Getting back to nature: a reality check for experiments in controlled environments