The item to note in DXO's testing is the 14.4 Ev's of dynamic range greater then the AD converters specifications. The file format and image processing does permit 16 bits and indicates what DXO is picking up is Nikon's software at work. In the end, they really can't provide a sensor performance test, but more of a system performance result. The initial result can not be technically above 14Ev, but after some processing, room remains in the file format and processing to make it happen.
In truth, all of these tests are subject to lens choice, software in camera, and system settings. I do not believe that any company computer display that can actually allow you to fully view the dynamic range and color depth afforded by either camera. Printers also tend to be a greater limit as their Dmax is somewhere closer to 10 stops worth and gamuts tend a bit narrow in comparison.
So while we can see the readouts from the files that explain the detail is there, we really can not print or see it on screen as that part of the workflow still has a bit of catching up to do. The situation is fine as I would always rather have a source that has greater depth then trying to extract from a lower quality to output a higher quality. The Nikon will be superior in well lit conditions with images intended for very large prints. The Canon will be able to match or surpass it in almost other conditions and will provide its user a bit more flexibility in more challenging lighting.
Now the limit of the quality of images is not often limited by the camera, but more in the photographers ability in the art and understanding of tools and technology.