Qn for population geneticists familiar with altruism research.
John Brady
genetics majors I spoke to thought "interesting and possibly new". If
you're a population geneticist familiar with research on social 'altruism',
please consider 1) and 2) below.
(This is an introduction to an earlier post. Sorry that I didn't outline
this introduction in the original post. I was concerned that geneticists
would think I was proposing 'group selection', and turn off immediately.
Wuh.)
1) I'd like to know whether behavioural scientists have ever observed a
non-Mendelian bounce-back in the incidence of animals showing an altruistic
phenotype (e.g. the 'heroic' defence of distant kin in the same tribe)
following the removal of all heroes from the breeding population. E.g. if
all heroic baboons in a tribe died due to cheetah attacks during one
unusually unlucky year, would the next generation of baboons show a return
('bounce-back') to roughly the same incidence of heroes, or would you be
left with a tribe of cowards? I appreciate that the aggression of wild
cattle can be bred out in just a few generations. But I wonder if
'altruistic' phenotypes are more resilient (perhaps due to the co-selection
described below). Certainly, if a group of individuals suffers the
selective forces of occasional 'hero wipe-outs' as above, then 'hero
bounce-back' would confer a substantial selective advantage.
2) I think the above phenomenon could be explained by the mechanism I
proposed in the first 1.5 pages of my 1st post to sci.bio.evolution titled
"Possible Mechanism to Explain Postulated 'Distribution Modification' of
Behavioural Phenotypes", 4/Nov/99. Is this mechanism feasible? The
experiment I propose to test for the existence of the above phenomenon is
not cheap, so I think it's reasonable to try to think how the phenomenon
might actually occur. (The post describes in detail how individual
selection could evolve a co-selected triple-allele that includes a
'distribution allele' that toggles between the other two co-selected alleles
for heroism and cowardice.)
If 2) above seems plausible, please also read the next page of my post.
From what I can gather of current altruism theory, this would be exciting
stuff if it tests true! It describes how the individual selection could
evolve the level of heroism of the above co-selected hero allele to display
more altruism than could evolve in a singly-selected hero allele. Despite
being evolved by individual selection, such an altruism phenotype would be
group adaptive but have the appearance of being 'individually
maladaptive'!!!
If this has already been discovered / accepted, it can only be quite
recently.
The remaining 5 pages of my post (from the title MAIN BODY down) has details
of suggested experiments and more speculative stuff. It can be read
independently of the first 1.5 or the first 2.5 pages. I naively threw it
all in together because I don't think my speculative stuff has any street
cred unless I score some runs with the Distribution Modification mechanism.
Thanks for your patience, folks.
CurtAdams
>1) I'd like to know whether behavioural scientists have ever observed a
>non-Mendelian bounce-back in the incidence of animals showing an altruistic
>phenotype (e.g. the 'heroic' defence of distant kin in the same tribe)
>following the removal of all heroes from the breeding population
You'll have to use some other test. Essentially *any trait whatsoever*
will "bounce back" the generation after selection. The reason is the some
(often most) of the variation in a trait is from non-additive and environmental
effects. Selection of those has no effect. Only selection on additive genetic
variance affects the trait average in the next generation (well, there's a
little drift too).
The way to demonstrate your idea of a "distributional allele" would be to show
variability runs in families: some families vary a lot, some a little. The
effect would need to be statistically significant and you'd have to control for
a bunch of things like environment and maternal effects.
Curt Adams (curt...@aol.com)
"It is better to be wrong than to be vague" - Freeman Dyson
John Brady
>bra...@bigpond.com writes:
>
>>1) I'd like to know whether behavioural scientists have ever observed a
>>non-Mendelian bounce-back in the incidence of animals showing an
altruistic
>>phenotype (e.g. the 'heroic' defence of distant kin in the same tribe)
>>following the removal of all heroes from the breeding population
>
>You'll have to use some other test. Essentially *any trait whatsoever*
>will "bounce back" the generation after selection. The reason is the some
>(often most) of the variation in a trait is from non-additive and
environmental
>effects. Selection of those has no effect. Only selection on additive
genetic
>variance affects the trait average in the next generation (well, there's a
>little drift too).
Hmmm. For a trait or an feature to evolve, it must be IBD and selectable.
I suspect height is also a function of the additive effects of many alleles
+ how much food the organism got, and yet afaik, if you kill off all the
tall animals in a tribe the next generation would be measureably smaller.
Unless there is an adaptive advantage in maintaining the same mix of heights
regardless of occasional culls of all tall animals!
A new thought: it may be that there are group selective forces favouring
inertia for some phenotype. E.g. consider a moth that lives amongst
vegetation that darkens with low rainfall and lightens with good rain. A
species that 'over-darkened' due to a dozen years of drought could find
itself uniformly too dark and maladapted when the good seasons returned.
Perhaps an arrangement that has some of the light alleles at locii that
contain other important alleles is adaptive. (This is distinct from the
mechanism that allows you to rapidly breed an animal for height up to a
certain point where you have all the alleles 'set to high', and then have to
wait for mutations before further height increases are possible at a much
slower rate.)
My Distribution Modification mechanism was postulated for
either/or-phenotypes, where it's clearly adaptive to be A1 or B1 but not
halfway between A1 & B1 (and further that A1 was individually-adaptive
(I-A), but almost group-adaptive / individually-maladaptive G-A/I-M). This
carried an assumption of singly selectable genotypes for A1 and B1 that
would allow predictions of Mendelian distribution. If you're correct in
saying there isn't any such binary trait, then without such clear-cut
phenotypes no prediction of Mendelian distribution is possible so you can't
see if that distribution is being modified.
(But is it possible to have a mutation introducing a binary-type trait?
Dawkins talks of a beaver with mutant neck muscles holding muddy logs higher
out of the water. Perhaps, but probably not a I-A_but_almost_G-A/I-M.)
That's the problem with postulating a mechanism for a postulated
itutation - if the situation doesn't actually occur... :-(. Still, I
reckon the Distribution Modification Mechanism might be a useful 'algebra'
for understanding the evolution of G-A/I-M phenotypes.
An example of my Distribution Modification came up, if I understood it
correctly. Apparently there is a protozoa that has a salt-resistant and a
freshwater-resistant form, with no adaptive intermediates, and no way of
determing what it's next environment will be. (Levins discussed this
phenotypic variation in the 60's???) Presumably this is adapted from a
still existing sea variety, and possibly there are freshwater variants in
different rivers, but this littoral or lagoon variety has a distribution of
salties and freshies. For an environment that is e.g. prodominantly
freshwater, you can sort of consider the minority salt-resistant form to be
an altruistic G-A/I-M. But clearly, there is some random toggling mechanism
that activates the alternative forms, and this would be the adaptive feature
for this species that lets it inhabit environments of variable salinity.
>The way to demonstrate your idea of a "distributional allele" would be to
show
>variability runs in families: some families vary a lot, some a little. The
>effect would need to be statistically significant and you'd have to control
for
>a bunch of things like environment and maternal effects.
Nice idea. Although various variablities would be IBD, I suspect they would
only coalesce into a variability large enough the measure and distinguish if
there are selective forces for variability at one end of a range but not at
the other. But this situation is quite possible for a species that has
unreliable seasons at one end of its range and reliable seasons at the
other. Ranges that include SE or SW Australia and extend into various
deserts may well suit. (I can't imagine any such traits that are
I-A_but_almost_G-A/I-M, so it's probably no use for testing my Distribution
Modification, but it's an interesting idea in its own right.)