Phylogenetic algorithms and the evolution of species communities in forest fragments

In forest fragmentation studies, low specific richness in small fragments and community nestedness are usually considered to result from species loss. However, except in the case of fragmentation experiments, these studies cannot distinguish between original low richness and secondary species loss, or between original high richness and secondary colonizations in fragments. To distinguish between these possibilities is a matter of historical inference for which phylogenetic algorithms are designed. The methods of phylogenetic analysis, and especially parsimony analysis, can be used to find a tree of relationships between communities from different forest fragments, taking the presence or absence of species among different communities as characters. Parsimony analysis searches if species subsets can be classified in a nested hierarchy, and also establishes how the communities evolved, polarizing species changes into either extinctions or colonizations. By re-analyzing two classical studies in this new and powerful way, we demonstrate that the differences between fragments and large continuous forests cannot be attributed to species loss in all cases, contrary to expectations from models.