition underlies the paradox of the genetic maintenance of SSB. Ganna et al. (2) previously showed that the number of children of an individual decreases with the individual’s fraction of same-sex sexual partners. Consistently, we found that, among the UKB participants of European ancestry (n = 452,557), a subset of the data used in (2), the mean number of children is lower for eSSB individuals than for BSB individuals, which is, in turn, lower than that for OSB individuals (Fig. 1). We linearly regressed the number of children against BSB and eSSB with sex, age, age2, and the first 10 genetic principal components as covariates. The resulting slopes showed that BSB and eSSB correspond to a reduction of 0.537 (SE = 0.013, P < 10−300) and 1.385 (SE = 0.025, P < 10−300) children, respectively, when compared with OSB individuals, who have 1.86 children on average. The reproductive costs of BSB and eSSB are huge.

Fig. 1. Average numbers of biological children of different groups of individuals in the UKB.

Fig. 1.

Open in a new tab

Only individuals of European ancestry (n = 452,557) are considered here. The error bar represents the standard error of the mean. OSB, individuals performing exclusive opposite-sex sexual behavior; BSB, individuals with bisexual behavior; eSSB, individuals with exclusive same-sex sexual behavior.

BSB and eSSB are genetically nonidentical

There is phenotypic evidence that sexual orientation is discrete (28, 29) and that BSB is distinct from both eSSB and OSB (31–33). A genetic discontinuity of SSB (stratified by the proportion of same-sex sexual partners) was recently suggested (2), but a direct comparison between the genetic architecture of BSB and that of eSSB has not been performed. To this end, we calculated the genetic correlation between BSB and eSSB using linkage disequilibrium score regression (LDSC) (37, 38). We found that this genetic correlation is significantly lower than 1 (rg = 0.281, SE = 0.173, P = 3 × 10−5; table S1) and is not significantly different from 0 (P > 0.1; table S1) despite the presence of significant heritability of both BSB and eSSB (table S2). Similar results were obtained when the two sexes were separately examined (table S1). More rigorously, as a control, we randomly assigned the SSB individuals to either BSB or eSSB under the constraint of the original numbers of BSB and eSSB individuals and then computed the genetic correlation between pseudo-BSB and pseudo-eSSB. The above analysis was repeated 20 times, yielding a mean genetic correlation of 1.04 and an SD of 0.41. All 20 of these genetic correlations are greater than that between actual BSB and eSSB (data S1), suggesting that BSB and eSSB are genetically nonidentical or distinct (P < 0.05). These findings justify the separation of BSB from eSSB in subsequent analysis.

Male BSB is genetically positively correlated with the number of children

We considered two BSB traits, male BSB and female BSB, and computed their respective genetic correlation (rg) with the number of children. We found rg to be significantly positive for male BSB (rg = 0.156, SE = 0.067, P = 0.019) but not significant for female BSB (Table 1). The above significant correlation remains significant after the consideration of multiple testing (adjusted P = 0.038). We also considered general BSB without specifying the sex but did not find it to have a significant rg with the number of children (Table 1).

Table 1. Genetic correlation (rg) between BSB and the number of children.

Male BSB Female BSB General BSB
rg (SE) P value rg (SE) P value rg (SE) P value
Number of children 0.156 (0.067) 0.019 −0.013 (0.068) 0.855 0.070 (0.057) 0.218
Number of children
among individuals
with children		 0.258 (0.085) 0.002 0.195 (0.091) 0.032 0.249 (0.073) 0.0006
Open in a new tab

Given the above findings, we further asked whether BSB-associated alleles predict childlessness, offspring number when there is at least one child, or both. We did not observe a significant genetic correlation between male BSB and childlessness but observed a significant positive genetic association between female BSB and childlessness that would become only marginally significant (adjusted P = 0.05) upon the control of multiple testing (table S3). When examining those with at least one child, we again observed a significant positive rg between male BSB and the number of children (rg = 0.258, SE = 0.085, P = 2 × 10−3; Table 1). This analysis also revealed a significant positive rg between female BSB and the number of children (rg = 0.195, SE = 0.091, P = 0.032; Table 1), but this significance disappeared upon the control of multiple testing (adjusted P = 0.064). We further observed a significant positive rg between general BSB and the number of children in this analysis (rg = 0.249, SE = 0.073, P = 6 × 10−4; Table 1). Hence, male BSB–associated alleles do not predict childlessness but predict more children when there is at least one child. By contrast, female BSB–associated alleles marginally predict childlessness and more children when there is at least one child, and the two potential influences on the reproductive output are more or less canceled out.

Alleles associated with risk-taking behavior underlie the positive genetic correlation between male BSB and the number of children

Given that BSB is associated with fewer children (Fig. 1), the above finding of the reproductive advantage of male BSB–associated alleles is unexpected. We therefore investigated the underlying mechanism of this phenomenon. Consistent with the previous report that SSB is genetically positively correlated with the number of sexual partners among OSB individuals (17), we found a strong, positive genetic correlation between BSB and the number of sexual partners among OSB individuals (rg = 0.872, SE = 0.156, P = 2.5 × 10−8; table S4). However, as mentioned, in modern societies, the number of sexual partners does not positively predict the number of children phenotypically or genetically (23). Consequently, the positive association with the number of sexual partners in OSB individuals alone may not confer reproductive advantages to BSB-associated alleles.

SSB is known to be genetically positively correlated with (self-reported) risk-taking behavior (17). We confirmed that male BSB and male risk-taking behavior are also positively genetically correlated (rg = 0.484, SE = 0.085, P = 1.4 × 10−8; Fig. 2A and table S5). In addition, male risk-taking behavior is both genetically (rg = 0.366, SE = 0.037, P = 4.2 × 10−23; Fig. 2A) and phenotypically (β = 0.182, SE = 0.007, P = 7.2 × 10−163) positively correlated with the number of children, while the same correlation for female risk-taking behavior is much weaker genetically (rg = 0.144, SE = 0.038, P = 0.0001) and even negative phenotypically (β = −0.045, SE = 0.007, P = 1.6 × 10−10; table S6). These results coincide with the pattern of genetic correlation between male/female BSB and the number of children (Table 1), suggesting the possibility that the positive genetic correlation between male BSB and the number of children is due to the horizontal pleiotropy of alleles associated with male risk-taking behavior (i.e., the genetic predisposition to male risk-taking behavior positively influences both BSB and the number of children through distinct pathways).

Fig. 2. Genomic SEM analysis shows that the positive genetic correlation between male BSB and the number of children can be explained by the horizontal pleiotropy of alleles associated with male risk-taking behavior.

Fig. 2.

Open in a new tab

(A) Raw genetic correlations calculated using LDSC. Pairwise genetic correlations (with SE in parentheses) and their P values are shown near each edge. (B) Partial genetic correlations. Standardized partial genetic correlations (with SE in parentheses) and their P values are shown near each edge. The value near the circular arrow represents the fraction of genetic variance of the trait that is not involved in its genetic correlation with male risk-taking behavior.

To assess the extent of the positive genetic correlation between male BSB and the number of children explainable by the above proposed horizontal pleiotropy, we conducted genomic structural equation modeling (genomic SEM) (39) to attain a “partial” genetic correlation between male BSB and the number of children while controlling the horizontal pleiotropic effects of alleles associated with male risk-taking behavior (see Materials and Methods). Specifically, we postulated that the number of children is potentially genetically affected by both male BSB and male risk-taking behavior, whereas male BSB is potentially genetically affected by male risk-taking behavior (Fig. 2B). The resulting partial genetic correlation between male BSB and the number of children becomes slightly but not significantly negative (rg = −0.024, SE = 0.082, P = 0.77; Fig. 2B and table S7). To confirm the above result, we also calculated LDSC-based genetic correlation between male BSB and the number of children while controlling risk-taking behavior in their GWAS. That is, we added risk-taking behavior as a covariate when conducting GWAS for male BSB and GWAS for the number of children and then used the resulting summary statistics to calculate the genetic correlation between male BSB and the number of children. As expected, the genetic correlation is no longer significant upon this control (rg = 0.11, SE = 0.07, P = 0.11). Therefore, the positive genetic correlation between male BSB and the number of children can be explained by the horizontal pleiotropic effects of the genetic variants underlying male risk-taking behavior.

eSSB is genetically negatively correlated with the number of children

In contrast to BSB, eSSB shows a negative genetic correlation with the number of children (rg = −0.404, SE = 0.132, P = 0.0022), which is primarily driven by the negative genetic correlation between male eSSB and the number of children (rg = −0.382, SE = 0.109, P = 0.0004) because female eSSB and the number of children do not show a significant genetic correlation (Table 2). Similar results were obtained among individuals with children (Table 2). Note that BSB and eSSB are genetically correlated with the number of children in opposite directions, further supporting the genetic distinction between BSB and eSSB. In addition, unlike BSB, eSSB is not genetically correlated with the number of sexual partners among OSB individuals (table S4).

Table 2. Genetic correlation (rg) between eSSB and the number of children.

Male eSSB Female eSSB General eSSB
rg (SE) P value rg (SE) P value rg (SE) P value
Number of children −0.382 (0.109) 0.0004 −0.068 (0.118) 0.56 −0.404 (0.132) 0.002
Number of children
among individuals
with children		 −0.268 (0.133) 0.043 0.083 (0.157) 0.60 −0.233 (0.149) 0.12
Open in a new tab

DISCUSSION

In this study, we provided statistical evidence that BSB and eSSB are genetically nonidentical, which prompted separate investigations of their (potential) genetic maintenances. We discovered that male BSB is genetically positively correlated with the number of children, which is explainable by the horizontal pleiotropy of the genetic variants underlying male risk-taking behavior. These findings suggest a scenario in which the current genetic maintenance of male BSB is a by-product of selection for male risk-taking behavior. If the mechanism discovered here from the UKB also worked in the past, then it could explain why male BSB has been genetically persistent. We note that while risk-taking behaviors in modern societies may differ from risk-taking behaviors in ancient times, the biological basis and thereby the genetic underpinning of potentially different risk-taking behaviors of different eras could be similar.

It is worth emphasizing that the positive genetic correlation between male BSB and the number of children does not mean that male BSB is positively selected because fitness has multiple components and the number of children is but one of them. For example, fitness is also influenced by viability and reproductive ages, which are not fully reflected in the number of children (e.g., fitness is drastically different with one child when the parent is 20 years old versus 30 years old at the birth of the child). UKB participants were recruited at the age of 40 to 69, so UKB data do not allow assessing genetic variants underlying viability up to the age of 40. Because SSB is known to be associated with increased mortality (40–42), it is possible that the reproductive benefits of SSB-associated alleles are offset by their potential costs to viability such that these alleles are neither favored nor disfavored by natural selection.

We discovered that eSSB is genetically negatively correlated with the number of children. This observation, coupled with the increased mortality associated with SSB (40–42), suggests that eSSB is likely under negative selection currently, unless the genetic variants underlying eSSB have unexpected pleiotropic effects that strongly improve other fitness components, which seems unlikely. Therefore, the heritability of eSSB may be a legacy from past genetic maintenance. We note, however, that the genetic correlation between eSSB and the number of sexual partners among OSB individuals is not significant (table S4), so the mechanism of the past genetic maintenance of eSSB is unknown. That said, Zietsch et al. (17) reported a significant, positive genetic correlation between eSSB and the number of sexual partners among OSB individuals. The inconsistency between their result and ours may be due to the use of different methods because Zietsch et al. inferred polygenic scores of eSSB and then calculated the genetic correlation, while we used LDSC to estimate the genetic correlation. Further studies are needed to clarify the genetic correlation between eSSB and the number of sexual partners among OSB individuals.

Because the variance of fitness is usually greater in males than in females of the same population (43), as is the case in the UKB in terms of the variance of the number of children (P < 2.6 × 10−308, F test), environmental and genetic factors likely have larger fitness effects in males than in females (44–46). Hence, it is expected that significant genetic correlations are often observed for males but not females in our study. For instance, only male BSB and male eSSB are genetically correlated with the number of children. Given the typical roles and strategies of the two sexes in reproduction, it is probable that the predisposition to risk-taking behavior increases the number of children for males more than that for females (table S6). Regarding eSSB, females generally have more opportunities to have (known) offspring than males via in vitro fertilization by donor sperm/eggs. Therefore, eSSB might cause a greater reduction of (known) reproduction in males than in females. The UKB data showed that the mean number of children of eSSB males is lower than that of eSSB females, while no such trend is observed among BSB or OSB individuals (Fig. 1).

Cross-trait assortative mating can create a genetic correlation between two traits that is not due to shared or linked genetic variants of the traits (47). However, to explain the positive genetic correlation between male BSB and the number of children by cross-trait assortative mating, one must assume preferential mating between BSB individuals and people of high fecundity, which has no empirical support and seems highly unlikely. Cross-trait assortative mating may partially explain the positive genetic correlation between risk-taking behavior and BSB and the positive genetic correlation between risk-taking behavior and the number of children, but the genomic SEM analysis is still valid regardless of whether the genetic correlation is caused by shared/linked genetic variants or cross-trait assortative mating. In other words, even if traits A and B are genetically correlated entirely because of assortative mating and A and C are also genetically correlated entirely because of assortative mating, B and C are still expected to be genetically correlated.

We note that the phenotypes analyzed here are self-reported, so they may contain substantial errors. Such phenotypic errors, when they are random, reduce the magnitude of estimated genetic correlations, rendering our conclusions conservative. However, if such errors are nonrandom, then they could create spurious genetic correlations. For example, if individuals carrying BSB-associated alleles tend to overreport their number of children, then the true genetic correlation may be between BSB and reporting more children rather than that between BSB and having more children. Such biased reporting has no empirical evidence and seems unlikely.

Note that, under our definition, a person must have at least two sexual partners to be identified as a BSB individual, while they need only one sexual partner to be identified as an OSB or eSSB individual. Therefore, under our definition, an average BSB individual has more sexual partners than does an average OSB or eSSB individual. However, because the number of sexual partners does not positively predict the number of children phenotypically or genetically (23), this potential bias cannot create a positive genetic correlation between BSB and the number of children. In addition, an exclusively homosexual individual may engage in sexual relationships with the opposite sex because of social pressures and thereby be identified as a BSB individual. The contrary is expected to be rarer. Regardless, given that eSSB is found to be genetically negatively correlated with the number of children, the potential misclassifications make the conclusion of the positive genetic correlation between BSB and the number of children conservative.

It is important to emphasize that our analysis of reproductive success and inference of selection are at the genetic level. That is, although eSSB-associated alleles are presently under negative selection, the fraction of eSSB individuals in the general population may not decrease over time because eSSB is determined by both genetic and environmental factors. The proportions of European ancestry participants in the UKB reporting BSB and eSSB both generally increased with the birth year (fig. S1), likely attributable to a growing openness of the society toward SSB (48) that could increase the probability of performing and/or the probability of reporting SSB. Therefore, eSSB may become more prevalent in the future even when eSSB-associated alleles are being selected against. It is also important to recognize that the present study is based on the British people of European ancestry; consequently, our results may or may not represent a general pattern across populations with diverse cultural, social, economic, and/or political environments.

The topic explored in this study intersects with sexuality and identity and potentially bears civil and political ramifications for sexual minority groups. We want to make it clear that our results predominantly contribute to the diversity, richness, and better understanding of human sexuality. They are not, in any way, intended to suggest or endorse discrimination on the basis of sexual behavior.

MATERIALS AND METHODS

UKB sample screening

The use of the UKB data in the present study was approved by the UKB (reference no. 48678). We generally followed the sample screening process previously described (2). To avoid spurious results caused by population stratification, we focused on UKB participants with European ancestry. We applied the K-means clustering algorithm on the first four genetic principal components precalculated by the UKB. The resulting four clusters were then visually inspected to identify the cluster corresponding to the European ancestry (all individuals in the cluster were labeled “white British”). We further excluded individuals who do not self-report as “white” (data-field 21000) and excluded those whose genotype missing rate is greater than 0.02. The resulting 452,557 individuals were used for GWAS (see data S2 for details).

UKB variant screening

We generally followed the variant screening process previously described (2). We excluded single-nucleotide polymorphisms (SNPs) that satisfy any of the following conditions: (i) Minor allele frequency is below 1%. (ii) Imputation quality is low (INFO <0.8). (iii) Genotyping missing rate exceeds 1%. (iv) Hardy-Weinberg equilibrium is rejected at P < 10−10. After the screening, 9,371,426 SNPs, including those on the X chromosome, were used in the analysis.

UKB phenotypes

The phenotypic information of SSB was obtained from the answers to the UKB question “Have you ever had sexual intercourse with someone of the same sex?” (data-field 2159); individuals who answered “yes” were coded 1 and “no” were coded 0. Those coded 1 for SSB were further divided into BSB or eSSB individuals according to their answers to the UKB questions “How many sexual partners of the same sex have you had in your lifetime?” (data-field 3669) and the question “About how many sexual partners have you had in your lifetime?” (data-field 2149). If an SSB individual had fewer same-sex sexual partners than sexual partners, then the individual was coded as 1 for BSB; otherwise, the person was coded 0 for BSB. If an SSB individual had equal numbers of same-sex sexual partners and sexual partners, then the individual was coded 1 for eSSB; otherwise, the person was coded 0 for eSSB.

The number of children of a UKB participant was obtained from the answers to the UKB questions “How many children have you fathered?” (data-field 2405) for males and “How many children have you given birth to? (Please include live births only)” (data-field 2734) for females.

Information of risk-taking behavior was obtained from the answers to the UKB question “Would you describe yourself as someone who takes risks?” (data-field 2040). Individuals who answered “yes” were coded 1, whereas those who answered “no” were coded 0. Individuals who answered “do not know” or refused to answer a question were excluded from the analysis on the relevant phenotype.

Genome-wide association analysis

We applied the state-of-the-art GWAS software, REGENIE (49), to conduct GWAS. REGENIE analysis consists of two steps. The first step fits a whole-genome regression model for trait predictions based on high-quality genotyped SNPs using the leave-one-chromosome-out scheme. The phenotypic predictions attained from the first step are used as offsets in the second step, which performs GWAS for all SNPs (both genotyped and imputed SNPs) using standard linear regression. For the first step, we used all genotyped SNPs (n = 595,864) simultaneously satisfying (i) minor allele frequency > 1%, (ii) minor allele count > 100, (iii) genotyping missing rate < 1%, and (iv) P > 10−10 in the test of the Hardy-Weinberg equilibrium. For the second step, we used previously screened European ancestry individuals (n = 452,557) and both genotyped and imputed SNPs (n = 9,371,426). The same covariates, including age, age2, genetic sex, and the first 10 genetic principal components, were used in regression models for both steps.

Genetic correlation

We used cross-trait LDSC based on GWAS summary statistics to estimate the genetic correlation between traits (37). Briefly, the genetic correlation is estimated by calculating the slope of the regression where the products of z-scores resulting from GWAS of two traits are regressed upon the LD score. We used LDSC (38) and followed the instructions on the GitHub website (https://github.com/bulik/ldsc/tree/v1.0.0; version 1.0.0, last accessed on 25 May 2023) to calculate genetic correlations.

Heritability estimation

To estimate the family-based heritability of SSB, BSB, and eSSB, we need to identify pairs of individuals with close kinship in the UKB. For example, first-degree relatives (e.g., mother and son) have a coefficient of relatedness of 0.5. Second-degree relatives (e.g., half siblings) have a coefficient of relatedness of 0.25, and third-degree relatives (e.g., first cousins) have a coefficient of relatedness of 0.125. UKB provides kinship coefficients and fractions of markers for which the pair shares zero alleles (IBS0), based on which we identified twins (N = 169) and the first-degree (N = 27,613), second-degree (N = 9622), and third-degree (N = 64,944) relatives among previously screened European ancestry individuals. Following the method previously described (2), we calculated the family-based heritability by modeling the additive genetic component and the shared and unshared environmental components of the phenotypic variance on the basis of different covariances between pairs with different degrees of relatedness (50). SNP-based heritability in observed or liability scale was estimated using LDSC (38).

Acknowledgments

We thank E. Long and D. Jiang for valuable comments. This research has been conducted using the UK Biobank Resource under Application Number 48678.

Funding: This work was supported by the NIH (grant R35GM139484 to J.Z.).

Author contributions: S.S. and J.Z. designed the study and wrote the paper. S.S. performed the analysis.

Competing interests: The authors declare that they have no competing interests.

Data and materials availability: All data and computer code needed to evaluate the conclusions in the paper are present in the paper, the Supplementary Materials, and/or Dryad (https://doi.org/10.5061/dryad.4b8gthtk9).

Supplementary Materials

This PDF file includes:

Fig. S1

Tables S1 to S7

Legends for data S1 and S2

Click here for additional data file. (520.7KB, pdf)

Other Supplementary Material for this manuscript includes the following:

Data S1 and S2

Click here for additional data file. (17.6KB, zip)

REFERENCES AND NOTES

  • 1.Bailey N. W., Zuk M., Same-sex sexual behavior and evolution. Trends Ecol. Evol. 24, 439–446 (2009). [DOI] [PubMed] [Google Scholar]
  • 2.Ganna A., Verweij K. J. H., Nivard M. G., Maier R., Wedow R., Busch A. S., Abdellaoui A., Guo S., Sathirapongsasuti J. F., Research T. M., Lichtenstein P., Lundstrom S., Langstrom N., Auton A., Harris K. M., Beecham G. W., Martin E. R., Sanders A. R., Perry J. R. B., Neale B. M., Zietsch B. P., Large-scale GWAS reveals insights into the genetic architecture of same-sex sexual behavior. Science 365, eaat7693 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Pillard R. C., Bailey J. M., Human sexual orientation has a heritable component. Hum. Biol. 70, 347–365 (1998). [PubMed] [Google Scholar]
  • 4.Langstrom N., Rahman Q., Carlstrom E., Lichtenstein P., Genetic and environmental effects on same-sex sexual behavior: A population study of twins in Sweden. Arch. Sex. Behav. 39, 75–80 (2010). [DOI] [PubMed] [Google Scholar]
  • 5.Hamer D. H., Hu S., Magnuson V. L., Hu N., Pattatucci A. M., A linkage between DNA markers on the X chromosome and male sexual orientation. Science 261, 321–327 (1993). [DOI] [PubMed] [Google Scholar]
  • 6.Sanders A. R., Beecham G. W., Guo S., Dawood K., Rieger G., Badner J. A., Gershon E. S., Krishnappa R. S., Kolundzija A. B., Duan J.; M. G. S. Collaboration, Gejman P. V., Bailey J. M., Martin E. R., Genome-wide association study of male sexual orientation. Sci. Rep. 7, 16950 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Mustanski B. S., DuPree M. G., Nievergelt C. M., Bocklandt S., Schork N. J., Hamer D. H., A genomewide scan of male sexual orientation. Hum. Genet. 116, 272–278 (2005). [DOI] [PubMed] [Google Scholar]
  • 8.Apostolou M., The direct reproductive cost of same-sex attraction: Evidence from two nationally representative U.S. samples. Arch. Sex. Behav. 51, 1857–1864 (2022). [DOI] [PubMed] [Google Scholar]
  • 9.Wells J. E., McGee M. A., Beautrais A. L., Multiple aspects of sexual orientation: Prevalence and sociodemographic correlates in a New Zealand national survey. Arch. Sex. Behav. 40, 155–168 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Camperio Ciani A., Cermelli P., Zanzotto G., Sexually antagonistic selection in human male homosexuality. PLOS ONE 3, e2282 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Camperio Ciani A., Battaglia U., Zanzotto G., Human homosexuality: A paradigmatic arena for sexually antagonistic selection? Cold Spring Harb. Perspect. Biol. 7, a017657 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Zietsch B., Morley K., Shekar S., Verweij K., Keller M., Macgregor S., Wright M., Bailey J., Martin N., Genetic factors predisposing to homosexuality may increase mating success in heterosexuals☆. Evol Hum Behav 29, 424–433 (2008). [Google Scholar]
  • 13.Miller E. M., Homosexuality, birth order, and evolution: Toward an equilibrium reproductive economics of homosexuality. Arch. Sex. Behav. 29, 1–34 (2000). [DOI] [PubMed] [Google Scholar]
  • 14.Bobrow D., Bailey J. M., Is male homosexuality maintained via kin selection? Evol Hum Behav 22, 361–368 (2001). [Google Scholar]
  • 15.Rahman Q., Hull M. S., An empirical test of the kin selection hypothesis for male homosexuality. Arch. Sex. Behav. 34, 461–467 (2005). [DOI] [PubMed] [Google Scholar]
  • 16.MacIntyre F., Estep K. W., Sperm competition and the persistence of genes for male homosexuality. Biosystems 31, 223–233 (1993). [DOI] [PubMed] [Google Scholar]
  • 17.Zietsch B. P., Sidari M. J., Abdellaoui A., Maier R., Langstrom N., Guo S., Beecham G. W., Martin E. R., Sanders A. R., Verweij K. J. H., Genomic evidence consistent with antagonistic pleiotropy may help explain the evolutionary maintenance of same-sex sexual behaviour in humans. Nat. Hum. Behav. 5, 1251–1258 (2021). [DOI] [PubMed] [Google Scholar]
  • 18.Cronk L., Wealth, status, and reproductive success among the Mukogodo of Kenya. Am. Anthropol. 93, 345–360 (1991). [Google Scholar]
  • 19.von Rueden C. R., Jaeggi A. V., Men's status and reproductive success in 33 nonindustrial societies: Effects of subsistence, marriage system, and reproductive strategy. Proc. Natl. Acad. Sci. U.S.A. 113, 10824–10829 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Gibson M. A., Mace R., Polygyny, reproductive success and child health in rural Ethiopia: Why marry a married man? J. Biosoc. Sci. 39, 287–300 (2007). [DOI] [PubMed] [Google Scholar]
  • 21.Fromonteil S., Marie-Orleach L., Winkler L., Janicke T., Sexual selection in females and the evolution of polyandry. PLoS Biol. 21, e3001916 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Scelza B. A., Female choice and extra-pair paternity in a traditional human population. Biol. Lett. 7, 889–891 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Song S., Zhang J., Contraception ends the genetic maintenance of human same-sex sexual behavior. Proc. Natl. Acad. Sci. U.S.A. 120, e2303418120 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Haslam N., Evidence that male sexual orientation is a matter of degree. J. Pers. Soc. Psychol. 73, 862–870 (1997). [DOI] [PubMed] [Google Scholar]
  • 25.Kinsey A. C., Pomeroy W. B., Martin C. E., Sexual behavior in the human male. J. Nerv. Ment. Dis. 109, 283 (1949). [Google Scholar]
  • 26.Savin-Williams R. C., An exploratory study of the categorical versus spectrum nature of sexual orientation. J. Sex Res. 51, 446–453 (2014). [DOI] [PubMed] [Google Scholar]
  • 27.Savin-Williams R. C., Sexual orientation: Categories or continuum? Commentary on Bailey et al. (2016). Psychol. Sci. Public Interest 17, 37–44 (2016). [DOI] [PubMed] [Google Scholar]
  • 28.Bailey J. M., Vasey P. L., Diamond L. M., Breedlove S. M., Vilain E., Epprecht M., Sexual orientation, controversy, and science. Psychol. Sci. Public Interest 17, 45–101 (2016). [DOI] [PubMed] [Google Scholar]
  • 29.Norris A. L., Marcus D. K., Green B. A., Homosexuality as a discrete class. Psychol. Sci. 26, 1843–1853 (2015). [DOI] [PubMed] [Google Scholar]
  • 30.Rieger G., Chivers M. L., Bailey J. M., Sexual arousal patterns of bisexual men. Psychol. Sci. 16, 579–584 (2005). [DOI] [PubMed] [Google Scholar]
  • 31.Safron A., Sylva D., Klimaj V., Rosenthal A. M., Li M., Walter M., Bailey J. M., Neural correlates of sexual orientation in heterosexual, bisexual, and homosexual men. Sci. Rep. 7, 41314 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Rullo J. E., Strassberg D. S., Miner M. H., Gender-specificity in sexual interest in bisexual men and women. Arch. Sex. Behav. 44, 1449–1457 (2015). [DOI] [PubMed] [Google Scholar]
  • 33.Jabbour J., Holmes L., Sylva D., Hsu K. J., Semon T. L., Rosenthal A. M., Safron A., Slettevold E., Watts-Overall T. M., Savin-Williams R. C., Sylla J., Rieger G., Bailey J. M., Robust evidence for bisexual orientation among men. Proc. Natl. Acad. Sci. U.S.A. 117, 18369–18377 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.A. Sharfman, P. Cobb, Sexual Orientation, UK: 2021 and 2022 (Office for National Statistics, 2023); www.ons.gov.uk/peoplepopulationandcommunity/culturalidentity/sexuality/bulletins/sexualidentityuk/2021and2022.
  • 35.G. J. Gates, How Many People Are Lesbian, Gay, Bisexual, and Transgender? (The Williams Institute, UCLA, 2011); https://escholarship.org/uc/item/09h684x2.
  • 36.Bycroft C., Freeman C., Petkova D., Band G., Elliott L. T., Sharp K., Motyer A., Vukcevic D., Delaneau O., O'Connell J., Cortes A., Welsh S., Young A., Effingham M., McVean G., Leslie S., Allen N., Donnelly P., Marchini J., The UK Biobank resource with deep phenotyping and genomic data. Nature 562, 203–209 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Bulik-Sullivan B., Finucane H. K., Anttila V., Gusev A., Day F. R., Loh P. R.; ReproGen Consortium; Psychiatric Genomics Consortium; Genetic Consortium for Anorexia Nervosa of the Wellcome Trust Case Control Consortium 3, Duncan L., Perry J. R., Patterson N., Robinson E. B., Daly M. J., Price A. L., Neale B. M., An atlas of genetic correlations across human diseases and traits. Nat. Genet. 47, 1236–1241 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Bulik-Sullivan B. K., Loh P. R., Finucane H. K., Ripke S., Yang J.; Schizophrenia Working Group of the Psychiatric Genomics Consortium, Patterson N., Daly M. J., Price A. L., Neale B. M., LD score regression distinguishes confounding from polygenicity in genome-wide association studies. Nat. Genet. 47, 291–295 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Grotzinger A. D., Rhemtulla M., de Vlaming R., Ritchie S. J., Mallard T. T., Hill W. D., Ip H. F., Marioni R. E., McIntosh A. M., Deary I. J., Koellinger P. D., Harden K. P., Nivard M. G., Tucker-Drob E. M., Genomic structural equation modelling provides insights into the multivariate genetic architecture of complex traits. Nat. Hum. Behav. 3, 513–525 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Erlangsen A., Drefahl S., Haas A., Bjorkenstam C., Nordentoft M., Andersson G., Suicide among persons who entered same-sex and opposite-sex marriage in Denmark and Sweden, 1989-2016: A binational, register-based cohort study. J. Epidemiol. Community Health 74, 78–83 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.R. Graham, B. Berkowitz, R. Blum, W. Bockting, J. Bradford, B. de Vries, H. Makadon, The Health of Lesbian, Gay, Bisexual, and Transgender People: Building a Foundation for Better Understanding (Institute of Medicine, 2011). [PubMed] [Google Scholar]
  • 42.Hottes T. S., Ferlatte O., Gesink D., Suicide and HIV as leading causes of death among gay and bisexual men: A comparison of estimated mortality and published research. Crit. Public Health 25, 513–526 (2014). [Google Scholar]
  • 43.T. H. Clutton-Brock, Reproductive Success: Studies of Individual Variation in Contrasting Breeding Systems (University of Chicago Press, 1988). [Google Scholar]
  • 44.Wade M. J., Sexual selection and variance in reproductive success. Am. Nat. 114, 742–747 (1979). [DOI] [PubMed] [Google Scholar]
  • 45.Wade M. J., Arnold S. J., The intensity of sexual selection in relation to male sexual behaviour, female choice, and sperm precedence. Anim. Behav. 28, 446–461 (1980). [Google Scholar]
  • 46.Wyman M. J., Rowe L., Male bias in distributions of additive genetic, residual, and phenotypic variances of shared traits. Am. Nat. 184, 326–337 (2014). [DOI] [PubMed] [Google Scholar]
  • 47.Border R., Athanasiadis G., Buil A., Schork A. J., Cai N., Young A. I., Werge T., Flint J., Kendler K. S., Sankararaman S., Dahl A. W., Zaitlen N. A., Cross-trait assortative mating is widespread and inflates genetic correlation estimates. Science 378, 754–761 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.T. W. Smith, J. Son, J. Kim, Public Attitudes Toward Homosexuality and Gay Rights Across Time and Countries (The Williams Institute, 2014), pp. 1–29. [Google Scholar]
  • 49.Mbatchou J., Barnard L., Backman J., Marcketta A., Kosmicki J. A., Ziyatdinov A., Benner C., O'Dushlaine C., Barber M., Boutkov B., Habegger L., Ferreira M., Baras A., Reid J., Abecasis G., Maxwell E., Marchini J., Computationally efficient whole-genome regression for quantitative and binary traits. Nat. Genet. 53, 1097–1103 (2021). [DOI] [PubMed] [Google Scholar]
  • 50.M. Lynch, B. Walsh, Genetics and Analysis of Quantitative Traits (Sinauer Sunderland, 1998), vol. 1. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Fig. S1

Tables S1 to S7

Legends for data S1 and S2

Click here for additional data file. (520.7KB, pdf)

Data S1 and S2

Click here for additional data file. (17.6KB, zip)

Articles from Science Advances are provided here courtesy of American Association for the Advancement of Science

RESOURCES