Similarity between partners leaves a detectable genetic signature that persists from generation to generation, according to a new study in Behavior Genetics. An analysis of large samples from the United States and the United Kingdom identified such traces for education, height and smoking. In the American sample, some were stronger among people born more recently, although the interpretation of this change remains cautious.
Assortative mating describes the formation of couples that is not random with respect to traits such as education or everyday habits. When these traits have a hereditary component, having children can, over successive generations, create statistical correlations between genetic variants even on different chromosomes. This does not require a conscious choice based on the particular trait: schools, workplaces and social networks already influence which people meet.
Gretchen R. B. Saunders, from the University of Minnesota, used data from the UK Biobank and the All of Us Research Program. The analysis included 82,398 likely opposite-sex married or long-term cohabiting couples and 16,273 pairs of full siblings from the British database. It also examined 333,491 unrelated people from the UK Biobank and 228,875 from All of Us. All samples were restricted to people of European genetic ancestry because the available genetic models had been developed mainly in corresponding populations.
The research examined eight characteristics: educational attainment, height, body mass index and five behaviors related to smoking and alcohol. These covered a history of regular smoking, age at initiation, the maximum daily number of cigarettes, smoking cessation and weekly alcohol consumption. Polygenic scores were calculated for each participant: statistical measures that combine many small genetic variants. The researcher compared both the traits themselves and these scores between partners, while comparing only polygenic scores between siblings.
Partners showed positive similarity across all eight characteristics, with the strongest correlations for education, at 0.45, and weekly alcohol consumption, at 0.42. On this scale, zero indicates no correlation and one a perfect match. People with higher educational attainment tended to have partners who were taller, had a lower body mass index and were less likely to smoke. Similarly, heavier smoking or alcohol consumption was associated with similar habits and a higher body mass index in the partner.
Similarities in polygenic scores generally followed the same pattern but were noticeably smaller: the correlation between partners reached 0.16 for education, 0.08 for height and 0.06 for smoking initiation. On average, genetic correlations were about 80% smaller than correlations for the traits themselves. Part of the difference is explained by the fact that the scores capture only a portion of genetic influences. For habits that can change within a relationship, the difference exceeded 90%, which Saunders interprets as evidence of shared living conditions and mutual influence.
To look for traces of previous generations, the study examined whether genetic similarities between full siblings exceeded the expected level of 0.50 under random mating, as they did for most traits. Among unrelated people, it compared genetic measures calculated separately from even- and odd-numbered chromosomes. In both databases, the correlation was about 0.07 for education and 0.04 for height and smoking initiation, instead of the expected zero. This method detects accumulated historical patterns without directly observing the formation of couples.
In the UK Biobank, whose participants were born between 1936 and 1970, these measures remained mostly stable, with an increase only for smoking cessation. In All of Us, which covered births from 1900 to 2005, increases appeared for education, smoking initiation and smoking cessation. The genetic signature for education rose from about 0.04 among those born in the 1930s to 0.09 among those born around 2000. Saunders notes that the broader age range of the American sample may make small changes easier to detect.
Norbert Meskó, a professor at the University of Pécs who was not involved in the study, finds the general evidence of similarity between partners more convincing than the evidence that it has strengthened over time. As he explains, differences in participation, survival and population composition can affect comparisons between birth cohorts. In addition, couples in the UK Biobank were identified through shared addresses and household characteristics, with the possibility of minor classification errors. The results also cannot automatically be generalized to other ancestries or cultural settings.
The findings do not mean that genes determine whom we fall in love with or that the polygenic score for education measures an unchanging educational potential, Meskó stresses. A possible widening of socioeconomic or health inequalities is a potential consequence, not an outcome demonstrated here. He suggests studies that follow people before and after they form a relationship, to distinguish initial similarity, the influence of a shared social environment and convergence during cohabitation. The study mainly shows that social patterns in relationships matter when interpreting genetic data.
Comments