Additive trees in the analysis of community data

The paper advocates a more extensive use of additive trees in community ecology. When the distance/dissimilarity coefficient is selected carefully, these trees can illuminate structural aspects that are not obvious otherwise. In particular, starting from squared distances based on presence/absence data, the resulting trees approximate relationships in species richness, a feature not available through other graphical techniques. The construction of additive trees is illustrated by three actual examples, representing different circumstances in the analysis of grassland community data.