Doug, what’s wrong with group selection is mostly that selection at the individual level works so much faster. If something’s harmful to individuals, it’s likely to have been wiped out by individual-level selection before it gets the chance to help the group.
It’s possible to concoct scenarios where group-level effects win. For instance: some allele has no effect at all when heterozygous, but when homozygous it causes its bearer to become astonishingly altruistic. By the time there’s much incidence of homozygosity in any given community, the chances are that the allele is (heterozygously) quite common, and then it’s possible that the individual’s altruism does more net good than harm to bearers of the allele. This is kin selection rather than group selection really, but on a different scale from the usual.
Or: some allele has a very slight deleterious effect on individual fitness in general—slight enough that it typically takes, say, 100 generations before natural selection becomes visible over genetic drift. If it then has some group-level effect that prevents rare but group-destroying incidents (say, once every 100 generations someone without it will go nuts and kill everyone around them) then it could be selected for on balance simply because groups where it doesn’t happen to get fixed in the population tend to die. Note that making this work is rather dependent on group size.
But it’s pretty hard to concoct such scenarios that are actually plausible, and pretty hard to argue that anything in the real world looks much like them.
Doug, what’s wrong with group selection is mostly that selection at the individual level works so much faster. If something’s harmful to individuals, it’s likely to have been wiped out by individual-level selection before it gets the chance to help the group.
It’s possible to concoct scenarios where group-level effects win. For instance: some allele has no effect at all when heterozygous, but when homozygous it causes its bearer to become astonishingly altruistic. By the time there’s much incidence of homozygosity in any given community, the chances are that the allele is (heterozygously) quite common, and then it’s possible that the individual’s altruism does more net good than harm to bearers of the allele. This is kin selection rather than group selection really, but on a different scale from the usual.
Or: some allele has a very slight deleterious effect on individual fitness in general—slight enough that it typically takes, say, 100 generations before natural selection becomes visible over genetic drift. If it then has some group-level effect that prevents rare but group-destroying incidents (say, once every 100 generations someone without it will go nuts and kill everyone around them) then it could be selected for on balance simply because groups where it doesn’t happen to get fixed in the population tend to die. Note that making this work is rather dependent on group size.
But it’s pretty hard to concoct such scenarios that are actually plausible, and pretty hard to argue that anything in the real world looks much like them.