Based on a compilation of OVI, CIV, SiIV and NV data from IUE, FUSE, GHRS, STIS, and COS, we derive an anti- correlation between the stellar temperature and the high ion velocity shift w.r.t. to the photosphere, with positive (resp. negative) velocity shifts for the cooler (resp. hotter) white dwarfs. This trend probably reflects more than a single process, however such a dependence on the WD's temperature again favors a CS origin for a very large fraction of those ion absorptions, previously observed with IUE, HST-STIS, HST-GHRS, FUSE, and now COS, selecting objects for which absorption line radial velocities, stellar effective temperature and photospheric velocity can be found in the literature. Interestingly, and gas in near-equilibrium in the star vicinity. It is also probably significant that the temperature that corresponds to a null radial velocity, i.e. \simeq 50,000K, also corresponds to the threshold below which there is a dichotomy between pure or heavy elements atmospheres as well as some temperature estimates for and a form of balance between radiation pressure and gravitation. This is consistent with ubiquitous evaporation of orbiting dusty material. Together with the fact that the fraction of stars with (red-or blue-) shifted lines and the fraction of stars known to possess heavy species in their atmosphere are of the same order, such a velocity-temperature relationship is consistent with quasi-continuous evaporation of orbiting CS dusty material, followed by accretion and settling down in the photosphere. In view of these results, ion measurements close to the photospheric or the IS velocity should be interpreted with caution, especially for stars at intermediate temperatures. While tracing CS gas, they may be erroneously attributed to photospheric material or to the ISM, explaining the difficulty of finding a coherent pattern of the high ions in the local IS 3D distribution.