Easily. Make the genetic machinery responsible for the trigger hideously complex, redundant and, if possible, self-repairing (perhaps via a gene drive), and have loads of other genes rely on specific bits of it, so that a single alteration in that machinery will make life suck for the mutant without much preventing the trigger.
Evolution can't make big changes, because there is no guiding intelligence behind it, only statistics; that's why our retinas are still backwards, and why a giraffe's laryngeal nerves take five metres to connect points ~30cm apart. Even though life without this massive lump of “junk DNA” that everything seems to rely on would work better, evolve faster, thrive more, waste fewer resources, reproduce more efficiently… it'll take a lot of mutations for it to unravel, none of which are selected for. Parts of it might get corrupted by sheer fluke, if the corruption also disables each anti-corruption mechanism and happens to coincide with a beneficial mutation, but that's what the redundancy's for.
How does anything "rely on specific bits of it" if it's noncoding? How does the gene drive remain intact across millions of generations' worth of mutation events? How are mutations in noncoding DNA selected against?
Easily. Make the genetic machinery responsible for the trigger hideously complex, redundant and, if possible, self-repairing (perhaps via a gene drive), and have loads of other genes rely on specific bits of it, so that a single alteration in that machinery will make life suck for the mutant without much preventing the trigger.
Evolution can't make big changes, because there is no guiding intelligence behind it, only statistics; that's why our retinas are still backwards, and why a giraffe's laryngeal nerves take five metres to connect points ~30cm apart. Even though life without this massive lump of “junk DNA” that everything seems to rely on would work better, evolve faster, thrive more, waste fewer resources, reproduce more efficiently… it'll take a lot of mutations for it to unravel, none of which are selected for. Parts of it might get corrupted by sheer fluke, if the corruption also disables each anti-corruption mechanism and happens to coincide with a beneficial mutation, but that's what the redundancy's for.
You'll never keep the phenotype the same, but you can keep a certain mostly-useless genetic mechanism intact, à la Robin Hood and Friar Tuck: http://catb.org/jargon/html/meaning-of-hack.html