The genetics of the timing of sex and reproduction

Genetic variants linked to the timing of first sex and first child, are related to externalising behaviour, reproductive biology and even longevity

Like Comment
Read the paper

Our Nature Human Behaviour article examines the onset of human reproductive behaviour – age at first sexual intercourse (AFS) and age at first birth (AFB) – identifying genetic variants, their biological function and their relationship with other traits and the socio-environment.

For decades, demographers and social scientists demonstrated that the timing of sexual debut and first birth were driven by personal, family and other contextual socio-environmental factors. This includes factors such as an individual’s personality, historical and national context, social norms, access to contraception or the ability to reconcile work and family life.

But there is also a biological and genetic component driving sexual and reproductive behaviour, necessitating not only a social but also a genetic approach, what we term sociogenomics. From twin studies, comparing those with virtually identical genetics (monozygotic) to those genetically akin to siblings (dyzogotic), we first learned that these traits have a heritability (amount we ascribe to genetics) of up to 25%. New statistical techniques soon allowed us to estimate the heritability of traits from unrelated individuals using whole-genome data and isolate the 'ceiling' of what current genetic data could predict, which is around 15% (AFB) and up to 17% for AFS. (For more depth, see my comment on same-sex sexual behaviour in Science, misconceptions of heritability, an exquisite Nature Human Behaviour paper unpacking the basics of genetics and behaviour or our accessible textbook aimed at social scientists as well as our FAQs with this paper.)

Although these previous studies found that there was a genetic basis of reproductive behaviour, they did not isolate the actual genetic variants. In 2016, we published two studies in Nature Genetics finding 10 genetic variants associated with AFB (N=251,151) and 38 with AFS (N=125,667). The current study expands this further and discovers not only a remarkably larger number of variants, but explores the complex interplay of genetics with individual and socio-environmental factors.

As shown below, the age at first sex has become younger, with around one-third of contemporary UK teenagers having sex by age 16. Conversely, there has been a postponement to later age at first birth in many countries to an average age at birth of around 30 years for women, with over 20% of women born after 1970 remaining childless. This means individuals in many countries now start to have children at ages where their fecundity (ability to conceive naturally) decreases, with increases in infertility. 

Age at first sexual intercourse by birth cohort, UKBiobank
Age at first birth by birth cohort, UK Biobank

Sex was previously more linked to marriage meaning that the timing of first sex used to coincide with first birth (correlation ~0.60), which has been decoupled in recent birth cohorts (correlation ~0.30).

Animated correlation plot between age at first birth and age at first sex by birth cohort, UK Biobank
Note: The small dots on the right are births reported before sex or 'immaculate conception.'
Considering that miracle is unlikely in our data, we removed the cases prior to analysis. 

We also uncovered genotypic changes, with heritability in women’s age at first birth shifting from ~10% for those born in 1940 to 23% for the 1965 birth cohort. Demographers such as Udry previously predicted that when social norms become overridden by choice, genetic and biological factors may come to the fore.

Our study is the largest GWAS (genome-wide association study) to date combining 36 datasets for AFS (N=397,338 pooled; N=214,547 women; N=182,791) and AFB (N=542,901 pooled; N=418,758 women; N=124,008 men). A GWAS examines each genetic locus to see if there is an association between a particular genetic variation and our outcomes. Humans are 99.9% identical to each other, and it is the 0.1% by which we differ that makes us all genetically unique.

We discovered a staggering 371 genetic variants of which:  282 were for AFS (272 pooled; 2 women; 8 men only) and 89 for AFB (84 pooled; 1 women; 4 X chromosome only). It is not one variant, but rather the combination of multiple ones that are useful to examine and for this reason, we combined them into what is called a polygenic score. Together genetic variants explained around 5% (AFB) and 6% (AFS) of the variability. Expressed differently, a one standard deviation change in the polygenic score as associated with a 7.3 (AFS) and 6.3 (AFB) month delay. It might at first glance seem small, but this is comparable to the direct effects of adding one strong ‘social’ predictor and we rarely examine one factor in isolation. 

Genetic effects were also sensitive to childhood socioeconomic status. Those growing up in higher socioeconomic households and having the highest polygenic scores (90%+, i.e., a genetic propensity for later age at first sex and birth), had visibly later ages at first sex and birth (AFB shown below). A difference between the highest and lowest socioeconomic decile in this group for AFS, for instance, translated to a 2.08 year delay.

Age at first birth polygenic score (PGS) by percentile groups and parent's education level

Examining the genetic correlations between 25 related traits by sex, we found that the strongest genetic overlaps were logically, for other reproductive traits, followed by educational attainment, externalizing behaviours (ADHD, risk taking), substance use (age at onset smoking) and personality (openness to experience).

Genetic correlations of age at first birth (AFB) and age at first sex (AFS) by related traits.
Horizontal bars are 95% confidence intervals. 

With such high genetic correlations between the GWAS of years of education and age at first birth, (also noted in 2016), the obvious question was: Could there be another trait driving both? Using the relatively new technique of GenomicSEM, we found that the strong genetic correlation was not attributed to a common identified cause driving both and was independent of risk tolerance, substance use and psychiatric disorders – with the exception of a partial mediation by age at smoking. There is a strong bi-directional relationship between the genetics of AFB and education or in other words, the causality goes both ways. Using bi-directional Mendelian Randomisation, we established causality between AFS, AFB, education, risk taking and age at smoking and only age at smoking was upstream in causal terms of AFS. Age at smoking is a unique window into adolescent behaviour. AFS/AFB appear to also capture metabolic effects protective against later life diseases (type 2 diabetes, coronary artery disease), serving as a more powerful predictive of later-life disease.

Testing whether there was a trade-off between the timing of reproduction and longevity, we also found that genetic variants linked with AFB were associated with a reduction in parental mortality. This is reminiscent of the disposable soma theory of aging, which hypothesizes an evolutionary trade-off between investments in somatic maintenance (e.g., remaining in education), which in turn reduces resources available for reproduction. Curious as to whether our genetic signals were from two sub-clusters of reproductive biology versus externalising behaviour, we separated them out. The majority seemed to be related to the genetics of externalising factors (e.g., ADHD, age smoking), followed by reproductive biology (age at menarche, menopause). 

We then wanted to understand the underlying biological function of the variants and undertook a variety of analyses, resulting in prioritisation of 386 unique genes, with genes for AFS related to follicle stimulating hormone, oocyte development,  implantation and placental growth. Those for AFB were linked to the follicle stimulating hormone and the Estrogen receptor. The study also uncovered interesting sex-specific relationships related to infertility, expressed at the protein level for AFS in men (e.g., spermatid differentiation, morphogenesis, binding acrosome-reacted sperm). And for AFB in women, linked to endometriosis and spontaneous abortion. Several genes prioritized in AFS were linked to risk-seeking behaviour, sociability and anxiety.

As with all studies, we have limitations as well. As also with 88% of genetic discoveries, we only included European ancestry populations.  After conducting a scientometric review of all GWAS and realizing 72% of genetic discoveries come from 3 countries, we set up the GWASDiversityMonitor described in our recent Nature Genetics article, so we know that more work is still required. We have also considered problematic biological race and genetic essentialism narratives when conducting these types of studies.

This study offers a unique intersection of discoveries and opens up new terrains of research in the area of reproductive behaviour, infertility, addiction and externalising behaviour. Read also our FAQs, the article and Supplementary Note for more detail.

Melinda C. Mills

Professor, Director , Leverhulme Centre for Demographic Science, University of Oxford