Parsimony analysis of endemicity describes but does not explain: an illustrated critique

Aim To demonstrate that parsimony analysis of endemicity (PAE) is not analogous to a cladistic biogeographical analysis. Location We used six data sets from previously published studies from around the world. Methods In order to test the efficiency of PAE in recovering historical relationships among areas, we performed an empirical comparison of nodes recovered with PAE, primary Brooks parsimony analysis (BPA), and an event‐based method using three models (maximum codivergence, reconciled trees, and the default model of the treefitter program) for six data sets. We measured the performance of PAE in recovering historical area relationships by counting the number and examining the content of nodes recovered by PAE and by historical methods. The dispersal/vicariance ratio was calculated to assess the prevalence of dispersal or vicariance in each reconstruction and its relationship to the performance of PAE. Results Our results show that PAE recovers an average of 17.25% of historical nodes. PAE and BPA tend to provide similar results; however, in relation to the event‐based models, PAE performance was poor under all the tested scenarios. Although in some cases PAE reconstructions are more resolved than historical reconstructions, this does not necessarily mean that PAE produces more informative answers. These additional nodes correspond to unsupported statements that are based solely on the distributional data of taxa and not on their phylogenetic history. In other words, these nodes were not found by the historical methods, which take phylogenetics into account. The number of historical nodes recovered using PAE was in general negatively correlated with the dispersal/vicariance ratio. Main conclusions Our results show that PAE is unable to recover historical patterns and therefore does not fit into the current paradigm of historical biogeography. These findings raise doubts regarding conclusions derived from biogeographical studies that interpret PAE trees as area cladograms. We acknowledge that PAE aims to describe but does not explain the current distribution of organisms. It is therefore a useful tool in other biogeographical or ecological analyses for exploring the distribution of taxa or for establishing hypotheses of primary homology between areas.