I need to correct my first comment. The stem pushes directly on the disk in this diagram.
The nut may be welded to the disk but neither are welded to the stem. If you look at your drawing closely there is a flange at the end of the stem. The disk nut sits above the flange and the disk below. The disk nut has a flange on top that bottoms out on the disk. There is some clearance left between the disk but and stem flange so that the disk will be loose on the end of the stem. The nut may get welded to the stem to keep it from unthreading.
The disc nut allows the disc and stem to be replaced separately. Depending on flow conditions and media it is likely that the stem can long outlast the disc. If the disc was welded to the stem it would need to be replaced with the disc. With the nut you are likely to get at least 2 or 3 trim changes with each stem.
It looks like the disc nut secures the flange on the end of the stem so the stem can pull the disc and disc nut up. The disc nut has it's own flange so that it cannot be tightened sufficiently to squeeze the flange on the stem.
This is what allows SPLD310's explanation to work. Plus, it allows the stem and disc to be different materials and means that the disc can have static friction against rotation at the final amount of force is applied by rotation of the stem, saving the sealing surface from being gouged if there is a small piece of debris that gets trapped.
