If you can imagine building a house, there are numerous subcontractors working on it, the framer, the electrician, the plumber, the bricklayer etc. Now each worker gets two sets of plans for his part of the job. Of course there are a number of other sets of plans or blueprints for this kind of job back at the main office, but for the job in this house, the worker only gets two of the many possible plans.
A gene is like a long blueprint given to the subcontractors in the body. If there is a mistake in the blueprint, the worker either does the job incorrectly or he doesn't finish the job. All the possible plans (or genes) that one might use for a particular job are termed alleles. In this case, the worker gets two sets of plans (genes) of the many possible ones (alleles), and he puts one set in each hip pocket. If one of these sets of plans gets damaged, the worker still has a good set, and the work can proceed on schedule.
The pocket is the analagous to the "gene locus". The locus is an actual location on each of a pair of chromosomes in the cell. Each cell has multiple pairs of chromosomes, - the number is species specific. Each chromosome contains many loci for all the jobs that have to be done in the body. There is one allele, specific for each job, residing at the identical locus on each of a pair of chromosomes, - just like the two sets of blueprints in the worker's hip pockets. The animal got one set of those paired chromosomes from each of it's parents. In the carrier state of a recessive genetic mutation, the animal has gotten one good copy of the gene in question, and one defective copy. In this case, one good copy is sufficient for the work to proceed and the animal to be healthy.
Genetic diversity in a population means that the population contains most of the possible alleles (alternate sets of plans) for a particular gene locus rather evenly distributed throughout the population. (Of course an individual animal can only have two of those alleles, for every gene locus, in his or her private collection.) This is where you start getting into trouble in an isolated population, such as in quarantine countries, or in very small countries. A very popular stud arrives in the country. Soon every bitch in the population is bred to him. Suddenly every puppy in the country has one or the other of his alleles for every gene locus. This would be bad enough in itself, but even worse, by breeding every bitch to the one male, you have selected AGAINST all the other alleles in the adult male population. These alleles will disappear unless they are handed down to offspring. This is genetic death. (death before having a chance to reproduce). Genetic diversity has been lost.
Let us suppose that there were 200 males, and 200 females in our hypothetical breeding population. But all 200 bitches were bred to Mr. Wonderful. Now, instead of having 200 different paternal gene collections represented in those 200 litters, we have only one. Now someone says, "We got such great type from this dog, let's breed his daughters to him as well." Now you are throwing away another whole set of alleles.
Even assuming Mr. Wonderful is perfect - an unlikely event that he carries no hidden defective alleles, since the average is around 7, many alleles carried in the former population have been ruthlessly discarded by overusing Mr. W. We have no idea what they are until something bad happens in the population. For instance, some genes MUST be carried in a heterozygous state for the animal to have an intact immune system. In an inbred population, this heterozygosity is lost. Modifiers that keep otherwise bad genes from being expressed can be lost. Genes responsible for subtle changes in longevity or fertility can be lost. Numerous studies have shown that declines in fertility are directly linked to the degree of inbreeding. So even Mr. Perfectly Wonderful is dangerous if he becomes the father of the majority of a population!
Now suppose this dog carries a hidden defect. In the first 200 litters sired by Mr. Wonderful, one half of the pups are carriers of that same gene. In selecting for the gene set that this dog carries, you have selected FOR the defective gene, and AGAINST a "good" one which may have been in the population before Mr. Wonderful came along.
Lets say that by now, (which is usually the case in popular sire effect), the overall carrier rate for the bad gene is 50%.This means that every other animal in the population carries at least one copy of Mr Wonderful's defective gene. The chances are 1/2 that any dog carries it and 1/2 that any bitch carries it. The chances are 1/2 x 1/2 or one in 4 that two carriers will be mated, even if we bred them randomly. But by now Mr. Wonderful's "look" has become a showring necessity. In all likelihood, we will be selecting, for our breeding, that dog and bitch who look most like Mr. Wonderful. The very plain pair have been neutered and placed as pets.
Again we have selected for the genes of Mr. Wonderful and against the other genes in the original population. But the pair resembling their famous sire are the very ones most likely to be carrying the heaviest helping of Mr. Wonderful genes, both good and bad. So the chances are that more than 3 out of four litters will produce carriers, and one out of 4 litters will now have affected puppies. In each of these affected litters, 3 out of 4 pups will carry the defective gene. Only one in four pups in each of these litters will be "clean".
What is the solution? The most obvious response is "don't breed any dogs carrying that gene". Now suppose we decide we will not permit breeding of affected dogs or carriers. Abruptly, we eliminate from our breeding the the 50% of the population which carry the defective gene. Which dogs are most likely to be carriers? The ones that look like Mr. Wonderful, of course. What is now left in this genetically decimated population? Precious little, perhaps not enough to go on with. There may be genetic problems that have surfaced because of the lack of "good" genes in the population, neonatal losses, failure to thrive, allergies, etc.
What happens, on the other hand, if we outcross to a line which has never had this problem? Suppose we assemble a number of unrelated dogs, - lets call them all "Mr.Clean" - and breed each of our bitches to a different Mr. Clean. Again 1/2 of the bitches which are bred to our Mr.Cleans, are carriers. This means that they have one "bad" copy of the gene, and one "good" copy. Since the sires have only good copies of the gene, half of the pups from the carrier mothers will be carriers, but NONE will be affected. In a late onset disease, identifying affected animals before they are bred will be a problem, but testing for the disease before breeding will reduce the number of litters an affected animal is likely to have. By using mainly outcrosses, even the accidental use of affected dogs will not increase the prevalence of affected offspring.
What about other problems? If this group of sires carry undesirable traits, they will each have a different collection, - alleles different from the inbred population and from each other. The likelihood of two "bad" copies of any gene getting together is thereby markedly reduced. And this infusion of new genes has introduced new genetic diversity into the impoverished line.
Put these two approaches together. Let's say we do not intentionally breed any animals who carry two copies of the defect, and we only use known carriers when there is a good reason to do so, and then ONLY to unrelated animals. This approach allows the gradual removal of the defective gene from the population, and the preservation of most of the good qualities of good old Mr. Wonderful. It restores a healthy degree of genetic vigor to the population, and lessens the likelihood that other "bad" problems will manifest themselves.
This is what the "genetic diversity" approach to breeding is all about. The "wholesale genetic slaughter" method may be appropriate in a genetically diverse population with only an occasional individual case of the disease in question. (Of course you rarely have this kind of problem in a genetically diverse population.) You get into the heavily "loaded" state only by inbreeding. Using the "wholesale genetic slaughter" method, even if it were practically possible, will not cure the problem in this inbred population, because the real culprit is not the defective gene, but the inbreeding. Besides you have to wonder what you will have left when you are finished. The dogs are still inbred, with all the problems that go along with inbreeding. In fact, they are twice as inbred as when you started, because you threw away half the genes. You won't have much of the genetic disease you select against, but maybe you won't have dogs either.
A reasonable course of action in any genetic disease demands that attention be paid both to removing the gene where possible without seriously degrading the genetic viability of the population, and taking steps to provide increased genetic diversity if it is needed.