I wish I could say I read Phillips answer, but I couldn't maintain focus. So, risking repeating elements of his analysis, I'm going to give this question a stab, but with a biological/evoluationary bent.
Evolutionary arguments have been offered for male homosexuality. For example, Symons (1979) argues that homosexuality results from a meme that takes advantage of the inborn male propensity to find sexual variety interesting. Normally this propensity would be adaptive because it would encourage males to attempt sexual relations with many women, which would result in more offspring. But a meme could parasitize this propensity, redirecting sexual interest to non-adaptive objects, as in fetishism or homosexuality.
Epigenetic arguments have been inspired by work on pre-natal hormones. Dîrner and his colleagues (cited in LeVay 1994) link stress during pregnancy to hormonal effects on the fetus that would lead to homosexuality. This could be part of a mother's adaptive reproductive strategy. That is, in times of stress, when it is difficult to raise children, it may be adaptive to have some homosexual children who could help their siblings raise offspring instead of having offspring of their own.
Memic and epigenetic arguments may explain some aspects of homosexuality, such as the cultural variations in practices or identities, or specific behaviors of homosexuals (femininity for example). However, the biological studies of homosexuality, together with the extraordinary cross-cultural similarities of exclusive homosexuals (Whitam 1983) suggest that some aspects of homosexuality (e.g. orientation) are at least partly determined by genes.
The basic problem for genetic explanations of homosexuality is to account for how a gene that does not lead to offspring could survive the pressures of natural selection. One possibility is that a maladaptive gene for homosexuality might re-occur repeatedly in a population if it results from the frequent mutation of a gene that is normally adaptive. However, people with maladaptive genes rarely exceed 1% of the population, while male homosexuality apparently occurs much more frequently (Whitam and Mathy 1986; Gadpaille 1980; Diamond 1993). Another possibility is that a recessive gene for homosexuality might have advantages that allow it to pass on to future generations. For example, homosexuals could directly help their relatives (who share the recessive gene) raise more children. This "kin selection" hypothesis predicts more homosexuality in those societies where homosexuals could most help their relatives (for example, in endogamous societies where homosexuals would live nearby). Nevertheless, cross-cultural research shows that this is not the case (Werner 1979). Another possibility is that a homosexual gene itself, when combined with non-homosexual genes, may provide reproductive advantages. This latter scenario is sometimes known as the "heterozygous" hypothesis, although it is really about any type of gene combination (LeVay 1994; Sommer 1990).
The heterozygous argument suggests that homosexuality results from genes for submissive behavior. An animal possessing only submissive genes would fail to reproduce for lack of trying. But an animal possessing only genes for dominance would also fail to reproduce because it would get killed for taking too many risks. It is the "heterozygous" males, who would most likely pass on their genes. By the genetic laws of probability, this would leave every generation with a certain percentage of individuals at the extremes. For example, if each of a pair of chromosomes had only one "homosexual" locus with only two possible alleles (one for dominance and one for submissiveness) this would leave a heterozygous couple with 25% of their offspring homozygous for submissiveness, 50% heterozygous, and 25% homozygous for dominance.