It's possible perhaps that there are evolutionary paths that, although dead-ends, are easily mutated into and thus always present in a species... The cornucopi

It's possible perhaps that there are evolutionary paths that, although dead-ends, are easily mutated into and thus always present in a species... The cornucopia of debilitating mental illnesses that we tend to suffer from implies that this is almost certainly the case, and there are a lot of things that go into this. Regarding traits like homosexuality, it's likely that the genetic "cause" ends up so widespread because the possession of it (or of a close precursor) is not necessarily that devastating to reproductive success, and may even be beneficial when it's not expressed as homosexuality. For instance, it's pretty well established at this point that whatever predisposes people to homosexuality genetically is not the whole story, a lot of it has to do with the embryonic environment, upbringing, and possibly sheer chance, so if there are other beneficial things about that mutation (or set of mutations), then as long as they outweigh the probability that homosexuality is expressed and prevents procreation, the gene stays around. There's a rule of thumb, "one mutation, one death", which implies that no matter how deleterious a mutation is, on average, it will cause one death (or more accurately in this case, one failure to reproduce) when it happens. Roughly speaking, this is because if it's a very mildly deleterious mutation, it will spread quite far through the population before there's enough evolutionary pressure to stamp it out, and if it's a really bad one, it will kill off the first person to have it before it spreads at all. This can be influenced by social factors, as well, i.e. eyesight problems are rectified by our building glasses, but that just reduces evolutionary pressure against bad eyesight, so we're letting it spread its (much lower) fitness penalty throughout the population. There's an element there that I'm glossing over, which is that in a population, you have to consider the probability of recreating the deleterious mutation, too - for instance, if a trait depends on a combination of a few mutations to be present, then depending on the number and prevalence of those mutated alleles in the population, the stable percentage of people displaying that trait will change. It may just be the case that for various social reasons, predisposition to homosexuality has not been responsible for enough of a decrease in reproduction to counterbalance the probability of randomly mixing genes in just the right way to recreate it "from scratch." This could be particularly likely if some of the involved genes are useful by themselves in other situations, and homosexuality is hitchhiking on the backs of those beneficial mutations.