PDS watches are issued for events that are expected to either be particularly widespread or particularly intense. For example, PDS tornado watches are issued when strong/violent tornadoes appear to be a good possibility. Likewise, PDS severe watches are issued when widespread, very intense damaging winds are expected. I think this tends to happen when the SPC expects thunderstorms winds at or greater than 95mph. They may also issue a PDS for widespread events, such as a derecho plowing through an entire state. Most of the time, however, they seem to be issued for very intense events rather than very widespread events. Though unofficial, I have heard that PDS watches tend to be issued when intense storms are fast-moving and likely to affect more developed (urban, suburban, etc) areas. Indeed, fast-moving storms have the potential to affect more people in that they will affect more physical land area.
So, what would yield PDS tornado watches? Say strong/extreme instability, strong vertical wind shear, low LCLs, and favorable alignment of the deep-layer shear vector relative to the boundary providing the surface convergence (dryline, front, etc). You can pull up the synoptic and mesoscale setups from days in which PDS tornado watches were issued and see a pretty big difference from those days in which 'regular' tornado watches were issued. Same holds for severe PDS watches...
What about in terms of 'upper-air' conditions? I think tornadic environment are most contrasted with nontornadic environments in the low-levels (e.g. 0-3km or 0-1km helicity, etc). I think if you'd examine the "upper-air" (which I consider to be 300-250mb region), you'd find it pretty hard to see many differences between PDS days and non-PDS days. Yes, jet strength, trough axis alignment (negative tilt?, etc), all can tell you something about the lower-levels (strong jet streak aloft indicates a strong low-level thermal boundary via thermal wind relationship). However, I'd suggest looking at the low-level wind profiles, low-level moisture profiles (low LCLs, etc) and instability when trying to find the environments that differentiate PDS days from "regular" days.
