Toward a physiological explanation of juvenile growth curves