A strong critique of On the Role of Males by Dawkins was, put simply, that it continues the wrongheaded pursuit of the notion of “multilevel selection” that confuses the vehicles and replicators in the selection process. In this case, selfish genetics precludes the application of a lossy filter for genetic defects because it becomes a species-level selective mechanism.
While largely a technical distinction in evolutionary theory, much remains to be explained if we presume that there are no selective pressures that operate at levels above the genetic. For instance, when human females are subject to environmental stressors, the sex ratio changes to favor boys. From Valerie Grant’s Wartime sex ratios: Stress, male vulnerability and the interpretation of atypical sex ratio data:
At the end of war, and other times of both chronic and acute stress, remarkable changes occur in the human secondary (birth) sex ratio. At the end of a long war, significantly more boys are born; after a short war, or disaster, fewer boys than usual are born six to nine months later. Since it is commonly held that the sex of the offspring is a matter of chance, these data provide an intriguing problem; but new findings in reproductive physiology, and an increased understanding of male vulnerability, could help resolve it. It appears the sex ratio of offspring may be influenced by variations in the mother’s follicular testosterone. Under conditions of chronic stress, maternal testosterone rises, resulting in an increase in male conceptions; but these same stressful conditions also exacerbate differential male vulnerability, so more males are lost during pregnancy. At the end of war, improving conditions temper male vulnerability, leaving higher sex ratios at birth. Conversely, normal conditions at conception followed by a severe stressor during pregnancy result in lower secondary sex ratios.
Here we have evidence of a proximate cause (stress triggers hormone changes) that leads to a group-level change that has selective advantage. How do we reconcile this with purist selfish genetics? We might be able to create a string of explanations that supports a selfish genetic interpretation, but it’s not clear why an immensely reticulated explanation helps improve on the interpretation.