My counter-argument, based just on intuition, was that the reproductive costs of rape outweigh its advantages. In a society without formal legal systems, the families of raped women will kill rapists, and the wider society will shun sexual transgressors (and being shunned is a big deal). There are of course destitute and unprotected women, without family to shelter them, but these women are in no position to raise babies. Destitute women will use abortion and infanticide, and of course lose babies at a much higher rate to disease and famine.
(ETA and there is rape in wars and in clan feuds, but these are not 'good reproductive opportunities' for child survival as we have seen in places like Rwanda. I hate to use these terms, but these are the terms that are used.)
And to be even blunter: reproductive success in humans is not mainly about 'getting women pregnant' because women are not passive receptacles and raising human children takes years of hard work, to which women must consent.
In contrast I think concubinage - keeping young women as extra 'second wives' - is actually a good reproductive strategy, although it's not great for women. So, I'm not all happy clappy saying 'nice things are reproductively successful and nasty things aren't'. Just that one particular nasty thing - rape - is only a good reproductive strategy in male fantasy land.
Anyway, some researchers studied the reproductive success rate of rape in a society with no anti-rape laws (no legal structures of any kind) - The Ache of Paraguay. They didn't observe any rapes, but did a what-if calculation based on other measurements within the society: for instance, the odds that a woman was able to conceive on any given day.
He and two colleagues therefore calculated how rape would affect the evolutionary prospects of a 25-year-old Ache. The scientists were generous to the rape-as-adaptation claim, assuming that rapists target only women of reproductive age [of course IRL this is not always the case]. Then they calculated rape's fitness costs and benefits. Rape costs a man fitness points if the victim's husband or other relatives kill him, for instance. He loses fitness points, too, if the mother refuses to raise a child of rape, and if being a known rapist (in a small hunter-gatherer tribe, rape and rapists are public knowledge) makes others less likely to help him find food. Rape increases a man's evolutionary fitness based on the chance that a rape victim is fertile (15 percent), that she will conceive (a 7 percent chance), that she will not miscarry (90 percent) and that she will not let the baby die even though it is the child of rape (90 percent). Hill then ran the numbers on the reproductive costs and benefits of rape. It wasn't even close: the cost exceeds the benefit by a factor of 10. "That makes the likelihood that rape is an evolved adaptation extremely low," says Hill. "It just wouldn't have made sense for men in the Pleistocene to use rape as a reproductive strategy, so the argument that it's preprogrammed into us doesn't hold up."
You can read the whole article here.
Modern science is limited by the imaginative and emotional limitations of practitioners. Men tend to emphasise the statistical importance of impregnation and underemphasise the importance of successful completion of pregnancy (for example) because of their social role. And the male social perspective is seen as 'objectivity' because of power imbalances in our culture.
ETA there is a good metafilter discussion of the article here. For example this comment:
What this "discipline" (evo psych) does is give the reader some sense of control of the very messy thing that is human sexuality. By providing reductionist hypotheses regarding our deepest sexual fears, we gain some level of control over our fear. It has nothing to do with the scientific method.
Though I think it is not just sexual anxiety which is addressed, but the whole anxiety of corporeality.