On the limits to secondary production