Trees have been published in the literature based on molecular sequence data which are not plausible to systematic biologists. We show that the method of "minimum conflict phylogeny estimation" (mcope) [Fuellen and Wägele; Fuellen] yields more plausible trees. It avoids to pull slowly evolving taxa into one group based on the "old" uneroded character states that they still share wherever the other taxa were subject to change.
Using data previously analyzed by Spears et al. [Spears et al., 1994] and Friedrich and Tautz [Friedrich and Tautz, 1995], we show that branchings which occur in 86% of maximum parsimonious resampled trees, or which enjoy 77% bootstrap support employing maximum likelihood, are not the ones displaying a minimum amount of conflict due to "novel" character states that can be found across the split, i.e. in taxa from both groups. Instead, the split with minimum conflict is the one that was deemed most plausible by systematists before.