On the nature of the Gardner transformation