"By selective breeding, we are seeing traits develop that 'might' have developed accidentally over time in random evolutionary ways, but can be controlled and found in just a few generations."

This indeed proves what can be done through selective breeding, but falls far short of proving what occued in a real-life, "free range" situation. Unfortunately, such breeding often reinforces meducal problems, making long term survival success (including the generated adnormalities) in the wild less likely.

One related quote I once heard was "if one let all the inbred dogs run loose and freely breed, in a few generations all of the inbred features would dissapear and the result would be dogs that mostly look like the dogs running wild in many third-world countries today."