Re: Hard science of evolution


"John Harshman" <jharshman.diespamdie@xxxxxxxxxxx> wrote in message news:fqmui.26976$RX.1591@xxxxxxxxxxxxxxxxxxxxxxxxxxxxx
Perplexed in Peoria wrote:

"UC" <uraniumcommittee@xxxxxxxxx> wrote in message news:uranium-1186520883.588047.221390@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

On Aug 7, 1:49 pm, "Perplexed in Peoria" <jimmene...@xxxxxxxxxxxxx>
wrote:

<geop...@xxxxxxxxxxx> wrote in messagenews:1186496797.481772.181440@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Occasionally Bimms makes the claim that evolution is not a rigorous
mathematical subject like, for example, physics, for which every
principle has a cool equation with squares and divides and stuff. I of
course have never believed him on this, since creationists are well
known for making pronouncements about the theory of evolution without
actually having looked at it. Nonetheless, I myself could not point to
any mathematics either. So just out of curiosity, I looked up
population genetics on wikipedia and was immediately confronted with
statistical equations and models of exactly the kind Bimms claims is
lacking.

What I want to know is, why isn't any of this stuff ever mentioned
much when people ask for the hard science behind evolution?

Well, one reason is that people's eyes glaze over when you show them the
math. Or they assume you are just trying to snow them. Or, if they are
really good, they look at the math carefully and notice the unrealistic
assumptions embedded therein.

All mathematical sciences involve unrealistic assumptions at some level.
Just think of the frictionless sliding blocks and unstretchable cables
running through inertia-less pulleys that you encounter in Physics 101.
Of course, they are there just to simplify the math. People don't object
too much, but that is partly because they are not working under a religiously
inspired agenda of discrediting the whole program of Newtonian mechanics.

But, as it happens, the unrealistic assumptions embedded in the basic
population genetics math are often far worse than those encountered in
elementary physics. One reason is because the math is far uglier than
that in physics to begin with - so if you put in very much realism, no one
can understand the math.

Another reason deals with problems of measurement. Physics is very much
an empirical science - you can measure the starting conditions, run the
experiment, measure the results, compare to theory, and determine whether
or not all is well with theory and experimental setup. But measurement,
in the evolutionary theory part of population genetics, is far more difficult,
and sometimes practically impossible. One good example is the heritability
coefficient for fitness. The concept is at the very heart of the idea
of evolution by natural selection. If you survive better and have more
children than your peers, then your children should inherit that special
something from you so that they too can survive better and have more children
than their peers. And, if the theory is true, then it ought to be possible
to show this - to measure it. Observe a natural population carefully over
two or more generations, keep careful records of survival and reproduction
and good genealogical charts. And show that fitness really is heritable.
But there are problems with this. Beyond the obvious one of knowing what
the survival rates would have been for the children of the first-generation
population that did not survive!

[snip]

Heritability of fitness can never be very high. You're always looking
backward, fighting the last war, not the next one. [snip example]

Actually, this strikes me as completely backward. You are not fighting the
last war, or the next one - you are always fighting the *current* war. What
you seem to be saying is that you are currently mostly adapted to the needs
of the last war. True enough, but then any variation that exists in the
population which is relevant to the needs of the current war ought to be
heritable variation in fitness. Assuming, that is, that the current war
lasts for more than a single generation.

The usual explanation ('excuse'?) given for the low measured heritability of
fitness is not that the environment is changing too fast, but rather that it
is changing too slowly. Thus, the population has already optimally adapted
to the environment - it is sitting near the top of a fitness peak and any
variation in fitness in the population is mostly a matter of non-heritable
luck rather than heritable difference in genetic endowment.

There's a second potential explanation, that fitness is related to a
great many alleles that all have gene x gene interactions and that are
reshuffled in every generation in diploid outcrossers. Or were you
talking about clones?

You raise an interesting technical point here. Geneticists distinguish between
'broad sense' heritability and 'narrow sense' heritability. The correlation
in phenotype (including realized fitness) between an organism and its MZ twin
or clone is all included in the 'broad sense' version of heritability. If
a trait is highly heritable in the broad sense, you can be pretty sure that
the trait is detertermined more by genes than by the environment. A clone
not only has all of its progenitor's genes, it also has all of the same cross
interactions beween genes.

But in normal sexual reproduction, offspring inherit only half of each parent's
genes and even less than half (typically) of the cross interactions. The
heritability coefficient for this situation is called 'narrow sense' heritability.
This is the version of heritability which goes into determining the response
to selection. And this is the version of heritability which is important for
the theory of evolution, and which I claimed was annoyingly difficult to
measure.

And the third explanation is that there is simply high stochastic
variance in realized fitness, which is related to but not identical to
your "broad peak" explanation.

This does, of course, make the heritability of fitness smaller. But that is
what we want as theorists, because it also makes adaptive evolution slower.
If all of the variance in realized fitness (i.e. reproductive success) was
due to stochastic factors and none was due to narrow-sense heritable genetics,
then natural selection wouldn't accomplish anything.

.

Relevant Pages

  • Re: Hard science of evolution
    ... they look at the math carefully and notice the unrealistic ... And show that fitness really is heritable. ... Heritability of fitness seems to be pretty small. ...
    (talk.origins)
  • Re: Complexity
    ... theory forcing organism fitness altruism ... The “heritability” of something in nature. ... selection: ... and the drift effect. ...
    (sci.bio.evolution)
  • Re: Heritability of fitness
    ... >>in the genome and the heritability of total fitness. ... >>enough to measure - at least over timescales in which the new environment ... > selection with fitness declining away from their optima. ...
    (sci.bio.evolution)
  • Re: Heritability of fitness
    ... >> summarize the conventional thinking (and also points out that heritability ... >> The heritability of fitness: ... > "Natural selection is the central concept of Darwinian theory?the fittest ... > "It is beginning to look as though what Darwin really discovered was nothing ...
    (sci.bio.evolution)
  • Re: Heritability of fitness
    ... >> the rate at which information 'about the environment' is accumulated ... >> in the genome and the heritability of total fitness. ... >> should be very close to zero. ...
    (sci.bio.evolution)