- There are heritable factors predisposing one to homosexuality
- These factors when expressed in heterosexual individuals increase their fitness
- Homosexual marriage will greatly reduce the number of offspring born to homosexual parents
#2 seems to be based on a study that showed that heterosexual twins with a homosexual twin had more sex partners than those without. I find that very weak evidence, since sexual incontinence is not what produces children, and the most reproductively fertile people are usually those that marry young, not the ones who spend their youth with multiple sex partners. In any case, it is a very indirect measure of reproductive success.
#1 is the main reason for the seeming "paradox" of homosexuality: the (partial) genetic basis of homosexuality, together with the reduced fitness of homosexuals would result in negative selection that would penalize the trait: how could such genetic factors evolve (hence the need to create mechanisms such as in #2).
The problem, however, is that the analysis is based on a common-allele-common variant model. Homosexuality is relatively common, hence, it is mysterious how causative alleles for it would have grown in a population to levels necessary to account for the observed occurrence.
A real problem for that theory is that there is absolutely no evidence for such alleles. As far as I know not a single locus has been identified as being associated with homosexuality.
How can this be reconciled with the evidence for homosexuality being heritable?
The short answer is that homosexuality occurs in each generation de novo, perhaps due to an accretion of deleterious mutations impeding normal sexual behavior, or the chance recombination of commodity alleles that -in themselves- are neutral, but that -in combination- impede normal sexual reproductive behavior.
Rather than imagining the existence hypothetical and completely unattested alleles for homosexuality that persist in populations for many generations, we only have to accept the much more reasonable proposition that mutation load or simply an unfortunate draw from the recombination lottery produces a genome that predisposes its bearer to homosexual behavior, and hence to reduced genetic fitness.
So, there really is no great mystery: homosexuality is not a trait that persists due to elaborate balancing selection mechanisms, but rather the side effect of mundane negative selection: deleterious mutations that impede heterosexual attraction routinely occur in a small number of individuals de novo and these are culled from the gene pool in a few generations.
Hence, I predict that the homosexual population will not decline as a result of the legalization of homosexual marriage. Rather, a few percent of the population will continue to express the trait as they always have. It's also quite possible that the more tolerant social atmosphere may boost the non-genetic contributing factors to the trait, so an increase is not out of the question.