Is natural selection a meaningless tautology?

In argument with creationists one often hears the point made, triumphantly, that natural selection is a tautology. The implication is that evolutionary biologists are so generally incompetent that they don't recognize this, and that when they invoke the concept of fitness they are engaging in a meaningless circular argument. Natural selection, creationists argue, is supposed to be "survival of the fittest". Yet how do we define fittest? Why, merely as those that survive. So it reduces to "survival of the survived", and thus cannot be more than a meaningless incantation. Gotcha!

But ...

... it's not a tautology. First of all, an aside: fitness is not merely "survival". For some reason creationists are fixated on survival and do not realize that evolutionary biologists also count in fitness the expected amount of reproduction of those parents that survive. But that technicality aside ...

I saw on a blog (here) a good analogy: Is it tautological or circular or meaningless to say that when we make up a track team we should choose runners who are faster than the others? Is it meaningless to look at earlier races or practice sessions to see who are the best runners? Obviously this is not meaningless, because ability to run fast often carries over from one race to another, even though this will not always be perfectly true.

It is when we expand our view beyond the single sample of individuals and the single generation that we can see that fitness is not a tautology. We can, for example:

1. Assess the fitness of different genotypes in our initial one-generation study and look at whether different genotypes have different fitnesses. And then we can validate the results by looking at how fit those genotypes are in additional samples of individuals in the same generation, or looking at samples in other populations or in later generations.
2. We can do the same for phenotypes: for example, are larger individuals more fit? Do hawks with longer wings catch more prey? Does this carry over into other samples, other populations, other generations?
3. We can ask what will be the consequences if the relative fitnesses (the ratios of the fitnesses) of different genotypes remain the same over many generations. Starting with initial gene frequencies, there are standard equations of theoretical population genetics that predict long-term outcomes. And we can calculate what will happen if fitnesses vary between populations, or vary across generations.

My colleagues in evolutionary biology are not idiots. They devote a lot of time and effort to measuring fitnesses of different genotypes and phenotypes, and in different populations and in different generations. Right now "experimental evolution" studies which mutate sequences and then ask how those genotypes change frequency in cultures of yeast are very popular, now that DNA sequencing of large samples can be done. These are popular methods for assessing the effect of changes in the sequence. There are good reasons for all of these studies, and calling fitness a tautology will not make any impression -- it certainly won't cause these studies to come to a screeching halt.

And in case you need more elaborate philosophical consideration of this issue, here are two discussions you can access online:

1. John Wilkins's discussion of the charge of tautology at the Talk.Origins Archive site,
2. In the Wikipedia article on "Survival of the fittest", the section on this issue.
`