Back in their DSLR days (and a few compacts, as well), Fujifilm made sensors that had dual photosites per pixel position: one gathered highlight info, one gathered shadow info. These were combined to make a final image. 200% in that era meant "one stop more dynamic range," and 400% meant "two stops more dynamic range."
The XF cameras use more conventional sensors with a single photosite per pixel position. Thus, they don't perform Dynamic Range changes using extra data. Instead, a setting of Dynamic Range 200% would mean that the camera underexposes by a stop, then uses in-camera processing to bring the values back into the correct exposure realm. Dynamic Range 400% is the same thing, but with a two-stop underexposure.
While useful for those that don't want to do post processing themselves, Fujifilm and the other camera companies seem to be missing the point. Many of us long-time digital shooters (I'm now finishing up my second decade of digital shooting) have long asked for ETTR (expose to the right) settings. What that means is that we want the camera to set an exposure that does not blow out any channel. If you then add Fujifilm's Dynamic Range processing to that image based upon how much the mean exposure should have been versus what it was, you'd get a very useful camera function. At least as long as the camera performed well enough to avoid noise in the shadows brought up to visibility.
Using Dynamic Range Auto on an XF-series camera is asking for some random noise propagation in the darker portions of your JPEG images. Images where the camera doesn't make this adjustment would be as noise-free as you'd expect. Images where the camera boosted to the max might have visible amounts of unexpected noise in them, especially if you were photographing at a high ISO value to begin with. My advice is to set this function to Off except when you need it and are willing to live with noise propagation, especially at higher ISO values.
