I think, this is because the measured capacitance is equal to the capacitance of the oxide in the cases of accumulation and inversion. If you want to compare the properties of the depletion layer beneath the oxide of two devices, it is much easier for the normalized graphs.