>Granting that there is some variation in the rate at which environments
>change, the question is: how much variation is that? If it is just one
>order of magnitude on average, then yes the puntuated equilibrium hypothesis
>is correct in some some sense, but the "small changes dominate" evolution
>view is also correct. It's just that "small" varies by an order of
>sometimes tiny and sometimes very tiny. If environment rate changes vary
>by ten orders of magnitude, on the other hand, then maybe it's a new world.
I should have been more precise by saying that punctuated equilibrium
in *speciation*. Different but closely related species differ in measurable traits, but the differences typically arise from only a few genes. Gradual changes in environment, by Orr's model, wouldn't produce the exponential trends in gene effects, but then they don't seem to produce new species either.
Orr's model really wouldn't be useful when the rate of onging environmental is comparable to the rate at which the gene pool adapts.
>I'd have to go look up those QTL data papers to see how many orders of
>magnitude their data varies over. But I'm not optimistic - Orr brags that
>one of the data sets covers 82 traits, which isn't even enough to see a
>half an order of magnitude.
The variation in effects is a variation in the effect of genes on a given trait, so hypothetically even a study of one trait could cover an arbitrarily wide range of magnitude. In practice, the sample sizes go up with the inverse of the minimum detectable effect, so I doubt *any* study covers a large range. However, it is known that the most important genes do roughly follow the exponential model, and the combined effects of the most important genes explains most of the variation between species. So even if small-effect genes are affected by environmental variation and don't follow Orr's model, they're not important for speciation, because the total effect of all minor genes combined is fairly small.