She did not know what the future held for her, but she resolved that if she managed to keep her job as a biology professor at Stanford University she would explore how widespread variation in gender and sexuality was in the animal kingdom. In the event she was forced to give up some administrative responsibilities and started to catalogue homosexuality in other species.

What she found astounded her. Studies document same-sex courtship rituals and mating in more than 300 species. Still more species have multiple genders, or exhibit gender reversal and hermaphroditism. Yet no one had collated them, no one had sought to explain this phenomenon. "Biologists know there is a problem there, they know there is a lot of same-sex sexuality, and it is in the back of everyone's mind that we are going to have to deal with it at some point," she says.

The problem is that dealing with it means challenging the master text in biology: Darwin's theory of evolution. Or more precisely, the part on the selection of sexual characteristics. In her book Evolution's Rainbow, due out next March, Roughgarden asserts that Darwin's theory is "false and inadequate" and that there is no patching it up.

Her main point of contention is over Darwin's notion that females select males for show, because their showy secondary sexual characteristics - the peacock's tail, for instance - reflect good genes. Because eggs are supposed to be costly to produce and sperm cheap, this in turn has led to the stereotypical - and, she believes, erroneous - depiction of males as promiscuous and females as coy and discerning. That false message has been picked up by evolutionary biologists, says Roughgarden, but you only have to look at animal societies to see that it is not true.

Take Japanese macaques, whose females are promiscuously gay. During the breeding season, they form lesbian consortships as well as heterosexual pairings. Among bonobos - the only primates apart from us to mate face-to-face - most females indulge in lesbian behaviour, rubbing their vulvas together, because, says Roughgarden, "If you did not do it, then you would not have any sisters. You would not have any buddies. It is absolutely necessary."

Bonobos perfectly illustrate the theory she offers up to replace Darwin's: social selection. According to this, much of the sexual behaviour observed in animals is not designed to propagate genes, at least not directly, but to make the protagonist socially acceptable to a powerful clique, thus ensuring him or her access to potential mates and a safe environment.

The penis of the female spotted hyena is very similar to the male's, although it contains the urethra and birth canal. This she erects and flashes about to other females, says Roughgarden, to advertise her eligibility to join their gang. "The party line is that genitals are used for the exchange of sperm," she says. "But the fact is that among mammals, they are often coloured very brightly and are bigger than they need to be." She believes oversized genitalia, the peacock's tail and perhaps even the enormous human brain evolved as a medium of communication, of body language between members of the same sex, because of this need for social inclusion.

The second part of her theory is that females do not choose males for their genes, as Darwin taught, but to avoid "deadbeat dads". She says females manage male power by selecting for good fathers rather than good sperm. This, she believes, creates a marketplace for reproductive opportunity.

Dominant males have a lot on their plate, maintaining their physical condition, controlling large territories and seeing off challengers. So it is in their interest to sub-contract out the task of finding a mate. The example she gives is the bluegill sunfish of North America, where a dominant male will recruit a smaller, feminine male - so-called because he sports female colours - in what looks like a homosexual courtship. They mate with a female in a menage ¿ trois. The conventional view is that the feminine male mimics the female to steal copulations. But Roughgarden says no one has proved the dominant male does not know the feminine male is male. She argues that he is negotiating a reproductive opportunity for both himself and the dominant male, that he may have been "schooled" with the females and therefore brings to the deal his prior rapport with them. Data show females prefer to enter territories containing dominant and feminine males.

She admits there is no direct evidence to support her hypothesis and has a get-out clause, arguing that as a theoretical ecologist it is her job to explain diversity and the job of experimentalists to gather the proof. Paul Vasey, a behavioural neurobiologist at the University of Lethbridge in Alberta, Canada, studies Japanese macaques in their habitat of Mount Fuji. He says their lesbian pairings in the breeding season do not promote social cohesion, because '- just as in heterosexual pairs - the females avoid incest. Dominant females will often also protect their subordinate lesbian partners against higher-ranking aggressors. In his opinion, their motivation is pure sensual pleasure.

Perhaps, in taking on Darwin, Roughgarden only wants to set the ball rolling. In the book she makes clear her personal and political standpoint, warding off criticism with the argument that throughout history those who have upheld Darwinian theory have had an axe to grind - whether it be to defend male philandering or to propagate the notion of a genetic elite.

She is also careful not to extrapolate her findings to humans, pointing out that patterns of homosexuality vary between species. "One can't draw parallels with humans other than to say that homosexuality is a regular part of nature and not some pathology," she says. In passing, however, she mentions the hijra, an ancient, caste-like group of transgendered people in India, and traces gender-crossing in European history from the Cybelean priestesses of the Roman Empire, through the transvestite saints of the middle ages right back to Joan of Arc.