Re: Levels of selection (Was: Can cognition override natural

Guy Hoelzer <[EMAIL PROTECTED]> wrote or quoted:
> Tim Tyler at [EMAIL PROTECTED] wrote on 11/7/03 8:44 AM:
>> Guy Hoelzer <[EMAIL PROTECTED]> wrote or quoted:
>>> [EMAIL PROTECTED] wrote:

[snip]

>>>> With this sort of perspective, genes and species make sense as
>>>> units which can be selected - whereas individuals and groups do
>>>> not - they usually mix themselves together too much in each
>>>> generation.
>>>> 
>>>> It is well established that group selection can only arise if
>>>> there are barriers to gene flow in the population - and that
>>>> those barriers neet to restrict gene flow a lot if there are
>>>> to be adaptations that can be said to be for the benefit of
>>>> groups.
>>> 
>>> I think this is an outrageous claim.  The only folks who think that any such
>>> thing is "well established" are the gene-selection advocates.  There is
>>> absolutely no way in which the process of group selection is dependent on
>>> genes or gene flow.
>> 
>> I don't think I was trying to say anything contraversial here.
>> 
>> For group selection you need heritable variation among the groups.
> 
> Yes, but it is "well established" that there are modes of inheritance other
> than genetics.  Population genetics has been a paradigm for modeling
> evolution based only on the genetic mode of inheritance, which allowed us to
> incorporate details of genetic mechanisms into those models.  This has been
> useful, because reliance on specific mechanisms constrains the evolutionary
> process in particular ways.  However, this also limits the inductive value
> of lessons learned from population genetic modeling, because those
> constraints to not apply to the mechanisms of inheritance through other
> modes, even when addressing selection at the individual level.  I would
> expect that inheritance at the group level would rely far less on the
> genetic mode than exists at the individual level, so inducing the lessons
> from population genetics to understand group selection would be a rather
> tenuous thing to do (IMHO).

Until the advent of human beings, non-genetic inheritance has been of
low significance and volume.

Until recenly about the biggest contributor has been location - the
offspring tend to share the same environment as their parents - since
they were at the same place when they were born.

Extra-genetic inheritance can strengthen the case for group selection:
if gene X is favoured in environment A and gene Y is favoured in
environment B, then there can be group-level selection between X and
Y based on the differential success of the groups in the different
environments.

This effect is no-doubt interesting - but it only really works for
genes that are selected differently in the different environments
(and perhaps their immediate bedfellows) - and so it is likely that
it won't touch most of the genome.

Human culture reprents substantial modern non-genetic inheritance.
However it isn't likely to make humanity much more group selected.
It is typically mixed up in each generation worse then genes ever
were by sex.

>> If groups share genes then that will tend to reduce variation
>> between the groups.  If they do it too much then it is going
>> to become hard to find heritable differences between the groups.
> 
> This is only important to the extent that the genetic mode of inheritance
> accounts for a substantial fraction of group level heritability.
> We don't know much about the relative roles of other modes of
> inheritance, especially at the group level.

Classically non-genetic inheritance is typically ignored as being
negligible.  I'd say that was usually about right - most of what
organisms inherit from their parents, they inherit through their
genes (and perhaps the genes of their shared parasites).

>> The critical threshold is somewhere around one individual per
>> generation.  I would describe that figure as a low one.
>> 
>> ``If fewer than one individual per generation moves from population to
>> population, so that the amount of gene flow is very low, then
>> populations will develop fixed differences through natural
>> selection. Fixed differences are what they sound like -- differences in
>> which alleles are fixed in different populations. That is, for a gene
>> with two alleles (say D and d) allele D could be fixed (the only allele
>> present) in one population and allele d could be fixed in the other
>> population, and this low level of gene flow would not be enough to
>> change that. 
>> 
>> In contrast, if large numbers of individuals move from population to
>> population , so that the amount of gene flow is very high, the
>> populations will essentially be like one single population and will
>> have the same alleles, in the same frequencies, even if they occur in
>> somewhat different environments so that differences might otherwise
>> tend to evolve through natural selection. Large amounts of gene flow
>> will thus mask the effects of other forms of evolutions and make
>> populations similar.''
>> 
>> - http://www.utm.edu/~rirwin/391DriftFlow.htm
> 
> This is taken from an evolution class website at a small branch campus of
> the University of Tennessee where the use the same textbook I use when I
> teach Evolution.  This textbook is rather weak in this area (IMHO) and
> presents only "soundbites" that are at least a decade out of date.  The "1
> migrant per generation" threshold has been strongly discredited for the same
> reasons that Fst has been shown to be a generally poor framework for
> estimating migration rates.

I won't defend the "one individual per generation" figure - but nontheless
the basic insight is correct - it takes suprisingly little migration
between groups for gene flow between them to effectively prevent the 
accumulation of heritable differences between them.

> To add a nod to the kind of work I am currently engaged in, it is easy to
> show with computational models that gene flow is necessary for the emergence
> of genetic divergence between subpopulations [...]

? They cannot diverge genetically if they are completely isolated from
one another...?

>> However the basic idea still seems sound - migration easily destroys
>> the identity of groups, very low levels of migration are all that
>> is needed, and these low levels of migration are not very common in
>> nature.''
> 
> This is ONLY the case if you limit your definition of what constitutes group
> boundaries to patterns of genetic distinctiveness, and the mode of
> inheritance (at the group level in the current context) to genetics.

When discussing group selection, we are usually talking about "biological"
groups - as characterised by their gene pools.

If you want to broaden things to discuss the evolution of (e.g. political
parties), then I'll be happy to admit that those may evolve via all
sorts of forces - possibly including "group" selection (selection
between the parties).

However the link back to the sort of evolution as described by
changes in allele frequencies is a tenuous one.

>>>> It's the same with individual selection.  For there to be
>>>> adaptations that favoured individuals, you would need there
>>>> to be restricted gene flow between individuals - and the
>>>> restriction would have to be pretty severe.
>>> 
>>> I would say that there is very rarely gene flow between individuals.  An
>>> exception would be tissue transplants.
>> 
>> I meant across generations.
> 
> Sorry for my mistake.

My fault for not being clear.

>> In an asexual population there is no gene flow between individuals
>> across generations - no mixing - but in sexual populations genes are
>> thoroughly mixed between two individuals in each generation - resulting
>> in substantial gene flow.
> 
> Interesting idea.  Do you know if this (there must be limited gene flow
> between individuals to permit selection at the individual level) appears in
> publication anywhere?

My comments above are - I believe - paraphrasing some of Dawkins
explanations of why the individual level is worthless.
He refers to it as the "internal fragmentation destroys copying
fidelity" issue [TEP, p.99] - and writes:``It is therefore better not to
speak of adaptations as being for the good of the organism'' [same].

>> I don't see how that can be true.  If individual selection can undo
>> the effects of group selection in the time that groups take to
>> reproduce, then it will destroy any heritable variation between the
>> groups in the process - with the result that selection between the
>> groups will have nothing to work on.
> 
> I see a few points here to take issue with.  First, I don't see how undoing
> the effects of group selection would necessarily reduce the heritability of
> group level traits. [...]

It removes group differences - with the result that group selection
has no variation to work on.

> Second, just because individual selection can interfere with group
> selection does not mean that it usually does, nor that it would be
> an important factor. [...]

Nontheless, individual selection is likely to care passionately about
most aspects of most organisms; there is little that is irrelevant to it.

Group selection will have the best chance when affecting traits that
individual selection acts weakly on - but I suspect this category
consists of lean pickings.
-- 
__________
 |im |yler  http://timtyler.org/  [EMAIL PROTECTED]  Remove lock to reply.