I have no idea about elision (that's well beyond my level of knowledge at the moment -- the Catullus excerpts towards the final chapters of Wheelock are giving me enough headaches as it is
), but that's not quite the rule I had learned from "A Latin Grammar" by Charles Bennett (available here at Textkit) for attaching clitics. That rule was this:
1. If, after attaching the clitic, the stress falls naturally on the syllable before the clitic, then that syllable is stressed. Example: discipulōque (stress on 'ō')
2. If, before attaching the clitic, stress was on the antepenult, it is moved to the syllable before the clitic. Example: discipulaque (stress on 'a')
3. Otherwise, stress does not change. Example: portaque (stress on first syllable, not second). By this rule, your 'servāreque' has no displacement at all!
I checked A&G, too, and it does indeed provide the rule that you gave, and given the examples in the book, it would indeed choose 'portáque', not 'pórtaque'. So it seems that nobody is fully sure how clitics affect stress, or at least, nobody was sure at the beginning of the 20th century. Maybe that's not the case now, but we'd probably have to venture into the world of academia to be sure.
Sadly, Wheelock doesn't explain how clitics affect stress at all, as far as I could find, which is unfortunate since it's one of the most popular Latin books, and the one I'm currently using.