Dog breeding - evidence for or against evolution?

Dog breeding - evidence for or against evolution?
Three features of dog breeding are obvious. Firstly, dogs vary greatly. Secondly, this variation arose in a short time within a single species. Thirdly, despite this variation, breeding doesn't lead to obvious speciation.

The American Kennel club recognizes 152 breeds of dogs. These all share common ancestry with the wolf within the last 15,000 to 135,000 years. The relatively recent origin of dogs means that DNA sequence variation in dogs is similar to that in humans. This is less variation than one might predict from known anatomical variation.

The DNA sequence of the dog genome is 0.5Gb shorter that of the human genome. The smaller size reflects less repeated DNA. Roughly 95% of dog genes share the same neighbor as corresponding genes in humans (synteny). The differences reflect rearrangement from a common ancestor and are consistent with known DNA rearrangement processes.
Is artificial selection the same as natural selection?
Charles Darwin recognized that artificial selection was analogous to, and a good model for, natural selection.

In the ordinary world, plants and animals stay pretty much the same. This reflects adaptation to stable environments. To see selective change in nature one must look to changing (unstable) environments. Two examples are the intestinal tract, which is at the mercy of diet, and agriculture, where new crops produce new ecosystems.

In artificial selection, new selection pressures are maintained on breeding populations. Ultimately progeny with altered anatomy and physiology appear either from genetic recombination or through mutation. Phenotypic changes may be directly selected or appear as correlates of adaptive changes. During selection neither nature nor the breeder need understand the mechanics of the process. Variations in DNA sequence produced by artificial selection are entirely consistent with those arising from natural selection.

The anatomical differences among dogs demonstrate the range of possibilities arising from limited genetic differences. They also show that the developmental program can accommodate and coordinate major anatomical and physiological changes. We don't yet know the relative roles of recombination and mutation in the process. Mutation in a single gene (IGF) is known to play a major role in producing small dogs.
Why don't dogs evolve into new species?
In everyday experience, major anatomical differences suggest speciation. This however is correlation, not a cause-effect relationship. There are many examples of similar species, downy and hairy woodpeckers for example.

While dog breeding introduces major anatomical changes it has little effect on processes that might encourage speciation. For example it has little effect on DNA sequence, or chromosome structure. It doesn't lead to hybrid mortality. In theory hybrid mortality is possible. For example one could breed a group of dogs to digest a certain food and a second group to digest another food. If hybrids between these groups could digest neither food, then interbreeding would be selected against.
Is the fact that dogs never evolve into cats evidence against macroevolution?
Imagining that dogs could evolve into cats indicates fundamental misunderstandings. Dogs and cats differ substantially in DNA sequence. Genome sequences evolve slowly (~0.5% per million years). Hundreds (or thousands) of years are insufficient time for changes of this magnitude.
Is the inability to produce this change by breeding evidence against macroevolution?
The fact that we can't reproduce something on demand isn't evidence that it didn't happen. In the case of dogs and cats we look at the fossil record and DNA sequences. Both lines of evidence are consistent with known mechanisms of evolutionary change. Most people accept that mountains can build up over long time periods because they can visualize the processes. Evolutionary processes are more abstract and difficult to visualize.

Dog evolution @ youtube

Genetics of dog behavior