Why is it not just a correlation? The sort of person who is motivated enough to do exercise is surely likely to also put more effort into what and how they eat. How could you even analyse for causation from a meta-analysis?

Yet they are strongly suggesting causation:

  Meaning: These results suggest that aerobic exercise training at least 150 minutes per week at moderate intensity or higher may be needed to achieve clinically important reductions in waist circumference and measures of body fat.