But I have never seen anyone's career affected by which journals they review for. It might happen, but if it does then it's news to me.
(BTW., for academics who feel they must publish in one of the tabloids, Science is much sounder than Nature because all their papers become freely accessible one year after publication, which is much better than nothing.)
Oh they absolutely are. Academic CVs are full of listings of which program committees you're on. You start being a member of a committee for a smallish conference (in programming language research, something like VEE or PLaS), then try to move up to a few medium size ones (CC, perhaps), and finally you make it to being on the program committees of the top top journals (PLDI or POPL).
I decided not to go down the academic career path after doing my PhD, but it was pretty plain even to a PhD student that being on good program committees is part of your career progression.
When I asked what motivates researches to volunteer their free time to do peer reviews, that was one of the answers I got. It's a small benefit.
One solution would be for universities taking stands like this is to somehow "give credit" to their academics who are asked to do peer review for non-open-access journals.