Genetic and Environmental Influences on Sexual Orientation: Moderation by Childhood Gender Nonconformity and Early-Life Adversity

Abstract

Existing evidence indicates genetic and non-genetic influences on sexual orientation; however, the possibility of gene-environment interplay has not been previously formally tested despite theories indicating this. Using a Finnish twin cohort, this study investigated whether childhood gender nonconformity and early-life adversities independently moderated individual differences in sexual orientation and childhood gender nonconformity, the relationship between them, and the etiological bases of the proposed moderation effects. Sexual orientation, childhood gender nonconformity, and early-life adversities were assessed using standard questionnaires. Structural equation twin model fitting was carried out using OpenMx. Childhood gender nonconformity was significantly associated with reduced phenotypic variance in sexual orientation (β = - 0.14, 95% CI - 0.27, - 0.01). A breakdown of the underlying influences of this moderation effect showed that this was mostly due to moderation of individual-specific environmental influences which significantly decreased as childhood gender nonconformity increased (β_E = - 0.38; 95% CI - 0.52, - 0.001) while additive genetic influences were not significantly moderated (β_A = 0.05; 95% CI - 0.30, 0.27). We also observed that the relationship between sexual orientation and childhood gender nonconformity was stronger at higher levels of childhood gender nonconformity (β = 0.10, 95% CI 0.05, 0.14); however, significance of the underlying genetic and environmental influences on this relationship could not be established in this sample. The findings indicate that beyond a correlation of their genetic and individual-specific environmental influences, childhood gender nonconformity is further significantly associated with reduced individual-specific influences on sexual orientation.

Keywords: Childhood gender nonconformity; Early-life adversity; Moderation; Sexual orientation; Twins.

Fig. 1

Bivariate moderation model with the moderator (childhood gender nonconformity—CGN) specified as a dependent variable. In this model, CGN is modeled as a dependent variable, allowing for its covariance with sexual orientation (SO) to be moderated by itself. A₁₁ and A₂₂ are unique additive genetic influences on CGN and SO, respectively; a₁₁ and a₂₂ denote their respective unmoderated coefficients, a₂₁ denotes the additive genetic coefficient of covariance between CGN and SO; (β₂₂ + M) and (β₂₁ + M) denote the respective moderation terms on the paths where M indicates the possible values of the moderator—CGN. Only Additive genetic (A) effects are depicted in this figure to facilitate clarity, similar parameters are specified for shared (C) and non-shared environmental (E) influences

Fig. 2

Trivariate moderation model with the moderator (early-life adversities—ELA) modeled as a dependent variable. In this model, ELA is modeled as a dependent variable which allows for its covariance with childhood gender nonconformity (CGN) and sexual orientation (SO) to be moderated by itself. A₁₁, A₂₂ and A₃₃ are unique additive genetic influences on ELA, CGN, and SO, respectively; a₁₁, a₂₂ and a₃₃ denote their respective unmoderated coefficients, a₂₁, a₃₁ and a₃₂ denote the additive genetic coefficients of covariances between ELA and CGN and SO, and between CGN and SO, respectively; (β₂₂ + M), (β₃₃ + M), (β₂₁ + M), (β₃₁ + M) and (β₃₂ + M) denote the respective moderated terms on the paths where M indicates the possible values of the moderator—ELA. Only Additive genetic (A) effects are depicted in this figure to facilitate clarity, similar parameters are specified for shared (C) and non-shared environmental (E) influences. The arrangement of the variables is based on ease of analyses rather than any causal inferences and the interpretation of the covariance paths is correlational rather than causal

Fig. 3

Plots showing changes in the means and variances of Childhood Gender Nonconformity (CGN) and Sexual Orientation (SO) and their covariance and correlation across different levels (mean ± 2 standard deviation units) of the moderators—CGN and Early-Life Adversities (ELA). Note: Plots 1, 3, and 2 indicate Genetic (A) and Unique Environmental (E) influences on the variances of Sexual orientation and CGN, and their covariance, respectively, Subscripts _C and _E indicate the specific moderator effect being depicted—CGN and ELA respectively, SD indicates Standard deviation, $\overline{x}$ indicates the mean (± standard deviation), Var Variance, r Correlation coefficient, Cov Covariance. The scales for the plots depicting moderation of A and E influences on SO variance and its covariance with CGN by ELA are halved to allow comparison with plots depicting moderation by CGN while facilitating visibility