"The potential hazards addressed in this EA center on the likelihood and consequences of AquAdvantage Salmon escaping, becoming established in the environment, and spreading to other areas. These hazards must be addressed for the production of eyed-eggs, grow-out to market size, and disposal (i.e., of fish & fish wastes)."
"As discussed in §184.108.40.206, the estimated escape rate of salmon from sea cages is about 1%. Sea cages, or net pens, have a direct connection with the aquatic environment."
1% of 50.000-90.000 fish in sea cage it's not something negligible, the first direct impact might be against the non-GE Atlantic salmon (salmo salar). Especially that salmon eats salmon eggs and so the dominant specie might become the GE one.
You might say, oh it's fine it's not touching the diversity of Salmon, it's just replacing the Atlantic salmon with another one. Wait, they took some coding sequence from the pacific salmon (and especially from the oncorhynchus tshawytscha) and added the "anti freezing" protein from zoarces americanus. So this "subspecie" got an interesting level of properties to find its place in the atlantic and/or in the pacific (where the diversity of salmons (Oncorhynchus) is much higher) region. Those risks are without any spreading of the genetic modification (assuming that triploid induction is effective with a probability of 1, another point where the scientific literature is lowering down the probability of effectiveness).
I suppose those risks are not really considered by the FDA as critical because the F is for Food in FDA. So the risk of changing the whole profile of wild salmon with such "GE" salmon is not negligible.