I've been reading this really interesting book called The Dog and it's Genome. One of the chapters is about how breed clubs have dealt with genetic disorders. They tell this story about the Portugese Water Dog in the US, which had a problem in the late '80s with a genetic disorder - with about 1/3 of dogs as carriers. A blood test was developed that allowed them to id carriers. The club, on good advice, recommended that breeders use a breed and replace approach - that they breed their good quality carriers to non-carriers and keep non-carrier progeny to replace the carrier parent while retaining their genes. But most breeders made what they no doubt felt was a morally correct decision and chose not to breed their carriers at all. Which reduced the gene pool markedly.
It was realised that this particular disorder came from one of the two main counrty of origin families in the breed, so breeders swung mainly to using the other line. But within a short period of time they realised that this family carried a kind of PRA - and that this occurred at a higher rate than the original condition.
They also talked about a family of Dalmations that were developed from Dal x Pointers to eliminate the kidney stone problem that Dals have (apparantly all Dals are homozygous for the gene that causes the purine absorbtion abnormality) - but after many generations (they started in the late 70s - the percentage of pointer DNA must be so tiny at this point) they haven't been able to get the big, round Dally spots - which suggests that the gene that causes the abnormality and the proper Dal spots are related.