I get that, but I feel like we've known for quite a while that people who exercise can run longer and have lower body fat, regardless of genetics. Genetics can affect how long you can run compared to others, but you will run longer if you exercise than if you don't.
If you take anybody in the world and have them exercise, they'll lower their body fat, everything else being equal, no? And that's not only the same genes, it's actually the same person.
I thought, too, that genetics only controls for something akin to a few percent of your potential, and that to bump up against that limit you would have to already be in the 99th percentile for a particular capability. So your endurance is controlled by past training almost exclusively.
You can control for genetics by running a controlled study, where you take a group of people and randomly assign some individuals to exercise and some individuals not to. Then, by construction, the exercise and no exercise groups are drawn from the same genetic distribution. I'd be very surprised if no one has tried to do this, but there's probably a limit on how long you could run it for.
There's also the possible "placebo effect" confound, which is a problem for both a controlled study and a twin study, but might be a bigger issue for the former (where participants are told to exercise) than the latter (where participants exercise or don't exercise of their own volition).
Twin Studies are often more feasible (and sometimes more ethical) than a random control study. A Twin Studies can have results that qualify as "interesting" even with just a dozen or two groups of twins; a random control study on the same sample size would be discounted for its small p value. And with a Twin Study, you can sometimes do post-hoc analysis on years of effect, without waiting the years for those to occur.
This study may even be one of the ones that's unethical to do in a random fashion. We know that lack of exercise is bad for your health; telling people not to exercise for a year could be seen as doing active harm to study participants.
Of course, Twin Studies suffer from some issues on their own. I would prefer random-control tests where they're possible. But sometimes Twin Studies are all that you're going to get. And sometimes the Twin Study of an effect is the economic small-scale pilot that generates interest and funding for the "real" study, and there's nothing wrong with that.
