Lecture 4. Group selection, individual selection,
gene level selection: evidence from analysis of the
evolution of sex ratios.
RA Fisher WD (Bill) Hamilton
Intro: In the last lecture we saw that we can model evolution by considering
what happens to allele frequencies over time. To do this we need to know how
the alleles are transmitted. We then ask under what conditions will the allele
invade. This will tell us what is going to happen in evolution. But when thinking
about (rather than modelling) evolution can we have a question, the answer to
which will give us, by logic a good estimation to the results of the mathematical
models? This would be useful to allow to develop an intuition as to what is
happening in evolution. Most people think that they have this intuition already.
The function of this lecture is to clarify how a well functioning intuition ought to
work.
So consider any given species and lets us think about a new mutation that
affects some aspect of the organisms biology. In principle when we ask
about evolution and want to guess the course of evolution
people might be tempted to ask:
1) Is the change going to be good for the species? – group selection
2) Is the change going to be good for the individual? – individual selection
3) Is the change going to be good for the allele? – gene selection
For the “good of the species”: A classic example would be the story of
lemmings throwing themselves off cliffs when food gets scarce. Wynne Edwards
would argue that this was done for the good of the species - to keep numbers
down some individuals would commit suicide, reducing the population density,
so benefitting the species. In fact lemmings don’t do this.
It is easy to see that there are many cases when the three do disagree.
Consider for example the problem of being nice to someone else – altruism.
Imagine I go out of my way to help a bunch of strangers. No doubt if I did this,
the species would be better off. But I would be worse off. So it is good for the
species but bad for the individual. So which is more important – an individual’s
loss or a group’s benefit?
Group selection predicts female biased sex ratios:
We can get a good handle on this problem by considering the evolution of sex
ratios. In most species with sperm and eggs, all investment into the next
generation comes from females (males are a waste of space – no
investment). So we can ask, what the sex ratio might be if a) group selection
was operating or b) if individual selection was operating.
Under group selection we expect selection to favour the trait that maximizes
the group’s growth rate. The sex ratio that does this is one in which there are
enough males to fertilize all the females but no more (as few males as
possible are produced). As one male can typically fertilize many females,
population growth rates will be highest under a strongly female biased sex ratio.
, RA Fisher (1930) pointed out that individual level selection does not favour
this – it is unsustainable. He argued that if there are loads of females about then
a mother should make loads of sons – this way her sons end up being very
successful fertilizing loads of females and leaving loads of grandchildren
(whereas a daughter would have fewer offspring). Likewise if there are loads of
males about, mothers should be favoured to make daughters as they would be
the limiting resource. So if individuals were left to choose the sex ratio they
should end up with a stable 1:1 ratio.
More precisely, he argued it should be a 1:1 ratio in investment into the
sexes. If males need to be big and, say, cost twice as much as daughters to
raise, then making a 1:1 body count sex ratio is not the best thing. When the
sex ratio is 1:1 a son has the same average number of children as a daughter.
But since sons cost twice as much they represent a bad return for the
investment. Each grandchild produced by a son is twice as costly as each
produced by a daughter. Factor in this and we find that the stable position is 1:1
investment not a 1:1 headcount. Naturally if the sexes are equally costly, then a
head count is also an investment count.
Importantly then, group selection and individual level selection end up making
different predictions. Which is found? Fisher’s model as we shall see has a
number of hidden assumptions (assumes panmictic population, autosomal
modifier of the sex ratio) but when these are met, we do indeed find 1:1
investment sex ratios. Notably in two species of Polistes wasps (P. metricus and
P. variatus) we find a female biased head count when males are expensive to
make (P. metricus), but a 1:1 headcount when they are the same size (P.
variatus). In both the investment ratio is 1:1. Importantly, we can
experimentally start a population off with a female biased sex ratio and see that
it returns to a 1:1 ratio because of selection favouring mothers that make more
of the minority sex. This has been done experimentally using Drosophila.
BIG CONCLUSION: GROUP SELECTION DOES NOT WORK.
If you ask “is a trait good for the group/species” the answer will not be
a good predictor of what we see in evolution. Therefore never use
group selective arguments to explain the evolution of traits.
This conclusion was further re-inforced by the work of WD Hamilton (1967).
He noticed an assumption in Fisher’s logic. The assumption was that a given
female could be mated by an unrelated male – panmictic population (any
individual can mate with any other individual in the species). That is Fisher’s
argument was that a mother should make sons because these sons could
fertilize the progeny of other mothers. But what if this is not true? What if we
are looking at some wasps, for example, in which the eggs are all laid, hatch and
reproduce in a butterfly larva? In this case the females will be mating with their
brothers – there are no other males about (no panmixis). This is known as
inbreeding. In this case Hamilton noticed, the best things for the individual is
gene level selection: evidence from analysis of the
evolution of sex ratios.
RA Fisher WD (Bill) Hamilton
Intro: In the last lecture we saw that we can model evolution by considering
what happens to allele frequencies over time. To do this we need to know how
the alleles are transmitted. We then ask under what conditions will the allele
invade. This will tell us what is going to happen in evolution. But when thinking
about (rather than modelling) evolution can we have a question, the answer to
which will give us, by logic a good estimation to the results of the mathematical
models? This would be useful to allow to develop an intuition as to what is
happening in evolution. Most people think that they have this intuition already.
The function of this lecture is to clarify how a well functioning intuition ought to
work.
So consider any given species and lets us think about a new mutation that
affects some aspect of the organisms biology. In principle when we ask
about evolution and want to guess the course of evolution
people might be tempted to ask:
1) Is the change going to be good for the species? – group selection
2) Is the change going to be good for the individual? – individual selection
3) Is the change going to be good for the allele? – gene selection
For the “good of the species”: A classic example would be the story of
lemmings throwing themselves off cliffs when food gets scarce. Wynne Edwards
would argue that this was done for the good of the species - to keep numbers
down some individuals would commit suicide, reducing the population density,
so benefitting the species. In fact lemmings don’t do this.
It is easy to see that there are many cases when the three do disagree.
Consider for example the problem of being nice to someone else – altruism.
Imagine I go out of my way to help a bunch of strangers. No doubt if I did this,
the species would be better off. But I would be worse off. So it is good for the
species but bad for the individual. So which is more important – an individual’s
loss or a group’s benefit?
Group selection predicts female biased sex ratios:
We can get a good handle on this problem by considering the evolution of sex
ratios. In most species with sperm and eggs, all investment into the next
generation comes from females (males are a waste of space – no
investment). So we can ask, what the sex ratio might be if a) group selection
was operating or b) if individual selection was operating.
Under group selection we expect selection to favour the trait that maximizes
the group’s growth rate. The sex ratio that does this is one in which there are
enough males to fertilize all the females but no more (as few males as
possible are produced). As one male can typically fertilize many females,
population growth rates will be highest under a strongly female biased sex ratio.
, RA Fisher (1930) pointed out that individual level selection does not favour
this – it is unsustainable. He argued that if there are loads of females about then
a mother should make loads of sons – this way her sons end up being very
successful fertilizing loads of females and leaving loads of grandchildren
(whereas a daughter would have fewer offspring). Likewise if there are loads of
males about, mothers should be favoured to make daughters as they would be
the limiting resource. So if individuals were left to choose the sex ratio they
should end up with a stable 1:1 ratio.
More precisely, he argued it should be a 1:1 ratio in investment into the
sexes. If males need to be big and, say, cost twice as much as daughters to
raise, then making a 1:1 body count sex ratio is not the best thing. When the
sex ratio is 1:1 a son has the same average number of children as a daughter.
But since sons cost twice as much they represent a bad return for the
investment. Each grandchild produced by a son is twice as costly as each
produced by a daughter. Factor in this and we find that the stable position is 1:1
investment not a 1:1 headcount. Naturally if the sexes are equally costly, then a
head count is also an investment count.
Importantly then, group selection and individual level selection end up making
different predictions. Which is found? Fisher’s model as we shall see has a
number of hidden assumptions (assumes panmictic population, autosomal
modifier of the sex ratio) but when these are met, we do indeed find 1:1
investment sex ratios. Notably in two species of Polistes wasps (P. metricus and
P. variatus) we find a female biased head count when males are expensive to
make (P. metricus), but a 1:1 headcount when they are the same size (P.
variatus). In both the investment ratio is 1:1. Importantly, we can
experimentally start a population off with a female biased sex ratio and see that
it returns to a 1:1 ratio because of selection favouring mothers that make more
of the minority sex. This has been done experimentally using Drosophila.
BIG CONCLUSION: GROUP SELECTION DOES NOT WORK.
If you ask “is a trait good for the group/species” the answer will not be
a good predictor of what we see in evolution. Therefore never use
group selective arguments to explain the evolution of traits.
This conclusion was further re-inforced by the work of WD Hamilton (1967).
He noticed an assumption in Fisher’s logic. The assumption was that a given
female could be mated by an unrelated male – panmictic population (any
individual can mate with any other individual in the species). That is Fisher’s
argument was that a mother should make sons because these sons could
fertilize the progeny of other mothers. But what if this is not true? What if we
are looking at some wasps, for example, in which the eggs are all laid, hatch and
reproduce in a butterfly larva? In this case the females will be mating with their
brothers – there are no other males about (no panmixis). This is known as
inbreeding. In this case Hamilton noticed, the best things for the individual is
