Levels of selection
Levels of selection
The issue of at what levels selection can be meaningfully be
said to act has been a contraversial one.
The question is important - since the agent that can be said
to be benefitting from selection underlies many explanations
in biology.
It really can make a difference what entitly can be said to
benefit from adaptations - since adaptations that benefit
the species can look quite different to adaptations that
benefit the individual.
Some have claimed that the main level within biology that
exbibits differential reproductive success is the gene [1].
Others advocate a range of different levels of selection,
with actors at each level [2,3]. Candidate levels include the gene, the
cell, the individual, the group, and the species.
Tim's view
I am sympathetic to the latter view - there are multiple
levels of selection - and biological theories that fail to
recognise this are likely to prove to be incomplete.
However it is also sometimes asserted that many of the
"higher" levels of selection are relatively impotent - and
that any adaptations that favour them tend to be rapidly
undone by selection at the level of the individual.
It appears that there may be some truth in this: individual
selection is very powerful and does
control much of the variation on which group-level selection
acts.
However, individual selection fails to drive out all
variation - and high level selection does have the
final word. Individual-level selection does not eliminate
all variation. We still witness competing species
driving other species extinct - as in the case of the fauna
of South America - or more recently in the case of
marsupials versus placental mammals. This suggests species-
level selection is indeed responsible for some macroscopic
features of the world.
So: while I am sympathetic to the notion that higher levels
of selection may prove to be relatively weak, I'm not convinced
that they can be written off for this reason - and feel
that their low significance is often overstated.
A particular problem comes from those individuals who hold
the view that high level selection is not just weak - but
that it doesn't exist at all - and that explatnations that
invoke high level selection are necessarily incorrect.
This view seems relatively widspread - but it is wrong. A
good argument can be made that high level selection is a
relatively weak force - but the argument that it
is non-existent is just plain incorrect.
A matter of definition
Unfortunately, some of the confusion and controversy
surrounding the level of selection issue is associated with
the usage of the terms "gene selection", "individual
selection", "group selection", and "species selection".
I'll briefly present my definitions of these terms - and
explain why I think they make the most sense.
- Gene selection - differential reproductive success of
genes on the basis of their traits;
- Individual selection - differential reproductive success of
individuals on the basis of their traits;
- Group selection - differential reproductive success of
groups on the basis of their traits;
- Species selection - differential reproductive success of
species on the basis of their traits;
I think these definitions are concise, clear, intuitive,
consistent - and make the most sense.
A remaining point of debate about such definitions is
whether producing a geographically-widely-distributed
species is somehow worth more - on the grounds that it
increases the chance of a species being a long-term
ancestor. However this is a relatively minor issue -
which I will discuss no further here.
I note that there are those who define group selection
differently. In particular they define it to
exclude examples of differential reproductive
success of groups that can be explained by individual-level
selection.
Folks who define group selection in these terms then usually
go on to argue that it is not a very significant force.
Those using this definition often cite G. C. Williams'
1966 book [4], in support of their definition.
Williams drew attention to the fact that differential
reproductive success of groups could sometimes be
explained on the basis of differential reproductive
success of individuals within those groups.
Under those circumstances, he argued that group selection
had no explanatory force, offered no novel predictions to
distinguish it from individual selection - and should thus
be discarded on grounds of Occam's razor - as a useless and
unnecessary theory.
This is all very well - but I don't consider that it
provides a good basis for defining group selection
to be something other than differential reproductive success
of groups on the basis of their traits.
There is no rule which states that group selection must
always produce different predictions from individual
selection - no more than there is a rule that says that
relativity must always produce different predictions from
Newtonian physics.
I argue that any attempt to define group selection as those
phenomena explicable by differential reproduction of groups
- but not by differential reproduction of
individuals - makes a complete mess out of the definition of
group selection.
Any attempt to define species selection along similar lines
would be even more of a mess - consider:
Species selection: the effects of differential reproduction
of species on the basis of their traits - which are not
explained by differential reproduction of groups on the
basis of their traits - or by differential reproduction of
individuals on the basis of their traits.
Indeed, maybe we should also include gene-level selection in
the definition:
Species selection: the effects of differential reproduction
of species on the basis of their traits - which are not
explained by differential reproduction of groups on the
basis of their traits - or by differential reproduction of
individuals on the basis of their traits - or by
differential reproduction of genes on the basis of their
traits.
Such definitions are ugly, counter-intutitive messes.
Williams does make a fine point - in order to
determine whether group selection has any explanatory power
not offered by individual selection, the cases of interest
are those where group selection and individual selection
make different predictions.
Only in those cases can the two theories possibly be
experimentally distinguished.
Howver those followers of Williams who define group
selection to consist only of the cases where it
makes different predictions from individual-level selection
go too far - much too far.
A sufficient condition for the existence of group selection
should be witnessing differential reproductive success of
groups with differing traits.
The questions of how and why that
differential reproductive success arises are irrelevant
issues: the observation of differential reproductive success
on the basis of group traits is sufficient to identify an
instance of group selection.
References
- Richard Dawkins - The Selfish Gene
- Leo W. Buss - The Evolution of Individuality
- Stephen J. Gould - The Structure of Evolutionary Theory
- G. C. Williams - Adaptation and Natural Selection
|