>I'm afraid it can't be done. Where genetic engeneering is used to
introduce
>these "killer genes", it is the F1 and subsequent generation of the
>introduced organism that have the lethal factor, so only
descentants of the
>introduced species would be effected. If they did cross with a
local
>population then the gene would be quickly selected out - it is
after all
>lethal.
Now you just need to find something that could eat all the gambusia
in one generation, then die off itself!