A new statistical approach for assessing similarity of species composition with incidence and abundance data