My bad.
So, that paper effectively has beneficial mutations happening more often, simply because they're beneficial?
If you're in a bad environment, then any mutation is more likely to be beneficial instead of um, not-beneficial, as compared to an okay or even good environment.
Like: if every one of your alleles is perfectly suited to the environment, any change will be harmful ('living fossil' species for example, perfectly adpated to their nche, so they remain unchanged for millenia). If none of your alleles are suited to the environment, you're dead.
But if you're only 70% suited, then there's a 70% chance a mutation will be bad (ignoring the whole "no effect" mutations that don't change the protein shape and stuff), which would mean a 30% chance of any mutation being beneficial.
If you're only 30% suited, then there's a 70% chance of a mutation being useful or beneficial. IOW something living in a bad environment/one it is less suited for is more likely to get a good mutation when compared to the same organism in a good environment.
So, the closer you are to being suited to the environment, the less likely you are to receive a good mutation. The further you are from being suited, the more likely you are to receive a good mutation. I recall Percy's evo algorithmy thing. That indicated the same effect: as you approach the target, you become less and less able to get good mutations.
Which means bacteria in the bad place will of course end up with more beneficial mutations. They hit rock-bottom already, so the only way through is up.
If none of that made sense, then great, I'm not a biologist anyway
if it did, I do get lucky sometimes.