On the number of categories in biological classifications