I’m super interested in this sort of study! However, it looks like n=62 here, which I think weakens the results —they’re probably just useful as suggestions of possible effects. Also, any food is expected to have similar effects on the microbiome. They didn’t test caffeine in isolation. In some ways that’s better (I don’t consume caffeine in isolation), but in some ways that’s less useful (it’s possible you get similar results from many random vegetables)