If later-born children are more likely to be homosexual, then it seems trivially true that the more children one is inclined to have, the higher percentage of them will be gay. If any genetic mechanism encourages having more children, we'd expect there to be an apparent genetic link as well.

Alternatively, if error-checking causes spontaneous abortions of homosexual children (assuming homosexuality is an error in construction rather than in the instructions), then we would expect more children to be born to those without that error-checking. (And false positives of whatever error-checking is encoded may make it prohibitively expensive to do so.)