If we know any two of the selection intensity, the heritability, and the response to selection, we can calculate the third:
Response = Intensity * h2
A population can stop responding to selection and then suddenly begin responding again, due to any of:
These events are difficult to predict.
Inbreeding is preferential mating with kin. The most severe form is self-fertilization.
The effect of inbreeding is to reduce heterozygosity. We can define an inbreeding coefficient f representing the strength of inbreeding, where f=1 is complete self-fertilization and f=0 is a randomly mating population.
The coefficient f can be thought of in terms of dividing the population into a self-fertilizing part and a random-mating part, but f works equally well in cases where no actual self-fertilization is possible. It is basically the proportion of offspring which are produced by pairing two genes recently derived from the same ancestor, as opposed to genes randomly drawn from the gene pool.
For a given f we can calculate the new genotype frequencies, which are different from H-W:
P = p2 + fpq
H = 2pq - 2fpq
Q = q2 + fpq
Inbreeding does not directly change allele frequencies, though it can change genotype frequencies drastically. However, selection on an inbred population may have different results than on a random-mating population, since selection acts on genotypes.
The most drastic case is heterozygote advantage. As we have seen, with random mating any heterozygote advantage defines a stable polymorphism. With increasing inbreeding, however, the population is pushed toward one of the homozygotes. At f=1 the more fit homozygote will sweep the population; only if the two homozygotes have exactly the same fitness can the polymorphism be maintained. Intermediate values of f make the area of stable polymorphism smaller and smaller; if one homozygote is too much better than the other the population will fix on that allele.
Inbreeding depression is the observed loss of fitness in inbred plants and animals. Several forces can cause inbreeding depression: heterozygote advantage is an obvious one, but there is also the possibility (Peter will cover this) that the inbred population fixes bad alleles. Inbreeding depression is most severe (sometimes lethally so) in organisms that do not normally inbreed, and less severe or even absent in organisms that normally inbreed. Inbreeding species may lose heterozygote-advantage alleles and recessive deleterious alleles so that they face no further difficulty inbreeding.
Inbred strains of mice, produced by brother/sister mating, can appear quite healthy once established. However, most lines are lost in the first 20 generations or so. The major problem faced by established strains of inbred mice is that they are prone to epidemics; any disease infecting one will tend to infect all.
It is controversial whether humans instinctively avoid inbreeding. Most cultures avoid brother/sister mating, but some actively prefer first-cousin mating (f=1/16).
The inbreeding coefficient f can be thought of as the proportion of an individual's alleles which are identical by descent from some recent ancestor (and thus must be homozygous).
You can calculate inbreeding from a pedigree by reckoning the proportion of gene sharing up one side and down the other, though this is tricky in large pedigrees. Be sure to make a circuit for each common ancestor. Also be sure to distinguish between the relatedness of two individuals and the inbredness of their offspring (the inbredness of the offspring is half of the relatedness of the parents).