Early in his career the famed evolutionary biologist William D. Hamilton had a strong interest in eugenics. In his autobiographical collection of papers Hamilton admits that he suspects these tendencies were the reason for the suspicion he aroused in some of the more senior scientists in Britain after World War II. But Hamilton later also admits that his earlier enthusiasms for social engineering through selection for "good traits" may have been wrong-headed, insofar as the selection pressures of evolution are protean, and what may be adaptive perfection in one age may be doom in another (or, in the world of international migration, you can substitute place for time). This does not mean that Hamilton abandoned his worry about increased "genetic load" in the human population (deleterious mutations accumulating in the human gene pool because the "unfit" now live and breed thanks to modern medicine). It is simply that such ideas and concerns can't be easily reduced into simple formulas and maxims, because evolutionary processes can vary in their implications over time and place. I thought of these issues when stumbling upon this curious comment over at Genetic Future in regards to preimplantation genetic diagnosis:
Genetic selection will help us to be more healthy in the mid-term, but more fragile and less adaptable in the long term. Take pure bred dogs for example. They are bred to exhibit certain traits, but can be more susceptible to certain diseases. It is said that if you want a healthy dog, make sure to get a mutt. We are all mutts right now. Genetic selection will make us all pure breds. May be seen as desirable, but there will be tradeoffs.
I've actually seen forms of this comment before, so it's not just one isolated confusion. There's an intelligible structure to the misimpression. The worry is that PGD is going to lead to genetic uniformity. For example, that everyone will wish to have children who are Nordic superman and the like. The confusion emerges because
the genetic uniformity is simply a side effect of the correlation between some genes and the phenotypes which we are trying to force to uniformity.
Therefore the sameness which is the aimed outcome of PGD and animal breeding programs is one of trait values, not genes necessarily. Inbreeding therefore is not an outcome of the drive for phenotypic similarity so much as a "quick & dirty" technique to generate lineages which express unique and distinctive characteristics. In the case of many animal breeds generations of forced inbreeding is a natural form of cloning, as mating siblings and parents and offspring repeatedly will read to rapid genetic and phenotypic homogeneity (more or less, I am aware of possibility of fluctuating asymmetry increasing with homozygosity). For many characteristics breeders don't know which region of the genome controls the trait value variation, and the most efficient way to perpetuate and amplify a character is selecting individuals which express the preferred trait. Because of the likelihood of the small size of the set of animals which are candidates for this process inbred lines naturally emerge (to be frank, some of the freaky looking dog varieties are clearly rare mutants). But in an ideal world with god-like powers you wouldn't do this. You'd look for individuals with the characteristics you wish to amplify, but make sure that they are not closely related. All things equal being inbred is not the best, as the lifespans of purebred dogs suggest. Selecting pairs based on a finite set of characteristics is termed assortative mating. In these pairs you have correlations of the value of the trait, and possibly the state of the genes (though remember that even at the same gene different variants can produce the same outcome). But you don't have to have widespread similarity across the whole genome which is the characteristic of inbred organisms to achieve your ends. Because individuals with the same genes who are related tend to have the same characteristics inbreeding and incest are by definition assortative mating outwardly, insofar as there is a correlation of phenotypes. But assortative mating is not necessarily inbreeding. So the confusion that the commenter above expressed is due to the fact that we intuitively use phenotype as a proxy for genotype, and so when you select for a very precise phenotype one presumes that one is selecting for the same genotype. Even granting high heritabilities, many traits are going to be controlled by only a small proportion of the genome. In other words the phenotype is not a perfect proxy for the whole genotype. And in the case of traits like height which may be controlled by variation at many genes and so is widely distributed across the genome the particular specific genetic architectures of two very tall individuals is likely to differ a great deal in the aggregate. Therefore I think it is a misconception to imagine that genetic screening for a finite set of traits is going to lead to a race of genetic clones. They may look the same, but the salient exterior physical characteristics we use to assess relatedness are only controlled by a tiny proportion of the tens of thousands of human genes. By and large PGD is going to sample the range of a couple's own fertilized eggs, so there will be a great deal of whole genome genetic variance across the Perfect Children.^TM