🚀 go-pugleaf

Jon Thaler wrote:
 
> Does the concept of natural selection have predictive power, or is it
> only good for retrodiction?  Please give examples.

	Natural selection is differential reproductive success of types of
genetic variants in a population. In short, some types of organisms
leave more offspring than others. Over time, the frequency of these
types of organisms increases within a population.

	The best sorts of predictions in science come from explicit
(mathematical) models. Natural selection is modelled in a variety of
ways. 

	There are population genetic models of natural selection. From these,
one can predict how fast an allele will increase or decrease in
frequency in a population. [An allele is a form of a gene. For example
A, B and O are three common (classes of) blood type alleles.] One can
also predict the probability an allele will fix (increase to a frequency
of one) in a population, how long an allele is likely to remain in a
population and etc. To make these predictions, you would need to measure
the relative fitnesses (reproductive rates in a given environment) of
individuals carrying different alleles. This is conceptually simple, but
getting good estimates of relative fitnesses in natural populations is
often difficult.

	There are optimality models of natural selection. From these, one can
predict what the optimal behavior for an organism should be. For
example, you may want to predict how long an organism should forage per
day when predators are present at some frequency (and therefore the
probability of getting caught is something per unit time during foraging
and something per unit time for hiding). If you further assume that
there has been sufficiently fine-grained genetic variation in the
population and enough time to reach equilibrium, you could predict that
the population should have evolved to the predicted optimum. There was a
review paper in Nature a few years ago assessing the usefulness of
optimality models.

	The are game theory models of selection. If you know all the strategies
being played in a population and have an accurate estimate of the
pay-off matrix, you can predict what the evolutionary stable strategy
is, whether a given strategy can invade other strategies and so forth.
IMHO, game theory models have limited application in biology. I could be
wrong. 

	There are also tests to see if natural selection has acted on a locus.
(A locus is the location of a gene.) These are based on inter- and
intra- specific comparisons on nucleotide diversity. Natural selection
depletes genetic variation at a locus. I explain a little more about
these types of tests (and give more information about natural selection
in my FAQ I wrote for the talk.origins archive. It's titled "An
Introduction to Evolutionary Biology." It's at:

http://www.talkorigins.org/faqs/faq-intro-to-biology.html

I've set the follow-ups to talk.origins. Please Cc me if you responding
but directing the response elsewhere. Thanks.

> Jon

Chris Colby