#1: Basically, you’re right. Heritability is population-dependent. Two different populations, living in different conditions, may have different values of heritability for the same trait.

That’s one of the reasons why height is so useful as an example, as Razib said. Everybody understands that malnutrition will dramatically affect your height, and yet height is 90% heritable in western countries – simply because, as the environmental variation is low, mechanically the genetic variation has a higher impact on total trait variation.

Notice that you can already infer this from the simple definition above.

Probably not. If height was controlled by a single gene, then yes, in a scenario like this, you might lose the ‘tall’ allele. But height is additive. There will still be variability in this new population, it will just be variation around a different mean.

Thanks for exposition. As one who has mental blocks in this area let me ramble a little. Let us take your height example. Suppose, let us say in a population of cattle, to make this only slightly less gory, before they are old enough to breed they pass under a killing fence. If they are tall enough they are culled and become veal. Let us further say average height of the herd will become smaller and if that herds genetics are compared to another there will be a definite variation. This whole thing really shouldn’t affect the amount of heritability of height in the cattle correct?

Let us say we do this for thousands of years such that the several of the genetic variations that cause tall cattle are lost from the population totally and nobody noticed that. According to the definition wouldn’t there now be less heritability of height measured in cattle by the folks who didn’t know there were currently non-existing possibilities?

I contrived this funny scenario because I feel like on the one hand the “heritability” factor could be partly caused by the environment, or rather it was caused by the environment in the past by the pressure that created the current genetic profile of a population, and it (heritability) only looks at the current state, not possible other genetic profiles that are not in the population under study but that might be “available” under correct conditions.

Does this make sense? These points make the heritability only really useful as a pointer for further looks. I think I get the base mechanics of the model: height is partially due to genetics, such that even if constraints like amount of food are removed, different genetic profiles are correlated with height in ways that have decent predictive value. But I guess I feel like there are always non-genetic constraints (or unconstraints, like the assumption of good nutrition) that can affect a trait. Thus the heritability correlations are just a little part of a bigger model to me, and I am not sure that they are the right way to split things out.

I this this is just my understanding issue.