Hardiness means a lot of things, but mostly it's about the ability to tolerate change. In aquariums, the main changes are temperature, water chemistry (pH/GH/KH/TDS), and water quality (ammonia/nitrite/nitrate). There's also disease immunity, but that's a lot harder to test or quantify.
There aren't really genes for "adaptability to X". You have genes that code for certain cell enzymes or a larger liver. Those genes are lost or mutated just as easily as others, so if the shrimp that produce the next generation happen to have lost the genes for a certain antibody-producer, then that trait will be lost.
Additionally, it's not as simple as one gene = one trait. Genes are often "tangled" together (genes that cause Trait X and Trait Y are side-by-side on the chromosome). For example, the gene that causes red legs may be right next to the gene that determines the shape of the intestines, affecting the shrimp's ability to absorb nutrients. Genes can also have effects throughout the body instead of just one spot. For example, the protein that's needed to produce brown color may be the same protein that's needed to excrete excess ammonia from the gills.
Shrimp breeders have sought the "ideal" conditions for shrimp, then bred shrimp in those conditions for hundreds of generations. The current shrimp are extremely inbred, so even though the "ideal" may have been wider than breeders thought, any genes that would allow them to adapt to other conditions are likely gone.
We know the key to adaptability on a large scale is population variation. Simply having a lot of shrimp from different sources increases your chances of different genes. We also know that mutations happen at every generation. In theory, if you take a large number of inbred individuals and let them breed freely for several generations, mutations can bring back some of the lost traits.
Let's say you want to create bulletproof red cherry shrimp. I would start by getting as large and as varied of a population as possible. Buy many shrimp from many sources, then breed them all together, letting them mix as much as possible. Then I would get some wild (not wild-type, but actually wild caught) shrimp. Breed them to the reds in a separate tank. The F1 generation will appear wild-type, but will have some of those extra genes for red color. Now, here's the tricky part: you need to breed those F1s to reds, but you can't let the population constrict. You might also need to split the population so you have multiple lines going at once.
I would do it by combining shrimp from the different lines based on sex: breed several dozen red females to several dozen F1 males, then take the adults out before the next generation has a chance to breed with their parents. Take the resulting F2 males and breed them to new red females. Repeat until the Fs are reliably red. Run several of these lines in parallel so you can cross between them later.
Of course, you will need to cull to some degree, but you have to be careful. You want lots of breeders and lots of babies.
While this is happening, don't baby the Fs in terms of temperature, water chemistry, and water quality. You'll probably need to baby the original reds, though. You'll have to keep the breeder tanks in ideal conditions while the red parents are there, but then vary the conditions after removing the parents.
Also, try to identify and breed your longest-lived shrimp. If you have a shrimp that seems to be tougher than most, try adding it back in for another round of breeding (this would be a backcross). Or you could just start a tank of your oldest/largest shrimp and let them breed freely, developing an "old-timers" line which you could use for occasional outcrossing.
***************
A small-scale version would be to get a few wild-caught shrimp and breed them with some reds (preferably purchased from different breeders). Take that cross, and split it in two: one line will be repeatedly outcrossed to more reds as above, while the other will be line bred for red color. So you'll need three tanks: a tank of red shrimp, a tank of wild-crosses to be bred with the red shrimp, and a tank of wild-crosses line bred for red color. I would be very curious to see if the wild crosses started producing reds without any additional red crosses, but even if they don't, they'll be useful for outcrossing.
***************
Or, if you have the money, just hire some scientists to figure out exactly which genes control the red color, and then genetically engineer a bunch of wild shrimp. Assuming the red genes themselves don't cause a loss of hardiness, you should have very hardy shrimp with red color. You could breed these shrimp as their own thing or cross-breed them with existing reds to invigorate them. (This is one huge advantage of GMOs that people don't know about - you get the traits you want without the problems associated with inbreeding. Unless, of course, your starting population is inbred and/or you inbreed the resulting modified organism.)
