In some ways, this isn't surprising. More and more studies are finding that, while we can have friendships and more serious relationships with people who are unlike us, generally speaking, we prefer to have friendships and marriages with people who are as similar to ourselves as possible in all of the most salient respects. And, mental health conditions are frequently salient to who we would like to have a relationship with in life. (In contrast, non-mental health conditions like diabetes mellitus and rheumatoid arthritis are apparently not salient as they do no show elevated levels of assortative mating.)
Individuals with other mental health conditions (e.g. "affective disorders") have much lower likelihoods of marrying someone else with the same or another mental health condition, and those conditions tend to be less hereditary.
The study also confirms that these conditions are influenced both by additive genetic variance, which is increased reliably by assortative mating (a kind of variance that is easily detected in genome wide association studies and effective caps the amount of heritability that those studies can identify) and non-additive genetic variance (involving factors like dominance, complex non-linear multiple gene effects, and gene x environment effects) which current genetic studies often miss.
This is clear from twin studies. If only additive genetic effects were at work, risk would be proportionate to relatedness.
However, twin studies suggest that nonadditive genetic influence is greater for these same 3 disorders than for other disorders in that dizygotic twins are much less than half as similar as monozygotic twins. Most notably, concordance for ASD is approximately 60% for monozygotic twins and approximately 5% for dizygotic twins. Nonetheless, it is possible that these 3 disorders are so highly heritable because they include injections of both additive genetic variance from assortative mating and unusually high nonadditive genetic variance. Without assortative mating, these disorders might show little additive genetic variance.Assortative mating could also be a major factor in explaining high rates of co-morbid mental health conditions. Many individuals may inherit one mental health condition from one or more ancestors, and another from another subset of ancestors.