Thanks. Very interesting link which definitely suggests that you get just under 1 stop less dynamic range in relation to the ISO value in DX mode. I wonder if it is as simple as that or if the reduced sensor area has an impact on the dynamic range measurement. If this dynamic range reduction is for real it would be good to know why it happens.
As I understand it you could say it is the crop factor squared times the original iso value. So for DX vs FX 1.5 squared is 2.25. So ISO 2000 in FX would have similar noise to iso 4500. Dynamic range is primarily lost due to noise in the shadows. So as you say a little more than one stop. This is just due to the area of the sensor, not pixel size or density.