Original paper

Interpretation of Bootstrap Values in Phylogenetic Analysis

Wiesemüller, Bernhard;Rothe, Hartmut


Bootstrap Analysis is a common tool in cladistics, and consequently many authors tend to believe that it could be close to a test of monophyly. In fact, it is only a procedure to calculate the redundancy of a certain character pattern among taxa. To demonstrate this, we set up a study with questionable data:Four skulls of great apes and humans were digitally photographed, and the pixels'brightness values were simply transformed to a one-zeromatrix, which was then used to calculate a Wagner tree with PHYLIP. As a rule, the higher the resolution of the photos is, the higher are the bootstrap values of supported taxa(and the lower are the bootstrap values of non-supported data). Redundancy of intertaxic information might indeed be an indicator of phylogenetic relationship, but can also be due to other reasons, like functional-adaptive needs in morphology, or semantic needs in a DNA-code. As a result, we tend to believe that high bootstrap values are actually less important than low ones. It is safer, based on a low bootstrap value, to claim that a certain taxon is not well supported by certain data. Therefore, we recommend discussions of low bootstrap values in future publications.