Autism is known to be highly heritable. Twin studies in children and adolescents suggest that subclinical expression of traits related to autism also shows strong genetic influences. Some researchers (Baron-Cohen, 2006; Constantino & Todd, 2005) have proposed that individuals may actively or passively select their partner based on their autistic traits, a phenomenon referred to as assortative mating. Assortative mating induces higher similarity between siblings and - when not taken into account - may attenuate heritability estimates. Previous studies examining assortative mating have yielded conflicting results, with spousal correlations ranging from non-significant (r=.03 to.05; Pollman et al., 2010; Hoekstra et al., 2007) to moderate (r=.26 to .38; Constantino & Todd, 2005; Virkud et al., 2009; Schwichtenberg et al., 2010).
Objectives:
This paper reports on the first community-based extended twin family study in adults, aiming to i) investigate the heritability of autistic traits in adulthood while taking into account assortative mating, and ii) elucidate the mechanisms behind potential assortative mating.
Methods:
305 twin pairs (mean age 46.14 years) and their parents, spouses, siblings and children (total N=1094) were asked to fill out the short version of the Autism-spectrum Quotient (AQ-short; Hoekstra et al., 2010). The genetic and environmental effects on individual differences in AQ-short scores were examined using structural equation modelling. It was tested whether spousal resemblance on AQ-short scores was due to active partner selection based on observable autistic traits (phenotypic assortment) or to a shared social background between the spouses (social homogamy).
Results:
Identical twins resembled each other more strongly on the AQ-short than fraternal twins, non-twin siblings, parents and their offspring, and other relatives with less genetic similarity, suggesting that genetic influences play a role. AQ-short scores of spouses correlated only modestly (r=.13). The very small effects of assortative mating were explained by social homogamy rather than phenotypic assortment. Genetic influences explained 55% of the variance in autistic traits, most of the remainder of the variance was explained by environmental influences unique to each individual in the family.
Conclusions:
In line with previous findings from studies in children (Ronald et al., 2006) and adolescent twins (Constantino et al., 2003; Hoekstra et al., 2007), individual differences in autistic traits show substantial heritability in adulthood. The effect of assortative mating is very limited; suggesting that these mechanisms have little influence on the risk for autism, and induce negligible bias in heritability estimates when using classical twin studies. Since autistic traits show considerable variation in the general population, future genetic studies may be aided by measuring autistic traits on a continuous measure such as the AQ.