The variance in dog breeds is a result of selection through protection. Humans observed variances occurring in the dog population and made sure that those variations survived and were replicated. Variation that would normally result in replicative failure was selected for and supported by human intervention. This protection, something seldom found in nature, along with the unnaturally small population size, is what produced the speed of variation. In other words, dog breeds are the result of selection for difference not survivability. Interestingly enough this human selection also restricted the 'difference' to the humanly 'desirable'. Any variation deemed 'desirable' was reproduced with (relatively) remarkable speed, yet any variation that would lead to any 'undesirable' change was/is terminated. This is why we have a great variety of subspecies but no obvious speciation.
Note: The speed of variation within Canus familiarus has spanned a greater expanse of time than the 6000 years of the Bible yet shown much less variety than would be necesssary in the Noachian flood scenario.
Right. That would be ideology speaking, that last line. Dating methodologies being what they are..