Similarity indices for spatial ecological data

We present a method for assessing similarity between species maps of presen ce and absence or abundance that emphasizes global features while ignoring minor local dissimilarities. The method arranges sites into small groups, o r cliques, and allows controlled changes to be made within cliques to reduc e the influence of local discrepancies. Resulting: measures of similarity a re visually more satisfactory than traditional indices. We show that the si milarity indices are useful for model selection by comparing observed spati al patterns with those predicted by different fitted models. Examples are p rovided for spatial distributions of oribatid mites (Acari, Oribatei), wood larks (Lullula arborea), and red deer (Cervus elaphus).